From owner-vxwexplo-process  Mon Feb  1 04:00:32 1999
From: daemon@csg.lbl.gov
Date: Mon Feb  1 04:00:36 PST 1999
Subject: comp.os.vxworks newsdigest

Comp.Os.Vxworks Daily Digest    Mon Feb  1 04:00:28 PST 1999

        Subject: Re: Looking for embedded HTTP servers
        Subject: Re: VxWorks port to Linux survey
        Subject: Ping Client
        Subject: Re: MPC860 and cache
        Subject: Re: Choosing a RTOS
        Subject: Proxies for SENS :any experience ?
        Subject: Re: absolute minimum start and init time
        Subject: Re: Interrupts despite intLock() - result
        Subject: Re: Choosing a RTOS
        Subject: compiling for 68060 on mvme172 board
        Subject: Re: WRS stock drop - what happened?
        Subject: Re: Interrupts despite intLock()

-------------------------------------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Looking for embedded HTTP servers
Date: Sun, 31 Jan 1999 11:53:25 GMT
From: sunnym@my-dejanews.com
Organization: Deja News - The Leader in Internet Discussion
Message-ID: <791g7j$kj6$1@nnrp1.dejanews.com>
References: <b5gr2.4$is3.117@news14.ispnews.com> <36B06E7F.6AEC626@ieee.org>

JAWS seems nice, but where are the sources?


In article <36B06E7F.6AEC626@ieee.org>,
  clarence m weaver <c.weaver@ieee.org> wrote:

[cut]

> Check out&nbsp; <A HREF="http://www.cs.wustl.edu/~jxh/research/">JAWS</A>
> a freeware http server that runs VxWorks, LynxOS, and others.
> <P>Tim Meagher wrote:

[cut]

- -----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: VxWorks port to Linux survey
Date: Fri, 29 Jan 1999 13:29:03 GMT
From: mcarr92365@my-dejanews.com
Organization: Deja News - The Leader in Internet Discussion
Message-ID: <78sd2m$dpe$1@nnrp1.dejanews.com>
References: <78qnh6$dd37@overload.lbl.gov>

Just for clarity, are you talking about a) the kernel, b) the host
developement tools, or c) both?

- --Mike

In article <78qnh6$dd37@overload.lbl.gov>,
  "Bruce O. Benson" <benson@sparta.com> wrote:
> In response to the recent lengthy thread on the hope (for some at least)
> that WRS would port VxWorks to Linux (or BSD for that matter), I am
> collecting names to present to WRS management as a show of strength for
> this.
>
> If you would like to be included as a supporter of this port, please drop me
> a note at mailto:benson@linux.sparta.com
> In fairness, if you think it's a dumb idea, I will collect your note and fw
> your name, too.
>
> I'll post the statistical results of the survey, and the WRS response in the
> next couple of weeks on http://linux.sparta.com/VxWorks
>
> Regards,
> Bruce.
> Bruce Benson, Senior Engineer,
> SPARTA (CPD)
> 7926 Jones Branch Drive, Suite 900
> McLean, VA  22102-3303
> USA
> +1 703-448-0210 x211
> benson@sparta.com
> http://www.mclean.sparta.com
>
>

- -----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

---------------------------

Newsgroups: comp.os.vxworks
Subject: Ping Client
Date: Thu, 28 Jan 1999 22:03:06 -0500
From: Derrick Longo <dlongo1@umbc.edu>
Organization: University of Maryland, Baltimore County
Message-ID: <Pine.SGI.3.96A.990128220212.992771B-100000@umbc9.umbc.edu>

I was looking around for a ping client for VxWorks 5.1. I haven't been
able to find one yet. Could anyone point me in the direction of source
code? Thanks, I would really appreciate it.


Derrick

....................................................
"...and I will touch the sky..."

		-- Rod Stewart, _Faith of the Heart_


---------------------------

Newsgroups: comp.os.vxworks,comp.sys.powerpc.tech
Subject: Re: MPC860 and cache
Date: 29 Jan 1999 06:25:44 -0800
From: andrew@user2.teleport.com (Andrew Klossner)
Message-ID: <78sgd8$a2n$1@user2.teleport.com>
References: <petekz-2701991950220001@dal-tsa31-56.cyberramp.net>
 <36B00C1A.414B@nospam.aracnet.com>
 <petekz-2801991927100001@dal-tsa12-61.cyberramp.net>

> if I enable the caches but the MMU is told that all memory areas are
> non-cacheable, how is that different that having the cache off?  It
> seems to behave differently.

I work with 603e, not 860, so this is of value only to 603e users who
happen to be reading the thread.

On 603e, the difference is: with MMU and cache off, stores can be
deferred until after a subsequent load.  When you turn on the MMU,
stores and loads are executed in order for cache-inhibited pages.
This can make a big difference in code like:

	stw	...		// select the other register bank
	lwz	...		// load from that bank

Performing the lwz before the stw in this case gives wrong results.

  -=- Andrew Klossner (andrew@teleport.com)

---------------------------

Newsgroups: comp.arch.embedded,comp.os.os9,comp.os.psos,comp.os.vxworks
Subject: Re: Choosing a RTOS
Date: Fri, 29 Jan 1999 04:45:25 GMT
From: gvarndell@hotmail.com
Organization: Deja News - The Leader in Internet Discussion
Message-ID: <78red3$l2v$1@nnrp1.dejanews.com>
References: <01be492d$f3e241c0$65020abe@ee02> <36ADFE9E.2BA991B5@ada.com>
 <78np50$46c@atlantis.metrowerks.com> <78nrkk$goq$1@news04.btx.dtag.de>
 <01be4a94$9be76140$0100007f@llga>

In article <01be4a94$9be76140$0100007f@llga>,
  "Llewellyn Griffiths" <nojunk@llga.com.mars> wrote:
> Didn't your mother ever teach you that it was rude to shout?
>
>  Perhaps you could enlighten us all with some of these 'rules' and your
> right to determine what is or is not acceptable.
>
> I personally found the article [to which you have lodged such an
> ill-mannered objection] both relevant & interesting, and I have filed it
> away for future reference.
>
> Since you appear to have a penchant for loud noise, perhaps your interests
> would be better served by directing your attention to one of the politics
> news groups or even comp.object.

Well-mannered and well written.
Still, Martin's reaction to the offensive, opportunistic, and YES
irrelevant advertisement (one of too many I've seen lately from this
particular company) was justified.
If you believe a plug for some also-ran, me-too compiler vendor is relevant
to RTOS selection, then you probably would consider it relevant to every
frickin' thread in all the offended groups.
Subtle indignant pomposity doesn't make you right.

gv

- -----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

---------------------------

Newsgroups: comp.os.vxworks
Subject: Proxies for SENS :any experience ?
Date: Fri, 29 Jan 1999 13:24:38 +0100
From: Sebastien Mallet <mallet@nortel.ca>
Organization: Nortel
Message-ID: <36B1A886.A859D7FF@nortel.ca>

Hi ,
I wrote an END  for the Sens Stack and I was wondering two things :
- - I can't find a proxy client for a non shared memory network...
anyone's seen this before ?
- - anyone ever ported  any proxy to vxworks ?
The aim of this developpment is to do multi-hosting with different
interfaces on the same network(not the gateway thing).
The target is a PPC board.
Any answers would be greatly appreciated.
Sebastien Mallet
mailto: mallet@nortelnetworks.com

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: absolute minimum start and init time
Date: Sun, 31 Jan 1999 12:25:43 -0500
From: "HP" <hmp@vrtelecom.com>
Organization: IBM.NET
Message-ID: <36b4929b.0@news1.ibm.net>
References: <36af4772.12810124@news-grtn> <ABHs2.5007$BS5.2971@news.rdc1.sfba.home.com>

You'll probably also want to customize your system configuration to strip
out all non-essential services, or at least defer their initialization until
later, if possible.

- -Henning

John Chann wrote in message ...
>I can't answer for the startup time but you might also try running one of
the
>ROM resident images, that will eliminate the ROM to RAM crossloading time
too.
>
>JRC
>
>In article <36af4772.12810124@news-grtn>, dsipe_NOT@oscsystems.com wrote:
>>Is there and app note or manual entry on how to minimize the time
>>vxworks takes to initialize?
>>
>>I am trying to make an embedded system that has a warm-start
>>requirement of less than one second. I know it depends on the speed of
>>the cpu, we would be using the fastest mpc860 available.
>>I believe this would include non-compressed code storage.  It might be
>>possible for us to keep the RAM powered during the off time so that
>>uncompressed code would already be present and ready to use.
>>



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Interrupts despite intLock() - result
Date: Fri, 29 Jan 1999 12:25:27 GMT
From: sa@ute.port.de (Dr. Peter Salewsky)
Organization: port GmbH, Bitterfeld
Message-ID: <F6BMIG.LJK@port.de>
References: <F5z20M.p2D@port.de>
Sender: news@port.de

Hi all,

Thanks to all who answered!

It turned out to be a CPU (MC68060) exception
(which is unmaskable) that caused the described behavior.

The intLockLevel was set correctly (I had checked this
previously but forgot to mention it - sorry)
and the intLock() code looks okay to me.

This exception (floating-point unimplemented data type)
was due to an uninitialized double variable.

However, I don't have even the slightest idea, why this
behavior

a) occured only occasionally (I would assume that a problem
   with an uninitialized variable occurs either only on the
   first or on every access),

b) depends on the running system clock.

P.S.: As John Finley suggested, also the cache has an influence
(some 4 us variation) - however, disabling the cache decreases
the performance significantly (up to factor 15 when a loop is
involved) so I decided to live with that variation.

- -- 
MfG/Regards

Peter Salewsky

- ------------------------------------------
Dr. Peter Salewsky
port GmbH            phone +49 3493 743-10
Antonienstr. 3       fax   +49 3493 743-15
D-06749 Bitterfeld   email service@port.de
Germany              http://www.port.de
- ------------------------------------------

In article <F5z20M.p2D@port.de>,
	sa@ute.port.de (Dr. Peter Salewsky) writes:

> I'm using VxWorks 5.3.1 on an MVME172.
> 
> I have connected a small routine to the AuxClk interrupt
> which performs the following:
> 
> intLock()
> trigger 1
> some stuff computing (basic math only, no kernel function involved)
> trigger 2
> intUnlock()
> 
> The trigger is an access to a VMEbus address watched by an analyzer.
> 
> Despite the intLock(), I found that often the time between
> trigger 1 and 2 was significantly higher than usual
> (normal time: 6.5 us, worst time: 90 us) - so obviously my ISR
> was delayed by something.
> 
> After disabling the system clock, this effect disappeared.

---------------------------

Newsgroups: comp.arch.embedded,comp.os.os9,comp.os.psos,comp.os.vxworks,comp.realtime
Subject: Re: Choosing a RTOS
Date: Fri, 29 Jan 1999 18:51:13 +0100
From: Martin.Raabe@t-online.de (Martin Raabe)
Organization: Integrated Systems GmbH
Message-ID: <78ssfo$gv4$1@news00.btx.dtag.de>
References: <01be492d$f3e241c0$65020abe@ee02> <36AE4520.B65CDF9@SPAMFREE.iol.ie> <wg4s2.262$Bg3.696@read1.inet.fi>

Hello KW,
yes this is true for the past.
The new pSOSystem V2.5 introduces two enhancements to that:
1. Service calls by function call instead of traps. (configurable)
2. FastIRQ (configurable)

So pSOsystem takes care of this observations from the user side.

Ciao
Martin Raabe

KW schrieb in Nachricht ...
>I've been working with pSOS/PowerPC with telecommunication applications and
>have found it easy to use and understand.
>
>The only negative thing to say is the performance. The system calls and
task
>switches seems to take quite a long time. If your application is not
pushing
>the HW limits, then I could recommend it, as it seems to be easy to write
>working applications, but if you're running close to the limits, do some
>benchmarking first.
>There is a new version coming, where they say things will be much faster,
>but I have no good knowledge on that.
>
>
>



---------------------------

Newsgroups: comp.os.vxworks
Subject: compiling for 68060 on mvme172 board
Date: 29 Jan 1999 19:46:18 GMT
From: pjb@cs.cmu.edu (Peter Berkelman)
Organization: School of Computer Science, Carnegie Mellon
Message-ID: <78t36a$n6u$1@goldenapple.srv.cs.cmu.edu>

I've recently upgraded to a MVME172 board from a MVME162 and the gcc
cross compiler we have with vxworks from before doesn't support the
68060 processor on the new board, although code compiled for the 68040
generally works ok on the MVME172.

Would I get much of a gain in speed if I got a compiler that optimized
for the 68060?  Has anybody tried this?  I'd like to get as much speed
as I can out of this board and I know that the most recent gcc
distribution is supposed to have 68060 support.

thanks,
- --Peter Berkelman


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: WRS stock drop - what happened?
Date: Mon, 01 Feb 1999 03:38:02 GMT
From: johnchann@home.deletethis.com (John Chann)
Organization: @Home Network
Message-ID: <uk9t2.6887$BS5.3316@news.rdc1.sfba.home.com>
References: <36a935de.0@news1.ibm.net> <01be4caf$56c9b500$f93d3ccf@LittleFoot>

The first response is simply ill informed and the second is the purest BS.

HP, the stock split does not become effective until next month.

Eric, if you bothered to do some research before you speculate for the entire 
world you'd read something like Tech Stocks and you'd know that their CEO 
briefed two analysts on that day and told them that earnings would be in line 
with expectations. "I heard..." typical sign of a rumor merchant, state your 
source. Wind have a recent history of always beating expectations, so in that 
light this could be seen as a shortfall, however they have not announced their 
results yet. The two analysts were unperturbed, one left his recommendation as 
strong buy and the other reduced from strong buy to buy. More interesting is 
that Schwab declared Wind as a volatile stock, despite that they are less 
volatile than another 100 that Schwab trades in... Schwab refuse to say why 
but there is speculation that it was a move to have their investors reduce 
their holding in Wind without saying as much.


JRC

In article <01be4caf$56c9b500$f93d3ccf@LittleFoot>, "Eric S. Riseman" 
<eriseman@tiac.net> wrote:
>I understand that there was a stock split.   I also heard that they missed
>their 4th Qtr numbers by a long shot.
>
>HP <hmp@vrtelecom.com> wrote in article <36a935de.0@news1.ibm.net>...
>> So, does anybody know what might have caused today's $17+ drop in WRS
>stock?
>> 
>> -hp
>> 
>> 
>> 
>> 

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Interrupts despite intLock()
Date: Mon, 01 Feb 1999 05:12:27 GMT
From: John or Jenn <lastimp@home.com>
Organization: Who, us organized???
Message-ID: <%Iat2.4876$RK5.3126@news.rdc1.az.home.com>
References: <36AFA1B3.73890A40@tc.fluke.com> <36B39A5D.C6EF9D94@fmi.uni-passau.de> <01be4caf$56c9b500$f93d3ccf@LittleFoot>

In article <uk9t2.6887$BS5.3316@news.rdc1.sfba.home.com> John Chann,
johnchann@home.deletethis.com writes:
>world you'd read something like Tech Stocks and you'd know that their CEO 

Ummmm, COO, actually, not CEO...

---------------------------

End of New-News digest
**********************


From owner-vxwexplo-process  Tue Feb  2 03:00:03 1999
From: Sasikumar S <ssasi_kumar@hotmail.com>
Date: Tue Feb  2 03:00:06 PST 1999
Subject: FAQ: Why it is called as Real Time?


--------------F0D806CF9340869E3DC8F617
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

When a OS can be called as Real Time OS?
or
What are ADDITIONAL Functionalities a Real Time OS has to possess ?
or
What makes the OS to be called as Real Time OS?

you can give relative examples w.r.t VxWorks RTOS and other os like
DOS,Window'95/NT,unix etc.

Pls do reply,
Thanking you,
SASi
--
-------------------------------------------------
SASIKUMAR S
EMAIL:  ssasi_kumar@hotmail.com
 sasikumar@ncoretech.com
-------------------------------------------------
Software Engineer
NCore Technology Pvt Ltd.
Leo complex 4th floor,
44 & 45 Residency cross Rd.
Bangalore - 560 025.
Ph:91-080-558 0405/8257
http://www/ncore.soft.net
-------------------------------------------------
C/o Chandran compound,
Ist main, Ist cross street,
V.P. Road,Madivala,Bangalore-560 068.
-------------------------------------------------


--------------F0D806CF9340869E3DC8F617
Content-Type: text/html; charset=us-ascii
Content-Transfer-Encoding: 7bit

<!doctype html public "-//w3c//dtd html 4.0 transitional//en">
<html>
When a OS can be called as <b>Real Time OS</b>?
<br>or
<br>What are <b>ADDITIONAL Functionalities </b>a <b>Real Time OS </b>has
to possess ?
<br>or
<br>What <b>makes </b>the OS to be called as <b>Real Time OS</b>?
<p>you can give relative examples w.r.t VxWorks RTOS and other os like
DOS,Window'95/NT,unix etc.
<p>Pls do reply,
<br>Thanking you,
<br>SASi
<br>--
<br>-------------------------------------------------
<br>SASIKUMAR S
<br>EMAIL:&nbsp; ssasi_kumar@hotmail.com
<br>&nbsp;sasikumar@ncoretech.com
<br>-------------------------------------------------
<br>Software Engineer
<br>NCore Technology Pvt Ltd.
<br>Leo complex 4th floor,
<br>44 &amp; 45 Residency cross Rd.
<br>Bangalore - 560 025.
<br>Ph:91-080-558 0405/8257
<br><A HREF="http://www/ncore.soft.net">http://www/ncore.soft.net</A>
<br>-------------------------------------------------
<br>C/o Chandran compound,
<br>Ist main, Ist cross street,
<br>V.P. Road,Madivala,Bangalore-560 068.
<br>-------------------------------------------------
<br>&nbsp;</html>

--------------F0D806CF9340869E3DC8F617--



From owner-vxwexplo-process  Tue Feb  2 04:00:19 1999
From: daemon@csg.lbl.gov
Date: Tue Feb  2 04:00:22 PST 1999
Subject: comp.os.vxworks newsdigest

Comp.Os.Vxworks Daily Digest    Tue Feb  2 04:00:15 PST 1999

        Subject: Re: VxWorks port to Linux survey
        Subject: Re: Interrupts despite intLock()
        Subject: Re: Interrupts despite intLock() - result
        Subject: Re: MPC860 and cache
        Subject: Re: VxWorks port to Linux survey
        Subject: Principal Software Engineer ISDN/ATM
        Subject: Re: C++ exception handling
        Subject: Re: compiling for 68060 on mvme172 board
        Subject: Re: MPC860 and cache
        Subject: Re: MPC860 and cache
        Subject: SPI on an MBX860??
        Subject: Re: SPI on an MBX860??
        Subject: Re: Choosing a RTOS
        Subject: Some puzzle thing about ads860

-------------------------------------------------------

Newsgroups: comp.os.vxworks
Subject: Re: VxWorks port to Linux survey
Date: Mon, 1 Feb 1999 15:03:05 +0200
From: "Saffi Hartal" <saffi@globaloop.com>
Organization: NetVision Israel	
Message-ID: <794a9e$hv0$1@news.netvision.net.il>
References: <78qnh6$dd37@overload.lbl.gov>

Hi Bruce
Could you elaborate of the meaning of "porting vxWorks to Linux" ?.
I work in vxWorks and I would like to port stuff from linux to vxWorks or
work with very light linux instead of working with vxWorks ( I would rather
do
that because linux is open sources and it's really big problem in vxWorks).
if you mean working in linux environment on vxWorks ... could be nice but I
rather for now be able to port linux code to vxWorks.
So I guess the best thing for developers is vxWorks will become open source.
        With Regards
                        Saffi.


Bruce O. Benson wrote in message <78qnh6$dd37@overload.lbl.gov>...
>In response to the recent lengthy thread on the hope (for some at least)
>that WRS would port VxWorks to Linux (or BSD for that matter), I am
>collecting names to present to WRS management as a show of strength for
>this.
>
>If you would like to be included as a supporter of this port, please drop
me
>a note at mailto:benson@linux.sparta.com
>In fairness, if you think it's a dumb idea, I will collect your note and fw
>your name, too.
>
>I'll post the statistical results of the survey, and the WRS response in
the
>next couple of weeks on http://linux.sparta.com/VxWorks
>
>Regards,
>Bruce.
>Bruce Benson, Senior Engineer,
>SPARTA (CPD)
>7926 Jones Branch Drive, Suite 900
>McLean, VA  22102-3303
>USA
>+1 703-448-0210 x211
>benson@sparta.com
>http://www.mclean.sparta.com
>
>





---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Interrupts despite intLock()
Date: Fri, 29 Jan 1999 16:29:26 -0600
From: petekz@cyberramp.net (Pete Kockritz)
Organization: posted via: CyberRamp.net, Dallas, TX (214) 343-3333/(817) 461-8484 for info
Message-ID: <petekz-2901991629260001@dal-tsa34-26.cyberramp.net>
References: <F5z20M.p2D@port.de> <36AFA1B3.73890A40@tc.fluke.com> <78rdkb$kag$1@nnrp1.dejanews.com>

In article <78rdkb$kag$1@nnrp1.dejanews.com>, john@kivala.com wrote:

> MSR-EE masks both external interrupts and the decrementer
> interrupt, even though they come in on different vectors
> (at least on the MPC860; I don't know about other PPCs).
> It says so in one of those big honkin' manuals somewhere.
> But I agree that the PIT would make a better system ticker.
> 
> John


But since the decrementer interrupt is not going through the SIU,
there is less overhead in servicing it. So the PIT is perhaps better
from the standpoint of precision, but not better from the standpoint
of CPU overhead.

Regards,
Pete
- --
+-----------------------------------------------------------------+
|       Pete Kockritz                mailto:petekz@cyberramp.net  |
|                       This space for rent.                      |
+-----------------------------------------------------------------+

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Interrupts despite intLock() - result
Date: Fri, 29 Jan 1999 17:47:08 -0500
From: "David J. Cacciaglia" <chico@sgss.com>
Organization: Naval Research Laboratory, Washington, DC
Message-ID: <36B23A6C.DF251FFA@sgss.com>
References: <F5z20M.p2D@port.de> <F6BMIG.LJK@port.de>
Reply-To: chico@sgss.com

Peter

The reason for it not happening consistently is unless the variable is
global the variable is allocated off the stack thus it would depend on 
when the routine was called and of course ends up being a game of
chance.

"Dr. Peter Salewsky" wrote:
> 
> Hi all,
> 
> Thanks to all who answered!
> 
> It turned out to be a CPU (MC68060) exception
> (which is unmaskable) that caused the described behavior.
> 
> The intLockLevel was set correctly (I had checked this
> previously but forgot to mention it - sorry)
> and the intLock() code looks okay to me.
> 
> This exception (floating-point unimplemented data type)
> was due to an uninitialized double variable.
> 
> However, I don't have even the slightest idea, why this
> behavior
> 
> a) occured only occasionally (I would assume that a problem
>    with an uninitialized variable occurs either only on the
>    first or on every access),
> 
> b) depends on the running system clock.
> 
> P.S.: As John Finley suggested, also the cache has an influence
> (some 4 us variation) - however, disabling the cache decreases
> the performance significantly (up to factor 15 when a loop is
> involved) so I decided to live with that variation.
> 
> --
> MfG/Regards
> 
> Peter Salewsky
> 
> ------------------------------------------
> Dr. Peter Salewsky
> port GmbH            phone +49 3493 743-10
> Antonienstr. 3       fax   +49 3493 743-15
> D-06749 Bitterfeld   email service@port.de
> Germany              http://www.port.de
> ------------------------------------------
> 
> In article <F5z20M.p2D@port.de>,
>         sa@ute.port.de (Dr. Peter Salewsky) writes:
> 
> > I'm using VxWorks 5.3.1 on an MVME172.
> >
> > I have connected a small routine to the AuxClk interrupt
> > which performs the following:
> >
> > intLock()
> > trigger 1
> > some stuff computing (basic math only, no kernel function involved)
> > trigger 2
> > intUnlock()
> >
> > The trigger is an access to a VMEbus address watched by an analyzer.
> >
> > Despite the intLock(), I found that often the time between
> > trigger 1 and 2 was significantly higher than usual
> > (normal time: 6.5 us, worst time: 90 us) - so obviously my ISR
> > was delayed by something.
> >
> > After disabling the system clock, this effect disappeared.

- -- 
__________Space Above and Beyond_________
/   Space Ground System Solutions  (SGSS) \
| David J. Cacciaglia a.k.a Chico         |
| chico@sgss.com                          |
| 301.870.3582 ph.      301.870.2065 fax  |
| 301.513.7137 pager                      |
| 301.934.5884 voice mail                 |
|_________________________________________|

---------------------------

Newsgroups: comp.os.vxworks,comp.sys.powerpc.tech
Subject: Re: MPC860 and cache
Date: Fri, 29 Jan 1999 23:18:09 GMT
From: pierre_olivier@srtelecom.com
Organization: Deja News - The Leader in Internet Discussion
Message-ID: <78tfj9$enn$1@nnrp1.dejanews.com>
References: <petekz-2701991950220001@dal-tsa31-56.cyberramp.net> <36B00C1A.414B@nospam.aracnet.com> <petekz-2801991927100001@dal-tsa12-61.cyberramp.net> <78sgd8$a2n$1@user2.teleport.com>

In article <78sgd8$a2n$1@user2.teleport.com>,
  andrew@user2.teleport.com (Andrew Klossner) wrote:
> > if I enable the caches but the MMU is told that all memory areas are
> > non-cacheable, how is that different that having the cache off?  It
> > seems to behave differently.
>
> I work with 603e, not 860, so this is of value only to 603e users who
> happen to be reading the thread.
>
> On 603e, the difference is: with MMU and cache off, stores can be
> deferred until after a subsequent load.  When you turn on the MMU,
> stores and loads are executed in order for cache-inhibited pages.
> This can make a big difference in code like:
>
> 	stw	...		// select the other register bank
> 	lwz	...		// load from that bank
>
> Performing the lwz before the stw in this case gives wrong results.
>
>   -=- Andrew Klossner (andrew@teleport.com)

In this case, you should perform an eieio between the stw and lwz.
>

- -----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: VxWorks port to Linux survey
Date: Mon, 1 Feb 1999 14:48:58 +0200
From: "Saffi Hartal" <saffi@globaloop.com>
Organization: NetVision Israel	
Message-ID: <7948ge$e0t$1@news.netvision.net.il>
References: <78qnh6$dd37@overload.lbl.gov>

Hi Bruce
Could you elaborate of the meaning of "porting vxWorks to Linux" ?.
I work in vxWorks and i would like to port stuff from linux to vxWorks or
work with very light linux instead of working in vxWorks ( i would rather do
that because i'll have the sources ).
if you mean working in linux environment on vxWorks ... could be nice but i
rather for now be able to port linux code to vxWorks.
So i guess my needs are a bit different then yours.
        With Regards
                        Saffi.


Bruce O. Benson wrote in message <78qnh6$dd37@overload.lbl.gov>...
>In response to the recent lengthy thread on the hope (for some at least)
>that WRS would port VxWorks to Linux (or BSD for that matter), I am
>collecting names to present to WRS management as a show of strength for
>this.
>
>If you would like to be included as a supporter of this port, please drop
me
>a note at mailto:benson@linux.sparta.com
>In fairness, if you think it's a dumb idea, I will collect your note and fw
>your name, too.
>
>I'll post the statistical results of the survey, and the WRS response in
the
>next couple of weeks on http://linux.sparta.com/VxWorks
>
>Regards,
>Bruce.
>Bruce Benson, Senior Engineer,
>SPARTA (CPD)
>7926 Jones Branch Drive, Suite 900
>McLean, VA  22102-3303
>USA
>+1 703-448-0210 x211
>benson@sparta.com
>http://www.mclean.sparta.com
>
>



---------------------------

Newsgroups: comp.os.vxworks
Subject: Principal Software Engineer ISDN/ATM
Date: Mon, 01 Feb 1999 19:17:14 GMT
From: "Robertson-Surrette" <ottawa@robsur.com>
Organization: Robertson-Surrette Inc.
Message-ID: <01be4e30$d8368140$ced0ba89@istar>

Communications Systems

Principal Software Engineer

If the prospect of applying leading-edge commercial voice and data
communications technology to highly reliable, ruggedized and muti-level
reduncy based military products sounds challenging, read on:

On behalf of one of the Ottawa region's most innovative system solutions
providers, we are currently conducting a search for a Principal Engineer,
Software Development.  This organization has established a worldwide
reputation for designing and manufacturing state-of-the-art advanced
communications products.  As part of an advanced development team, this
individual will play a key role in defining the "next generation" of the
company's leading-edge system.

To be ideally suited for this role, you have a strong communications and
telephony background with extensive ATM and ISDN design experience.  Your
network management expertise will serve you well in designing multimedia
delivery capabilities using state-of-the-art technologies such as JAVA, for
use in very specialized military environments.  Experience with ADA
Programming and/or VxWorks will be considered an assets.

You possess a bachelor's degree in either Computer Science or Engineering. 
Having worked in the technology field for at least 7 to 8 years, you are
ready for a challenging role where a small team setting fosters mentoring
and learning.  Here, every team member understands the difference his or
her contribution makes to the product.  The position requires an individual
who loves to be challenged.  

If you, or someone you know is interested in pursuing this unique
opportunity further, please contact Jim Harmon at 'jharmon@robsur.com' or
by fax at (613) 749-9599.  We would also welcome your call at (613)
749-9909.  You can find more information about this role on our web site at
'www.robsur.com'.




---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: C++ exception handling
Date: Fri, 29 Jan 1999 19:07:42 -0500
From: default <Mark_R_Fay@res.raytheon.com>
Organization: Software Development & Integration Resources
Message-ID: <36B24D4E.252FE458@res.raytheon.com>
References: <789fk6$4pv$1@fleetstreet.Austria.EU.net>

I talked to Green Hills about this issue. As described by Don Bowman,
the Green Hills compiler, not VxWorks will perform the exception handling.
Thread save exception handling for VxWorks is supported in MUTLI for C++
version 1.8.9 (we have not tested this version for exception handling at
Raytheon).
Also, Green Hills field support is available through our maintenance
contract
for telephone and/or on-site visits for problems such as your compilation
problem.


From message:
 Newsgroups: comp.os.vxworks
 Subject: Re: C++ exception handling
 Date: Wed, 6 Jan 1999 22:31:35 -0500
 From: "Don Bowman" <don@pixstream.com>
 Organization: UUNET Canada News Transport
 Message-ID: <771a5p$n0i$2@nntp2.uunet.ca>
 References: <770550$5bg$1@swlmail.msd.ray.com>

Don Bowman wrote:

>  Regina DuBord {84134} wrote in message
> <770550$5bg$1@swlmail.msd.ray.com>...
>  >Hi,
>  >
>  >I have a question regarding VxWorks and C++ exception
>  >handling. In various places on the Wind River web site,
>  >I've read that VxWorks itself doesn't accomodate C++
>  >exception handling, even if one has a compiler that will
>  >produce VxWorks object files from C++ source that contains
>  >exception handling code.
>  >
>  >Is this true? If so, then what's the value of having a
>  >Green Hills C++ compiler that's advertised as supporting
>  >exception handling? Has anyone had any luck getting that
>  >compiler to handle C++ exceptions? I haven't been able
>  >to get it to even compile anything, but I'm wondering if
>  >it's worth the struggle.
>
>
>   Hi.
>
>   We've been using the egcs release of GCC, and the
>   exception handling is working fine. You need the new
>   version of libgcc.a that comes with that compiler.
>   The OS doesn't really have to be involved in the exception
>   handling.


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: compiling for 68060 on mvme172 board
Date: 29 Jan 1999 22:11:55 GMT
From: chc@nasa2.ksc.nasa.gov (Charles H. Chapman)
Organization: NASA, Kennedy Space Center
Message-ID: <slrn7b4cd2.s0f.chc@nasa2.ksc.nasa.gov>
References: <78t36a$n6u$1@goldenapple.srv.cs.cmu.edu>

On 29 Jan 1999 19:46:18 GMT, Peter Berkelman <pjb@cs.cmu.edu> wrote:
>I've recently upgraded to a MVME172 board from a MVME162 and the gcc
>cross compiler we have with vxworks from before doesn't support the
>68060 processor on the new board, although code compiled for the 68040
>generally works ok on the MVME172.
>
>Would I get much of a gain in speed if I got a compiler that optimized
>for the 68060?  Has anybody tried this?  I'd like to get as much speed
>as I can out of this board and I know that the most recent gcc
>distribution is supposed to have 68060 support.

I have very little knowledge of the differences in architectures of the
040 vs. the 060 and how much of an advantage upgrading the gcc compiler
would be (although I thought gcc had had 060 support for a long time --
you may already have an 060 capable compiler).  But another thing to
consider is the VxWorks libraries -- there are specific libraries for
each architecture so even if you could compile your source with an 060
capable compiler, you'd still have to find 060 libraries to take full
advantage.

Chuck

---------------------------

Newsgroups: comp.os.vxworks,comp.sys.powerpc.tech
Subject: Re: MPC860 and cache
Date: 1 Feb 1999 06:50:36 -0800
From: andrew@user2.teleport.com (Andrew Klossner)
Message-ID: <794evs$f6s$1@user2.teleport.com>
References: <petekz-2701991950220001@dal-tsa31-56.cyberramp.net> <petekz-2801991927100001@dal-tsa12-61.cyberramp.net> <78sgd8$a2n$1@user2.teleport.com> <78tfj9$enn$1@nnrp1.dejanews.com>

>> I work with 603e ...
>> 	stw	...		// select the other register bank
>> 	lwz	...		// load from that bank

> In this case, you should perform an eieio between the stw and lwz.

Nope.  From the PowerPC 603e Microprocessor Programming Manual, page 2-39:

	"The Enforce In-Order Execution of I/O (eieio) instruction
	is used to ensure memory reordering of noncacheable memory access.
	Since the 603e does not reorder noncacheable memory accesses,
	the eieio instruction is treated as a no-op."

In the case under discussion, boot ROM code, the MMU is not enabled so
the 603e thinks that all locations are write-through cacheable.  The
cache is disabled (via the HID0 register) and no actual caching
occurs, yet the CPU feels free to reorder the accesses.  I use sync
instructions to preserve order.

  -=- Andrew Klossner (andrew@teleport.com)

---------------------------

Newsgroups: comp.os.vxworks,comp.sys.powerpc.tech
Subject: Re: MPC860 and cache
Date: Mon, 01 Feb 1999 21:00:18 -0600
From: Pierre Olivier <polivier@netcom.ca>
Organization: Netcom Canada
Message-ID: <36B66A42.622@netcom.ca>
References: <petekz-2701991950220001@dal-tsa31-56.cyberramp.net> <36B00C1A.414B@nospam.aracnet.com> <petekz-2801991927100001@dal-tsa12-61.cyberramp.net> <78sgd8$a2n$1@user2.teleport.com> <78tfj9$enn$1@nnrp1.dejanews.com> <slrn7b4s98.1h8g.skipper@web1.calweb.com> <slrn7b5724.273k.skipper@web1.calweb.com>
Reply-To: polivier@netcom.ca

Skipper Smith wrote:
> 
> Skipper Smith <skipper@no-spam.calweb.com> wrote:
> >pierre_olivier@srtelecom.com <pierre_olivier@srtelecom.com> wrote:
> >>In article <78sgd8$a2n$1@user2.teleport.com>,
> >>> This can make a big difference in code like:
> >>>
> >>>     stw     ...             // select the other register bank
> >>>     lwz     ...             // load from that bank
> >>>
> >>> Performing the lwz before the stw in this case gives wrong results.
> >>
> >>In this case, you should perform an eieio between the stw and lwz.
> >
> >Won't work.  When the MMU is turned off, the default caching mode is
> >cacheable-writeback.  Therefore, the eieio is ignored.  You would have to
> >use a sync instruction to even hope that it would work with or without the
> >MMU turned on.                                              ^
>                                                              |
>                                                              |
>                                                              |
> Ooops, I meant AND--------------------------------------------
> 


Unless the 603, like the 860, also has an command to put the whole data
cache in writethrough mode (by a write to DC_CST of 0x02000000, if I
remember well).

Pierre.


> --
> Skipper Smith                         Helpful Knowledge Consulting
> Worldwide         Microprocessor       Architecture       Training
> PowerPC, ColdFire, 68K, CPU32                Hardware and Software
> 
> /* Remove no-spam. from the reply address to send mail directly */

- -- 
- ---------------------------------------------
Pierre Olivier
Montreal (QC) Canada
polivier@netcom.ca               (. .)
- ------------------------------oOO-(_)-OOo----

---------------------------

Newsgroups: comp.os.vxworks
Subject: SPI on an MBX860??
Date: Mon, 1 Feb 1999 21:58:42 -0500
From: "Dale" <dwebb@stny.lrun.com>
Organization: TWC Syracuse NY
Message-ID: <DNtt2.3614$EE2.2379@newsr1.twcny.rr.com>


Has anyone tried to implement SPI on a MBX860 board.

I have basically duplicated the example code in the PPC 860 manual and/or
training CD to no avail.   The SPICLK or SPIMOSI pins never move, and the
SPI registers / dual port ram buffers show now signs of life.

The MBX860 BSP does support the DSP microcode patch that relocates the I2C /
SPI dual port ram buffers to allow I2C and ethernet to function
simultaneously.   Does this patch impact the location / functionality of SPI
/ SPI dual port ram?

Is there a problem using SPI with I2C and ethernet??

Any ideas / pointers / code snippets would be greatly appreciated.  I'm
assuming that I am fundamentally missing something since there should be
some error indications in the SPI registers / buffers.

Thanks,

Dale



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: SPI on an MBX860??
Date: Mon, 1 Feb 1999 22:01:02 -0500
From: "Dale" <dwebb@stny.lrun.com>
Organization: TWC Syracuse NY
Message-ID: <SPtt2.3616$EE2.2635@newsr1.twcny.rr.com>
References: <DNtt2.3614$EE2.2379@newsr1.twcny.rr.com>

By the way,

My work E-Mail is:   webbda@corning.com

Thanks for any messages.



Dale wrote in message ...
>
>Has anyone tried to implement SPI on a MBX860 board.
>
>I have basically duplicated the example code in the PPC 860 manual and/or
>training CD to no avail.   The SPICLK or SPIMOSI pins never move, and the
>SPI registers / dual port ram buffers show now signs of life.
>
>The MBX860 BSP does support the DSP microcode patch that relocates the I2C
/
>SPI dual port ram buffers to allow I2C and ethernet to function
>simultaneously.   Does this patch impact the location / functionality of
SPI
>/ SPI dual port ram?
>
>Is there a problem using SPI with I2C and ethernet??
>
>Any ideas / pointers / code snippets would be greatly appreciated.  I'm
>assuming that I am fundamentally missing something since there should be
>some error indications in the SPI registers / buffers.
>
>Thanks,
>
>Dale
>
>



---------------------------

Newsgroups: comp.arch.embedded,comp.os.os9,comp.os.psos,comp.os.vxworks,comp.realtime,comp.os.lynx
Subject: Re: Choosing a RTOS
Date: Tue, 26 Jan 1999 11:50:29 -0700
From: Eric Price <eprice@phx.mcd.mot.com>
Message-ID: <36AE0E75.DA6E3FD@phx.mcd.mot.com>
References: <01be492d$f3e241c0$65020abe@ee02> <36ADC1A0.F1E02DC7@realtime-info.be>

Will/was lynxOS be included in the evaluation?

Alexander Teetaert wrote:

> Thomas, Real-Time Magazine is releasing next quarter their RTOS Evaluation
> Reports for Tornado, pSOS and QNX. OS-9 is not included in the RTOS Evaluation
> Project as Microware did not want to co-operate, all other RTOS vendors were
> eager to have their product tested.



- --
There are no simple solutions,
        only intelligent choices.


                Alexander Graham Bell.



---------------------------

Newsgroups: comp.os.vxworks
Subject: Some puzzle thing about ads860
Date: Tue, 2 Feb 1999 14:20:15 +0800
From: "SRD Zha Junhai" <svczjh@sbell.com.cn>
Organization: SBELL
Message-ID: <36b69b1b.0@dnews.sbell.com.cn>

Hi, everybody

    The ram size in the standard configuration of ads860
is 4MB. When like this Anything is correct include the
terminal window. I change the ram to 8MB and modified the
RAM_HIGH_ADRS to 0x00600000 and LOCAL_MEM_SIZE to
0x00800000 in config.h and makefile, but the things is
so puzzle that when I program the bootrom_uncmp.hex to
flash and restart, everything is correct(I can launch
the target server and start the shell and windview) except
that there is nothing at the terminal windows. Is the
serial port mapped memory addr overlayed by the ram addr?






---------------------------

End of New-News digest
**********************


From owner-vxwexplo-process  Tue Feb  2 04:48:10 1999
From: "Joe W." <joew@wtc.gov>
Date: Tue Feb  2 04:48:14 PST 1999
Subject: VxWorks: SCSI boot of MVME1604

Hello,

I am unable to boot a MVME1604 board from the SCSI disk.  I have
burned new boot roms, and turned on SCSI debug.  Here is the output:

CPU: Motorola MVME1600 - PowerPC 604
Version: 5.3
BSP version: 1.1/5
Creation Date: Feb 1, 1999, 14:14:50

Press any key to stop auto-boot...
 0
auto-booting...

boot device          : scsi
unit number          : 0
processor number     : 0
host name            : appserver1
file name            : /sd0/vxWorks
inet on ethernet (e) : 79.0.2.215:ffffff00
host inet (h)        : 79.0.2.213
user (u)             : ast
flags (f)            : 0x0
target name (tn)     : test

Calling scsi2IfInit...
Returned from scsi IF init
Calling sysScsiInit
sysScsiInit: Turned on scsiDebug
0x7f6510 (tBoot) : Found an NCR810 Controller
Calling scsiBusReset
0x7fed48 (tscsiTask) : ncr810Event: received event 52 (thread =
0x00000000 )
calling scsiAutoConfig
0x7f6510 (tBoot) : scsiAutoConfig: bus ID = 0, LUN = 0
0x7f6510 (tBoot) : scsiReqSense:
0x7f6510 (tBoot) : scsi2Transact:
0x7fed48 (tScsiTask): scsimgrRequestProc: client request: 0
0x7fed48 (tScsiTask): scsiMgrActivateRequest: thread 0x007fd480
0x7fed48 (tScsiTask): ncr810TheadActivitate: thread
0x007fd480: activating
0x7fed48 (tScsitask): ncr810XferParamsCvt: converted to: offset      =
0, period
    = 0
xferParams = 0, clockDivide = 3
0x7fed48 (tScsitask): ncr810XferParamsCvt: converted to: offset      =
0, period
    = 0
xferParams = 0, clockDivide = 3
0x7fed48 (tScsiTask): ncr810ThreadStateSet: thread 0x007fd480: 1 -> 6
0x7fed48 (tScsiTask): scsiMgrTimeoutProc: thread 0x007fd480 (state = 6)
timed out
0x7fed48 (tScsiTask): ncr810ThreadAbort: thread 0x007fd480 (stat =
6) aborting
0x7fed48 (tScsiTask): ncr810ThreadStateSet: thread 0x007fd480: 6 -> 9


and that's as far as the thing will go.

It appears to be getting stuck in scsiPhysDevCreate; it gets stuck there

also if I
explicitly set which SCSI dev to boot off of by setting the boot device
as "scsi=0,0".

I read in one of the FAQs or from the newsgroup that a delay before the
scsiPhysDevCreate sometimes helps, so I added a 30 second delay, but it
didn't help.  I also added some code we had to add for our MVME167
boards which did a scsiBusReset, but that also didn't help.

If anyone out there has any suggestions, they'd be most appreciated!

Thanks in advance.

--Joe W (joew@wtc.gov)



From owner-vxwexplo-process  Tue Feb  2 09:51:38 1999
From: David Laight <dsl@tadpole.co.uk>
Date: Tue Feb  2 09:51:42 PST 1999
Subject: Ordering of memory cycles

There have been a lot of deep discussions about the following code snippet:

> 	stw	...		// select the other register bank
> 	lwz	...		// load from that bank

I don't know about the 603e, but on most modern processors the above code 
doesn't work - in many with or without the cache enabled.

The easiest solution (works in every case I've seen, and is implementation 
independant) is to follow the 'write' with a 'read' of the same location.
This has the effect of forcing the write out of ALL the places where it might 
have got buffered - there can be many - before the read is done.

Clearing the write from ALL the places it might be buffered is particularly 
important when clearing an interrupt request.  If you don't then the o/s might 
report 'unexpected interrupt 0xxx' (vxworks interrupt code doesn't attemp to be 
that clever).

Unlike eieio (old MacDonald?) and obscure cache controller instructions, the 
read-back can be done in C.  Just ensure everything is volatile.

Buffering of writes first became a problem with the good old 286.  Allowing it 
to break the inter cycle recovery time of the 8259 interrupt controller and 
putting garbage into all of the controllers registers.


	David


----------------------------------------------------------------
David Laight                          email: dsl@tadpole.co.uk
Tadpole Technology plc                phone: +44 1223 278 256
Cambridge, UK                         fax:   +44 1223 278 201



From owner-vxwexplo-process  Tue Feb  2 11:56:29 1999
From: rcw@DynRes.com
Date: Tue Feb  2 11:56:33 PST 1999
Subject: NFS filesize limit?

VxWorks
Tornado

Does anyone know about a filesize limit when using NFS?

I am using Distinct NFS on a Win95 machine, connecting to a 
VxWorks machine. I can drag and drop, edit, etc., to my heart's
content. But if I try to copy a large file (maybe greater than
500K bytes), I get an error. Something like: 'no room on disk'.

The same Win95 machine *will* perform the large file copy to 
a genuine UNIX machine.

I've looked through the docs, but find no mention of a size limit.

Any ideas?

Ron Wagner


From owner-vxwexplo-process  Wed Feb  3 00:08:30 1999
From: Johan Bostedt <qsabost@rsa.ericsson.se>
Date: Wed Feb  3 00:08:34 PST 1999
Subject: Booting from flash

Hi VxWorkers,

We are designing a custom board based on the MCP860 processor. I have a
question concerning
the loading of the boot code and the VxWorks image. We would like to
have it all in flash memory.

Is it possible to acheive this without a boot PROM?

What do you need to program the boot monitor and the VxWorks image into
flash?
We have an ethernet port and an RS232 port connected to the
microprocessor.

What about sizes? How big is the bootrom code and how
big is a typical VxWorks image?


Best regards,
Johan Bostedt
_________________________________________________
Johan Bostedt     phone: +46(0)8-404 87 96
RSA/B/UD          cellular: +46(0)70-657 64 70
                  email: qsabost@rsa.ericsson.se




From owner-vxwexplo-process  Wed Feb  3 02:32:38 1999
From: Ian.Kerr@bae.co.uk
Date: Wed Feb  3 02:32:42 PST 1999
Subject: Update to NewsDigest for 29.1.99

CORRIGENDA TO TRUNCATED NEWS DIGEST OF 29 Jan 1999 1200 UTC
@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
P.S. no periods "." in 1st col in any line, please.

From - Tue Feb 02 13:10:22 1999
Path: overload.lbl.gov!dog.ee.lbl.gov!newsfeed.berkeley.edu!news.maxwell.
syr.edu!news.mel.connect.com.au!not-for-mail
From: "Llewellyn Griffiths" <nojunk@llga.com.mars>
Newsgroups: comp.arch.embedded,comp.os.os9,comp.os.psos,comp.os.vxworks
Subject: Re: Choosing a RTOS
Date: 29 Jan 1999 09:57:18 GMT
Organization: Llew Griffiths & Associates
Lines: 68
Message-ID: <01be4b6b$9e165480$0100007f@llga>
References: <01be492d$f3e241c0$65020abe@ee02> <36ADFE9E.2BA991B5@ada.com>
 <78np50$46c@atlantis.metrowerks.com> <78nrkk$goq$1@news04.btx.dtag.de> 
<01be4a94$9be76140$0100007f@llga> <78red3$l2v$1@nnrp1.dejanews.com>
NNTP-Posting-Host: acc10-ppp82.mel.enternet.com.au
X-Trace: perki.connect.com.au 917603838 15658 210.8.9.210 (29 Jan 1999 
 09:57:18 GMT)
X-Complaints-To: abuse@connect.com.au
NNTP-Posting-Date: 29 Jan 1999 09:57:18 GMT
X-Newsreader: Microsoft Internet News 4.70.1161
Xref: overload.lbl.gov comp.arch.embedded:26219 comp.os.os9:9669
 comp.os.vxworks:16220



gvarndell@hotmail.com wrote

[earlier stuff snipped]

> If you believe a plug for some also-ran, me-too compiler vendor is
relevant
> to RTOS selection, then you probably would consider it relevant to every
> frickin' thread in all the offended groups.

Actually, the selection of an appropriate tool set is a part of the RTOS
selection process. When debugging code running in an RTOS environment, a
debugger that has good kernel awareness features will make the development
cycle a lot easier. Have you ever wondered why the RTOS vendors make the
effort to form cooperative associations with the organisations that produce
compilers and debuggers? Do you ever think about the tools that you use?
Perhaps you aren't foreman material.

> Subtle indignant pomposity doesn't make you right.

Subtle? Did I use words that you didn't understand?

Indignant? Never! You, however, could very well be so.

Pompous? I obviously expressed an opinion different to yours.

Time to get back on track. The individual who started this thread may still
be looking for a serious answer. The comp.realtime FAQ contains a list of
RTOS manufacturers that is posted to that news group from time-to-time. The
following links may also be of interest:

http://www.atinucleus.com

Nucleus. Haven't used it but I saw an evaluation copy of the source code
last year. Looked OK.

http://www.expresslogic.com

ThreadX. Developed by Bill Lamie who also had a lot to do with Nucleus. We
are currently using it for an MPC860 project with no problems encountered.
Small footprint. Express Logic has also give us good support.

http://www.kadak.com

AMX. I was involved with a project that used this RTOS several years ago. I
don't recall any problems.

http://www.psti.com

Precise/MQX. Again, I have only seen an evaluation copy running a demo
application. Didn't strike any problems.

I could extend this list for several pages since there are a lot of RTOS
products around. A lot of them come with source licences and quite a few
are royalty free. Most of the vendors also offer add-ons like file systems,
TCP/IP etc and an increasing number are now offering GUI support. 


Llew Griffiths
-- 
LLEWELLYN GRIFFITHS
Llew Griffiths & Associates Pty Limited
Melbourne, Australia
info AT llga DOT com DOT au
The reply address is a black hole for spam

----------------------------------------------------------------------------
From - Tue Feb 02 13:12:57 1999
Path: overload.lbl.gov!dog.ee.lbl.gov!newsfeed.berkeley.edu!newsfeed.tli.de
!news-fra.pop.de!newsfeed.eurocyber.net!newsfeed.ecrc.net!newsfeed2.ecrc.net
!news.siemens.de!news.mchh.siemens.de!news
From: Marco Mele <Marco.Mele.GP@icn.siemens.de>
Newsgroups: comp.os.vxworks
Subject: ATM !!! ATM !!! ATM !!! ATM !!!
Date: Fri, 29 Jan 1999 10:47:14 +0100
Organization: Siemens AG
Lines: 7
Message-ID: <36B183A2.7D55BDF6@icn.siemens.de>
NNTP-Posting-Host: mhw12147.mchh.siemens.de
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.04 [en] (WinNT; I)

Hi all !

Has anybody written something similar to an ATM device driver for
VxWorks ?
I really need this information.
Thank you !

-----------------------------------------------------------------------------

From - Tue Feb 02 13:20:24 1999
Path: overload.lbl.gov!dog.ee.lbl.gov!newsfeed.berkeley.edu!news
.maxwell.syr.edu!nntp2.dejanews.com!nnrp1.dejanews.com!not-for-mail
From: cxj@gsta.com
Newsgroups: comp.os.vxworks
Subject: PCI device interrupt problem
Date: Fri, 29 Jan 1999 09:55:16 GMT
Organization: Deja News - The Leader in Internet Discussion
Lines: 7
Message-ID: <78s0i2$417$1@nnrp1.dejanews.com>
NNTP-Posting-Host: 202.96.150.4
X-Article-Creation-Date: Fri Jan 29 09:55:16 1999 GMT
X-Http-User-Agent: Mozilla/4.0 (compatible; MSIE 4.01; Windows 98)
X-Http-Proxy: 1.0 x8.dejanews.com:80 (Squid/1.1.22) for client 202.96.150.4

Hi,all
   How to determine which device caused PCI bus interrupt ? I am using mcp750
BSP.
   Thanks in advance.

-----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

-----------------------------------------------------------------------------

From - Tue Feb 02 13:24:04 1999
Path: overload.lbl.gov!dog.ee.lbl.gov!newshub.nosc.mil!nntp!oanews!news.
primate.wisc.edu!srvr1.engin.umich.edu!news.cc.ukans.edu!newscon05
 !prodigy.com!skynet.be!remarQ-easT!supernews.com
 !news.maxwell.syr.edu!nntp2.dejanews.com!nnrp1.dejanews.com!not-for-mail
From: ksamavedam@hns.com
Newsgroups: comp.os.vxworks
Subject: Changing ping timeout
Date: Mon, 25 Jan 1999 20:46:22 GMT
Organization: Deja News - The Leader in Internet Discussion
Lines: 9
Message-ID: <78il6r$7cr$1@nnrp1.dejanews.com>
NNTP-Posting-Host: 139.85.22.26
X-Article-Creation-Date: Mon Jan 25 20:46:22 1999 GMT
X-Http-User-Agent: Mozilla/3.04 (X11; U; HP-UX B.10.20 9000/712)
X-Http-Proxy: 1.0 x8.dejanews.com:80 (Squid/1.1.22) for client 139.85.22.26

Hi,
        How can I change the "ping" timeout from the vxWorks
        system default of 5 secs ?

regards
krishna

-----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    
@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ THE END @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@


From owner-vxwexplo-process  Wed Feb  3 02:37:07 1999
From: Roddy Pratt <roddy.pratt@oxtel.com>
Date: Wed Feb  3 02:37:11 PST 1999
Subject: Vxworks 486 BSP - reclaiming the bottom meg?

Hi

We're using Vxworks on an Advantech 486 cpu card, booting off HDD. It
seems the bottom meg of memory is only used during this boot process,
and is effectively wasted during program execution.

I'd like to just add it to the heap using something like...

memAddToPool(BASE, (640 * 1024) - BASE);

I'm assuming the memory I/O space over 640K is effectively inacessible.

I'm also assuming that BASE must be non-zero, a) so I don't get confused
with null pointers, and b) as there may be 'low-memory' variables/vector
tables that I shouldn't stamp on. But I don't know what it should
actually be.

Does anybody have any experience of this technique in the 486/386
environment, or any words of wisdom to impart?

Thanks
        Roddy

...VxWorks Tornado...



From owner-vxwexplo-process  Wed Feb  3 04:00:19 1999
From: daemon@csg.lbl.gov
Date: Wed Feb  3 04:00:22 PST 1999
Subject: comp.os.vxworks newsdigest

Comp.Os.Vxworks Daily Digest    Wed Feb  3 04:00:16 PST 1999

        Subject: Re: wdTimer & EPICS sequencer
        Subject: Program Flash on MV2604 via program
        Subject: x86 VxWorks -- Ethernet support
        Subject: x86 VxWorks -- Ethernet support
        Subject: Re: x86 VxWorks -- Ethernet support

-------------------------------------------------------

Newsgroups: comp.os.vxworks
Subject: Re: wdTimer & EPICS sequencer
Date: 30 Jan 1999 15:18:30 GMT
From: chc@nasa2.ksc.nasa.gov (Charles H. Chapman)
Organization: NASA, Kennedy Space Center
Message-ID: <slrn7b68i4.ska.chc@nasa2.ksc.nasa.gov>
References: <78tiu9$dd49@overload.lbl.gov>

On Fri, 29 Jan 1999 18:56:40 -0500, Russell J. Page, Sr. <rjpage@mit.edu> wrote:
>
>I'm presently trying to resolve a problem in an EPICS Sequencer Program.
>I need to create a delay and the SNL delay function is unsuitable for
>what I need to do. Using the escape mechanism with a usr function, I am
>able to create a vxWorks watchdog timer okay, but when the timer
>expires, I need to return to the sequencer code. At present, my test
>timer returns a message to my debug terminal as I expect (I use the
>printf function). I tried every permutation I can think of on the idea
>of exit(), wdDelete, wdCancel, etc. but all that happens is either a
>reboot of the target or the terminal hangs. ^C, ^Z etc. won't break it.

I have no idea what an EPICS Squencer or an SNL delay function are but
the reason your program is crashing is probably because of using the
printf statement in your watchdog timer.  Watchdog timer functions are
run at interrupt level and printf's are not allowed (they rely on
semaphores and possibly other interrupts which are -bad- things to
do in an ISR).  Try using the logMsg function instead.

Chuck


---------------------------

Newsgroups: comp.os.vxworks
Subject: Program Flash on MV2604 via program
Date: Tue, 26 Jan 1999 21:20:28 -0500
From: "David Garner" <daveg@progeny.net>
Message-ID: <78m0mg$qnq$1@winter.news.rcn.net>

Does anyone have a program, with source code, that we can run from the
target shell that will take a file name as an argument and program the flash
memory with the contents of the file?

Thanks,
Dave Garner




---------------------------

Newsgroups: comp.os.vxworks
Subject: x86 VxWorks -- Ethernet support
Date: Tue, 02 Feb 1999 17:41:37 -0800
From: LEE SYLVIA BO MING <sbmlee@ee.ubc.ca>
Organization: University of BC, Dept. of Electrical Engineering
Message-ID: <36B7A951.6459@ee.ubc.ca>

Hi all:

	I'm working on setting up the target and host communication on PC. 
Does anyone know whether the Ethernet card: intel etherexpress pro/10+
is supported by VxWorks?  If not, if there already a driver written for
it.

	Any help would be greatly appreciated.
- -- 
Yours,
Sylvia

---------------------------

Newsgroups: comp.os.vxworks
Subject: x86 VxWorks -- Ethernet support
Date: Tue, 02 Feb 1999 17:46:50 -0800
From: Lee Sylvia Bo Ming <j9o1@ugrad.cs.ubc.ca>
Organization: UBC Computer Science Undergraduate Laboratory
Message-ID: <36B7AA8A.A3260517@ugrad.cs.ubc.ca>

Hi all:

 I'm working on setting up the target and host communication on PC.
Does anyone know whether the Ethernet card: intel etherexpress pro/10+
is supported by VxWorks?  If not, if there already a driver written for
it.

 Any help would be greatly appreciated.
- --
Yours,
Sylvia



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: x86 VxWorks -- Ethernet support
Date: Wed, 03 Feb 1999 11:34:40 GMT
From: parag@cmcltd.com
Organization: Deja News - The Leader in Internet Discussion
Message-ID: <799c8g$5bg$1@nnrp1.dejanews.com>
References: <36B7AA8A.A3260517@ugrad.cs.ubc.ca>

VxWorks support Intel Ether Express PRO100B PCI. If you are talking about
that then answer is yes.

>  I'm working on setting up the target and host communication on PC.
> Does anyone know whether the Ethernet card: intel etherexpress pro/10+
> is supported by VxWorks?  If not, if there already a driver written for
> it.

- -----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

---------------------------

End of New-News digest
**********************


From owner-vxwexplo-process  Wed Feb  3 05:04:17 1999
From: "Gundula Blohm" <blohm@vs.dasa.de>
Date: Wed Feb  3 05:04:21 PST 1999
Subject: Re:ServerBasedTornado

Hi Tim,


what i did, was 

- installing Tornado on a WinNt4.0/SP3 Server

- exporting the created registry subtree
  [HKEY_CURRENT_USER\Software\Wind River Systems]
   via regedit to tornado.reg

- changing paths in the exported tornado.reg file in a text editor to match
  drive mappings on client machines

- changing paths in the tornado.lnk file to match drive
  mappings on client machines

- providing tornado.reg file and tornado.lnk file for the clients

now the clients do just have to doubleclick the tornado.reg file and to copy 
the tornado.lnk to their desktop.

this worked fine for me.

HTH,

dula

 
BTW:the registry and license was installed on the same server too.


===================================================
Gundula Blohm
Fon  +49 (731) 392-4731
Fax  +49 (731) 392-4958
mailto:blohm@vs.dasa.de
===================================================


From owner-vxwexplo-process  Wed Feb  3 05:17:02 1999
From: Charlie Grames <cgrames@mdc.com>
Date: Wed Feb  3 05:17:06 PST 1999
Subject: Ordering of memory cycles -Reply

David,

Perhaps I missed the origin of this thread, but what you suggest will do nothing useful on the 603 and 604 in Motorola's MVME2603 and 2604 implementations.  A write followed by a read from the same location will probably just happily return the value you just wrote from cache.

To quote from PowerPC Microprocessor Family: The Programming Environments:

"The PowerPC architecture provides for relaxed memory coherency.  Features such as write-back caching and out-of-order execution allow software engineers to exploit the performance benefits of weakly-ordered memory access.  The architecture also provides the means to control the order of accesses for order-critical operations."

The "means to control the order of accesses" is in the form of the "eieio" (yes, I expect someone had Old MacDonald on their mind when they dreamed up that mnemonic) and "sync" instructions.  These instructions are essential for ensuring data integrity in critical sections of code.

The whole memory coherency issue has been a bane of mine, recently, so if I have spoken out of turn, I apologize.  However, I don't want to leave the 603/604 community with the idea that following a store with a load will solve their problems--it won't.

Charlie Grames
The Boeing Company
(314) 233-1956
Charles.R.Grames@boeing.com

>>> the vxWorks Users Group Exploder <vxwexplo@lbl.gov> 02/02/99 11:51am >>>
Submitted-by owner-vxwexplo-process  Tue Feb  2 09:51:38 1999
Submitted-by: David Laight <dsl@tadpole.co.uk>

There have been a lot of deep discussions about the following code snippet:
*
> 	stw	...		// select the other register bank
> 	lwz	...		// load from that bank

I don't know about the 603e, but on most modern processors the above code 
doesn't work - in many with or without the cache enabled.

The easiest solution (works in every case I've seen, and is implementation 
independant) is to follow the 'write' with a 'read' of the same location.
This has the effect of forcing the write out of ALL the places where it might 
have got buffered - there can be many - before the read is done.

Clearing the write from ALL the places it might be buffered is particularly 
important when clearing an interrupt request.  If you don't then the o/s might 
report 'unexpected interrupt 0xxx' (vxworks interrupt code doesn't attemp to be 
that clever).

Unlike eieio (old MacDonald?) and obscure cache controller instructions, the 
read-back can be done in C.  Just ensure everything is volatile.

Buffering of writes first became a problem with the good old 286.  Allowing it 
to break the inter cycle recovery time of the 8259 interrupt controller and 
putting garbage into all of the controllers registers.


	David


----------------------------------------------------------------
David Laight                          email: dsl@tadpole.co.uk
Tadpole Technology plc                phone: +44 1223 278 256
Cambridge, UK                         fax:   +44 1223 278 201



From owner-vxwexplo-process  Wed Feb  3 07:27:10 1999
From: Hakan Karakas <h.karakas@ieee.org>
Date: Wed Feb  3 07:27:14 PST 1999
Subject: Re: VxWorks: SCSI boot of MVME1604

Hi,

Try assigning different SCSI id then 0.. As far as I know valid SCSI id
ranges 1 to 7 for one controller.
Regards,

Hakan Karakas

the vxWorks Users Group Exploder wrote:

> Submitted-by owner-vxwexplo-process  Tue Feb  2 04:48:10 1999
> Submitted-by: "Joe W." <joew@wtc.gov>
>
> Hello,
>
> I am unable to boot a MVME1604 board from the SCSI disk.  I have
> burned new boot roms, and turned on SCSI debug.  Here is the output:
>
> CPU: Motorola MVME1600 - PowerPC 604
> Version: 5.3
> BSP version: 1.1/5
> Creation Date: Feb 1, 1999, 14:14:50
>
> Press any key to stop auto-boot...
>  0
> auto-booting...
>
> boot device          : scsi
> unit number          : 0
> processor number     : 0
> host name            : appserver1
> file name            : /sd0/vxWorks
> inet on ethernet (e) : 79.0.2.215:ffffff00
> host inet (h)        : 79.0.2.213
> user (u)             : ast
> flags (f)            : 0x0
> target name (tn)     : test
>
> Calling scsi2IfInit...
> Returned from scsi IF init
> Calling sysScsiInit
> sysScsiInit: Turned on scsiDebug
> 0x7f6510 (tBoot) : Found an NCR810 Controller
> Calling scsiBusReset
> 0x7fed48 (tscsiTask) : ncr810Event: received event 52 (thread =
> 0x00000000 )
> calling scsiAutoConfig
> 0x7f6510 (tBoot) : scsiAutoConfig: bus ID = 0, LUN = 0




From owner-vxwexplo-process  Wed Feb  3 09:17:58 1999
From: David Laight <dsl@tadpole.co.uk>
Date: Wed Feb  3 09:18:01 PST 1999
Subject: Re: Intel PRO/10+ support

As I recall, the PRO/10+ PCI card has the same chipset as the PRO/100B (82557).
The device driver shouldn't need to know the difference.

Drivers for the PRO/100+ (82558) eil also work.

Beware of drivers for the original PRO/100 (82556).  These are incompatible with 
the later cards.

There may also be a PRO/10 ISA card - it is entirely different.

I'm not sure which of the these VxWorks supports...

	David
	
(I hope I've got the part numbers right, I'm most sure of the last digit.)


----------------------------------------------------------------
David Laight                          email: dsl@tadpole.co.uk
Tadpole Technology plc                phone: +44 1223 278 256
Cambridge, UK                         fax:   +44 1223 278 201



From owner-vxwexplo-process  Wed Feb  3 14:33:21 1999
From: Alexander_R_Povolotsky@res.raytheon.com
Date: Wed Feb  3 14:33:24 PST 1999
Subject: FPU control register problem - Intel Pentium VxWorks BSP related 

VxWorks Tornado

To all Intel (Pentium?) based BSP users - has anybody seen this happening ?

Subject :  FPU control register problem  - Intel Pentium VxWorks BSP
related ?!

Debugging Tool used:

Pentium Intel "assembly" level emulator from Microtek.

Summary:

It seems that a Floating Point exception has
occurred, and the CPU is blocking
all floating point operations until the exception is handled.  VxWorks
apparently has missed the floating point
exception (or it does not do any floating point exception handling).

Details:

In the "good" data sample, all of the floating point exception conditions
are masked (CW register).  This means that if an exception
occurs, the FPU will ignore it and provide an appropriate data value.  In
this "good" sample, three exceptions have occurred
that the FPU has ignored:
     Precision,
     Underflow,
     Denormalized Operation

In the "stalled" data sample, 5 of the 6 floating point exception
conditions have been unmasked. The status register indicates
that an error is present waiting to be handled, along with the same three
exception conditions and a condition code C3C2C1C0=0001.
At this point, the processor will not allow more floating point
instructions until the exception has cleared and hangs on "Fld"
assembly instruction.

The board is working in "DOS-compatible" floating point error handling
mode.  That means that two CPU pins FERR# and IGNFPE# are
used along with the system chipset to tell the processor when to stall if
exceptions are pending.  A write to I/O location 0xf0 will assert the
Ignore Floating Point Exception hardware line and allow further floating
point instructions to occur.  You can test this in the lab by issuing the
following emulator commands:

size byte;
write 0xf0p 0 byte io;

The next step is to figure out why the exception masks get changed in the
FPU control register.

Workaround: U






From owner-vxwexplo-process  Wed Feb  3 16:58:33 1999
From: ktsubota@keck.hawaii.edu (Kevin Tsubota)
Date: Wed Feb  3 16:58:36 PST 1999
Subject: spy and shell problems



Hi,

I've just started running vxWorks on a mvme2307 (powerPC) board
and am having the following problems (I've included an excerpt
from my target/config/all/configAll.h file at the end):

1) spy

   I've defined INCLUDE_SPY in config/all/configAll.h but when I
   type spy from the target shell is says: spy not configured.
   However when I do a lkup "spy" its built into my vxWorks.sym

   Can anyone tell me what I did wrong or didn't do?

-> spy
spy not configured into this system.
value = 37 = 0x25 = '%'

-> lkup "spy"
_func_spyClkStart         0x001b2674 bss      (vxWorks.sym)
spy                       0x00150410 text     (vxWorks.sym)
_func_spyReport           0x001b26c8 bss      (vxWorks.sym)
_func_spy                 0x001b2620 bss      (vxWorks.sym)
spyHelp                   0x0015054c text     (vxWorks.sym)
spyTask                   0x001503bc text     (vxWorks.sym)
spyClkStop                0x001504bc text     (vxWorks.sym)
spyStop                   0x00150504 text     (vxWorks.sym)
_func_spyTask             0x001b27e4 bss      (vxWorks.sym)
_func_spyClkStop          0x001b26f8 bss      (vxWorks.sym)
spyClkStart               0x00150464 text     (vxWorks.sym)
_func_spyStop             0x001b2554 bss      (vxWorks.sym)
spyReport                 0x00150368 text     (vxWorks.sym)
value = 0 = 0x0
-> 

2) target shell and use of "<"

   We've always used the "<" sign to execute script from the
   target shell but now when I do I get a syntax error.
   Is there something new that I need to include now?
   It does work from a windshell.   

-> 
-> <nfs.hq
syntax error
-> 



**** excerpt from target/config/all/configAll.h ****

I've added these #define files to the configAll.h files and nothing else:

/*                      INCLUDED SOFTWARE FACILITIES                          */

#define INCLUDE_LOADER          /* object module loading */
#define INCLUDE_SHELL           /* interactive c-expression interpreter */
#define INCLUDE_SHOW_ROUTINES   /* show routines for system facilities*/
#define INCLUDE_SPY             /* spyLib for task monitoring */
#define INCLUDE_STARTUP_SCRIPT  /* execute start-up script */
#define INCLUDE_STAT_SYM_TBL    /* create user-readable error status */
#define INCLUDE_SYM_TBL         /* symbol table package */
#define INCLUDE_TELNET          /* telnet-style remote login */
#define INCLUDE_UNLOADER        /* object module unloading */

#define INCLUDE_WDB		/* tornado debug agent */
/* allows use of windsh but also disables native debugging */
#define INCLUDE_SYM_TBL_SYNC	/* synchronize host and target symbol tables */

/*                         INCLUDED NETWORK FACILITIES                        */

#define INCLUDE_NET_SHOW        /* network info and status facilities */
#define INCLUDE_NET_SYM_TBL     /* load symbol table from network */
#define INCLUDE_NFS             /* nfs package */
#define INCLUDE_PING		/* ping() utility */
#define INCLUDE_RDB             /* remote debugging package */
#define INCLUDE_RLOGIN          /* remote login */
#define INCLUDE_TFTP_SERVER	/* tftp server */



----- End Included Message -----



From owner-vxwexplo-process  Wed Feb  3 22:46:16 1999
From: Miri Epstein <mirie@telegate.co.il>
Date: Wed Feb  3 22:46:20 PST 1999
Subject: DEC 21140 driver on MPMC 202

Hi,

I was wondering if someone is working with this PMC card, and has the
driver for it.

We have a mv2304 (ppc603) board with the DEC 21140 on it. We are working

with SENS 1.0 and with the driver object  supplied for this chip that
seems to be working
(not the END model).
We have added a second ethernet interface by plugging in the MPMC 202.
We took the
unsupported vxworks driver for the DEC 21140 and ported it to 4.4BSD.
This driver ported works in 10Mbps and does NOT work in 100Mbps.

Does anybody has the MPMC 202 working in auto sense mode ?

Any help will be appreciated,

TIA

Miri




From owner-vxwexplo-process  Wed Feb  3 23:41:44 1999
From: "Michael Lawnick" <lawnick@softec.de>
Date: Wed Feb  3 23:41:47 PST 1999
Subject: Re: spy and shell problems

Kevin Tsubota wrote:
> Hi,
> 
> I've just started running vxWorks on a mvme2307 (powerPC) board
> and am having the following problems (I've included an excerpt
> from my target/config/all/configAll.h file at the end):
> 
> 1) spy
> 
>    I've defined INCLUDE_SPY in config/all/configAll.h but when I
>    type spy from the target shell is says: spy not configured.
>    However when I do a lkup "spy" its built into my vxWorks.sym
> 
>    Can anyone tell me what I did wrong or didn't do?
> 
--snip --

Hi Kevin,

as far as I know, a mvme2307 has no auxiliary clock, but SPY needs 
it, so starting it up will fail.
Nevertheless, I wonder how you could compile your BSP, as there is 
implemented a controll mechanism in config.h and you should get an 
error message like
vxWorks.tmp: In function `spyClkStartCommon':
vxWorks.tmp(.text+0x3cee4): undefined reference to `sysAuxClkConnect'
vxWorks.tmp(.text+0x3cf10): undefined reference to `sysAuxClkRateGet'
vxWorks.tmp(.text+0x3cf20): undefined reference to `sysAuxClkRateSet'
vxWorks.tmp(.text+0x3cf24): undefined reference to `sysAuxClkEnable'
vxWorks.tmp: In function `spyClkStopCommon':
vxWorks.tmp(.text+0x3cf74): undefined reference to `sysAuxClkRateSet'
vxWorks.tmp(.text+0x3cf78): undefined reference to `sysAuxClkDisable'
make.exe: *** [vxWorks] Error 0x1
on building your BSP.

Are you sure that your BSP makefile contains the correct constant for 
your board type ?
I think it should be
BOARD		= MVME2300
(in config.h it will be transformed in some def's and undefs)

I hope this help

Michael
MfG
Lawnick, SOFTEC GmbH

==============================================
SOFTEC GmbH               Tel +49-731-96600-0
Promenade 17              Fax +49-731-96600-23
D-89073 Ulm               Michael Lawnick
Germany                   lawnick@softec.de
==============================================


From owner-vxwexplo-process  Thu Feb  4 04:00:15 1999
From: daemon@csg.lbl.gov
Date: Thu Feb  4 04:00:18 PST 1999
Subject: comp.os.vxworks newsdigest

Comp.Os.Vxworks Daily Digest    Thu Feb  4 04:00:11 PST 1999

        Subject: Re: Why my target server can not connect to target?
        Subject: Re: PCI Interrupt Handler question.
        Subject: Re: Booting from flash
        Subject: Re: Booting from flash
        Subject: How to write an interrupt routine in C with a 68K GNU2.7 ? #pragma interrupt
        Subject: Re: PCI Interrupt Handler question.
        Subject: Disabling ICMP redirects on a per interface basis
        Subject: Re: PCI Interrupt Handler question.
        Subject: Problem with Motorola FADS860 boot code
        Subject: Re: Booting from flash

-------------------------------------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Why my target server can not connect to target?
Date: Mon, 1 Feb 1999 08:25:06 +0800
From: "Wang Jian" <scewjian@sbell.com.cn>
Organization: Alcatel/Bell
Message-ID: <36b4f607.0@dnews.sbell.com.cn>

Wang Jian wrote:
>
> Hi, all
>
> If I boot VxWorks via Ethernet, the target server can always connect to
> target.
> But If I use the rom-resident image, the target server cannot connect to
> target.
> It seems that the WDB agent in target didn't start up. Why ? My BSP is
> ads860.

I've seen this happen if I didn't name the ethernet device driver in
the "other (o)" field and place valid info in the "Inet on ethernet (e)"
field.  The WDB agent tries to start during the boot, but if the
network hasn't been started WDB terminates.  When you boot over the
network the network is started automatically, but if you boot from
ROM you have to tell it by filling-in the "other" field.

Good luck
Frank.




---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: PCI Interrupt Handler question.
Date: Mon, 1 Feb 1999 03:53:02 +0100
From: "arve sollie" <arve.sollie@drammen.mail.telia.com>
Message-ID: <36b51754.0@d2o201.telia.com>
References: <78tmae$onc@news.or.intel.com>

This is not neccessarily  a PCI related problem.  On most hardware you are
required to reset the interrupt status register
to enable a new interrupt. This is often done by writing the
content of the IS req. back to itself.

rvsaripa wrote in message <78tmae$onc@news.or.intel.com>...
>I am new to vxworks and I am trying to figure out the following about
>interrupt processing on vxworks.
>
>I have a PCI card and I have hooked my own interrupt handler using
>pciIntconnect().  My interrupt handler is getting called. But the device
>interrupts only once. My target is a 486-class machine.
>
>In my interrupt handler, do I have to do something to enable further
>interrupts or is it done automatically by vxworks when my handler returns
>back to the base routine installed in the interrupt vector?.
>
>Any answers will be appreciated.
>
>Thanks,
>rk
>
>



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Booting from flash
Date: Wed, 03 Feb 1999 10:15:57 +0000
From: Ian Love <ijl@wireplay.bt.co.uk>
Organization: BT Labs - but speaking only for myself
Message-ID: <36B821DD.22AE@wireplay.bt.co.uk>
References: <7990ii$hl53@overload.lbl.gov>

Johan Bostedt wrote:
> We are designing a custom board based on the MCP860 processor. I have a
> question concerning
> the loading of the boot code and the VxWorks image. We would like to
> have it all in flash memory.
> 
> Is it possible to acheive this without a boot PROM?
> 
> What do you need to program the boot monitor and the VxWorks image into
> flash?
> We have an ethernet port and an RS232 port connected to the
> microprocessor.
> 
> What about sizes? How big is the bootrom code and how
> big is a typical VxWorks image?
> 
> Best regards,
> Johan Bostedt

We have achieved a single step boot straight into vxworks: the 'boot'
prom version is basically 'only' vxWorks with bootConfig.c instead of
usrConfig.c....  We added our own startup bits (ethernet etc) into 
usrConfig.c, modifying a few things  along the way:

for example, the main difference between the 2 boots is the call
to usrBootLineInit(): for single step boot must be BOOT_CLEAR to 
force a read of the boot string from NVRAM (if you have it) or 
DEFAULT_BOOT_LINE

hence:
    usrBootLineInit (BOOT_CLEAR);	/* crack the bootline */

We have also done our own boot prompt in the middle of our boot to 
allow parameter changes, and we interpret most of the bootline ourselves
to configure things our way....

For programming flash - see my previous postings (eg look on DEJA news)
for notes and a code example for the chips we have on our MVME2305 
boards: the code is vxWorks and oriented around that board, but 
the concepts will be useful (sectors etc). basically you need to select
a chip-set, get hold of the manuals (usually on the net in pdf form),
and work out how to apply the appropriate programming commands...
(they are all different!)

Your main problem is how to accomplish the first programming of the 
flash: either get a board with the same chips and use that as a 
programmer, or you will have to go the whole hog and get yourself
a programmer. Unless that is you can persuade your board to boot
vxWorks off the serial port in the first instance, in which case 
you're sorted (SLIP?)

As for size of image - depends what's in it/Processor: our last 
kernel (1-step-boot) for the PPC (MPC604) was under 400k _compressed_ - 
including some basic startup of our devising, + TCP/IP stack 
(pre-SENS), target shell, standalone symbol table etc 
- - will vary with other processors
- - uncompressed kernel is more like 700kb
- - your code may or may not add significantly to this....

hope this helps - if you need any more detail, get in touch direct

regards,
Ian Love

The views expressed above are my own and not those of my employer

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Booting from flash
Date: Wed, 03 Feb 1999 08:13:31 -0500
From: "James A. Littlefield" <jal@alum.mit.edu>
Organization: Shore.Net/Eco Software, Inc; (info@shore.net)
Message-ID: <36B84B7B.67203602@alum.mit.edu>
References: <7990ii$hl53@overload.lbl.gov>

We are using a modified version of the VxWorks bootproms in flash (about
160k).   Our application code is about 500k compressed.
The '860 board I work on has a port for the Background Debug Mode (BDM) and
we use a product from EST (VisionIce) which works very nicely.   Using this
device you can program flash on the target as well as single step, step
breakpoints etc.   If you contemplate any serious customization of the boot
process or integration with special hardware (ie you write the drivers), I
would recommend getting an ICE.

Regards,
Jim

Johan Bostedt wrote:

> Hi VxWorkers,
>
> We are designing a custom board based on the MCP860 processor. I have a
> question concerning
> the loading of the boot code and the VxWorks image. We would like to
> have it all in flash memory.
>
> Is it possible to acheive this without a boot PROM?
>
> What do you need to program the boot monitor and the VxWorks image into
> flash?
> We have an ethernet port and an RS232 port connected to the
> microprocessor.
>
> What about sizes? How big is the bootrom code and how
> big is a typical VxWorks image?
>
> Best regards,
> Johan Bostedt
> _________________________________________________
> Johan Bostedt     phone: +46(0)8-404 87 96
> RSA/B/UD          cellular: +46(0)70-657 64 70
>                   email: qsabost@rsa.ericsson.se




---------------------------

Newsgroups: comp.os.vxworks
Subject: How to write an interrupt routine in C with a 68K GNU2.7 ? #pragma interrupt
Date: Wed, 3 Feb 1999 11:56:03 -0800
From: "Herve Pellan" <hpellan@rose.hp.com>
Organization: Hewlett Packard Cupertino Site
Message-ID: <79a9lg$b1a$1@ocean.cup.hp.com>

I don't want to use intConnect to write some very critical interrupt (very
frequent and very small piece of code to be executed in there) in C.

I tried to use the    #pragma interrupt    before my interrupt routine but
when I look at the assembly code generated there is still a "RTS" at the end
of the routine instead of a "RTE" assembly code.

Does anybody know how to generate from a piece of C code an interrupt
routine for a Motorola 68k environment ?

Thanks
Herve




---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: PCI Interrupt Handler question.
Date: Mon, 01 Feb 1999 11:50:03 GMT
From: parag@cmcltd.com
Organization: Deja News - The Leader in Internet Discussion
Message-ID: <7944da$n5m$1@nnrp1.dejanews.com>
References: <78tmae$onc@news.or.intel.com>

Hi,
   It is not at all needed to do anything since your ISR has executed once.
My doubt is you might have used some printf statement in your ISR. If
it so then remove it.
Bye
Parag

In article <78tmae$onc@news.or.intel.com>,
  "rvsaripa" <ramakrishna_saripalli@intel.com> wrote:
> I am new to vxworks and I am trying to figure out the following about
> interrupt processing on vxworks.
>
> I have a PCI card and I have hooked my own interrupt handler using
> pciIntconnect().  My interrupt handler is getting called. But the device
> interrupts only once. My target is a 486-class machine.
>
> In my interrupt handler, do I have to do something to enable further
> interrupts or is it done automatically by vxworks when my handler returns
> back to the base routine installed in the interrupt vector?.
>
> Any answers will be appreciated.
>
> Thanks,
> rk
>
>

- -----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

---------------------------

Newsgroups: comp.os.vxworks
Subject: Disabling ICMP redirects on a per interface basis
Date: Wed, 03 Feb 1999 17:47:39 -0600
From: Andy Mauro <amauro@broadbandnetworks.com>
Organization: Nortel Networks
Message-ID: <36B8E01B.976722AA@broadbandnetworks.com>

Howdy all.

I need to turn off ICMP redirects only on certain interfaces.  Does
anyone know how to do this?  I realize that I merely need to set the IP
flags to disable ICMP redirects, but this is a global change.

Unfortunately I cannot check every packet to see if it is an ICMP
redirect, and cannot filter it at the destination.

Thanks in advance,
- -Andy

Andy Mauro
Software Designer
Nortel Networks
Broadband Wireless Access Unit
amauro@broadbandnetworks.com



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: PCI Interrupt Handler question.
Date: Mon, 01 Feb 1999 10:02:51 -0600
From: gcox <gcox@sed.jsc.nasa.gov>
Organization: NASA, Kennedy Space Center
Message-ID: <36B5D02B.ABB36AAC@sed.jsc.nasa.gov>
References: <78tmae$onc@news.or.intel.com> <7944da$n5m$1@nnrp1.dejanews.com>
Reply-To: gacox@hal-pc.org

I had a 1553 card which required me to clear a bit in a register after I
received an interrupt. See if your card has something like that.

parag@cmcltd.com wrote:

> Hi,
>    It is not at all needed to do anything since your ISR has executed once.
> My doubt is you might have used some printf statement in your ISR. If
> it so then remove it.
> Bye
> Parag
>
> In article <78tmae$onc@news.or.intel.com>,
>   "rvsaripa" <ramakrishna_saripalli@intel.com> wrote:
> > I am new to vxworks and I am trying to figure out the following about
> > interrupt processing on vxworks.
> >
> > I have a PCI card and I have hooked my own interrupt handler using
> > pciIntconnect().  My interrupt handler is getting called. But the device
> > interrupts only once. My target is a 486-class machine.
> >
> > In my interrupt handler, do I have to do something to enable further
> > interrupts or is it done automatically by vxworks when my handler returns
> > back to the base routine installed in the interrupt vector?.
> >
> > Any answers will be appreciated.
> >
> > Thanks,
> > rk
> >
> >
>
> -----------== Posted via Deja News, The Discussion Network ==----------
> http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own




---------------------------

Newsgroups: comp.os.vxworks
Subject: Problem with Motorola FADS860 boot code
Date: Thu, 04 Feb 1999 11:13:24 +0200
From: Rami Gideoni <rami@chromatis.com>
Organization: Chromatis
Message-ID: <36B964B4.5375CDCC@chromatis.com>

Hi,

I had installed Tornado for MPC860 and I'm using the Motorola's FADS860
evaluation board.

Please note : The board is FADS860 - not ADS860.

I burned down the bootrom.hex over the flash and find out a strange phenomena :

When I use MPC860EN or MPC860T derivatives in the board, the boot code functions
as expected.

- -- BUT ---

when I use MPC860SAR derivative - No response from the board (acctually, the
transmit LED of SMC 1 is turning as usuall - but no output on the terminal).

I changed the frequency of the boot code from 4 Mhz crystal/24 Mhz system clock
to  5 Mhz/50 Mhz system clock and I was suprised to find out that the MPC860SAR
started to work !

Any ideas ?


Thanks in advance

Rami Gideoni

[WindRiver techincal support : TSR# 124446, For Mr. Gideon Chan]



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Booting from flash
Date: Thu, 4 Feb 1999 11:56:39 +0200
From: "Saffi Hartal" <saffi@globaloop.com>
Organization: NetVision Israel	
Message-ID: <79bqsn$mc6$1@news.netvision.net.il>
References: <7990ii$hl53@overload.lbl.gov>


Johan Bostedt wrote in message <7990ii$hl53@overload.lbl.gov>...
>Hi VxWorkers,
>
>We are designing a custom board based on the MCP860 processor. I have a
>question concerning
>the loading of the boot code and the VxWorks image. We would like to
>have it all in flash memory.
>
>Is it possible to acheive this without a boot PROM?
>
>What do you need to program the boot monitor and the VxWorks image into
>flash?
>We have an ethernet port and an RS232 port connected to the
>microprocessor.
>
>What about sizes? How big is the bootrom code and how
>big is a typical VxWorks image?
>
>
>Best regards,
>Johan Bostedt
>_________________________________________________
>Johan Bostedt     phone: +46(0)8-404 87 96
>RSA/B/UD          cellular: +46(0)70-657 64 70
>                  email: qsabost@rsa.ericsson.se
>
>


Hi
We have 2Meg flash. We have enhanced boot supporting flash file Redding and
writing (size of 230K).
then we load image compressed or from network or from flash (the file size
730K the regular vxWorks.st is 3.8 M I think the no_sym should be about
1.6M ).
So we can have too versions complete in the flash including boot :-).
farther more ... the loading of compressed file by network is faster. better
for upgrading software.
We burn boot but we can download running version into flash and reload
system.

        Bye Saffi





---------------------------

End of New-News digest
**********************


From owner-vxwexplo-process  Thu Feb  4 09:50:44 1999
From: "Staub, Phil" <Phils@audioprecision.com>
Date: Thu Feb  4 09:50:48 PST 1999
Subject: RE: FPU control register problem - Intel Pentium VxWorks BSP rela

A couple of things here:

1. I don't remember where I read it, (it may have been in some code) but
VxWorks does *not* provide handlers for floating point exceptions. One would
assume that if you want (or need) to handle them, it's up to you to develop
code to do so, presumably since such handling is beyond the scope of
something that can be done in a generic way that is acceptable to all
applications.

2. About 6 months ago, I debugged a floating point problem that sounds quite
similar to what you're seeing. You didn't include quite enough information
for me to determine if it's exactly what I was seeing or not, and if you
did, I'm not sure I would recognize the register bits as the same as I saw.
However, I *do* remember that, after sorting through all of the status bits,
I determined that the exception I saw was a floating point stack overflow.
After that point, operations that tried to load a value on the floating
point stack (virtually anything) would fail. 

The problem turned out to be an incorrect function prototype(!). I had a (C)
function that returned a double, but a header file containing its prototype
claimed it returned int. So the function returns the value on the floating
point stack, but the caller never popped the value off the stack. After a
few calls, the stack filled up (it is only 8 entries deep, anyway), and
future attempts at loading values on the stack failed. 

I was ready to report a compiler bug until I looked at the prototype. 

I guess the moral to the story is something I have found quite often: before
I blame the compiler, I make sure I'm not doing something wrong. Once I'm
sure I'm not doing something wrong, I check my code again. Twice. Then I
*might* blame the compiler.

This may or may not be specifically related to your problem, but for what
it's worth, I know how difficult it can be to debug floating point problems
(particularly since Crosswind in Tornado 1.0.1 does not correctly report
floating point register contents, and I didn't have a hardware debugger to
use).

Hope this gives you something to think about. 

Phil Staub, phils@audioprecision.com
Senior Software Engineer
Audio Precision, Inc.
(503) 627-0832
(800) 231-7350

> -----Original Message-----
> From:	vxwexplo@lbl.gov [SMTP:vxwexplo@lbl.gov]
> Sent:	Wednesday, February 03, 1999 2:33 PM
> To:	vxworks_users@csg.lbl.gov
> Subject:	FPU control register problem - Intel Pentium VxWorks BSP
> related 
> 
	[SNIP] 
	  
> Summary:
> 
> It seems that a Floating Point exception has
> occurred, and the CPU is blocking
> all floating point operations until the exception is handled.  VxWorks
> apparently has missed the floating point
> exception (or it does not do any floating point exception handling).
> 
> 


From owner-vxwexplo-process  Thu Feb  4 15:10:40 1999
From: Charlie Grames <cgrames@mdc.com>
Date: Thu Feb  4 15:10:43 PST 1999
Subject: RMW on MVME2604/2700

VxWorks 5.3.1
MVME2700 1.1/4

Has anyone out there tried to use the read-modify-write feature of the
Universe II chip on the MVME2604 or 2700?  Since the PowerPC doesn't have a
true atomic read-modify-write capability, I don't see how the Universe
implementation could work.  If someone out there knows otherwise, please let
me know.  Thanks.

Charlie Grames
The Boeing Company
(314) 233-1956
Charles.R.Grames@boeing.com


From owner-vxwexplo-process  Fri Feb  5 02:07:54 1999
From: Patrick.Soulard@alcatel.fr
Date: Fri Feb  5 02:07:58 PST 1999
Subject: RE: Vxworks 486 BSP - reclaiming the bottom meg?


--openmail-part-03442a86-00000001
Content-Type: text/plain; charset=US-ASCII; name="BDY.TXT"
Content-Disposition: inline; filename="BDY.TXT"
Content-Transfer-Encoding: 7bit

Roddy,

There are 2 methods to reclaim the PC low memory below 640 Ko :

1) Create, with memPartCreate(), a new memory partition starting at 0x5000 and 
ending at 0x9FFFF (see VxWorks 5.3.1 Programmer's Guide p 554)
2) Add to system memory partition, with memAddToPool(), the memory block 
starting at 0x5000 and ending at 0x9FFFF 

You can also use this area as a RAM Disk (see ramDrv() and ramDevCreate() 
functions)


------------------------------------------------------------------------------
 Patrick SOULARD
 patrick.soulard@alcatel.fr
 patrick_soulard@csi.com   COMPUSERVE: 100635,1717
 Alcatel CGA Transport  BP 57  91229 Bretigny-sur-Orge  FRANCE
------------------------------------------------------------------------------


--openmail-part-03442a86-00000001
Content-Type: application/x-openmail-1734; name="WINMAIL.DAT"
Content-Disposition: attachment; filename="WINMAIL.DAT"
Content-Transfer-Encoding: base64

eJ8+IgiJAQaQCAAEAAAAAAABAAEAAQeQBgAIAAAA5AQAAAAAAADoAAEIgAcAGAAAAElQTS5N
aWNyb3NvZnQgTWFpbC5Ob3RlADEIAQ2ABAACAAAAAgACAAEEgAEAMQAAAFJFOiBWeHdvcmtz
IDQ4NiBCU1AgLSByZWNsYWltaW5nIHRoZSBib3R0b20gbWVnPwDyDwEFgAMADgAAAM8HAgAF
AAsACQAQAAUABgEBAIAAAEIAAAAEAEIAEgAgAFNPVUxBUkQsIFBBVFJJQ0sAAE9QRU5NQUlM
OlBBVFJJQ0sgU09VTEFSRCAvQ0dBSDMAAAAAAAAAAABYDQEggAMADgAAAM8HAgAFAAkAKAAD
AAUAFgEBCYABACEAAAA4REY4MkU4QjFGQkNEMjExOTRGRDAwMTA1QUJFRjk4OABjBwEDkAYA
KAMAACAAAAALAAIAAQAAAAsAIwAAAAAAAwAmAAAAAAALACkAAAAAAAMALgAAAAAAAwA2AAAA
AABAADkAsBFmlu9QvgEeAEIAAQAAABEAAABTT1VMQVJELCBQQVRSSUNLAAAAAEAASACw2RWX
71C+AR4AcAABAAAAMQAAAFJFOiBWeHdvcmtzIDQ4NiBCU1AgLSByZWNsYWltaW5nIHRoZSBi
b3R0b20gbWVnPwAAAAACAXEAAQAAABYAAAABvlDvlk2LLviOvB8R0pT9ABBavvmIAAAeABoM
AQAAABEAAABTT1VMQVJELCBQQVRSSUNLAAAAAAIBHQwBAAAAIAAAAE9QRU5NQUlMOlBBVFJJ
Q0sgU09VTEFSRCAvQ0dBSDMAHgAeDAEAAAAJAAAAT1BFTk1BSUwAAAAAHgAfDAEAAAAXAAAA
UEFUUklDSyBTT1VMQVJEIC9DR0FIMwAAHgBCEAEAAAABAAAAAAAAAAMAgBD/////QAAHMBB4
DiDjUL4BQAAIMBB4DiDjUL4BCwAAgAggBgAAAAAAwAAAAAAAAEYAAAAAA4UAAAAAAAADAASA
CCAGAAAAAADAAAAAAAAARgAAAABShQAAdBAAAB4ABYAIIAYAAAAAAMAAAAAAAABGAAAAAFSF
AAABAAAABQAAADguMDIAAAAAAwAGgAggBgAAAAAAwAAAAAAAAEYAAAAAAYUAAAAAAAALAA+A
CCAGAAAAAADAAAAAAAAARgAAAAAOhQAAAAAAAAMAEIAIIAYAAAAAAMAAAAAAAABGAAAAABCF
AAAAAAAAAwARgAggBgAAAAAAwAAAAAAAAEYAAAAAEYUAAAAAAAADABSACCAGAAAAAADAAAAA
AAAARgAAAAAYhQAAAAAAAB4AJIAIIAYAAAAAAMAAAAAAAABGAAAAADaFAAABAAAAAQAAAAAA
AAAeACWACCAGAAAAAADAAAAAAAAARgAAAAA3hQAAAQAAAAEAAAAAAAAAHgAmgAggBgAAAAAA
wAAAAAAAAEYAAAAAOIUAAAEAAAABAAAAAAAAAB4APQABAAAABQAAAFJFOiAAAAAAAwANNP03
AAB/bA==

--openmail-part-03442a86-00000001--



From owner-vxwexplo-process  Fri Feb  5 04:00:32 1999
From: daemon@csg.lbl.gov
Date: Fri Feb  5 04:00:37 PST 1999
Subject: comp.os.vxworks newsdigest

Comp.Os.Vxworks Daily Digest    Fri Feb  5 04:00:29 PST 1999

        Subject: Routing Problem
        Subject: DOS-like format
        Subject: VxWorks PPC860
        Subject: Re: MPC860 and cache
        Subject: Re: Routing Problem
        Subject: Question about taskTcb ()
        Subject: Servlet & Wind Web Server ?
        Subject: Re: VxWorks PPC860
        Subject: Re: Get time from an NT Host?
        Subject: Re: What is RT-11 file system?
        Subject: use of tt () inside code
        Subject: Disabling console output for a ROMed application

-------------------------------------------------------

Newsgroups: comp.os.vxworks
Subject: Routing Problem
Date: Thu, 04 Feb 1999 13:44:37 +0100
From: christoph.nadig@stest.ch (Christoph Nadig)
Organization: Wavetek Wandel & Goltermann
Message-ID: <christoph.nadig-0402991344370001@yoda.alcatel.ch>

Hello,

we've ancountering problems with VxWorks 5.3.1 and IP routing
of the standard (not SENS) TCP/IP stack.
We like to use private network addresses according to RFC1918 for our
VxWorks boards. The default router, that should be used by these
boards, however, have legal IP addresses inside our company network.

Obviously VxWorks does not accept default routes that do bot
belong to the network number of the interface.
Any Clues?

Thanks, Christoph

---------------------------

Newsgroups: comp.os.vxworks
Subject: DOS-like format
Date: Thu, 4 Feb 1999 08:29:05 -0500
From: "Joe Zabaga" <Joseph.Zabaga@unisys.com>
Organization: ECDC Unisys Corp. Tredyffrin
Message-ID: <79c70m$lg3$1@trsvr.tr.unisys.com>

I am looking for a Vxworks/i960 version of a format program
that creates a file system just like DOS 5.0 (checks for bad blocks, writes
a partition boot sector, FAT, and root directory).

Thanks in advance.

Joe

gjzabaga@mcs.drexel.edu



---------------------------

Newsgroups: comp.os.vxworks
Subject: VxWorks PPC860
Date: 4 Feb 1999 14:12:51 GMT
From: "Clive Walley" <cwalley@ndsuk.com>
Organization: NDS Broadcast, Eastleigh, UK
Message-ID: <01be5048$7222c020$43c310ac@ndsuk1310.stone.ndsuk.com>

I am using vxWorks on PowerPC microprocessor, and am having particular
problems with my program running on it. Within a certain function, the
program crashes without warning. However, just recompiling and changing the
program slightly causes the problem to go away. I am beginning to suspect
the microprocessor. Does anyone know if there are any problems with release
A3 of the silicon? If so, are they documented anywhere?

Many thanks ....in anticipation.

Clive

---------------------------

Newsgroups: comp.os.vxworks,comp.sys.powerpc.tech
Subject: Re: MPC860 and cache
Date: 1 Feb 1999 14:51:11 +0800
From: skipper@no-spam.calweb.com (Skipper Smith)
Organization: Helpful Knowledge Consulting
Message-ID: <slrn7bcbuv.1f92.skipper@web1.calweb.com>
References: <petekz-2701991950220001@dal-tsa31-56.cyberramp.net> <petekz-2801991927100001@dal-tsa12-61.cyberramp.net> <78sgd8$a2n$1@user2.teleport.com> <78tfj9$enn$1@nnrp1.dejanews.com> <794evs$f6s$1@user2.teleport.com>
Reply-To: skipper@no-spam.calweb.com

Andrew Klossner <andrew@user2.teleport.com> wrote:
>>> I work with 603e ...
>>> 	stw	...		// select the other register bank
>>> 	lwz	...		// load from that bank
>
>> In this case, you should perform an eieio between the stw and lwz.
>
>Nope.  From the PowerPC 603e Microprocessor Programming Manual, page 2-39:
>
>	"The Enforce In-Order Execution of I/O (eieio) instruction
>	is used to ensure memory reordering of noncacheable memory access.
>	Since the 603e does not reorder noncacheable memory accesses,
>	the eieio instruction is treated as a no-op."
>
>In the case under discussion, boot ROM code, the MMU is not enabled so
>the 603e thinks that all locations are write-through cacheable.  The

Everything is probably correct here, except when the MMU is off, all
accesses are treated as write-BACK, not write-THROUGH, cacheable.

>cache is disabled (via the HID0 register) and no actual caching
>occurs, yet the CPU feels free to reorder the accesses.  I use sync
>instructions to preserve order.
>
>  -=- Andrew Klossner (andrew@teleport.com)


- -- 
Skipper Smith                         Helpful Knowledge Consulting
Worldwide         Microprocessor       Architecture       Training
PowerPC, ColdFire, 68K, CPU32                Hardware and Software

/* Remove no-spam. from the reply address to send mail directly */

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Routing Problem
Date: Thu, 04 Feb 1999 11:35:55 -0500
From: Rich Boyd <rlboyd@innocon.com>
Organization: Innovative Concepts, Inc
Message-ID: <36B9CC6B.AA0FC2DC@innocon.com>
References: <christoph.nadig-0402991344370001@yoda.alcatel.ch>

Try adding a route to the gateway via the interface's IP address.

You should be able to use a host or network route.

For example, if your interface is 200.1.1.1, and your gateway is
201.1.1.1:

   routeAdd "201.1.1.1", "200.1.1.1"
   routeAdd "0", "201.1.1.1"



Christoph Nadig wrote:
> 
> Hello,
> 
> we've ancountering problems with VxWorks 5.3.1 and IP routing
> of the standard (not SENS) TCP/IP stack.
> We like to use private network addresses according to RFC1918 for our
> VxWorks boards. The default router, that should be used by these
> boards, however, have legal IP addresses inside our company network.
> 
> Obviously VxWorks does not accept default routes that do bot
> belong to the network number of the interface.
> Any Clues?
> 
> Thanks, Christoph

- -- 
Richard L Boyd
Innovative Concepts, Inc         rlboyd@innocon.com
8200 Greensboro Dr, Suite 801    Phone: 703-893-2007 x277
McLean, VA 22102                 Fax:   703-893-5890

---------------------------

Newsgroups: comp.os.vxworks
Subject: Question about taskTcb ()
Date: Thu, 04 Feb 1999 12:01:12 -0800
From: Brian <bmcgonig@cisco.com>
Organization: Cisco Systems, Inc.
Message-ID: <36B9FC88.A52D9E72@cisco.com>

hi vxworksers:

Im working with 5.2 on a MIPS 4650.

Unfortunately I'm getting an ALE expcetion (bad address) inside the
taskTcb() function.  However, Im not calling that function anywhere in
the code.
Does anyone know if  the os calls this function, and when?

On a second note.. does anyone have any stack trace code for the
MIPS whereby I could find out the sequence of calls leading up to
the exception?


Thanx a lot

Brian


---------------------------

Newsgroups: comp.os.vxworks
Subject: Servlet & Wind Web Server ?
Date: Thu, 4 Feb 1999 11:17:37 +0100
From: "Sven Longolius" <sven.longolius@detewe.de>
Organization: News from a DeTeWe Group
Message-ID: <36b97355.0@ctb.detewe.de>

hi everyone,

Is there a support for servlets in the Wind Web Server ? I know that VxWorks
is able to run Java Bytecode.

Best Regards
- --
- ---------------------------------------
| Sven Longolius
| CorTech A/S , Buero Berlin
| Koepenicker Str. 178 , 10997 Berlin
| voice: +49 30 6104 4235
| fax:     +49 30 6104 4567
| eMail: sven.longolius@detewe.de
- ---------------------------------------




---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: VxWorks PPC860
Date: Thu, 04 Feb 1999 21:24:00 -0600
From: Pierre Olivier <polivier@netcom.ca>
Organization: Netcom Canada
Message-ID: <36BA6450.7E9D@netcom.ca>
References: <01be5048$7222c020$43c310ac@ndsuk1310.stone.ndsuk.com>
Reply-To: polivier@netcom.ca

Clive Walley wrote:
> 
> I am using vxWorks on PowerPC microprocessor, and am having particular
> problems with my program running on it. Within a certain function, the
> program crashes without warning. However, just recompiling and changing the
> program slightly causes the problem to go away. I am beginning to suspect
> the microprocessor. Does anyone know if there are any problems with release
> A3 of the silicon? If so, are they documented anywhere?
> 
> Many thanks ....in anticipation.
> 
> Clive

There are indeed some problems with the A3 version of the CPU.
You can get the errata information at

http://www.mot.com/SPS/RISC/netcomm/support/index.html#errata_group



- ---------------------------------------------
Pierre Olivier
Montreal (QC) Canada
polivier@netcom.ca               (. .)
- ------------------------------oOO-(_)-OOo----

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Get time from an NT Host?
Date: Thu, 04 Feb 1999 19:03:51 -0800
From: Stephen Macmanus <stephen@wrs.com>
Organization: Wind River Systems
Message-ID: <36BA5F97.2A6B@wrs.com>
References: <36A396C7.5D862297@boeing.com> <36BA1C45.95BC4DB1@ibm.net>

Armando Paz wrote:
> 
> Ken Ostaszewski wrote:
> 
> > Hi everyone,
> >
> > I was wondering if there was a way to get the host system time & date
> > from an NT machine.  I know of the way to do it from an Unix host, but
> > this method doesn't work for NT.
> >
> > Any help would be appreciated.
> >
> > Ken Ostaszewski
> 
> I didn't check this, but there's an ICMP request normally used to measure
> the transit time of packets in a network. From the information that I've
> got, the reply for the request carries the current value for the time of
> day in the machine. And as far as I know NT answers ICMP requests, at
> least ECHO request :-)
> 
> You should check the format of the reply.
> 
> Bye,
> Armando


     You can also get the time over the network using SNTP. Presumably
     NT will/does support that protocol.

- -- 
- ------------------
Stephen Macmanus                         #include <std_disclaimer.h>
Software Engineer
Wind River Systems                       stephenm@wrs.com

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: What is RT-11 file system?
Date: Fri, 05 Feb 1999 08:44:12 GMT
From: John or Jenn <lastimp@home.com>
Organization: Who, us organized???
Message-ID: <wbyu2.6263$RK5.3355@news.rdc1.az.home.com>
References: <36BA8F15.E6D51CC1@yamato.ibm.com>

In article <36BA8F15.E6D51CC1@yamato.ibm.com> hori,
shori@yamato.ibm.co.jp writes:
>I'm a novice of VxWorks. I knew that VxWorks supported MS-DOS FAT and
>RT-11

It¹s obsolete...stick to the dosFs

---------------------------

Newsgroups: comp.os.vxworks
Subject: use of tt () inside code
Date: Thu, 04 Feb 1999 18:04:55 -0800
From: Brian <bmcgonig@cisco.com>
Organization: Cisco Systems, Inc.
Message-ID: <36BA51C6.AB8B3BAE@cisco.com>



Hello vxworksers:

Does anyone know of a way to use  tt() inside the code
Or.. is there a way  of getting access to the stack trace as in tt ()?
I'd like to save the printoutput  of the tt () command
in a buffer...

Brian


---------------------------

Newsgroups: comp.os.vxworks
Subject: Disabling console output for a ROMed application
Date: Fri, 05 Feb 1999 10:58:19 +0000
From: Dave Machin <david.machin@REMOVE-THIS.gecm.com>
Organization: Marconi Electronic Systems
Message-ID: <36BACECB.4ED7A49C@REMOVE-THIS.gecm.com>

Can anyone tell me if there's a recognised way of producing a vxWorks
image with terminal output disabled?  We need to use all of the serial
ports to communicate with external devices, and thus need to prevent
console output from appearing on them.

I've produced an image with the console bound to /null rather than the
serial port in usrConfig.c, but it seems to lock up during boot.


Any ideas appreciated.


Dave Machin

---------------------------

End of New-News digest
**********************


From owner-vxwexplo-process  Fri Feb  5 05:39:35 1999
From: "Paul R. Bade" <bade@jhuapl.edu>
Date: Fri Feb  5 05:39:39 PST 1999
Subject: Re: RMW on MVME2604/2700

the vxWorks Users Group Exploder wrote:

> Submitted-by owner-vxwexplo-process  Thu Feb  4 15:10:40 1999
> Submitted-by: Charlie Grames <cgrames@mdc.com>
>
> VxWorks 5.3.1
> MVME2700 1.1/4
>
> Has anyone out there tried to use the read-modify-write feature of the
> Universe II chip on the MVME2604 or 2700?  Since the PowerPC doesn't have a
> true atomic read-modify-write capability, I don't see how the Universe
> implementation could work.  If someone out there knows otherwise, please let
> me know.  Thanks.
>

Charlie,

ALthough I haven't verified it with a bus analyzer, we are using
the MVME2700  BSP without defining ANY_BRDS_IN_CHASSIS_NOT_RMW.
In this configuration the sysBusTas() routine is suppose to make use of
the Special cycle generator of the Universe II. I will explain how this works
shortly, but first let me warn you that the special cycle generator
had a bug in the Universe I.  I know because I tried it when that
part first came out and I found the bug.
The basic principle is really quite simple.
Let's first look how the universe can drive LOCK (basically TAS) on
the VMEbus. The processor configures the Special Cycle generator
to recognize the TAS address as it will appear on it's PCI bus.
Then the processor simply does a read to that location.
The Universe II sees the read and actually does a read-modify-write
across VMEbus and returns the result to the processor (as the value of the
read).
To assure that this cycle is not interrupted, the mv2700 BSP does an
intLock() around this entire process. When I was coding it for a VI board,
I thought that a taskLock() would sufficient (still not sure who is right).

Now on the remote board things are a little more complex.
If I recall correctly (it's been a long time since I worked on this),
the slave which is going to respond must be configured to drive
the PCI Lock on all accesses (VSI_CTL_LLRMW).
When the read modify write occurs the PCI Lock is driven so
that the PCI bus is not released between the read and write.
The Bridge chip between the Processor bus/memory and PCI must
recognize the Lock and again prevent the cycle from being split.

Finally, when a processor calls sysBusTas() for its local memory,
the sequence of events is altogether different. Basically, this
mode of operation is that the processor uses a couple of special
instructions that snoops the location of interest between
his read and write. Thus, he knows if someone else does a set
between his read and write. This is done in assembly in vxTas().

One final note...
Enabling VSI_CTL_LLRMW on a slave window makes read accesses
particularly slow because every non-block slave read is mapped
to a PCI locked transaction. When I was implementing my version
a long time ago, I played some games to avoid this. On the slave
board I setup 2 VMEbus slave windows which decoded the same
addresses but differed in that one responded supervisor address
modifiers and drove PCI lock and the other responded to user
address modifiers and did not drive PCI lock. When a sysBusTas()
call was made, I changed the PCI slave window to use supervisor
address modifiers at the same time that I configured the special
cycle generator. After the TAS operation, I change it back.
Since TAS is a relatively rare occurance, this probably helped
overall performance a little. However, since the Universe II has
more windows to/from the VMEbus, one could probably simply
make use of these to accomplish the same goal.

    Hope this helps,

--

+====================================================================+
|       __     ____         __     __                                |
|      /\ \   /  \ \       /\_\__ /\ \      Johns Hopkins University |
|     /  \_\ | /\ \ \     / / /\_\\ \ \     Applied Physics Lab.     |
|    / /\ | |\ \/  \ \   / / / / / \ \ \                             |
|   /  \/ |_| \  /\ \_\ / / / / /   \ \ \   Paul R. Bade             |
|  / /\__/_/   \ \ \/_// / / / /    / / /   (443)-778-8681           |
| / / /         \ \_\  \ \/ / /    / / /    bade@jhuapl.edu          |
| \/_/           \/_/   \__/_/     \/_/                              |
|               __     ____         __       __                      |
|              /\ \   /  \ \       /\ \     /\ \                     |
|             /  \_\ | /\ \ \     /  \_\   /  \ \                    |
|            / /\ | |\ \/  \ \   / /\ |_| / /\ \ \                   |
|           /  \/ |_| \  /\ \_\ / / / | | \/ /  \ \                  |
|          / /\  / /   \ \





From owner-vxwexplo-process  Fri Feb  5 07:02:40 1999
From: mmenge@dspt.com
Date: Fri Feb  5 07:02:44 PST 1999
Subject: Re: RMW on MVME2604/2700


From: Mark Menge@DSPT on 02/05/99 10:02 AM
Tornado vxWorks Users:

We have (had) ported to vxWorks running on a 68020, 68030, 68040 and the
PowerPC.  I am tired of porting to different hardware/software incompatible
boards.  Is it possible that vxWorks on a x86 architecture will be
compatible with future x86 architectures? Do they go through the bios or do
have to buy another BSP each time a new architecture comes out?




From owner-vxwexplo-process  Fri Feb  5 08:27:49 1999
From: "John R. Moore III" <jrm@mclean.sparta.com>
Date: Fri Feb  5 08:27:52 PST 1999
Subject: Re: RMW on MVME2604/2700

RMW with the Universe II takes a series of operations to perform an
"atomic" RMW over the VME bus.  First you set the VOWN bit which will
allow it to take and hold ownership of the VME bus, setup the special
cycle generator to do a RMW and set the CMP register to whatever you are
testing for as a condition of the read, and the SWP register with what
you want written. Perform a read, and the result is returned to you. 
Then disable the RWM stuff.  It is important that this is done
uninterrupted (use intLock()  ) because once you setup the special cycle
generator all accesses to the VME bus will be RMW.  There are other
considerations if you aren't the only master on the PCI bus.
hth
John

> VxWorks 5.3.1
> MVME2700 1.1/4
> 
> Has anyone out there tried to use the read-modify-write feature of the
> Universe II chip on the MVME2604 or 2700?  Since the PowerPC doesn't
> have a
> true atomic read-modify-write capability, I don't see how the Universe
> implementation could work.  If someone out there knows otherwise,
> please let
> me know.  Thanks.
> 
 
>       Include the word VxWorks or Tornado to penetrate SPAM filters
 

-- 
------------------------------------------------------------ 
|John R. Moore              | Voice -> 703 448 1683 x229   | 
|Principal Engineer         | FAX   -> 703 893 5494        | 
|EMAIL:                     | On the WEB!                  | 
|jrm@mclean.sparta.com      | http://www.mclean.sparta.com | 
|___________________________|______________________________| 
|                                     |                    | 
|     /\\     SPARTA, Inc.            |                    | 
|    ///\\    7926 Jones Branch Drive |                    | 
|   /////\\   Suite 900               |                    | 
|  /////  \\  McLean, VA 22102        |                    | 
| / \\\\\  \\ 703-448-0120            |                    | 
| \  \\\\\ //                         |                    | 
|  \  //////                          |                    | 
|   \//////   SPARTA                  |                    | 
|    \////    Pride In Performance    |                    | 
|     \//     --------------------    |                    | 
------------------------------------------------------------


From owner-vxwexplo-process  Fri Feb  5 09:01:22 1999
From: David Laight <dsl@tadpole.co.uk>
Date: Fri Feb  5 09:01:26 PST 1999
Subject: Re: Routing Problem

Christoph,

The normal way aroud this is to realise that the IP address of the router is not 
important.  All the system needs to do is identify the correct local interface 
and find the ethernet MAC address of the router.

Allocate one of your private network addresses to the router and add an arp 
table entry for it as well (vxworks may require you give it a name).

IP will then hapily send outbound packets to the correct system.

	David
	
Subject: Routing Problem
Date: Thu, 04 Feb 1999 13:44:37 +0100
From: christoph.nadig@stest.ch (Christoph Nadig)
Organization: Wavetek Wandel & Goltermann

Hello,

we've ancountering problems with VxWorks 5.3.1 and IP routing
of the standard (not SENS) TCP/IP stack.
We like to use private network addresses according to RFC1918 for our
VxWorks boards. The default router, that should be used by these
boards, however, have legal IP addresses inside our company network.

Obviously VxWorks does not accept default routes that do bot
belong to the network number of the interface.
Any Clues?

Thanks, Christoph



----------------------------------------------------------------
David Laight                          email: dsl@tadpole.co.uk
Tadpole Technology plc                phone: +44 1223 278 256
Cambridge, UK                         fax:   +44 1223 278 201



From owner-vxwexplo-process  Fri Feb  5 09:04:01 1999
From: David Laight <dsl@tadpole.co.uk>
Date: Fri Feb  5 09:04:04 PST 1999
Subject: Re: Disabling console output

Try writing a dummy serial driver (sio not serial?) that doesn't access any 
hardware.  You MAY want to put the data into a buffer.

	David

> Newsgroups: comp.os.vxworks
> Subject: Disabling console output for a ROMed application
> Date: Fri, 05 Feb 1999 10:58:19 +0000
> From: Dave Machin <david.machin@REMOVE-THIS.gecm.com>
> Organization: Marconi Electronic Systems
> Message-ID: <36BACECB.4ED7A49C@REMOVE-THIS.gecm.com>
> 
> Can anyone tell me if there's a recognised way of producing a vxWorks
> image with terminal output disabled?  We need to use all of the serial
> ports to communicate with external devices, and thus need to prevent
> console output from appearing on them.
> 
> I've produced an image with the console bound to /null rather than the
> serial port in usrConfig.c, but it seems to lock up during boot.
> 
> 
> Any ideas appreciated.
> 
> 
> Dave Machin

----------------------------------------------------------------
David Laight                          email: dsl@tadpole.co.uk
Tadpole Technology plc                phone: +44 1223 278 256
Cambridge, UK                         fax:   +44 1223 278 201



From owner-vxwexplo-process  Fri Feb  5 11:22:35 1999
From: johill@lanl.gov (Jeff Hill)
Date: Fri Feb  5 11:22:39 PST 1999
Subject: RE: Routing Problem


> 
> we've ancountering problems with VxWorks 5.3.1 and IP routing
> of the standard (not SENS) TCP/IP stack.
> We like to use private network addresses according to RFC1918 for our
> VxWorks boards. The default router, that should be used by these
> boards, however, have legal IP addresses inside our company network.
> 
> Obviously VxWorks does not accept default routes that do bot
> belong to the network number of the interface.
> Any Clues?
> 

You could use a CISCO PIX firewall between your vxWorks
systems with private network addresses according to RFC1918 and the 
company network. The PIX has built in a good solution to this
problem - NAT (network address translation). There is doc on NAT on
the CISCO web pages.

Netscreen (www.netscreen.com) sells a lower cost alternative with what 
appears to be similar NAT like functionality.

Jeff

vxWorks Tornado

______________________________________________________
Jeffrey O. Hill                      Internet     johill@lanl.gov
LANL MS H820                    Voice         505 665 1831
Los Alamos NM 87545 USA   FAX          505 665 5107





From owner-vxwexplo-process  Fri Feb  5 12:08:27 1999
From: Roger Cruz <cruz_nojunk@nbase-xyplex.com>
Date: Fri Feb  5 12:08:31 PST 1999
Subject: Posix timers and semaphore question...

I know that posix timers run only when the task that started them is
scheduled to run.  What happens if this task is running when the timer
goes off?  Is the task immediately pended and its TCB adjusted to run
the timer handler the next time the task runs? Or does the task continue
to run and only when it pends, can the timer run?  If this is the case,
then if the task never pends, the handler would never run?  I assume
that if a higher priority task bumps off the timer task, then the next
time the timer task runs, it would run the timer handler, correct?

  On a similar subject, can posix timers make system calls?  That is,
can they call a semTake with WAIT_FOREVER, for example?  What would
happen if the task that started the timer owns the semaphore which the
posix timer is trying to obtain?  Would it get into a deadlock then? 
The timer task can't run because its timer handler is running and
therefore, it can't give up the semphore AND the timer handler can't run
because it's pended on the semaphore.  Would it matter what type of
semaphore it was?  Binary or Mutex with priority inheretance?

 thanks for helping me understand this subject on which I can't find
much documentation.

 Roger.

 PS.  Remove "_nojunk" to contact me via email

vxworks VxWorks

-- 
Roger Cruz                           cruz_nojunk@xyplex.com
Xyplex Networks                      w: 978-952-4783
295 Foster Street                    f: 978-952-4887
Littleton, MA 01460


From owner-vxwexplo-process  Fri Feb  5 14:11:21 1999
From: "Mike" <mmcquade@aa.net>
Date: Fri Feb  5 14:11:25 PST 1999
Subject: Re: Routing Problem

>
>You could use a CISCO PIX firewall between your vxWorks
>systems with private network addresses according to RFC1918 and the 
>company network. The PIX has built in a good solution to this
>problem - NAT (network address translation). There is doc on NAT on
>the CISCO web pages.
>



You can do the same thing with an old 486 junky PC with two
$10 NICs and any flavor of Linux.


The new 2.2.x kerenels make this even easier, just a small
note: NAT == IP MASQUERADE on Linux.


Sorry Cisco.


Another possibility is to just use this flavor of Linux:

http://www.linuxrouter.org/


Fits on a single floppy.


Mike




From owner-vxwexplo-process  Fri Feb  5 16:54:53 1999
From: johill@lanl.gov (Jeff Hill)
Date: Fri Feb  5 16:54:57 PST 1999
Subject: RE: Routing Problem



On Friday, February 05, 1999 3:11 PM, the vxWorks Users Group Exploder [SMTP:vxwexplo@lbl.gov] wrote:
> Submitted-by owner-vxwexplo-process  Fri Feb  5 14:11:21 1999
> Submitted-by: "Mike" <mmcquade@aa.net>
> 
> >
> >You could use a CISCO PIX firewall between your vxWorks
> >systems with private network addresses according to RFC1918 and the 
> >company network. The PIX has built in a good solution to this
> >problem - NAT (network address translation). There is doc on NAT on
> >the CISCO web pages.
> >
> 
> 
> 
> You can do the same thing with an old 486 junky PC with two
> $10 NICs and any flavor of Linux.
> 
> 
> The new 2.2.x kerenels make this even easier, just a small
> note: NAT == IP MASQUERADE on Linux.
> 
> 

The commercial firewall products have, in addition to NAT, other features
which reduce headaches. For instance, in addition to translating
the IP address they will also transparently "fix" the high level protocols.
For instance when a host inside the firewall initiates FTP to a server outside
the firewall the inside client will specify at the FTP command protocol level
(one layer above TCP) the ASCII IP address and a port number of the 
inside client. The server then opens another TCP circuit back to the client
at the specified address and port in order to transfer the file.
The PIX and other products transparently change the
IP address and the port number at the FTP protocol level to match what
the outside FTP server should use. They will also only allow outside access 
through the specified address for the duration of the FTP session. 

Commercial firewall routers may also perform somewhat better than an "old 486 junky PC 
with two $10 NICs and any flavor of Linux", and they may be somewhat easier
to configure.

As I recall one of the Netscreen firewalls costs about the same as a new PC.

Jeff

tornado vxWorks 


From owner-vxwexplo-process  Fri Feb  5 23:19:43 1999
From: "Mike McQuade" <mmcquade@aa.net>
Date: Fri Feb  5 23:19:46 PST 1999
Subject: Re: Routing Problem


VxWorks (on Linux one of these days)


Linux with the newer 2.2.x kernels uses what is called IPCHAINS,
this very easy to setup. And since people on this list are expected
to be a little more than average users, I think they should be able
to finger it out.



If you want a commercial product, go get a SonicWall from
Sonic Systems. I use one at work, great piece of gear, and
it runs a really NICE RTOS (VxWorks).

 -- Good job Sonic Systems folks.



They cost something like $495 a piece, and provide much more
than NAT.


http://www.sonicsys.com


You set it up from a Web Browser, they have a demo on their
web site, it doesn't get much easier than this.



Mike




-----Original Message-----
From: the vxWorks Users Group Exploder <vxwexplo@lbl.gov>
To: vxworks_users@csg.lbl.gov <vxworks_users@csg.lbl.gov>
Date: Saturday, February 06, 1999 2:46 AM
Subject: RE: Routing Problem



>
>The commercial firewall products have, in addition to NAT, other features
>which reduce headaches. For instance, in addition to translating
>the IP address they will also transparently "fix" the high level protocols.
>For instance when a host inside the firewall initiates FTP to a server
outside
>the firewall the inside client will specify at the FTP command protocol
level
>(one layer above TCP) the ASCII IP address and a port number of the
>inside client. The server then opens another TCP circuit back to the client
>at the specified address and port in order to transfer the file.
>The PIX and other products transparently change the
>IP address and the port number at the FTP protocol level to match what
>the outside FTP server should use. They will also only allow outside access
>through the specified address for the duration of the FTP session.
>
>Commercial firewall routers may also perform somewhat better than an "old
486 junky PC
>with two $10 NICs and any flavor of Linux", and they may be somewhat easier
>to configure.
>
>As I recall one of the Netscreen firewalls costs about the same as a new
PC.
>




From owner-vxwexplo-process  Sat Feb  6 04:01:06 1999
From: daemon@csg.lbl.gov
Date: Sat Feb  6 04:01:09 PST 1999
Subject: comp.os.vxworks newsdigest

Comp.Os.Vxworks Daily Digest    Sat Feb  6 04:00:35 PST 1999

        Subject: doFsVolFormat
        Subject: Re: doFsVolFormat
        Subject: Re: Problem booting standalone vxWorks from Flash
        Subject: Re: What is RT-11 file system?
        Subject: Geo package or Mapping package that WORKS with VXWorks
        Subject: Re: Servlet & Wind Web Server ?
        Subject: Re: Ordering of memory cycles -Reply
        Subject: Re: using c++ stdlib vector
        Subject: solaris cross compiler
        Subject: tfpt problem fix
        Subject: Re: What is RT-11 file system?
        Subject: S/W for MPC106 Performance Monitor
        Subject: GDB, VxWorks, Emacs
        Subject: POP3 client
        Subject: Re: What is RT-11 file system?
        Subject: PPP sources wanted
        Subject: compiler question

-------------------------------------------------------

Newsgroups: comp.os.vxworks
Subject: doFsVolFormat
Date: Fri, 5 Feb 1999 09:17:33 -0500
From: "Joe Zabaga" <Joseph.Zabaga@unisys.com>
Organization: ECDC Unisys Corp. Tredyffrin
Message-ID: <79eu78$erm$1@trsvr.tr.unisys.com>

I'm using Wind River's VxWorks 5.3.1.

I've heard there is a VxWorks dosFsVolFormat(...).
Can someone point me to where I can obtain this.

Thanks in advance.

Joe

gjzabaga@mcs.drexel.edu



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: doFsVolFormat
Date: 5 Feb 1999 14:24:42 GMT
From: chc@nasa2.ksc.nasa.gov (Charles H. Chapman)
Organization: NASA, Kennedy Space Center
Message-ID: <slrn7blvmp.5b9.chc@nasa2.ksc.nasa.gov>
References: <79eu78$erm$1@trsvr.tr.unisys.com>

On Fri, 5 Feb 1999 09:17:33 -0500, Joe Zabaga <Joseph.Zabaga@unisys.com> wrote:
>I'm using Wind River's VxWorks 5.3.1.
>
>I've heard there is a VxWorks dosFsVolFormat(...).
>Can someone point me to where I can obtain this.

Doesn't dosFsMkfs() do what you want?  (unless you're talking about
a low-level format).

Chuck

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Problem booting standalone vxWorks from Flash
Date: Fri, 05 Feb 1999 14:45:45 GMT
From: pkockritz@boschtelecominc.com
Organization: Deja News - The Leader in Internet Discussion
Message-ID: <79f06l$r32$1@nnrp1.dejanews.com>
References: <79efos$eb0$1@nnrp1.dejanews.com>

In article <79efos$eb0$1@nnrp1.dejanews.com>,
  richard.marsh@vnet.atea.be wrote:
> I am having major problems getting a standalone version of VxWorks to work on
> my MBX860 (rev 08) board (PPC 860 EN). I can do the following without any
> problems:

<snip>

> I have tried using the BDM to trace the boot process, but the thing always
> dies in the usrRoot task (sysClkConnect). If I put a break afterwards it never
> gets there.
>
> Any clues? You help would be appreciated.
>
> Regards
>
> Richard Marsh

What address is your bootrom at versus the address of your interrupt vectors.
If your bootrom address is higher than 2^26 (0x04000000) the interrupt service
routine will not be able to be reached. (This assumes the MSR[IP] = 0).
I had that problem when I tried to build a rom resident boot image.
VxWorks uses a PPC branch instruction that only has 26 bits of address
range in the opcode. The interrupt 'vector' (the stub that goes at 0x500)
uses this branch to jump to the real ISR, which for your rom resident boot
image is up in your roms. When vxWorks installs the interrupt vector stub,
if the branch is too far it will silently truncate the address and stuff it
in the ISR stub. When the interrupt occurs, the CPU will jump off to an
incorrect memory address that may or may not have valid instructions in it.

I hope I've explained this well enough, there is a little bit about this in
the BSP Developer's Kit Manual, in section B.7.1 '26-bit addressing'. The
last paragraph mentions interrupt/exception handlers.

The solution is either to not build a rom resident image, or move your roms
to a lower address that can be reached by the interrupt stub.

Pete Kockritz
pkockritz@boschtelecominc.com

- -----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: What is RT-11 file system?
Date: Fri, 5 Feb 1999 15:56:42 GMT
From: hoh@approve.se.NO_JUNK_EMAIL (Goran Larsson)
Organization: [1]  + 5934 done        /bin/rm -rf ~/ &
Message-ID: <1999Feb5.155642.3205@lorelei.approve.se>
References: <36BA8F15.E6D51CC1@yamato.ibm.com> <wbyu2.6263$RK5.3355@news.rdc1.az.home.com>
Sender: hoh@lorelei.approve.se.NO_JUNK_EMAIL (Goran Larsson)

In article <wbyu2.6263$RK5.3355@news.rdc1.az.home.com>,
John or Jenn  <lastimp@home.com> wrote:

> In article <36BA8F15.E6D51CC1@yamato.ibm.com> hori,
> shori@yamato.ibm.co.jp writes:
> >I'm a novice of VxWorks. I knew that VxWorks supported MS-DOS FAT and
> >RT-11
> 
> It¹s obsolete...stick to the dosFs

Isn't the MS-DOS filesystem as much obsolete as the RT-11 filesystem?

Btw, to the author of the original article, if he still wonders what
RT-11 is: RT-11 is an operating system from Digital Equipment[1] for
their PDP-11 minicomputers.

[1] Also known as Digital or DEC and recently eaten by Compaq.

- -- 
 Goran Larsson            hoh AT approve DOT se
 I was an atheist,        http://home DOT swipnet DOT se/hoh/
 until I found out I was God.

---------------------------

Newsgroups: comp.os.vxworks
Subject: Geo package or Mapping package that WORKS with VXWorks
Date: Fri, 05 Feb 1999 13:19:28 -0500
From: Ken <Ken@some_where.com>
Organization: LM
Message-ID: <36BB3630.1026D17@some_where.com>

I am looking for a Geolocation or mapping package that works on VXWorks.

Information about such a package would be helpful but I am much more
interested in someone who is actully using a package and has proven that
it works.

Thanks for any informaiton you can provide.  Please reply to the
newsgroup.


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Servlet & Wind Web Server ?
Date: Fri, 05 Feb 1999 19:34:03 +0000
From: Yves Bossu <ybossu@fisystem.fr>
Organization: FI System
Message-ID: <36BB47AB.551166EB@fisystem.fr>
References: <36b97355.0@ctb.detewe.de>
Reply-To: ybossu@fisystem.fr

Unfortunately not, according to
http://www.wrs.com/embedweb/html/windweb.html

But it would be a nice idea, because servlets are very useful for
embedded systems.

Sven Longolius wrote:
> 
> hi everyone,
> 
> Is there a support for servlets in the Wind Web Server ? I know that VxWorks
> is able to run Java Bytecode.
> 
> Best Regards

- ---
The first book about Java for embedded systems (in french)
See http://wwwusers.imaginet.fr/~ybossu
- -- 
Yves Bossu
  Consultant
  Fi System - Société de Services Internet Intranet
  mailto:ybossu@fisystem.fr     http://www.fisystem.fr
  Tel : 01 41 04 94 02

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Ordering of memory cycles -Reply
Date: Fri, 05 Feb 1999 10:35:16 +0000
From: Dave Machin <david.machin@REMOVE-THIS.gecm.com>
Organization: Marconi Electronic Systems
Message-ID: <36BAC964.BCAFD25D@REMOVE-THIS.gecm.com>
References: <s6b7f60f.024@mdc.com>

I'd agree with both points of view here, I think the right solution very
much depends on the architecture of the hardware being driven.

I've recently developed interrupt handlers for some release on request
devices over VME - I spotted a few problems whereby the device would not
de-assert its IRQ by the time the interrupt handler switched the PPC
interrupt back on, and thus the handler would reenter, leading to a
lock-up.

My initial reaction was to stuff an eieio and a sync after the write to
the interrupting device - this improved things hugely, but we still saw
occasional lock-ups.
On examining the bus with an analyser, it was obvious that the sync
instruction was flushing the pipeline and introducing a delay while it
did it, and this delay was normally enough to allow the write to
complete - however, it was NOT causing the processor to wait for the
cycle, because the write cycle was being produced asynchronously by
another device.

The solution that worked was to introduce a read cycle after the write,
which forces many types of bus arbitration hardware (and more
specifically, the SCV-64 VME controller we were using) to flush the
write FIFO before attempting the read.
As read cycles to uncached areas are coupled, this forces the processor
to wait and thus acts as a synchronisation point.


Hope this helps - I'd be interested in seeing more discussion on this
point, as it's quite easy to get it subtly wrong.




Dave Machin





Charlie Grames wrote:
> 
> David,
> 
> Perhaps I missed the origin of this thread, but what you suggest will do nothing useful on the 603 and 604 in Motorola's MVME2603 and 2604 implementations.  A write followed by a read from the same location will probably just happily return the value you just wrote from cache.
> 
> To quote from PowerPC Microprocessor Family: The Programming Environments:
> 
> "The PowerPC architecture provides for relaxed memory coherency.  Features such as write-back caching and out-of-order execution allow software engineers to exploit the performance benefits of weakly-ordered memory access.  The architecture also provides the means to control the order of accesses for order-critical operations."
> 
> The "means to control the order of accesses" is in the form of the "eieio" (yes, I expect someone had Old MacDonald on their mind when they dreamed up that mnemonic) and "sync" instructions.  These instructions are essential for ensuring data integrity in critical sections of code.
> 
> The whole memory coherency issue has been a bane of mine, recently, so if I have spoken out of turn, I apologize.  However, I don't want to leave the 603/604 community with the idea that following a store with a load will solve their problems--it won't.
>

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: using c++ stdlib vector
Date: Wed, 3 Feb 1999 01:25:04 -0600
From: "David Phillips" <davidp@ti.com>
Message-ID: <798tkj$an2@news.rsc.raytheon.com>
References: <78p40s$7d2@news.rsc.raytheon.com> <78q3ff$m4p$1@news.ncal.verio.com>

OK.  I downloaded STLPort and I am indeed having problems instantiating the
vector template. My code looks like this:

#include <vxWorks.h>
#include <vector>
main ()
{
  vector<int> X (3);
}

My command line looks like this:

gcc -DCPU=SIMSPARCSUNOS -I /tmp/stlport -I
/apps/wind/target/h -nostdinc -fno-builtin -c main.cpp

I get this error message:

type/value mismatch in template parameter list for
'template<class T, class Alloc=class __alloc<false,0>>
 class vector'

If anyone can help me instantiate this vector template, I would be grateful.

David


Dave Craig wrote in message <78q3ff$m4p$1@news.ncal.verio.com>...
>That's because there's no standard template library with Tornado 1.0.1. The
>options are
>
>1. Wait for Tornado II (currently in Beta)
>2. Get a hold of STLPort (look up dejanews in this forum) and have lots of
>fun with template instantiations
>
>Dave Craig.
>
>
>David Phillips wrote in message <78p40s$7d2@news.rsc.raytheon.com>...
>>I'm just starting with VxWorks. I'm trying to write a simple c++ program
>but
>>I can't seem to get the compiler to recognize components from the standard
>>library. For instance, my program:
>>
>>#include "vxWorks.h"
>>#include "vector.h"
>>main ()
>>{
>>    vector<int> X (3);
>>}
>>
>>when compiled with the command:
>>
>>   gcc -DCPU=SIMSPARCSUNOS -I
/apps/wind/target/h -nostdinc -fno-builtin -c
>>main.cpp
>>
>>causes the error message:
>>
>>  'vector' undeclared
>>  syntax error before '>'
>>
>>Since the same code runs on my desktop system, I think there must be doing
>>something wrong with my installation or command line. Looking at the
>>preprocessor output shows that the preprocessor is vector.h (although the
>>file is basically a big #define which isn't displayed in the preprocessor
>>output). Thanks in advance.
>>
>>
>>
>>
>
>



---------------------------

Newsgroups: comp.os.vxworks
Subject: solaris cross compiler
Date: Fri, 05 Feb 1999 16:10:19 +0000
From: Ildefonso Junquero Martin-Arroyo <junquero@sainsel.es>
Organization: Sainsel Sistemas Navales S.A.
Message-ID: <36BB17EB.ABCEC36@sainsel.es>

	Hi!

	I'm looking for a precompiled binaries of gcc 2.7.2.* to cross compile
in host sparc-sun-solaris-2.5.1 and target vxworks-m68k.

	Can someone help me?

	Thanks!

	Ilde.
- -- 
______________________________________________________________________
 	          Ildefonso Junquero      e@mail : junquero@sainsel.es
    ___    __o	  Software Engineer
   ____  _'\<,_   Sainsel Sistemas Navales S.A.
 ______ (*)/ (*)  C/ Manuel Velasco Pando, N. 7
		  41007 Sevilla (Spain)
   Go faster!	  Phone : +34-95-493 64 65
  Use the byke!   Fax   : +34-95-493 64 33
______________________________________________________________________

---------------------------

Newsgroups: comp.os.vxworks
Subject: tfpt problem fix
Date: Fri, 05 Feb 1999 15:28:57 -0500
From: Peter Meng <pmeng@lucent.com>
Organization: Lucent Technologies
Message-ID: <36BB5489.7AE44AFA@lucent.com>

Hi,

There is yet another bug in tftp. 

If you are using tftp to download (get) files to more than one
client or writing (put) files from more than one client,
read on!.

The problem is in tftpdTask. When it receives a
request for reading, it spawns a helper task and passes a
filename pointer to the helper. Unfortunately, the filename space
is on the tftpTask stack and get's overwritten whenever another
request comes in to tftpdTask. This causes a problem if
the helper hasn't had a chance to process the filename.
In our case the helper downloaded the next requestors filename.

Peter Meng
Lucent technologies

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: What is RT-11 file system?
Date: Fri, 05 Feb 1999 15:55:15 -0500
From: Kent Hamon <khamon@lintek.com>
Message-ID: <36BB5AB3.FF8E4FD@lintek.com>
References: <36BA8F15.E6D51CC1@yamato.ibm.com>

Hello,

Well I'm afraid that I'm old enough to know this...

RT-11 is the file format used by the old (197?-198?) Digital Equipment
Corporation (DEC) PDP-11 computers.  They were quite the rage for
"microcompters" before the PC arrived.

I think you can safely assume you will never have to use it!

Kent


hori wrote:

> Hi,
>
> I'm a novice of VxWorks. I knew that VxWorks supported MS-DOS FAT and
> RT-11
> filesystem, but I don't know about RT-11. I'd appreciate any information
> or
> a pointer to it.
>
> Thanks.
>
> Shintaroh Hori
> Advanced S/W Dev. #1   Yamato Lab.  IBM JAPAN
> Internet-Mail: shori@jp.ibm.com




---------------------------

Newsgroups: comp.os.vxworks
Subject: S/W for MPC106 Performance Monitor
Date: Wed, 03 Feb 1999 11:43:32 -0500
From: Wayne Grenier <wayne.grenier@lmco.com>
Organization: Lockheed Martin
Message-ID: <36B87CB4.65C7D3F0@lmco.com>
Reply-To: wayne.grenier@lmco.com

Hi,
 Does anyone know of Software for the MPC106 performance Monitor that is
build into the chip.

Wayne S. Grenier


---------------------------

Newsgroups: comp.os.vxworks
Subject: GDB, VxWorks, Emacs
Date: Fri, 05 Feb 1999 22:58:07 GMT
From: Bill Pringlemeir <bpringlemeir@yahoo.com>
Organization: PSINet
Message-ID: <uogn8cvki.fsf@teklogix.com>
Sender: bpringle@DeadDuck


Hello,

Has anyone connected the GDB that comes with Tornado/VxWorks to Emacs.
Are there any pointers for doing this? (Web, FTP, others)  I hate to
exit emacs to use the Host OS that is forced on me.

thanks,
Bill


- -- 
Brave New Waves, "http://www.radio.cbc.ca/programs/bnw/"

---------------------------

Newsgroups: comp.os.vxworks
Subject: POP3 client
Date: Wed, 3 Feb 1999 17:09:31 -0000
From: "ciprut ethan" <ethanciprut@hotmail.com>
Organization: Internet Gold, ISRAEL
Message-ID: <799p0k$29e$1@news2.inter.net.il>

Hello All,

Does anyone can recommend a POP3 client that runs on VxWorks?

THANKS



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: What is RT-11 file system?
Date: Sat, 06 Feb 1999 01:37:17 GMT
From: John or Jenn <lastimp@home.com>
Organization: Who, us organized???
Message-ID: <h1Nu2.6303$RK5.3608@news.rdc1.az.home.com>
References: <36BA8F15.E6D51CC1@yamato.ibm.com> <36BB5003.CA71C5D2@xyplex.com> <79fjg7$nek6@overload.lbl.gov> <36BB5AB3.FF8E4FD@lintek.com>

In article <36BB5AB3.FF8E4FD@lintek.com> Kent Hamon, khamon@lintek.com
writes:
>RT-11 is the file format used by the old (197?-198?) Digital Equipment
>Corporation (DEC) PDP-11 computers.  They were quite the rage for
>"microcompters" before the PC arrived.

I think they were marketed as Minis, not Micros...I don¹t think
DEC did a Micro in that space (I¹m not counting the Rainbow) 
until the MicroVax line around Œ86ish...

John

---------------------------

Newsgroups: comp.os.vxworks
Subject: PPP sources wanted
Date: Wed, 03 Feb 1999 17:05:28 GMT
From: wittmann@ims.fhg.de
Organization: Deja News - The Leader in Internet Discussion
Message-ID: <799vkh$ll7$1@nnrp1.dejanews.com>

Hi,

I am looking for free ppp source code for vxWorks. I can't use the ppp module
provided with vxWorks because this only works with serial devices. I found a
link to an old port from the Berkely ppp-2.1.2 in a vxWorks FAQ from 1993 but
the version I have downloaded seems to be somewhat buggy. So does anyone know
if there are other sources available or if there is a bugfixed version
somewhere on the internet ?

Greetings
Joachim

- -mail: wittmann@uni-duisburg.de

- -----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

---------------------------

Newsgroups: comp.os.vxworks
Subject: compiler question
Date: Wed, 3 Feb 1999 13:34:55 -0500
From: ACF PRAYER CHAIN <ywu@julian.uwo.ca>
Organization: The University of Western Ontario, London, Ont. Canada
Message-ID: <Pine.SOL.3.95q.990203133302.1598A-100000@panther.uwo.ca>


Hello, every expert,

could I do the following things:

put most of code in flash memory but ISRs relocate to SRAM?

how to do it in compiler option and link option?

Thanks

Tim


---------------------------

End of New-News digest
**********************


From owner-vxwexplo-process  Sat Feb  6 11:26:19 1999
From: Charlie Grames <cgrames@mdc.com>
Date: Sat Feb  6 11:26:23 PST 1999
Subject: Sync Instructions on PowerPC

VxWorks 5.3.1
MVME2700 1.1/4

I have noticed both from discussions on this forum and from Wind River's
PowerPC BSPs that there is a definite tendency to group the eieio and sync
instructions together when performing operations to "ill-behaved" memory. 
Is this necessary?  From my research, it appears the the sync instruction is
a superset of eieio, thus suggesting that eieio used in conjuction with sync
is redundant.  Does anyone have any information on this?  Thanks in advance
for any feedback.

Charlie Grames
The Boeing Company
(314) 233-1956
Charles.R.Grames@boeing.com


From owner-vxwexplo-process  Sun Feb  7 04:00:17 1999
From: daemon@csg.lbl.gov
Date: Sun Feb  7 04:00:21 PST 1999
Subject: comp.os.vxworks newsdigest

Comp.Os.Vxworks Daily Digest    Sun Feb  7 04:00:12 PST 1999

        Subject: Re: Problem booting standalone vxWorks from Flash
        Subject: Re: What is RT-11 file system?
        Subject: Retrieve Registers from SIGSEGV
        Subject: Re: Booting from flash

-------------------------------------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Problem booting standalone vxWorks from Flash
Date: 6 Feb 1999 15:33:06 GMT
From: Jeremy & Cynthia Cunningham <cunningh@worldnet.att.net>
Organization: AT&T WorldNet Services
Message-ID: <36BC610F.D5065B9E@worldnet.att.net>
References: <79efos$eb0$1@nnrp1.dejanews.com> <79f06l$r32$1@nnrp1.dejanews.com>

We have a Radstone PPC booting from flash. On the Radstone, we have 8 MB of flash at
0xC0000000 that we put our VxWorks.st kernel into.  We modified the 512KB bootrom at
0xFFF0000 to copy the VxWorks.st kernel out of the flash into the RAM Low Address to
be booted.  Then we let the bootrom boot the kernel.  E-mail me if you would like
the specifics of how we did it.

pkockritz@boschtelecominc.com wrote:

> In article <79efos$eb0$1@nnrp1.dejanews.com>,
>   richard.marsh@vnet.atea.be wrote:
> > I am having major problems getting a standalone version of VxWorks to work on
> > my MBX860 (rev 08) board (PPC 860 EN). I can do the following without any
> > problems:
>
> <snip>
>
> > I have tried using the BDM to trace the boot process, but the thing always
> > dies in the usrRoot task (sysClkConnect). If I put a break afterwards it never
> > gets there.
> >
> > Any clues? You help would be appreciated.
> >
> > Regards
> >
> > Richard Marsh
>
> What address is your bootrom at versus the address of your interrupt vectors.
> If your bootrom address is higher than 2^26 (0x04000000) the interrupt service
> routine will not be able to be reached. (This assumes the MSR[IP] = 0).
> I had that problem when I tried to build a rom resident boot image.
> VxWorks uses a PPC branch instruction that only has 26 bits of address
> range in the opcode. The interrupt 'vector' (the stub that goes at 0x500)
> uses this branch to jump to the real ISR, which for your rom resident boot
> image is up in your roms. When vxWorks installs the interrupt vector stub,
> if the branch is too far it will silently truncate the address and stuff it
> in the ISR stub. When the interrupt occurs, the CPU will jump off to an
> incorrect memory address that may or may not have valid instructions in it.
>
> I hope I've explained this well enough, there is a little bit about this in
> the BSP Developer's Kit Manual, in section B.7.1 '26-bit addressing'. The
> last paragraph mentions interrupt/exception handlers.
>
> The solution is either to not build a rom resident image, or move your roms
> to a lower address that can be reached by the interrupt stub.
>
> Pete Kockritz
> pkockritz@boschtelecominc.com
>
> -----------== Posted via Deja News, The Discussion Network ==----------
> http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: What is RT-11 file system?
Date: Sat, 06 Feb 1999 15:28:05 -0800
From: Hwa Jin Bae <hjb@pso.com>
Organization: PSO Systems Inc
Message-ID: <36BCD005.5AB84C36@pso.com>
References: <36BA8F15.E6D51CC1@yamato.ibm.com> <36BB5003.CA71C5D2@xyplex.com> <79fjg7$nek6@overload.lbl.gov> <36BB5AB3.FF8E4FD@lintek.com> <h1Nu2.6303$RK5.3608@news.rdc1.az.home.com>

Well... LSI-11 would qualify as a micro.  I have a couple of LSI-11's
that are in enclosure of the size similar to desktop PC's.  Also,
even though RT-11 is no longer supported, it is useful for generic
realtime data logging.  Its simplicity allows simple modifications
to achieve better reliability and performance, especially since
VxWorks DOS-FS is so buggy.

John or Jenn wrote:
> 
> I think they were marketed as Minis, not Micros...I don¹t think
> DEC did a Micro in that space (I¹m not counting the Rainbow)
> until the MicroVax line around Œ86ish...

- -- 
Hwa Jin Bae
PSO Systems Inc
mailto:hjb@pso.com
http://www.pso.com

---------------------------

Newsgroups: comp.os.vxworks
Subject: Retrieve Registers from SIGSEGV
Date: Sat, 06 Feb 1999 21:57:54 -0600
From: Glenn Hofmeister <glennh@uswest.net>
Message-ID: <36BD0F41.9840C31D@uswest.net>
References: <s6bc42cf.071@mdc.com>

Greetings,

I am a new user of vxsim (so bear with me).  I am using vxsim 5.3.1 on a
Solaris.  I have a signal handler to catch a segmentation violation
(SIGSEGV) set up as follows:

static void SPARCsigSegv(int signo, siginfo_t *infop, void *pContext)

My signal handler is working and catches the SIGSEGV.  Is there a way to
retrieve values of the registers at the time of the SIGSEGV?  I have used
taskRegsGet, but the registers are for the signal handler, I need the
registers values for the function that caused the SIGSEGV.  It appears that
I need to move up a "function" stack.  Any advice on how to retrieve the
registers for function causing the SIGSEGV would be greatly appreciated by me.

Thank you
Glenn
glennh@uswest.net



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Booting from flash
Date: Thu, 04 Feb 1999 08:26:17 GMT
From: john@kivala.com
Organization: Kivala Systems, Inc.
Message-ID: <79blj3$1op$1@nnrp1.dejanews.com>
References: <7990ii$hl53@overload.lbl.gov>

Our 860 board runs with VxWorks resident in flash; we build
vxWorks.res_rom_nosym. I don't remember how much tweaking if
any I did to get it working. For development, we run with
1 MB flash and 1 MB of RAM. Production systems should work
with 1 MB flash and 512 KB of RAM; one of the ram chips is
hand soldered on the back of the board so it's not there in
normal production, without leaving a big blank space. From
memory, our VxWorks is roughly 400-500 KB, with no network,
and the application is maybe 300 KB (so far). If you are
planning a small-memory system, try to figure a way to have
more memory during development. And pad that board for the
next size up flash, too. We use the Abatron BDM just for
flash programming these days, but it and I were real good
buddies bringing up the BSP and some "early init" drivers.
I recommend them as a good choice, although I think EST
would be a good choice too. I respectfully disagree with
Jim L in that I don't think you need an ICE unless you're
really blazing new trails with your hardware (maybe that's
what he meant). I would guess that if such a thing exists
for the 860, it's not real cheap.
Abatron is http://www.abatronag.ch

John

- --------------------------------------------------
 John Finley                 Kivala Systems, Inc.
 Project Manager                   (619) 689-0032
 john@kivala.com            http://www.kivala.com
- --------------------------------------------------
 VxTool - Build Tornado Apps with MS Visual Basic
- --------------------------------------------------

In article <7990ii$hl53@overload.lbl.gov>,
  Johan Bostedt <qsabost@rsa.ericsson.se> wrote:
> Hi VxWorkers,
>
> We are designing a custom board based on the MCP860 processor. I have a
> question concerning
> the loading of the boot code and the VxWorks image. We would like to
> have it all in flash memory.
>
> Is it possible to acheive this without a boot PROM?
>
> What do you need to program the boot monitor and the VxWorks image into
> flash?
> We have an ethernet port and an RS232 port connected to the
> microprocessor.
>
> What about sizes? How big is the bootrom code and how
> big is a typical VxWorks image?
>
> Best regards,
> Johan Bostedt
> _________________________________________________
> Johan Bostedt     phone: +46(0)8-404 87 96
> RSA/B/UD          cellular: +46(0)70-657 64 70
>                   email: qsabost@rsa.ericsson.se
>
>

- -----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

---------------------------

End of New-News digest
**********************


From owner-vxwexplo-process  Mon Feb  8 01:44:07 1999
From: MacroSystem_FE_trahn@t-online.de (Thomas Rahn)
Date: Mon Feb  8 01:44:10 PST 1999
Subject: Re: Disabling console output for a ROMed application


Dave asked:

> Can anyone tell me if there's a recognised way of producing a vxWorks
> image with terminal output disabled?  We need to use all of the serial
> ports to communicate with external devices, and thus need to prevent
> console output from appearing on them.
> 
> I've produced an image with the console bound to /null rather than the
> serial port in usrConfig.c, but it seems to lock up during boot.
> 
> 

There are three (or more) sources for the VxWorks output:
	printErr()
	printf()
	logMsg()

They print to stderr / stdout.  These file descriptors are initialized
in bootConfig.c, resp. usrConfig.c via

    ioGlobalStdSet (STD_IN,  consoleFd);
    ioGlobalStdSet (STD_OUT, consoleFd);
    ioGlobalStdSet (STD_ERR, consoleFd);

The consoleFd is calculated at runtime, based on the BSP configuration
(with r w/o CONSOLE, PC_CONSOLE, TTY, etc.)  

To turn off the output, try: 
1)  modify the #define CONSOLE_TTY entry (somewhere in 
config.h or pc.h).

2) use a call to ioGlobalStdSet () to change the stdio file descriptors

3) wrap/delete the lines in ???config.c  


Thomas Rahn



From owner-vxwexplo-process  Mon Feb  8 01:45:14 1999
From: MacroSystem_FE_trahn@t-online.de (Thomas Rahn)
Date: Mon Feb  8 01:45:18 PST 1999
Subject: Re: dosFsVolFormat


Hi Joe,

> I saw a posting in comp.os.vxworks referencing dosFsVolFormat(...)
> for VxWorks.  I was wondering where you got this function.
> 

The function is from a dosFS 2.0 EAR (=early access release =
software for someone really troublesome who calls the support 
every other week ;-)

Anyway, the function is similar to dosFsMkfs() or dosFsDevinit().
As far as I remember, the problem described in the newsgroup 
(hard disk MBR / partition boot block data and booting with VxLd) 
is the same with the standard dosFsLib.

The EAR (dated september) contains functions for disk consistency 
checking and partition table reading, but does not yet support 
partition table creation....

Bye

Thomas Rahn



From owner-vxwexplo-process  Mon Feb  8 04:00:14 1999
From: daemon@csg.lbl.gov
Date: Mon Feb  8 04:00:18 PST 1999
Subject: comp.os.vxworks newsdigest

Comp.Os.Vxworks Daily Digest    Mon Feb  8 04:00:10 PST 1999

        Subject: Re: FAQ: Why it is called as Real Time?
        Subject: Re: Get time from an NT Host?
        Subject: Re: VxWorks:  booting from IDE and using startup scripts
        Subject: Re: Problem booting standalone vxWorks from Flash
        Subject: Re: Problem booting standalone vxWorks from Flash
        Subject: Re: Booting from flash
        Subject: Try to make "bootrom_uncmp"
        Subject: What is RT-11 file system?
        Subject: Problem booting standalone vxWorks from Flash
        Subject: Re: doFsVolFormat
        Subject: Re: Accessing CD-ROM from VxWorks via SCSI

-------------------------------------------------------

Newsgroups: comp.os.vxworks
Subject: Re: FAQ: Why it is called as Real Time?
Date: Thu, 04 Feb 1999 15:42:37 GMT
From: b.van.beneden@realtime-info.be
Organization: Deja News - The Leader in Internet Discussion
Message-ID: <79cf5a$mer$1@nnrp1.dejanews.com>
References: <796mnn$hl51@overload.lbl.gov>

Go to the real-time encyclopedia (www.realtime-info.be). Real-Time Consult
has launched an RTOS evaluation program. You can register free of charge.
Those who are registered can obtain several papers and evaluation reports.
One of the papers available is "What makes a good RTOS". This paper describes
in great detail what ingredients an operating system needs to qualify as a
real-time operating system.

Bart.


In article <796mnn$hl51@overload.lbl.gov>,
  Sasikumar S <ssasi_kumar@hotmail.com> wrote:
>
> --------------F0D806CF9340869E3DC8F617
> Content-Type: text/plain; charset=us-ascii
> Content-Transfer-Encoding: 7bit
>
> When a OS can be called as Real Time OS?
> or
> What are ADDITIONAL Functionalities a Real Time OS has to possess ?
> or
> What makes the OS to be called as Real Time OS?
>
> you can give relative examples w.r.t VxWorks RTOS and other os like
> DOS,Window'95/NT,unix etc.
>
> Pls do reply,
> Thanking you,
> SASi
> --
> -------------------------------------------------
> SASIKUMAR S
> EMAIL:  ssasi_kumar@hotmail.com
>  sasikumar@ncoretech.com
> -------------------------------------------------
> Software Engineer
> NCore Technology Pvt Ltd.
> Leo complex 4th floor,
> 44 & 45 Residency cross Rd.
> Bangalore - 560 025.
> Ph:91-080-558 0405/8257
> http://www/ncore.soft.net
> -------------------------------------------------
> C/o Chandran compound,
> Ist main, Ist cross street,
> V.P. Road,Madivala,Bangalore-560 068.
> -------------------------------------------------
>
> --------------F0D806CF9340869E3DC8F617
> Content-Type: text/html; charset=us-ascii
> Content-Transfer-Encoding: 7bit
>
> <!doctype html public "-//w3c//dtd html 4.0 transitional//en">
> <html>
> When a OS can be called as <b>Real Time OS</b>?
> <br>or
> <br>What are <b>ADDITIONAL Functionalities </b>a <b>Real Time OS </b>has
> to possess ?
> <br>or
> <br>What <b>makes </b>the OS to be called as <b>Real Time OS</b>?
> <p>you can give relative examples w.r.t VxWorks RTOS and other os like
> DOS,Window'95/NT,unix etc.
> <p>Pls do reply,
> <br>Thanking you,
> <br>SASi
> <br>--
> <br>-------------------------------------------------
> <br>SASIKUMAR S
> <br>EMAIL:&nbsp; ssasi_kumar@hotmail.com
> <br>&nbsp;sasikumar@ncoretech.com
> <br>-------------------------------------------------
> <br>Software Engineer
> <br>NCore Technology Pvt Ltd.
> <br>Leo complex 4th floor,
> <br>44 &amp; 45 Residency cross Rd.
> <br>Bangalore - 560 025.
> <br>Ph:91-080-558 0405/8257
> <br><A HREF="http://www/ncore.soft.net">http://www/ncore.soft.net</A>
> <br>-------------------------------------------------
> <br>C/o Chandran compound,
> <br>Ist main, Ist cross street,
> <br>V.P. Road,Madivala,Bangalore-560 068.
> <br>-------------------------------------------------
> <br>&nbsp;</html>
>
> --------------F0D806CF9340869E3DC8F617--
>
>

- -----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Get time from an NT Host?
Date: Thu, 04 Feb 1999 20:16:38 -0200
From: Armando Paz <apaz001@ibm.net>
Organization: IBM.NET
Message-ID: <36BA1C45.95BC4DB1@ibm.net>
References: <36A396C7.5D862297@boeing.com>

Ken Ostaszewski wrote:

> Hi everyone,
>
> I was wondering if there was a way to get the host system time & date
> from an NT machine.  I know of the way to do it from an Unix host, but
> this method doesn't work for NT.
>
> Any help would be appreciated.
>
> Ken Ostaszewski

I didn't check this, but there's an ICMP request normally used to measure
the transit time of packets in a network. From the information that I've
got, the reply for the request carries the current value for the time of
day in the machine. And as far as I know NT answers ICMP requests, at
least ECHO request :-)

You should check the format of the reply.

Bye,
Armando


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: VxWorks:  booting from IDE and using startup scripts
Date: Thu, 4 Feb 1999 23:04:19 +0100
From: "Mario Sangalli" <mariosan@tin.it>
Organization: TIN
Message-ID: <79d60e$6j$1@nslave1.tin.it>
References: <78odmm$dd42@overload.lbl.gov>

>I have a problem that I have a hard time believing is
>unique.  I'm trying to separate out our application code
>from our VxWorks image for release and upgrade reasons.
>The goal is to have both images resident on our IDE drive,
>load the vxWorks image, load the application code at the
>end of usrRoot(), and then kick off a call to a function
>in the application code.  Seems like an feasible thing to
>me.


It's Ok, it the way we are using during the prototyping of our
application...

>
>I've heard that you can't use startup scripts when booting
>from anything other than LAN, but I find that hard to believe.
>In practice, this seems to be true since the definition of
>INCLUDE_NET_SYM_TBL causes dosFsMkFs to fail and leaves the
>target shell dumb.  Here's the output from the serial port:
>
>  Attaching network interface lo0... done.
>  NFS client support not included.
>  Attaching shared memory objects at 0x4100... done
>  usrIdeConfig: dosFsDevInit failed, assuming new disk and running
>dosFsMkfs. d0006
>  usrIdeConfig: dosFsMkfs failed.
>  KAM:  starting usrRoot()
>  -> i
>  undefined symbol: i
>     ^^^^^^^^^^^^^^^^^ cute, huh?


I thing that You do not have placed the vxworks.sym file on the IDE on the
same directory where
You have placed the vxWorks file.
If this file is missing, the on-board shell do not recognize any symbols and
it is not able to run the
script because the commands are not recognized.

Mario Sangalli
Instrumentation Laboratory
Paderno Dugnano (MI)
Sangalli@itpad.ilww.com







---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Problem booting standalone vxWorks from Flash
Date: Mon, 8 Feb 1999 00:19:04 -0000
From: "Geoff" <geoff.westwood@virgin.net>
Organization: Virgin Net Usenet Service       
Message-ID: <79lal9$nsg$1@nclient1-gui.server.virgin.net>
References: <79efos$eb0$1@nnrp1.dejanews.com> <79f06l$r32$1@nnrp1.dejanews.com>

Hi

I hit this problem on an PPC 821 board with a project we did about a year
ago. The details are a bit hazy now, but I thought the ROM address threshold
was 16Megs i.e
24 bits ? IIRC the default build for vxworks put the ROMS at 0x2800000 this
crashed till we lowered the ROM address to below 16Megs.
If it had been 26 Bits usable range, this should have been ok @ 0x2800000.
The limiting factor is how the branch instruction is encoded, being RISC the
instruction is only 32 bits max, hence the first bits of the opcode denote
the instruction, the remaining bits determine the branch address.
Can anyone clarify this, is the limit set by 24 or 26 bits ? Thanks.

                                                  Regards   Geoff





pkockritz@boschtelecominc.com wrote in message
<79f06l$r32$1@nnrp1.dejanews.com>...
>In article <79efos$eb0$1@nnrp1.dejanews.com>,
>  richard.marsh@vnet.atea.be wrote:
>> I am having major problems getting a standalone version of VxWorks to
work on
>> my MBX860 (rev 08) board (PPC 860 EN). I can do the following without any
>> problems:
>
><snip>
>
>> I have tried using the BDM to trace the boot process, but the thing
always
>> dies in the usrRoot task (sysClkConnect). If I put a break afterwards it
never
>> gets there.
>>
>> Any clues? You help would be appreciated.
>>
>> Regards
>>
>> Richard Marsh
>
>What address is your bootrom at versus the address of your interrupt
vectors.
>If your bootrom address is higher than 2^26 (0x04000000) the interrupt
service
>routine will not be able to be reached. (This assumes the MSR[IP] = 0).
>I had that problem when I tried to build a rom resident boot image.
>VxWorks uses a PPC branch instruction that only has 26 bits of address
>range in the opcode. The interrupt 'vector' (the stub that goes at 0x500)
>uses this branch to jump to the real ISR, which for your rom resident boot
>image is up in your roms. When vxWorks installs the interrupt vector stub,
>if the branch is too far it will silently truncate the address and stuff it
>in the ISR stub. When the interrupt occurs, the CPU will jump off to an
>incorrect memory address that may or may not have valid instructions in it.
>
>I hope I've explained this well enough, there is a little bit about this in
>the BSP Developer's Kit Manual, in section B.7.1 '26-bit addressing'. The
>last paragraph mentions interrupt/exception handlers.
>
>The solution is either to not build a rom resident image, or move your roms
>to a lower address that can be reached by the interrupt stub.
>
>Pete Kockritz
>pkockritz@boschtelecominc.com
>
>-----------== Posted via Deja News, The Discussion Network ==----------
>http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Problem booting standalone vxWorks from Flash
Date: Sun, 07 Feb 1999 19:19:51 -0600
From: petekz@cyberramp.net (Pete Kockritz)
Organization: posted via: CyberRamp.net, Dallas, TX (214) 343-3333/(817) 461-8484 for info
Message-ID: <petekz-0702991919510001@dal-tsa2-23.cyberramp.net>
References: <79efos$eb0$1@nnrp1.dejanews.com> <79f06l$r32$1@nnrp1.dejanews.com> <79lal9$nsg$1@nclient1-gui.server.virgin.net>

In article <79lal9$nsg$1@nclient1-gui.server.virgin.net>, "Geoff"
<geoff.westwood@virgin.net> wrote:

> Hi
> 
> I hit this problem on an PPC 821 board with a project we did about a year
> ago. The details are a bit hazy now, but I thought the ROM address threshold
> was 16Megs i.e
> 24 bits ? IIRC the default build for vxworks put the ROMS at 0x2800000 this
> crashed till we lowered the ROM address to below 16Megs.
> If it had been 26 Bits usable range, this should have been ok @ 0x2800000.
> The limiting factor is how the branch instruction is encoded, being RISC the
> instruction is only 32 bits max, hence the first bits of the opcode denote
> the instruction, the remaining bits determine the branch address.
> Can anyone clarify this, is the limit set by 24 or 26 bits ? Thanks.
> 
>                                                   Regards   Geoff

I think it is 26 bits, but it is a signed offset, so the range is approx.
+/- 2^25 (+/- 0x0200_0000). Here is the description of the branch instruction:

    Branch
    b target_addr (AA = 0 LK = 0)

    0 .. 5 |  6 .. 29 | 30 | 31
      18        LI      AA   LK

    LI
    Immediate field specifying a 24-bit signed two's complement integer that
    is concatenated on the right with 0b00 and sign-extended to 32 bits.

Of course, this doesn't guarantee that you get the full range. I remember
looking at what vxWorks was doing to create the ISR stub at 0x500. It
calcuated the offset of the real ISR from the stub at 0x500 and masked it
down to some number of bits (I don't remember if it was 24 or 26) and
saved the modified instruction in the ISR stub.

Regards,
Pete
- --
+-----------------------------------------------------------------+
|       Pete Kockritz                mailto:petekz@cyberramp.net  |
|                       This space for rent.                      |
+-----------------------------------------------------------------+

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Booting from flash
Date: Thu, 04 Feb 1999 21:44:48 -0600
From: Pierre Olivier <polivier@netcom.ca>
Organization: Netcom Canada
Message-ID: <36BA6930.471@netcom.ca>
References: <7990ii$hl53@overload.lbl.gov>
Reply-To: polivier@netcom.ca

Johan Bostedt wrote:
> 
> Hi VxWorkers,
> 
> We are designing a custom board based on the MCP860 processor. I have a
> question concerning
> the loading of the boot code and the VxWorks image. We would like to
> have it all in flash memory.
> 
> Is it possible to acheive this without a boot PROM?
> 
> What do you need to program the boot monitor and the VxWorks image into
> flash?
> We have an ethernet port and an RS232 port connected to the
> microprocessor.
> 
> What about sizes? How big is the bootrom code and how
> big is a typical VxWorks image?
> 
> Best regards,
> Johan Bostedt
> _________________________________________________
> Johan Bostedt     phone: +46(0)8-404 87 96
> RSA/B/UD          cellular: +46(0)70-657 64 70
>                   email: qsabost@rsa.ericsson.se

Hi There,

I would believe we have one onf the most complex setup of Vxworks
around.... Our cards boot up with a standalone (e.g. no VxWorks)
bootstrap that is originally programmed into the Flash using the
BDM port of the MPC860 to download a Flash burner program in RAM and 
program the Flashes through a serial port.

This bootstrap then executes a Power-On self-test, and offers the
possibility to do software download (through the same serial port, but
without BDM involvment) of the main application code (hosted on VxWorks)
Altough this bootstrap is copied from Flash to RAM (code + data), the
main VxWorks images runs from Flash and only the data segment is copied
to
RAM.

The image we build is vxworks.res_rom_nosym_res_low, and required
** major ** rework from the distribution files. Bootinit.c was thrown
out
and completely rewritten, usrConfig.c also had a major face lift to
cleanup
the unused architectures (it has #defines for almost every single
CPU on the planet), and the libraries also had to be tweaked
to handle our memory map (they originally assumed MSR[IP] to be 0)
Several bug reports were submitted to WindRiver, but I don't think the
fixes
we suggested (and ourselves implemented) have yet been included in any
new
patch available from WRS.

As for the image size, we typically have a breakdown of 400 KBytes for
the core
Vxworks system (no app code), and roughly 1 MByte of application code. 
Everything is of course uncompressed, since we run from Flash.

- ---------------------------------------------
Pierre Olivier
Montreal (QC) Canada
polivier@netcom.ca               (. .)
- ------------------------------oOO-(_)-OOo----

---------------------------

Newsgroups: comp.os.vxworks
Subject: Try to make "bootrom_uncmp"
Date: Thu, 04 Feb 1999 20:17:50 -0800
From: Arlin <akalench@ee.ubc.ca>
Organization: The University of British Columbia
Message-ID: <36BA70EE.40217B17@ee.ubc.ca>

Hi all:

    I try to make "bootrom_uncmp" so that I can make a boot disk to boot
up my target.  However, when I try to compile bootrom_uncmp, the
following error messages show up:

cc386 -BC:\Tornado/host/x86-win32/lib/gcc-lib/ -c -m486 -ansi -nostdinc
- -O -fvolatile -nos
tdlib -fno-builtin -fno-defer-pop -Wall -I/h   -I.
- -IC:\Tornado\target\config\all -IC:\Tor
nado\target/h -IC:\Tornado\target/src/config -IC:\Tornado\target/src/drv
- -DCPU=I80486
 -o version.o C:\Tornado\target\config\all/version.c
ld386 -X -N  -e _romInit -Ttext 00008000  \
    -o bootrom_uncmp romInit.o bootInit_uncmp.o version.o bootConfig.o \

    sysALib.o sysLib.o    C:\Tornado\target/lib/libI80486gnuvx.a
C:\Tornado\target/lib/libI80486gnuvx.a(if_ultra.o)(.text+0xc2b):
undefined reference to `m
_freem'
C:\Tornado\target/lib/libI80486gnuvx.a(uipc_mbuf.o)(.text+0x10d):
undefined reference to `
clDescTblNumEnt'
C:\Tornado\target/lib/libI80486gnuvx.a(uipc_mbuf.o)(.text+0x113):
undefined reference to `
clDescTbl'
C:\Tornado\target/lib/libI80486gnuvx.a(uipc_mbuf.o)(.text+0x118):
undefined reference to `
mClBlkConfig'
C:\Tornado\target/lib/libI80486gnuvx.a(uipc_mbuf.o)(.text+0x12f):
undefined reference to `
sysClDescTblNumEnt'
C:\Tornado\target/lib/libI80486gnuvx.a(uipc_mbuf.o)(.text+0x135):
undefined reference to `
sysClDescTbl'
C:\Tornado\target/lib/libI80486gnuvx.a(uipc_mbuf.o)(.text+0x13a):
undefined reference to `
sysMclBlkConfig'
C:\Tornado\target/lib/libI80486gnuvx.a(if_loop.o)(.text+0xcb): undefined
reference to `mfr
ee'
C:\Tornado\target/lib/libI80486gnuvx.a(if_loop.o)(.text+0xfe): undefined
reference to `mbs
tat'
C:\Tornado\target/lib/libI80486gnuvx.a(if_loop.o)(.text+0x105):
undefined reference to `mb
stat'
C:\Tornado\target/lib/libI80486gnuvx.a(if_loop.o)(.text+0x110):
undefined reference to `mf
ree'
C:\Tornado\target/lib/libI80486gnuvx.a(if_loop.o)(.text+0x135):
undefined reference to `m_
more'
C:\Tornado\target/lib/libI80486gnuvx.a(if_loop.o)(.text+0x1dd):
undefined reference to `m_
freem'
C:\Tornado\target/lib/libI80486gnuvx.a(if_loop.o)(.text+0x290):
undefined reference to `m_
freem'
make.exe: *** [bootrom_uncmp] Error 0x1

Done.
- -----------------
We use the Ethernet Card: Intel Ether Express PRO100B PCI

We have the following lines in the config.h file:
#if CPU==I80486

#define DEFAULT_BOOT_LINE \
"fd=0,0(0,0)host:/fd0/vxWorks.st h=XXX.XX.XX.XX e=XXX.XX.XX.XX u=vxworks
pw=vw tn=target o=fei"
#else
#define DEFAULT_BOOT_LINE \
"fd=0,0(0,0)host:/fd0/vxWorks.st h=XXX.XX.XX.XX e=XXX.XX.XX.XX u=vxworks
pw=vw tn=target o=fei"
#endif /* CPU==I80486 */

...

#ifdef INCLUDE_FEI
#define IO_ADRS_FEI 0x240
#define INT_LVL_FEI 0x0a
#define INT_VEC_FEI (INT_NUM_IRQ0 + INT_LVL_FEI)
#define NRF_FEI  0x00
#define CONFIG_FEI 3 /* 0=EEPROM 1=AUI  2=BNC  3=RJ45 */
#endif

==========
Q. 1. : Is there anything wrong with the parameters set in config.h?
(esp. the parameters set for the Ethernet card?)

Q2.: How to solve the above compilation errors?

Any hints or help would be greatly appreciated.

Sylvia


---------------------------

Newsgroups: comp.os.vxworks
Subject: What is RT-11 file system?
Date: Fri, 05 Feb 1999 15:26:29 +0900
From: hori <shori@yamato.ibm.co.jp>
Organization: IBM
Message-ID: <36BA8F15.E6D51CC1@yamato.ibm.com>
Reply-To: Shintaroh@yamato.ibm.co.jp, Hori@yamato.ibm.co.jp,
        ">"@yamato.ibm.co.jp


Hi,

I'm a novice of VxWorks. I knew that VxWorks supported MS-DOS FAT and
RT-11
filesystem, but I don't know about RT-11. I'd appreciate any information
or
a pointer to it.

Thanks.

Shintaroh Hori
Advanced S/W Dev. #1   Yamato Lab.  IBM JAPAN
Internet-Mail: shori@jp.ibm.com


---------------------------

Newsgroups: comp.os.vxworks
Subject: Problem booting standalone vxWorks from Flash
Date: Fri, 05 Feb 1999 10:05:19 GMT
From: richard.marsh@vnet.atea.be
Organization: Deja News - The Leader in Internet Discussion
Message-ID: <79efos$eb0$1@nnrp1.dejanews.com>

I am having major problems getting a standalone version of VxWorks to work on
my MBX860 (rev 08) board (PPC 860 EN). I can do the following without any
problems:

1. make clean bootrom_res_high.hex
2. Blow the flash using AMC BDM
3. Peform a boot
4. Configure the bootline (for file created next)
5. make clean vxWorks
6. Reboot

And all works are planned. I include the following additional defines at the
end of my config.h

#define INCLUDE_CPLUS
#define INCLUDE_CPLUS_IOSTREAMS
#define INCLUDE_CPLUS_HEAP
#define INCLUDE_SHELL
#define INCLUDE_LOADER
#define INCLUDE_UNLOADER
#define INCLUDE_SHOW_ROUTINES
#define INCLUDE_SW_FP
#define INCLUDE_PING
#define INCLUDE_NET_SYM_TBL

I get the shell, symbols loaded and can ping the board. I can then create a
target server from the launch window under Solaris, and start a debug session
using CrossWind.


- ---- BUT ---

What I need to do is combine the two steps into a single boot from flash (no
network loading). I am doing the following:

1. make clean vxWorks.res_rom_res_low.hex
2. Burn the flash using the BDM
3. Reboot

... and I get nothing, not a sausage.

I have included the following defines in my config.h

#define INCLUDE_CPLUS
#define INCLUDE_CPLUS_IOSTREAMS
#define INCLUDE_CPLUS_HEAP
#define INCLUDE_SHELL
#define INCLUDE_LOADER
#define INCLUDE_UNLOADER
#define INCLUDE_SHOW_ROUTINES
#define INCLUDE_SW_FP
#define INCLUDE_PING
#define INCLUDE_NET_INIT
#define STANDALONE_NET
#define INCLUDE_SYM_TBL
#define INCLUDE_STAT_SYM_TBL
#define INCLUDE_DEMO

and modified the usrBootLineInit in usrConfig.c

I have tried using the BDM to trace the boot process, but the thing always
dies in the usrRoot task (sysClkConnect). If I put a break afterwards it never
gets there.

Any clues? You help would be appreciated.

Regards

Richard Marsh

- -----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: doFsVolFormat
Date: Mon, 08 Feb 1999 08:56:15 +0300
From: Leonid Rosenboim <leonid@bitband.com>
Organization: BitBand Technologies Ltd.
Message-ID: <36BE7C7F.9F46C12C@bitband.com>
References: <79eu78$erm$1@trsvr.tr.unisys.com> <slrn7blvmp.5b9.chc@nasa2.ksc.nasa.gov>


"Charles H. Chapman" wrote:

> On Fri, 5 Feb 1999 09:17:33 -0500, Joe Zabaga <Joseph.Zabaga@unisys.com> wrote:
> >I'm using Wind River's VxWorks 5.3.1.
> >
> >I've heard there is a VxWorks dosFsVolFormat(...).
> >Can someone point me to where I can obtain this.
>
> Doesn't dosFsMkfs() do what you want?  (unless you're talking about
> a low-level format).
>
> Chuck

dosFsVolFormat() is part of a newly developed DosFs 2.0 file system product,
which has not been released yet, and is in many respects different from current
dosFsMkfs() and dosFsDevInit() functions.

I had the honor of leading the development of DosFs 2.0 at WRS, but now I am
no longer employed by WRS so I can not comment on release schedules.

Leonid Rosenboim
BitBand Technologies Ltd.

P.S. For two years I have not been active on this newsgroup, I was not
allowed to do so as a WRS employee, my apologies.



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Accessing CD-ROM from VxWorks via SCSI
Date: Mon, 08 Feb 1999 08:58:56 +0300
From: Leonid Rosenboim <leonid@bitband.com>
Organization: BitBand Technologies Ltd.
Message-ID: <36BE7D20.DC26FC05@bitband.com>
References: <36B0E15D.4B89E6F0@emw.ericsson.se>


Andreas Johnsson wrote:

> Is it possible to access a SCSI CR-ROM from a PPC board with
> scsi port running VxWorks 5.3 ?
> /Andreas

What you need is a CD-ROM file System which is provided by WRS now on
the
more recent Tornado 1.0.1 CD-ROMs as a separate free-of-charge
component.

Leonid Rosenboim
BitBand Technologies Ltd.


---------------------------

End of New-News digest
**********************


From owner-vxwexplo-process  Mon Feb  8 07:08:14 1999
From: ksamavedam@hns.com (Krishna Samavedam)
Date: Mon Feb  8 07:08:18 PST 1999
Subject: TFTP Tuning - vxWorks



hi,

This is regarding TFTP usage in VxWorks. We evaluated the VxWorks TFTP performance compared to FTP
performance. Looks like TFTP file transfer is taking double the amount of the time compared to that of FTP.  I
was expecting the performance to be in the reverse way as TFTP uses UDP (no acks from peer). 

I think TFTP server task is implementing some backoff mechanism based on the network capacity. Can some
body help me in tuning TFTP server task to achieve optimal performance??

Thanks in advance,
-- 
Krishna Samavedam



From owner-vxwexplo-process  Mon Feb  8 08:05:58 1999
From: jmk@alum.mit.edu (Joel Katz)
Date: Mon Feb  8 08:06:02 PST 1999
Subject: Re: RT-11

>In article <36BB5AB3.FF8E4FD@lintek.com> Kent Hamon, khamon@lintek.com
>writes:
>>RT-11 is the file format used by the old (197?-198?) Digital Equipment
>>Corporation (DEC) PDP-11 computers.  They were quite the rage for
>>"microcompters" before the PC arrived.
>
>I think they were marketed as Minis, not Micros...I don¼t think
>DEC did a Micro in that space (I¼m not counting the Rainbow) 
>until the MicroVax line around å86ish...
>
>John
>

John-
        Sorry, I normally hate to pick nits, but I couldn't resist this one. You are right in saying that Digital marketed the PDP-11's as mini-computers. (I always found the distinctions between mainframes, mini's, and micro's a bit silly, but anyway). However, I believe that the venerable MINC-11 laboratory computer based on the PDP-11 instruction set and the Q-bus would probably qualify as a PDP-11 micro-computer. I mention it only for the sake of nostalgia for other "old timers" like myself. I saw my first MINC in 1979 but I don't recall exactly when they were introduced.

        RT-11 was a Disk Operating System. The RT-11 file system in vxWorks is based on the file system that the RT-11 operating system used. Gee, thinking about it though, I'd take RT-11 over DOS any day.

--joel


Joel Katz, Principal Engineer  --- (603)-577-5895
Acuity Imaging, 9 Townsend West, Nashua, NH 03063 --- (603)-577-5965 (FAX)
visit our Web Site at     >>> www.acuityimaging.com <<<




From owner-vxwexplo-process  Mon Feb  8 08:33:57 1999
From: ksamavedam@hns.com (Krishna Samavedam) (by way of Loren Shalz <LLShalz@lbl.gov>)
Date: Mon Feb  8 08:34:00 PST 1999
Subject: TFTP Tuning..



hi,

This is regarding TFTP usage in VxWorks. We evaluated the VxWorks TFTP
performance compared to FTP
performance. Looks like TFTP file transfer is taking double the amount of
the time compared to that of FTP.  I
was expecting the performance to be in the reverse way as TFTP uses UDP (no
acks from peer). 

I think TFTP server task is implementing some backoff mechanism based on
the network capacity. Can some
body help me in tuning TFTP server task to achieve optimal performance??

Thanks in advance,
-- 
Krishna Samavedam(x1103)



From owner-vxwexplo-process  Mon Feb  8 10:17:20 1999
From: MacroSystem_FE_trahn@t-online.de (Thomas Rahn)
Date: Mon Feb  8 10:17:24 PST 1999
Subject: AW: Try to make "bootrom_uncmp" / undefined symbols

Hello,

Sylvia reported a problem about undefined symbols 
> 
>     I try to make "bootrom_uncmp" so that I can make a boot disk to boot
> up my target.  However, when I try to compile bootrom_uncmp, the
> following error messages show up:
...
> C:\Tornado\target/lib/libI80486gnuvx.a(if_ultra.o)(.text+0xc2b):
> undefined reference to `m_freem'
....
> C:\Tornado\target/lib/libI80486gnuvx.a(uipc_mbuf.o)(.text+0x10d):
> undefined reference to `
> clDescTblNumEnt'
....

We had a similar problem when we upgrated VxWorks.  
As far as I remember, in our setup only m_freem, mbstat, m_more 
were missing.

WRS support provided a solution: SENS had not been installed 
correct (though we do not use SENS, sigh).  The isntalltion seems
to mangle the library.

Having another setup run, installing the SENS package 
solved the problem.  

But:  be sure to install only SENS (they say, it is written down 
somewhere...), do not install SENS together with any other 
package (... but why does not the setup program check this?).


Thomas Rahn


#=================================================#
# MS MacroSystem GmbH                 Thomas Rahn #
# Forschung und            tel: +49(0)6151-272576 #
# Entwicklung              fax: +49(0)6151-272577 #
#                     email: trahn@macrosystem.de #
#=================================================#






From owner-vxwexplo-process  Tue Feb  9 04:00:16 1999
From: daemon@csg.lbl.gov
Date: Tue Feb  9 04:00:20 PST 1999
Subject: comp.os.vxworks newsdigest

Comp.Os.Vxworks Daily Digest    Tue Feb  9 04:00:12 PST 1999

        Subject: Please HELP to make a copy of disk!
        Subject: Re: GDB, VxWorks, Emacs
        Subject: Re: Please HELP to make a copy of disk!
        Subject: Re: TFTP Tuning - vxWorks
        Subject: Re: GDB, VxWorks, Emacs
        Subject: Re: doFsVolFormat
        Subject: Re: Please HELP to make a copy of disk!
        Subject: Re: Try to make "bootrom_uncmp"
        Subject: TFTP question
        Subject: [Q] ei, bp, sm
        Subject: VxWorks and PC-104 form module
        Subject: VxWorks OS and PC104 module.
        Subject: Another basic question...
        Subject: Q: STL port for vxworks5.3/Tornado1.0.1
        Subject: Q: template instanciation
        Subject: Re: TFTP question
        Subject: Re: Q: STL port for vxworks5.3/Tornado1.0.1
        Subject: Re: Q: template instanciation

-------------------------------------------------------

Newsgroups: comp.os.vxworks
Subject: Please HELP to make a copy of disk!
Date: Mon, 8 Feb 1999 15:12:19 +0300
From: "Valeri Soldatov" <ua3dmy@qsl.net>
Message-ID: <36bed51f.0@news.ptt.ru>

Hello everybody!
Please, please help!
How can I make a copy of hard drive with VX WORKS (SCSI) on non-VX Works
machine (NT, FreeBSD, Win95, DOS). Is it possible?

With all the best
Valeri Soldatov




---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: GDB, VxWorks, Emacs
Date: 08 Feb 1999 14:34:20 +0100
From: Martin Lorentzon <emwson@emw.ericsson.se>
Organization: Ericsson Microwave Systems AB, Molndal
Message-ID: <vmhfsxaujn.fsf@breeze.mo.emw.ericsson.se>
References: <uogn8cvki.fsf@teklogix.com>

Bill Pringlemeir <bpringlemeir@yahoo.com> writes:

> Has anyone connected the GDB that comes with Tornado/VxWorks to Emacs.

Yes, to VxSim but I guess the procedure is similar for a target
environment.  I ran M-x gdb<RET> gdbsimso<RET>.  Then I used the gdb
`target--0' command to attach my-self to the session using `wtx'.

It never even occurred to me that a debugging session outside my Emacs
was a viable solution ;-)

/Martin Lorentzson

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Please HELP to make a copy of disk!
Date: Mon, 08 Feb 1999 10:18:44 -0800
From: Bob Irwin <bobi@systemsinterface.com>
Organization: Systems Interface Software
Message-ID: <36BF2A84.2C870B5A@systemsinterface.com>
References: <36bed51f.0@news.ptt.ru>

Valeri Soldatov wrote:
> 
> Hello everybody!
> Please, please help!
> How can I make a copy of hard drive with VX WORKS (SCSI) on non-VX Works
> machine (NT, FreeBSD, Win95, DOS). Is it possible?
> 
> With all the best
> Valeri Soldatov

from non-vx machine, ftp to VxWorks target, and 'get' or 'mget' files.
This requires that vx machine be configured for ftp access.
- -- 
Bob Irwin bobi@systemsinterface.com 281-484-6661 Fx.281-484-8988 Houston

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: TFTP Tuning - vxWorks
Date: Mon, 08 Feb 1999 16:26:27 GMT
From: Bill Pringlemeir <bpringlemeir@yahoo.com>
Organization: PSINet
Message-ID: <uvhhcn9y1.fsf@teklogix.com>
References: <990208100719.ZM3468@dcn49.hns.com>
Sender: bpringle@DeadDuck

TFTP is based on UDP.  However, it must use its own ACKing mechanism.  
This method is simple - send/ACK (or timeout).  The TCP/IP used by FTP
has a sliding window protocol.  It is not necessary for the sender
to wait for ACKs all the time.  It can send several packets before
having to halt and wait for an ACK.

It is not at all surprising the FTP is faster than TFTP.  TFTP is
meant to be simple to implement (ie low resource).  The purpose is
to have a small ROM containing TFTP that can retrieve the main image.

If your main goal is performance, FTP is better than TFTP.

hth,
Bill

VxWorks, Tornado

>>>>> "Krishna" == Krishna Samavedam <ksamavedam@hns.com> writes:

    Krishna> This is regarding TFTP usage in VxWorks. We evaluated the
    Krishna> VxWorks TFTP performance compared to FTP
    Krishna> performance. Looks like TFTP file transfer is taking
    Krishna> double the amount of the time compared to that of FTP.  I
    Krishna> was expecting the performance to be in the reverse way as
    Krishna> TFTP uses UDP (no acks from peer).

    Krishna> I think TFTP server task is implementing some backoff
    Krishna> mechanism based on the network capacity. Can some body
    Krishna> help me in tuning TFTP server task to achieve optimal
    Krishna> performance??


- -- 
Brave New Waves, "http://www.radio.cbc.ca/programs/bnw/"

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: GDB, VxWorks, Emacs
Date: Mon, 08 Feb 1999 19:30:31 GMT
From: Bill Pringlemeir <bpringlemeir@yahoo.com>
Organization: PSINet
Message-ID: <uogn4n1f9.fsf@teklogix.com>
Sender: bpringle@DeadDuck


    James> When emacs asks you for the gdb command line, use
    James> gdbppc/gdb68k/gdbi86/gdb{whatever} as the executable.  Use
    James> 'target wtx yourtargetname' to attach, and you're all set.

That is so easy I fell stupid.  Supposedly the variable 'gdb-command-name'
can be set to refer to the correct executable.  I looked at gud.el and I
couldn't find a reference to this.  I think I will try to find the most
up to date gud.el.

Here are some complaints that gdbarm gave:

        "file" command is not supported with a VxWorks target.
        Use the "load" command to load an object module incrementally.
        Use the "add-symbol-file" command to load symbols from a file.

After loading an object, and checking my directory variables, my source
file still wasn't pulled up in a split frame.  Occasionally the step command
would print out a line number with source.  Do I have to set a language
mode or is something foobar with the combo of emacs19.34, gdb 4.16.  They
both are fairly antiquated programs... but I doubt if I can upgrade the gdb
version myself.

    Martin> Yes, to VxSim but I guess the procedure is similar for a
    Martin> target environment.  I ran M-x gdb<RET> gdbsimso<RET>.
    Martin> Then I used the gdb `target--0' command to attach my-self
    Martin> to the session using `wtx'.

Those are very similar to what is needed to connect to the wtx agent.
Unfortunately I have NT.  I will have to wait for Tornado 2.0 to get
VxSim.

    James> You might also be interested in ddd; it's a gdb GUI that's
    James> nicer than CrossWind.  I'll use it about one time in ten
    James> when I want a gdb front end (the rest of the time I do
    James> things under emacs.)
    James> http://mumm.ibr.cs.tu-bs.de/softech/ddd/

This does appear to be a nice tool... The bottom of the page also refers
to several other debugging agents.  thanks for the link.

thanks for the help,
Bill

- -- 
Brave New Waves, "http://www.radio.cbc.ca/programs/bnw/"

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: doFsVolFormat
Date: Mon, 8 Feb 1999 14:54:58 -0500
From: "Joe Zabaga" <Joseph.Zabaga@unisys.com>
Organization: ECDC Unisys Corp. Tredyffrin
Message-ID: <79nf41$ais$1@trsvr.tr.unisys.com>
References: <79eu78$erm$1@trsvr.tr.unisys.com> <slrn7blvmp.5b9.chc@nasa2.ksc.nasa.gov>

I think dosFsMkfs() does not do two things I need:
   1. Respect partitions as laid out by a Master Boot
      record.
   2. Check for bad blocks in the disk data area, and then create
      2 FATs that reflect the findings.

(working on a project where at some times an i960 will
 see a SCSI disk, and other times a x86 will see the
 disk.)

i don't believe dosMkfs does either of these things,
am i wrong?

Thanks,

Joe

Charles H. Chapman wrote in message ...
>On Fri, 5 Feb 1999 09:17:33 -0500, Joe Zabaga <Joseph.Zabaga@unisys.com>
wrote:
>>I'm using Wind River's VxWorks 5.3.1.
>>
>>I've heard there is a VxWorks dosFsVolFormat(...).
>>Can someone point me to where I can obtain this.
>
>Doesn't dosFsMkfs() do what you want?  (unless you're talking about
>a low-level format).
>
>Chuck



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Please HELP to make a copy of disk!
Date: Mon, 08 Feb 1999 11:42:13 -0800
From: Hwa Jin Bae <hjb@pso.com>
Organization: PSO Systems Inc
Message-ID: <36BF3E15.5BA4DDCA@pso.com>
References: <36bed51f.0@news.ptt.ru> <36BF2A84.2C870B5A@systemsinterface.com>

You can duplicated the whole disk (assuming original and duplicate
are the same type of disks) via Unix dd command.

Bob Irwin wrote:
> 
> Valeri Soldatov wrote:
> >
> > Hello everybody!
> > Please, please help!
> > How can I make a copy of hard drive with VX WORKS (SCSI) on non-VX Works
> > machine (NT, FreeBSD, Win95, DOS). Is it possible?
> >
> > With all the best
> > Valeri Soldatov
> 
> from non-vx machine, ftp to VxWorks target, and 'get' or 'mget' files.
> This requires that vx machine be configured for ftp access.
> --
> Bob Irwin bobi@systemsinterface.com 281-484-6661 Fx.281-484-8988 Houston

- -- 
Hwa Jin Bae
PSO Systems Inc
mailto:hjb@pso.com
http://www.pso.com

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Try to make "bootrom_uncmp"
Date: Mon, 08 Feb 1999 12:46:27 -0800
From: Stephen Macmanus <stephen@wrs.com>
Organization: Wind River Systems
Message-ID: <36BF4D23.357A@wrs.com>
References: <36BA70EE.40217B17@ee.ubc.ca>

Arlin wrote:
> 
> Hi all:
> 
>     I try to make "bootrom_uncmp" so that I can make a boot disk to boot
> up my target.  However, when I try to compile bootrom_uncmp, the
> following error messages show up:
> 
> cc386 -BC:\Tornado/host/x86-win32/lib/gcc-lib/ -c -m486 -ansi -nostdinc
> -O -fvolatile -nos
> tdlib -fno-builtin -fno-defer-pop -Wall -I/h   -I.
> -IC:\Tornado\target\config\all -IC:\Tor
> nado\target/h -IC:\Tornado\target/src/config -IC:\Tornado\target/src/drv
> -DCPU=I80486
>  -o version.o C:\Tornado\target\config\all/version.c
> ld386 -X -N  -e _romInit -Ttext 00008000  \
>     -o bootrom_uncmp romInit.o bootInit_uncmp.o version.o bootConfig.o \
> 
>     sysALib.o sysLib.o    C:\Tornado\target/lib/libI80486gnuvx.a
> C:\Tornado\target/lib/libI80486gnuvx.a(if_ultra.o)(.text+0xc2b):
> undefined reference to `m
> _freem'
> C:\Tornado\target/lib/libI80486gnuvx.a(uipc_mbuf.o)(.text+0x10d):
> undefined reference to `
> clDescTblNumEnt'
> C:\Tornado\target/lib/libI80486gnuvx.a(uipc_mbuf.o)(.text+0x113):
> undefined reference to `
> clDescTbl'
> C:\Tornado\target/lib/libI80486gnuvx.a(uipc_mbuf.o)(.text+0x118):
> undefined reference to `
> mClBlkConfig'
> C:\Tornado\target/lib/libI80486gnuvx.a(uipc_mbuf.o)(.text+0x12f):
> undefined reference to `
> sysClDescTblNumEnt'
> C:\Tornado\target/lib/libI80486gnuvx.a(uipc_mbuf.o)(.text+0x135):
> undefined reference to `
> sysClDescTbl'
> C:\Tornado\target/lib/libI80486gnuvx.a(uipc_mbuf.o)(.text+0x13a):
> undefined reference to `
> sysMclBlkConfig'
> C:\Tornado\target/lib/libI80486gnuvx.a(if_loop.o)(.text+0xcb): undefined
> reference to `mfr
> ee'
> C:\Tornado\target/lib/libI80486gnuvx.a(if_loop.o)(.text+0xfe): undefined
> reference to `mbs
> tat'
> C:\Tornado\target/lib/libI80486gnuvx.a(if_loop.o)(.text+0x105):
> undefined reference to `mb
> stat'
> C:\Tornado\target/lib/libI80486gnuvx.a(if_loop.o)(.text+0x110):
> undefined reference to `mf
> ree'
> C:\Tornado\target/lib/libI80486gnuvx.a(if_loop.o)(.text+0x135):
> undefined reference to `m_
> more'
> C:\Tornado\target/lib/libI80486gnuvx.a(if_loop.o)(.text+0x1dd):
> undefined reference to `m_
> freem'
> C:\Tornado\target/lib/libI80486gnuvx.a(if_loop.o)(.text+0x290):
> undefined reference to `m_
> freem'
> make.exe: *** [bootrom_uncmp] Error 0x1
> 
> Done.
> -----------------
> We use the Ethernet Card: Intel Ether Express PRO100B PCI
> 
> We have the following lines in the config.h file:
> #if CPU==I80486
> 
> #define DEFAULT_BOOT_LINE \
> "fd=0,0(0,0)host:/fd0/vxWorks.st h=XXX.XX.XX.XX e=XXX.XX.XX.XX u=vxworks
> pw=vw tn=target o=fei"
> #else
> #define DEFAULT_BOOT_LINE \
> "fd=0,0(0,0)host:/fd0/vxWorks.st h=XXX.XX.XX.XX e=XXX.XX.XX.XX u=vxworks
> pw=vw tn=target o=fei"
> #endif /* CPU==I80486 */
> 
> ...
> 
> #ifdef INCLUDE_FEI
> #define IO_ADRS_FEI 0x240
> #define INT_LVL_FEI 0x0a
> #define INT_VEC_FEI (INT_NUM_IRQ0 + INT_LVL_FEI)
> #define NRF_FEI  0x00
> #define CONFIG_FEI 3 /* 0=EEPROM 1=AUI  2=BNC  3=RJ45 */
> #endif
> 
> ==========
> Q. 1. : Is there anything wrong with the parameters set in config.h?
> (esp. the parameters set for the Ethernet card?)
> 
> Q2.: How to solve the above compilation errors?
> 
> Any hints or help would be greatly appreciated.
> 
> Sylvia


     Sylvia,

          The undefined references usually indicate an installation
     problem. All of those symbols are used to access the network 
     stacks' memory pool. The most common cause of this problem is 
     installing the SENS component release in the wrong order. It
     must be installed last, after any other Tornado products.

           None of the undefined references are related to the driver
     configuration definitions.

                                                        Stephen
- -- 
- ------------------
Stephen Macmanus                         #include <std_disclaimer.h>
Software Engineer
Wind River Systems                       stephenm@wrs.com

---------------------------

Newsgroups: comp.os.vxworks
Subject: TFTP question
Date: Mon, 8 Feb 1999 21:59:02 -0000
From: "Geoff" <geoff.westwood@virgin.net>
Organization: Virgin News Service
Message-ID: <79nmru$mgq$1@nclient3-gui.server.virgin.net>

Hi All

Hope you dont mind a slightly off topic question but we have just started
using TFTP with
a PowerPC target board running as the TFTP client. To my surprise I did a
web search and couldn't find a good freeware TFTP Server program that runs
on a PC (32 bit Windows OS)

The best I could find was a shareware TFTP Server from Walusoft. This looks
quite nice but has a nasty bug that causes it to run out of resources after
a few file transfers.

Could anyone recommend a freeware Windows TFTP Server program , any ideas
appreciated thanks.

                           Regards Geoff





---------------------------

Newsgroups: comp.os.vxworks
Subject: [Q] ei, bp, sm
Date: Mon, 08 Feb 1999 17:16:18 -0600
From: Kyoil Kim <kikim@cerc.utexas.edu>
Organization: University of Texas at Austin
Message-ID: <36BF7042.A19F2BDD@cerc.utexas.edu>

I'm running VxWorks on Motorola board.

In the help message, it has 3 booting devices, ei, bp and sm.

What's the meaning of these?

Please let me know it. Thanks.



---------------------------

Newsgroups: comp.os.vxworks
Subject: VxWorks and PC-104 form module
Date: Mon, 08 Feb 1999 20:51:03 -0500
From: Cluster User <gt2869a@prism.gatech.edu>
Organization: Gideon R. Saroufiem. Georgia Institute of Technology
Message-ID: <36BF9486.3869D2B2@prism.gatech.edu>

My name is
Gideon Saroufiem and I am one of the researchers at Georgia Tech who
are developping a next generation UAV (unmanned aerial vehicle).

My role on the team is to choose the onboard computer system.
We are in agreement that we would like to implement
a VxWorks OS to handle the event scheduling.

The key question is to find the right PC-104-form module that
is compatible with VxWorks OS.

The module need also have

             Intel Based Processor(s) (For cross platform compilation)
             Payload (Weight, Dimentions, etc)
                       - smaller/lighter the better
             Ethernet Capabilities (Wireless)
                       - I believe most modules come
                         with an ethernet connection but
                         not wireless. We may need an adapter
                         to the board.
             Processor Speed
                         The faster the better, thus far
                         I have found speeds of around 333 MHz Intel
                         Based.
             Flash Memory (Boot off chip capability)
             Power Module (the unit will be connected to a
                           external chargeable battery while
                           the aircraft is in the air)
             D/A and A/D module (work the external actuators on the
                                 aircraft)


The control algorithms will be complex. While the vehicle
is in flight, the avionics need to be such that the module
responds to real-time feedback from the external sensors.

All said and done, the less complicated solution the better.

We have choosen VxWorks OS. Perhaps your experience with different
manufacturers could help us narrow down the right module for our
specific application.

Respectfully Yours,

Gideon R. Saroufiem.
Georgia Institute of Technology
Phone: 404 894 8909 (lab)
Phone: 404 206 4479 (home - ok to leave messages)
Email: gt2869a@prism.gatech.edu


---------------------------

Newsgroups: comp.os.vxworks
Subject: VxWorks OS and PC104 module.
Date: Mon, 08 Feb 1999 20:54:46 -0500
From: Cluster User <gt2869a@prism.gatech.edu>
Organization: Gideon R. Saroufiem. Georgia Institute of Technology
Message-ID: <36BF9566.9AC2F8B8@prism.gatech.edu>

Hello Everyone,

We here at Georgia Tech are trying build aircrafts that fly themselves.

My role on the team is to choose the onboard computer system.
We are in agreement that we would like to implement
a VxWorks OS to handle the event scheduling.

The key question is to find the right PC-104-form module that
is compatible with VxWorks OS.

The module need also have

             Intel Based Processor(s) (For cross platform compilation)
             Payload (Weight, Dimentions, etc)
                       - smaller/lighter the better
             Ethernet Capabilities (Wireless)
                       - I believe most modules come
                         with an ethernet connection but
                         not wireless. We may need an adapter
                         to the board.
             Processor Speed
                         The faster the better, thus far
                         I have found speeds of around 333 MHz Intel
                         Based.
             Flash Memory (Boot off chip capability)
             Power Module (the unit will be connected to a
                           external chargeable battery while
                           the aircraft is in the air)
             D/A and A/D module (work the external actuators on the
                                 aircraft)


The control algorithms will be complex. While the vehicle
is in flight, the avionics need to be such that the module
responds to real-time feedback from the external sensors.

All said and done, the less complicated solution the better.

We have choosen VxWorks OS. Perhaps your experience with different
manufacturers could help us narrow down the right module for our
specific application.

Respectfully Yours,

Gideon Saroufiem.

Georgia Institute of Technology
Phone: 404 894 8909 (lab)
Phone: 404 206 4479 (ok to leave messages)
Email: gt2869a@prism.gatech.edu


---------------------------

Newsgroups: comp.os.vxworks
Subject: Another basic question...
Date: Tue, 09 Feb 1999 13:39:18 +1100
From: Ian Fitzgerald <ian@syseng.anu.edu.au>
Organization: Australian National University
Message-ID: <36BF9FD6.63042A53@syseng.anu.edu.au>


Hi folks

Im using a pc486 BSP, trying to just send some Visca packets to and fro
down the serial line. I figure there have already been provisions made
for writing serial drivers etc - such that I dont have to write a serial
driver of my own - surely not!  :)

Having looked at the source tree etc, I see a few files that I think
look like they might provide me with a simple way of accessing the
serial port etc. The files are sysSerial.c, i8250Sio.c/h and sioLib.h.

sysSerial is initialised at bootup, and in turn initialised structures
in i8250Sio/sioLib.

sioLib.h has declarations for a bunch of macros:
    sioIoctl
    sioTxStartup
    sioCallbackInstall
    sioPollInput
    sioPollOutput

So Ive used sysSerialChannelGet(channel) to get a ptr to the SIO_CHAN
and hence I8250_CHAN structure. Then I set the registers as I want them
- - then I figure all I do is use the macros (Im trying to use polled
mode) to write and read the packets to/from the device. (I read and
write with the poll macros).

Well, its not working - god knows why. I can write (it seems), but I am
reading junk - and I was wondering if there is some <termios.h> file in
vxWorks which would make things so much easier.


Thanks in advance
Ian


---------------------------

Newsgroups: comp.os.vxworks
Subject: Q: STL port for vxworks5.3/Tornado1.0.1
Date: 09 Feb 1999 02:38:56 GMT
From: hiranabe@esm.co.jp (Kenji Hiranabe)
Organization: Eiwa System Management, Inc. /Fukui,Japan
Message-ID: <HIRANABE.99Feb9113856@monk.esm.co.jp>


Hi,

Has anybody tried porting STL for vxworks5.3/Tornado1.0.1 ?
(which has gcc2.7.2 in it)

I'm trying with

  http://corp.metabyte.com/~fbp/stl/

And it was quite successful.

Is it possible to compile libg++2.7.2 (which includes string and
STL as libstdc++) for vxworks5.3/Tornado1.0.1 ? Has somebody tried ?

I'm trying but not succeeded yet.
- --
  Java3D vecmath implementor    http://www.esm.co.jp/java/vecmath/
  Eiwa System Management, Inc.  http://www.esm.co.jp/
  Kenji Hiranabe                E-Mail: hiranabe@NOSPAMesm.co.jp

---------------------------

Newsgroups: comp.os.vxworks
Subject: Q: template instanciation
Date: 09 Feb 1999 02:31:28 GMT
From: hiranabe@esm.co.jp (Kenji Hiranabe)
Organization: Eiwa System Management, Inc. /Fukui,Japan
Message-ID: <HIRANABE.99Feb9113128@monk.esm.co.jp>


Hi,

I'm wondering how to configure C++ source code for templates
in Tornado 1.0.1 for i386.

To avoid the code bloat, I'd like to use the compiler option 
- -fno-implement-inlines with explicit template instantiation
syntax.

But I heard from a support about "PSR8483" dated Dec01,1997 saying;

- ----<QUOTE>---
ANALYSIS OF THE PROBLEM

     The C++ compiler fails to make constructor globally visible when it
     was generated by explicit instantiation of a template class.  Instead,
     it attempts to make it a `weak' symbol.  On machines without support
     for weak symbols, this strategy is incorrect.

SCOPE OF THE PROBLEM

     All targets except the ones with support for `weak' symbols will be
     affected.  (Such symbols is supported by SVR4 derived systems, ELF 
     systems, and a few others.)
- ---------------

Is there no solution ?
- --
  Java3D vecmath implementor    http://www.esm.co.jp/java/vecmath/
  Eiwa System Management, Inc.  http://www.esm.co.jp/
  Kenji Hiranabe                E-Mail: hiranabe@NOSPAMesm.co.jp

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: TFTP question
Date: 9 Feb 1999 11:32:04 GMT
From: "Kelly Hornsby" <hornsbyk@worldnet.att.net>
Organization: AT&T WorldNet Services
Message-ID: <79p6bk$iop@bgtnsc01.worldnet.att.net>
References: <79nmru$mgq$1@nclient3-gui.server.virgin.net>

There is a freebie version of a tftp server inside of the WRS distribution
now.  We purchased walusoft's tftp program and used it too but did not see
your problem.  From now on we will use the freeware version on the WRS
distribution.  It is freeware... contact me if you need a copy of it.





---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Q: STL port for vxworks5.3/Tornado1.0.1
Date: 09 Feb 1999 20:00:46 +0900
From: Kenji Hiranabe <hiranabe@esm.co.jp>
Organization: Eiwa System Management, Inc. /Fukui,Japan
Message-ID: <19990209200046R.hiranabe@esm.co.jp>
References: <199902091012.LAA32498@se-46.wpa.wtb.tue.nl>


Hi, thanks for your reply.

On Tue, 9 Feb 1999 11:12:57 +0100,
hat@se-46.wpa.wtb.tue.nl said:

>> Hi,
>> 
>> Has anybody tried porting STL for vxworks5.3/Tornado1.0.1 ?
>> (which has gcc2.7.2 in it)
>> 
>> I'm trying with
>> 
>> http://corp.metabyte.com/~fbp/stl/
>> 
>> And it was quite successful.

 > Not so succesful for me.
 > I managed to download, configure and compile,
 > but linking didn't work (with multiple files).

I managed to link and  run the regresstion test suite.

   http://corp.metabyte.com/~fbp/stl/testsuite.html

And it was OK.

 > I used set and map in 2 different c++ files. Both compiled, but linking
 > together gave a multiple definition conflict of the memory routines used
 > in both structures.
 > It appears that the static methods in the alloc template were included
 > in both object files.
 > Since I have no idea how to get around this problem, I left STL :-(

Yes. you need _PUT_STATIC_DATA_MEMBERS_HERE in one of your
compilation unit like;

#define  __DECLARE_GLOBALS_HERE
#define  __PUT_STATIC_DATA_MEMBERS_HERE
#include <vector>

and also you should define

#  define __STL_WEAK_ATTRIBUTE 0 /* for vxworks */

Would you like me to send my "stl_conf.h" ?

>> Is it possible to compile libg++2.7.2 (which includes string and
>> STL as libstdc++) for vxworks5.3/Tornado1.0.1 ? Has somebody tried ?

 > I am not sure why you would want to do this. STL and the string template
 > are both source templates, and it should thus be possible to have STL
 > and string without having to compile anything. Also, I think that this
 > library assumes several facilities to be available (like a file system
 > for example, or a full-blown process administration).

As you say, it was very hard. so I extracted only 'string'
related part and compile(hard again) and now it's woking
with STL.


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Q: template instanciation
Date: 09 Feb 1999 20:02:23 +0900
From: Kenji Hiranabe <hiranabe@esm.co.jp>
Organization: Eiwa System Management, Inc. /Fukui,Japan
Message-ID: <19990209200223L.hiranabe@esm.co.jp>
References: <199902091017.LAA32519@se-46.wpa.wtb.tue.nl>


On Tue, 9 Feb 1999 11:17:29 +0100,
hat@se-46.wpa.wtb.tue.nl said:

 > In comp.os.vxworks, you wrote:
>> To avoid the code bloat, I'd like to use the compiler option 
>> -fno-implement-inlines with explicit template instantiation

 > Imho this option prevents inlining of functions, it does not prevent
 > implicit template instantiations.
 > (use -fno-implicit-template (or something along this lines) for that)

You're right, my mistake.

>> Is there no solution ?
 > Get a better C++ compiler I think.
 > gcc 2.7.2 is very, very, very, very outdated wrt templates in C++

Does vxworks 5.3/Tornado1.0.1 support any other compilers ?

---------------------------

End of New-News digest
**********************


From owner-vxwexplo-process  Tue Feb  9 09:56:12 1999
From: Shuchi Chawla <shuchi@oresis.com>
Date: Tue Feb  9 09:56:15 PST 1999
Subject: Network Interface Extensions in VxWorks (based on RFC2233)


Tornado 1.0/VxWorks 5.3.1

Hello,

I'm relatively new to the VxWorks environment and am
hoping to get some pointers/information on extending the
network interface support as currently provided in VxWorks

We need to provide support for non-IP interfaces, and also 
support layering/stacking of interfaces.  We hope to model
the stacking of interfaces sort of similar to the model as defined 
by RFC2233's Interfaces MIB, regardless of whether we decide 
to support SNMP or not.

The VxWorks documentation mentions support built-in for
MIB-II (RFC1213), and hence the ifTable in MIB-II.  However, 
I don't see any mention of support for later MIBs (RFC1573 or 
RFC2233).  I had a question regarding this:

     Has anyone had some experience extending the network interface
     support?  Is there information available on how to add such support 
     without access to the source code for the TCP/IP stack and network 
     interface layer therein?  Are there hooks that one can use?  Is there 
     a mechanism available to extend the network interface structures to 
     support layering.  Here's  an example to illustrate the above:

	When an IMA group (for ATM) is formed, individual DS1 links 
	are grouped together to form a higher level interface.

	     IF#4
	    +------------------------------+
		 IMA Group
	    +------------------------------+
	      /              |            \
	     /               |             \
	+-------+    +--------+     +--------+
	   DS1         DS1          DS1
	+-------+    +--------+     +--------+
	IF#1          IF#2           IF#3

	In the above case, DS1 Links 1, 2 and 3 are individual
	interfaces that are grouped to form a logical interface
	IF#4.  So IF#4 is stacked on IF#1, IF#2 and IF#3.
	IF#4 may have in addition an IP interface stacked on top
	of it.

Any information/pointers on that anyone has on this will be 
very helpful.

Thanks.

Regards,
Shuchi
-----------------------------------
Shuchi Chawla
Oresis Communications, Inc.
schawla@oresis.com 




From owner-vxwexplo-process  Tue Feb  9 18:04:38 1999
From: "Venkat Tanikella" <venkat@trinc.com>
Date: Tue Feb  9 18:04:41 PST 1999
Subject: vxWorks Raw Sockets (Tornado)

vxWorks, tornado

Hi,
    I have been trying to write the traceroute utility for VxWorks.

    For some time, I have been facing problems using the IPPROTO_RAW
option.
    This is what I do:
    a) create a socket with SOCK_RAW, IPPROTO_RAW
    b) use setsockopt with IP_HDRINCL to force IP to use my IP header
    c) Create my own IP header with TTL set to 1,2... and so on
    d) send the packet using sendto() to send it to the destination.

    I observe that the behaviour is slightly varied during different
tests but the
    end result is the same. The NetTask hangs. When I run my program
from
    the shell, I see the task has incurred a program fault and shows a
trace
    of calls made at the lower levels. The trace goes through
bsdSendto(),
    sosend(), rip_usrreq(), rip_output(), ip_output, ether_output()...
some more
    routines, then copyFromMbufs(), netMblkClChainFree() and some
routine
    ahead of ip_forward() for which there is symbol information.

    Does anyone have an idea why this might be happening?

Thanks in advance,
Venkat




From owner-vxwexplo-process  Wed Feb 10 04:00:16 1999
From: daemon@csg.lbl.gov
Date: Wed Feb 10 04:00:19 PST 1999
Subject: comp.os.vxworks newsdigest

Comp.Os.Vxworks Daily Digest    Wed Feb 10 04:00:12 PST 1999

        Subject: Re: Booting vxworks on PC (Dell Pentium II)
        Subject: Re: Another basic question...
        Subject: Re: [Q] ei, bp, sm
        Subject: Re: Booting vxworks on PC (Dell Pentium II)
        Subject: Re: When to recompile bootrom?
        Subject: Graphics under VxWorks on MVME2604
        Subject: TCP/IP Performance on MVME2302 ?

-------------------------------------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Booting vxworks on PC (Dell Pentium II)
Date: Tue, 09 Feb 1999 12:20:33 +0000
From: Phil Ashby <phil.ashby@wireplay.bt.co.uk>
Organization: BT Wireplay
Message-ID: <36C02811.2302@wireplay.bt.co.uk>
References: <79neme$fsr$1@nnrp1.dejanews.com>

raons@ornl.gov wrote:
>         I am trying to boot vxworks on a PC using dikette. But I am not able to
>         get it connected to the server. The PC has 3COM Ethernet II card. I
>         specified the boot device as fd. Unfortunately, the documentation is not
>         helpful in specifying how to specify the network parameters. I entered
>         the IP addresses of the host and target and other entries that are
>         direct. I entered elt on the line "other(o)". The system boots vxworks
>         but th  network commands (such as ping) do not work.

Are you booting from the default bootrom disk or have you built your own
bootrom.sys?

The default bootrom is unlikely to have the correct drivers or
configuration
for your ethernet board. If you have built your own bootrom.sys then you
must
check that you have the correct drivers and configuration in config.h

HTH,
- -- 
!--------------------------- Phil "Phlash" Ashby ---------------------!
! BT Wireplay - Lead Programmer and general technical trivia supplier !
! Phone: 01473-644348                   WWW: http://www.wireplay.com/ !
! Snail: B81/G41, BT Labs, Martlesham Heath, Ipswich IP5 3RE, England.!

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Another basic question...
Date: 9 Feb 1999 13:33:02 GMT
From: chc@nasa2.ksc.nasa.gov (Charles H. Chapman)
Organization: NASA, Kennedy Space Center
Message-ID: <slrn7c0e6u.a49.chc@nasa2.ksc.nasa.gov>
References: <36BF9FD6.63042A53@syseng.anu.edu.au>

On Tue, 09 Feb 1999 13:39:18 +1100, Ian Fitzgerald <ian@syseng.anu.edu.au>
 wrote:

>Im using a pc486 BSP, trying to just send some Visca packets to and fro
>down the serial line. I figure there have already been provisions made
>for writing serial drivers etc - such that I dont have to write a serial
>driver of my own - surely not!  :)

I don't know how things are set up on the pc486 BSP, but on the Motorola
680X0 BSP's I'm using here, there are already VxWorks devices defined for
the four on-board serial ports:

/tyCo/0             
/tyCo/1             
/tyCo/2             
/tyCo/3             

To use these, all I have to do is do an "open" on them and use the
standard "read" and "write" functions on the resulting file descriptor.

Chuck

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: [Q] ei, bp, sm
Date: 9 Feb 1999 13:36:57 GMT
From: chc@nasa2.ksc.nasa.gov (Charles H. Chapman)
Organization: NASA, Kennedy Space Center
Message-ID: <slrn7c0ee9.a49.chc@nasa2.ksc.nasa.gov>
References: <36BF7042.A19F2BDD@cerc.utexas.edu>

On Mon, 08 Feb 1999 17:16:18 -0600, Kyoil Kim <kikim@cerc.utexas.edu> wrote:
>I'm running VxWorks on Motorola board.
>
>In the help message, it has 3 booting devices, ei, bp and sm.
>
>What's the meaning of these?

ei is the Ethernet interface.  bp is for backplane networking (I'm
not sure but I think that's obsolete now).  sm is for shared memory
networking which I believe replaced backplane networking in later
versions of VxWorks.

Chuck

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Booting vxworks on PC (Dell Pentium II)
Date: Tue, 09 Feb 1999 08:12:18 -0600
From: Frank Minich <fminich@ti.com>
Organization: NCP
Message-ID: <36C04242.7B1A@ti.com>
References: <79neme$fsr$1@nnrp1.dejanews.com>
Reply-To: fminich@ti.com

raons@ornl.gov wrote:
> 
>         I am trying to boot vxworks on a PC using dikette.

Dikette?

>         The PC has 3COM Ethernet II card.

Stop the boot sequence during the countdown, then "help" to see
what boot devices are available.  The "elt" device supports the
3C509 (an older ISA card), not the 3C590 (i.e., the EtherLink II).

>         I specified the boot device as fd. Unfortunately, the 
>         documentation is not helpful in specifying how to specify 
>         the network parameters. 

Actually, you can't specify the network parameters.  The boot floppy
was built with a specific I/O address and IRQ level specified for the
driver.  If your 509 card hasn't been DOS-program configured for that
I/O address and IRQ, the boot floppy can't use it.  You can either 
acquire a network card supported by your boot floppy, or build a
new boot floppy with a network driver for your card.

>         the IP addresses of the host and target and other entries 
>         that are direct. I entered elt on the line "other(o)". 

"elt" is the driver for the 509, not the 590.

>         The system boots vxworks

It is booting from floppy if you have "fd" as the boot device

>         but th  network commands (such as ping) do not work.

Good luck,
Frank Minich
fminich@ti.com

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: When to recompile bootrom?
Date: Tue, 09 Feb 1999 08:20:00 -0600
From: Frank Minich <fminich@ti.com>
Organization: NCP
Message-ID: <36C04410.759@ti.com>
References: <79mvsv$39e$1@nnrp1.dejanews.com>
Reply-To: fminich@ti.com

jyl97@my-dejanews.com wrote:
> 
> Hi,
> 
> Can somebody tell me when to recompile bootrom (or boot.bin)?

We recompile if we:
o  move the location of the BOOT_LINE in memory (bootrom and booted
    code have to agree on where it is)
o  add or modify an I/O driver from which we could potentially boot
    (e.g., disk, network)
Regards,
Frank.

---------------------------

Newsgroups: comp.os.vxworks
Subject: Graphics under VxWorks on MVME2604
Date: Tue, 9 Feb 1999 23:22:40 -0000
From: "Mark Setchell" <dpl@dial.pipex.com>
Organization: UUNET WorldCom server (post doesn't reflect views of UUNET WorldCom)
Message-ID: <79qh7g$9rn$1@plug.news.pipex.net>

Has anyone done any graphics with either RtX-Windows or Zinc on a MVME2604
with VxWorks and are you able to recommend a good, compatible PMC graphics
card ?

Thanks
Mark



---------------------------

Newsgroups: comp.os.vxworks
Subject: TCP/IP Performance on MVME2302 ?
Date: Wed, 10 Feb 1999 07:35:58 GMT
From: wittig@vs.dasa.de
Organization: Deja News - The Leader in Internet Discussion
Message-ID: <79rcst$sj7$1@nnrp1.dejanews.com>



Hi all,

we intend to use a plain Tornado 1.0, vxWorks 5.3 kernel on a
MVME2302 board.

I have been told, that vxWorks 5.3 will support the DEC21140
with 100 MBit. It this correct ?

I'm certain, that someone has already done this, so here's a second question:

What would be the typically achievable TCP/IP throughput between a
MVME2302 board and a 350 MHz PII running NT4.0 ?

If possible, could you please respond by email.

Thanks in advance,

Juergen Wittig


- -----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

---------------------------

End of New-News digest
**********************


From owner-vxwexplo-process  Wed Feb 10 06:05:54 1999
From: John Ford <jford@bara.gb.nrao.edu>
Date: Wed Feb 10 06:05:57 PST 1999
Subject: ntp daemon and SENS

Hi.

We have installed the SENS stack, and our implementation of ntp,
derived from version 3.1, has quit working.  No other hosts can talk
to the server running on vxWorks.  Does anyone know why this may have
happened?  I can't see what would cause it.  Backing up to version
5.3.1 without SENS works just fine.

Version: VxWorks 5.3.1 
SENS version: 1.0 + a patch 

Thanks for any insight!

-- 
John Ford
National Radio Astronomy Observatory
Green Bank, WV 24944-0002
jford@nrao.edu



From owner-vxwexplo-process  Wed Feb 10 06:23:58 1999
From: RobertK54@aol.com
Date: Wed Feb 10 06:24:02 PST 1999
Subject: Tornado 1.0.1 - ARM - NT

Can anyone explain why compiling with the '-g' (debug) option, the resulting
arm-coff file contains .stab and .stabstr sections that the Tornado tools have
trouble with?  I get the many of the following warning from the Target Server:

	Warning: Symbol .stabstr from section of type STYP_REG. Ignored.

It seems the Debugger generates similar warnings along with many anoying
dialog windows that need to be 'Ok'ed away.

It seems as though -g is the same as -gstabs.  I tried specifying -gcoff but
it causes an NT application error. Using -ggdb also seems to generate the same
as -gstabs.

Thanks,
Bob


From owner-vxwexplo-process  Wed Feb 10 12:31:49 1999
From: Mike Cuddy <mcuddy@fensende.com>
Date: Wed Feb 10 12:31:53 PST 1999
Subject: Re: Tornado 1.0.1 - ARM - NT 


> Can anyone explain why compiling with the '-g' (debug) option, the resulting
> arm-coff file contains .stab and .stabstr sections that the Tornado tools have
> trouble with?  I get the many of the following warning from the Target Server:
> 
> 	Warning: Symbol .stabstr from section of type STYP_REG. Ignored.
> 
> It seems the Debugger generates similar warnings along with many anoying
> dialog windows that need to be 'Ok'ed away.

I got this error, too.  The only way that I found around it was to 
use strip to remove the .stab and .stabstr sections.

(on the other hand, I didn't try real hard to get around it, either ;-)

--
Mike Cuddy (mcuddy@FensEnde.com, MC312), Programmer, Daddy, Human.
Fen's Ende Software, Redwood City, CA, USA, Earth, Sol System, Milky Way.
I remember asking why ... Let it rain, and protect us from this Cruel Sun.

       Join CAUCE: The Coalition Against Unsolicited Commercial E-mail.  
                          <http://www.cauce.org/>
 


From owner-vxwexplo-process  Thu Feb 11 03:41:53 1999
From: "Land Systems LSA2" <lsa2@dera.gov.uk>
Date: Thu Feb 11 03:41:57 PST 1999
Subject: RE: Graphics on 2604 vxworks



VxWorks
Tornado


We are using a 2604 with RtX Windows but the configuration is 
a bit of a problem. We are using the CETIA P9100
PMC but I have two problems logged at the moment.
(1) No mouse support on 2604 (there is a keyboard patch 
apparently but I havent tried it yet).
(2) Lack of RtX configuration under EXTENDED_VME.

We have managed to get mouse, keyboard, screen
working by exporting the graphics to a Sun though.

Pete Gardiner
DERA UK
lsa2@taz.dra.hmg.gb


From owner-vxwexplo-process  Thu Feb 11 04:00:13 1999
From: daemon@csg.lbl.gov
Date: Thu Feb 11 04:00:17 PST 1999
Subject: comp.os.vxworks newsdigest

Comp.Os.Vxworks Daily Digest    Thu Feb 11 04:00:10 PST 1999

        Subject: Patcher on vxworks
        Subject: Re: MPC860 and cache
        Subject: symFindByName()
        Subject: Re: symFindByName()
        Subject: Registry Server
        Subject: Re: GDB, VxWorks, Emacs
        Subject: Can't read memory across page boundaries with PII

-------------------------------------------------------

Newsgroups: comp.os.vxworks
Subject: Patcher on vxworks
Date: Wed, 10 Feb 1999 13:38:16 +0100
From: "Baudouin, Sebastien (EXCHANGE:MLV:0000)" <mlvsxb04@europem01.nt.com>
Organization: Nortel
Message-ID: <36C17DB8.2DC4544B@europem01.nt.com>

i'm using tornado 1.0.1 for 486 CPU, and
i'd like too know if it is possible to patch a procedure.
I'm looking for documentation or existing tools to
study that.

Cheers,

Sebastien BAUDOUIN

---------------------------

Newsgroups: comp.os.vxworks,comp.sys.powerpc.tech
Subject: Re: MPC860 and cache
Date: 07 Feb 1999 15:34:47 -0500
From: James Carlson <carlson@ironbridgenetworks.com>
Organization: UltraNet Communications , an RCN Company http://www.ultranet.com/
Message-ID: <86n22q7y1k.fsf@ironbridgenetworks.com>
References: <petekz-2701991950220001@dal-tsa31-56.cyberramp.net>

petekz@cyberramp.net (Pete Kockritz) writes:
> If I enable the I-cache (either with or without the D-cache) the board
> won't boot. I get some garbage out of the console port and it hangs. I
> tried enabling the caches and making all the memory sections non-cacheable
> in sysPhysMemDesc[], but that didn't work either. I can enable the D-cache
> without enabling the I-cache and it works fine. I am running at 48MHz
> with SDRAM and flash memory. I have the internal registers of the 860 
> mapped as non-cacheable and guarded.
> 
> I tried leaving the bootrom compiled without enabling the cache, but
> compiled the vxWorks kernel with both caches on... and it works! Is there
> some problem with the I-cache running from flash? The flash is controlled
> by the GPCM, so it is not a burst issue.

Even without real burst cycles, the cache controller will seize the
bus for a long period of time when reading from slow external devices.
The cache will do many back-to-back cycles and won't let go to allow
the SDMA to run, even when SDMA is in "aggressive" mode.  (We ran into
ugly problems in Ethernet mode SCCs with the icache enabled using an 8
bit FLASH.)

Another possible problem is that icache powers up corrupted.  You have
to do an invalidation manually first before turning it on.

- -- 
James Carlson, Consulting S/W Engineer  <carlson@ironbridgenetworks.com>
IronBridge Networks / 55 Hayden Avenue  71.246W    Vox:  +1 781 372 8132
Lexington MA  02421-7996 / USA          42.423N    Fax:  +1 781 372 8090
"PPP Design and Debugging" --- http://people.ne.mediaone.net/carlson/ppp

---------------------------

Newsgroups: comp.os.vxworks
Subject: symFindByName()
Date: Wed, 10 Feb 1999 17:14:24 +0100
From: Leif Thuresson <leif.thuresson@tsl.uu.se>
Organization: TSL
Message-ID: <36C1B060.80284FDE@tsl.uu.se>
Reply-To: leif.thuresson@tsl.uu.se

I have problem getting the function symFindByName()
to work. I have configured vxWorks with

INCLUDE_SHELL
INCLUDE_NET_SYM_TBL
INCLUDE_SYM_TBL

I load my object modules from the target shell and
  lkup
from the target shell can finds my symbols but

  symFindByName(sysSymTbl, "test_sym", &symValue, &symType)

can not !

Any ideas what I have missed ?
- -- leif

- ---------------------------------------------------------
Leif Thuresson               Tel:    +46-18-4713860
Uppsala University           Fax:    +46-18-4713833
The Svedberg Laboratory      E-mail: leif.thuresson@tsl.uu.se
P.O.Box 533
S-751 21 Uppsala
Sweden



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: symFindByName()
Date: 10 Feb 1999 16:22:43 +0000
From: Chris Berry <cberry@acorn.com>
Organization: Element 14 Ltd
Message-ID: <05btj244a4.fsf@acorn.com>
References: <36C1B060.80284FDE@tsl.uu.se>

Leif Thuresson <leif.thuresson@tsl.uu.se> writes:

> 
>   symFindByName(sysSymTbl, "test_sym", &symValue, &symType)
> 
> Any ideas what I have missed ?

Just a guess: a leading underscore?

 symFindByName(sysSymTbl, "_test_sym", &symValue, &symType)


Chris.

---------------------------

Newsgroups: comp.os.vxworks
Subject: Registry Server
Date: Wed, 10 Feb 1999 09:26:35 -0500
From: Tim Sohacki <sohacki@nortelnetworks.com>
Organization: Nortel (Northern Telecom)
Message-ID: <36C1971B.20E305B3@nortelnetworks.com>

Does everyone have problems with the registry server going down,
or is it just us?

We have Win95 and HP-UX tornado installations using a registry
server on a HP-UX 10.20 box.  About once per week, the registry
server (wtxregd) dies without warning, usually under load.

Just curious if this is typical behavior ... thanks in advance.

Cheers,
Tim


- -- 
Tim Sohacki, Nortel Networks
email: sohacki@nortelnetworks.com
phone: (919) 991-2228 [ESN 35-12228]
  fax: (919) 991-4126 [ESN 35-14126]

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: GDB, VxWorks, Emacs
Date: Sun, 7 Feb 1999 19:01:00 -0800
From: "seanet" <james@banshee.com>
Organization: Seanet Online Services, Seattle WA
Message-ID: <79lkb4$nom@q.seanet.com>
References: <uogn8cvki.fsf@teklogix.com>

When emacs asks you for the gdb command line, use
gdbppc/gdb68k/gdbi86/gdb{whatever} as the executable.  Use 'target wtx
yourtargetname' to attach, and you're all set.

You might also be interested in ddd; it's a gdb GUI that's nicer than
CrossWind.  I'll use it about one time in ten when I want a gdb front end
(the rest of the time I do things under emacs.)
http://mumm.ibr.cs.tu-bs.de/softech/ddd/

James Moore
Seattle, WA, USA
james@banshee.com

Bill Pringlemeir wrote in message ...
>
>Hello,
>
>Has anyone connected the GDB that comes with Tornado/VxWorks to Emacs.
>Are there any pointers for doing this? (Web, FTP, others)  I hate to
>exit emacs to use the Host OS that is forced on me.
>
>thanks,
>Bill
>
>
>--
>Brave New Waves, "http://www.radio.cbc.ca/programs/bnw/"



---------------------------

Newsgroups: comp.os.vxworks
Subject: Can't read memory across page boundaries with PII
Date: Thu, 11 Feb 1999 04:33:25 GMT
From: William Ross <wrr@empireone.net>
Organization: DIGEX, Inc. - Beltsville, MD - http://www.digex.net
Message-ID: <36C25DD6.F038F5B1@empireone.net>

This is a multi-part message in MIME format.
- --------------B482C43A8DFE58CCBEA31B5C
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit


Hi All,

	I am running VxWorks 5.3.1 on a PII-based system with a
	2 GByte PCI memory card mapped in four 512 MByte segments.
	The lowest segment is mapped from 0x40000000 to 0x5fffffff.
	Whenever I try to read memory from anywhere in this region,
	and my read crosses a 0x1000 page boundary, my machine hangs.
	What I typically see happen is that the read will stop at 
	8 bytes before the page boundary,and hang the system.

	The code that fails looks something like this:

	long line[1024];
	memcpy((char*)line,0x40000000,1024*4);

	However, if I truncate the read to 
	
	long line[1024];
	memcpy((char*)line,0x40000000,(1024-3)*4);

	then everything works perfectly.

	Is this some weird PII effect?  Are DMAs impossible with 
	VxWorks?  I would appreciate hearing from anyone who has seen
	anything like this.

					Thanks

					Bill Ross
					GE Corporate R&D
- --------------B482C43A8DFE58CCBEA31B5C
Content-Type: text/x-vcard; charset=us-ascii;
 name="wrr.vcf"
Content-Transfer-Encoding: 7bit
Content-Description: Card for William Ross
Content-Disposition: attachment;
 filename="wrr.vcf"

begin:vcard 
n:Ross;William
x-mozilla-html:TRUE
version:2.1
email;internet:wrr@empireone.net
x-mozilla-cpt:;0
fn:William Ross
end:vcard

- --------------B482C43A8DFE58CCBEA31B5C--


---------------------------

End of New-News digest
**********************


From owner-vxwexplo-process  Thu Feb 11 07:03:39 1999
From: brett.smith@wg.com
Date: Thu Feb 11 07:03:43 PST 1999
Subject: problems with dosFs


     We have been experiencing I/O problems - they seem to have something 
     to do with dosFs.
     
     We are running the Java Virtual Machine on vxWorks, x86 target.  We 
     have some Java code that backs-up the vxworks executable file as well 
     as some *.jar files.  The code usually works, but once in a while, the 
     disk SEEMS to get corrupted.  Once this happens, an "open/create" 
     native call from the JVM appears to hang.  Reboot does not clear this 
     up.  Once the open hangs, FTP cannot perform dir or put.  The recovery 
     process requires a Format of the HD.
     
     DOS scandisk does not detect any disk errors, but the vxWorks impl is 
     slightly different isn't it?
     
     Has anyone experienced this type of problem?
     
     Does anyone know of any good "show" routines or debugging aids for 
     dosFs?
     
     I would really like to has something dump the FATs for me.  Does 
     anyone know how to see the FATs?
     
     Any help is greatly appreciated,
     
     Brett Smith
     brett.smith@wg.com




From owner-vxwexplo-process  Thu Feb 11 09:37:06 1999
From: Pritesh Patel <patel@sentientnet.com>
Date: Thu Feb 11 09:37:09 PST 1999
Subject: NAT implementation for VxWorks

Does anyone out there know of an available Network

Address Translation implementation that runs under

VxWorks?

Thanks,

Pritesh Patel

patel@sentientnet.com





From owner-vxwexplo-process  Thu Feb 11 21:07:41 1999
From: bob schulman <bob@seaweed.com>
Date: Thu Feb 11 21:07:44 PST 1999
Subject: Re: Graphics under VxWorks on MVME2604

Mark Setchell wrote:
> Has anyone done any graphics with either RtX-Windows or Zinc on a MVME2604
> with VxWorks and are you able to recommend a good, compatible PMC graphics
> card ?

Seaweed has a product extremely similar to RtX Windows that runs on a
2604 against a variety of PMC boards. Please visit our web site at
www.seaweed.com for more information.

bob


From owner-vxwexplo-process  Fri Feb 12 03:21:47 1999
From: Harry Stox <harry@ellips.nl>
Date: Fri Feb 12 03:21:56 PST 1999
Subject: UniROM debugging support for Tornado WDB?

Hello,

Does anyone have experience creating a WDB debugging channel for VxWorks
using the UniROM eprom simulator from Techtools? This device claims to
have
the same capabilities as AMC's NetRom, but at a fraction of the price.
The only
disadvantage is the lack of officially supported VxWorks drivers.

If anyone has experience in using it as debugging tool with WDB, I'm
very
interested in hearing about it.

Thanks in advance for your reactions.

Best regards,
Harry Stox - Ellips B.V.

---
Harry Stox (harry@ellips.nl) -- Software Engineer -- Ellips B.V.
Woenselsestraat 352A -- 5623 EG -- Eindhoven -- The Netherlands
phone: +31-40-2456540 -- fax: +31-40-2467183 -- http://www.ellips.nl



From owner-vxwexplo-process  Fri Feb 12 04:00:18 1999
From: daemon@csg.lbl.gov
Date: Fri Feb 12 04:00:22 PST 1999
Subject: comp.os.vxworks newsdigest

Comp.Os.Vxworks Daily Digest    Fri Feb 12 04:00:14 PST 1999

        Subject: Re: Problem with Motorola FADS860 boot code
        Subject: Re: Problem with Motorola FADS860 boot code
        Subject: Proxy server for VxWorks??
        Subject: VxWorks internet proxy server
        Subject: When to recompile bootrom?
        Subject: Re: problems with dosFs
        Subject: UDP Performance again
        Subject: Booting vxworks on PC (Dell Pentium II)
        Subject: Booting from a serial connection
        Subject: Raw sockets in SENS
        Subject: How to change default (ethernet) interface used by bootrom_uncmp to ATM interface
        Subject: Re: UDP Performance again
        Subject: Re: NAT implementation for VxWorks
        Subject: Re: How to change default (ethernet) interface used by bootrom_uncmp to 
        Subject: Re: When to recompile bootrom?
        Subject: Re: UDP Performance again
        Subject: How to get IP of Gateway and own IP?
        Subject: VxWorks & Intel Ultra-Low Power 486SX Eval.

-------------------------------------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Problem with Motorola FADS860 boot code
Date: Thu, 11 Feb 1999 06:57:57 -0500
From: Travis Sawyer <tsawyer@northc.com>
Organization: Newbridge Networks Incorporated
Message-ID: <36C2C5C5.9673BB7F@northc.com>
References: <36B964B4.5375CDCC@chromatis.com>

Funny... that's happening with me also.

My original board had a 40Mhz 860.  My new one is 50Mhz.  I wasn't able to get
anything from the terminal until I switched to the 5Mhz crystal.  Now I get
'garbage' on the terminal.

I changed the 'cristal' frequency in the config.h file under the ads860 config
directory to reflect 5 Mhz and not 4Mhz but now I get nothing from the terminal!?!

Please help...  you'd figure MOT would put the 5Mhz oscillator in at manufacture...
go figure!

thanx in advance,

travis sawyer

Rami Gideoni wrote:

> Hi,
>
> I had installed Tornado for MPC860 and I'm using the Motorola's FADS860
> evaluation board.
>
> Please note : The board is FADS860 - not ADS860.
>
> I burned down the bootrom.hex over the flash and find out a strange phenomena :
>
> When I use MPC860EN or MPC860T derivatives in the board, the boot code functions
> as expected.
>
> -- BUT ---
>
> when I use MPC860SAR derivative - No response from the board (acctually, the
> transmit LED of SMC 1 is turning as usuall - but no output on the terminal).
>
> I changed the frequency of the boot code from 4 Mhz crystal/24 Mhz system clock
> to  5 Mhz/50 Mhz system clock and I was suprised to find out that the MPC860SAR
> started to work !
>
> Any ideas ?
>
> Thanks in advance
>
> Rami Gideoni
>
> [WindRiver techincal support : TSR# 124446, For Mr. Gideon Chan]




---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Problem with Motorola FADS860 boot code
Date: Thu, 11 Feb 1999 09:12:53 -0500
From: Travis Sawyer <tsawyer@northc.com>
Organization: Newbridge Networks Incorporated
Message-ID: <36C2E565.3F1F466D@northc.com>
References: <36B964B4.5375CDCC@chromatis.com> <36C2C5C5.9673BB7F@northc.com>

Folks:

Here's what I had to change to get the boards up and running:

1.  Change to 5 Mhz Crystal
2.  Change Tornado/target/config/ads860/config.h
        CRISTAL_FREQ 5000000
        SPLL_FREQ_REQUESTED    FREQ_50_MHZ

Reblasted the FLASH, rebuilt the kernel.

Then all is good.

Hope this helps, Rami.

- -travis sawyer

Travis Sawyer wrote:

> Funny... that's happening with me also.
>
> My original board had a 40Mhz 860.  My new one is 50Mhz.  I wasn't able to get
> anything from the terminal until I switched to the 5Mhz crystal.  Now I get
> 'garbage' on the terminal.
>
> I changed the 'cristal' frequency in the config.h file under the ads860 config
> directory to reflect 5 Mhz and not 4Mhz but now I get nothing from the terminal!?!
>
> Please help...  you'd figure MOT would put the 5Mhz oscillator in at manufacture...
> go figure!
>
> thanx in advance,
>
> travis sawyer
>
> Rami Gideoni wrote:
>
> > Hi,
> >
> > I had installed Tornado for MPC860 and I'm using the Motorola's FADS860
> > evaluation board.
> >
> > Please note : The board is FADS860 - not ADS860.
> >
> > I burned down the bootrom.hex over the flash and find out a strange phenomena :
> >
> > When I use MPC860EN or MPC860T derivatives in the board, the boot code functions
> > as expected.
> >
> > -- BUT ---
> >
> > when I use MPC860SAR derivative - No response from the board (acctually, the
> > transmit LED of SMC 1 is turning as usuall - but no output on the terminal).
> >
> > I changed the frequency of the boot code from 4 Mhz crystal/24 Mhz system clock
> > to  5 Mhz/50 Mhz system clock and I was suprised to find out that the MPC860SAR
> > started to work !
> >
> > Any ideas ?
> >
> > Thanks in advance
> >
> > Rami Gideoni
> >
> > [WindRiver techincal support : TSR# 124446, For Mr. Gideon Chan]




---------------------------

Newsgroups: comp.os.vxworks
Subject: Proxy server for VxWorks??
Date: Thu, 11 Feb 1999 15:38:16 GMT
From: rrebo@flash.net (Rick Rebo)
Organization: FlashNet Communications, http://www.flash.net
Message-ID: <IPCw2.60$9_.172@news.flash.net>

        Does anyone know of any internet proxy server code available for 
VxWorks?   We are doing a development and would rather not write it ourselves.

---------------------------

Newsgroups: comp.os.vxworks
Subject: VxWorks internet proxy server
Date: Thu, 11 Feb 1999 16:11:28 GMT
From: rrebo@flash.net (Rick Rebo)
Organization: FlashNet Communications, http://www.flash.net
Message-ID: <QiDw2.67$9_.540@news.flash.net>

Does anyone know of a VxWorks internet proxy server that is available?  We are 
doing a development and would like to include a proxy server and would rather 
not write our own.

---------------------------

Newsgroups: comp.os.vxworks
Subject: When to recompile bootrom?
Date: Mon, 08 Feb 1999 15:29:46 GMT
From: jyl97@my-dejanews.com
Organization: Deja News - The Leader in Internet Discussion
Message-ID: <79mvsv$39e$1@nnrp1.dejanews.com>

Hi,

Can somebody tell me when to recompile bootrom (or boot.bin)? If I change some
INCLUDE_s, do I need to recompile both bootrom and vxWorks?

Thanks a lot.

- -- JYL

- -----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: problems with dosFs
Date: 11 Feb 1999 17:44:29 GMT
From: davel@zeppelin.tgf.tc.faa.gov (Dave Lonkart)
Organization: FAA Technical Center, Pomona, NJ
Message-ID: <79v4tt$jc9$1@faatcrl.faa.gov>
References: <9902119187.AA918743134@smtp.wg.com>

Is this "backup" filling up the disk ?  I have found that my disks under
vxWorks 5.2 get corrupted when the capacity is exceeded.  There are no 
warnings or errors generated by this, either.  Also, and this is a wild
guess since I know nothing about the vxWorks VM, are you using the
DOS_OPT_LONGNAMES options when creating the DOSFS volume ?  Java file names
tend to get long and have four letter extensions.  Just a guess.  Good luck.
> 
>      We have been experiencing I/O problems - they seem to have something 
>      to do with dosFs.
      
>      Brett Smith
>      brett.smith@wg.com
> 
> 

- -- 


- ---
Dave Lonkart
Raytheon Service Support Company
davel@tgf.tc.faa.gov

---------------------------

Newsgroups: comp.os.vxworks
Subject: UDP Performance again
Date: Thu, 11 Feb 1999 12:12:56 -0500
From: Tim Sohacki <sohacki@nortelnetworks.com>
Organization: Nortel (Northern Telecom)
Message-ID: <36C30F98.923393C2@nortelnetworks.com>

At first blush, measurements with Windview 2.0 suggest our target 
is taking on the order of 10 ms to send 10-500 byte UDP packets.
This is with a 200 Mhz 603 PPC, an AMD 79C972 ethernet device,
and SENS 1.1.  I realize Windview is intrusive, but 1 ms per 
datagram???

Not long ago someone complained about performance that was ten
times as fast.  Might there be something significantly wrong with
our configuration?

Also, does anyone have any network stack performance numbers
for other processors, stacks, etc.?  Is there a benchmark suite 
of any sort?   

Thanks,
Tim


- -- 
Tim Sohacki, Nortel Networks
email: sohacki@nortelnetworks.com

---------------------------

Newsgroups: comp.os.vxworks
Subject: Booting vxworks on PC (Dell Pentium II)
Date: Mon, 08 Feb 1999 19:42:08 GMT
From: raons@ornl.gov
Organization: Deja News - The Leader in Internet Discussion
Message-ID: <79neme$fsr$1@nnrp1.dejanews.com>

	I am trying to boot vxworks on a PC using dikette. But I am not able to
	get it connected to the server. The PC has 3COM Ethernet II card. I
	specified the boot device as fd. Unfortunately, the documentation is not
	helpful in specifying how to specify the network parameters. I entered
	the IP addresses of the host and target and other entries that are
	direct. I entered elt on the line "other(o)". The system boots vxworks
	but th  network commands (such as ping) do not work.

	I would greatly appreciate if some one can point me in the right
	direction (e.g. website, or archive).

	Thank you very much.

	Nagi Rao

- -----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

---------------------------

Newsgroups: comp.os.vxworks
Subject: Booting from a serial connection
Date: Thu, 11 Feb 1999 20:08:33 GMT
From: "Bernard O. Debbasch" <debbasch@ultimacom.com>
Organization: Ultima Commmunication
Message-ID: <36C338B9.2B58919B@ultimacom.com>

We have tried to boot VxWorks from a serial connection:

We changed WDB_COMM_TYPE as follows and burned a new EPROM.

#define WDB_COMM_TYPE WDB_COMM_SERIAL

And correctly set the baud rate and WDB_TTY_CHANNEL

Despite this, we never get WDB READY on the serial port.

Any clue ? Thanks.
Bernard

PS: If it helps: platform is ARM PID7T_T

---------------------------

Newsgroups: comp.os.vxworks
Subject: Raw sockets in SENS
Date: Mon, 08 Feb 1999 16:59:15 -0800
From: Pradeep Kanyar <pradeep@trinc.com>
Organization: The Internet MainStreet
Message-ID: <36BF8863.BC0B0BA7@trinc.com>

Hi,
    I have been trying to write the traceroute utility for VxWorks.

    For some time, I have been facing problems using the IPPROTO_RAW
option.
    This is what I do:
    a) create a socket with SOCK_RAW, IPPROTO_RAW
    b) use setsockopt with IP_HDRINCL to force IP to use my IP header
    c) Create my own IP header with TTL set to 1,2... and so on
    d) send the packet using sendto() to send it to the destination.

    I observe that the behaviour is slightly varied during different
tests but the
    end result is the same. The NetTask hangs. When I run my program
from
    the shell, I see the task has incurred a program fault and shows a
trace
    of calls made at the lower levels. The trace goes through
bsdSendto(),
    sosend(), rip_usrreq(), rip_output(), ip_output, ether_output()...
some more
    routines, then copyFromMbufs(), netMblkClChainFree() and some
routine
    ahead of ip_forward() for which there is symbol information.

    Does anyone have an idea why this might be happening?

Thanks in advance,
Pradeep


---------------------------

Newsgroups: comp.os.vxworks
Subject: How to change default (ethernet) interface used by bootrom_uncmp to ATM interface
Date: 11 Feb 1999 21:43:27 GMT
From: jcl@apl.washington.edu (Jim Luby)
Organization: APL-UW
Message-ID: <79vitv$8ju$1@nntp6.u.washington.edu>

Hi,
I'm running Tornado/VxWorks 5.3.1 and a Motorola PowerPC single board computer.  
My current setup includes a boot loader (bootrom_uncmp) in the SBC flash which 
in turn downloads the actual kernel (vxWorks) via ethernet upon SBC reset.  I 
would like to replace the ethernet interface with an ATM interface such that 
the kernel is downloaded via the ATM interface and not the ethernet interface.  
In the long run I want to get rid of the ethernet interface entirely although 
if it can coexist with the ATM interface initially that would be fine.  My main 
question is what routine or routines do I need to modify such that when I use 
Tornado's "make bootrom_uncmp" command my ATM driver stuff gets added and my 
ATM initialization functions get called?  I'm not sure whether I should do this 
by modifying bootconfig.c or whether there is a cleaner way that doesn't 
require me to muck around a whole lot with the baseline Wind River code.

Thanks for any hints!

Jim Luby
Applied Physics Lab, UW


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: UDP Performance again
Date: 11 Feb 1999 15:01:48 PST
From: Spero Koulouras <spero@wacos.com>
Organization: Concentric Internet Services
Message-ID: <36C36189.9E1CB64E@wacos.com>
References: <36C30F98.923393C2@nortelnetworks.com>

We have been playing with this on an x86 extensively.  Our environment
is different, but performance is still poor.

Environment:

- -  200 Mhz x86 CPU
- -  21140 Digital Ethernet device
- -  Dedicated 100 Mbps LAN
- -  SENS 1.1 and non-SENS both tested

In the non-SENS application:

- -  a 128 byte sendto() requires roughly 150 us inside of the sendto()
API call and 10 us at interrupt level
- -  receiving 128 byte packets requires about 10 us at interrupt,  then
50 us within nettask
- -  sending/receiving 4k packets adds to the application level times
(300us total time) but does not impact interrupt
- -  result:  packet rate of 220 us/packet (~ 5000 packet/second) and
transfer rate of roughly 4.8 Mbps with CPU at 100% on 100 Mbps LAN

Under SENS,  with END driver

- -  Driver has unexplained 16ms periods in which it "locks" CPU.  This
must be a bug
- -  NetTask appears to get involved in data TRANSMISSION, not just
receipt, adding 50ms latency

Numbers are pretty consistent with and without Windview instrumented in
the kernel,  there is no major benefit so we have decided to keep
windview in.

Using wvEvent() helps a lot to isolate what is going on.  We now bracket
critical sections with wvEvent() calls so that the Windview traces make
it obvious.

There are network performance numbers available from a varety of
places,  but not for vxWorks.  The biggest collection of benchmarks that
I found was at the netperf website (netperf.hp.com ?).  Additional
numbers were buried somewhere in the Linux Router project
(www.linuxrouterproject.org).  Mentat has a very good benchmark
comparing thier stack against Unixware (somewhere under
www.mentat.com).  The netperf benchmark source code is available at no
cost.

Spero

Tim Sohacki wrote:

> At first blush, measurements with Windview 2.0 suggest our target
> is taking on the order of 10 ms to send 10-500 byte UDP packets.
> This is with a 200 Mhz 603 PPC, an AMD 79C972 ethernet device,
> and SENS 1.1.  I realize Windview is intrusive, but 1 ms per
> datagram???
>
> Not long ago someone complained about performance that was ten
> times as fast.  Might there be something significantly wrong with
> our configuration?
>
> Also, does anyone have any network stack performance numbers
> for other processors, stacks, etc.?  Is there a benchmark suite
> of any sort?
>
> Thanks,
> Tim
>
> --
> Tim Sohacki, Nortel Networks
> email: sohacki@nortelnetworks.com


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: NAT implementation for VxWorks
Date: 11 Feb 1999 15:02:44 PST
From: Spero Koulouras <spero@wacos.com>
Organization: Concentric Internet Services
Message-ID: <36C361C1.C78405D9@wacos.com>
References: <79v6n6$sf53@overload.lbl.gov>

You may want to try the Linux version from www.linuxrouterproject.com.

Spero

Pritesh Patel wrote:

> Does anyone out there know of an available Network
>
> Address Translation implementation that runs under
>
> VxWorks?
>
> Thanks,
>
> Pritesh Patel
>
> patel@sentientnet.com


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: How to change default (ethernet) interface used by bootrom_uncmp to 
 ATM interface
Date: 11 Feb 1999 23:32:36 GMT
From: Jeremy & Cynthia Cunningham <cunningh@worldnet.att.net>
Organization: AT&T WorldNet Services
Message-ID: <36C368FD.EB668C70@worldnet.att.net>
References: <79vitv$8ju$1@nntp6.u.washington.edu>

Hello,
  I have a Radstone PPC board that I have on an ATM network.  The ATM drivers that
I have for my ATM card are too big to fit in the bootrom, so I had to make a kernel
that has the ATM drivers and put it into flash.  I then had to modify the bootrom
so that it booted the kernel that was in flash.  Once the kernel in flash boots, it
loads my application object module off of a UNIX host on the ATM network using
loadModule() by calling the USER_APPL macro in config.h.  This way, to test my
kernel I can boot my it from the ethernet. Once I get the kernel loading my
application correctly across the ATM, I can write it to the flash and not have to
modify it any more.  It may not be exactly what you're looking for, but that is how
I got around not being able to boot a kernel from the ATM like I can boot it from
the ethernet.

Jim Luby wrote:

> Hi,
> I'm running Tornado/VxWorks 5.3.1 and a Motorola PowerPC single board computer.
> My current setup includes a boot loader (bootrom_uncmp) in the SBC flash which
> in turn downloads the actual kernel (vxWorks) via ethernet upon SBC reset.  I
> would like to replace the ethernet interface with an ATM interface such that
> the kernel is downloaded via the ATM interface and not the ethernet interface.
> In the long run I want to get rid of the ethernet interface entirely although
> if it can coexist with the ATM interface initially that would be fine.  My main
> question is what routine or routines do I need to modify such that when I use
> Tornado's "make bootrom_uncmp" command my ATM driver stuff gets added and my
> ATM initialization functions get called?  I'm not sure whether I should do this
> by modifying bootconfig.c or whether there is a cleaner way that doesn't
> require me to muck around a whole lot with the baseline Wind River code.
>
> Thanks for any hints!
>
> Jim Luby
> Applied Physics Lab, UW


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: When to recompile bootrom?
Date: Mon, 08 Feb 1999 13:08:11 -0800
From: Stephen Macmanus <stephen@wrs.com>
Organization: Wind River Systems
Message-ID: <36BF523B.7969@wrs.com>
References: <79mvsv$39e$1@nnrp1.dejanews.com>

jyl97@my-dejanews.com wrote:
> 
> Hi,
> 
> Can somebody tell me when to recompile bootrom (or boot.bin)? If I change some
> INCLUDE_s, do I need to recompile both bootrom and vxWorks?
> 
> Thanks a lot.
> 
> -- JYL
> 
> -----------== Posted via Deja News, The Discussion Network ==----------
> http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own

     When to recompile depends on the purpose of the bootrom image. 

     It is often used only to obtain the full-size VxWorks runtime
     image. This model may be the most common case, especially during
     application development. Since the bootrom image is replaced by
     the retrieved runtime application, it is only necessary to rebuild
     it when changing configuration items related to that task (for
     example, adding or removing support for a particular storage 
     device. It is advisable to keep a backup copy of the configuration
     file if it differs from the runtime image.

     Alternatively, the bootrom could contain the entire application,
     bypassing the replacement with a runtime image. In that case, it
     would be necessary to rebuild it whenever an OS component was
     added or removed for the change to take effect.

- -- 
- ------------------
Stephen Macmanus                         #include <std_disclaimer.h>
Software Engineer
Wind River Systems                       stephenm@wrs.com

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: UDP Performance again
Date: Fri, 12 Feb 1999 03:23:25 GMT
From: netperf@my-dejanews.com
Organization: Deja News - The Leader in Internet Discussion
Message-ID: <7a06r5$5ht$1@nnrp1.dejanews.com>
References: <36C30F98.923393C2@nortelnetworks.com> <36C36189.9E1CB64E@wacos.com>


> I found was at the netperf website (netperf.hp.com ?).  Additional

try http://www.netperf.org/

rick jones

- -----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

---------------------------

Newsgroups: comp.os.vxworks
Subject: How to get IP of Gateway and own IP?
Date: Thu, 11 Feb 1999 23:28:11 -0800
From: "gto" <overney@cemvax.cem.msu.edu>
Organization: SBC Internet Services
Message-ID: <NGQw2.1665$Qf5.12593@typhoon-sf.pbi.net>

I am using VxWorks 5.3.1 on a PPC860. Once the board is up and running, I
would like to get its own IP as well as the IP of the Gateway. In VxWorks,
neither gethostid() nor gethostbyname() exists. What system calls can
provide me with this info?

Thanks,

Gregor




---------------------------

Newsgroups: comp.os.vxworks
Subject: VxWorks & Intel Ultra-Low Power 486SX Eval.
Date: Fri, 12 Feb 1999 08:40:33 GMT
From: Peter.Smith@tde.alstom.com
Organization: Deja News - The Leader in Internet Discussion
Message-ID: <7a0pe0$le2$1@nnrp1.dejanews.com>

Hi,

Has anyone out there used VxWorks with the Intel EV486ULP Evaluation Board ?

As far as I can tell this board is PC/AT compatible and should work with the
generic BSP, can anyone confirm this ?

Peter Smith
ALSTOM T&D P&C Ltd
Stafford
England

- -----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

---------------------------

End of New-News digest
**********************


From owner-vxwexplo-process  Fri Feb 12 04:17:37 1999
From: David Laight <dsl@tadpole.co.uk>
Date: Fri Feb 12 04:17:43 PST 1999
Subject: Serial line device drivers

I'm trying to decide which vxworks serial driver to use for Exar st16C2550 duart 
and st16c554D quart chips.

Anyone know any REAL differences between the ns16550Sio, st16552Sio and 
evbNs16550Sio drivers?  Or do they just do the same things with slightly 
different code.

To confuse things, the source for the st16552Sio driver is absent from the 
drivers development kit.

Maybe I should write another clone :-)

Also what is the difference between 'Sio' drivers and 'serial' drivers?

	David

----------------------------------------------------------------
David Laight                          email: dsl@tadpole.co.uk
Tadpole Technology plc                phone: +44 1223 278 256
Cambridge, UK                         fax:   +44 1223 278 201



From owner-vxwexplo-process  Fri Feb 12 07:42:11 1999
From: "Paul R. Bade" <bade@jhuapl.edu>
Date: Fri Feb 12 07:42:14 PST 1999
Subject: Re: UDP Performance Again

Tim Sohacki writes:

> At first blush, measurements with Windview 2.0 suggest our target
> is taking on the order of 10 ms to send 10-500 byte UDP packets.
> This is with a 200 Mhz 603 PPC, an AMD 79C972 ethernet device,
> and SENS 1.1.  I realize Windview is intrusive, but 1 ms per
> datagram???
>
Seems very slow to me.
While I don't recall the exact numbers, I believe that we were getting
on the
order of 8000 (200 byte msgs/sec) on a 166 MHZ 604e with a
100 Mbit/sec ethernet (probably was full duplex).
However, we had to optimize the dec21140 driver for small packets.
The original driver would post a job to clean up the transmit
descriptor ring. We changed the ISR to clean up the descriptor ring
in the ISR. It significantly helped performance. Other things that you
need to consider are the number of buffers that you have defined
in netBufLib.h. You should probably at least double the defaults.
Do you know if you are losing packets?
That is the more typical problem with UDP when you try to
send at the max rate, but have not setup your buffers
correctly or you have your task priorities incorrect.

--


+====================================================================+
|       __     ____         __     __                                |
|      /\ \   /  \ \       /\_\__ /\ \      Johns Hopkins University |
|     /  \_\ | /\ \ \     / / /\_\\ \ \     Applied Physics Lab.     |
|    / /\ | |\ \/  \ \   / / / / / \ \ \                             |
|   /  \/ |_| \  /\ \_\ / / / / /   \ \ \   Paul R. Bade             |
|  / /\__/_/   \ \ \/_// / / / /    / / /   (443)778-8681            |
| / / /         \ \_\  \ \/ / /    / / /    bade@jhuapl.edu          |
| \/_/           \/_/   \__/_/     \/_/                              |
|               __     ____         __       __                      |
|              /\ \   /  \ \       /\ \     /\ \                     |
|             /  \_\ | /\ \ \     /  \_\   /  \ \                    |
|            / /\ | |\ \/  \ \   / /\ |_| / /\ \ \                   |
|           /  \/ |_| \  /\ \_\ / / / | | \/ /  \ \                  |
|          / /\  / /   \ \ \/_// / / / /    / /\ \_\                 |
|          \ \/ / /     \ \_\  \ \/ / /     \/ / / /                 |
|           \__/_/       \/_/   \__/_/        /_/_/                  |
|                                                                    |
+====================================================================+
vxWorks




From owner-vxwexplo-process  Fri Feb 12 07:49:09 1999
From: Kevin Kernin <kernin@research.moore.com>
Date: Fri Feb 12 07:49:13 PST 1999
Subject: Stethoscope Problem

Hello vxWorks Users,

We are using Tornado 1.0.1 with vxWorks 5.3.1 on a MVME2306 board.
We recently purchased Stethoscope 5.0 and installed it in our WIND_BASE.

I am trying to run their demo but I don't have
WIND_BASE/rti/scope.5.0c/lib/ppcVx5.3.1/libutilsip.so.
I cannot run anything because I get unresolved refernces which I assume
are
in libsutilsip.so.

Is anyone using Stethoscope with the ppcVx5.3.1 architecture?
If so,  would you please let me know what I need to do.
Thanks,
Kevin

=================================
Kevin Kernin
Moore Research Center, Inc.
(716)773-0658
FAX  : (716)773-0462
email: kernin@research.moore.com
=================================



From owner-vxwexplo-process  Fri Feb 12 10:57:19 1999
From: Charlie Grames <cgrames@mdc.com>
Date: Fri Feb 12 10:57:23 PST 1999
Subject: Ordering of Memory Cycles and Related Stuff

VxWorks 5.3.1
MVME2700 1.1/4

My sincere apologies to David Laight, whose 2 February posting suggesting
the use of read-after-write I rebuked.  While the use of instructions like
"eieio" and "sync" cannot be discounted on the PowerPC processor for proper
_ordering_ of memory cycles, they do nothing to flush the FIFOs of
intermediate bridge devices that use write posting.  The read-after-write
mechanism, however, ensures that a written value is properly placed before
the read completes.

In my case, a combination of "sync" and read-after-write solved my
synchronization problems.  Thanks to David for sparking the discussion, and
thanks to Dave Machin for providing his related case history.

Charlie Grames
The Boeing Company
(314) 233-1956
Charles.R.Grames@boeing.com


From owner-vxwexplo-process  Fri Feb 12 10:58:20 1999
From: Charlie Grames <cgrames@mdc.com>
Date: Fri Feb 12 10:58:24 PST 1999
Subject: Sync Instructions on PowerPC

VxWorks 5.3.1
MVME2700 1.1/4

I have noticed both from discussions on this forum and from Wind River's
PowerPC BSPs that there is a definite tendency to group the eieio and sync
instructions together when performing operations to "ill-behaved" memory. 
Is this necessary?  From my research, it appears the the sync instruction is
a superset of eieio, thus suggesting that eieio used in conjuction with sync
is redundant.  Does anyone have any information on this?  Thanks in advance
for any feedback.

Charlie Grames
The Boeing Company
(314) 233-1956
Charles.R.Grames@boeing.com


From owner-vxwexplo-process  Fri Feb 12 13:08:05 1999
From: Jeff Hay <jrhay@lanl.gov>
Date: Fri Feb 12 13:08:08 PST 1999
Subject: sysInWordString() on pc486


I am having problems with sysInWordString() on a pc486 BSP. 
Speifically, I am reading only 0x00, no matter what port or how much I
try to read (I want 4 bytes; this seems the only way to do it...)  I
imagine I'm missing something entirely obvious, so anybody brighter then
me out there want to clue me in?  Thanks!

Tornado 1.0.1
VxWorks 5.3

-- 
Jeffrey R. Hay
Los Alamos National Laboratory
CIC-5, Network Engineering
email: jrhay@lanl.gov
phone: (505)667-2599, fax: (505)665-7793


From owner-vxwexplo-process  Sat Feb 13 04:00:41 1999
From: daemon@csg.lbl.gov
Date: Sat Feb 13 04:00:45 PST 1999
Subject: comp.os.vxworks newsdigest

Comp.Os.Vxworks Daily Digest    Sat Feb 13 04:00:38 PST 1999

        Subject: Re: How to get IP of Gateway and own IP?
        Subject: Re: Shared Memory Network problems
        Subject: Re: UDP Performance again: NOT SENT
        Subject: Re: Real Time Data Compression Algorithms ?
        Subject: Re: Booting vxworks on PC (Dell Pentium II)
        Subject: Re: How to get IP of Gateway and own IP?
        Subject: AMD ELAN SC400
        Subject: Problems Rebuilding Kernel w/ mv2700 BSP
        Subject: rommable vxworks uncompressed
        Subject: Re: compression source code
        Subject: Re: compression source code
        Subject: Re: Sharing memory between PPC and 68K fails
        Subject: WE NEED TIPS!!!

-------------------------------------------------------

Newsgroups: comp.os.vxworks
Subject: Re: How to get IP of Gateway and own IP?
Date: 12 Feb 1999 13:04:46 GMT
From: chc@nasa2.ksc.nasa.gov (Charles H. Chapman)
Organization: NASA, Kennedy Space Center
Message-ID: <slrn7c89mn.f41.chc@nasa2.ksc.nasa.gov>
References: <NGQw2.1665$Qf5.12593@typhoon-sf.pbi.net>

On Thu, 11 Feb 1999 23:28:11 -0800, gto <overney@cemvax.cem.msu.edu> wrote:
>I am using VxWorks 5.3.1 on a PPC860. Once the board is up and running, I
>would like to get its own IP as well as the IP of the Gateway. In VxWorks,
>neither gethostid() nor gethostbyname() exists. What system calls can
>provide me with this info?

You can use ifAddrGet() to get your own IP address but I'm not sure how
you could find out the address of the gateway.

Chuck

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Shared Memory Network problems
Date: Fri, 12 Feb 1999 14:45:19 +0000
From: Phil Ashby <phil.ashby@wireplay.bt.co.uk>
Organization: BT Wireplay
Message-ID: <36C43E7F.31A3@wireplay.bt.co.uk>
References: <7a1c19$488$1@nnrp1.dejanews.com>

john.collis@rrds.co.uk wrote:
> executing. The slave 177 card detects the Shared Memory Network 'heartbeat'
> when it starts to boot, a 'Loading...' message is displayed, then after some
> time an errno=0xd0003 message is displayed and boot is aborted. Whilst the
> slave is attempting to load, invoking smNetShow on the master shows that the
> Slave is there but with no activity. Has anyone in this forum used the Shared
> Memory network facility and got it working? What is errno 0xd0003?

This sort of problem can be caused by memory caches on the master board
not reflecting slave updates correctly (ie: check the cache
configuration
of the shared memory in sysLib.c), or non-functional test & set (TAS)
operations (which the heartbeat does not rely on, but the rest of the
sm driver does... check the SM_TAS_TYPE in all the board configs, ensure
that they are the same, and that all boards support the chosen type.
We normally err on the side of caution and always use SOFTware TAS (as
we run a non-homoengous network of PPC/Intel boards).
- -- 
!--------------------------- Phil "Phlash" Ashby ---------------------!
! BT Wireplay - Lead Programmer and general technical trivia supplier !
! Phone: 01473-644348                   WWW: http://www.wireplay.com/ !
! Snail: B81/G41, BT Labs, Martlesham Heath, Ipswich IP5 3RE, England.!

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: UDP Performance again: NOT SENT
Date: Fri, 12 Feb 1999 16:53:28 +0000
From: Ian Love <ijl@wireplay.bt.co.uk>
Organization: BT Labs - but speaking only for myself
Message-ID: <36C45C88.4749@wireplay.bt.co.uk>
References: <36C30F98.923393C2@nortelnetworks.com>

Tim Sohacki <sohacki@nortelnetworks.com>

ijlTim Sohacki wrote:
> 
> At first blush, measurements with Windview 2.0 suggest our target
> is taking on the order of 10 ms to send 10-500 byte UDP packets.
> This is with a 200 Mhz 603 PPC, an AMD 79C972 ethernet device,
> and SENS 1.1.  I realize Windview is intrusive, but 1 ms per
> datagram???
> 
> Not long ago someone complained about performance that was ten
> times as fast.  Might there be something significantly wrong with
> our configuration?
> 
> Also, does anyone have any network stack performance numbers
> for other processors, stacks, etc.?  Is there a benchmark suite
> of any sort?

We have MVME2305 (300Mhz MOC604) with DEC  21140. We run 100Base-T
full duplex. 

We do NOT yet run SENS - doesn't seem to be that stable from what I've
heard here (and from very limited experience: install, crashed boards
instantly, uninstalled....), and recently WRS engineers couldn't even 
get it to build a generic kernel for my BSP from a fresh install from 
a recent CD....

We do _NOT_ use windview as we have found that it can seriously
interfere
with the network performance of vxWorks (especially if you are
connecting windview over the network!). For the same reason we do not
use windSh either.....

I have a test suite that is geared to measuring individual round-trip
(latency) of packets bouncing between 2 of the above boards (or any
other board under test). Basically a number (user selectable) of TCP
streams are opened, and then 'tick-stamped' packets are sent over each
stream, bounced back down the stream by the other end and then the
'tick' in the packet compared with the current processor tick.

Thus we measure the ENTIRE time that the packet has spent in the network
and being bounced - to the nearest tick. We run our clock at 1000Hz,
thus we get timing to the nearest millisecond.

Our normal test is loads of streams running fairly small packets, but
for the purposes of investigating what you reported I did the following:

on 100 streams, send 1000 packets of 500 bytes size at 100ms intervals
	: that equates to 1000 500 byte packets per second 
		- ie the same as what you are doing

		Result:most packets arrived back within the same tick
			a few packets arrived 'in the next tick'
			processor running 84% idle

Implication: it does not take 1ms to send the packet: for starters
it's been sent twice- and received - within 1ms, but also the processor
was doing nothing most of the time......

But: Although theoretically I should be able to do ten times that
much (according to processor load AND network bandwidth), I find in
practise that I can't. 2 things spring to mind:
	1) Interrupt handling/delivery   ie driver problem
	2) TCP stack problem
		(I already have a TSR open with WR regarding occasional
			300ms 'hang' condition on individual streams,
			so it is certainly on the cards that there is
			something else wrong as well)

I tried 1000 byte packets with the same configuration as above, and 
it worked - sort of, although latency was higher and I began to see
latency spikes as per my TSR.

500 byte packets on 150 streams was the most telling: 

regards
Ian Love,

The views expressed above are my own and not those of my employer

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Real Time Data Compression Algorithms ?
Date: Fri, 12 Feb 1999 16:32:05 +0000
From: octec <gordy@octec.co.uk>
Organization: Octec Ltd
Message-ID: <TS23UGAFeFx2Ew6G@octec.co.uk>
References: <918761614.12412.0.nnrp-10.9e98f04e@news.demon.co.uk>

In article <918761614.12412.0.nnrp-10.9e98f04e@news.demon.co.uk>,
Systems <systems@mass.co.uk> writes
>We want to perform lossless compression of  binary data which is arriving at
>about 50MB/sec.
>Does anyone know of any algorithms/source code which would help us ?
>
>Allan MacDonald
>Mass Consultants Ltd.
>amcdonald@mass.co.uk

Ask in comp.compression

A>


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Booting vxworks on PC (Dell Pentium II)
Date: Tue, 09 Feb 1999 08:29:31 -0600
From: Frank Minich <fminich@ti.com>
Organization: NCP
Message-ID: <36C0464B.3D39@ti.com>
References: <79neme$fsr$1@nnrp1.dejanews.com> <36C04242.7B1A@ti.com>
Reply-To: fminich@ti.com

Correction to my previous post:

> >         The PC has 3COM Ethernet II card.
> 
Stop the boot sequence during the countdown, then "help" to see
what boot devices are available.  The "elt" device supports the
3C509 (an older ISA card, also known as the "EtherLink III").
The "EtherLink II" is a predecessor to the 509 card.  
I don't know whether or not the "elt" driver will work with the
EtherLink II.

Good luck,
Frank Minich
fminich@ti.com

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: How to get IP of Gateway and own IP?
Date: Fri, 12 Feb 1999 10:44:22 -0700
From: Mike Stimpson <mstimp@utsci.com>
Organization: Utah Scientific, Inc.
Message-ID: <36C46876.41C6@utsci.com>
References: <NGQw2.1665$Qf5.12593@typhoon-sf.pbi.net> <slrn7c89mn.f41.chc@nasa2.ksc.nasa.gov>

Charles H. Chapman wrote:
> 
> On Thu, 11 Feb 1999 23:28:11 -0800, gto <overney@cemvax.cem.msu.edu> wrote:
> >I am using VxWorks 5.3.1 on a PPC860. Once the board is up and running, I
> >would like to get its own IP as well as the IP of the Gateway. In VxWorks,
> >neither gethostid() nor gethostbyname() exists. What system calls can
> >provide me with this info?
> 
> You can use ifAddrGet() to get your own IP address but I'm not sure how
> you could find out the address of the gateway.
> 
> Chuck

If the bootline contains the *current* gateway address, you can do 

BOOT_PARAMS bootParams;
bootStringToStruct(BOOTLINE_ADDRESS, &bootParams)

and then the gateway address is in bootParams.gad (for this to work, if
you *change* the gateway address, you have to also store the new one in
the bootline...)

Hope that helps.
- -- 
Mike Stimpson (mstimp@utsci.com)

Opinions expressed are not necessarily those of the management...

The more I work as an engineer, the more I disbelieve the theory of
evolution.

---------------------------

Newsgroups: comp.os.vxworks
Subject: AMD ELAN SC400
Date: Fri, 12 Feb 1999 19:18:26 +0000
From: Peter Smith <Peter.Smith@salerio.demon.co.uk>
Message-ID: <36C47E82.9350A222@salerio.demon.co.uk>

Can anyone give me any tips about using VxWorks on the AMD ELAN
SC400/SC410 Evaluation board ? Does the standard pc486 BSP work out of
the box or do I need to get the AMD BSP ? As far as I can see, the AMD
BSP is rather old and may not support the latest version of VxWorks. On
another subject, what Ethernet cards's does the latest version of the
generec pc BSP support, and are END drivers supplied. Thanks.

Peter Smith

---------------------------

Newsgroups: comp.os.vxworks
Subject: Problems Rebuilding Kernel w/ mv2700 BSP
Date: Tue, 09 Feb 1999 19:21:04 GMT
From: rhead@virtc.com
Organization: Deja News - The Leader in Internet Discussion
Message-ID: <79q1qj$ner$1@nnrp1.dejanews.com>

Dear vxWorks Gurus,

I'm having a problem rebuilding the kernel "out of the
box" (i.e., with no changes to the standard set of included options) in the
following configuration.

VxWorks 5.3.1
Motorola MVME2700 (MPC 604e) BSP
Relase 1.1/5
Tornado 1.0.1 on Solaris

I've tried rebuilding using the
WindCFG applet and by running the makefiles from the command line, with the
same result:

ccppc -B/home/tornado/host/sun4-solaris2/lib/gcc-lib/
- -mstrict-align -ansi -nostdinc -O2 -fvolatile -fno-builtin -fno-for-scope
- -Wall -I/h  -I. -I/home/rhead/vxworks/target/config/all
- -I/home/rhead/vxworks/target/h -I/home/rhead/vxworks/target/src/config
- -I/home/rhead/vxworks/target/src/drv -DCPU=PPC604  -DMV2700
- -DTARGET_DIR="\"mv2700\""  -c sysLib.c
sysLib.c: In function `sysHwInit':
sysLib.c:864: `_func_vxMemProbeHook' undeclared (first use this function)
sysLib.c:864: (Each undeclared identifier is reported only once
sysLib.c:864:
for each function it appears in.)
sysLib.c: In function `sysProbeErrClr':
sysLib.c:2241: warning: implicit declaration of function `vxDarSet'
sysLib.c:2242: warning: implicit declaration of function `vxDsisrSet'
sysLib.c:2243: warning: implicit declaration of function `vxSrr0Set'
sysLib.c:2244: warning: implicit declaration of function `vxSrr1Set'
sysLib.c: In function `sysBusProbe':
sysLib.c:2473: warning: implicit
declaration of function `vxMemArchProbe'
make: *** [sysLib.o] Error 1
Press
Return to Quit

I did a grep for "_func_vxMemProbeHook" in the entire Tornado
directory hierarchy and couldn't find any declarations for this symbol --
only usages in "sysLib.c" and "symTbl.c".

I'm beginning to suspect that this
BSP isn't compatible with Tornado 1.0.1, and that I need to get the mv2604
BSP instead. (We actually have an mvme2604 board but got sent the mv2700 BSP
for some reason -- it seemed to be working fine until I tried to rebuild the
kernel.) Here's why:

1. In the release notes for Tornado 1.0.1 (section
8.1), mv2604 is listed as a "BSP Supported for Tornado 1.0.1", but not
mv2700.

2. A while back in this newsgroup, someone reported this same
compile error after he attempted to turn a mv2604 BSP into a mv2700 BSP by
applying some kind of patch from Motorola. "sysLib.c" was one of the files
that changed.

Can anyone confirm whether or not the mv2700 BSP is supported
by Tornado 1.0.1? (I've put in a TSR to Wind Rivers but haven't received a
reply.) As I said, I have not made any changes to the source or the included
options; I am simply trying to rebuild the kernel as it comes "out of the
box".

Thanks very much,
Robert V. Head
Software Engineer, Virtual Technology
Corporation

- -----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

---------------------------

Newsgroups: comp.os.vxworks
Subject: rommable vxworks uncompressed
Date: Fri, 12 Feb 1999 21:19:50 -0500
From: "David Garner" <daveg@progeny.net>
Message-ID: <7a2nik$90g$1@winter.news.rcn.net>

Hi all,
I have been compiling vxWorks.st_rom for quite some time and everything is
working fine.  I am now trying to compile vxWorks.res_rom_res_low (for the
MV2604 PowerPC) to allow the VxWorks Text to remain in rom.  When I compile
& link I get the following error during the link step:

"Seg1: Expected load address 0xfff00100 but file address is 0x0013931c"

The following is set in config.h

#define RAM_LOW_ADRS 0x100000
#define RAM_HIGH_ADRS 0x500000
#ROM_TEXT_ADRS 0xfff00100

I did not alter any settings to make the new image; just specified
vxWorks.res_rom_res_low as the rule instead of vxWorks.st_rom.

Can someone please tell me what I'm doing wrong.

Thanks,
Dave Garner

vxworks tornado



---------------------------

Newsgroups: comp.arch.embedded,comp.os.vxworks,comp.realtime
Subject: Re: compression source code
Date: Wed, 10 Feb 1999 14:10:21 -0500
From: srh@iname.killspam.com (Stephen R Holmes)
Organization: TNS, Inc
Message-ID: <srh-1002991410210001@user-37ka9lc.dialup.mindspring.com>
References: <79s8hi$3r2$1@news.netway.at>

In article <79s8hi$3r2$1@news.netway.at>, "mash" <cybermash@gmx.net> wrote:

> Does anyone has a source code for ASCII Data compression or decompression ??

Since 'gzip' is distributed under the GNU public license, I believe, you
ought to be able to find its source on the 'net; a good place to start
might be:

ftp://prep.ai.mit.edu/pub/gnu/

/s.holmes
- -- 
Stephen R. Holmes   | srh@iname.killspam.com
TNS, Inc.           |   (remove the middle part of the domain name
Reston, VA  USA     |    to form my actual e-mail address)

---------------------------

Newsgroups: comp.arch.embedded,comp.os.vxworks,comp.realtime
Subject: Re: compression source code
Date: Sat, 13 Feb 1999 05:35:39 GMT
From: Kevin@Quitt.net (Kevin D. Quitt)
Organization: If was organized, would I be reading Net News?
Message-ID: <36cb0ee6.907923124@news.pacificnet.net>
References: <79s8hi$3r2$1@news.netway.at>
Reply-To: Kevin@Quitt.net

You might also want to look at BZIP2.  It's a little unstable, in the sense that
while it usually does a terrific job, sometimes it does a lousy job.  But when
it's on, it'll squeeze files to 50% of what gzip can do.

- -- 
#include <standard.disclaimer>
 _
Kevin D Quitt  USA 91351-4454           96.37% of all statistics are made up
Per the FCA, this email address may not be added to any commercial mail list

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Sharing memory between PPC and 68K fails
Date: Sat, 13 Feb 1999 10:03:18 +0100
From: UFranke@t-online.de (Ulf Franke)
Message-ID: <36C53FD6.7AEE@t-online.de>
References: <36C4533F.1C9E@lin.foa.se>
Reply-To: ufranke@t-online.de

John Andersson wrote:
> 
> I'm trying to use shared memory semaphores on a MVME2700 and a MVME177.
> The 177 is processor #0. This is what happens on the PPC in a windsh:
> 
> -> semBSmCreate(0, 0)
> value = 16411921 = 0xfa6d11
> -> semShow(0xfa6d11, 0)
> can't read "offset(SM_MEM,objType)": no such element in array
> 
> The semaphore does however show up on the 68k. The above also works in
> a C-program. Why doesn't it work in a shell?
> 
> If the shell was the only problem everything would be fine, but it
> isn't.
> If I do semTake on the PPC and semGive on the 68k, semGive returns with
> an
> error, but errno is not set. semTake never returns. semShow on both 68k
> and
> PPC shows that the semaphore has been given, but the task hanging in
> semTake
> never gets this.
> 
> Is this a bug or have I missed something?
> 
> /John

Hello John,

the PPC CPU must be processor #0 to export shared memory.

See "VxWorks Programmers Guide" chapter PPC.

Best regards

Ulf

=================
Ulf.Franke@men.de

---------------------------

Newsgroups: comp.os.vxworks
Subject: WE NEED TIPS!!!
Date: Tue, 09 Feb 1999 07:20:26 GMT
From: joenav@navosoft.com
Organization: Email Platinum v.3.1b
Message-ID: <_kRv2.9087$V55.1643@news.flash.net>

We are starting a new discussion forum and need your tips. Do you have a tip you can share with us concerning problems with your computer, printer, fax machine or copier. Please visit www.navosoft.com discussion and post your tips under the correct category.


---------------------------

End of New-News digest
**********************


From owner-vxwexplo-process  Sat Feb 13 23:39:54 1999
From: Adi Molkho <adi@EnsembleCom.co.il>
Date: Sat Feb 13 23:39:57 PST 1999
Subject: PowerQuick-II library for vxWorks

Hi
We are using a PowerQuick-II(8260) as our processor and as Windriver
will have a special library for this processor 
only in mid May (Beta version) we are trying to find another solution.
The PowerQuick-II is a little different from the
860 and from the 603 so we can not use either of those.
Has some one faced or still facing a similar problem. Every tip will
help
The 

Adi Molkho
Ensemble Communications
Tel : 972-3-9027924 ext 204
Fax : 972-3-9027944



From owner-vxwexplo-process  Sun Feb 14 04:00:32 1999
From: daemon@csg.lbl.gov
Date: Sun Feb 14 04:00:36 PST 1999
Subject: comp.os.vxworks newsdigest

Comp.Os.Vxworks Daily Digest    Sun Feb 14 04:00:29 PST 1999

        Subject: compression source code
        Subject: Re: compression source code
        Subject: Re: compression source code
        Subject: Re: compression source code
        Subject: Re: Survey of peripherals used under VxWorks
        Subject: re: Tornado 1.0.1 - ARM - NT
        Subject: Principal Software Engineering Position

-------------------------------------------------------

Newsgroups: comp.arch.embedded,comp.os.vxworks,comp.realtime
Subject: compression source code
Date: Wed, 10 Feb 1999 16:25:54 +0100
From: "mash" <cybermash@gmx.net>
Organization: NETWAY AG
Message-ID: <79s8hi$3r2$1@news.netway.at>

Does anyone has a source code for ASCII Data compression or decompression ??

Thanx for any help
mash



---------------------------

Newsgroups: comp.arch.embedded,comp.os.vxworks,comp.realtime
Subject: Re: compression source code
Date: Sat, 13 Feb 1999 09:06:33 -0500
From: "James A. Littlefield" <jal@alum.mit.edu>
Organization: Shore.Net/Eco Software, Inc; (info@shore.net)
Message-ID: <36C586E9.F81E7603@alum.mit.edu>
References: <79s8hi$3r2$1@news.netway.at>

Also look at the link mentioned under the inflate() vxworks manpage.   This
points to zlib which I have found works well and can be adapted to do useful
things on a target.
Jim


mash wrote:

> Does anyone has a source code for ASCII Data compression or decompression ??
>
> Thanx for any help
> mash




---------------------------

Newsgroups: comp.arch.embedded,comp.os.vxworks,comp.realtime
Subject: Re: compression source code
Date: 13 Feb 1999 15:12:42 GMT
From: dsf@frontiernet.net (Dan Foster)
Organization: Frontier Globalcenter
Message-ID: <7a44pa$1rja$1@node17.cwnet.frontiernet.net>
References: <79s8hi$3r2$1@news.netway.at> <36cb0ee6.907923124@news.pacificnet.net>

In article <36cb0ee6.907923124@news.pacificnet.net>,
Kevin D. Quitt <Kevin@Quitt.net> wrote:
>You might also want to look at BZIP2.  It's a little unstable, in the sense tha
>while it usually does a terrific job, sometimes it does a lousy job.  But when
>it's on, it'll squeeze files to 50% of what gzip can do.

Actually, my understanding of BZIP2 is that it usually can compress about 20%
more than gzip for binary data (and data, in general) but at the expense of
incurring much more CPU overhead. This may be important to someone who needs
a deterministic way of knowing how much time their code will need. Generally
not a problem for high speed desktop systems and folks who downloads huge
files over a slow modem link, but may not be as appealing in other situations
unless you're really short on RAM or storage space and am willing to trade a
little CPU power for it.

- -Dan

---------------------------

Newsgroups: comp.arch.embedded,comp.os.vxworks,comp.realtime
Subject: Re: compression source code
Date: Sun, 14 Feb 1999 02:33:11 GMT
From: Kevin@Quitt.net (Kevin D. Quitt)
Organization: If was organized, would I be reading Net News?
Message-ID: <36d83522.983308052@news.pacificnet.net>
References: <79s8hi$3r2$1@news.netway.at> <36cb0ee6.907923124@news.pacificnet.net> <7a44pa$1rja$1@node17.cwnet.frontiernet.net>
Reply-To: Kevin@Quitt.net

On 13 Feb 1999 15:12:42 GMT, dsf@frontiernet.net (Dan Foster) wrote:
>Actually, my understanding of BZIP2 is that it usually can compress about 20%
>more than gzip for binary data (and data, in general) but at the expense of
>incurring much more CPU overhead.

Twenty percent is a more usual number, but I see the 50% figure (or close to it)
with some regularity.  Normally I compress data for three reasons: to increase
what I can carry with me, to reduce transmission times of large files, and for
large (collections of) files of archival material that I want to keep available
but may not touch for long periods of time.

- -- 
#include <standard.disclaimer>
 _
Kevin D Quitt  USA 91351-4454           96.37% of all statistics are made up
Per the FCA, this email address may not be added to any commercial mail list

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Survey of peripherals used under VxWorks
Date: Sun, 14 Feb 1999 09:11:26 +0300
From: Leonid Rosenboim <leonid@bitband.com>
Organization: BitBand Technologies Ltd.
Message-ID: <36C6690E.22A4C88@bitband.com>
References: <36C470B9.2924CEF8@jhuapl.edu>


Andy Helten wrote:

> I am interested in any information about peripherals that are being used
> under VxWorks.  For instance, we currently use Exabyte 8mm SCSI tape
> drives, but are interested in finding a replacement for these somewhat
> unreliable drives.  It doesn't have to be a SCSI peripheral -- USB,
>

Andy, Exabyte 8mm tapes are as reliable as tape drives can be these days.
With current technology and bit densities which are incredible, reliability
is naturally worse then in the 9-track tape era, but most people live with
that
just fine.

> FibreChannel, Ethernet, Firewire, etc. -- are all possibilities.  I am
> really just interested in knowing what is out there and any
> personal/professional opinions on what is available.
>
> Thanks,
> Andy Helten

The alternatives are other tapes, e.g. DAT, ,Magneto-Optical (e.g.
Maxoptics) and
magnetical-removable (e.g. Zip/Jazz). Of these I would vote for the MOs as
being
the most rugged and easy to maintain. Still you may want to address the
current
tape reliability issues by contacting Exabyte, they are usually very
helpful.
It may be as simple as cleaning your drive regularly...

Leonid


---------------------------

Newsgroups: comp.os.vxworks
Subject: re: Tornado 1.0.1 - ARM - NT
Date: Sun, 14 Feb 1999 07:24:07 GMT
From: johnchann@home.net (johnchann)
Organization: @Home Network
Message-ID: <rSux2.1260$yv3.841@news.rdc1.sfba.home.com>
References: <199902102014.MAA08732@fensende.com>

IIRC this is just a junk warning because some empty segments have been 
included.

JRC

In article <199902102014.MAA08732@fensende.com>, Mike Cuddy 
<mcuddy@fensende.com> wrote:
>
>> Can anyone explain why compiling with the '-g' (debug) option, the resulting
>> arm-coff file contains .stab and .stabstr sections that the Tornado tools
> have
>> trouble with?  I get the many of the following warning from the Target
> Server:
>> 
>>       Warning: Symbol .stabstr from section of type STYP_REG. Ignored.
>> 
>> It seems the Debugger generates similar warnings along with many anoying
>> dialog windows that need to be 'Ok'ed away.
>
>I got this error, too.  The only way that I found around it was to 
>use strip to remove the .stab and .stabstr sections.
>
>(on the other hand, I didn't try real hard to get around it, either ;-)
>

---------------------------

Newsgroups: comp.os.vxworks
Subject: Principal Software Engineering Position
Date: Wed, 10 Feb 1999 15:00:50 GMT
From: "Robertson-Surrette" <ottawa@robsur.com>
Organization: Robertson-Surrette Inc.
Message-ID: <01be551f$82acea70$c6d0ba89@istar>

Communications Systems

Principal Software Engineer

If the prospect of applying leading-edge commercial voice and data
communications technology to highly reliable, ruggedized and muti-level
reduncy based military products sounds challenging, read on:

On behalf of one of the Ottawa region's most innovative system solutions
providers, we are currently conducting a search for a Principal Engineer,
Software Development.  This organization has established a worldwide
reputation for designing and manufacturing state-of-the-art advanced
communications products.  As part of an advanced development team, this
individual will play a key role in defining the "next generation" of the
company's leading-edge system.

To be ideally suited for this role, you have a strong communications and
telephony background with extensive ATM and ISDN design experience.  Your
network management expertise will serve you well in designing multimedia
delivery capabilities using state-of-the-art technologies such as JAVA, for
use in very specialized military environments.  Experience with ADA
Programming and/or VxWorks will be considered an assets.

You possess a bachelor's degree in either Computer Science or Engineering. 
Having worked in the technology field for at least 7 to 8 years, you are
ready for a challenging role where a small team setting fosters mentoring
and learning.  Here, every team member understands the difference his or
her contribution makes to the product.  The position requires an individual
who loves to be challenged.

If you, or someone you know is interested in pursuing this unique
opportunity further, please contact Jim Harmon at 'jharmon@robsur.com' or
by fax at (613) 749-9599.  We would also welcome your call at (613)
749-9909.  You can find more information about this role on our web site at
'www.robsur.com'.


---------------------------

End of New-News digest
**********************


From owner-vxwexplo-process  Sun Feb 14 14:32:19 1999
From: "Keith Buchanan" <Keith_Buchanan@SPARTA.com>
Date: Sun Feb 14 14:32:23 PST 1999
Subject: RE: comp.os.vxworks newsdigest

> We do _NOT_ use windview as we have found that it can seriously
> interfere
> with the network performance of vxWorks (especially if you are
> connecting windview over the network!). For the same reason we do not
> use windSh either.....

All Tornado tools consume some network bandwidth, after all nothing comes
for free.  How and when they affect you application timing can be mitigated
by altering the relative priority of the tasks that handle the WDB traffic.
WindView is not the best tool to measure network bandwidth.  Measuring
throughput is easily done, just load up a generic blaster/blastee test
program and let it rip.  Measuring the response time of individual packets
is done best by a tool lke the HP Internet Advisor.

> Thus we measure the ENTIRE time that the packet has spent in the network
> and being bounced - to the nearest tick. We run our clock at 1000Hz,
> thus we get timing to the nearest millisecond.

Big mistake!!  Do you realize that you have increased the time spent in the
kernel by an order of magnitude?  This alone will degrade measurements of
any kind.  Try using your auxiliary clock.

> Implication: it does not take 1ms to send the packet: for starters
> it's been sent twice- and received - within 1ms, but also the processor
> was doing nothing most of the time......

On a lowly 100Mhz 486, I have tuned an application that spends < 700us
processing a TCP segment, parsing an application command, performing the
work requested, preparing a response and sending a TCP segment in response.
I estimate that less than 250us of that time was spent in the network stack.
When processing as many remote commands as possible, the < 700us timing was
still true for over 90% of all commands!!

In addition, using blaster/blastee programs off the net I measured TCP
throughput at over 850KBytes/s.  Blasting datagrams easily saturated the
bandwidth.  The big secret to increasing max throughput is make the socket
buffers larger.

> But: Although theoretically I should be able to do ten times that
> much (according to processor load AND network bandwidth), I find in
> practise that I can't. 2 things spring to mind:
> 	1) Interrupt handling/delivery   ie driver problem

If you lower the clock rate back to 100Hz, use large socket buffers (32K+)
and enable the TCP_NODELAY socket option and still get the numbers you
mnetioned above, something is wrong with the driver and/or your network
media.

> 	2) TCP stack problem
> 		(I already have a TSR open with WR regarding occasional
> 			300ms 'hang' condition on individual streams,
> 			so it is certainly on the cards that there is
> 			something else wrong as well)

TCP uses a 2Hz timer for all of the timeout and retransmission algorithms.
Your 300ms could be attributable to a variety of events.  Are you sure that
there are no problems on your network? Do you have the TCP_NODELAY option
set to defeat the Nagle algorithm?

Moral of the story... if you cannot saturate a 10Base network with any
VxWorks target faster than a 33Mhz 68020, something is wrong with your
hardware, network media, application code, test tools or all of the above.
Any diagnostic tool which also uses the network will affect network
bandwidth measurements, sometimes drastically.


HTH
------------------------------------
      /\       Thomas Keith Buchanan
     /**\      Principal Engineer
    /****\     SPARTA, Inc.
   /****/..    205 East IH 30
  /****/....   Rockwall, TX 75087
 .\****\__...
...\**\  /.... Voice: 972 772 4487
....\**\/\.... FAX:   972 722 3201
 ....\****\..  PCS:   972 672 9657
  ....\****/   WWW:   www.sparta.com
   .../***/
    ./***/     SPARTA
     \**/      ======
      \/       Pride In Performance



From owner-vxwexplo-process  Mon Feb 15 01:21:59 1999
From: Parag Badge <parag@cmcdak.cmc.stph.net>
Date: Mon Feb 15 01:22:03 PST 1999
Subject: Adding net driver to VxWorks

This is a multi-part message in MIME format.
--------------2BC76383FDF6C13B64EA4600
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

Hi,
    I have new network card. I want to write driver for it
and install it in VxWorks. Can anybody tell me how to add
new drivers in VxWorks?
Thanx
Parag

--------------2BC76383FDF6C13B64EA4600
Content-Type: text/x-vcard; charset=us-ascii;
 name="parag.vcf"
Content-Transfer-Encoding: 7bit
Content-Description: Card for Parag Badge
Content-Disposition: attachment;
 filename="parag.vcf"

begin:vcard 
n:Badge;Parag
tel;fax:91-040-3000509
tel;home:91-040-7833615 (pp)
tel;work:91-040-3000401/501 (x2259)
x-mozilla-html:FALSE
url:www.cmcltd.com
org:CMC Ltd;Real Time Systems
version:2.1
email;internet:parag@cmcltd.com
adr;quoted-printable:;;CMC Centre,=0D=0AOld Mumbai Highway,=0D=0AGachibowli,;Hyderabad,;Andhra Pradesh,;500019;INDIA
fn:Parag R.Badge
end:vcard

--------------2BC76383FDF6C13B64EA4600--



From owner-vxwexplo-process  Mon Feb 15 04:00:18 1999
From: daemon@csg.lbl.gov
Date: Mon Feb 15 04:00:22 PST 1999
Subject: comp.os.vxworks newsdigest

Comp.Os.Vxworks Daily Digest    Mon Feb 15 04:00:15 PST 1999

        Subject: TFFS compatibility
        Subject: Justifying new Development tools
        Subject: Re: semaphore information
        Subject: MVME162-533 available
        Subject: Real Time Data Compression Algorithms ?
        Subject: IEEE 1394 drivers available?
        Subject: Re: Tornado 1.0.1 - ARM - NT
        Subject: Re: problems with dosFs
        Subject: semaphore information
        Subject: Re: TFTP question
        Subject: Re: Survey of peripherals used under VxWorks
        Subject: Re: semaphore information
        Subject: Re: semaphore information
        Subject: Re: Sharing memory between PPC and 68K fails
        Subject: Re: VxWorks port to Linux survey
        Subject: How to get address of second IP interface ?

-------------------------------------------------------

Newsgroups: comp.os.vxworks
Subject: TFFS compatibility
Date: Wed, 10 Feb 1999 16:58:54 GMT
From: john@kivala.com
Organization: Kivala Systems, Inc.
Message-ID: <79sds7$pvk$1@nnrp1.dejanews.com>

Hello!

Does anyone have any good or bad stories to tell about
compatibility between Wind River's TFFS and the one
you get from M-Systems for a PC? I understand they came
from the same source, but the WR one needed substantial
fixing to get it working under VxWorks.

What I wanna do is stick a flash card in my PC, write
some files to it, stick it in the VxWorks system, read
the files and write new ones, then go back to the PC.
Formatting will be done on the PC, although it would be
interesting to know if there are problems doing it the
other way.

If you can answer with a simple "it works", then email
me and I'll summarize a vote count to the group.

Thanks!
John

- --------------------------------------------------
 John Finley                 Kivala Systems, Inc.
 Project Manager                   (619) 689-0032
 john@kivala.com            http://www.kivala.com
- --------------------------------------------------
 VxTool - Build Tornado Apps with MS Visual Basic
- --------------------------------------------------

- -----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

---------------------------

Newsgroups: comp.os.vxworks
Subject: Justifying new Development tools
Date: 14 Feb 1999 21:34:48 GMT
From: "Eric S. Riseman" <eriseman@tiac.net>
Organization: The Internet Access Company, Inc.
Message-ID: <01be5861$45995020$32f23ccf@LittleFoot>

Hello out there.....

I am trying to put together some statistics that I can use for an ROI
(Return On Investment).

This ROI is needed so that I can justify to management, the need to
purchase a cots Rtos as well as 20th Century development tools.

I am curious as to feedback on the following items 
1) Typical cost to a company to fix a bug,  AFTER the product has shipped
(FCS) ?
- ---bugs found after the fact might be hard to find ones like,  Stack
Overflows, Memory Leaks, Array Bound Reads/Writes, Critical paths never
executed - but should have been, etc.

2) Will the cost of the bug (engineers time to fix it, patch it, redeploy
it) vary depending upon the TYPE of bug found ?

3) What is the cost of memory ?
ie  using an RTOS that has a small footprint 30kb VS a large footprint
150kb.

4) How do you place a dollar amount on "fast/efficient" code ?
ie using a compiler that has a high degree of optimizations VS one that
does not.

Did I miss any other components/pieces that could be used to create an ROI
?

Pls respond to eriseman@intrinsa.com.

Thanks for your input and opinions.
Eric

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: semaphore information
Date: Thu, 11 Feb 1999 11:34:31 -0700
From: Mike Stimpson <mstimp@utsci.com>
Organization: Utah Scientific, Inc.
Message-ID: <36C322B7.41C6@utsci.com>
References: <36C2F747.B6374CF9@stratus.com>

Wendy McNaughton wrote:
> 
> Does anyone know if there is a way for a task to detect if it holds a
> semaphore?  There is a function available (semInfo) that tells me if
> there are other tasks waiting on a semaphore but it doesn't indicate who
> may be holding a semaphore.  I have a function where I want to detect if
> the caller is holding a particular mutual exclusion semaphore but I
> can't find the means to do this.
> 
> Does anyone know if this can be done?

Well, one way to do this is to simply have the task use a variable to
keep track of whether it holds the semaphore or not...

If the function can be called within more than one task, you could have
each task use the same global variable to store the task ID of the task
that takes the semaphore, and have the function compare that variable to
the task ID of the running task (there is a timing window, though, if a
task switch happens just as the function reads the variable...)
- -- 
Mike Stimpson (mstimp@utsci.com)

Opinions expressed are not necessarily those of the management...

The more I work as an engineer, the more I disbelieve the theory of
evolution.

---------------------------

Newsgroups: comp.os.vxworks
Subject: MVME162-533 available
Date: Wed, 10 Feb 1999 21:56:02 -0700
From: Marty Norton <Marty@innovative-research.com>
Organization: Innovative Research Technologies
Message-ID: <36C262E2.4EF15378@innovative-research.com>
Reply-To: Marty@innovative-research.com

We have a Motorola MVME162-533 and MVME162-02
board available.  E-mail if interested.





---------------------------

Newsgroups: comp.os.vxworks
Subject: Real Time Data Compression Algorithms ?
Date: Thu, 11 Feb 1999 19:28:56 -0000
From: "Systems" <systems@mass.co.uk>
Message-ID: <918761614.12412.0.nnrp-10.9e98f04e@news.demon.co.uk>

We want to perform lossless compression of  binary data which is arriving at
about 50MB/sec.
Does anyone know of any algorithms/source code which would help us ?

Allan MacDonald
Mass Consultants Ltd.
amcdonald@mass.co.uk



---------------------------

Newsgroups: comp.os.vxworks
Subject: IEEE 1394 drivers available?
Date: Fri, 12 Feb 1999 08:45:43 +0100
From: "J. Jansen" <jj@ekf.de>
Organization: EKF Elektronik GmbH
Message-ID: <36C3DC27.4921D633@ekf.de>

Does anybody know if there are VxWorks drivers and other software
available for the TI PCILynx TSB12LV21A IEEE 1394 link layer
controller? We see the need to integrate our CompactPCI 
FireWire Board (http://www.ekf.de/c/cfiw/cf1/cf1_e.html)
into a VxWorks environment.

Thanks to all - J. Jansen

- -- 
EKF Elektronik GmbH Industrial Computers & Information Technology
Joachim Jansen         mailto:jj@ekf.de         http://www.ekf.de
Philipp Reis Str. 4                      fax: +49 (0)2381-6890-90
D-59065 Hamm (Germany)                  phone: +49 (0)2381-6890-0

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Tornado 1.0.1 - ARM - NT
Date: Sun, 14 Feb 1999 22:36:17 -0500
From: "Patrick Drolet" <patrick.drolet@videotron.ca>
Message-ID: <XxMx2.715$qb4.587@weber.videotron.net>
References: <199902102014.MAA08732@fensende.com> <rSux2.1260$yv3.841@news.rdc1.sfba.home.com>

One way to get rid of the .stabstr warning in the debugger (they'll remain
in the target server) is to declare a label .stabstr in a file (I use
sysalib.s).  You won't get the anoying popups afterwards...

    .glob    .stabstr

.stabstr:

Patrick.

johnchann wrote in message ...
>IIRC this is just a junk warning because some empty segments have been
>included.
>
>JRC
>
>In article <199902102014.MAA08732@fensende.com>, Mike Cuddy
><mcuddy@fensende.com> wrote:
>>
>>> Can anyone explain why compiling with the '-g' (debug) option, the
resulting
>>> arm-coff file contains .stab and .stabstr sections that the Tornado
tools
>> have
>>> trouble with?  I get the many of the following warning from the Target
>> Server:
>>>
>>>       Warning: Symbol .stabstr from section of type STYP_REG. Ignored.
>>>
>>> It seems the Debugger generates similar warnings along with many anoying
>>> dialog windows that need to be 'Ok'ed away.
>>
>>I got this error, too.  The only way that I found around it was to
>>use strip to remove the .stab and .stabstr sections.
>>
>>(on the other hand, I didn't try real hard to get around it, either ;-)
>>



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: problems with dosFs
Date: Sun, 14 Feb 1999 23:12:14 -0500
From: Ryan O'Connell <ryan_deb@mediaone.net>
Organization: Northeast Region--MediaOne
Message-ID: <36C79E9E.76E2D8CD@mediaone.net>
References: <9902119187.AA918743134@smtp.wg.com>

brett.smith@wg.com wrote:

>      We have been experiencing I/O problems - they seem to have something
>      to do with dosFs.
>
>      We are running the Java Virtual Machine on vxWorks, x86 target.  We
>      have some Java code that backs-up the vxworks executable file as well
>      as some *.jar files.  The code usually works, but once in a while, the
>      disk SEEMS to get corrupted.  Once this happens, an "open/create"
>      native call from the JVM appears to hang.  Reboot does not clear this
>      up.  Once the open hangs, FTP cannot perform dir or put.  The recovery
>      process requires a Format of the HD.
>
>      DOS scandisk does not detect any disk errors, but the vxWorks impl is
>      slightly different isn't it?
>
>      Has anyone experienced this type of problem?
>
>      Does anyone know of any good "show" routines or debugging aids for
>      dosFs?
>
>      I would really like to has something dump the FATs for me.  Does
>      anyone know how to see the FATs?
>
>      Any help is greatly appreciated,
>
>      Brett Smith
>      brett.smith@wg.com

You may want to ensure you have the DosFS file patch.  This patch fixed lots of
problems for us some similar to the ones you are describing.  I forget the
patch ID and they changed the windsurf password (sorry).

Ryan


---------------------------

Newsgroups: comp.os.vxworks
Subject: semaphore information
Date: Thu, 11 Feb 1999 10:29:11 -0500
From: Wendy McNaughton <wendy_mcnaughton@stratus.com>
Organization: Stratus Computer
Message-ID: <36C2F747.B6374CF9@stratus.com>

Does anyone know if there is a way for a task to detect if it holds a
semaphore?  There is a function available (semInfo) that tells me if
there are other tasks waiting on a semaphore but it doesn't indicate who
may be holding a semaphore.  I have a function where I want to detect if
the caller is holding a particular mutual exclusion semaphore but I
can't find the means to do this.

Does anyone know if this can be done?


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: TFTP question
Date: Thu, 11 Feb 1999 09:50:51 -0500
From: Tim Sohacki <sohacki@nortelnetworks.com>
Organization: Nortel (Northern Telecom)
Message-ID: <36C2EE4B.C7BEA895@nortelnetworks.com>
References: <79nmru$mgq$1@nclient3-gui.server.virgin.net>

> 
> The best I could find was a shareware TFTP Server from Walusoft. This looks
> quite nice but has a nasty bug that causes it to run out of resources after
> a few file transfers.
> 
> Could anyone recommend a freeware Windows TFTP Server program , any ideas
> appreciated thanks.

Geoff,

I don't have any success stories.

SolarWinds is another tftp server (SolarWinds.net?)

Walusoft often locks up on WinNT, and seems to have trouble with 
multiple file transfers on WinNT.  It works far better on Win95 (?).

SolarWinds works better than Walusoft on NT with multiple file 
connections.  It has a problem with sometimes not sending anything 
to the target, and sometimes it continues to send even beyond the 
end of the file (?), but overall seems to be more robust.

Keep me posted if you have any luck.

Thanks,
Tim

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Survey of peripherals used under VxWorks
Date: Mon, 15 Feb 1999 09:12:50 +0300
From: Leonid Rosenboim <leonid@bitband.com>
Organization: BitBand Technologies Ltd.
Message-ID: <36C7BAE2.5EF9E5DF@bitband.com>
References: <36C470B9.2924CEF8@jhuapl.edu> <36C6690E.22A4C88@bitband.com> <7a7cn0$11q$1@sun20.ccd.bnl.gov>

Kostia,

you may be right, but I have heard opinions to the contrary, claiming the that
tape movement speed of DLT is to high and wears out the tape too quickly,
while 8mm tapes can be used hundred of times. I do agree with you however that
the 8mm mechanics are complex, but so are swiss watches, and they work like ...
a swiss watch, nevertheless.

I too had pretty bad experience with my first EXB 8200, bu later modes showed
much improved reliability.

Anyway, this thread is getting too tape specific for this newsgroup, there is
nothing
VxWorks-specific to this discussion.

Leonid

Konstantin Olchanski wrote:

> > > I am interested in any information about peripherals that are being used
> > > under VxWorks.  For instance, we currently use Exabyte 8mm SCSI tape
> > > drives, but are interested in finding a replacement for these somewhat
> > > unreliable drives. ...
> >
> > Andy, Exabyte 8mm tapes are as reliable as tape drives can be these days...
>
> In my experience, Exabytes are not as reliable as DLTs, and it is easy to
> see why.
>
> The Exabyte drive mechanism, with all the gears, rollers and pullyes, is
> like a Swiss watch, compared to the simple and crude DLT drive mechanism.
> In a light-duty, clean-room environment it may not matter, but in
> a heavy-duty, semi-industrial environment it makes a heck of a difference.
> The delicate Exabyte mechanics are just... well... too delicate.
>
> For about three years (1993-1995) we have used EXB-8500 drives for
> heavy-duty data acquisition (round-the-clock operation at full bandwidth,
> 2-4 months per year), and the reliablity performance was abysmal. When
> new, each drive would work fine for 1-2 months, then it would "slow down"
> (write speed drops from 500 Kbytes/sec to 300-400 Kbytes/sec), record data
> that can't be read on other drives and (worse of all) cause mysterious
> SCSI bus timeouts and would have to be taken off the system (causing
> down time in a round-the-clock operation) and sent out for rebuilding.
> Out of a pool of 10 drives, all dthe rives had to be rebuilt many times.
>
> Since then, we had switched to DLT-2000 drives, and the same two DLT drives
> have served us for two years (1997 and 1998) without a single failure.
>
> As I understand, the newer half-height Exabyte mechanisms (EXB-8505 and
> later) are somewhat more durable. We have not used them for data collection,
> so I have no direct reliability data, but my gut feeling is that they
> are still way too complicated and delicate for their own good and would not
> be as reliable as DLTs in a heavy-duty, un-clean-room environment.
>
> YMMV.
>
> --
> Konstantin Olchanski
> Physics Department, Brookhaven National Laboratory, Long Island, New York
> olchansk@bnl.gov


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: semaphore information
Date: 11 Feb 1999 14:34:03 -0800
From: ggg+news@best.com (GGG)
Organization: rest
Message-ID: <79vlsr$t8v$1@shell4.ba.best.com>
References: <36C2F747.B6374CF9@stratus.com>
Sender: ggg@best.com

Take a look at $WIND_BASE/target/h/private/semLibP.h

In there the semaphore structure is defined.  Your SEM_ID is just a
pointer to one.  So, comparing the task's ID with semId->semOwner should
do the trick.


In article <36C2F747.B6374CF9@stratus.com>,
Wendy McNaughton  <wendy_mcnaughton@stratus.com> wrote:
>Does anyone know if there is a way for a task to detect if it holds a
>semaphore?  There is a function available (semInfo) that tells me if
>there are other tasks waiting on a semaphore but it doesn't indicate who
>may be holding a semaphore.  I have a function where I want to detect if
>the caller is holding a particular mutual exclusion semaphore but I
>can't find the means to do this.
>
>Does anyone know if this can be done?
>



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: semaphore information
Date: Thu, 11 Feb 1999 16:11:40 -0700
From: Mike Stimpson <mstimp@utsci.com>
Organization: Utah Scientific, Inc.
Message-ID: <36C363AC.41C6@utsci.com>
References: <36C2F747.B6374CF9@stratus.com> <79vlsr$t8v$1@shell4.ba.best.com>

GGG wrote:
> 
> Take a look at $WIND_BASE/target/h/private/semLibP.h
> 
> In there the semaphore structure is defined.  Your SEM_ID is just a
> pointer to one.  So, comparing the task's ID with semId->semOwner should
> do the trick.

True.  However, this is a *private* data structure, so there is no
guarantee that it won't change between revs of vxWorks.

Also, this technique has a timing window vulnerability:  If the task
that is checking gets pre-empted just after it read who had the
semaphore, and another task runs and takes/gives the semaphore, and then
goes to sleep, the task that was suspended thinks something that is no
longer true...

> In article <36C2F747.B6374CF9@stratus.com>,
> Wendy McNaughton  <wendy_mcnaughton@stratus.com> wrote:
> >Does anyone know if there is a way for a task to detect if it holds a
> >semaphore?  There is a function available (semInfo) that tells me if
> >there are other tasks waiting on a semaphore but it doesn't indicate who
> >may be holding a semaphore.  I have a function where I want to detect if
> >the caller is holding a particular mutual exclusion semaphore but I
> >can't find the means to do this.
> >
> >Does anyone know if this can be done?
> >

- -- 
Mike Stimpson (mstimp@utsci.com)

Opinions expressed are not necessarily those of the management...

The more I work as an engineer, the more I disbelieve the theory of
evolution.

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Sharing memory between PPC and 68K fails
Date: Mon, 15 Feb 1999 10:55:27 +0100
From: John Andersson <johand@lin.foa.se>
Organization: FOA
Message-ID: <36C7EF0F.7A92@lin.foa.se>
References: <36C4533F.1C9E@lin.foa.se> <36C53FD6.7AEE@t-online.de>
Reply-To: johand@lin.foa.se

Ulf Franke wrote:
> 
> John Andersson wrote:
> >
> > I'm trying to use shared memory semaphores on a MVME2700 and a MVME177.
> > The 177 is processor #0. This is what happens on the PPC in a windsh:
> >
> > -> semBSmCreate(0, 0)
> > value = 16411921 = 0xfa6d11
> > -> semShow(0xfa6d11, 0)
> > can't read "offset(SM_MEM,objType)": no such element in array
> >
> > The semaphore does however show up on the 68k. The above also works in
> > a C-program. Why doesn't it work in a shell?
> >
> > If the shell was the only problem everything would be fine, but it
> > isn't.
> > If I do semTake on the PPC and semGive on the 68k, semGive returns with
> > an
> > error, but errno is not set. semTake never returns. semShow on both 68k
> > and
> > PPC shows that the semaphore has been given, but the task hanging in
> > semTake
> > never gets this.
> >
> > Is this a bug or have I missed something?
> >
> > /John
> 
> Hello John,
> 
> the PPC CPU must be processor #0 to export shared memory.
> 
> See "VxWorks Programmers Guide" chapter PPC.
> 
> Best regards
> 
> Ulf
> 
> =================
> Ulf.Franke@men.de


The problem still persists. I now use the PPC as processor #0, but the
symptoms are exactly the same.

/Regards, John

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: VxWorks port to Linux survey
Date: Mon, 15 Feb 1999 03:44:54 -0500
From: Vic Mulyk <mulykv@cadvision.com>
Organization: CADVision Development Corporation (http://www.cadvision.com/)
Message-ID: <36C7DE86.6EF9112D@cadvision.com>
References: <78qnh6$dd37@overload.lbl.gov> <7948ge$e0t$1@news.netvision.net.il>

Wouldn't this just be taking the VxWorks headers and libs
and creating a Linux hosted - cross compiler?  
For me, PowerPC 8XX cross.

- -Vic

ps. this would be a very nice alternative to using it on NT.

Saffi Hartal wrote:
> 
> Hi Bruce
> Could you elaborate of the meaning of "porting vxWorks to Linux" ?.
> I work in vxWorks and i would like to port stuff from linux to vxWorks or
> work with very light linux instead of working in vxWorks ( i would rather do
> that because i'll have the sources ).
> if you mean working in linux environment on vxWorks ... could be nice but i
> rather for now be able to port linux code to vxWorks.
> So i guess my needs are a bit different then yours.
>         With Regards
>                         Saffi.
> 
> Bruce O. Benson wrote in message <78qnh6$dd37@overload.lbl.gov>...
> >In response to the recent lengthy thread on the hope (for some at least)
> >that WRS would port VxWorks to Linux (or BSD for that matter), I am
> >collecting names to present to WRS management as a show of strength for
> >this.
> >
> >If you would like to be included as a supporter of this port, please drop
> me
> >a note at mailto:benson@linux.sparta.com
> >In fairness, if you think it's a dumb idea, I will collect your note and fw
> >your name, too.
> >
> >I'll post the statistical results of the survey, and the WRS response in
> the
> >next couple of weeks on http://linux.sparta.com/VxWorks
> >
> >Regards,
> >Bruce.
> >Bruce Benson, Senior Engineer,
> >SPARTA (CPD)
> >7926 Jones Branch Drive, Suite 900
> >McLean, VA  22102-3303
> >USA
> >+1 703-448-0210 x211
> >benson@sparta.com
> >http://www.mclean.sparta.com
> >
> >

- -- 
- -Vic

               \\\\\\\\\\\
                  / ^   ^ \
                 { (o) (o) }
- ---------oOOOo--------U-------oOOOo----------
                 Vic Mulyk
             victorm@synapse.net
        http://www.synapse.net/~victorm
                                Oooo.
- -------------.oooO--------------(   )--------
             (   )               ) /
              \ (               (_/ 
               \_)

---------------------------

Newsgroups: comp.os.vxworks
Subject: How to get address of second IP interface ?
Date: Mon, 15 Feb 1999 12:27:39 +0100
From: Sebastien Mallet <mallet@nortel.ca>
Organization: Nortel
Message-ID: <36C804AB.675AB20E@nortel.ca>

Hi
 I'm looking for an API to be able to retrieve adress of second IP
interface...
ifShow gives all of them on the Display, any other ideas ?
I'm working with Sens on a mv2303 BSP with two  or three TCP/IP
interface. (ethernet, VME, PCI).

Any Help would be greatly appreciated.
Sebastien Mallet

---------------------------

End of New-News digest
**********************


From owner-vxwexplo-process  Mon Feb 15 17:22:28 1999
From: Fred Roeber <froeber@BBN.COM>
Date: Mon Feb 15 17:22:31 PST 1999
Subject: RE: Using WindView to measure network performance


On Sun, 14 Feb 1999, Keith Buchanan responded to a post:

> > We do _NOT_ use windview as we have found that it can seriously
> > interfere
> > with the network performance of vxWorks (especially if you are
> > connecting windview over the network!). For the same reason we do not
> > use windSh either.....
> 
> All Tornado tools consume some network bandwidth, after all nothing comes
> for free.  How and when they affect you application timing can be mitigated
> by altering the relative priority of the tasks that handle the WDB traffic.
> WindView is not the best tool to measure network bandwidth.  Measuring
> throughput is easily done, just load up a generic blaster/blastee test
> program and let it rip.  Measuring the response time of individual packets
> is done best by a tool lke the HP Internet Advisor.

I agree that using WindView in its "normal" operational mode (where it
uploads event data to an analysis host in real time) can significantly
perturb the application that you are using WindView to analyze.
However, there are some solutions that let you collect event data with
WindView without affecting application performance significantly. One
is to use WindView in its "post mortem" mode where it collects event
data but doesn't upload the event data until event collection is
complete. The drawbacks to this approach are that figuring out how to use
the post mortem mode can be a bit tricky and the amount of data you can
collect in this mode is limited by the size of the event buffer you
allocate on your target.

If you have an application running in a VME system, then another way to
collect event data without affecting the application much is to use our
TraceMaker tool. Our website www.tracemaker.bbn.com has information on
using TraceMaker with WindView in the section on WindView Support.
There are also some WindView displays that show the affect that the
normal event data upload can have on your application if you aren't
using TraceMaker.

One thing to realize is that WindView provides great insight into how
an application is operating. You lose a lot of potential help in
debugging complex application problems by not using WindView.  Fred

   |      Fred J Roeber,  BBN Systems & Technologies      |
   |    4 John Clarke Road   Middletown, RI  02842-5202   |
   |          froeber@bbn.com     401-848-3548            |
   | TraceMaker product manager -> www.tracemaker.bbn.com |


From owner-vxwexplo-process  Tue Feb 16 04:00:14 1999
From: daemon@csg.lbl.gov
Date: Tue Feb 16 04:00:18 PST 1999
Subject: comp.os.vxworks newsdigest

Comp.Os.Vxworks Daily Digest    Tue Feb 16 04:00:10 PST 1999

        Subject: Re: Adding net driver to VxWorks
        Subject: Re: Shared Memory Network problems
        Subject: Experiences with Ethernet-Chip SMC91C94
        Subject: VME-PCI Bridge Problems with Cypress Universe
        Subject: File System Compatibility
        Subject: Shared Memory Network problems
        Subject: Target Server Error Report
        Subject: Virtual I/O problems
        Subject: Re: ntp daemon and SENS
        Subject: Survey of peripherals used under VxWorks
        Subject: Re: VxWorks execute out of ROM on MV2604
        Subject: MV162-763 memory problem
        Subject: RAIMA DB or POLYHEDRA DB on VxWork.
        Subject: Re: NAT implementation for VxWorks

-------------------------------------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Adding net driver to VxWorks
Date: Mon, 15 Feb 1999 08:26:24 -0600
From: Frank Minich <fminich@ti.com>
Organization: NCP
Message-ID: <36C82E90.5F2F@ti.com>
References: <7a8s37$4jh2@overload.lbl.gov>
Reply-To: fminich@ti.com

Parag Badge wrote:
> 
> Hi,
>     I have new network card. I want to write driver for it
> and install it in VxWorks. Can anybody tell me how to add
> new drivers in VxWorks?
> Thanx
> Parag
> 

Get the BSP Porting Kit from WRS.

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Shared Memory Network problems
Date: Mon, 15 Feb 1999 12:53:26 GMT
From: "Thomas Fuda" <fudath@mail.northgrum.com>
Organization: Northrop Grumman Corporation, Baltimore, MD
Message-ID: <F775I3.Crz%mother@mdnews.md.essd.northgrum.com>
References: <7a1c19$488$1@nnrp1.dejanews.com>
Sender: news%mother@mdnews.md.essd.northgrum.com


john.collis@rrds.co.uk wrote in message <7a1c19$488$1@nnrp1.dejanews.com>...
>I am using VxWorks on multiple DY4 SVME177 PowerPC 603 SBCs in a VME rack.
The
>Shared Memory Network has been incorporated into VxWorks. There is an
Ethernet
>connection from the Master 177 card connected via a hub to a Sun
SPARCstaion.
>The VxWorks image downloads successfully to the Master 177 and starts
>executing. The slave 177 card detects the Shared Memory Network 'heartbeat'
>when it starts to boot, a 'Loading...' message is displayed, then after
some
>time an errno=0xd0003 message is displayed and boot is aborted. Whilst the
>slave is attempting to load, invoking smNetShow on the master shows that
the
>Slave is there but with no activity. Has anyone in this forum used the
Shared
>Memory network facility and got it working? What is errno 0xd0003?



What mechanism are you using to download the image (RSH or FTP)? Try forcing
FTP by specifying a user ID and password in your boot parameters (if you
haven't done so already).



---------------------------

Newsgroups: comp.os.vxworks
Subject: Experiences with Ethernet-Chip SMC91C94
Date: Mon, 15 Feb 1999 16:43:48 +0100
From: Volkhau@t-online.de (Heribert Volkhausen)
Message-ID: <7a9f6j$tdj$1@news03.btx.dtag.de>

Hi,

does somebody have made experiences in using an ethernet adapter based on
SMC's 91C94.

Please give me a comment.

Thanks in advance
Heribert Volkhausen




---------------------------

Newsgroups: comp.os.vxworks
Subject: VME-PCI Bridge Problems with Cypress Universe
Date: Mon, 15 Feb 1999 17:19:00 +0000
From: Matt Cowley <Matthew.Cowley@cambridge.simoco.com>
Organization: Simoco Europe Cambridge
Message-ID: <36C85704.5ADE6BF@cambridge.simoco.com>


- --------------672DF9DB914823315A9F8BE5
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

I'm writing a device driver for a VMETRO MIDAS20  PCI expansion card.
My VME system
controler is an Artesyn BAJA (MIPS R4700) and its VME interface device
is the VIC64.
I've mapped the PCI I/O, Memory and configuration space into the BAJAs
VME space, which
works fine.  However, when I map the BAJAs 64M of DRAM into the PCI
space and I access
the BAJAs DRAM I get aprox 40 to 50 target terminations without data,
followed by a target abort, which reboots the BAJA.  Also, when I try to
initialise my PMC SCSI device, which resides on
the MIDAS PCI bus, my BAJA reboots.

Does anyone have any experience with writing vxWorks drivers for the
Cypress Universe and
VIC64 and do you have any ideas what causes these problems?

Thanks

Matt

- --
- --------------------------------------------------------------
                        Matthew Cowley

Simoco Europe Ltd.              Email Matthew.Cowley@cambridge.simoco.com
P.O. Box 24                     Web Site: http://www.simoco.com
St Andrews Road                 Tel. +44 1223 877544
CAMBRIDGE CB4 1DP               Fax. +44 1223 314812
ENGLAND           (This is not an official communication)



- --------------672DF9DB914823315A9F8BE5
Content-Type: text/html; charset=us-ascii
Content-Transfer-Encoding: 7bit

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML>
I'm writing a device driver for a VMETRO MIDAS20&nbsp; PCI expansion card.&nbsp;
My VME system
<BR>controler is an Artesyn BAJA (MIPS R4700) and its VME&nbsp;interface
device is the VIC64.
<BR>I've mapped the PCI I/O, Memory and configuration space into the BAJAs
VME&nbsp;space, which
<BR>works fine.&nbsp; However, when I map the BAJAs 64M of DRAM into the
PCI space and I access
<BR>the BAJAs DRAM I get aprox 40 to 50 target terminations without data,
followed by a target abort, which reboots the BAJA.&nbsp; Also, when I
try to initialise my PMC&nbsp;SCSI device, which resides on
<BR>the MIDAS PCI bus, my BAJA reboots.
<P>Does anyone have any experience with writing vxWorks drivers for the
Cypress Universe and
<BR>VIC64 and do you have any ideas what causes these problems?
<P>Thanks
<P>Matt
<PRE>--
- --------------------------------------------------------------
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Matthew Cowley

Simoco Europe Ltd.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Email Matthew.Cowley@cambridge.simoco.com
P.O. Box 24&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Web Site: <A HREF="http://www.simoco.com">http://www.simoco.com</A>
St Andrews Road&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Tel. +44 1223 877544
CAMBRIDGE CB4 1DP&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Fax. +44 1223 314812
ENGLAND&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; (This is not an official communication)</PRE>
&nbsp;</HTML>

- --------------672DF9DB914823315A9F8BE5--


---------------------------

Newsgroups: comp.os.vxworks
Subject: File System Compatibility
Date: Mon, 15 Feb 1999 12:56:10 -0500
From: "David P. Stickney" <stickney@kodak.com>
Organization: Eastman Kodak Company
Message-ID: <7a9n6a$qh5$1@news.kodak.com>

Hi all.

Is it possible for several NT4 SP3 systems to share storage with a VxWorks
system?
Can a VxWorks system read/write to an NTFS partition?

Thanks for any related information!



---------------------------

Newsgroups: comp.os.vxworks
Subject: Shared Memory Network problems
Date: Fri, 12 Feb 1999 13:58:07 GMT
From: john.collis@rrds.co.uk
Organization: Deja News - The Leader in Internet Discussion
Message-ID: <7a1c19$488$1@nnrp1.dejanews.com>

I am using VxWorks on multiple DY4 SVME177 PowerPC 603 SBCs in a VME rack. The
Shared Memory Network has been incorporated into VxWorks. There is an Ethernet
connection from the Master 177 card connected via a hub to a Sun SPARCstaion.
The VxWorks image downloads successfully to the Master 177 and starts
executing. The slave 177 card detects the Shared Memory Network 'heartbeat'
when it starts to boot, a 'Loading...' message is displayed, then after some
time an errno=0xd0003 message is displayed and boot is aborted. Whilst the
slave is attempting to load, invoking smNetShow on the master shows that the
Slave is there but with no activity. Has anyone in this forum used the Shared
Memory network facility and got it working? What is errno 0xd0003?

Regards,
John Collis
Software Engineer
Racal Radar Defence Systems Limited
Leicester
Leics
UK

- -----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

---------------------------

Newsgroups: comp.os.vxworks
Subject: Target Server Error Report
Date: Mon, 15 Feb 1999 16:56:25 -0700
From: "Gregory L. Wickstrom" <glwicks@sandia.gov>
Organization: Sandia National Laboratories
Message-ID: <36C8B429.72EC@sandia.gov>
Reply-To: glwicks@sandia.gov

I'm using MV2304 BSP hosted solely on Windows NT machines.

The basic problem seems to be in running development tools on one NT
Workstation while executing the target server on another, and trying to
load files off the workstation that the Tornado tools are running on. 
More specific details are listed below.

My development system configuration is the following:

	Machine GLWICKS2 is running the target server launched from the
'Tornado/Tools/Target Server' menu 
	Machine GLWICKS is running the Tornado development tools and the shell

With this configuration doing an 'ld < xxx.o' command at the shell will
attempt to pull the xxx.o file from machine GLWICKS.  However, in the
shell on machine GLWICKS showing the following error:

	-> ld < xxx.o
	WTX Error 0x2 (no such file or directory)
	value = -1 = 0xffffffff

The target server window on machine GLWICKS2 shows the following
message:

	tgtsvr.exe (Fred@glwicks2): Thu Feb 11 07:55:45 1999
 	   Error: Can't open object module C:/Data/Software/VxWorks/XXX/xxx.o
  	  It could be a permission problem or the target server
  	  has no access to the module.

I have given the card it's own user id and password on GLWICKS and given
that user id full permissions on the xxx.o file.  I'm also logged into
both GLWICKS and GLWICKS2 as the same user, so I don't think it's a
permission problem.  The target server error message also states that
the target server may have no access to the module.  What does this mean
and how could that occur?  Do you think that's what my problem is?  If
not, what else could be causing it?


Thanks,

Greg Wickstrom
Sandia National Laboratories
(505) 844-7708

---------------------------

Newsgroups: comp.os.vxworks
Subject: Virtual I/O problems
Date: Mon, 15 Feb 1999 17:04:09 -0700
From: "Gregory L. Wickstrom" <glwicks@sandia.gov>
Organization: Sandia National Laboratories
Message-ID: <36C8B5F9.6B87@sandia.gov>
Reply-To: glwicks@sandia.gov

I'm using MV2304 BSP hosted on Windows NT machines.

I seem to be having a virtual I/O problem.  I'd like to use the host
based shell as the I/O interface to a simple application.  Of course it
defaults to wanting to use the serial port on the Motorola card, so I
used the following redirector to a virtual I/O port that I thought would
be going to the shell:

vf0=open("/vio/0",2,0)
ioGlobalStdSet(0,vf0)
ioGlobalStdSet(1,vf0)
ioGlobalStdSet(2,vf0)
logFdAdd(vf0)

However, although this does seem to direct the output to the shell, the
input doesn't seem to work.  In fact, it also doesn't event work at the
serial port either.  I'm using C++ with 'cin' and 'cout' statements for
some simple I/O.  Any ideas?


Thanks,

Greg Wickstrom
Sandia National Laboratories
(505) 844-7708

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: ntp daemon and SENS
Date: Fri, 12 Feb 1999 17:50:47 -0800
From: Ric Claus <claus@slac.stanford.edu>
Organization: Stanford Linear Accelerator Center
Message-ID: <36C4DA76.8821A3@slac.stanford.edu>
References: <199902101405.JAA07704@bara.gb.nrao.edu>

John Ford wrote:

> Hi.
>
> We have installed the SENS stack, and our implementation of ntp,
> derived from version 3.1, has quit working.  No other hosts can talk
> to the server running on vxWorks.  Does anyone know why this may have
> happened?  I can't see what would cause it.  Backing up to version
> 5.3.1 without SENS works just fine.
>
> Version: VxWorks 5.3.1
> SENS version: 1.0 + a patch
>
> Thanks for any insight!
>
> --
> John Ford
> National Radio Astronomy Observatory
> Green Bank, WV 24944-0002
> jford@nrao.edu

SENS 1.0 is buggy.  Get SENS 1.1.  In particulary, SENS 1.0 won't allow
you to open more than 15 or so sockets before running out of resources
without tweakage and recvmsg () corrupts it's second argument.  After
SENS 1.1 things were better, but recvmsg () had a habit of returning a
negative result (< -1).  There's a patch to fix this.  See WindSurf.
It's still not clear to us that recvmsg behaves entirely correctly as we
seem to see that it unexpectedly returns zero on occasion.

- --

    Ric Claus    (claus@slac.stanford.edu, (650) 926-3480)




---------------------------

Newsgroups: comp.os.vxworks
Subject: Survey of peripherals used under VxWorks
Date: Fri, 12 Feb 1999 13:19:37 -0500
From: Andy Helten <Andy.Helten@jhuapl.edu>
Organization: JHU/APL (http://www.jhuapl.edu)
Message-ID: <36C470B9.2924CEF8@jhuapl.edu>

I am interested in any information about peripherals that are being used
under VxWorks.  For instance, we currently use Exabyte 8mm SCSI tape
drives, but are interested in finding a replacement for these somewhat
unreliable drives.  It doesn't have to be a SCSI peripheral -- USB,
FibreChannel, Ethernet, Firewire, etc. -- are all possibilities.  I am
really just interested in knowing what is out there and any
personal/professional opinions on what is available.

Thanks,
Andy Helten


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: VxWorks execute out of ROM on MV2604
Date: Mon, 15 Feb 1999 21:44:58 -0600
From: petekz@cyberramp.net (Pete Kockritz)
Organization: posted via: CyberRamp.net, Dallas, TX (214) 343-3333/(817) 461-8484 for info
Message-ID: <petekz-1502992144590001@dal-tsa13-52.cyberramp.net>
References: <7aajb2$l5e$1@winter.news.rcn.net>

In article <7aajb2$l5e$1@winter.news.rcn.net>, "David Garner"
<daveg@progeny.net> wrote:

> I normally compile the VxWorks target; vxWorks.st_rom and burn it into flash
> after running it through elfToBin.  I can also successfully burn the flash
> online using a program that Ian Love wrote.  That program uses the elf
> version of the program: vxWorks.st_rom so that elfToBin is no longer
> necessary.  These files are compressed so that only about 500K of the flash
> is used.  Our MV2604s have 4 meg of flash memory installed.  All of this is
> working perfectly.  Now here's the problem:
> 
>  I want to generate an uncompressed VxWorks that will run out of ROM with
> only the data segment moved to RAM.  According to the manual this can be
> done on the PowerPC by specifying the rule: vxWorks.res_rom_res_low.  I
> compile this successfully and load it (prior to elfToBin) into the flash
> using the online burn program.  When I restart the processor the BFL amber
> light stays lit constantly and nothing happens.  I can not do the ppcBug
> niop burn because the output of elfToBin indicates that the load address is
> wrong.  According to the wrs database there is a TSR on this and the problem
> has been found in the elfToBin program.  The fix is not in distribution yet.
> I figure since I'm burning online with the burnFlash function that this
> won't affect me since I don't use elfToBin anyway.

Can you verify that the initialized data section is relocated to a ROM 
address? It should be immediately after your text section in the flash.
At initialization time, the ROM resident image should copy this section
to its runtime location in RAM.

I had a problem with a debugger not relocating the data section when I was
simulating a rom resident vxWorks image. You should be able to recognize
some of your initialized data, look for some strings that you know should
be there.

Regards,
Pete
- --
+-----------------------------------------------------------------+
|       Pete Kockritz                mailto:petekz@cyberramp.net  |
|                       This space for rent.                      |
+-----------------------------------------------------------------+

---------------------------

Newsgroups: comp.os.vxworks
Subject: MV162-763 memory problem
Date: Tue, 16 Feb 1999 18:24:53 +1100
From: Robert Nedelkovski <r.nedelkovski@rlmsystems.com.au>
Organization: RLM Systems Pty Ltd
Message-ID: <36C91D45.DE0766D0@rlmsystems.com.au>

We have VxWorks 5.1.1 with the MV162 board support package running on
Sparc 20.

We have tried to access the 16MB of ECC memory installed on the board.
The kernel dies whenever we change LOCAL_MEM_SIZE in config.h and
recompile to any size above 4 Meg. We have changed sysPhysMemDesc to
allow for the memory but that doesn't get us home either. Interestingly,
if we modify sysLib.c such that the VMECHIP2 is set to see all 16Megs,
but LOCAL_MEM_SIZE is set to 4 Meg, then we can access the memory above
the 4 Meg boundary.

Any clues on how to fix this so the kernel can see the 16 Meg as well ?

Thanks

Robert


---------------------------

Newsgroups: comp.os.vxworks
Subject: RAIMA DB or POLYHEDRA DB on VxWork.
Date: Tue, 16 Feb 1999 11:14:23 +0100
From: Marco Bernasconi <Marco.Bernasconi.GP@icn.siemens.de>
Organization: Siemens AG
Message-ID: <36C944FF.758F2A5D@icn.siemens.de>

I would like to use a Data Base, Raima , Polyhedra. or .. I don´t know,
on VxWork.
Someone know something? I need to "replicate" data on different boards,
can these DB help me or exist a client/server  real time DB on VxWork?
May be free source.

Thanks in advance,
Marco




---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: NAT implementation for VxWorks
Date: Tue, 16 Feb 1999 05:34:18 -0500
From: P Pulley <113001.723@CompuServe.COM>
Organization: Penny & Giles Aerospace
Message-ID: <#ZxkpfZW#GA.153@ntdwwaaw.compuserve.com>
References: <79v6n6$sf53@overload.lbl.gov>

We looked into this last year.
Apparently an implementation is available from Interniche 
Technologies
see www.iniche.com




Phil Pulley
Penny & Giles Aerospace Ltd (UK)
AIS DIvision
(ppulley@pgaero.co.uk)

- -- 







---------------------------

End of New-News digest
**********************


From owner-vxwexplo-process  Tue Feb 16 11:12:59 1999
From: Larry Sinn <sinnl@sd-star.com>
Date: Tue Feb 16 11:13:03 PST 1999
Subject: Re:Shared Memory Network problems

John Collis asked:

What is errno 0xd0003?

-------------------------------

VxWorks uses an error number reporting scheme similar to Unix.
Given that an error number is 32 bits, Unix only uses the lower 16 bits,
leaving the upper 16 bits to be used as a Module Number, so you can have
64k modules (-1 for Unix), each with 64k error code.

If you look in $WIND_BASE/target/h/vwModNum.h you will see that 0xd (
same as (13 << 16) ) corresponds to M_iosLib. Now in iosLib.h you see
(M_iosLib | 3) which corresponds to  S_iosLib_INVALID_FILE_DESCRIPTOR.

See man on errnoLib.
This is a really neat feature to integrate your error number with
VxWorks and it is especially usefull during debug.

Larry.


Larry Sinn
Spectral Dynamics               (408)474-1746  voice
1983 Concourse Drive            (408)474-1780  fax
San Jose, Ca. 95131-1708        sinnl@sd-star.com




From owner-vxwexplo-process  Tue Feb 16 11:23:12 1999
From: Jack Gold <jack.gold@lmco.com>
Date: Tue Feb 16 11:23:16 PST 1999
Subject: vxMemProbe concurrency problem

This is a multi-part message in MIME format.
--------------6BED996D5111AE87DF972F7A
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

I am currently in the process of linking an Ada application to the
VxWorks OS for a PPC and the following problem arose:

How do you test the write capability of RAM where the application
already exists?  This is a serious problem.  If you attempt to read a
memory address and then write to the memory address, there is always a
question of whether you were interrupted by a higher priority task which
has changed the location's value.

We considered what would happen if, instead of reading the value from an
address, storing it in some temporary space and then writing it back, we
chose to call vxMemProbe using the following parameter setup:
	vxMemProbe(storage_address, 1, 4, storage_address)

where the "storage_address" is both the address we want to write to and
the address we obtain the write value from.

Does anyone know if vxMemProbe will protect the "storage_address" from
being overwritten by a preempting task?


--Jack Gold 
--Lockheed Martin Aerospace Electronics
--------------6BED996D5111AE87DF972F7A
Content-Type: text/x-vcard; charset=us-ascii; name="vcard.vcf"
Content-Transfer-Encoding: 7bit
Content-Description: Card for Jack Gold
Content-Disposition: attachment; filename="vcard.vcf"

begin:          vcard
fn:             Jack Gold
n:              Gold;Jack
org:            Lockheed Martin Fairchild Systems
email;internet: jack.gold@lmco.com
title:          Analyst
x-mozilla-cpt:  ;0
x-mozilla-html: TRUE
version:        2.1
end:            vcard


--------------6BED996D5111AE87DF972F7A--



From owner-vxwexplo-process  Tue Feb 16 15:22:38 1999
From: "Bollhorst, Robert J" <robert.j.bollhorst@lmco.com>
Date: Tue Feb 16 15:22:42 PST 1999
Subject: RE: Shared Memory Network problems

I find an easier method to look up errno values is to
print them from the shell:

-> printErrno 0xd0003
0xd0003 = S_iosLib_INVALID_FILE_DESCRIPTOR
value = 0 = 0x
->

Bob
 <<...>> 

Robert J. Bollhorst
Software Engineering

Lockheed Martin
P.0. Box 4931, 103 Chesapeake Park Plaza  
Baltimore, MD 21220-0931
Telephone 410.682.0779  Facsimile 410.682.1678
E-Mail: mailto://robert.j.bollhorst@lmco.com


> ----------
> From: 	vxwexplo@lbl.gov[SMTP:vxwexplo@lbl.gov]
> Sent: 	Tuesday, February 16, 1999 2:13 PM
> To: 	vxworks_users@csg.lbl.gov
> Subject: 	Re:Shared Memory Network problems
> 
> Submitted-by owner-vxwexplo-process  Tue Feb 16 11:12:59 1999
> Submitted-by: Larry Sinn <sinnl@sd-star.com>
> 
> John Collis asked:
> 
> What is errno 0xd0003?
> 
> -------------------------------
> 
> VxWorks uses an error number reporting scheme similar to Unix.
> Given that an error number is 32 bits, Unix only uses the lower 16 bits,
> leaving the upper 16 bits to be used as a Module Number, so you can have
> 64k modules (-1 for Unix), each with 64k error code.
> 
> If you look in $WIND_BASE/target/h/vwModNum.h you will see that 0xd (
> same as (13 << 16) ) corresponds to M_iosLib. Now in iosLib.h you see
> (M_iosLib | 3) which corresponds to  S_iosLib_INVALID_FILE_DESCRIPTOR.
> 
> See man on errnoLib.
> This is a really neat feature to integrate your error number with
> VxWorks and it is especially usefull during debug.
> 
> Larry.
> 
> 
> Larry Sinn
> Spectral Dynamics               (408)474-1746  voice
> 1983 Concourse Drive            (408)474-1780  fax
> San Jose, Ca. 95131-1708        sinnl@sd-star.com
> 


From owner-vxwexplo-process  Wed Feb 17 04:00:18 1999
From: daemon@csg.lbl.gov
Date: Wed Feb 17 04:00:22 PST 1999
Subject: comp.os.vxworks newsdigest

Comp.Os.Vxworks Daily Digest    Wed Feb 17 04:00:14 PST 1999

        Subject: C++, overloading global new/delete
        Subject: timeouts with the VxWorks msgq utils
        Subject: Industrial Computing Society - update
        Subject: How to get "arpAdd", "arpDelete" work
        Subject: SENS error?
        Subject: SENS drivers and BSPs
        Subject: Re: vxMemProbe concurrency problem
        Subject: Re: Target Server Error Report
        Subject: Re: How to get IP of Gateway and own IP?
        Subject: Re: SENS error?
        Subject: Re: SENS error?

-------------------------------------------------------

Newsgroups: comp.os.vxworks
Subject: C++, overloading global new/delete
Date: Tue, 16 Feb 1999 13:59:35 +0100
From: Chris Varlese <no@email.com>
Organization: Ascom Switzerland, Bern, Switzerland
Message-ID: <36C96BB7.DEEEF022@email.com>

My setup is Tornado v1.0.1 with a MC68040 cross compiler.

I have some trouble to build a ROM image when I have overloaded the
global C++ new and delete operators.  The linker reports a clash between

the new/delete we have defined and the standard new/delete in the gnu
C++ library.

Does anyone know how to solve this problem?  I looked at the manual
for linker options but I don't see how to tell the linker to ignore the
_builtin_new and builtin_delete symbols from the gnu library.  I wonder
how normal compilers deal with this?  (e.g. GNU cygwin can happily
link code where you have overloaded new/delete;  but you don't see
what it is doing in the linking process).

Chris Varlese
Ascom Transmission AG



---------------------------

Newsgroups: comp.os.vxworks
Subject: timeouts with the VxWorks msgq utils
Date: Tue, 16 Feb 1999 08:28:15 -0500
From: Paul Allum <Pallum@harris.com>
Organization: Harris Electronic System Sector
Message-ID: <36C9726F.91B2A75B@harris.com>

Does anyone know how the msgQReceive function returns a timeout
indication?


---------------------------

Newsgroups: comp.lang.java,comp.os.ms-windows.ce,comp.os.qnx,comp.os.vxworks,comp.protocols.kerberos,comp.robotics.misc,comp.sys.m68k
Subject: Industrial Computing Society - update
Date: Tue, 16 Feb 1999 15:26:55 -0500
From: Mark Wells <mark@ics.org>
Organization: Industrial Computing Society
Message-ID: <36C9D48F.E8886E26@ics.org>

This is a multi-part message in MIME format.
- --------------9AEB1B1597A6465851C3850E
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

The Industrial Computing Society is pleased to report important new
benefits for the industrial computing community.  Find out about our
upcoming tutorials on Windows CE, Real-Time Java, Internet Security for
embedded devices, and object oriented databases.

Complete details are at http://www.ics.org/icspresident/update0299.html

Also, don't forget to read the first in a series of monthly articles by
Armin Steinhoff on "The World of Open Standards" at
http://www.ics.org/columns/armin_feb.html.

Sincerely,

Mark Wells
President, Industrial Computing Society
mark@ics.org

- --------------9AEB1B1597A6465851C3850E
Content-Type: text/html; charset=us-ascii; name="ICS.html"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline; filename="ICS.html"
Content-Base: "file:///I|/ICS/ICS.html"

<html>
<body>
<a href="http://www.ics.org/" 
target="_top"><img width=414 height=155 border=0 
src="http://www.ics.org/images/header2.gif"></a>
</body>
</html>
- --------------9AEB1B1597A6465851C3850E--


---------------------------

Newsgroups: comp.os.vxworks
Subject: How to get "arpAdd", "arpDelete" work
Date: Tue, 16 Feb 1999 21:35:07 GMT
From: "Yie Wang" <ywang@cs.dal.ca>
Organization: Magma Communications Ltd.
Message-ID: <fwly2.997$8c4.4413352@news.magma.ca>

Hi all,
   We are running vxWorks 5.3.1 and use mpc860 as our processor. We find
that the "arpAdd" does not work in the shell. We use "lkup" command to check
"arpAdd" and this symbol does not exist.

   What INCLUDE OPTIONS should we add in the WindConfig BSP configuration
dialog box to make "arpAdd", "arpDelete" work? Currently, arpShow works
well.

  Thank you very much.
   Yie  Wang






---------------------------

Newsgroups: comp.os.vxworks
Subject: SENS error?
Date: Tue, 16 Feb 1999 18:34:00 -0600 (CST)
From: Bob Robison <brobison@swri.org>
Organization: Southwest Research Institute
Message-ID: <XFMail.990216183400.brobison@swri.org>
Reply-To: brobison@swri.org
Sender: robison@narnia.electro.swri.edu

I am using NT-hosted Tornado Ver 1.0.1 (would prefer Linux....) and VxWorks
5.3.1.  I am aware of the install problem that you cannot install SENS at the
same time you install the rest of the tools.  I had a problem a while back
building a vxWorks kernel, and my local rep suggested that I do a clean
install, and NOT install SENS at the same time.  I did this and the problem
seemed to go away.  (I did not install SENS at all).  But since then, I needed
to install SENS to get some other network code to compile.  Anyway, I just now
tried to re-compile the kernel because I needed to make a minor configuration
change, and now I am getting the same error I was getting before:

   'undefined reference to m_freem'

Has anyone else seen this error?  Is it possibly related to SENS?  Any
suggestions on where to go from here?  I know I haven't given a lot of detail
here, but I wanted to keep the post fairly short.

thanks,
bob

- -----------
This message was sent by XFMail at 18:33:55 on 16-Feb-99 by:
Bob Robison          Southwest Research Institute,http://www.swri.org
Staff Engineer             Signal Exploitation & Geolocation Division
brobison@swri.org               Phone:(210)522-3935/Fax:(210)522-2709
<><    <><    <><    <><    <><    <><    <><    <><    <><   Col 3:2

---------------------------

Newsgroups: comp.os.vxworks
Subject: SENS drivers and BSPs
Date: Tue, 16 Feb 1999 21:22:52 -0500
From: "Neil Lefebvre" <neil.lefebvre@sympatico.ca>
Organization: Bell Solutions
Message-ID: <KFpy2.3173$iQ3.4211@news21.bellglobal.com>

Hi everyone.

Can an END driver written as described in the SENS manual be used as the
main network interface for retreiving VxWorks after booting, and for
debugging?  Or is the BSP network driver very different from a SENS driver?

Neil Lefebvre

please reply to me at neil.lefebvre@sympatico.ca





---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: vxMemProbe concurrency problem
Date: Tue, 16 Feb 1999 19:13:09 -0800
From: sjq@aracqnet.com
Organization: National Association for the Advancement of Computer Geeks
Message-ID: <36CA33C5.B650C04C@aracqnet.com>
References: <7acjk7$4jh5@overload.lbl.gov>

Jack Gold wrote:

> This is a multi-part message in MIME format.
> --------------6BED996D5111AE87DF972F7A
> Content-Type: text/plain; charset=us-ascii
> Content-Transfer-Encoding: 7bit
>
> I am currently in the process of linking an Ada application to the
> VxWorks OS for a PPC and the following problem arose:
>
> How do you test the write capability of RAM where the application
> already exists?  This is a serious problem.  If you attempt to read a
> memory address and then write to the memory address, there is always a
> question of whether you were interrupted by a higher priority task which
> has changed the location's value.
>

The usual way to handle this is with semaphores.  If you have a critical
region
of memory, accesses to it are guarded with calls to semTake() and
semGive().

Other ways include pre-emption lockouts ( taskLock() and taskUnlock() ) and

interrupt disabling ( intLock() and intUnlock() ).  Semaphores should be
preferred.

Can you modify the Ada app to use semaphores?  Since you are worried about
this,
one assumes that there will be some memory shared with the Ada app and the
rest
of the system.

>
> We considered what would happen if, instead of reading the value from an
> address, storing it in some temporary space and then writing it back, we
> chose to call vxMemProbe using the following parameter setup:
>         vxMemProbe(storage_address, 1, 4, storage_address)
>
> where the "storage_address" is both the address we want to write to and
> the address we obtain the write value from.
>
> Does anyone know if vxMemProbe will protect the "storage_address" from
> being overwritten by a preempting task?

No.  vxMemProbe is for diagnostics and things like that.  It is to test and
see if
there is any memory at a particular address.

Normally, if you try to access invalid memory, you get a bus error (a
"machine
check exception" in PowerPC lingo); which is Not A Good Thing.  vxMemProbe
"hooks" the bus error vector such that if the access is bad, and a bus
error does
occur, it is harmlessly trapped.  (vxMemProbe also "unhooks" the vector
when it
is done, so normal bus error handling is restored.)

It sound like what you are thinking of is an equivalent to the PPC
instructions
"store with reservation" (stwcx) and "load with reservation"  (lwarx).
There
are no vxWorks functions which provide direct access
to these two instructions, so you will have  to use assembly language if
you
want to use these...and for small scalars, these might work.  (They do
cause
a performance hit if used to frequently...)

Hope this helps,

Scott

>
> --Jack Gold
> --Lockheed Martin Aerospace Electronics
> --------------6BED996D5111AE87DF972F7A
> Content-Type: text/x-vcard; charset=us-ascii; name="vcard.vcf"
> Content-Transfer-Encoding: 7bit
> Content-Description: Card for Jack Gold
> Content-Disposition: attachment; filename="vcard.vcf"
>
> begin:          vcard
> fn:             Jack Gold
> n:              Gold;Jack
> org:            Lockheed Martin Fairchild Systems
> email;internet: jack.gold@lmco.com
> title:          Analyst
> x-mozilla-cpt:  ;0
> x-mozilla-html: TRUE
> version:        2.1
> end:            vcard
>
> --------------6BED996D5111AE87DF972F7A--

- --
- ---------------------------------------------------------------------
Scott Johnson        Beer-Drinker, Geek, Citizen          GO BLAZERS!
If you want to send mail to me, remove the letters 'twixt "r" and "p"
- ---------------------------------------------------------------------




---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Target Server Error Report
Date: Tue, 16 Feb 1999 22:02:47 -0800
From: "John Finley" <john@kivala.com>
Organization: Kivala Systems, Inc.
Message-ID: <7adlrq$qkk@dfw-ixnews9.ix.netcom.com>
References: <36C8B429.72EC@sandia.gov>

Gregory L. Wickstrom wrote in message <36C8B429.72EC@sandia.gov>...
....
>The target server window on machine GLWICKS2 shows the following
>message:
>
> tgtsvr.exe (Fred@glwicks2): Thu Feb 11 07:55:45 1999
>    Error: Can't open object module C:/Data/Software/VxWorks/XXX/xxx.o
>    It could be a permission problem or the target server
>    has no access to the module.
....

When you do a ld at the host shell, the shell tells the
target server to do the load, and gives it the file name
you typed as the file to load from. So that all gets sent
over the network to your target server running on a
different machine. That target server tries to open the
file, but can't, because the file is at the specified
location only as viewed on the first machine. The path
you give to ld has to be the path as viewed by the target
server, wherever it is running.

You can do one of the following:

- - Copy any files to be loaded over to the target server
machine. Yuk.

- - Map/mount the drive where the file exists on the machine
where the target server is running, then give it that path.

- - Run the target server where you are building your files.

If you can't do the third, the second is usually feasible.

John

- --------------------------------------------------
 John Finley                 Kivala Systems, Inc.
 Project Manager                   (619) 689-0032
 john@kivala.com            http://www.kivala.com
- --------------------------------------------------
 VxTool - Build Tornado Apps with MS Visual Basic
- --------------------------------------------------




---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: How to get IP of Gateway and own IP?
Date: Tue, 16 Feb 1999 23:17:26 -0800
From: "gto" <overney@cemvax.cem.msu.edu>
Organization: SBC Internet Services
Message-ID: <G_ty2.4693$bP2.46071@typhoon-sf.pbi.net>
References: <NGQw2.1665$Qf5.12593@typhoon-sf.pbi.net> <slrn7c89mn.f41.chc@nasa2.ksc.nasa.gov> <36C46876.41C6@utsci.com>

Thanks. Works great. I found out that bootParams.ead returns the network
interface IP and, as you have mentioned, bootParams.gad returns the Gateway
IP.

But where can I find the complete description of this BOOT_PARAMS structure?
I looked under bootLib and bootStringToStruct() in my VxWorks reference
manual but could not find anything useful regarding BOOT_PARAMS.

Gregor


Mike Stimpson wrote in message <36C46876.41C6@utsci.com>...
>Charles H. Chapman wrote:
>>
>> On Thu, 11 Feb 1999 23:28:11 -0800, gto <overney@cemvax.cem.msu.edu>
wrote:
>> >I am using VxWorks 5.3.1 on a PPC860. Once the board is up and running,
I
>> >would like to get its own IP as well as the IP of the Gateway. In
VxWorks,
>> >neither gethostid() nor gethostbyname() exists. What system calls can
>> >provide me with this info?
>>
>> You can use ifAddrGet() to get your own IP address but I'm not sure how
>> you could find out the address of the gateway.
>>
>> Chuck
>
>If the bootline contains the *current* gateway address, you can do
>
>BOOT_PARAMS bootParams;
>bootStringToStruct(BOOTLINE_ADDRESS, &bootParams)
>
>and then the gateway address is in bootParams.gad (for this to work, if
>you *change* the gateway address, you have to also store the new one in
>the bootline...)
>
>Hope that helps.
>--
>Mike Stimpson (mstimp@utsci.com)
>
>Opinions expressed are not necessarily those of the management...
>
>The more I work as an engineer, the more I disbelieve the theory of
>evolution.



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: SENS error?
Date: Wed, 17 Feb 1999 10:34:57 +0200
From: Alef0 <Alef0@127.0.0.1>
Organization: NetVision Israel	
Message-ID: <36CA7F30.892139DD@127.0.0.1>
References: <XFMail.990216183400.brobison@swri.org>

We've encountered the same problem with our setup (NT host, x86 target).
After some calls to WRS support, we found out

The problem seems to be this:  the older network stack contained a function
named m_freem.  The newer SENS stack does not have this function, but
rather a macro of the same name.  This leads to linker errors when you try
to build the kernel, because the object file cacheArchLib.o tries to access
this function.  Interestingly, according to the man page, this file is only
relevant to the Motorola 68K targets, NOT to x86, but it was included
in our build.  For some reason, it apparently solves some BSP bug, making
our target machine boot much faster.

Possible workarounds (worked for us...):
1. Removing the line "MACH_EXTRA     = c:\tornado\target\lib\cacheArchLib.o"
from our /target/config/pc486/Makefile solved this issue, leaving
us without the functionality we gained from the cacheArchLib, but with SENS.

2. Disabling SENS and reverting to the original stack also works.

3. WRS rep suggested that newer x86 BSPs (for Pentium / PentiumPro / PentiumII)
due shortly may also solve this problem.

Hope this helps,

Alef0

- ---
Replace 127.0.0.1 with usa.net when replying by e-mail
C programmers never die, they are just cast into void.


Bob Robison wrote:

> I am using NT-hosted Tornado Ver 1.0.1 (would prefer Linux....) and VxWorks
> 5.3.1.  I am aware of the install problem that you cannot install SENS at the
> same time you install the rest of the tools.  I had a problem a while back
> building a vxWorks kernel, and my local rep suggested that I do a clean
> install, and NOT install SENS at the same time.  I did this and the problem
> seemed to go away.  (I did not install SENS at all).  But since then, I needed
> to install SENS to get some other network code to compile.  Anyway, I just now
> tried to re-compile the kernel because I needed to make a minor configuration
> change, and now I am getting the same error I was getting before:
>
>    'undefined reference to m_freem'
>
> Has anyone else seen this error?  Is it possibly related to SENS?  Any
> suggestions on where to go from here?  I know I haven't given a lot of detail
> here, but I wanted to keep the post fairly short.
>
> thanks,
> bob
>
> -----------
> This message was sent by XFMail at 18:33:55 on 16-Feb-99 by:
> Bob Robison          Southwest Research Institute,http://www.swri.org
> Staff Engineer             Signal Exploitation & Geolocation Division
> brobison@swri.org               Phone:(210)522-3935/Fax:(210)522-2709
> <><    <><    <><    <><    <><    <><    <><    <><    <><   Col 3:2






---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: SENS error?
Date: Wed, 17 Feb 1999 12:46:28 +0100
From: Sebastien Mallet <mallet@nortel.ca>
Organization: Nortel
Message-ID: <36CAAC14.C2E71692@nortel.ca>
References: <XFMail.990216183400.brobison@swri.org> <36CA7F30.892139DD@127.0.0.1>

Hi,
This problem has been fixed by some various patches available from the
support (or Windsurf).
The fix is very easy, see :
http://www.wrs.com/search97cgi/s97_cgi?action=View&VdkVgwKey=%2E%2E%2Ftext%2FFAQs%5Fno%5Fauthor%2F114464%2E2&DocOffset=8&DocsFound=8&QueryZip=m%5Ffreem&Collection=FAQs&Collection=known+problems&Collection=VxWorks+demo&Collection=technotes&Collection=compOs+News&Collection=manualsTor1%2E0%2E1&Collection=manualsTor1%2E0&Collection=manuals5%2E3&Collection=manuals5%2E2&Collection=manuals5%2E1&Collection=manuals5%2E0&SortField=CollectionName&SortOrder=Desc&ScoreThreshold=0&hlnavigate=all&ViewTemplate=wrsExterna
lView.template&name=FAQs

or search for m_feem at the support site
HTH
Sebastien Mallet

---------------------------

End of New-News digest
**********************


From owner-vxwexplo-process  Wed Feb 17 08:01:56 1999
From: "Kohorn, Judy" <judy.kohorn@lmco.com>
Date: Wed Feb 17 08:01:59 PST 1999

Has any one had problems with TaskSpawn under AdaMulti?  My code compiles
under Tornado - but not AdaMulti.

Thanks,
Judy Kohorn
Lockheed
judy.kohorn@lmco.com
(805)572-3162


From owner-vxwexplo-process  Wed Feb 17 08:12:06 1999
From: "Williams, Scott" <scott.williams@lmco.com>
Date: Wed Feb 17 08:12:10 PST 1999
Subject: NFS using SENS on PowerPC

Has anyone used NFS (client) with SENS on PowerPC to access remote files on
a WinNT machine?  This is the first time I have used VxWorks NFS support, so
would appreciate any lessons learned from those who have trod this path
before me.

Thanks.  :-)

Scott Williams
Senior Group Engineer
Lockheed Martin Launching Systems
103 Chesapeake Park Plaza,  MS 54
Baltimore, MD  21220
Ph (410) 682-2522
Fx (410) 682-1742
Scott.williams@lmco.com




From owner-vxwexplo-process  Thu Feb 18 04:00:14 1999
From: daemon@csg.lbl.gov
Date: Thu Feb 18 04:00:19 PST 1999
Subject: comp.os.vxworks newsdigest

Comp.Os.Vxworks Daily Digest    Thu Feb 18 04:00:11 PST 1999

        Subject: Problem with MPC860SAR
        Subject: Re: Target Server Error Report
        Subject: Re: IEEE 1394 drivers available?
        Subject: Using ZIP or LS-120 Drive
        Subject: find stuff for your os
        Subject: Ctrl-X sequence?
        Subject: Books & Docs

-------------------------------------------------------

Newsgroups: comp.os.vxworks
Subject: Problem with MPC860SAR
Date: Wed, 17 Feb 1999 14:18:49 +0100
From: Tricarico Giuseppe <tricaric@sisun9.cselt.it>
Organization: Centro Servizi Interbusiness
Message-ID: <36CAC1B9.85865712@sisun9.cselt.it>

Hi everyone

I have problem with MPC860SAR when I try to use the SAR functions for
UTOPIA interface.
The following is a description of my environment and settings:

- -EST MDP8XX PRO Development Platform with MPC860SAR.
- -VxWorks O.S.
- -Driver for UTOPIA interface, provided by
Motorola.(www.mot.com/SPS/RISC/netcomm/tools/)
- -I haven't still an interface that connect the phisical layer with
UTOPIA interface, but I have     connected the TxCAV signal at VCC, and
RxCAV signal at GND.

I'm testing the driver to adapt it for my environment, so I'm using
Tornado tools.
When I try to set the PDPAR register with the value 0xDF7F (enabling the
UTOPIA and SAR finction) the debugger on the remote host stops
debugging, because it cannot communicate with the target.
I think that there could be a conflict with the Ethernet driver, when
the UTOPIA SAR function are Enabled.
If  I set  this register with EST's visionICE (that use the development
port on the platform), the Tornado tools blocks, after I clear the PDPAR
register with visionICE and the Tornado tools return to run. What happen
when the UTOPIA interface is enabled?

Thanks for help

  Tricarico Giuseppe.


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Target Server Error Report
Date: Wed, 17 Feb 1999 11:13:37 +0000
From: Stephen A Walsh <stephen.walsh@gecm.com>
Organization: Marconi Electronic Systems
Message-ID: <36CAA461.218D@gecm.com>
References: <36C8B429.72EC@sandia.gov>

Gregory L. Wickstrom wrote:
> 
> I'm using MV2304 BSP hosted solely on Windows NT machines.
> 
> The basic problem seems to be in running development tools on one NT
> Workstation while executing the target server on another, and trying to
> load files off the workstation that the Tornado tools are running on.
> More specific details are listed below.
> 
> My development system configuration is the following:
> 
>         Machine GLWICKS2 is running the target server launched from the
> 'Tornado/Tools/Target Server' menu
>         Machine GLWICKS is running the Tornado development tools and the shell
> 
> With this configuration doing an 'ld < xxx.o' command at the shell will
> attempt to pull the xxx.o file from machine GLWICKS.  However, in the
> shell on machine GLWICKS showing the following error:
> 
>         -> ld < xxx.o
>         WTX Error 0x2 (no such file or directory)
>         value = -1 = 0xffffffff
> 
> The target server window on machine GLWICKS2 shows the following
> message:
> 
>         tgtsvr.exe (Fred@glwicks2): Thu Feb 11 07:55:45 1999
>            Error: Can't open object module C:/Data/Software/VxWorks/XXX/xxx.o
>           It could be a permission problem or the target server
>           has no access to the module.
> 
> I have given the card it's own user id and password on GLWICKS and given
> that user id full permissions on the xxx.o file.  I'm also logged into
> both GLWICKS and GLWICKS2 as the same user, so I don't think it's a
> permission problem.  The target server error message also states that
> the target server may have no access to the module.  What does this mean
> and how could that occur?  Do you think that's what my problem is?  If
> not, what else could be causing it?
> 

I think that if you start the Target Server in GLWICKS this would solve
your problem. 
If this dosn't solve the problems I may have to think a bit harder.

Hope This Helps,

Steve
- -- 

*******************************************
Stephen A Walsh
Senior Software Engineer
Radar and Countermeasures Systems
Marconi Electronic Systems
*******************************************
Opinions are usually mine and mine alone
*******************************************































.

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: IEEE 1394 drivers available?
Date: 18 Feb 1999 01:27:29 GMT
From: Ken Hayber <khayber@kstc.konicabt.com>
Organization: Konica Business Technologies, Inc.
Message-ID: <7afqa1$duv$1@news-2.news.gte.net>
References: <36C3DC27.4921D633@ekf.de>

Award (now Phoenix) the BIOS company is working on vxWorks IEEE 1394 
drivers.  Sorry I don't have contact information.

J. Jansen <jj@ekf.de> wrote in <36C3DC27.4921D633@ekf.de>:

> Does anybody know if there are VxWorks drivers and other software
> available for the TI PCILynx TSB12LV21A IEEE 1394 link layer
> controller? We see the need to integrate our CompactPCI 
> FireWire Board (http://www.ekf.de/c/cfiw/cf1/cf1_e.html)
> into a VxWorks environment.
> 
> Thanks to all - J. Jansen
> 
> -- 
> EKF Elektronik GmbH Industrial Computers & Information Technology
> Joachim Jansen         mailto:jj@ekf.de         http://www.ekf.de
> Philipp Reis Str. 4                      fax: +49 (0)2381-6890-90
> D-59065 Hamm (Germany)                  phone: +49 (0)2381-6890-0

---------------------------

Newsgroups: comp.os.vxworks
Subject: Using ZIP or LS-120 Drive
Date: 18 Feb 1999 01:30:02 GMT
From: Ken Hayber <khayber@kstc.konicabt.com>
Organization: Konica Business Technologies, Inc.
Message-ID: <7afqeq$duv$2@news-2.news.gte.net>

I am looking for information about anyone who has successfully got an IDE 
(ATAPI) Zip or LS-120 drive working under vxWorks.

It seems to me that they should appear as hard disks.  I know that Linux 
supports IDE Zip drives and fdisk shows a partition table for them.


Thanks.

---------------------------

Newsgroups: comp.os.os2.programmer.tools,comp.os.os2.scitech,comp.os.os2.setup.misc,comp.os.os2.setup.storage,comp.os.os2.setup.video,comp.os.os2.utilities,comp.os.os9,comp.os.parix,comp.os.plan9,comp.os.qnx,comp.os.research,comp.os.rsts,comp.os.v,comp.os.vms,comp.os.vxworks,comp.os.xinu
Subject: find stuff for your os
Date: Wed, 17 Feb 1999 21:00:59 -0400
From: crhedges@hotmail.com (John Willson)
Organization: none
Message-ID: <crhedges-1702992101000001@circ-ras2-3-cs-21.dial.bright.net>

find it here  http://www.eads.com/adserve/adserve.dll/link?CN19607,0,3
if that doesnt work then try 
 http://www.eads.com/adserve/adserve.dll/link?CN19607

---------------------------

Newsgroups: comp.os.vxworks
Subject: Ctrl-X sequence?
Date: Thu, 18 Feb 1999 13:07:15 +0900
From: Aaron Jang <aaron@integra.co.kr>
Organization: Integra Telecom Co., Ltd. Seoul KOREA
Message-ID: <36CB91F2.D4A389C3@integra.co.kr>

Hello there?

I want to insert some function between
Ctrl-X key detect and system reset.
What can I do for this work?
My target system is a MC68360 SBC.

Thanks.

- --

Aaron Jang (aaron@integra.co.kr,"ÀåÀÎÁø")

Integra Telecom Co., Ltd.              http://www.integra.co.kr
Hwayoung Bdg., 112-2, Garak-Dong,      Tel : +82-2-449-6851
Songpa-Gu, Seoul, 138-160, Korea       Fax : +82-2-431-3741



---------------------------

Newsgroups: comp.os.vxworks
Subject: Books & Docs
Date: Wed, 17 Feb 1999 18:39:14 +0000
From: Erich MARTIN <emartin@mpisa.francenet.fr>
Organization: MPI-tech, France
Message-ID: <36CB0CD2.BE9CA4CC@mpisa.francenet.fr>


Where can I find informations about vxworks programming ? I do not find
references to vxworks at amazon.com ... Can someone give me a few titles
? (I do not have VxWorks yet, so I want books I can buy separatly from
VxWorks)

- -- 
Erich MARTIN

---------------------------

End of New-News digest
**********************


From owner-vxwexplo-process  Thu Feb 18 06:37:11 1999
From: "Rasty Slutsker (STX)" <rslutsker@stx.kollmorgen.com>
Date: Thu Feb 18 06:37:14 PST 1999
Subject: vxworks: problem with proadcast

This message is in MIME format. Since your mail reader does not understand
this format, some or all of this message may not be legible.

------_=_NextPart_001_01BE5B4C.131EB818
Content-Type: text/plain;
	charset="windows-1255"
Content-Transfer-Encoding: quoted-printable

Hello,
I=92ve faced with the strange problem with broadcast messages. I CANNOT =
send
UDP with broadcast (255.255.255.255) address. The IP layer converts =
this
address into something like xxx.255.255.255 where xxx is my subnet.
IfBroadcastSet() does not help, moreover, =
IfBroadcastSet(=93255.255.255.255=94)
returns with =961.

Does somebody know BOOTP works if the target refuses to send UDP to
255.255.255.255 ?

Thanks,

------_=_NextPart_001_01BE5B4C.131EB818
Content-Type: text/html;
	charset="windows-1255"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV=3D"Content-Type" CONTENT=3D"text/html; =
charset=3Dwindows-1255">
<META NAME=3D"Generator" CONTENT=3D"MS Exchange Server version =
5.5.2232.0">
<TITLE>vxworks: problem with proadcast</TITLE>
</HEAD>
<BODY>

<P ALIGN=3DLEFT><FONT SIZE=3D2 FACE=3D"Arial">Hello,</FONT></P>

<P ALIGN=3DLEFT><FONT SIZE=3D2 FACE=3D"Arial">I</FONT><FONT SIZE=3D2 =
FACE=3D"Arial">=92</FONT><FONT SIZE=3D2 FACE=3D"Arial">ve faced =
w</FONT><FONT SIZE=3D2 FACE=3D"Arial">ith the strange problem =
with</FONT> <FONT SIZE=3D2 FACE=3D"Arial">broadcast</FONT><FONT =
SIZE=3D2 FACE=3D"Arial"></FONT> <FONT SIZE=3D2 =
FACE=3D"Arial">messages.</FONT> <FONT SIZE=3D2 FACE=3D"Arial">I =
CANNOT</FONT> <FONT SIZE=3D2 FACE=3D"Arial">send UDP with =
broadcast</FONT> <FONT SIZE=3D2 FACE=3D"Arial">(255.255.255.255)</FONT> =
<FONT SIZE=3D2 FACE=3D"Arial">address</FONT><FONT SIZE=3D2 =
FACE=3D"Arial">.</FONT><FONT SIZE=3D2 FACE=3D"Arial"></FONT> <FONT =
SIZE=3D2 FACE=3D"Arial">T</FONT><FONT SIZE=3D2 FACE=3D"Arial">he</FONT> =
<FONT SIZE=3D2 FACE=3D"Arial">IP layer con</FONT><FONT SIZE=3D2 =
FACE=3D"Arial">v</FONT><FONT SIZE=3D2 FACE=3D"Arial">erts</FONT><FONT =
SIZE=3D2 FACE=3D"Arial"></FONT> <FONT SIZE=3D2 FACE=3D"Arial">this =
address into somethi</FONT><FONT SIZE=3D2 FACE=3D"Arial">ng</FONT><FONT =
SIZE=3D2 FACE=3D"Arial"></FONT> <FONT SIZE=3D2 =
FACE=3D"Arial">like</FONT> <FONT SIZE=3D2 =
FACE=3D"Arial">xxx</FONT><FONT SIZE=3D2 FACE=3D"Arial">.255</FONT><FONT =
SIZE=3D2 FACE=3D"Arial">.255.255 where xxx is my subnet.</FONT> <FONT =
SIZE=3D2 FACE=3D"Arial">I</FONT><FONT SIZE=3D2 =
FACE=3D"Arial">f</FONT><FONT SIZE=3D2 FACE=3D"Arial">B</FONT><FONT =
SIZE=3D2 FACE=3D"Arial">roadcastSet(</FONT><FONT SIZE=3D2 =
FACE=3D"Arial">)</FONT><FONT SIZE=3D2 FACE=3D"Arial"> does not help, =
moreover,</FONT> <FONT SIZE=3D2 FACE=3D"Arial">I</FONT><FONT SIZE=3D2 =
FACE=3D"Arial">f</FONT><FONT SIZE=3D2 FACE=3D"Arial">B</FONT><FONT =
SIZE=3D2 FACE=3D"Arial">roadcastSet</FONT><FONT SIZE=3D2 =
FACE=3D"Arial">(</FONT><FONT SIZE=3D2 FACE=3D"Arial">=93</FONT><FONT =
SIZE=3D2 FACE=3D"Arial">255.255.255</FONT><FONT SIZE=3D2 =
FACE=3D"Arial">.255</FONT><FONT SIZE=3D2 FACE=3D"Arial">=94</FONT><FONT =
SIZE=3D2 FACE=3D"Arial">) returns with</FONT> <FONT SIZE=3D2 =
FACE=3D"Arial">=96</FONT><FONT SIZE=3D2 FACE=3D"Arial">1.</FONT></P>

<P ALIGN=3DLEFT><FONT SIZE=3D2 FACE=3D"Arial">Does som</FONT><FONT =
SIZE=3D2 FACE=3D"Arial">e</FONT><FONT SIZE=3D2 =
FACE=3D"Arial">body</FONT> <FONT SIZE=3D2 FACE=3D"Arial">know</FONT> =
<FONT SIZE=3D2 FACE=3D"Arial">BOOTP</FONT><FONT SIZE=3D2 =
FACE=3D"Arial"> works if the target</FONT> <FONT SIZE=3D2 =
FACE=3D"Arial">refuses to send</FONT> <FONT SIZE=3D2 FACE=3D"Arial">UDP =
to</FONT>&nbsp;<FONT SIZE=3D2 FACE=3D"Arial"></FONT> <FONT SIZE=3D2 =
FACE=3D"Arial">255.255.255</FONT><FONT SIZE=3D2 =
FACE=3D"Arial">.255</FONT><FONT SIZE=3D2 FACE=3D"Arial"> ?</FONT></P>

<P ALIGN=3DLEFT><FONT SIZE=3D2 FACE=3D"Arial">Thanks,</FONT></P>

</BODY>
</HTML>
------_=_NextPart_001_01BE5B4C.131EB818--




From owner-vxwexplo-process  Thu Feb 18 06:58:01 1999
From: Charlie Grames <cgrames@mdc.com>
Date: Thu Feb 18 06:58:05 PST 1999
Subject: vxMemProbe concurrency problem -Reply

Jack,

What you propose doing probably depends on the architecture.  I believe it
will work on the Motorola MVME2604/2700, because sysMemProbeBus(), one of
the routines used in the vxMemProbe() implementation, uses intLock() and
intUnlock() on either side of the actual load and store operations.

I have to admit that I'm a little curious why you would want to test the
read and write attributes of RAM in which an application is already loaded.

Charlie Grames
The Boeing Company
(314) 233-1956
Charles.R.Grames@boeing.com


>>> the vxWorks Users Group Exploder <vxwexplo@lbl.gov> 02/16/99 01:23pm >>>
Submitted-by owner-vxwexplo-process  Tue Feb 16 11:23:12 1999
Submitted-by: Jack Gold <jack.gold@lmco.com>

This is a multi-part message in MIME format.
--------------6BED996D5111AE87DF972F7A
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

I am currently in the process of linking an Ada application to the
VxWorks OS for a PPC and the following problem arose:

How do you test the write capability of RAM where the application
already exists?  This is a serious problem.  If you attempt to read a
memory address and then write to the memory address, there is always a
question of whether you were interrupted by a higher priority task which
has changed the location's value.

We considered what would happen if, instead of reading the value from an
address, storing it in some temporary space and then writing it back, we
chose to call vxMemProbe using the following parameter setup:
	vxMemProbe(storage_address, 1, 4, storage_address)

where the "storage_address" is both the address we want to write to and
the address we obtain the write value from.

Does anyone know if vxMemProbe will protect the "storage_address" from
being overwritten by a preempting task?


--Jack Gold 
--Lockheed Martin Aerospace Electronics
--------------6BED996D5111AE87DF972F7A
Content-Type: text/x-vcard; charset=us-ascii; name="vcard.vcf"
Content-Transfer-Encoding: 7bit
Content-Description: Card for Jack Gold
Content-Disposition: attachment; filename="vcard.vcf"

begin:          vcard
fn:             Jack Gold
n:              Gold;Jack
org:            Lockheed Martin Fairchild Systems
email;internet: jack.gold@lmco.com
title:          Analyst
x-mozilla-cpt:  ;0
x-mozilla-html: TRUE
version:        2.1
end:            vcard


--------------6BED996D5111AE87DF972F7A--



From owner-vxwexplo-process  Thu Feb 18 09:33:50 1999
From: Jim Caprio <caprio@research.moore.com>
Date: Thu Feb 18 09:33:54 PST 1999
Subject: VxWorks memory partitions

BSP: mv2300 (604e PPC)

I'm interested in some general information on how to optimize memory
usage in VxWorks.  We're running into a memory fragmentation problem,
which is caused in part to multiple threads concurrently allocating
memory from the same memory partition (the default heap).  Due to
VxWorks "first fit" (vs. best fit) allocation algorithm, the result is
inefficient placement of objects in memory, and resultant fragmentation.

Several candidate solutions have been proposed:
(a) Assign (pre-allocated?) memory partitions on a per task basis
(b) Create memory buffers for homogenous object size
(c) Modify the VxWorks memory allocation algorithm to make better use of
available space

To your knowledge, are any of these options possible in VxWorks?  A
brief explanation of how to implement the required modifications, and
some motivation behind the "best" approach would be greatly
appreciated.   Thanks for the help.


--
=================================
Jim Caprio
Moore Research Center, Inc.
(716)773-0333
FAX  : (716)773-0462
email: caprio@research.moore.com
=================================





From owner-vxwexplo-process  Thu Feb 18 22:44:34 1999
From: ksamavedam@hns.com (Krishna Samavedam)
Date: Thu Feb 18 22:44:37 PST 1999
Subject: IP address change..


HI VxWorkers,
	I have a basic Question, Can the IP address be changed when vxWorks up and running, 
	with out rebooting ? if so how ?

thanx in advance, 
Krishna Samavedam


From owner-vxwexplo-process  Fri Feb 19 00:53:53 1999
From: "Keith Buchanan" <Keith_Buchanan@SPARTA.com>
Date: Fri Feb 19 00:53:56 PST 1999
Subject: RE: IP address change..

> 	I have a basic Question, Can the IP address be changed when
> vxWorks up and running,
> 	with out rebooting ? if so how ?

Unlike Windoze, it sure can.  Use the ifAddrSet and ifMaskSet API functions.

HTH
------------------------------------
      /\       Thomas Keith Buchanan
     /**\      Principal Engineer
    /****\     SPARTA, Inc.
   /****/..    205 East IH 30
  /****/....   Rockwall, TX 75087
 .\****\__...
...\**\  /.... Voice: 972 772 4487
....\**\/\.... FAX:   972 722 3201
 ....\****\..  PCS:   972 672 9657
  ....\****/   WWW:   www.sparta.com
   .../***/
    ./***/     SPARTA
     \**/      ======
      \/       Pride In Performance





From owner-vxwexplo-process  Fri Feb 19 01:26:02 1999
From: David Laight <dsl@tadpole.co.uk>
Date: Fri Feb 19 01:26:06 PST 1999
Subject: Re: VxWorks memory partitions

IMHO the memory allocater in vxworks (memPartAlignedAlloc does the work)
seems designed to fragment memory as much as humanly possible.                 
As well as using a 'first fit' algorythm it compounds things by using a LIFO 
free list.  
The effect is that if you free a large piece of memory the next allocate will be 
from (the end of) that piece.
(If the free list were in address order, then memory would not get as fragmented
- but would still get fragmented.)

I think this behaviour could be considerd a bug...

	David

> Subject: VxWorks memory partitions

> Submitted-by: Jim Caprio <caprio@research.moore.com>
> 
> BSP: mv2300 (604e PPC)
> 
> I'm interested in some general information on how to optimize memory
> usage in VxWorks.  We're running into a memory fragmentation problem,
> which is caused in part to multiple threads concurrently allocating
> memory from the same memory partition (the default heap).  Due to
> VxWorks "first fit" (vs. best fit) allocation algorithm, the result is
> inefficient placement of objects in memory, and resultant fragmentation.
> 
> Several candidate solutions have been proposed:
> (a) Assign (pre-allocated?) memory partitions on a per task basis
> (b) Create memory buffers for homogenous object size
> (c) Modify the VxWorks memory allocation algorithm to make better use of
> available space
> 
> To your knowledge, are any of these options possible in VxWorks?  A
> brief explanation of how to implement the required modifications, and
> some motivation behind the "best" approach would be greatly
> appreciated.   Thanks for the help.
> 
> Jim Caprio

----------------------------------------------------------------
David Laight                          email: dsl@tadpole.co.uk
Tadpole Technology plc                phone: +44 1223 278 256
Cambridge, UK                         fax:   +44 1223 278 201



From owner-vxwexplo-process  Fri Feb 19 04:00:40 1999
From: daemon@csg.lbl.gov
Date: Fri Feb 19 04:00:44 PST 1999
Subject: comp.os.vxworks newsdigest

Comp.Os.Vxworks Daily Digest    Fri Feb 19 04:00:37 PST 1999

        Subject: Questions on features and performance
        Subject: Re: Ctrl-X sequence?
        Subject: MPC860SAR & VxWorks
        Subject: Re: MPC860SAR & VxWorks
        Subject: Classless Subnetting (Bug?)
        Subject: Re: IEEE 1394 drivers available?
        Subject: How do I hook the Timer Interrupt on the Intel BSP?
        Subject: re: Help!  Cannot see BREAKPOINTS in Tornado 1.0.1
        Subject: Re: Help! Cannot see BREAKPOINTS in Tornado 1.0.1
        Subject: How to read CMOS into VxWorks clock?
        Subject: Re: timeouts with the VxWorks msgq utils
        Subject: Frame's structure in SENS
        Subject: two ethernets and vxworks

-------------------------------------------------------

Newsgroups: comp.os.vxworks
Subject: Questions on features and performance
Date: Thu, 18 Feb 1999 12:04:32 GMT
From: Geurt <G.Vos@rohill.nl>
Organization: Deja News - The Leader in Internet Discussion
Message-ID: <7agvkc$3pv$1@nnrp1.dejanews.com>

Hello,

For a project at Rohill Enginreening BV, we have to determine
which RTOS to use for a mainboard with an Pentium II and an I/O
board with an i960 (it is not yet certain if we will use
these two specific processors).

We are researching several RTOSes and right now VxWorks seems
to have the best chances. There are a few unanswered questions
on the features and performance of VxWorks:

1. Does VxWorks support:
	- TCP/IP routing
	- PCI Plug & Play (determine memory/port ranges, etc).
        - Hard drive mirroring

2. What is the performance of VxWorks:
	- Guarenteed maximum interrupt latency
        - Context switching time
   The target processors are a P2 and an i960. Could you please
   specify the times for these two processors (or related
   processors). For P2 I would prefer 333Mhz.

3. What are the typical ROM and RAM sizes for VxWorks.

4. What is the availability of device drivers (serial ports,
   PCI, H110, ...).


Could someone please answer (part of) these question?
Thanks in advance!

Geurt Vos
Rohill Engineering BV
http://www.rohill.nl

- -----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Ctrl-X sequence?
Date: Thu, 18 Feb 1999 15:27:50 +0000
From: Ian Love <ijl@wireplay.bt.co.uk>
Organization: BT Labs - but speaking only for myself
Message-ID: <36CC3176.686E@wireplay.bt.co.uk>
References: <36CB91F2.D4A389C3@integra.co.kr>

Aaron Jang wrote:
> 
> Hello there?
> 
> I want to insert some function between
> Ctrl-X key detect and system reset.
> What can I do for this work?
> My target system is a MC68360 SBC.
> 
> Thanks.

rebootHookAdd() would seem a good place to start.....
	rebootHookAdd((FUNCPTR)rebootHook);

function prototype:
	 void rebootHook(int startType)

"VxWorks Reference Manual 5.3.1", Edition 1 page 2-481


regards
Ian Love,

The views expressed above are my own and not those of my employer

---------------------------

Newsgroups: comp.os.vxworks
Subject: MPC860SAR & VxWorks
Date: Thu, 18 Feb 1999 18:53:31 +0100
From: Tricarico Giuseppe <tricaric@sisun9.cselt.it>
Organization: Centro Servizi Interbusiness
Message-ID: <36CC539B.B291297A@sisun9.cselt.it>

Hi everyone

I wrote a message because I had conflict problem with MPC860SAR.
Now I think that the proble arise because the UTOPIA use the DREQ0
signal as Rxcav.
If the Ethernet controller use IDMA1 channel the conflict is inevitable!

Do someone know if the ethernet controller of VxWorks use IDMA1?

Thanks for help
                Tricarico Giuseppe


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: MPC860SAR & VxWorks
Date: Thu, 18 Feb 1999 18:23:58 GMT
From: td@imd.m.isar.de (Thomas Doerfler)
Organization: IMD
Message-ID: <36cc59e1.35453138@news.isar.net>
References: <36CC5353.7F03702B@sisun9.cselt.it>

On Thu, 18 Feb 1999 18:52:19 +0100, Tricarico Giuseppe
<tricaric@sisun9.cselt.it> wrote:

Hi,

>Hi everyone
>
>I wrote a message because I had conflict problem with MPC860SAR.
>Now I think that the proble arise because the UTOPIA use the DREQ0
>signal as Rxcav.

This sound a bell for me...

>If the Ethernet controller use IDMA1 channel the conflict is inevitable!
>
>Do someone know if the ethernet controller of VxWorks use IDMA1?

No, this does not make sense. Each SCC has it's own (virtual) DMA
channel and is not bound to the IDMAx pins. But all SCCs and the
SAR/UTOPIA interface share the same CP....

I have been told that the SAR microcode is quite sensitive to inproper
initialization sequences concerning the Ports. And some earlier
versions of the SAR manual supplement  (and also the current sample
software available on Motorolas web server) specify/use the wrong
order for port initialization. 

Maybe this hint helps a bit...

Bye,
	Thomas.


>
>Thanks for help
>                Tricarico Giuseppe
>

- --------------------------------------------
IMD Ingenieurbuero fuer Microcomputertechnik
Thomas Doerfler           Herbststrasse 8
D-82178 Puchheim          Germany
email:    td@imd.m.isar.de

---------------------------

Newsgroups: comp.os.vxworks
Subject: Classless Subnetting (Bug?)
Date: 18 Feb 1999 16:44:42 -0500
From: Henry Houh <hhh@wasabi.lcs.mit.edu>
Organization: MIT Laboratory for Computer Science
Message-ID: <qw93e43v12t.fsf_-_@wasabi.lcs.mit.edu>


Hi,

I have found that the IP stack on VxWorks enforces classful
subnetting.  That is, if the IP address of the system is a Class C
address (say 192.168.1.X), then I cannot set the netmask to be
anything less than a 24 bit subnet mask, ie, I can set the subnet mask
to be 255.255.255.0 or 255.255.255.128 but not 255.255.240.0.

RFC1878 (Informational - Variable Length Subnet Table For IPv4), which
describes the current best practice for subnetting, specifically
allows any subnet mask for any address.  In addition, no router or
other devices that I know of enforces classful subnetting.

Does anyone know how to get around this limitation?  We need our
system to work in classlessly (no pun intended) subnetted networks.  I
have tried first setting the IP address to a temporary Class A
address, then setting the subnet mask, then setting the IP address to
a Class C address, but this does not work.

Does anything in SENS fix this, and is SENS currently bug-free enough
to add?  I have heard that 1.0 is a bit buggy, but 1.1 fixes many of
them.

Thanks,

Henry Houh
hhh@lcs.mit.edu


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: IEEE 1394 drivers available?
Date: Thu, 18 Feb 1999 14:03:50 -0800
From: Sathyan Iyengar <sathyan@intotoinc.com>
Organization: The Internet MainStreet
Message-ID: <36CC8E46.EEEA7A8D@intotoinc.com>
References: <36C3DC27.4921D633@ekf.de> <36C9D7B4.9574DB8E@signtech.com>

Yes. Intoto, Inc. a spin-off from Technology Rendezvous Inc. has
complete software
solution for 1394 supported on VxWorks. The software stack supports
various 1394
Link Controllers, processors and protocols(IP1394, SBP-2, AV/C, DPP, CP,
etc.)

Please visit http://www.intotoinc.com or call 408-844-0480 or send
e-mail to info@intotoinc.com

Thanks,
- -Sathyan

Allen Walker wrote:

> Intoto (http://www.intotoinc.com) provides VxWorks drivers for the
> PCILynx.
>
> J. Jansen wrote:
> >
> > Does anybody know if there are VxWorks drivers and other software
> > available for the TI PCILynx TSB12LV21A IEEE 1394 link layer
> > controller? We see the need to integrate our CompactPCI
> ...
>
> --
> Allen Walker                Signtech USA Ltd.
> http://www.signtech.com     4669 Highway 90 West
> +1 210.436.4777 x341        San Antonio, TX 78237




---------------------------

Newsgroups: comp.os.vxworks
Subject: How do I hook the Timer Interrupt on the Intel BSP?
Date: Thu, 18 Feb 1999 17:33:49 -0500
From: "Craig Gartside" <gartside@plc.com>
Organization: Programmed Logic
Message-ID: <7ai42s$fst$1@prologic.plc.com>

I want to put together a quick and dirty code profiler. I
would like to hook the timer interrupt (PIT 0) or PIT 1
so I can gather statistics on where my code is spending
it's time. I'm looking for finer granularity than the spyLib().

I need access to the interrupt stack so I can see the
address that was interrupted.

thx
Craig Gartside
gartside@plc.com



---------------------------

Newsgroups: comp.os.vxworks
Subject: re: Help!  Cannot see BREAKPOINTS in Tornado 1.0.1
Date: Fri, 19 Feb 1999 02:12:04 GMT
From: John or Jenn <lastimp@home.com>
Organization: Who, us organized???
Message-ID: <UL3z2.2990$7P6.1082@news.rdc1.az.home.com>
References: <rSux2.1260$yv3.841@news.rdc1.sfba.home.com> <F7D48H.nn@ecf.toronto.edu> <qw93e43v12t.fsf_-_@wasabi.lcs.mit.edu> <36CC8E46.EEEA7A8D@intotoinc.com>

In article <F7D48H.nn@ecf.toronto.edu> Ned Ning Ren, ren@ecf.toronto.edu
writes:
>I'm running Tornado 1.0.1 with MBX860 board. Whenever I set a breakpoint,
>it doesn't show the triangle mark as described in the manual, however, the
>break point is actually there since excution does stop. I have already
>enabled "attribute pane" in editor options. Would this be BSP related?
>some extra option I have to include in my BSP?

Are you on NT?  This is a known bug (which will be fixed by switching
to Tornado II).  If you switch to ³Mixed mode display², then back to
³source² it should show up, btw.

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Help! Cannot see BREAKPOINTS in Tornado 1.0.1
Date: Thu, 18 Feb 1999 18:50:01 -0800
From: "Steve Doiel" <nospam_steved@pacifier.com>
Message-ID: <36ccd001.0@news.pacifier.com>
References: <F7D48H.nn@ecf.toronto.edu>

You don't say what type of host you are using, but on NT Tornado gets
confused if there is an inconsistancy between the case on the file names.
Of course on UNIX you'll catch this kind of problem much earlier since the
OS is case sensitive.

I hope this helps,
SteveD

Ned Ning Ren wrote in message ...
>Hi,
>I'm running Tornado 1.0.1 with MBX860 board. Whenever I set a breakpoint,
>it doesn't show the triangle mark as described in the manual, however, the
>break point is actually there since excution does stop. I have already
>enabled "attribute pane" in editor options. Would this be BSP related?
>some extra option I have to include in my BSP?
>
>THank you!!
>



---------------------------

Newsgroups: comp.os.vxworks
Subject: How to read CMOS into VxWorks clock?
Date: Fri, 19 Feb 1999 03:41:26 GMT
From: "Marc Yaxley" <ohmy-gosh@home.com>
Organization: MySoftworks Co.
Message-ID: <G35z2.226$No.581@news.rdc1.sdca.home.com>

Has anyone out there read the PC's CMOS clock data into a program
then set VxWorks's system time/date with this data?   I didn't see any
thing to do this built in to the PC486 BSP or otherwise?

Thanks for any ideas you may have that will help me,
Marc





---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: timeouts with the VxWorks msgq utils
Date: Tue, 16 Feb 1999 15:39:11 GMT
From: bruce@CUTMEOUTscimisys.com (Bruce Wedding)
Organization: NeoSoft, Inc.  
Message-ID: <36c990f9.3202481@news.neosoft.com>
References: <36C9726F.91B2A75B@harris.com>

On Tue, 16 Feb 1999 08:28:15 -0500, Paul Allum <Pallum@harris.com>
wrote:

>Does anyone know how the msgQReceive function returns a timeout
>indication?

It returns ERROR with errno set to S_objLib_OBJ_TIMEOUT.



Bruce Wedding
Software Manager
bruce@scimisys.com
Scientific MicroSystems, Inc.
http://www.scimisys.com
Tomball, Texas

---------------------------

Newsgroups: comp.os.vxworks
Subject: Frame's structure in SENS
Date: Fri, 19 Feb 1999 08:59:34 +0100
From: "Fabio Bertoldi" <fbertoldi@mael.it>
Organization: Customer of Flashnet S.p.A. - http://www.flashnet.it
Message-ID: <7aj5jq$9ha$1@news.flashnet.it>

Hi everyone.

I have a problem with SENS MUX.
The purposes of the MUX is:

    1) decouple the protocol and network driver, thus making the network
driver and protocol nearly independent of each another (SENS for Tornado -
Component Release Supplement - Edition 1; pag.189);
    2) act as a MUX using the same link layer with more than one protocol.

Receiving a packet, the driver calls the function pointed by receiveRtn
(muxReceive()); this routine determines the protocol type of the packet.

My question is:

if frame's structure in link layer is neither an Ethernet nor 802.3
structure, how does muxReceive()  determine the protocol type of the packet
? Does SENS stack support ONLY Ethernet and 802.3 packets ?

Could someone please answer this question ?

Thanks,

    Fabio Bertoldi




---------------------------

Newsgroups: comp.os.vxworks
Subject: two ethernets and vxworks
Date: 19 Feb 1999 09:58:16 GMT
From: joerg@hera-b.desy.de (Joerg Rieling)
Organization: Deutsches Elektronen-Synchrotron (DESY)
Message-ID: <7ajcjo$m3i$1@claire.desy.de>

hi all,

  in vxworks, only one ethernet connection is foreseen. how do i
  proceed if i want to use two or more ethernet cards simultaneoulsy?

  thanks in advance for hints,

j. 

- -------------------------------------------------------------------
J. Rieling                    Tel:                 +49 421 539 5643 
DaimlerChrysler Aerospace   eMail:         joerg.rieling@ri.dasa.de
Bremen / Germany              PGP: www.desy.de/~joerg/mypubkey.html
- -------------------------------------------------------------------
(PGP signed message follows.)

- -----BEGIN PGP MESSAGE-----
Version: 2.6.3i
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=I79z
- -----END PGP MESSAGE-----

---------------------------

End of New-News digest
**********************


From owner-vxwexplo-process  Fri Feb 19 09:34:01 1999
From: Jim Caprio <caprio@research.moore.com>
Date: Fri Feb 19 09:34:05 PST 1999
Subject: Re: VxWorks memory partitions

From the number and tone of the responses my question has generated, it seems that
there's a strong current of opinion (if not an outright consensus) that VxWorks
does not implement memory allocation efficiently.  This is forcing my project to
evaluate a $6000 third party product for a feature which should be optimized for a
so-called real-time operating system.

I agree with David's comments below: this behavior should be considered a bug.
Does anyone know why Wind River doesn't just redesign their implementation?

> Submitted-by owner-vxwexplo-process  Fri Feb 19 01:26:02 1999
> Submitted-by: David Laight <dsl@tadpole.co.uk>
>
> IMHO the memory allocater in vxworks (memPartAlignedAlloc does the work)
> seems designed to fragment memory as much as humanly possible.
> As well as using a 'first fit' algorythm it compounds things by using a LIFO
> free list.
> The effect is that if you free a large piece of memory the next allocate will be
> from (the end of) that piece.
> (If the free list were in address order, then memory would not get as fragmented
> - but would still get fragmented.)
>
> I think this behaviour could be considerd a bug...
>
>         David



From owner-vxwexplo-process  Fri Feb 19 12:36:54 1999
From: "Bruce O. Benson" <benson@sparta.com>
Date: Fri Feb 19 12:36:57 PST 1999
Subject: VxWorks->Linux:  Last call for votes

Hello VxUsers,

I am forwarding all the votes for the VxWorks to Linux port survey today
to WRS, for inclusion in a meeting they are holding next week.

I have 95 votes now...if you haven't voted yet, please drop me a note
directly and help get the tally over 100. 
mailto:benson@mclean.sparta.com?subject=VxWorks->Linux

The results will be posted on http://linux.sparta.com/VxWorks

Thanks Again,
Bruce.
Bruce Benson, Senior Engineer,
SPARTA (CPD)
7926 Jones Branch Drive, Suite 900
McLean, VA  22102-3303
USA
+01 703-448-0210 x211
benson@sparta.com
http://www.mclean.sparta.com


From owner-vxwexplo-process  Fri Feb 19 13:04:24 1999
From: "Keith Buchanan" <Keith_Buchanan@SPARTA.com>
Date: Fri Feb 19 13:04:28 PST 1999
Subject: RE: VxWorks memory partitions

> >From the number and tone of the responses my question has 
> generated, it seems that there's a strong current of opinion 
> (if not an outright consensus) that VxWorks
> does not implement memory allocation efficiently.  This is 
> forcing my project to evaluate a $6000 third party product for 
> a feature which should  be optimized for a
> so-called real-time operating system.

I vehemently disagree, it is a tried and true, hard-core, deadline-driven,
real-time OS.  I have delivered many life and death types of systems where
VxWorks has performed in a deterministic manner w/o fail.  If you can't 
solve your allocation problems with existing tools, the design is suspect
not the operating system.

> I agree with David's comments below: this behavior should be 
> considered a bug.
> Does anyone know why Wind River doesn't just redesign their 
> implementation?

Geez, this is laying it on a little thick don't you think?  I tend to 
believe any designer who allocates and frees memory willy-nilly in a 
real-time application is a bug ;o).

You cannot get something for nothing.  The more elaborate the memory
allocation algorithms, the slower the functions that allocate and free
memory and the larger the footprint of the OS.  Allocate your resources
when your application starts up.  Use memory partitions for applications 
that must often allocate and free data structures.  Don't dynamically 
create and kill tasks for no good purpose.  Memory managment routines are 
slow in any API.  Why on earth would someone call them so often that 
you fragment memory?  If you have that kind of requirement, the buffers 
should be managed more efficiently.  If you want garbage collection, 
use SmallTalk running under Windoze.

Wind River had to compromise between complex alogrithms with a 
large footprint and the current implementation.  In addition, many of 
the choices for algorithms are not deterministic.  The functionality 
provided is reasonable and my opinion of it more typical.


adios

------------------------------------
      /\       Thomas Keith Buchanan 
     /**\      Principal Engineer
    /****\     SPARTA, Inc. 
   /****/..    205 East IH 30
  /****/....   Rockwall, TX 75087 
 .\****\__...  
...\**\  /.... Voice: 972 772 4487
....\**\/\.... FAX:   972 722 3201
 ....\****\..  PCS:   972 672 9657
  ....\****/   WWW:   www.sparta.com
   .../***/         
    ./***/     SPARTA              
     \**/      ======              
      \/       Pride In Performance




From owner-vxwexplo-process  Fri Feb 19 13:55:00 1999
From: bwaite@cspi.com
Date: Fri Feb 19 13:55:04 PST 1999
Subject: 2603 Booting from SCSI


VxWorks


I am having a problem reliably booting from a scsi disk of the MVME2603. We
find that when booting from SCSI occasionally (1 out of 10-20 times) the
kernel fails to load. Watching the boot sequence via a serial link we are
seeing that the kernel gets to the point of saying: Starting at 0x10000000
and hangs. The CPU light is solid green and never blinks. We have looked
into the go() routine within bootConfig.c and found that the hang occurs
when taskDelay(sysClkRateGet()) is called. We have insured that clock
interrupts are still being generated. By adding a taskSwitchHook,  I print
out when a task is switched and what the new task is. I only see one task
switch and that is a switch to tScsiTask. I never see anything else happen.
Also, by changing the taskDelay(sysClkRateGet()) to taskDelay(0) the system
boots correctly and no task are switched. I am not totally sure the problem
lies in the task switch but it looks that way. Has anyone else seen this
problem or anything similar? Any pointers on what to look at or how to fix
this would be greatly appreciated. We are using VxWorks 5.3.1 FYI. The
following is some details of the hang state and the code responsible:

---------------------------------------------------------------------------
----------------------------------------
Here is the output from a hung system:

Attaching network interface lo0... done.

boot device          : scsi=0,0
processor number     : 0
file name            : /sd0/tftpboot/vxWorks2610_bjw
inet on ethernet (e) : 60.0.0.111:ff000000
host inet (h)        : 60.0.0.254
user (u)             : cspiboot
flags (f)            : 0xc8
startup script (s)   : bootHost:/tftpboot/st2610_scsi

Waiting for scsi to spin up ...............
Attaching to scsi device... done.
Loading /sd0/tftpboot/vxWorks2610_bjw...1304596
Finish loading /sd0/tftpboot/vxWorks2610_bjw...
Starting at 0x100000...

Swap from fe64f8 to fc2738
Swap from  to tScsiTask

---------------------------------------------------------------------------
----------------------------------------
Here is the taskSwitchHook routine:
void PrintSwitch(WIND_TCB *pOldTcb, WIND_TCB *pNewTcb)
{
  printf("Swap from %x to %x\n",  pOldTcb, pNewTcb);
  fflush(stdout);

  printf("Swap from %s to %s\n" , taskName((int)pOldTcb), taskName((int)
pNewTcb));
   fflush(stdout);
 }

---------------------------------------------------------------------------
----------------------------------------
Here is the code from bootConfig.c go() routine:


    printf ("Starting at 0x%x...\n\n", (int) entry);
    fflush (stdout);

    taskSwitchHookAdd((FUNCPTR)PrintSwitch);

/* Commented out for sanity check. Insures sysClkRateGet is not the culprit
 */
/*    taskDelay (sysClkRateGet ());     /* give the network a moment to
close */
/* Replaces taskDelay(syClkRateGet()) with same delay time  (1 second) */
   taskDelay (60);     /* give the network a moment to close */

---------------------------------------------------------------------------
----------------------------------------


Thank You
Brian Waite
CSPI Customer Support
Email: bwaite@cspi.com




From owner-vxwexplo-process  Fri Feb 19 14:35:32 1999
From: Scott Peters <Scott.Peters@AGAI.com>
Date: Fri Feb 19 14:35:36 PST 1999
Subject: vxworks network IP address just started failing.

Hello vx-ers:

We  just started to getting the following message on all our machines:
0xfc95b8 (tNetTask): arp: ether address is broadcast for IP address a010547!
We have not change the software and we even disconnect from our company Lan.
We just started seeing this as of 2/18/97.  

Has anybody seen this?  

Please email me directly, I am not current on the exploder.

Thank you very much
Scott.Peters@agai.com




From owner-vxwexplo-process  Fri Feb 19 19:24:23 1999
From: Fred Roeber <froeber@BBN.COM>
Date: Fri Feb 19 19:24:27 PST 1999
Subject: Re: VxWorks memory partitions

On Fri, 19 Feb 1999, Jim Caprio <caprio@research.moore.com> wrote:
  
> From the number and tone of the responses my question has generated,
> it seems that there's a strong current of opinion (if not an outright
> consensus) that VxWorks does not implement memory allocation efficiently.
> This is forcing my project to evaluate a $6000 third party product for
> a feature which should be optimized for a so-called real-time operating
> system.
> 
> I agree with David's comments below: this behavior should be considered a bug.
> Does anyone know why Wind River doesn't just redesign their implementation?

Jim was commenting about a "bug" in the VxWorks memory allocation algoritm.
Having analyzed memory allocation algorithms in some depth and having gone
through the VxWorks memory allocation code, I feel qualified to make a
comment.

There are numerous memory allocation algorithms floating around with
significantly different execution and memory efficiencies. The Wind River
one is not elegant but it is well implemented and fairly widely used.
It isn't optimized for use in a time critical system, however. The normal
use of memory allocation in real-time systems is to allocate all dynamic
memory areas during initialization or as part of a background task. If
you have an application that allocates and frees a lot of memory during
operation then that isn't the type of application that the VxWorks
allocator is designed for. There are some alternatives you might consider.

If you don't care about efficient memory use then one of the most time
efficient allocators is a fixed block size allocator. There is code for
such an algorithm in the VxWorks archive (called poolLib if I remember).
If you need to handle different size memory blocks then an algorithm
we have had good success with is the "binary buddy algorithm". A variant
of it is used in various GNU programs so is easily available. We found
that it could be written in such a fashion that it was usable by multiple
tasks concurrently with a deterministic maximum run time (the VxWorks
allocator is single threaded and non deterministic).

Getting an allocation algorithm to work correctly and robustly is not
trivial, however. You may be better off buying one unless you have time
to check out algorithms in the literature (some good sources are "Data
Structures & Their Algorithms" by Lewis and Denenberg and one of the Knuth
algorithm books; can't remember which volume as someone has borrowed it).
I have heard good things about allocation code for VxWorks from Seaweed
Systems but have never used it (query info@seaweed.com).

I guess my main point is that the VxWorks algorithm may not be optimal but,
then, I don't think there is any one algorithm that is optimal for all
applications. Real time systems that do significant dynamic memory allocation
during operation need careful design to avoid fragmentation problems. Most
applications are well advised to avoid the problem by preallocating memory.

Well, that's my $.02.  Fred

   |      Fred J Roeber,  BBN Systems & Technologies      |
   |    4 John Clarke Road   Middletown, RI  02842-5202   |
   |          froeber@bbn.com     401-848-3548            |
   | TraceMaker product manager -> www.tracemaker.bbn.com |


From owner-vxwexplo-process  Sat Feb 20 04:01:04 1999
From: daemon@csg.lbl.gov
Date: Sat Feb 20 04:01:08 PST 1999
Subject: comp.os.vxworks newsdigest

Comp.Os.Vxworks Daily Digest    Sat Feb 20 04:01:00 PST 1999

        Subject: Re: VxWorks memory partitions
        Subject: Any experience with large vxWorks.st_rom.hex ?
        Subject: PCI bus and device drivers....
        Subject: Re: RAIMA DB or POLYHEDRA DB on VxWork.
        Subject: Re: C++, overloading global new/delete
        Subject: Target-based loader problem
        Subject: Re: determining how many bytes currently in TCP buffer
        Subject: Re: How do I hook the Timer Interrupt on the Intel BSP?
        Subject: Re: How to read CMOS into VxWorks clock?
        Subject: Re: Target-based loader problem
        Subject: Re: VxWorks memory partitions
        Subject: Ata drive setup
        Subject: tTelnetOutTask hogs CPU ....
        Subject: Re: How to read CMOS into VxWorks clock?
        Subject: System mode debugging:  NT and  BSP486
        Subject: Re: determining how many bytes currently in TCP buffer
        Subject: Re: How to get IP of Gateway and own IP?
        Subject: priorityText Example
        Subject: schedPxLib
        Subject: re: mcp750 bsp broblem - vxWorks 5.3.1
        Subject: Re: priorityText Example
        Subject: Re: Frame's structure in SENS
        Subject: Re: PCI bus and device drivers....
        Subject: Are linear addresses physically contiguous in vxworks 486 BSP?.
        Subject: Re: IP address change..
        Subject: Re: Target Server Error Report
        Subject: Re: VxWorks memory partitions
        Subject: Re: Are linear addresses physically contiguous in vxworks 486 BSP?.
        Subject: Re: Problem with MPC860SAR
        Subject: Help! Cannot see BREAKPOINTS in Tornado 1.0.1

-------------------------------------------------------

Newsgroups: comp.os.vxworks
Subject: Re: VxWorks memory partitions
Date: Fri, 19 Feb 1999 14:56:58 +0300
From: Leonid Rosenboim <leonid@bitband.com>
Organization: BitBand Technologies Ltd.
Message-ID: <36CD5189.28EB49ED@bitband.com>
References: <199902190926.JAA08729@dsl-2.tadpole.co.uk>

This behaviour of VxWorks memory allocator is know for a long time,
increasing number of customers find this not acceptable:
In the past, mot hard real-time systems where designed such that all
object allocations are done during system initialization, and very
little or no allocations of memory at all happen during the critical operations
time.

This is changing with the advent of OO techniques, hence the allocation
mechanism which used to be satisfactory, is not so anymore.

WRS seems to ignore this trend, rightly claiming the the primary competitor
(i.e. pSOS) is no better...

There is a solution from SeaWeed systems, who sell a replacement library for
memPartLib which does much more stuff.
Some customers have devised some smart techniques which create sub-partitions
for specific block sizes and add a layer on top of memAlloc() and malloc()
which chooses a partition according to block size, and that in many cases solves
the issue too.

HTH,
 Leonid

David Laight wrote:

> IMHO the memory allocater in vxworks (memPartAlignedAlloc does the work)
> seems designed to fragment memory as much as humanly possible.
> As well as using a 'first fit' algorythm it compounds things by using a LIFO
> free list.
> The effect is that if you free a large piece of memory the next allocate will be
> from (the end of) that piece.
> (If the free list were in address order, then memory would not get as fragmented
> - but would still get fragmented.)
>
> I think this behaviour could be considerd a bug...
>
>         David
>
> > Subject: VxWorks memory partitions
>
> > Submitted-by: Jim Caprio <caprio@research.moore.com>
> >
> > BSP: mv2300 (604e PPC)
> >
> > I'm interested in some general information on how to optimize memory
> > usage in VxWorks.  We're running into a memory fragmentation problem,
> > which is caused in part to multiple threads concurrently allocating
> > memory from the same memory partition (the default heap).  Due to
> > VxWorks "first fit" (vs. best fit) allocation algorithm, the result is
> > inefficient placement of objects in memory, and resultant fragmentation.
> >
> > Several candidate solutions have been proposed:
> > (a) Assign (pre-allocated?) memory partitions on a per task basis
> > (b) Create memory buffers for homogenous object size
> > (c) Modify the VxWorks memory allocation algorithm to make better use of
> > available space
> >
> > To your knowledge, are any of these options possible in VxWorks?  A
> > brief explanation of how to implement the required modifications, and
> > some motivation behind the "best" approach would be greatly
> > appreciated.   Thanks for the help.
> >
> > Jim Caprio
>
> ----------------------------------------------------------------
> David Laight                          email: dsl@tadpole.co.uk
> Tadpole Technology plc                phone: +44 1223 278 256
> Cambridge, UK                         fax:   +44 1223 278 201


---------------------------

Newsgroups: comp.os.vxworks
Subject: Any experience with large vxWorks.st_rom.hex ?
Date: Fri, 19 Feb 1999 14:17:55 +0100
From: Sebastien Mallet <mallet@nortel.ca>
Organization: Nortel
Message-ID: <36CD6483.4D417524@nortel.ca>

Hi all,
I'm using Motorola 860 custom BSP.
Previously I used a minimal configuration with no problem at all.
Now with adding a consequent amount of code, I crash the board.
My problem is when inflating the BSP into RAM I got a bus error.
I tried different memory configuration with no success (changing
RAM_HIGH_ADRS).
Any help would be really appreciated.
Sebastien Mallet

---------------------------

Newsgroups: comp.os.vxworks,comp.unix.programmer
Subject: PCI bus and device drivers....
Date: Fri, 19 Feb 1999 15:23:25 +0200
From: Bahri Okuroglu <bahrio@rnd.netas.com.tr>
Organization: Netas (Northern Electric Telecom AS)
Message-ID: <36CD65CC.77B86A3B@rnd.netas.com.tr>

Hi,

I have devices on the PCI bus. Both the cards and the board is custom,
and I have to develope all the required drivers.

I think I need to develope drivers for the cards on the PCI bus. If so,
can you please tell me what a PCI driver does? (There exist a driver
(named PCI Driver : pciLocalBus.c) in the list of "WRS Device Driver
List".)

Can I control all the cards on my PCI bus only with a PCI bus driver.
(without developing specific drivers for the cards)???

I would be please if you can help me.

- --

=====================================================
Bahri OKUROGLU

Netas R&D                       RT6 - Software Design
mailto:bahrio@netas.com.tr    http://www.netas.com.tr
mailto:bahrio@yahoo.com
http://193.140.184.5/r748/member/bahrio    (internal)
Netas   Alemdag Cad.   Umraniye 81244 ISTANBUL TURKEY
=====================================================




---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: RAIMA DB or POLYHEDRA DB on VxWork.
Date: Wed, 17 Feb 1999 12:30:51 -0800
From: "Dave Morse" <davem@polyhedra.com>
Organization: Posted via RemarQ, http://www.remarQ.com - Discussions start here!
Message-ID: <7af8p8$77d$1@remarQ.com>
References: <36C944FF.758F2A5D@icn.siemens.de>


Dear Marco,
Depends on exactly what you are doing but Polyhedra can do FT for you on
VxWorks and supports multiple servers running together.  Raima doesn't
support replication on RDM for VxWorks and it is not a real-time C/S
database.
Dave Morse
Marco Bernasconi <Marco.Bernasconi.GP@icn.siemens.de> wrote in message
news:36C944FF.758F2A5D@icn.siemens.de...
>I would like to use a Data Base, Raima , Polyhedra. or .. I don´t know,
>on VxWork.
>Someone know something? I need to "replicate" data on different boards,
>can these DB help me or exist a client/server  real time DB on VxWork?
>May be free source.
>
>Thanks in advance,
>Marco
>
>
>



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: C++, overloading global new/delete
Date: Wed, 17 Feb 1999 01:22:57 GMT
From: coryrad@my-dejanews.com
Organization: Deja News - The Leader in Internet Discussion
Message-ID: <7ad5l9$qv8$1@nnrp1.dejanews.com>
References: <36C96BB7.DEEEF022@email.com>

Have you tried to remove the C++ cores from the library that define the
symbols you want? Granted, you will have to provide the rest of the functions
that exist in the object yourself, but it seems like the safest thing to do.
Use the utility "ar" to extract an object from the library, and insert your
own.

In article <36C96BB7.DEEEF022@email.com>,
  Chris Varlese <no@email.com> wrote:
> My setup is Tornado v1.0.1 with a MC68040 cross compiler.
>
> I have some trouble to build a ROM image when I have overloaded the
> global C++ new and delete operators.  The linker reports a clash between
>
> the new/delete we have defined and the standard new/delete in the gnu
> C++ library.
>
> Does anyone know how to solve this problem?  I looked at the manual
> for linker options but I don't see how to tell the linker to ignore the
> _builtin_new and builtin_delete symbols from the gnu library.  I wonder
> how normal compilers deal with this?  (e.g. GNU cygwin can happily
> link code where you have overloaded new/delete;  but you don't see
> what it is doing in the linking process).
>
> Chris Varlese
> Ascom Transmission AG
>
>

- -----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

---------------------------

Newsgroups: comp.os.vxworks
Subject: Target-based loader problem
Date: Fri, 19 Feb 1999 09:02:54 -0500
From: Ron Daubenspeck <dauben@nortelnetworks.com>
Organization: Nortel Networks, Raleigh, NC
Message-ID: <36CD6F0E.265B761E@nortelnetworks.com>

VxWorks is running on PPC platform.

I'm trying to load a utility from local disk using either loadModule()
or ld() from my application, but I get undefined references for all the
basic I/O stuff; ie, printf, open, close, etc.

BUT...
I can load this very same utility from windsh using ld() and it works
great.  What's the difference in loading from windsh versus loading
from an application program running on the target?  Do I need to build
the utility differently somehow?  It's currently being built using ldppc
with the -r option (relocatable).

Any help is greatly needed, and appreciated.

- -- 
Ron Daubenspeck                      dauben@nortelnetworks.com
Nortel Networks                      919/905-4683
Research Triangle Park, NC           919/905-8313 FAX

---------------------------

Newsgroups: comp.unix.programmer,comp.os.vxworks
Subject: Re: determining how many bytes currently in TCP buffer
Date: 19 Feb 1999 16:10:25 GMT
From: peter@icke-reklam.netman.se.nu
Organization: CR&T / Network Management
Message-ID: <7ak2dh$cqp$1@nyheter.carlstedt.se>
References: <36cf7a2f.343908023@sislnews.csc.ti.com>

In comp.unix.programmer Gene Bruce <gbruce@ti.com> wrote:


> I have a TCP socket connection between two computers. In the case of a physical
> disconnection of the connection, how can I tell how much data is backed up in my
> TCP send buffer? My write calls will eventually block when the buffer fills, but
> I really need to be able to detect and handle the problem before this happens,
> preferably by somehow flushing the buffer to delete messages which are time
> sensitive and cause more problems than they solve by being sent too late. (Once
> the connection is physically re-established, all messages in the buffer are
> transmitted - but they are now out of date.)

> I know that physical disconnection is not likely, but I need by application to
> be very robust to keep the radar system that this code resides in running very
> smoothly.

> I've looked at the Stevens book without success for answers. But maybe someone
> knows chapter and verse in the book that I missed or knows of a different source
> of information concerning this problem.

If you write with ioctl(,FIONBIO, ) (page 322 in stevens book!) you can test if the
write is successful. If not errno might contain EWOULDBLOCK.



> Gene Bruce                       Ph/Voice Mail: (972)952-4937
> Raytheon Systems Company         FAX: (972)952-4275
> 2501 W. University, MS 8056      McKinney, Texas 75070

> "The inherent vice of capitalism is the unequal sharing of
> blessings; the inherent virtue of socialism is the equal
> sharing of miseries." -- Winston Churchill

- -- 
- --
Peter Håkanson            Phone +46 0708 39 23 04
Network Management AB     Fax   +46 031 779 7844
Email : use peter (at) gbg (dot) netman (dot) se  No copy to sanford wallace!

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: How do I hook the Timer Interrupt on the Intel BSP?
Date: Fri, 19 Feb 1999 17:07:43 +0000
From: Phil Ashby <phil.ashby@wireplay.bt.co.uk>
Organization: BT Wireplay
Message-ID: <36CD9A5F.13C0@wireplay.bt.co.uk>
References: <7ai42s$fst$1@prologic.plc.com>

Craig Gartside wrote:
> I want to put together a quick and dirty code profiler. I
> would like to hook the timer interrupt (PIT 0) or PIT 1
> so I can gather statistics on where my code is spending
> it's time. I'm looking for finer granularity than the spyLib().

I would start by cranking the system clock rate up to the
resolution you need (eg: 1000), then attach a routine to
the right interrupt (intHandlerCreate/intVecSet). The register
set should be available on the stack (you will have to do
some reverse engineering on intHandlerCreate to see what the
stack frame is).

You will need to retain the original interrupt handler and
call out to it or the kernel will freeze (use intVecGet).

HTH,
- -- 
!--------------------------- Phil "Phlash" Ashby ---------------------!
! BT Wireplay - Lead Programmer and general technical trivia supplier !
! Phone: 01473-644348                   WWW: http://www.wireplay.com/ !
! Snail: B81/G41, BT Labs, Martlesham Heath, Ipswich IP5 3RE, England.!

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: How to read CMOS into VxWorks clock?
Date: Fri, 19 Feb 1999 17:18:40 +0000
From: Phil Ashby <phil.ashby@wireplay.bt.co.uk>
Organization: BT Wireplay
Message-ID: <36CD9CF0.3EE2@wireplay.bt.co.uk>
References: <G35z2.226$No.581@news.rdc1.sdca.home.com>

Marc Yaxley wrote:
> 
> Has anyone out there read the PC's CMOS clock data into a program
> then set VxWorks's system time/date with this data?   I didn't see any
> thing to do this built in to the PC486 BSP or otherwise?

It is certainly possible to read the CMOS clock. Use the I/O space
byte transfer routines (sysInByte/sysOutByte) to access the CMOS at
I/O address 0x70/0x71. You will need a PC hardware reference to get
all the info you need about the CMOS..

Does VxWorks have a system time/date? I can only find the POSIX
routines (clock_settime) which are optional.

I tried calling time() on my PC board which has been up for a few
hours. I got 0. I tried again. I got 3..... useful huh!
- -- 
!--------------------------- Phil "Phlash" Ashby ---------------------!
! BT Wireplay - Lead Programmer and general technical trivia supplier !
! Phone: 01473-644348                   WWW: http://www.wireplay.com/ !
! Snail: B81/G41, BT Labs, Martlesham Heath, Ipswich IP5 3RE, England.!

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Target-based loader problem
Date: Fri, 19 Feb 1999 10:42:25 -0800
From: Hwa Jin Bae <hjb@pso.com>
Organization: PSO Systems Inc
Message-ID: <36CDB091.B79B8E26@pso.com>
References: <36CD6F0E.265B761E@nortelnetworks.com>

The main difference is that the host ld has access to symbol table
information locally hosted, while the target ld has to have
in-core symbol table access, which must be enabled.  Look in
your configdb.h or config.h.

- -- 
Hwa Jin Bae
PSO Systems Inc
mailto:hjb@pso.com
http://www.pso.com

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: VxWorks memory partitions
Date: Fri, 19 Feb 1999 11:10:26 -0800
From: Hwa Jin Bae <hjb@pso.com>
Organization: PSO Systems Inc
Message-ID: <36CDB722.1E7B6549@pso.com>
References: <199902190926.JAA08729@dsl-2.tadpole.co.uk> <36CD5189.28EB49ED@bitband.com>

As Leonid said this memory fragmentation problem has been with VxWorks
for a long time, and it is in general prevalent in most non VM oriented
systems.  

The strategy for avoiding problems in this area has always been: know
what you are allocating and freeing.  By careful coding you can avoid
memory fragmentation.   This is the view that was expressed to me by the
original author of VxWorks memLib when I complained about memLib
fragmentation issue over 10 years ago.  In fact, malloc/free, etc. are
considered a useful add-on, not as a core facility to be used in hard
realtime systems.  My views obviously differ on this.

The problem is that this issue can be eased with simple modifications to
the memLib itself without adding costly penalties.  And another issue is
that this is a *real* problem.  I have seen numerous projects that
duplicate needless hours of coding to avoid this problem, or fix this
problem in their own way.

One of the solutions I have come up with over the years is to replace
the memLib with another implementation.  A good alogorithm that seems to
behave much better than VxWorks memLib is described in 
http://g.oswego.edu/dl/html/malloc.html   which is a public domain 
implementation.   This has been ported in VxWorks and has served well
for at least one project that did massive malloc/free's (due to ported
Unix code behavior that could not be prevented).  

Unfortunately, the way the port is done requires VxWorks source code,
this is due to many problems that has to do with link time errors --
they do not make it easy to replace the memLib!  So this port is not
availble PD.  Perhaps someday.

The partitioning scheme, as Leonid suggests, solves some problems and is
quite useful as well.  Unfortunately, VxWorks memory partitions, once
created cannot be destroyed.  This makes it less desirable.  Plus it
requires a lot of changes in ported Unix code base and is not as easy to
do.

- -- 
Hwa Jin Bae
PSO Systems Inc
mailto:hjb@pso.com
http://www.pso.com

---------------------------

Newsgroups: comp.os.vxworks
Subject: Ata drive setup
Date: Fri, 19 Feb 1999 11:17:33 -0800
From: "Glenn Dicus" <gdicus@nomadix.com>
Organization: Nomadix LLC
Message-ID: <7akd2i$1im$1@la-mail4.digilink.net>

I am currently trying to install a 1443 MB Toshiba  ata drive.  After
defining INCLUDE_ATA I call

>usrAtaConfig (0, 0, "/");

I get the following error.

0x1f21ff4 (t11): memPartAlloc: block too big - 889091160 in partition
0x1abb38.
Error during dosFsDevInit: 380001

The drive is formatted for FreeBSD, does this matter?  Not that I need to
save the data, but can I do it this way and then format the drive under
VxWorks?



---------------------------

Newsgroups: comp.os.vxworks
Subject: tTelnetOutTask hogs CPU ....
Date: Fri, 19 Feb 1999 14:14:27 -0500
From: Tom Kam <tkam@newbridge.com>
Organization: Newbridge Networks Corporation
Message-ID: <36CDB813.6ACE9DFA@smt.net>

when a telnet session to the shell is killed and
has a priority at least that of the tShell task. 
If I lower tTelnetOutTask priority, everything is
fine.

When telnet is killed, tExecTask is running at
almost 100% CPU, executing the function qJobGet. 
Does anybody know why this is happening?  How can
I fix it other than changing the task priority? 
I'm using VxWorks V5.2, Kernel V2.4.


Thanks
Tom Kam
tkam@smt.net

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: How to read CMOS into VxWorks clock?
Date: Fri, 19 Feb 1999 14:30:08 -0500
From: "Steve Robertson" <Srobertson@automationintelligence.com>
Organization: PSINet
Message-ID: <7ake9o$7lr$1@client2.news.psi.net>
References: <G35z2.226$No.581@news.rdc1.sdca.home.com>

Marc,

WRS sent me the file ds12887rtc.c which I used as the basis for reading the
real-time clock.   Armed with this and a spec sheet for Dallas
Semiconductors' 12887 chip (downloaded from their web page), I was able to
read and write to the RTC.  It seems like I made a few changes to their code
but I don't recall what they were right now.

Next you need to set the VxWorks clock so that all the "time" functions in
the ansiTime library will work correctly.  With all disclaimer's that could
possibly apply, here is what I did:

void SysTimeInit( void )
{
 struct tm  tmTime;
 struct timespec tsTime;

 sysRtcGet( &tmTime );         /* get time from RTC  */

 tsTime.tv_sec = mktime( &tmTime );      /* convert to timespec  */
 tsTime.tv_nsec = 0;

 clock_settime( CLOCK_REALTIME, &tsTime );    /* update VxWorks clock  */
}

Note that "sysRtcGet" is a function in ds12887rtc.c.

Good Luck
Steve Robertson


Marc Yaxley wrote in message ...
>Has anyone out there read the PC's CMOS clock data into a program
>then set VxWorks's system time/date with this data?   I didn't see any
>thing to do this built in to the PC486 BSP or otherwise?
>
>Thanks for any ideas you may have that will help me,
>Marc
>
>
>
>



---------------------------

Newsgroups: comp.os.vxworks
Subject: System mode debugging:  NT and  BSP486
Date: Fri, 19 Feb 1999 14:49:41 -0500
From: "Steve Robertson" <Srobertson@automationintelligence.com>
Organization: PSINet
Message-ID: <7akfeb$80a$1@client2.news.psi.net>

I am trying to use system mode debugging (as opposed to task level
debugging).  I can successfully do this with one module; however, it is
unsuccessful if I download more than one module.  I get the same error no
matter what method I try.  It is:

"The instruction at 0x7801CD84 referenced memory at 0x00000000.  The memory
could not be read."

I have tried a number of things but this is the simplest:

(gdb) load file1                        // load file1
(gdb) load file2                        // load file2
(gdb) attach system                // enable system mode debugging
(gdb) b file1Sub1                    // set a breakpoint in one of file1's
functions
(gdb) wind-sp 0x12345678   // 0x12345678 = file1's entry point
(gdb) c                                     // start execution at file1's
entry point
(gdb) wind-sp 0x87654321  // 0x87654321 = file2's entry point
(gdb) c                                    // start execution at file2's
entry point

This is when I get the error.  BTW, it is not really feasible to link file1
and file2 together.  Thanks for any help.

Steve Robertson



---------------------------

Newsgroups: comp.unix.programmer,comp.os.vxworks
Subject: Re: determining how many bytes currently in TCP buffer
Date: 19 Feb 1999 17:21:03 GMT
From: jhalpin@nortel.com.fava.beans
Organization: Bell-Northern Research, Richardson, TX
Message-ID: <7ak6hv$42m$1@crchh14.us.nortel.com>
References: <36cf7a2f.343908023@sislnews.csc.ti.com>

In article <36cf7a2f.343908023@sislnews.csc.ti.com>, Gene Bruce
<gbruce@ti.com> wrote: 

>I have a TCP socket connection between two
>computers. In the case of a physical disconnection of the connection,
>how can I tell how much data is backed up in my TCP send buffer?

There isn't any way to do this, short of maybe poking around in the
kernel. Once you call write(), the data is owned by the protocol
stack. I don't think it would help you anyway, for the reasons I give
below. 

>My write calls will eventually block when the buffer fills, but I
>really need to be able to detect and handle the problem before this
>happens,

The reason that the send buffer fills up is that 1) the receiver
has told the sender (through its window advertisements) that it
doesn't have any more room in its receive buffer, or 2) network
problems are preventing the receiver's acks from reaching the sender. 

In the first case it won't do any good to flush the send buffer (even
if you could) because the receiver isn't reading its input fast
enough. 

The second case is pretty much the same, only for a different
reason. The sender will only send a max of a window to the receiver
until the receiver's acks show that it has more room in its receive
buffer. If the sender isn't getting those acks, he won't send more
even if the buffer is flushed. It doesn't sound like this is the case
though, as you say later that closing and reopening the connection
will get more data through.

If you run tcpdump and monitor the connection, you can see if the
receiver is sending 0 window advertisements. If it starts doing that,
the problem is that the receiver just isn't keeping up with the
sender. If that's the case, it might be better to concentrate on the
receiver, and find out what the problem is there. It might be that an
optimization in that program would get it to keep up with the sender.

Joe

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: How to get IP of Gateway and own IP?
Date: Wed, 17 Feb 1999 09:50:00 -0700
From: Mike Stimpson <mstimp@utsci.com>
Organization: Utah Scientific, Inc.
Message-ID: <36CAF338.41C6@utsci.com>
References: <NGQw2.1665$Qf5.12593@typhoon-sf.pbi.net> <slrn7c89mn.f41.chc@nasa2.ksc.nasa.gov> <36C46876.41C6@utsci.com> <G_ty2.4693$bP2.46071@typhoon-sf.pbi.net>

gto wrote:
> 
> Thanks. Works great. I found out that bootParams.ead returns the network
> interface IP and, as you have mentioned, bootParams.gad returns the Gateway
> IP.
> 
> But where can I find the complete description of this BOOT_PARAMS structure?
> I looked under bootLib and bootStringToStruct() in my VxWorks reference
> manual but could not find anything useful regarding BOOT_PARAMS.
> 
> Gregor

I find it in bootLib.h in the vxWorks include directory (for me,
/usr/wind/target/h - your location may vary).

HTH.
- -- 
Mike Stimpson (mstimp@utsci.com)

Opinions expressed are not necessarily those of the management...

The more I work as an engineer, the more I disbelieve the theory of
evolution.

---------------------------

Newsgroups: comp.os.vxworks
Subject: priorityText Example
Date: Fri, 19 Feb 1999 16:49:35 -0800
From: "Joseph J. Lindula" <jlindula@ecst.csuchico.edu>
Organization: California State University, Chico
Message-ID: <Pine.HPP.3.95.990219164418.9840A-100000@steroid.ecst.csuchico.edu>

Hello, I'm new to the VxWork system and need some help. I'm tying to do
the
example on page 81 of the WindRiver VxWorks Programmer's Guide 5.2. When I
load the object file priorityText.o onto the target machine I get this
error message.

Undefined symbols:
_sched_setparam
_sched_getparam
Warning: object module may not be usable because of undefined symbols.
value = 3170864 = 0x306230


I have included sched.h which should have defined these symbols.

Does anyone have any suggestions?

Thanks,

Joe Lindula 

heyjoejoe@yahoo.com


---------------------------

Newsgroups: comp.os.vxworks
Subject: schedPxLib
Date: Fri, 19 Feb 1999 17:02:29 -0800
From: "Joseph J. Lindula" <jlindula@ecst.csuchico.edu>
Organization: California State University, Chico
Message-ID: <Pine.HPP.3.95.990219165858.9840B-100000@steroid.ecst.csuchico.edu>

Hello, I'm a novice with VxWorks  so this maybe a stupid question. Is
schedPxLib a library and if it is can I download it? The problem I'm
having that I'm getting undefined symbols when I try to run the examples
in the VxWorks Programmer's Guide 5.2 pages 81 and 83.

If anyone has any suggestion, please let me know.



heyjoejoe@yahoo.com


---------------------------

Newsgroups: comp.os.vxworks
Subject: re: mcp750 bsp broblem - vxWorks 5.3.1
Date: Wed, 17 Feb 1999 04:13:56 GMT
From: stevebj@inxpress.net
Organization: Deja News - The Leader in Internet Discussion
Message-ID: <7adflt$3gc$1@nnrp1.dejanews.com>
References: <78o6lm$9uo35@overload.lbl.gov>

Larry,

We have a PCI driver that interrogates the cPCI bus/device/function and maps
everything out from the mcp750 controller.  Granted it was alot of extra work,
but we've been working with the board for 6 months or so.

SBJ

www.cp.artesyn.com


In article <78o6lm$9uo35@overload.lbl.gov>,
  Larry Sinn <sinnl@sd-star.com> wrote:
> Bill wrote:
>
> > Submitted-by owner-vxwexplo-process  Fri Jan 22 15:08:42 1999
> > Submitted-by: Bill Brown <wbrown@brownie.als.lbl.gov>
> >
> > I'm just starting out at working with Compact PCI.  At the moment, I'm
trying
> > to get a Motorola mcp750 cpu to talk to (or even admit the existance of!) a
> > GreenSpring cPCI-200 I.P. carrier which is plugged into the slot next to the
cpu
> >
> > After boot, the console tells me the following:
> >
> >     -> pciDeviceShow 0
> >     Scanning function 0 of each PCI device on bus 0
> >     Using configuration mechanism 1
> >     bus       device    function  vendorID  deviceID  class
> >     00000000  00000000  00000000  00001057  00004801  00060000
> >     00000000  0000000b  00000000  00001106  00000586  00060100
> >     00000000  0000000e  00000000  00001011  00000009  00020000
> >     00000000  00000014  00000000  00001011  00000026  00060400
> >     value = 0 = 0x0
>
> I'm working with a TechnoBox 2044 PMC SCSI on a Motorola mvme2606 and it shows
up on
> Bus 0.
> Have you done a pciDeviceShow without any IP boards installed?
>
> Larry.
>
> Larry Sinn
> Spectral Dynamics               (408)474-1746  voice
> 1983 Concourse Drive            (408)474-1780  fax
> San Jose, Ca. 95131-1708        sinnl@sd-star.com
>
>

- -----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: priorityText Example
Date: Fri, 19 Feb 1999 17:21:17 -0800
From: "David Wiggins" <david.k.wiggins@cpmx.saic.com>
Message-ID: <74995351@NEWS.SAIC.COM>
References: <Pine.HPP.3.95.990219164418.9840A-100000@steroid.ecst.csuchico.edu>

Joe,

The first important thing to realize (if you don't already) is that vxWorks
dynamically links your applications when you DOWNLOAD them to your target.
At compile/link time, undefined external references are NOT resolved.  So,
the fact that you included sched.h only allows the compiler to find function
prototypes and any types, etc. that are needed, but this does NOT link in
the object code for any system functions that are prototyped in the '.h'.

Having said all that, the reason you're getting those two undefined symbols
is because they aren't built into your vxWorks image.  To do that, and I'm
just making an educated guess here, try doing a '#define
INCLUDE_POSIX_SCHED' in your config.h or configdb.h file (you may only have
config.h in 5.2, I'm running 5.3.1 and it's in configdb.h in that revision)
in your target bsp directory.  Then re-make your vxWorks image and reboot
your target.  That should (I think) build these two functions in your
vxWorks image so that when you download your '.o' file it can resolve the
references.

I hope this helps.  Good luck.
Dave Wiggins - SAIC
david.k.wiggins@cpmx.saic.com

Joseph J. Lindula wrote in message ...
>Hello, I'm new to the VxWork system and need some help. I'm tying to do
>the
>example on page 81 of the WindRiver VxWorks Programmer's Guide 5.2. When I
>load the object file priorityText.o onto the target machine I get this
>error message.
>
>Undefined symbols:
>_sched_setparam
>_sched_getparam
>Warning: object module may not be usable because of undefined symbols.
>value = 3170864 = 0x306230
>
>
>I have included sched.h which should have defined these symbols.
>
>Does anyone have any suggestions?
>
>Thanks,
>
>Joe Lindula
>
>heyjoejoe@yahoo.com
>



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Frame's structure in SENS
Date: Sat, 20 Feb 1999 01:43:09 GMT
From: SomeGuy@SomethingOrOther.com (Some Guy)
Organization: Cisco Systems, Inc.
Message-ID: <36cf1219.202252884@news.cisco.com>
References: <7aj5jq$9ha$1@news.flashnet.it>

On Fri, 19 Feb 1999 08:59:34 +0100, "Fabio Bertoldi"
<fbertoldi@mael.it> wrote:

>? Does SENS stack support ONLY Ethernet and 802.3 packets ?

WindRiver admitted to us that this is so.

So what we do now is we emulate ethernet over our hardware (one shared
memory interface and one proprietary bus). When SENS gives us an
ethernet packet we put a wrapper around it and push that down. When a
wrapped packet comes in we unwrap it and give it to SENS.


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: PCI bus and device drivers....
Date: Sat, 20 Feb 1999 01:38:20 GMT
From: SomeGuy@SomethingOrOther.com (Some Guy)
Organization: Cisco Systems, Inc.
Message-ID: <36ce10c6.201913556@news.cisco.com>
References: <36CD65CC.77B86A3B@rnd.netas.com.tr>

On Fri, 19 Feb 1999 15:23:25 +0200, Bahri Okuroglu
<bahrio@rnd.netas.com.tr> wrote:

>I have devices on the PCI bus. Both the cards and the board is custom,
>and I have to develope all the required drivers.
>
>I think I need to develope drivers for the cards on the PCI bus. If so,
>can you please tell me what a PCI driver does? (There exist a driver
>(named PCI Driver : pciLocalBus.c) in the list of "WRS Device Driver
>List".)
>
>Can I control all the cards on my PCI bus only with a PCI bus driver.
>(without developing specific drivers for the cards)???

No.

If you have custom PCI boards you can use the supplied functions to
find and configure your devices, but you still have to write the
drivers.


---------------------------

Newsgroups: comp.os.vxworks
Subject: Are linear addresses physically contiguous in vxworks 486 BSP?.
Date: Fri, 19 Feb 1999 18:45:54 -0800
From: "rvsaripa" <ramakrishna_saripalli@intel.com>
Organization: Intel Corporation
Message-ID: <7al9qs$qa0@news.or.intel.com>

I have a 486 BSP question.

I know paging is not enabled by default.

Does this mean that if I have a piece of memory that is contiguous in linear
address space, it is also contiguous in physical address space?.

I ask because my hardware needs to be programmed with starting address of a
memory block and expects it to be contiguous in physical address space ( for
DMA reasons, I think ).

Thanks for any responses,
Ramakrishna Saripalli

all opinions are mine & mine only



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: IP address change..
Date: Fri, 19 Feb 1999 20:33:39 -0800
From: "Steve Doiel" <nospam_steved@pacifier.com>
Message-ID: <36ce39c4.0@news.pacifier.com>
References: <7aja2p$9i4@overload.lbl.gov>

Keith Buchanan wrote in message <7aja2p$9i4@overload.lbl.gov>...
>> I have a basic Question, Can the IP address be changed when
>> vxWorks up and running,
>> with out rebooting ? if so how ?
>
>Unlike Windoze, it sure can.  Use the ifAddrSet and ifMaskSet API
functions.
>
Gee.  I often change the IP address of my windows system without re-booting.
But I didn't manage to figure out the right combinations of ifAddrSet,
ifMaskSet functions to use after VxWorks was up and running to change the IP
address under VxWorks.

Hmmm...
SteveD



---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Target Server Error Report
Date: Tue, 16 Feb 1999 18:55:47 +0100
From: Konrad Schwarz <konrad_schwarz@europe.mcd.mot.com>
Organization: Motrola Computer Group/Technical Products Division
Message-ID: <36C9B122.2F324218@europe.mcd.mot.com>
References: <36C8B429.72EC@sandia.gov>

Gregory L. Wickstrom wrote:

> [...] Problems relating to running the target server and windSh on different
> servers.

The Tornado end-user tool passes the file to be loaded to the target server ``by
name''.The target server can't find the corresponding file on its machine.  You
may want enable
network filesystem sharing (creating network ``shares'' in Windows NT parlance)
and
name the files using universal naming conventions
(//glwicks/share/bla/bla/bla.o).  There are
other alternatives, too.

Konrad


- --
Konrad Schwarz
Motorola Computer Group
81829 M"unchen  Schatzbogen 7           Tel: +49 89 92103 828
81809 M"unchen  Postfach 820960         Fax: +49 89 92103 266




---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: VxWorks memory partitions
Date: Fri, 19 Feb 1999 22:16:07 -0800
From: Hwa Jin Bae <hjb@pso.com>
Organization: PSO Systems Inc
Message-ID: <36CE5327.9C379D8F@pso.com>
References: <7alcs5$9l11@overload.lbl.gov>

One thing that needs to be considered, but often forgotten, is the
fact that a vast majority of VxWorks applications and products
out there really are *not* hard real-time required applications
at all.   There are a lot of people that use VxWorks simply as
an embedded OS.  And many applications use ported Unix software
as basis in lots of products.  As VxWorks is used more in information
appliance type of devices, a good malloc/free is essential.
Besides, having a good malloc/free that does not fragment as badly
is not compromising realtime performance at all because if your
realtime requirement is strict, you wouldn't be malloc/free'ing
all the time anyway.  The point is that you should be able to
choose which way to go: realtime preallocated memory usage or
generic embedded systems that have legacy code that expects
decent malloc/free that won't chop up memory into useless little
bits after a day of running.

I should add that in all my years of doing VxWorks consulting
for the past 10 years, I have seen less than 10% of the projects
that really do use VxWorks as hard realtime system.  The rest
simply use VxWorks because it is a good embeddable OS, and it
runs on a wide variety of platforms.

- -- 
Hwa Jin Bae
PSO Systems Inc
hjb@pso.com   
http://www.pso.com

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Are linear addresses physically contiguous in vxworks 486 BSP?.
Date: Fri, 19 Feb 1999 22:21:23 -0800
From: Scott Johnson <sj_nospam@nospam.aracnet.com>
Organization: National Association for the Advancement of Computer Geeks
Message-ID: <36CE5463.7EB@nospam.aracnet.com>
References: <7al9qs$qa0@news.or.intel.com>
Reply-To: sj_nospam@nospam.aracnet.com

rvsaripa wrote:
> 
> I have a 486 BSP question.
> 
> I know paging is not enabled by default.
> 
> Does this mean that if I have a piece of memory that is contiguous in linear
> address space, it is also contiguous in physical address space?.

Yup.

Actually, it depends on what you put in sysPhysMemDesc in your BSP
(It's a big honking array of structs that gets defined in sysLib.c).

If you specify physical addresses to be the same as logical addresses
in all cases--that is the behavior that you will get.


> I ask because my hardware needs to be programmed with starting address > of a memory block and expects it to be contiguous in physical address > space ( for DMA reasons, I think ).

But scatter-gather DMA (how OS's with virtual memory solve the problem
of DMAing into a logical address space which is scattered to the four
winds in physical address space) is so much FUN!!!!!

Not.

Even with you DID do address translations done in sysPhysMemDesc, 
you would still get large contiguous chunks of memory.  "Virtual memory"
arrangements with arbitrary paging are not supported in vxWorks unless
you get VX_VMI.


Scott

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Problem with MPC860SAR
Date: Wed, 17 Feb 1999 15:03:11 GMT
From: aerdogan@hc.aselsan.com.tr
Organization: Deja News - The Leader in Internet Discussion
Message-ID: <7aeln9$2ls$1@nnrp1.dejanews.com>
References: <36CAC1B9.85865712@sisun9.cselt.it>

In article <36CAC1B9.85865712@sisun9.cselt.it>,
  Tricarico Giuseppe <tricaric@sisun9.cselt.it> wrote:
> Hi everyone
>
> I have problem with MPC860SAR when I try to use the SAR functions for
> UTOPIA interface.
> The following is a description of my environment and settings:
>
> -EST MDP8XX PRO Development Platform with MPC860SAR.
> -VxWorks O.S.
> -Driver for UTOPIA interface, provided by
> Motorola.(www.mot.com/SPS/RISC/netcomm/tools/)
> -I haven't still an interface that connect the phisical layer with
> UTOPIA interface, but I have     connected the TxCAV signal at VCC, and
> RxCAV signal at GND.
>
> I'm testing the driver to adapt it for my environment, so I'm using
> Tornado tools.
> When I try to set the PDPAR register with the value 0xDF7F (enabling the
> UTOPIA and SAR finction) the debugger on the remote host stops
> debugging, because it cannot communicate with the target.
> I think that there could be a conflict with the Ethernet driver, when
> the UTOPIA SAR function are Enabled.
> If  I set  this register with EST's visionICE (that use the development
> port on the platform), the Tornado tools blocks, after I clear the PDPAR
> register with visionICE and the Tornado tools return to run. What happen
> when the UTOPIA interface is enabled?
>
> Thanks for help
>
>   Tricarico Giuseppe.
>
>

The ethernet may be running on SCC3 or SCC4, creating a conflict on PortD.
While setting utopia bits, try not to disturb other bits on PDPAR by using a
|= or &= statement. If that doesn't work, it may be necessary to move
ethernet to another SCC.

Alper Erdogan

- -----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

---------------------------

Newsgroups: comp.os.vxworks
Subject: Help! Cannot see BREAKPOINTS in Tornado 1.0.1
Date: Thu, 18 Feb 1999 18:19:29 GMT
From: ren@ecf.toronto.edu (Ned Ning Ren)
Organization: University of Toronto, Engineering Computing Facility
Message-ID: <F7D48H.nn@ecf.toronto.edu>
Followup-To: ned.ren@leitch.com
Sender: news@ecf.toronto.edu (News Administrator)

Hi,
I'm running Tornado 1.0.1 with MBX860 board. Whenever I set a breakpoint,
it doesn't show the triangle mark as described in the manual, however, the
break point is actually there since excution does stop. I have already
enabled "attribute pane" in editor options. Would this be BSP related?
some extra option I have to include in my BSP?

THank you!!


---------------------------

End of New-News digest
**********************


From owner-vxwexplo-process  Sat Feb 20 04:24:03 1999
From: LJR <leer@voicenet.com>
Date: Sat Feb 20 04:24:07 PST 1999
Subject: Re: VxWorks->Linux:  Last call for votes

My vote is no for Linux support. I offer the following article excerpt in
support:

This is from Jack Ganssle's most recent "Embedded Muse" newsletter. 

---------------------------- 
The Trouble with Open Source 
---------------------------- 
One of the reputedly great things about a lot of public domain software is 
that it often comes with the source code. 
 
I don't want the source code. What am I going to do with it? Fix bugs? Most 
of us are far too busy just doing our own work to worry about maintaining 
someone else's code. I personally have written half a dozen compilers, yet 
I'd never, in a million years, want to dig into the innards of a freebie 
compiler to find some obscure optimization bug, or into an OS to figure out 
why context switches corrupt registers. No way. 
 
80% of all embedded systems are delivered late. If we figure that 
maintaining public domain software is now another required part of 
developing our own systems, well, disaster looms. 
 
Now, the argument that open source means lots of people - *other* people - 
are correcting bugs is certainly valid. Just don't let it be me, or anyone 
else concerned with getting a product to market on time. 
 
A healthy software world means we buy development products for cold hard 
cash. That money goes into other developers' pockets; part of the service 
they provide is a useful support group that efficiently deals with bugs. 
Admittedly many commercial products don't live up to this ideal. I'm 
convinced, though, that as long as we're all mucking around in each others' 
code the software crisis will never go away. 
 
This philosophy goes to modules that we include in our own products, like 
communications packages and real time OSes. Yeah, a lot of these 
off-the-shelf products are less than perfect. But as long as we demand 
access to the source we'll never be truly productive developers. 
 
Fact: Time to market pressures grow daily; it's almost impossible to get a 
product out the door on-time. 
 
Fact: Buggy tools and software packages are schedule-killers. 
 
Fact (well, at least this is my contention): If we are forced to maintain 
our tools we're doomed. 
 
Too many developers have bought into the "I can fix it" school of thought 
for dealing with the reality of poor tools. Better solutions exist. 
 
GNU tools show the yin and yang of the "open source" model. Sure, you can 
download all of the GNU source and build your own compilers. Better, send 
Cygnus some money. Let them provide you with the tools, and then you get to 
hold them responsible for maintaining the products. Cygnus has leveraged 
the best of open source while still providing support. 
 
Another option might be better communication between developers. If the 
problem is that people don't trust commercial products, perhaps we need a 
repository for success and failure stories. A registry of tools versus real 
developer experiences. One sailing magazine tracks hundreds of boats and 
their owners, so that a prospective buyer can contact a dozen or more 
owners to get the real scoop on the product. I'd like to see something 
similar for commercial products so we can get a vendor-independent, real 
"in the trenches" view of how well a particular product stands up in real 
se. 
 
But don't give me the source code. I want to go home at nights.



From owner-vxwexplo-process  Sat Feb 20 10:07:47 1999
From: "Keith Buchanan" <Keith_Buchanan@SPARTA.com>
Date: Sat Feb 20 10:07:51 PST 1999
Subject: RE: changing the IP address

> >> I have a basic Question, Can the IP address be changed when
> >> vxWorks up and running, with out rebooting ? if so how ?
> >
> >Unlike Windoze, it sure can.  Use the ifAddrSet and ifMaskSet API
> functions.
> >
> Gee.  I often change the IP address of my windows system without
> re-booting.
> But I didn't manage to figure out the right combinations of ifAddrSet,
> ifMaskSet functions to use after VxWorks was up and running to
> change the IP address under VxWorks.

I have WindowsNT 4.0 and Windows95 OSR2 running on machines in my lab.
When I change the IP address using the network applet in the control panel,
they both prompt for the CD and display a dialog which says that I should
reboot.  Is this not necessary?  Is there a less intrusive way to change
network
parameters under Windoze w/o rebooting?

I just typed 'man ifLib' & 'man ifAddrSet' and the info to change the IP
address in VxWorks is available.  The man pages even tell you to call
ifMaskSet first.  The
functions worked the first I tried them from a shell prompt.

adios

------------------------------------
      /\       Thomas Keith Buchanan
     /**\      Principal Engineer
    /****\     SPARTA, Inc.
   /****/..    205 East IH 30
  /****/....   Rockwall, TX 75087
 .\****\__...
...\**\  /.... Voice: 972 772 4487
....\**\/\.... FAX:   972 722 3201
 ....\****\..  PCS:   972 672 9657
  ....\****/   WWW:   www.sparta.com
   .../***/
    ./***/     SPARTA
     \**/      ======
      \/       Pride In Performance



From owner-vxwexplo-process  Sun Feb 21 04:00:16 1999
From: daemon@csg.lbl.gov
Date: Sun Feb 21 04:00:20 PST 1999
Subject: comp.os.vxworks newsdigest

Comp.Os.Vxworks Daily Digest    Sun Feb 21 04:00:13 PST 1999

        Subject: Re: PCI bus and device drivers....
        Subject: TORONTO - SOFTWARE PROJECT LEADER - VxWorks
        Subject: Re: VxWorks->Linux:  Last call for votes
        Subject: Re: IEEE 1394 drivers available?
        Subject: Seeking VxWorks Software Contracts
        Subject: Re: VxWorks->Linux:  Last call for votes
        Subject: Embedded Java ?
        Subject: Re: Frame's structure in SENS
        Subject: Free stuff on the net

-------------------------------------------------------

Newsgroups: comp.os.vxworks,comp.unix.programmer
Subject: Re: PCI bus and device drivers....
Date: Sat, 20 Feb 1999 12:09:10 GMT
From: George Varndell <gvarndell@hotmail.com>
Organization: Deja News - The Leader in Internet Discussion
Message-ID: <7am8l3$l24$1@nnrp1.dejanews.com>
References: <36CD65CC.77B86A3B@rnd.netas.com.tr>

In article <36CD65CC.77B86A3B@rnd.netas.com.tr>,
  Bahri Okuroglu <bahrio@rnd.netas.com.tr> wrote:
> I have devices on the PCI bus. Both the cards and the board is custom,
> and I have to develope all the required drivers.
>
> I think I need to develope drivers for the cards on the PCI bus. If so,
> can you please tell me what a PCI driver does? (There exist a driver
> (named PCI Driver : pciLocalBus.c) in the list of "WRS Device Driver
> List".)
>
> Can I control all the cards on my PCI bus only with a PCI bus driver.
> (without developing specific drivers for the cards)???

Hi Bahri,

You are about to ascend a very larger mountain, armed only with
a pair of golf shoes and a light sweater.
Get some books on PCI. Study the code you mentioned above.
You need a grasp of first principles.

gv
- ----------------------------------------------------------------
If I had a quarter for every man-hour wasted fussing with,
rebooting, reinstalling, and otherwise recovering from crashes
in MS Windows, I'd be Bill Gates.
- ----------------------------------------------------------------

- -----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

---------------------------

Newsgroups: alt.computer.consultant.ads,alt.computer.consultants,alt.computer.consultants.ads,aol.neighborhood.ont.jobs,can.jobs,ont.jobs,tor.jobs,tor.ieee,comp.os.vxworks,prg.jobs
Subject: TORONTO - SOFTWARE PROJECT LEADER - VxWorks
Date: 20 Feb 99 11:04:47 -0500
From: Karen Berends, Internet Recruiter, Alumni-Network.com <karen@alumni-network.com>
Organization: Alumni-Network.com
Message-ID: <8d73.9bcf.1b8@karen_berends>
Reply-To: karen@alumni-network.com

POSITION: SOFTWARE PROJECT LEADER - VxWorks
LOCATION: Toronto, ONTARIO CANADA

POSITION RESPONSIBILITIES:  Our client is a household word in the field
of data
communications.  Their world class hardware development group  is
currently in
need of a VxWorks guru to lead and mentor a small team of software
development
engineers.  The successful applicant will:
Develop, integrate, and maintain initialization, kernel, memory, and
debug
software for Motorola 683xx and Power PC based systems using the VxWorks
operating system.
Maintain existing low-level kernel and io_drivers services for data
communications products using Motorola processors.
Develop and maintain compilation and linking scripts that integrate
existing
software into VxWorks development/debugging environment.
The environment comprises "C", assembly language, real-time and Motorola
processors.  Protocols include X25, FRI and IP.

SALARY RANGE:

EDUCATION:

EXPERIENCE:
solutions in real-time, C, assembly language, VxWorks.
Send your resume to Karen Berends, Alumni-Network Recruiter at mailto:karen@alumni-network.com
(tel) 1/905/785-0844
(fax) 1/905/785-0877


---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: VxWorks->Linux:  Last call for votes
Date: Sat, 20 Feb 1999 13:49:33 -0700
From: jcl_home@yonext.apl.washington.edu (Jim Luby)
Organization: University of Washington
Message-ID: <jcl_home-2002991349330001@cs101-2.fsmodem.washington.edu>
References: <4.1.19990220071526.00a4be80@hardrock.isi.com>

In article <4.1.19990220071526.00a4be80@hardrock.isi.com>, LJR
<leer@voicenet.com> wrote:

> My vote is no for Linux support. I offer the following article excerpt in
> support:
> 
> This is from Jack Ganssle's most recent "Embedded Muse" newsletter. 
> 
> ---------------------------- 
> The Trouble with Open Source 
> ---------------------------- 
> One of the reputedly great things about a lot of public domain software is 
> that it often comes with the source code. 
>  
> I don't want the source code. What am I going to do with it? Fix bugs? Most 
> of us are far too busy just doing our own work to worry about maintaining 
> someone else's code. I personally have written half a dozen compilers, yet 
> I'd never, in a million years, want to dig into the innards of a freebie 
> compiler to find some obscure optimization bug, or into an OS to figure out 
> why context switches corrupt registers. No way. 
>  
> 80% of all embedded systems are delivered late. If we figure that 
> maintaining public domain software is now another required part of 
> developing our own systems, well, disaster looms. 
>  
> Now, the argument that open source means lots of people - *other* people - 
> are correcting bugs is certainly valid. Just don't let it be me, or anyone 
> else concerned with getting a product to market on time. 
>  
> A healthy software world means we buy development products for cold hard 
> cash. That money goes into other developers' pockets; part of the service 
> they provide is a useful support group that efficiently deals with bugs. 
> Admittedly many commercial products don't live up to this ideal. I'm 
> convinced, though, that as long as we're all mucking around in each others' 
> code the software crisis will never go away. 
>  
> This philosophy goes to modules that we include in our own products, like 
> communications packages and real time OSes. Yeah, a lot of these 
> off-the-shelf products are less than perfect. But as long as we demand 
> access to the source we'll never be truly productive developers. 
>  
> Fact: Time to market pressures grow daily; it's almost impossible to get a 
> product out the door on-time. 
>  
> Fact: Buggy tools and software packages are schedule-killers. 
>  
> Fact (well, at least this is my contention): If we are forced to maintain 
> our tools we're doomed. 
>  
> Too many developers have bought into the "I can fix it" school of thought 
> for dealing with the reality of poor tools. Better solutions exist. 
>  
> GNU tools show the yin and yang of the "open source" model. Sure, you can 
> download all of the GNU source and build your own compilers. Better, send 
> Cygnus some money. Let them provide you with the tools, and then you get to 
> hold them responsible for maintaining the products. Cygnus has leveraged 
> the best of open source while still providing support. 
>  
> Another option might be better communication between developers. If the 
> problem is that people don't trust commercial products, perhaps we need a 
> repository for success and failure stories. A registry of tools versus real 
> developer experiences. One sailing magazine tracks hundreds of boats and 
> their owners, so that a prospective buyer can contact a dozen or more 
> owners to get the real scoop on the product. I'd like to see something 
> similar for commercial products so we can get a vendor-independent, real 
> "in the trenches" view of how well a particular product stands up in real 
> se. 
>  
> But don't give me the source code. I want to go home at nights.

I think you may have missed the mark as regards advocacy for a Linux
- -based version of Tornado/VxWorks.  I don't recall seeing much if any
discussion about access to operating system source code and so forth. 
Instead, the discussion has centered on the desire to use Tornado/VxWorks
on inexpensive PC hardware running the Linux OS and thus not having to put
up with various deficiencies of the Windows NT environment or the high
cost of a Sun or HP workstation.  Also I'd be amazed if the majority of
those advocating a Linux-based release of Tornado didn't fully expect that
such a release would be fully supported by Wind River.  

Cheers,

Jim

PS: I suspect that the vast majority of programmers would agree with the
theme of Jack Ganssle's article.

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: IEEE 1394 drivers available?
Date: Tue, 16 Feb 1999 20:40:09 GMT
From: Allen Walker <awalker@signtech.com>
Organization: Signtech USA Ltd. (http://www.signtech.com)
Message-ID: <36C9D7B4.9574DB8E@signtech.com>
References: <36C3DC27.4921D633@ekf.de>

Intoto (http://www.intotoinc.com) provides VxWorks drivers for the
PCILynx.

J. Jansen wrote:
> 
> Does anybody know if there are VxWorks drivers and other software
> available for the TI PCILynx TSB12LV21A IEEE 1394 link layer
> controller? We see the need to integrate our CompactPCI
...

- -- 
Allen Walker                Signtech USA Ltd.
http://www.signtech.com     4669 Highway 90 West
+1 210.436.4777 x341        San Antonio, TX 78237

---------------------------

Newsgroups: comp.os.vxworks
Subject: Seeking VxWorks Software Contracts
Date: Wed, 17 Feb 1999 03:55:35 GMT
From: realtime4u@yahoo.com
Organization: Deja News - The Leader in Internet Discussion
Message-ID: <7adejl$2k3$1@nnrp1.dejanews.com>

A small team of very talented VxWorks embedded software designers is seeking
software development contracts. Each of us has 10+ years of experience
developing  real-time embedded software. Our expertise is in developing
software using WindRiver's VxWorks suite of development tools and real-time
operating system. Some of our areas of exepertise are as follows:

* Custom Board Support Packages (BSPs).
* Communication drivers (Ethernet, serial and IEEE 1394 (FireWire))
* FLASH device drivers and FLASH file systems.
* SNMP agents.
* HTTP servers.
* Internet protocols (UDP, TCP/IP, PPP, BOOTP, DHCP, OSPF, etc.)

We are currently accepting contract requests. We will provide a complete
software solution or specific software components, such as BSPs or device
drivers for custom hardware. We are located in Ottawa, Ontario Canada and
currently we are only seeking contracts for which the work can be completed
from our location in Ottawa.

For more information please send email to realtime4u@yahoo.com

- -----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own    

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: VxWorks->Linux:  Last call for votes
Date: Sat, 20 Feb 1999 22:47:41 -0700
From: Douglas Jerome <jerome@primenet.com>
Organization: www.primenet.com/~jerome
Message-ID: <36CF9DFD.7B7DA9C7@primenet.com>
References: <4.1.19990220071526.00a4be80@hardrock.isi.com>

Well, here goes the begining of what might be a long thread...

LJR wrote:
> 
> My vote is no for Linux support. I offer the following article excerpt in
> support:
> 
> This is from Jack Ganssle's most recent "Embedded Muse" newsletter.
> 
> ----------------------------
> The Trouble with Open Source
> ----------------------------
> One of the reputedly great things about a lot of public domain software is
> that it often comes with the source code.
> 
> I don't want the source code. What am I going to do with it? Fix bugs? Most
> of us are far too busy just doing our own work to worry about maintaining
> someone else's code. I personally have written half a dozen compilers, yet
> I'd never, in a million years, want to dig into the innards of a freebie
> compiler to find some obscure optimization bug, or into an OS to figure out
> why context switches corrupt registers. No way.
> 
> 80% of all embedded systems are delivered late. If we figure that
> maintaining public domain software is now another required part of
> developing our own systems, well, disaster looms.
> 
> Now, the argument that open source means lots of people - *other* people -
> are correcting bugs is certainly valid. Just don't let it be me, or anyone
> else concerned with getting a product to market on time.
> 
> A healthy software world means we buy development products for cold hard
> cash. That money goes into other developers' pockets; part of the service
> they provide is a useful support group that efficiently deals with bugs.
> Admittedly many commercial products don't live up to this ideal. I'm
> convinced, though, that as long as we're all mucking around in each others'
> code the software crisis will never go away.
> 
> This philosophy goes to modules that we include in our own products, like
> communications packages and real time OSes. Yeah, a lot of these
> off-the-shelf products are less than perfect. But as long as we demand
> access to the source we'll never be truly productive developers.
> 
> Fact: Time to market pressures grow daily; it's almost impossible to get a
> product out the door on-time.
> 
> Fact: Buggy tools and software packages are schedule-killers.
> 
> Fact (well, at least this is my contention): If we are forced to maintain
> our tools we're doomed.

Forced???  The use of the word *forced* here is an attempt to obliterate
rational, logical thought.  You are changing the subject to something no
one is going to argue against, and subversively use that to draw some
conclusion about the original subject.  This is classic subterfuge.  All
laws should be good does not mean all good should be laws.  You know
that
no one wants to force you (or force anyone else) to maintain your own
tools.
And that has nothing to do with the quality of open source tools.

> Too many developers have bought into the "I can fix it" school of thought
> for dealing with the reality of poor tools. Better solutions exist.

Prove that better solutions exist.  It *IS* the contention of open
source
advocates that the open source model generates the best tools.  Who but
shoe makers better know what tools are best for shoe making?  Who would
make
better shoe making tools than the shoe makers that use them?  There is a
great concept that the producers of a product should use their product,
and
that is the heart of the open source model.  You don't have to be a part
of
that, but why limit others?

> GNU tools show the yin and yang of the "open source" model. Sure, you can
> download all of the GNU source and build your own compilers. Better, send
> Cygnus some money. Let them provide you with the tools, and then you get to
> hold them responsible for maintaining the products. Cygnus has leveraged
> the best of open source while still providing support.
> 
> Another option might be better communication between developers. If the
> problem is that people don't trust commercial products, perhaps we need a
> repository for success and failure stories. A registry of tools versus real
> developer experiences. One sailing magazine tracks hundreds of boats and
> their owners, so that a prospective buyer can contact a dozen or more
> owners to get the real scoop on the product. I'd like to see something
> similar for commercial products so we can get a vendor-independent, real
> "in the trenches" view of how well a particular product stands up in real
> se.
> 
> But don't give me the source code. I want to go home at nights.

- -- 
Douglas Jerome

---------------------------

Newsgroups: comp.os.vxworks
Subject: Embedded Java ?
Date: 20 Feb 1999 11:30:28 GMT
From: "Eric S. Riseman" <eriseman@tiac.net>
Organization: The Internet Access Company, Inc.
Message-ID: <01be5cc3$df4b8880$941f77ce@LittleFoot>

Hi all,
While I understand that Java for embedded is still in it's infancy, I also
understand that Sun (pJava) and HP (whatever their VM is called) are poised
for
battle.

Apparently HP is already shipping and Sun pJava for embedded is and has
been in Beta forever.
==Anyone know when pJava will be out of Beta ?
==When will a Sun pJava port to VxWorks be available as "FCS" (not the beta
marketing hype)  ?
- ----Again this assumes that pJava ever comes out of Beta.......

I am assuming that since HP knows the embedded marketplace much better
than sun does (or ever will), that the HP solution is and will be a far
better embedded solution.

Any comments ??

ESR

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Frame's structure in SENS
Date: Sun, 21 Feb 1999 10:28:42 GMT
From: johnchann@home.net (johnchann)
Organization: @Home Network
Message-ID: <udRz2.8694$yv3.3231@news.rdc1.sfba.home.com>
References: <7aj5jq$9ha$1@news.flashnet.it> <36cf1219.202252884@news.cisco.com>

In article <36cf1219.202252884@news.cisco.com>, SomeGuy@SomethingOrOther.com (Some Guy) wrote:
>On Fri, 19 Feb 1999 08:59:34 +0100, "Fabio Bertoldi"
><fbertoldi@mael.it> wrote:
>
>>? Does SENS stack support ONLY Ethernet and 802.3 packets ?
>
>WindRiver admitted to us that this is so.
>
>So what we do now is we emulate ethernet over our hardware (one shared
>memory interface and one proprietary bus). When SENS gives us an
>ethernet packet we put a wrapper around it and push that down. When a
>wrapped packet comes in we unwrap it and give it to SENS.

Since WindRiver also delivers PPP and PPP is neither 802.3 not Ethernet and 
since there are third party drivers for ATM which also isn't either Ethernet 
or 802.3 then I'd say that it is neither true that the SENS stack is so 
limited not that Wind Rivers delivered product is so limited.

JRC

---------------------------

Newsgroups: comp.os.vxworks
Subject: Free stuff on the net
Date: 21 Feb 1999 10:39:25 GMT
From: Grant<grant@globetrotter.qc.ca>
Organization: GlobeTrotter
Message-ID: <7aonot$58k$10147@news.quebectel.com>

Please check out my free stuff page, magazines,coupons, CDs etc.
http://www.corosswinds.net/montreal/~electgra/freestuf.html  While you are there check out my Electronic and computer web page.

---------------------------

End of New-News digest
**********************


From owner-vxwexplo-process  Sun Feb 21 15:08:23 1999
From: "Mike McQuade" <mmcquade@aa.net>
Date: Sun Feb 21 15:08:26 PST 1999
Subject: Re: VxWorks->Linux:  Last call for votes


VxWorks.

I believe that Jack has a problem with
Linux in general, as so does Plauger
if you read Embedded Systems Magazine.

I don't doubt Ganssle's vast knowledge of
things embedded, but I question weather
he or Plauger have ever used Linux, or 
if they have other motives for their comments.

"I often wonder if FREE software is
not some of the most expensive code
around" -- Ganssle ESP Jan 1999.

"Before you write, please note that
I cannot afford FREE software" 
-- PJ Plauger ESP Jan 1999.


What makes the O/S so powerful is the intense
peer review that goes into every line of
OPEN SOURCE code.


I sure am glad that the decision makers at
Wind River we smart enough to use some 
FREE software in their product. I am equally
sure that they (like SGI, IBM, DELL, HP, Compaq,
Oracle, Sybase, etc...) will embrace the
FREE and superior operating system.


Maybe in the future we will have an operating
system that is FREE, robust, and runs
on multiple architectures. A true level playing
field for people to compete upon.


And maybe the local bookstore will have some
excellent books on gardening by Ganssle.


I run both Linux and NT, my Linux boxes (Pentium PRO
SMP, and two PowerPC's (MVME-2603, and PowerStack)
run for months on end. Reboots of NT are a required
fact of life.


Start to write some network or serial code on NT,
and you will put some cycles on that power switch.


Then you will see why the community is running nearly
100:1 in favor of a VxWorks Tornado port.


Mike




-----Original Message-----
From: the vxWorks Users Group Exploder <vxwexplo@lbl.gov>
To: vxworks_users@csg.lbl.gov <vxworks_users@csg.lbl.gov>
Date: Saturday, February 20, 1999 5:41 AM
Subject: Re: VxWorks->Linux: Last call for votes


>Submitted-by owner-vxwexplo-process  Sat Feb 20 04:24:03 1999
>Submitted-by: LJR <leer@voicenet.com>
>
>My vote is no for Linux support. I offer the following article excerpt in
>support:
>
>This is from Jack Ganssle's most recent "Embedded Muse" newsletter. 
>
snip....




From owner-vxwexplo-process  Sun Feb 21 17:04:41 1999
From: John Ford <jford@bara.gb.nrao.edu>
Date: Sun Feb 21 17:04:44 PST 1999
Subject: Re: VxWorks->Linux:  Last call for votes


Well, to further flog this dead horse:

> Submitted-by owner-vxwexplo-process  Sun Feb 21 15:08:23 1999
> Submitted-by: "Mike McQuade" <mmcquade@aa.net>
> 
> 
> VxWorks.
> 
> I believe that Jack has a problem with
> Linux in general, as so does Plauger
> if you read Embedded Systems Magazine.
> 
> I don't doubt Ganssle's vast knowledge of
> things embedded, but I question weather
> he or Plauger have ever used Linux, or 
> if they have other motives for their comments.
> 
> "I often wonder if FREE software is
> not some of the most expensive code
> around" -- Ganssle ESP Jan 1999.
> 
> "Before you write, please note that
> I cannot afford FREE software" 
> -- PJ Plauger ESP Jan 1999.

<snip>

I was rather taken aback by these comments in ESP, and then when this particular
"embedded muse" came out, I was forced to reply.  All that I said then
is still true, with the exception that I have actually gotten good
results from WRS tech support lately!  So, mentally delete them from
the following:

----->8-  Cut here ---->8---------

Date: Thu, 7 Jan 1999 16:01:56 -0500
From: John Ford <jford@bara.gb.nrao.edu>
To: jack@ganssle.com
Subject: latest embedded muse.
Reply-to: jford@sadira.gb.NRAO.EDU


After reading your article (and one by P.J. Plauger) in ESP, and
noting your woes with Microsoft's products, I was intrigued by your
essay on open source products.  I disagree with your conclusions, but
only slightly.  I don't know if you have experience with any of these
products, but I'll give you my views.

I think your premise is wrong about spending lots of time debuggng
your tools.  I have been using the GNU compilers for Sun Solaris,
Intel Linux, and VxWorks 68K systems for many years.  These compilers
are bulletproof.  Compiling one is a one-liner.  It takes me 5 minutes
of my time to install GCC on a Sun or Linux system.  The reason that
people use them is that they work!  No muss, no fuss.  As you point
out below, you can find any number of people to maintain the code for
you, just as you can buy a maintenance agreement for other software.

Interestingly. the tools that come with VxWorks (Itself non-free, to
be sure!)  are the GNU toolset.  I have never spent time debugging the
tools.  I built a companion compiler/debugger suite so that I can
compile my source for different platforms using ONE C++ compiler.
This is an unbelievable win.

As far as your essay goes, it's correct.  You ought to spend your time
working on your own problems.  But I can honestly tell you that I, and
my co-workers, spend less time on our computers, tools, and operating
systems than do the engineers that I supervise who use Windows95/NT
and shrink-wrapped tools. 

> Fact (well, at least this is my contention): If we are forced to maintain
> our tools we're doomed.

I contend that if you rely on a vendor to fix bugs, you're doomed.  I
have NEVER had a vendor fix anything for me.  I have found bugs in
VxWorks, Solaris, and Microsoft products.  I was never given a fix.
I'm waiting right now (2 weeks!) for help on a bug in one of our
purchased non-source products.  If I had the source, I could fix it
myself in a few hours.

I think the feeling of safety you get from buying something already
compiled and with a "warranty" (HA! have you read your shrink-wrap
license lately!) is way overblown.  Usually, if you call tech support
with a problem, the best you can hope for is that they'll cut you off
before you waste much time hanging on the line too long.    

>Too many developers have bought into the "I can fix it" school of thought
>for dealing with the reality of poor tools. Better solutions exist.

Typical programmer hubris!
 
>GNU tools show the yin and yang of the "open source" model. Sure, you can
>download all of the GNU source and build your own compilers. Better, send
>Cygnus some money. Let them provide you with the tools, and then you get to
>hold them responsible for maintaining the products. Cygnus has leveraged
>the best of open source while still providing support.

This is IMHO the best of both worlds.  Wind River actually uses the
Cygnus toolset.

>Another option might be better communication between developers. If the
>problem is that people don't trust commercial products, perhaps we need a
>repository for success and failure stories.

Your article and the one by P.J. Plaugher show that this does not
work.  It has become sort of a badge of honor to be the loser in a
wrestling match with WindowsXXXX.  It does not seem to be a
deterrent.  Maybe it has to do with the typical programmer hubris
mentioned above!

> But don't give me the source code. I want to go home at nights.
 
Me too.  I sleep well knowing that if there is a problem, I can fix
it.  I also am confident that my system will be usable in the morning,
and will not crash, taking my hard work with it.  The down side is I
don't get to tell any horror stories about how my computer ate my
homework :). 

Thanks for provoking me to write this!

-- 
John Ford
National Radio Astronomy Observatory
Green Bank, WV 24944-0002
jford@nrao.edu

------------ Cut Here -----------


John


From owner-vxwexplo-process  Sun Feb 21 19:45:48 1999
From: "Bill W.G. Shi" <wgshi@bbti.com.cn>
Date: Sun Feb 21 19:45:51 PST 1999
Subject: RE:How to get "arpAdd", "arpDelete" work

Hi, Wang:
        I have been through this problem before.
        What I did was:
        1.) using "ar" to extract the arpLib.o from the target library.
        2.) manually add it to the VxWorks image to be built.
        or
            put it in the vxworks image ftp directory on the host, which is
the "/" in the shell, then "ld" it into memory.

Hope this helps.

Bill W.G. Shi at BBT Inc.
E-Mail Address: wgshi@bbti.com.cn





From owner-vxwexplo-process  Mon Feb 22 00:41:27 1999
From: "Michael Lawnick" <lawnick@softec.de>
Date: Mon Feb 22 00:41:30 PST 1999
Subject: PPP configuration

Hi all,

It seems I have got a NT problem, but perhaps somebody can help me 
anyway :

I am trying to configure my vxWorks target to support PPP 
over a modem line.
On my first attempt, I tried to make the target the passive server, 
called by a NT Web Browser. After the connect the target sends 
configuration requests, but the NT isn't answering.
After a call to WRS (they said that the NT-machine should be called 
and function as server) I tried it the other way, but I could not get 
the RAS-server to accept a call. On starting the machine, my modem 
sets the AA light for automatic answer, but on logging in the light 
turns off and all calls on the telephone line are leading to a simple 
light show on the modem, but not an accept of the call by NT.
I have enabled both incoming and out going connects in RAS setup.

Any hints would be welcome.


MfG
Lawnick, SOFTEC GmbH

==============================================
SOFTEC GmbH               Tel +49-731-96600-0
Promenade 17              Fax +49-731-96600-23
D-89073 Ulm               Michael Lawnick
Germany                   lawnick@softec.de
==============================================


From owner-vxwexplo-process  Mon Feb 22 02:33:35 1999
From: David Laight <dsl@tadpole.co.uk>
Date: Mon Feb 22 02:33:39 PST 1999
Subject: Re: VxWorks memory partitions

Fred Roeber and Keith Buchanan are not incorrect in saying that a true 'real 
time' application will not use malloc/free during its normal operation.  However 
it is also true that many of use are using VxWorks as a (small) O/S for a single 
application system.  We don't necessarily care exactly how long malloc/free take 
(within reason) but do wish that the system will not fail to allocate memory 
just because it has fragmented the free list.


Fred said:
> 
> There are numerous memory allocation algorithms floating around with
> significantly different execution and memory efficiencies.

True...

> The Wind River one is not elegant but it is well implemented and fairly widely 
> used.
My observation is that the code paths through malloc/free could be halved in 
size without much effort.

> It isn't optimized for use in a time critical system, however. The normal
> use of memory allocation in real-time systems is to allocate all dynamic
> memory areas during initialization or as part of a background task. If
> you have an application that allocates and frees a lot of memory during
> operation then that isn't the type of application that the VxWorks
> allocator is designed for. There are some alternatives you might consider.

The time critical parts of the system will not (normallY) allocate memory on the 
fly - if only because you don't want to code the error paths.

> If you don't care about efficient memory use then one of the most time
> efficient allocators is a fixed block size allocator. There is code for
> such an algorithm in the VxWorks archive (called poolLib if I remember).

Allocators that issue fixed size blocks can be quite efficient in cpu time and 
even memory use.  A quite good example of this is the allocater in UnixWare 7.

> If you need to handle different size memory blocks then an algorithm
> we have had good success with is the "binary buddy algorithm". A variant
> of it is used in various GNU programs so is easily available. We found
> that it could be written in such a fashion that it was usable by multiple
> tasks concurrently with a deterministic maximum run time (the VxWorks
> allocator is single threaded and non deterministic).

The 'binary buddy' algorithm is horrid.  While fine in principle, the difficulty 
in determining whether the 'buddy' is free requires additional data structures 
which can easily get out of step, attempts to speed up the code just make it 
more complicated (and take longer and be buggy).  Getting it efficient for 
multi-cpu systems is almost impossible (all of the control structures must be 
locked for every access).
 
> Getting an allocation algorithm to work correctly and robustly is not
> trivial, however.

There are simple algorithms that work fine and are easy the check.  Marginal 
inefficencies in the algorithm can be made up by simplicity of design and 
reduced execution time.

> You may be better off buying one unless you have time
> to check out algorithms in the literature (some good sources are "Data
> Structures & Their Algorithms" by Lewis and Denenberg and one of the Knuth
> algorithm books; can't remember which volume as someone has borrowed it).
> I have heard good things about allocation code for VxWorks from Seaweed
> Systems but have never used it (query info@seaweed.com).

It took me less than a day to write a replacement memPartAlignedAlloc...
Its significantly better (in terms of fragmentation) the that supplied by 
WindRiver - even though it has to use their free-list structure.
> 
> I guess my main point is that the VxWorks algorithm may not be optimal but,
> then, I don't think there is any one algorithm that is optimal for all
> applications. Real time systems that do significant dynamic memory allocation
> during operation need careful design to avoid fragmentation problems. Most
> applications are well advised to avoid the problem by preallocating memory.

Ok, try running the java startup code...

My basic point is that you can do an awful lot better without trying very hard.
Allocating 'best fit' (instead of 'first fit') will solve most fragmentation 
problems.  Putting the 'free list' into a binary tree will reduce search time. 
Balancing the binary tree would be useful and is the only tricky bit.
The memory overhead per allocated area can be reduced to 1 word (from 2 words) 
without changing the algorithm.

	David

----------------------------------------------------------------
David Laight                          email: dsl@tadpole.co.uk
Tadpole Technology plc                phone: +44 1223 278 256
Cambridge, UK                         fax:   +44 1223 278 201



From owner-vxwexplo-process  Mon Feb 22 04:00:15 1999
From: daemon@csg.lbl.gov
Date: Mon Feb 22 04:00:19 PST 1999
Subject: comp.os.vxworks newsdigest

Comp.Os.Vxworks Daily Digest    Mon Feb 22 04:00:12 PST 1999

        Subject: Re: two ethernets and vxworks
        Subject: Re: Questions on features and performance
        Subject: Re: VxWorks->Linux:  Last call for votes
        Subject: semTake problem using semBSmCreate
        Subject: MPC860SAR & VxWorks
        Subject: MPC860SAR & VxWorks
        Subject: can i get some documents?
        Subject: VxWorks 5.3 Interrupt Service Routine
        Subject: Last Day's Prophecy Website
        Subject: WindFoundationClass (WFC)

-------------------------------------------------------

Newsgroups: comp.os.vxworks
Subject: Re: two ethernets and vxworks
Date: Fri, 19 Feb 1999 08:11:36 -0600
From: Frank Minich <fminich@ti.com>
Organization: NCP
Message-ID: <36CD7118.362B@ti.com>
References: <7ajcjo$m3i$1@claire.desy.de>
Reply-To: fminich@ti.com

Joerg Rieling wrote:
> 
> hi all,
> 
>   in vxworks, only one ethernet connection is foreseen. how do i
>   proceed if i want to use two or more ethernet cards simultaneoulsy?
> 

Caveat:  I don't know if SENS adds any complications;  following is
for the older architecture.

No problem if the cards are of the same type (e.g., both are
3Com 3C509 cards).  I've never tried with different types of card.

It's true that the default boot sequence will only start one,
but your application can call the driver's "attach" function for the
second card (with the second card's base address and IRQ), and then
ifMaskSet() for the second card, then ifAddrSet() for the second
card.  Then you are up and going.

If you want to assign both cards IP addresses which appear to be
on the same network, you will have problems, though.  You will have
to use the etherOutputHook to make sure that outbound packets are
actually sent on the card you want them sent from, since the 
default VxWorks blob of code at the bottom of the TCP/IP stack uses
the destination IP address to determine which card to send the
packet on.

Regards,
Frank Minich
fminich@ti.com

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: Questions on features and performance
Date: Thu, 18 Feb 1999 12:50:34 +0000
From: Phil Ashby <phil.ashby@wireplay.bt.co.uk>
Organization: BT Wireplay
Message-ID: <36CC0C9A.5066@wireplay.bt.co.uk>
References: <7agvkc$3pv$1@nnrp1.dejanews.com>

There will be other (probably better informed :) answers than mine,
however I'll get the ball rolling..

Geurt wrote:
> 1. Does VxWorks support:
>         - TCP/IP routing

Yes. It uses a version of the standard BSD4.3/4.4 network stack.

>         - PCI Plug & Play (determine memory/port ranges, etc).

It depends. VxWorks support PCI devices and has the ability to
read/write PCI config space. The rest depends on the device
driver(s) themselves. Many device drivers re-map PCI devices to
suit a particular memory model.

>         - Hard drive mirroring

I have not come across this myself, anyone else know?

> 3. What are the typical ROM and RAM sizes for VxWorks.

A fully-operational standalone kernel with network drivers for
a couple of chipsets, VME support & a small application is 700k
uncompressed (RAM size), 380k compressed (ROM size) including
the decompressor. This is for a typical RISC CPU (PPC 604).
For an Intel CPU (386 codebase) the corresponsing sizes are
600k uncompressed, 280k compressed.

HTH,
- --
!--------------------------- Phil "Phlash" Ashby ---------------------!
! BT Wireplay - Lead Programmer and general technical trivia supplier !
! Phone: 01473-644348                   WWW: http://www.wireplay.com/ !
! Snail: B81/G41, BT Labs, Martlesham Heath, Ipswich IP5 3RE, England.!

---------------------------

Newsgroups: comp.os.vxworks
Subject: Re: VxWorks->Linux:  Last call for votes
Date: Sun, 21 Feb 1999 10:21:30 -0500
From: Tim Sohacki <sohacki@nortelnetworks.com>
Organization: Nortel (Northern Telecom)
Message-ID: <36D02479.7C18929D@nortelnetworks.com>
References: <4.1.19990220071526.00a4be80@hardrock.isi.com> <36CF9DFD.7B7DA9C7@primenet.com>

A bit far off the original topic, but Jack's follow-up Muse 
regarding Open Source follows.

A very good point was made on this thread.  The Open Source 
movement has little to do with whether WRS supports Linux 
or not.  The request isn't for WRS to provide source, but to 
support another host OS.

- ------------------------------------------------------------
- ------------------------------------------------------------

The Embedded Muse

- ------------------------------------------------------------
Embedded Muse 32     Copyright 1999 TGG     February 1, 1999
- ------------------------------------------------------------
You may redistribute this newsletter for noncommercial purposes. For
commercial use contact info@ganssle.com. 

EDITOR: Jack Ganssle, jack@ganssle.com

CONTENTS:
- - Embedded Seminar in San Jose
- - The Trouble with Open Source (again!)
- - Thought for the Week
- - About The Embedded Muse


Embedded Seminar in San Jose
- ----------------------------
I'll present the seminar "The Best Ideas for Developing Better Firmware
Faster" in San Jose on February 18, 1999. 

The focus is uniquely on embedded systems. I'll talk about ways to link
the
hardware and software, to identify and stamp out bugs, to manage risk,
and
to meet impossible deadlines. If you're interested reserve early as
these
seminars fill completely.

For more information check out http://www.ganssle.com or email
info@ganssle.com. 


The Trouble with Open Source
- ----------------------------
Last issue I went off about issues I have with the "open source"
movement,
generating a staggering volume of replies.

Several folks asked if the article was a troll, an attempt to get folks
riled up. Well, *of course* it was a troll - there's nothing more fun
than
hearing strong opinions from smart people... though that doesn't mean I
don't think that some of the reasoning used for distributing source code
is
flawed. 

Others accused me of being a Microsoft stooge, which is funny since my
Windows 95 machine crashed three times in reading these emails! 

Of the respondents, about a third agreed with me; two thirds were firmly
at
odds. There was no middle ground.

Many people made a number of very good points. To summarize the most
common
threads:
- - Few developers trust the vendors. As one reader put it 
  "I feel better trusting part-time hobbyists 
  than the vendor to get it right."
- - No one forces you to modify the source, so there's no
  downside to having it.
- - Open source means you can port the code to another 
  CPU or platform. 
- - The code from vendors is buggy. Developers need the
  code to fix it. One reader mentioned that 90% of 
  coding time is debugging, especially with purchased
  code, so there's a quality Vs source issue. Also, 
  documentation errors often can't be resolved without
  the source.

And so, since it's only fair to give equal time to others, here are some
excerpts from readers' comments. All are reprinted with permission of
the
authors.

And thanks to everyone for writing. I wish there was room to share more
of
people's thoughts.


Randy Brei writes:
The other day, one of my co-workers was singing the praises of Linux ...
and one of (supposed) virtues was that you get all the source code. I'm
thinking to myself, "that's the last thing I want to have: zillions of
lines of source code". I've written an OS and I've studied a handful of
other OSes. I don't want to work on another OS. I just want to be a
user.
Also, it is not a comforting feeling to me ... knowing that a thousands
...
or millions ... of people have been messing about with the source code.


I particularly liked Bob Kodner's rephrasing of a Freak Brothers saying:
Source and No Money will get you through the hard times better than
Money
and No Source.


Gary Bergstrom said:
I'm aware of the trap that many engineers fall into: NIH.  and two
corollaries:  I can do it better  &  everyone else does it wrong. I try
hard to not fall into these.  What is important is getting the job done.
If it's faster to work on my tools, so be it.  If it's not, then I hope
that my experience will let me know that before I waste time.


Donald Kerns and numerous others said:
The engineers get much better "kicks" releasing the "free" LINUX version
(just because it's COOL) than the death marched commercial release.


John Ford spoke for a lot of readers:
I contend that if you rely on a vendor to fix bugs, you're doomed.  I
have
NEVER had a vendor fix anything for me.  I have found bugs in VxWorks,
Solaris, and Microsoft products.  I was never given a fix. I'm waiting
right now (2 weeks!) for help on a bug in one of our purchased
non-source
products.  If I had the source, I could fix it myself in a few hours.


>From Scott Finneran:
I do however have a reason for wishing that I had the source to the
cross-compiler that we use here. I am working on a project which has
been
going for about 10 years now (customers just keep asking for more
functions... you know the story). Our cross-compiler (status: NO LONGER
SUPPORTED) runs under an ancient version of an OS (status: NO LONGER
SUPPORTED) on an old underpowered machine (status: NO LONGER SUPPORTED).

The answer is obvious... go back to the compiler vendor and get them to
port it to the latest version of their OS. Their response to our
requests
is "that product is no longer supported". As they no longer sell the
compiler, we requested the source so that we could port it ourselves. We
were willing to sign NDAs and agree to only use it in house. The
response:
"that product is no longer supported". As we don't have the source to
the
compiler nor the budget to port our code to a new compiler (a massive
effort due to compiler "quirks"), we are stuck. If however we had chosen
for example EGCS - the Cygnus supported variant of GCC, we would be in
far
better shape. We could simply get any qualified contractor (for example
Cygnus) to port the compiler to the new platform. Not only that, but we
would be getting EXACTLY THE SAME COMPILER on our new blindingly fast
platform. This also avoids the other problem of vendors coming back with
"we have a new version which runs on your new platform..... their are
just
a couple of MINOR differences".

I guess what I am getting at is, no I don't want to tinker with the guts
of
my compiler but having the source is an insurance policy against my
product
outliving the tools that I use to build it.


Andrew Mayo wrote:
As someone who has been watching the recent emergence of the Open Source
movement, I think you have been a little harsh on the idea of access to
the
source code.

While some misguided souls might indeed consider that they can hack into
their compiler etc. and improve it, most of us don't want the source
code
for that reason.

Instead (and if you've ever tried to get support through Microsoft, for
example, you'll see the point), the availability of Open Source means
that
support organisations will be able to properly resolve issues in a much
more effective way. Suppose you have a bug in your compiler. You ring
the
support organisation (you are paying either per incident or via a
contract,
so they get to make some money, of course). The experienced techie at
the
other end can examine the source code to determine the problem and then
either patch it, have you patch it, and/or notify the author(s) so that
the
next release will have the fix.

Whereas with Microsoft, the source code is locked up in Redmond. The
support people in Singapore don't have access to it. So most problems
take
a lot longer to resolve. Would you rather have your bug fixed today or
next
year?

Open Source will clearly result in higher-quality software and quicker
release cycles. In addition, access to the source will allow budding
programmers to examine work done by professionals (hopefully). I submit
that if you look, say, at the source for PGP or Perl, you are looking at
quality software developed by experts. You will do well to copy their
styles and techniques. Also, here are a million wheels already invented
for
you so you can just take the routines and re-use them. Want a really
good
random number generator? A hashing algorithm? A set of B-Tree routines?.
Here they are.

Oh, and perhaps you'd like something running on another platform.
Microsoft
might not choose to do that for marketing reasons. For example, Internet
Explorer 4 has been ported to Solaris but will we ever see it on Linux?.
Not likely, I'd say. But Netscape's browser is publicly available in
source
form so of course it's already been ported.

Open Source is already a concept most embedded systems developers are
familiar with, anyway. For example, take the PIC chip. Look at all the
stuff Microchip provide on their web site in source form - clearly, this
encourages developers to get started quickly. The Open Source movement
simply takes this concept to an extreme.

Finally you have to look at the cost. With Linux, you could set up an
engineering workstation a lot cheaper, spending the money on better
design
tools etc. Isn't this a good thing, or do you want Mr. Gates to earn
even
more than he does?


Thought for the Week
- --------------------
See http://www.duh-2000.com for the latest on Y2K.


About The Embedded Muse
- -----------------------
The Embedded Muse is an occasional newsletter sent via email by Jack
Ganssle. Send complaints, comments, and contributions to him at
jack@ganssle.com. 

To subscribe, send a message to majordomo@ganssle.com, with the words
"subscribe embedded email-address" in the body. To unsubscribe, change
the
message to "unsubscribe embedded email-address". BUT - please use YOUR
email address in place of "email-address".

The Embedded Muse is supported by The Ganssle Group, whose mission is to
help embedded folks get better products to market faster.

---------------------------

Newsgroups: comp.os.vxworks
Subject: semTake problem using semBSmCreate
Date: Sun, 21 Feb 1999 12:37:51 -0500
From: "Dave Rhodes" <dlrhodes@injersey_no_spam.com>
Organization: InfiNet
Message-ID: <7apg6v$tkn$1@nw001t.infi.net>

Hi, I am new to VxWorks and have a PC target and PC host.
I am using the newest versions vxworks 5.3.1. and tornado 1.0.1
I have things working, got signals working, tasking, etc. I have
set shared mem base address/size etc. I am experimenting
using a single target, but want to use the shared semaphores
and memory.

But I have an error I can't figure out. The error is from semTake,
it gives a bad SEMID but I have printed the id # out when created
(in another task) and when I give and when I take and they all agree,
I am using the local <-> global translations correctly too. the take
is with WAIT_FOREVER, but even with no waiting it doesn't work
either. Any ideas or suggestions welcome!

Dave




---------------------------

Newsgroups: comp.os.vxworks
Subject: MPC860SAR & VxWorks
Date: Thu, 18 Feb 1999 18:52:34 +0100
From: Tricarico Giuseppe <tricaric@sisun9.cselt.it>
Organization: Centro Servizi Interbusiness
Message-ID: <36CC5362.6E13E175@sisun9.cselt.it>

Hi everyone

I wrote a message because I had conflict problem with MPC860SAR.
Now I think that the proble arise because the UTOPIA use the DREQ0
signal as Rxcav.
If the Ethernet controller use IDMA1 channel the conflict is inevitable!

Do someone know if the ethernet controller of VxWorks use IDMA1?

Thanks for help
                Tricarico Giuseppe


---------------------------

Newsgroups: comp.os.vxworks
Subject: MPC860SAR & VxWorks
Date: Thu, 18 Feb 1999 18:52:19 +0100
From: Tricarico Giuseppe <tricaric@sisun9.cselt.it>
Organization: Centro Servizi Interbusiness
Message-ID: <36CC5353.7F03702B@sisun9.cselt.it>

Hi everyone

I wrote a message because I had conflict problem with MPC860SAR.
Now I think that the proble arise because the UTOPIA use the DREQ0
signal as Rxcav.
If the Ethernet controller use IDMA1 channel the conflict is inevitable!

Do someone know if the ethernet controller of VxWorks use IDMA1?

Thanks for help
                Tricarico Giuseppe


---------------------------

Newsgroups: comp.os.vxworks
Subject: can i get some documents?
Date: Mon, 22 Feb 1999 15:04:04 +0900
From: =?iso-8859-1?Q?=B1=E8=BF=F8=B1=B9?= <kugi@cs.kyungpook.ac.kr>
Organization: System Engineering Research Institute (SERI)
Message-ID: <36D0F354.E0AD5819@cs.kyungpook.ac.kr>

hello.
I am trying to development video-player in VxWorks.
but I have not gotton VxWorks. May be It will given to me after 1 month.

So, I cannot process my jobs.
Can i get some VxWorks documents(system API and Programming Guide)
and Some Source code ???
thanks for reading..
bye.



---------------------------

Newsgroups: comp.os.vxworks
Subject: VxWorks 5.3 Interrupt Service Routine
Date: Sun, 21 Feb 1999 22:28:57 -0800
From: MAXi <cheong@cs.ucdavis.edu>
Organization: University of California, Davis
Message-ID: <36D0F929.D122363B@cs.ucdavis.edu>

Hi,

Just a brief question about the function in VxWorks:

        inConnect(INUM_TO_IVEC (some_num), myISR, some_val);

I used this to connect to hardware interrupt but I get compile error as
the type "INUM_TO_IVEC" is not found.

Does anyone know which .h file this is declared in?

Thanks in advance.


Regards,

Maxwell
==================================================================
                                   Maxwell CHEONG
 University of California (Davis)    Address:
 Computer Engineering, Junior           Rm 302, Pierce J Building
 Studnet ID#: 991-023926                   Tercero Hall Circle
 Phone: (530) 754 0352                       University of California
(Davis)
 email: cheong@cs.ucdavis.edu          DAVIS, CA 95616
==================================================================



---------------------------

Newsgroups: comp.os.vxworks
Subject: Last Day's Prophecy Website
Date: 22 Feb 1999 09:05:58 GMT
From: Shawn Astels<sorel@globetrotter.qc.ca>
Organization: GlobeTrotter
Message-ID: <7ar6lm$oeh$16588@news.quebectel.com>

Hello, I hope that everyone will stop in and visite my Last Day's Prophecy Bible Website,
 My website includes information on False Bible Versions, Salvation, The Antichrist, Y2K Year 2000 Bug,  And Alot More.
Everything to do with the Second Coming of our Lord Jesus Christ in general.

Also you can subscribe to my Last Day's Prophecy Newsletter!

Please don't forget to sign my guestbook.

MAY GOD BLESS YOU ALL.

URL OF MY SITE: http://www.geocities.com/Athens/Agora/7197/
MY E-MAIL: sorel@globetrotter.qc.ca

---------------------------

Newsgroups: comp.os.vxworks
Subject: WindFoundationClass (WFC)
Date: Mon, 22 Feb 1999 12:39:20 +0100
From: "Markus Mitterer" <markus_mitterer@sbu1.storkgroup.com>
Organization: Customer of EUnet Austria
Message-ID: <7arfne$d5p$1@fleetstreet.Austria.EU.net>

Hi,
I have bought the WindFoundationClasses (WFC) due to the fact that I want to
work with C++. With the WFC-packet I also got a book named "C++ Language
System" from WindRiver with a chapter "The Task Library".
They create task-objects and let the task run in the constructor of the
object and do other strange things (for me).
Please let me know if there is anybody who has used this class and can tell
me about his experience.
I also could not find the h-Files they are using???

thanks in advance,

Markus Mitterer, STK



---------------------------

End of New-News digest
**********************


From owner-vxwexplo-process  Mon Feb 22 05:01:27 1999
From: Michel LESPINASSE <walken@wrs.com>
Date: Mon Feb 22 05:01:31 PST 1999
Subject: Re: VxWorks->Linux:  Last call for votes


vxWorks

> Submitted-by: LJR <leer@voicenet.com>
>
> My vote is no for Linux support. I offer the following article excerpt in
> support:
>
> This is from Jack Ganssle's most recent "Embedded Muse" newsletter.
>
> ----------------------------
> The Trouble with Open Source
> ----------------------------
> [clueless article deleted]

Err, how is this article relevant to the question ? There has been some
talks about porting Tornado to Linux here, but I didnt heard anything
about open sourcing this port.

By the way I didnt found that the quoted article was particularly good
either. The author goes to great lenghts explaining why he does not want
to look at source code. But he does not have to ! Do you use GCC ? Is it a
bad compiler, just because its source code is available ???

P.S. I dont speak for my employer here

--
Michel "Walken" LESPINASSE - http://www.via.ecp.fr/~walken/



From owner-vxwexplo-process  Mon Feb 22 06:21:14 1999
From: ksamavedam@hns.com (Krishna Samavedam)
Date: Mon Feb 22 06:21:17 PST 1999
Subject: RE:VxWorks 5.3 Interrupt Service Routine


The hdr file is located in ../target/h/arch/iv<archType>.h
ex: for mips :
	../target/h/arch/mips/ivMips.h

HTH
krishna




---------------------------

Newsgroups: comp.os.vxworks
Subject: VxWorks 5.3 Interrupt Service Routine
Date: Sun, 21 Feb 1999 22:28:57 -0800
From: MAXi <cheong@cs.ucdavis.edu>
Organization: University of California, Davis
Message-ID: <36D0F929.D122363B@cs.ucdavis.edu>

Hi,

Just a brief question about the function in VxWorks:

        inConnect(INUM_TO_IVEC (some_num), myISR, some_val);

I used this to connect to hardware interrupt but I get compile error as
the type "INUM_TO_IVEC" is not found.

Does anyone know which .h file this is declared in?

Thanks in advance.


Regards,

Maxwell
==============================================
-- 
Krishna Samavedam(x1103)


From owner-vxwexplo-process  Mon Feb 22 08:09:59 1999
From: Harald Grundner <harald.grundner@ri.dasa.de>
Date: Mon Feb 22 08:10:02 PST 1999
Subject: Memory leak on open()/close()?

Dear fellow VxWorks'ers,

why am I losing 304 bytes after opening a file and closing it again?
Here's a shell-based example (parts of memShow() output and "zero"
return values omitted):

-> file = "blah.c"

-> fd = open(file)
fd = 0x23df0a8: value = 5 = 0x5

-> close(fd)

-> memShow
   free   3167608
              ^^^

-> fd = open(file)
fd = 0x23df0a8: value = 5 = 0x5

-> close(fd)

-> memShow
   free   3167304
              ^^^

-> fd = open(file)
fd = 0x23df0a8: value = 5 = 0x5

-> close(fd)

-> memShow
   free   3167000
              ^^^

Anybody having an idea what's going on here?

AdThanksvance for any clue you're able to give,

    Harald




From owner-vxwexplo-process  Mon Feb 22 09:42:08 1999
From: Friedrich Scheurer / ZFS <zfssrr@zfssun6.tz.rus.uni-stuttgart.de>
Date: Mon Feb 22 09:42:12 PST 1999
Subject: Re: VxWorks->Linux:  Last call for votes


> > Submitted-by: LJR <leer@voicenet.com>
> >
> > My vote is no for Linux support. I offer the following article excerpt in
> > support:
> >
> > This is from Jack Ganssle's most recent "Embedded Muse" newsletter.
> >
> > ----------------------------
> > The Trouble with Open Source
> > ----------------------------
> > [clueless article deleted]
> 
> Err, how is this article relevant to the question ? There has been some
> talks about porting Tornado to Linux here, but I didnt heard anything
> about open sourcing this port.
> 
> By the way I didnt found that the quoted article was particularly good
> either. The author goes to great lenghts explaining why he does not want
> to look at source code. But he does not have to ! Do you use GCC ? Is it a
> bad compiler, just because its source code is available ???
> 
> P.S. I dont speak for my employer here
> 
> --
> Michel "Walken" LESPINASSE - http://www.via.ecp.fr/~walken/

Hi Michel, hi VxWorkers,

totally right! I say YES to TORNADO for Linux!

I think that now perhaps a few hundred 
Linux-Users out of the millions are developpers
at the same time. And all the rest 
- as someone posted before -
wants to use an inexpensive, well configurable  
platform with a plenty of fine concepts. 
One big aspect is the weight on continuous development
of stability more than long-distance announcements for new features
on coloured paper, that are implemented and running stable on
UNIX-platforms for decades. This is what professionals
think.

It should not be that problem to port and 
support this new product. 

I even think, that in the future it will
be inevitable to support this platform
if they want or not, because it
comes to a standard and the market
calls for it.

Regards

These are my own opinions!
+-------------------------------------------------+
|                                                 |
| Dipl.-Ing. F. Scheurer                          |
|                                                 |
| c/o Zentrum Fertigungstechnik Stuttgart         |
| Arbeitsbereich Steuerungstechnik                |
| Nobelstrasse 15                                 |
| D-70569 Stuttgart                               |
|                                                 |
| Telefon +49 711 13162-41(-0)                    |
| Fax     +49 711 13162-11                        |
| E-Mail  scheurer@zfs.rus.uni-stuttgart.de       |
|         zfssrr@zfssun6.tz.rus.uni-stuttgart.de  |
+-------------------------------------------------+
| http://zfs.rus.uni-stuttgart.de                 |
+-------------------------------------------------+
|                                                 |
| privat:                                         |
| Telefon +49 7161 686362                         |
| Fax     +49 7161 686301                         |
| E-Mail  Friedrich.Scheurer@t-online.de          |
|                                                 |
+-------------------------------------------------+



From owner-vxwexplo-process  Mon Feb 22 10:25:59 1999
From: "Bruce O. Benson" <benson@sparta.com>
Date: Mon Feb 22 10:26:03 PST 1999
Subject: Fwd: Re: VxWorks->Linux:  Last call for votes

Submitted-by: LJR <leer@voicenet.com>

>My vote is no for Linux support. I offer the following article excerpt in
>support:

>This is from Jack Ganssle's most recent "Embedded Muse" newsletter. 
>
[...insane spin deleted...]

I am flabbergasted that someone posted this to a list of deep kernel
hackers and device-driver geeks.
The article does not a merit a point-by-point response.

On behalf of the 129 respondents to the survey poll:  Rubbish!

WRS is well aware that we are not asking them to open source VxWorks. 
eCos, RTEMS, RTLinux, and KURT currently fulfill that role.  He who
writes the code dictates the license.

Like Eric Raymond asked me once:  Do you want the software [host OS in
our case] to follow your business plan, or your vendor's?

</flame>

BB.
Bruce Benson, Senior Engineer,
SPARTA (CPD)
7926 Jones Branch Drive, Suite 900
McLean, VA  22102-3365
USA
+01 703-448-0210 x211
benson@sparta.com
http://www.mclean.sparta.com


From owner-vxwexplo-process  Mon Feb 22 15:07:39 1999
From: john.c.calvin@exgate.tek.com
Date: Mon Feb 22 15:07:43 PST 1999
Subject: RE: Embedded Java ?

Greetings:

Re:
>Apparently HP is already shipping and Sun pJava for embedded is and has
>been in Beta forever.
>==Anyone know when pJava will be out of Beta ?
>==When will a Sun pJava port to VxWorks be available as "FCS" (not the beta
>marketing hype)  ?

I'd like to know a bit more about this HP pJava implementation as well.
Like, who's class libraries are they using?  It's been my experience that
finding a embeddable "java" VM is not so much an issue these days(in
particular with Kaffe -aka- TransVirtual, Neumonics, NSI, etc.  The issue is
what do you want to call your package when your done with it, and who's
Class libraries are you going to build your Java code against. 

If you have a VM derived from the SUN distribution and it passes the FULL
JCK test suite, you can call it "Java" or "Embedded Java" depending on a few
particulars of the VM.  If you can't satisfy the above criteria, (I belive
HP, Neumonics, TransVirtual fall into this category, but am not sure) you
don't have a "Java" Virtual Machine, at least from the name licensing
standpoint. You have a "Java Compatible" VM.

If your going to use Java, the only sensible thing to do is to build on top
of the Class Libraries or at least pieces of it.  After all this is where
the real value added is from the standpoint of module re-use and leveraging
off of the 100's of thousands of hours of collective development time which
the SUN class libraries represent.  Again, it's been my experience that you
cannot license the use of the SUN class libraries to use in your product
unless you use a SUN derived VM.  You then have to persue something like
TransVirtual which has developed a clean-room class library reproduction for
Kaffe and derivatives.  

WRS Engineering Services just completed a E-Java solution for us, on the 68K
architecture and we started shipping product based on this work product last
month.  I know they were really busy working on similar derivates of E-Java
at last contact, so I would not be surprised if they had some form of formal
announcement soon.

FYI: check out 
http://java.sun.com/products/embeddedjava/spec/1.0/eJavaTechnicalOverview.ht
ml


Cheers,

  John Calvin, MBD, Oscilloscope Application Engineering, Tektronix Inc.
  14150 S.W.  Karl Braun Drive  P.O. Box 500,  M.S. 39-732
  Beaverton OR 97077  U.S.A.
  John.C.Calvin@tek.com, Tel:503-627-4465, Page:503-497-1729,
Fax:503-627-5548 


From owner-vxwexplo-process  Mon Feb 22 18:21:06 1999
From: Roger Cruz <cruz_nojunk@nbase-xyplex.com>
Date: Mon Feb 22 18:21:10 PST 1999
Subject: Can't get out of the idle loop...

We've been experiencing a 2-3 "hangs" every few weeks.  The shell task is hung and 
the LCD/LEDs in our bsp486 product, controlled by a separate task, is also dead.  We
take an NMI dump when this occurs and we have observed a very peculiar behavior that 
I hope some of you may shed some light on.

All indications is that the VxWorks kernel is not receving "events" (semGives, 
tickAnnounce, etc).  We're sure interrupts are enabled (the NMI exception frame 
shows the FLAGs register and it is enabled).  Interrupts are also enabled at the 
PIC (we read the PIC during the dump).  You may wonder how I know tickAnnounces 
are not being received by the kernel?  Easy, vxTicks which is incremented in every
tickAnnounce can be used to compute what the current time is IF you know the boot 
time, which we do record.  The derived time matches the last time shown on the LCD
when it froze. 

When we view all 70 tasks in our system, they are all pending or delayed.  
taskLock has not been called. The tasks priority are OK and the shell, with a 
priority of 1, and waiting on input from the serial port, is also shown pending.  
The work queue is empty and shows mostly deferred windWdStart() events in it.  
We do have a watchdog timer which goes off every tick and self arms itself again 
in the handler (this may be a clue). 

I have exhausted all possibilities and I'm looking for any new ideas from the 
group.  I've included all the info I have on this crash below.   Our idleTask 
is nothing more than a while loop keeping a counter to figure out how much time 
we're idle.


When the crash took place, we know the following about the OS state:
 
Interpretation of CPU flags register: 0x202
  - CF (Carry Flag)       : 0
  - 1  (Always 1)         : 1
  - PF (Parity Flag)      : 0
  - 0  (Always 0)         : 0
  - AF (Aux. Carry Flag)  : 0
  - 0  (Always 0)         : 0
  - ZF (Zero Flag)        : 0
  - SF (Sign Flag)        : 0
  - TF (Trap Flag)        : 0
  - IF (Int. Enable Flag) : 1 (interrupts enabled)
  - DF (Direction Flag)   : 0
  - OF (Overflow Flag)    : 0
 
The task that was running was:
  - taskIdCurrent (0x54d7fc) = 0x1ed4bc8 (tIdleTask)
 
If current task's lockCnt is not zero, then taskLock() has been
called as many times as indicated by the number.
  - taskIdCurrent->lockCnt = 0
 
If kernelState = 1, then we were executing VxWorks kernel code
  - kernelState (0x54bcf0) = 0. 
 
If kernelIsIdle = 1, then we were in the kernel idle loop
  - kernelIsIdle (0x56d2f4) = 0. 
 
If intCnt > 0 then we were at ISR level.  A value of 1, means we
were in a nested interrupt.
  - intCnt (0x4f6098) = 0.
 
The readyQHead queue (0x5763e8) links tasks in the READY state, in priority
order.  The nodes for the link list are in each tasks' TCB.  The tasks 
ready to run are found by following the first node: pFirstNode.
 
  Task #1 0x1ed4bc8 Prio: 255(tIdleTask)
 
The Priority of the current task (tIdleTask) is: 255
 
Timer events currently present in tickQHead. The current tick # is 0x37ad7a.
--------------------------------------------------------------------------------
| Timer  | Expires   | Exp. in |  Object  |    Task      | Timer    | Timer    |
| Number | in tick # | N ticks |  Type    |    Name      | Argument | Handler  |
--------------------------------------------------------------------------------
|      1 |    37ad7b |       1 | Delayed  |         Leds |   N/A    |    N/A   |
|      2 |    37ad7b |       1 | Watchdog |     N/A      |        0 | 0x32276c <tlib_tmr_tick> |
|      3 |    37ad7b |       1 | Delayed  |   t1e1OpTask |   N/A    |    N/A   |
|      4 |    37ad7c |       2 | Watchdog |     N/A      |        0 | 0x26b994 <OspfTimerHandler> |
|      5 |    37ad7c |       2 | Delayed  | frdlLmiProce |   N/A    |    N/A   |
|      6 |    37ad7c |       2 | Delayed  | frdlAnsiProc |   N/A    |    N/A   |
|      7 |    37ad7c |       2 | Watchdog |     N/A      |        0 | 0x1d40e4 <IpMgmtTimerHandler> |
|      8 |    37ad7c |       2 | PosixTm  |     tcircuit |        0 | 0x3c37c8 <CircTimeoutHandler> |
|      9 |    37ad7c |       2 | PosixTm  |          PPP |        0 | 0x389e40 <PppProcTimerSend> |
|     10 |    37ad7c |       2 | PosixTm  | frdlTimedPro |        0 | 0x1b1210 <frdlTimedProcHdle> |
|     11 |    37ad7c |       2 | PosixTm  |      TcpProc |        0 | 0x2388e8 <TcpSlowTimerHandler> |
|     12 |    37ad7c |       2 | PosixTm  |      twsApps |        0 | 0x3ce324 <WsTimeoutHandler> |
|     13 |    37ad8d |      19 | Watchdog |     N/A      |   5765e0 | 0x168248 <EdcLinkStatus> |
|     14 |    37ad9d |      35 | PosixTm  |   tRadMaster |        0 | 0x39b930 <RadTimeoutHandler> |
|     15 |    37ad9d |      35 | PosixTm  |   tDcpMaster |        0 | 0x14c720 <DcpTimeoutHandler> |
|     16 |    37adab |      49 | PosixTm  |    ISDNTIMER |        0 | 0x2dda28 <IsdnTimerTaskHdl> |
|     17 |    37adaf |      53 | PosixTm  |       twsmgt |        0 | 0x3fa650 <UtilCounterTimeoutHandler> |
|     18 |    37ade6 |     108 | PosixTm  |      mdmTask |        0 | 0x3301f8 <MdmTimerTaskHdl> |
|     19 |    37ae3f |     197 | PosixTm  |       twsmgt |        0 | 0x3d51f8 <WsmgtTimeoutHandler> |
|     20 |    37af5c |     482 | Watchdog |     N/A      |        0 | 0x122670 <CpuMonitorHandler> |
|     21 |    37b0f4 |     890 | Watchdog |     N/A      |        0 | 0x1f89dc <IpProcTimerHandler> |
|     22 |    37bb35 |    3515 | PosixTm  |   tDcpMaster |        0 | 0x14b680 <DhcpLeaseTimerHandler> |

The current time up is contained in vxTicks (0x56c7dc).  Each tick
represents 1/60th of second.  The up-time is displayed below:
  - Total ticks: 3648890 (0x37ad7a)
  -  in seconds: 60814.83
  -  in minutes: 1013.58
  -  in hours  : 16.89
  -  in days   : 0.70
 
The 64-bit tick counter vxAbsTicks (0x547088):
{lower = 0x37ad7a, upper = 0x0}

The read & write indexes are used to extract and add information to queue.
 - workQReadIx  (0x5440e8) = 0xf4 Entry #61
 - workQWriteIx (0x54bf00) = 0xf4 Entry #61
 - There are 0 pending events in the workQ.
 - The state of the variable workQIsEmpty (0x579990) is: 1
 
The deferred work is stored in the array pJobPool of size 64*16 bytes.
Here are ALL the entries in that array

         ----------------------------------------------------------------
         | Entry  | Argument | Argument | Argument | Function Name of   |
         | Number |     1    |     2    |     3    | Deferred Work      |
         ----------------------------------------------------------------
         |     0  |  114598c |        1 |        0 | 0x465f24 <windWdStart> | 
         |     1  |  114598c |        1 |        0 | 0x465f24 <windWdStart> | 
 ..........
Rd=WrIx=>|    61  |  114598c |        1 |        0 | 0x465f24 <windWdStart> | 
         |    62  |  114598c |        1 |        0 | 0x465f24 <windWdStart> | 
         |    63  |  114598c |        1 |        0 | 0x465f24 <windWdStart> | 
         ----------------------------------------------------------------

 Info when dump took place:

  systemStartTime = "FEB 19 21:43:43:1999", '\000' <repeats 20 times>, 
  crashTime = "FEB 20 14:37:15:1999", '\000' <repeats 20 times>}
  lastIsr = 0x166928 <EdcInt>, lastWatchdog = 0x32276c <tlib_tmr_tick>
 
  exceptionPc = 0x1225ee

0x1225ee is in IdleProcess (system_server.c:186).
169      */
170     GLOBAL VOID IdleProcess()
171     {
172             while (TRUE)
173             {
174             /* Ideally, should always stay in this loop. */
175             macIdleTimingLoop();
...     
184             cpuMonitor.bWaitForMonitor = TRUE;
185     
186             while (cpuMonitor.bWaitForMonitor);     /* Wait for timer */
187     
188             cpuMonitor.idleLoopMax++;       /* Adjust the idle loop */
189