Problem: Commands failing.

Hello:

I have a v4.22 system that is exhibiting some strange behaviour.

We got called by our operator about a system that had its
main application fail after its data files were updated.
We found that it was unable to start some of the application
programs successfully.

While out at the site, we tried to compare some of the data
file versions using “diff”, but “diff” would fail with the
message:

//65/bin32/diff terminated (SIGSEGV) at 0007:0000D6BC
%1 29294 Memory fault diff /data/data1 /data/data2

The failure address doesn’t change. We tried comparing
two identical files. diff would find no differences and
exit with no errors. If we compared an editted copy of
one file with its own original, diff would fail.

We replaced the RAM (because of the error message) but
this made no difference. Neither did scrubbing the HD
and reinstalling the OS.

Later, we happened to try using “freeze” to recompress
a tar file. freeze exitted with the following error:

Segmentation violation occurred.

I’m probably going to try changing the SBC next, and try
to review the BIOS setting, but I’m also looking for other
suggestions. Anyone?

Thanks in advance.
Kevin Beauchamp

“Kevin.Beauchamp” <Kevin.Beauchamp@ualberta.ca> wrote in message
news:3A5F40BD.DDFC1A8C@ualberta.ca

Hello:

I have a v4.22 system that is exhibiting some strange behaviour.

We got called by our operator about a system that had its
main application fail after its data files were updated.
We found that it was unable to start some of the application
programs successfully.

While out at the site, we tried to compare some of the data
file versions using “diff”, but “diff” would fail with the
message:

//65/bin32/diff terminated (SIGSEGV) at 0007:0000D6BC
%1 29294 Memory fault diff /data/data1 /data/data2

The failure address doesn’t change. We tried comparing
two identical files. diff would find no differences and
exit with no errors. If we compared an editted copy of
one file with its own original, diff would fail.

Where the files ASCII, diff is I beleive meant for ASCII files not
binaries.

We replaced the RAM (because of the error message) but
this made no difference. Neither did scrubbing the HD
and reinstalling the OS.

Later, we happened to try using “freeze” to recompress
a tar file. freeze exitted with the following error:

Segmentation violation occurred.

I’m probably going to try changing the SBC next, and try
to review the BIOS setting, but I’m also looking for other
suggestions. Anyone?

I’d say you are on the right track changing the SBC.
You might also be looking at HD problem, run dcheck.

If program are loaded from the network the NIC could
potentionaly be the cause.




Thanks in advance.
Kevin Beauchamp

Previously, Mario Charest wrote in qdn.public.qnx4:

“Kevin.Beauchamp” <> Kevin.Beauchamp@ualberta.ca> > wrote in message
news:> 3A5F40BD.DDFC1A8C@ualberta.ca> …
Hello:

I have a v4.22 system that is exhibiting some strange behaviour.

We got called by our operator about a system that had its
main application fail after its data files were updated.
We found that it was unable to start some of the application
programs successfully.

While out at the site, we tried to compare some of the data
file versions using “diff”, but “diff” would fail with the
message:

//65/bin32/diff terminated (SIGSEGV) at 0007:0000D6BC
%1 29294 Memory fault diff /data/data1 /data/data2

The failure address doesn’t change. We tried comparing
two identical files. diff would find no differences and
exit with no errors. If we compared an editted copy of
one file with its own original, diff would fail.


Where the files ASCII, diff is I beleive meant for ASCII files not
binaries.

diff shouldn’t crash like that… but in any event, you might want to try using “cmp” to compare the files.

  • Pete

Mario Charest wrote:

“Kevin.Beauchamp” <> Kevin.Beauchamp@ualberta.ca> > wrote:

…“diff” would fail with the message:

//65/bin32/diff terminated (SIGSEGV) at 0007:0000D6BC
%1 29294 Memory fault diff /data/data1 /data/data2

The failure address doesn’t change. We tried comparing
two identical files. diff would find no differences and
exit with no errors. If we compared an editted copy of
one file with its own original, diff would fail.


Where the files ASCII, diff is I beleive meant for ASCII files not
binaries.

The ones I was testing in the shop were ASCII. In fact, I created
two versions of /etc/profile to run comparisons on.

We replaced the RAM (because of the error message) but
this made no difference. Neither did scrubbing the HD
and reinstalling the OS.

Later, we happened to try using “freeze” to recompress
a tar file. freeze exitted with the following error:

Segmentation violation occurred.

I’m probably going to try changing the SBC next, and try
to review the BIOS setting, but I’m also looking for other
suggestions. Anyone?

I’d say you are on the right track changing the SBC.
You might also be looking at HD problem, run dcheck.

I’ll give dcheck a try.

If program are loaded from the network the NIC could
potentionaly be the cause.

At one point, I thought that it could be related, since I was
comparing a file on the local HD to one on another node’s
HD, but I later went to local/local comparisons and the problem
persisted.

Thanks
Kevin B.

Pete DiMarco wrote:

diff shouldn’t crash like that… but in any event, you might want to try using “cmp” to compare the files.

diff runs fine on every other node on our net. The problem I’m having in on node 65.
If I am logged onto node 65 locally, typing:

diff /etc/config/sysinit.65 /etc/config/sysinit

will cause diff to fail. Typing:

on -n 1 diff /etc/config/sysinit.65 /etc/config/sysinit

will succeed and return the expected results.

Thanks
Kevin B.

This is not exactly the same but I had a problem with commands in my sysinit failing. It turned out to be fixed by
running chkfsys on the drive. I now have chkfsys -Pqrsu / in my sysinit.

My guess valued at 1/2 cent.

KenR

David Gibbs wrote:

Kevin.Beauchamp <> Kevin.Beauchamp@ualberta.ca> > wrote:

Pete DiMarco wrote:

diff shouldn’t crash like that… but in any event, you might want to try using “cmp” to compare the files.

diff runs fine on every other node on our net. The problem I’m having in on node 65.
If I am logged onto node 65 locally, typing:

diff /etc/config/sysinit.65 /etc/config/sysinit

Hm… corrupted shared library on that machine, maybe?
Very odd.

-David

QNX Training Services
dagibbs@qnx.com

Kevin.Beauchamp <Kevin.Beauchamp@ualberta.ca> wrote:


Pete DiMarco wrote:

diff shouldn’t crash like that… but in any event, you might want to try using “cmp” to compare the files.

diff runs fine on every other node on our net. The problem I’m having in on node 65.
If I am logged onto node 65 locally, typing:

diff /etc/config/sysinit.65 /etc/config/sysinit

Hm… corrupted shared library on that machine, maybe?
Very odd.

-David

QNX Training Services
dagibbs@qnx.com

“Kevin.Beauchamp” <Kevin.Beauchamp@ualberta.ca> wrote in message
news:3A5F6591.28157045@ualberta.ca

Mario Charest wrote:

“Kevin.Beauchamp” <> Kevin.Beauchamp@ualberta.ca> > wrote:

…“diff” would fail with the message:

//65/bin32/diff terminated (SIGSEGV) at 0007:0000D6BC
%1 29294 Memory fault diff /data/data1 /data/data2

The failure address doesn’t change. We tried comparing
two identical files. diff would find no differences and
exit with no errors. If we compared an editted copy of
one file with its own original, diff would fail.


Where the files ASCII, diff is I beleive meant for ASCII files not
binaries.

The ones I was testing in the shop were ASCII. In fact, I created
two versions of /etc/profile to run comparisons on.

Do you mean that this happends at different location?

If so maybe your new program is corrupting Slib32 or Slib16.

We replaced the RAM (because of the error message) but
this made no difference. Neither did scrubbing the HD
and reinstalling the OS.

Later, we happened to try using “freeze” to recompress
a tar file. freeze exitted with the following error:

Segmentation violation occurred.

I’m probably going to try changing the SBC next, and try
to review the BIOS setting, but I’m also looking for other
suggestions. Anyone?

I’d say you are on the right track changing the SBC.
You might also be looking at HD problem, run dcheck.

I’ll give dcheck a try.

If program are loaded from the network the NIC could
potentionaly be the cause.

At one point, I thought that it could be related, since I was
comparing a file on the local HD to one on another node’s
HD, but I later went to local/local comparisons and the problem
persisted.

Thanks
Kevin B.

Previously, Kevin.Beauchamp wrote in qdn.public.qnx4:

Pete DiMarco wrote:

diff shouldn’t crash like that… but in any event, you might
want to try using “cmp” to compare the files.

diff runs fine on every other node on our net. The problem I’m
having in on node 65.
If I am logged onto node 65 locally, typing:

diff /etc/config/sysinit.65 /etc/config/sysinit

will cause diff to fail. Typing:

on -n 1 diff /etc/config/sysinit.65 /etc/config/sysinit

will succeed and return the expected results.

We had a similar problem where we had two sticks of RAM in the
machine, and the BIOS was incorrectly set to use parity RAM on the
second one when it was in fact non-parity RAM. We saw programs work
perfectly until, presumably, they started using RAM from the second
stick. Then they crashed with all manner of strange data and
executable errors.

Just another 2 cents…

Andrew

I don’t know if this is related, but I’ve had a strange
problem ever since i upgraded to a 650Mhz PIII w 256meg.

Very occaisionally I’ll run a standard utility, sin
comes to mind, and it will SIGSEGV immediately. If
I repeat the command it may fail a few more times, but
it eventually works ok. While disconcerting, I’ve never
seen any other behavior that might be related, such as
disk/file corruption, or running programs exit.

Mitchell Schoenbrun --------- maschoen@pobox.com