Debian Bug report logs - #743885
octave: Octave triggers an assertion in Mesa

Display info messages

Message #5 received at submit@bugs.debian.org (full text, mbox, reply):

From: Thomas Weber <tweber@debian.org>

To: Debian Bug Tracking System <submit@bugs.debian.org>

Subject: octave: Octave triggers an assertion in Mesa

Date: Mon, 7 Apr 2014 23:51:46 +0200

[Message part 1 (text/plain, inline)]

Package: octave
Version: 3.8.1-1+b1
Severity: normal

I found this bug first in the normal packaged version of Octave in
Debian. The attached stack traces were created with a local build
without optimization to get meaningful stack traces. The code orginates
from matlab2tikz's test suite.

================================================================
$ cat mesatest.m 
x = -pi:pi/10:pi;
y = tan(sin(x)) - sin(tan(x));
y(3) = NaN;
y(7) = Inf;
y(11) = -Inf;
plot(x,y,'--o', 'Color', [0.6,0.2,0.0], ...
                'LineWidth', 1*360/127,...
                'MarkerEdgeColor','k',...
                'MarkerFaceColor',[0.3,0.1,0.0],...
                'MarkerSize', 5*360/127 );
================================================================

$ gdb -q  --args /usr/bin/octave --quiet --persist --eval "mesatest"
Reading symbols from /usr/bin/octave...Reading symbols from /usr/lib/debug/.build-id/02/aaaba0a5a74803e4a94b21f0fd9873fb6f6a8b.debug...done.
done.
(gdb) r
Starting program: /usr/bin/octave --quiet --persist --eval mesatest
warning: Could not load shared library symbols for linux-vdso.so.1.
Do you need "set solib-search-path" or "set sysroot"?
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
process 27392 is executing new program: /usr/bin/octave-cli
warning: Could not load shared library symbols for linux-vdso.so.1.
Do you need "set solib-search-path" or "set sysroot"?
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
octave-cli: ../../../../../../../src/mesa/drivers/dri/i965/brw_eu_emit.c:267: validate_reg: Assertion `execsize >= width' failed.

Program received signal SIGABRT, Aborted.
0x00007ffff46183a9 in __GI_raise (sig=sig@entry=6) at ../nptl/sysdeps/unix/sysv/linux/raise.c:56
56	../nptl/sysdeps/unix/sysv/linux/raise.c: Datei oder Verzeichnis nicht gefunden.

I am filing the bug right now in the hope that someone has an idea on
how to continue with this - I have no clue whatsoever about Mesa.

	Thomas

[bt-full.txt.gz (application/octet-stream, attachment)]

[bt.txt.gz (application/octet-stream, attachment)]

Message #10 received at 743885@bugs.debian.org (full text, mbox, reply):

From: Mike Miller <mtmiller@debian.org>

To: 743885@bugs.debian.org

Subject: Re: Bug#743885: octave: Octave triggers an assertion in Mesa

Date: Tue, 8 Apr 2014 12:07:18 -0400

On Mon, Apr 7, 2014 at 23:51:46 +0200, Thomas Weber wrote:
> I am filing the bug right now in the hope that someone has an idea on
> how to continue with this - I have no clue whatsoever about Mesa.

I am unable to reproduce on my laptop with Intel integrated graphics,
and I'm using the same i965 driver that your backtrace lists. I don't
know much about mesa either, but it could be that this is hardware
dependent since the assertion seems to be coming specifically from the
i965 driver.

Do you have a different video card you can test on?

Any difference when you run the test with LIBGL_ALWAYS_INDIRECT=y in
your environment?

I see this listed in the mesa upstream bug tracker:
  https://bugs.freedesktop.org/show_bug.cgi?id=36193

-- 
mike

Message #15 received at 743885@bugs.debian.org (full text, mbox, reply):

From: Thomas Weber <tweber@debian.org>

To: 743885@bugs.debian.org

Subject: Re: [Pkg-octave-devel] Bug#743885: octave: Octave triggers an assertion in Mesa

Date: Fri, 11 Apr 2014 13:57:35 +0200

On Tue, Apr 08, 2014 at 12:07:18PM -0400, Mike Miller wrote:
> On Mon, Apr 7, 2014 at 23:51:46 +0200, Thomas Weber wrote:
> > I am filing the bug right now in the hope that someone has an idea on
> > how to continue with this - I have no clue whatsoever about Mesa.
> 
> I am unable to reproduce on my laptop with Intel integrated graphics,
> and I'm using the same i965 driver that your backtrace lists. I don't
> know much about mesa either, but it could be that this is hardware
> dependent since the assertion seems to be coming specifically from the
> i965 driver.
> 
> Do you have a different video card you can test on?

I am running Debian unstable only on my laptop (Lenovo T61). I might be
able to test it on some other hardware, but that would require quite an
investment of time (installing Debian, while keeping my 2.5 year old from
"helping" Daddy :)). And I am lacking time already.

> Any difference when you run the test with LIBGL_ALWAYS_INDIRECT=y in
> your environment?
Yes, my X server crashes! So kids, don't try this at home.

> I see this listed in the mesa upstream bug tracker:
>   https://bugs.freedesktop.org/show_bug.cgi?id=36193
I saw this too, but considered it unrelated as it mentioned VmWare. Hmm,
maybe I should add myself to the bug report, as I can reproduce it with
100% reliability.

	Thomas

Message #20 received at 743885@bugs.debian.org (full text, mbox, reply):

From: Mike Miller <mtmiller@debian.org>

To: Thomas Weber <tweber@debian.org>, 743885@bugs.debian.org

Subject: Re: [Pkg-octave-devel] Bug#743885: Bug#743885: octave: Octave triggers an assertion in Mesa

Date: Tue, 7 Oct 2014 09:57:11 -0400

On Fri, Apr 11, 2014 at 13:57:35 +0200, Thomas Weber wrote:
> On Tue, Apr 08, 2014 at 12:07:18PM -0400, Mike Miller wrote:
> > On Mon, Apr 7, 2014 at 23:51:46 +0200, Thomas Weber wrote:
> > > I am filing the bug right now in the hope that someone has an idea on
> > > how to continue with this - I have no clue whatsoever about Mesa.
> > 
> > I am unable to reproduce on my laptop with Intel integrated graphics,
> > and I'm using the same i965 driver that your backtrace lists. I don't
> > know much about mesa either, but it could be that this is hardware
> > dependent since the assertion seems to be coming specifically from the
> > i965 driver.
> > 
> > Do you have a different video card you can test on?
> 
> I am running Debian unstable only on my laptop (Lenovo T61). I might be
> able to test it on some other hardware, but that would require quite an
> investment of time (installing Debian, while keeping my 2.5 year old from
> "helping" Daddy :)). And I am lacking time already.

Can you still reproduce this on unstable now?

Someone claims on the Ubuntu equivalent bug report [1] that Mesa 10.2.7
fixed it. I see Mesa 10.2.8 is in unstable now. It's a shot in the dark,
but any chance it's fixed for you?

[1] https://bugs.launchpad.net/ubuntu/+source/octave/+bug/1313952/comments/5

-- 
mike

Message #25 received at 743885-done@bugs.debian.org (full text, mbox, reply):

From: Thomas Weber <tweber@debian.org>

To: Mike Miller <mtmiller@debian.org>

Cc: 743885-done@bugs.debian.org

Subject: Re: [Pkg-octave-devel] Bug#743885: Bug#743885: octave: Octave triggers an assertion in Mesa

Date: Mon, 26 Jan 2015 10:16:36 +0100

On Tue, Oct 07, 2014 at 09:57:11AM -0400, Mike Miller wrote:
> On Fri, Apr 11, 2014 at 13:57:35 +0200, Thomas Weber wrote:
> > On Tue, Apr 08, 2014 at 12:07:18PM -0400, Mike Miller wrote:
> > > On Mon, Apr 7, 2014 at 23:51:46 +0200, Thomas Weber wrote:
> > > > I am filing the bug right now in the hope that someone has an idea on
> > > > how to continue with this - I have no clue whatsoever about Mesa.
> > > 
> > > I am unable to reproduce on my laptop with Intel integrated graphics,
> > > and I'm using the same i965 driver that your backtrace lists. I don't
> > > know much about mesa either, but it could be that this is hardware
> > > dependent since the assertion seems to be coming specifically from the
> > > i965 driver.
> > > 
> > > Do you have a different video card you can test on?
> > 
> > I am running Debian unstable only on my laptop (Lenovo T61). I might be
> > able to test it on some other hardware, but that would require quite an
> > investment of time (installing Debian, while keeping my 2.5 year old from
> > "helping" Daddy :)). And I am lacking time already.
> 
> Can you still reproduce this on unstable now?
> 
> Someone claims on the Ubuntu equivalent bug report [1] that Mesa 10.2.7
> fixed it. I see Mesa 10.2.8 is in unstable now. It's a shot in the dark,
> but any chance it's fixed for you?
> 
> [1] https://bugs.launchpad.net/ubuntu/+source/octave/+bug/1313952/comments/5

I can no longer reproduce this issue with the current mesa in unstable
(10.4.2 at the time of writing), so I am closing this bug.

Thanks Mike for helping with it and sorry for the long delay - I just
overlooked your email in October.

	Thomas

Send a report that this bug log contains spam.