[Mesa-dev] [Bug 109532] ir_variable has maximum access out of bounds -- but it's not out of bounds

bugzilla-daemon at freedesktop.org bugzilla-daemon at freedesktop.org
Mon Feb 25 14:02:04 UTC 2019


--- Comment #41 from asimiklit <andrey.simiklit at gmail.com> ---
(In reply to andrii simiklit from comment #40)
> (In reply to Mark Janes from comment #39)
> > I just noticed that this new test passes for 32 bit builds.
> > 
> > Does that surprise anyone else?
> > 
> > http://mesa-ci.01.org/mesa_master_daily/builds/4806/group/
> > e60513df13ade427f01bb7334bd5174e
> Thanks that pointed that out.
> I didn't see any platform specific code there.
> I will post an update here as far as I figure out the reason.

Unfortunately I can't to reproduce this behavior locally.
I built the 32-bit debug mesa + 32-bit debug piglit and have the following
   asimiklit at asimiklit-pc:~/projects/piglit32$ bin/glslparsertest
   /unused-array-element.comp pass 4.50 
   ir_variable has maximum access out of bounds (1 vs 0)
   Aborted (core dumped)

glslparsertest and mesa 100% have 32-bit architecture:
   asimiklit at asimiklit-pc:~/projects/piglit32$ file bin/glslparsertest
   bin/glslparsertest: ELF 32-bit LSB shared object, Intel 80386, 
   version 1 (SYSV), dynamically linked, interpreter /lib/ld-linux.so.2, 
   for GNU/Linux 3.2.0, BuildID[sha1]=0243d96a1caf7c54189447516b066e5582ef86aa,
   with debug_info, not stripped

   asimiklit at asimiklit-pc:~/projects/piglit32$ ldd bin/glslparsertest
   libgbm.so.1 => /home/.../mesa_versions/1802_32_dbg/lib/libgbm.so.1
   libGL.so.1 => /home/.../mesa_versions/1802_32_dbg/lib/libGL.so.1
   libEGL.so.1 => /home/.../mesa_versions/1802_32_dbg/lib/libEGL.so.1
   libglapi.so.0 => /home/.../mesa_versions/1802_32_dbg/lib/libglapi.so.0

   asimiklit at asimiklit-pc:~/projects/piglit32$ file /home/.../mesa_versions
   /home/.../mesa_versions/1802_32_dbg/lib/libGL.so.1.2.0: ELF 32-bit LSB
shared object, 
   Intel 80386, version 1 (SYSV), dynamically linked,    
   BuildID[sha1]=8708ca7f174b267838c5b1ffe71ca68d5307f62d, with debug_info, not

   PS: I know that 1802 in the folder name isn't a correct name
       for the latest mesa folder name but I just forgot to update it in prefix

Mark could you please clarify which mesa configuration is used
with 32-bit piglit, debug or release? Because I see the same behavior
just with a release mesa configuration.

You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/mesa-dev/attachments/20190225/6291fffb/attachment.html>

More information about the mesa-dev mailing list