Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 03 Oct 2017 16:27:49 +0000
From:      bugzilla-noreply@freebsd.org
To:        x11@FreeBSD.org
Subject:   maintainer-feedback requested: [Bug 222757] Some problems in mesa ports related to OpenCL
Message-ID:  <bug-222757-7141-qDrAuSVVkI@https.bugs.freebsd.org/bugzilla/>
In-Reply-To: <bug-222757-7141@https.bugs.freebsd.org/bugzilla/>
References:  <bug-222757-7141@https.bugs.freebsd.org/bugzilla/>

next in thread | previous in thread | raw e-mail | index | archive | help
shamaz.mazum@gmail.com has reassigned Bugzilla Automation
<bugzilla@FreeBSD.org>'s request for maintainer-feedback to x11@FreeBSD.org:
Bug 222757: Some problems in mesa ports related to OpenCL
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=3D222757



--- Description ---
Created attachment 186882
  --> https://bugs.freebsd.org/bugzilla/attachment.cgi?id=3D186882&action=
=3Dedit
Gallium pipe-loader patch

Hello. I've tried OpenCL using lang/clover provider and encountered some
problems in mesa ports (can be fixed patching graphics/mesa-dri). I use Fre=
eBSD
11.1 and have the port tree updated with portsnap and packages updated with
portupgrade

1) OpenCL applications do not work in X11 environment. I have exactly the s=
ame
problem as described here: bug #214204 (*) The fix is still the same, maybe
this is a regression. I attach patch which is combined from an old patch in
graphics/mesa-dri/files and the fix. I still do not know how to link -lxcb =
to
opencl targets. Autoconf is very complicated for me, so I do not know what =
to
do.

2) When I call clReleaseContext() from my program, it finishes OK, but prog=
ram
crashes with segmantation fault when calls atexit handlers, having very str=
ange
backtrace:

(gdb) bt
#0  0x0000000808534500 in ?? ()
#1  0x0000000800b754d5 in __cxa_finalize (dso=3D0x0) at
/usr/src/lib/libc/stdlib/atexit.c:237
#2  0x0000000800b063d1 in exit (status=3D0) at /usr/src/lib/libc/stdlib/exi=
t.c:72
#3  0x0000000000400da6 in _start ()
#4  0x0000000800625000 in ?? ()
#5  0x0000000000000000 in ?? ()

clinfo has the same behaviour. I do not know the solution. Any suggestions?

(*) https://lists.freebsd.org/pipermail/freebsd-x11/2016-November/018240.ht=
ml



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?bug-222757-7141-qDrAuSVVkI>