Comment 45 for bug 1492037

I have probably the same issue, the backtrace on the crash is a bit different, but my card might not have fast_clear.

I have the

Failed to open BO for returned DRI2 buffer

message before the crash. And it doesn't happen with uxa or sna+dri3.

I have an apitrace trace that makes it happen for me in 4 out of 5 times.

run with (after decompressing)
$ glretrace replay test_case.trace

It's 14MB so I uploaded it on an external site. (https://mega.co.nz/#!O0MHRByI!z7GywnuO8Ai_9633pmj5FY8ejzvKYBsDAri1lXFSXN0)

Software versions:
------------------
Linux 4.1.4 (Archlinux)
mesa 10.6.3
libdrm 2.4.62+106+gc8df9e7-1 (git master at the time of writing)
xf86-video-intel 2.99.917+426+g611ec7d-1 (git master)

Hack to fix it in gdb
----------------------
After drm_intel_bo_gem_create_from_name fails(brw_context.c:1421), if I run (from gdb) getBuffersWithFormat and drm_intel_bo_gem_create_for_name again
then it succeds.
(This is what has to be done when dri2 buffers got invalidated)

Backtrace when drm_intel_bo_gem_create_from_name fails
------------------------------------------------------
#0 intel_process_dri2_buffer (buffer_name=0x7ffff21ddf75 "dri2 back buffer", rb=0xaf2cb0, buffer=0xaf2b50, drawable=0xa76ad0, brw=0x7ffff7fd1038)
    at brw_context.c:1423
#1 intel_update_dri2_buffers (drawable=0xa76ad0, brw=0x7ffff7fd1038) at brw_context.c:1226
#2 intel_update_renderbuffers (context=context@entry=0xbaea90, drawable=drawable@entry=0xa76ad0) at brw_context.c:1248
#3 0x00007ffff203d5b1 in intel_prepare_render (brw=brw@entry=0x7ffff7fd1038) at brw_context.c:1267
#4 0x00007ffff2031290 in brw_clear (ctx=0x7ffff7fd1038, mask=18) at brw_clear.c:234
#5 0x00000000004e8c96 in ?? ()
#6 0x000000000040bd1d in ?? ()
#7 0x000000000040c37c in ?? ()
#8 0x0000000000407b05 in ?? ()
#9 0x00007ffff61e0790 in __libc_start_main () from /usr/lib/libc.so.6
#10 0x00000000004095e9 in _start ()