1999232 - Crash in [@ begin_rendering] using mesa/zink 25.2.x drivers on NVidia hardware

Status: NEW

(Core :: Graphics, defect)

Description

[Gabriele Svelto [:gsvelto]]

Crash report: https://crash-stats.mozilla.org/report/index/3fdce655-92ea-404a-83f9-cae960251110

Reason:

SIGSEGV / SEGV_MAPERR

Top 10 frames:

0  libgallium-25.2.5-1.so  begin_rendering  src/gallium/drivers/zink/zink_context.c:3101
1  libgallium-25.2.5-1.so  zink_batch_rp  src/gallium/drivers/zink/zink_context.c:3322
2  libgallium-25.2.5-1.so  zink_draw<(zink_multidraw)1, (zink_dynamic_state)5, false, false>(pipe_contex...  src/gallium/drivers/zink/zink_draw.cpp:452
3  libgallium-25.2.5-1.so  tc_call_draw_single  src/gallium/auxiliary/util/u_threaded_context.c:3814
3  libgallium-25.2.5-1.so  batch_execute  src/gallium/auxiliary/util/u_threaded_context_calls.h:11
3  libgallium-25.2.5-1.so  tc_batch_execute  src/gallium/auxiliary/util/u_threaded_context.c:5331
4  libgallium-25.2.5-1.so  util_queue_thread_func  src/util/u_queue.c:294
5  libgallium-25.2.5-1.so  impl_thrd_routine  src/c11/impl/threads_posix.c:43
6  firefox-esr  set_alt_signal_stack_and_start(PthreadCreateParams*)  mozglue/interposers/pthread_create_interposer.cpp:81
7  libc.so.6  start_thread  nptl/nptl/pthread_create.c:448

The volume is significant for being a Linux crash but it doesn't look like something we can fix on our side. I'll see if there is a corresponding bug in mesa's tracker. Note that the ESR population may seem more affected than release, but that's partly caused by how we filter incoming crash reports, we're likely undercounting the affected users in the release channel.

Comment 1

[Gabriele Svelto [:gsvelto]]

Filed a bug in mesa. The upstream code has changed a lot so it's hard to tell if the issue has already been fixed in the current mainline or if it's still present.

Comment 2

[Ashley Hale [:ahale]]

Triage note - I'm not sure how we rate severity for a crash volume like this on Linux?

Comment 3

[Gabriele Svelto [:gsvelto]]

It seems half of the crashes have this entry logged among the graphics critical errors:

RenderCompositorSWGL failed mapping default framebuffer, no dt

Could this be related to the crash? The mesa developers were asking if we had some other info about what happened before the crash.

Comment 4

[Glenn Watson [:gw]]

That warning is not relevant here