Control: reassign -1 libegl-mesa0 24.1.0-2
Control: affects -1 firefox
Control: tags -1 fixed-upstream patch
Control: forwarded -1
https://gitlab.freedesktop.org/mesa/mesa/-/issues/11352
On Tue, 04 Jun 2024 02:22:58 -0400 Kevin Israel <pleasest...@live.com>
wrote:
> Commands to reproduce:
>
> rm -rf freshprofile && mkdir freshprofile
> firefox --no-remote --profile $PWD/freshprofile
https://www.irs.gov/pub/irs-pdf/f1040.pdf
>
> Resulting crash report:
>
>
https://crash-stats.mozilla.org/report/index/e8c25505-10d1-4faf-8c23-eab6b0240604
>
> Also, a full stack trace from a separate run using gdb is attached.
Yesterday, I encountered a crash in the same libegl-mesa0 function even
when opening other apps such as gnome-text-editor. That crash also ended
my user session, unlike the firefox crash. Turns out this was reported
upstream on June 17:
https://gitlab.freedesktop.org/mesa/mesa/-/issues/11352
The version of libegl-mesa0 this time was 24.1.3-2. At the time I filed
this bug report, I had just upgraded to 24.1.0-2.
I rebuilt the mesa packages to add the four patches that were merged
into the staging/24.1 branch:
radeonsi: fix buffer_size in si_compute_shorten_ubyte_buffer
https://gitlab.freedesktop.org/mesa/mesa/-/commit/7d35a5d5023b702a72bad194a3978a591a341e41
Revert "ac, radeonsi: remove has_syncobj, has_fence_to_handle"
https://gitlab.freedesktop.org/mesa/mesa/-/commit/3340392efa86786efd07ae6206803ded8e9192fd
winsys/radeon: fill lds properties
https://gitlab.freedesktop.org/mesa/mesa/-/commit/15a02d9814567b9118c1698603b4646daacf0659
radeonsi: fix crash in si_update_tess_io_layout_state for gfx8 and earlier
https://gitlab.freedesktop.org/mesa/mesa/-/commit/cecc809ab476cf08b0d1bb0e97aec0a30216dddc
Not only did that problem go away, but also I cannot reproduce this
crash any more. So I think that this crash probably was caused by the
mesa bug.
