Bug#1077800: mutter: flaky test basic-x11: In pixman_region32_init_rect: Invalid rectangle passed; cogl_texture_2d_new_from_data: assertion 'data != NULL' failed

[Simon McVittie]

On Fri, 02 Aug 2024 at 15:02:14 +0100, Simon McVittie wrote:
> Please consider other
> failures to be out-of-scope here unless there is evidence that they share
> a root cause.

Actually, the first few of the other ppc64el failures that I looked at
(in the "stacking" test suite) seem like they might have the same root
cause after all, for example:

> # mutter-test-runner-DEBUG: closed-transient-no-input-no-take-focus-parent.metatest:3: 'show 1/1'
> # mutter-test-runner-DEBUG: closed-transient-no-input-no-take-focus-parent.metatest:5: 'create 1/2 csd'
> *** BUG ***
> In pixman_region32_init_rect: Invalid rectangle passed
> Set a breakpoint on '_pixman_log_error' to debug
> 
> *** BUG ***
> In pixman_region32_init_rect: Invalid rectangle passed
> Set a breakpoint on '_pixman_log_error' to debug
> 
> Bail out! FATAL-CRITICAL: cogl_texture_2d_new_from_data: assertion 'data != NULL' failed
> (EE) failed to read Wayland events: Connection reset by peer
> ――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――
> 152/185 mutter:core+mutter/stacking / closed-transient-no-input-no-take-focus-parent                   FAIL             2.69s   (exit status 251 or signal 123 SIGinvalid)

So I'd suggest that our debugging efforts concentrate on basic-x11
for now, because this failure mode has been seen in that test even on
amd64, and its name suggests that it might be quite ... basic. And then
hopefully if we can figure out a solution for that one, it will also
fix the intermittent failures in the stacking suite on ppc64el. If not,
we can look at any remaining failures separately.

    smcv