GNOME Bugzilla – Bug 680214
Spurious window resize of new windows to another window's size.
Last modified: 2021-07-05 14:04:03 UTC
Created attachment 219174 [details] [review] 0001-util-Warn-and-report-SKIP-when-we-get-a-window-resiz.patch Environment: debian unstable. ii gnome-shell 3.4.1-8 amd64 ii freeglut3:amd64 2.6.0-4 The GL regression testing suite, piglit, is unfortunately somewhat reliant on getting its windows at the size it requested, or some tests will report failure. This is not a problem on most wms (such as metacity), but gnome-shell sometimes produces a resize event from freeglut, and the size happens to be that of another piglit window that was created. Steps to reproduce: git clone git://anongit.freedesktop.org:/git/piglit cd piglit # flail around to get dependencies # I'm so sorry about the cmake here. cmake . git am 0001-util-Warn-and-report-SKIP-when-we-get-a-window-resiz.patch make while ./bin/scissor-stencil-clear -auto | grep "pass"; do; done # the 100x100 test above will happily run until you get bored. now, open a new tab and start this: while ./bin/copy-pixels -auto | grep "pass"; do; done # this one is 32x32. eventually, one of the two fails, and the size was the other one's size. I've seen failure when I'm not interacting with the system, but failure rate may be correlated with moving the mouse or alt-tabbing. Note that while I'm running these loops, tail -f ~/.xsession-errors is sometimes reporting: Window manager warning: Log level 16: STACK_OP_ADD: window 0x2e00002 already in stack
Full disclosure: I'm running the alternatetab, frippery panel favorites, remove accessibility, and remove user name extensions. However, given that QA has been reporting these spurious failures under gnome-shell, I don't expect them to be part of the problem.
Do these issues happen under mutter or metacity, which have the same constraints code? Are you marking your window as override-redirect to prevent resizes? Otherwise, it's fully within the ICCCM to allow WMs to deny client resize requests.
Regardless of whether it's ICCCM-allowed, do you think it's reasonable to assign a brand new window the size of somebody else's brand new window, just because somebody else happened to be making a window? This problem hasn't been seen under metacity.
Comment on attachment 219174 [details] [review] 0001-util-Warn-and-report-SKIP-when-we-get-a-window-resiz.patch tests/util/piglit-framework-glut.c does not exist anymore in git master of gnome-shell. Hence setting 'needs-rework' as patch does not apply cleanly against git master.
GNOME is going to shut down bugzilla.gnome.org in favor of gitlab.gnome.org. As part of that, we are mass-closing older open tickets in bugzilla.gnome.org which have not seen updates for a longer time (resources are unfortunately quite limited so not every ticket can get handled). If you can still reproduce the situation described in this ticket in a recent and supported software version, then please follow https://wiki.gnome.org/GettingInTouch/BugReportingGuidelines and create a new ticket at https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/ Thank you for your understanding and your help.