| Summary: | NULL deref in busy_cursor_grab_focus causes SEGV | ||
|---|---|---|---|
| Product: | Wayland | Reporter: | U. Artie Eoff <ullysses.a.eoff> |
| Component: | weston | Assignee: | Wayland bug list <wayland-bugs> |
| Status: | VERIFIED FIXED | QA Contact: | |
| Severity: | critical | ||
| Priority: | high | ||
| Version: | unspecified | ||
| Hardware: | Other | ||
| OS: | All | ||
| Whiteboard: | |||
| i915 platform: | i915 features: | ||
| Attachments: | weston gdb backtrace | ||
Hi Artie, I tried running that test but it didn't trigger the problem for me. Re you using Clutter 1.16? Possible solution patch posted to list. Rob, ah yes I forgot to list my s/w: wayland (master) heads/master-0-g508dd69 fontconfig (master) heads/master-0-g0f9aa87 drm (master) heads/master-0-gacfbf39 mesa (master) heads/master-0-g4f518e1 libxkbcommon (master) heads/master-0-ga392d26 pixman (master) heads/master-0-gcb5d131 cairo (master) heads/master-0-g631bf29 weston (master) heads/master-0-geb1e130 harfbuzz (master) heads/master-0-gfa3d0a0 glib (master) heads/master-0-gad7959c atk (master) heads/master-0-gc71edfa cogl (cogl-1.16) heads/cogl-1.16-0-gc08fe74 clutter (clutter-1.16) heads/clutter-1.16-0-g323ec19 Try running it on x11-backend if you haven't done so... it's harder to trigger on the drm-backend. Our CI agrees with that. Also, try running the entire clutter conform suite all at once since it may be intermittently triggered by running back-to-back tests. I'll grab the patch you mentioned and see if that helps. Rob, the efl/egl/ListUserMouseScrollTest in wayland-fits triggers this issue 100%. https://github.com/01org/wayland-fits Run: CK_FORK=no wfits --filter efl/egl/ListUserMouseScrollTest Note: you'll need upstream EFL Wayland |
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.
Created attachment 79444 [details] weston gdb backtrace Weston crashes due to a NULL deref in busy_cursor_grab_focus() on "grab->shsurf" (see attached gdb backtrace). This is triggered with various clutter and efl client test cases that I've tried. If you run the following clutter conform test, you can trigger the crash about 95% of the time: ./test-conformance -k --verbose -p /conform/actor/actor_pick When weston crashes, it prints the following message to the console: [15:12:54.169] caught signal: 11 [15:12:54.170] [000000000040e176] -- (weston) [15:12:54.170] [000000000040e25e] -- (weston) [15:12:54.170] [00007f70b2a03fe0] -- (/lib64/libpthread.so.0) [15:12:54.170] [00007f70ad63b00c] -- (/home/uartie/Work/intel/wayland/install/lib/weston/desktop-shell.so) [15:12:54.170] [000000000040f67d] -- (weston) [15:12:54.170] [000000000040a86a] -- (weston) [15:12:54.170] [000000000040b834] -- (weston) [15:12:54.170] [000000000040ba1d] weston_output_finish_frame (weston) [15:12:54.170] [00007f70b0fdd438] -- (/home/uartie/Work/intel/wayland/install/lib/weston/x11-backend.so) [15:12:54.170] [00007f70b0fdd736] -- (/home/uartie/Work/intel/wayland/install/lib/weston/x11-backend.so) [15:12:54.170] [00007f70b349d29b] -- (/home/uartie/Work/intel/wayland/install/lib/libwayland-server.so.0) [15:12:54.170] [00007f70b349dba4] wl_event_loop_dispatch (/home/uartie/Work/intel/wayland/install/lib/libwayland-server.so.0) [15:12:54.170] [00007f70b349b673] wl_display_run (/home/uartie/Work/intel/wayland/install/lib/libwayland-server.so.0) [15:12:54.170] [000000000040ee88] -- (weston) [15:12:54.170] [0000003763221735] __libc_start_main (/lib64/libc.so.6) [15:12:54.170] [0000000000407659] -- (weston) It appears that the client-side crashes before the server-side. First bad Weston commit is be6403ed5c4fdab884d391778e2572aae109f1a0