[Bug 99887] Random segfault in intel driver in __kgem_busy() with Xorg 1.19.x

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=99887

--- Comment #4 from Chris Wilson <chris at chris-wilson.co.uk> ---
Hmm, the valgrind log should be less noisy if the ddx is compiled with
--enable-debug. It should mention successfully detecting vg in the compile and
when using the new intel_drv.so under valgrind.

As I understand the test, it creates a new client, changes the pointer, waits
for the cursor to move, then exits and repeats. To clarify, the
XDefinteCursor() step is critical for reproducing? The forking should be
irrelevant, the client id should be recycled so that too should not be an
issue. Suggests that perhaps the test is just to rapidly change the cursor
whilst processing pointer events.

Or it is nothing to do with the pointer, and is the window map/unmapping.
Neither are particularly rare paths :| Didn't spot anything in the vg log to
suggest otherwise, my first suspicion is the noise is masking a real
out-of-bounds write.

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/intel-gfx-bugs/attachments/20170221/52b33662/attachment.html>