For any bugs showing mysterious Intel-graphics crashes with libdrm 2.4.17,
please encourage the bug reporters to re-try with libdrm 2.4.18 (just
released today). It includes the commit below.
I'll mark this as NEEDINFO for now. Please update the bug (and clear
NEEDINFO) after adding results of testing with libdrm 2.4.18.
-Carl
commit 4f0f871730b76730ca58209181d16725b0c40184
Author: Chris Wilson <email address hidden>
Date: Wed Feb 10 09:45:13 2010 +0000
intel: Handle resetting of input params after EINTR during SET_TILING
The SET_TILING is pernicious in that it overwrites the input arguments
following an error in order to report the current tiling state of the
buffer. This caught us by surprise as we then fed those arguments back
into to the ioctl unmodified following an EINTR and so the kernel then
reported success for the no-op. We interpreted this success as meaning
that the tiling on the buffer had changed so updated our state and
started using the buffer incorrectly in the new tiled/untiled manner.
This lead to all sorts of random corruption and GPU hangs, even though
the batch buffers would look sane (when the GPU had not wandered off
into forbidden territory).
Hi Bryce,
For any bugs showing mysterious Intel-graphics crashes with libdrm 2.4.17,
please encourage the bug reporters to re-try with libdrm 2.4.18 (just
released today). It includes the commit below.
I'll mark this as NEEDINFO for now. Please update the bug (and clear
NEEDINFO) after adding results of testing with libdrm 2.4.18.
-Carl
commit 4f0f871730b7673 0ca58209181d167 25b0c40184
Author: Chris Wilson <email address hidden>
Date: Wed Feb 10 09:45:13 2010 +0000
intel: Handle resetting of input params after EINTR during SET_TILING
The SET_TILING is pernicious in that it overwrites the input arguments
following an error in order to report the current tiling state of the
buffer. This caught us by surprise as we then fed those arguments back
into to the ioctl unmodified following an EINTR and so the kernel then
reported success for the no-op. We interpreted this success as meaning
that the tiling on the buffer had changed so updated our state and
started using the buffer incorrectly in the new tiled/untiled manner.
This lead to all sorts of random corruption and GPU hangs, even though
the batch buffers would look sane (when the GPU had not wandered off
into forbidden territory).
References:
Bug 25475 - [i915] Xorg crash / Execbuf while wedged bugs.freedeskto p.org/show_ bug.cgi? id=25475
http://
Bug 25554 - i830_uxa_ prepare_ access: gtt bo map failed: Input/output error bugs.freedeskto p.org/show_ bug.cgi? id=25554
http://
(And probably every other weird bug in the last few months.)
Signed-off-by: Chris Wilson <email address hidden>