drm/i915/execlists: Reset RING registers upon resume
authorChris Wilson <chris@chris-wilson.co.uk>
Wed, 21 Sep 2016 13:51:08 +0000 (14:51 +0100)
committerChris Wilson <chris@chris-wilson.co.uk>
Wed, 21 Sep 2016 15:57:48 +0000 (16:57 +0100)
commitbafb2f7d4755bf1571bd5e9a03b97f3fc4fe69ae
tree3ca26e9ec91a1ce03ce516a0e9b565e6e2b7b29d
parent6a800eabba34945c2986d70114b41d564bad52a8
drm/i915/execlists: Reset RING registers upon resume

There is a disparity in the context image saved to disk and our own
bookkeeping - that is we presume the RING_HEAD and RING_TAIL match our
stored ce->ring->tail value. However, as we emit WA_TAIL_DWORDS into the
ring but may not tell the GPU about them, the GPU may be lagging behind
our bookkeeping. Upon hibernation we do not save stolen pages, presuming
that their contents are volatile. This means that although we start
writing into the ring at tail, the GPU starts executing from its HEAD
and there may be some garbage in between and so the GPU promptly hangs
upon resume.

Testcase: igt/gem_exec_suspend/basic-S4
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=96526
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20160921135108.29574-3-chris@chris-wilson.co.uk
drivers/gpu/drm/i915/intel_lrc.c