drm/i915: Remove wait priority boosting
authorChris Wilson <chris@chris-wilson.co.uk>
Thu, 7 May 2020 15:23:38 +0000 (16:23 +0100)
committerChris Wilson <chris@chris-wilson.co.uk>
Thu, 7 May 2020 19:08:58 +0000 (20:08 +0100)
commiteec39e441c29dd636bbdf5390af765cd9db6b380
treebaa4f09e885695ae81109880b4e7e02025eb62f7
parent6b6cd2ebd8d071e55998e32b648bb8081f7f02bb
drm/i915: Remove wait priority boosting

Upon waiting a request (when asked), we gave that request a small
priority boost, not enough for it to cause preemption, but enough for it
to be scheduled next before all equals. We also used that bit to give
new clients a small priority boost, similar to FQ_CODEL, such that we
favoured short interactive tasks ahead of long running streams.

However, this is causing lots of complications with timeslicing where we
both want to honour the boost and yet ignore it. Those complications
cause unexpected user behaviour (tasks not being timesliced and run
concurrently as epxected), and the easiest way to resolve that is to
remove the boost. Hopefully, we can find a compromise again if we need
to, but in theory timeslicing itself and future more advanced schedulers
should give us the interactivity boost we seek.

Testcase: igt/gem_exec_schedule/lateslice
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200507152338.7452-3-chris@chris-wilson.co.uk
drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c
drivers/gpu/drm/i915/gt/intel_lrc.c
drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c
drivers/gpu/drm/i915/i915_priolist_types.h
drivers/gpu/drm/i915/i915_request.c
drivers/gpu/drm/i915/i915_scheduler.c