Re: [Intel-gfx] [PATCH] drm/i915: Use seqlock in engine stats

2018-02-15 Thread Chris Wilson
Quoting Tvrtko Ursulin (2018-02-15 11:59:06) > > On 15/02/2018 11:47, Chris Wilson wrote: > > Quoting Tvrtko Ursulin (2018-02-15 11:13:33) > >> From: Tvrtko Ursulin > >> > >> We can convert engine stats from a spinlock to seqlock to ensure interrupt > >> processing is

Re: [Intel-gfx] [PATCH] drm/i915: Use seqlock in engine stats

2018-02-15 Thread Tvrtko Ursulin
On 15/02/2018 11:47, Chris Wilson wrote: Quoting Tvrtko Ursulin (2018-02-15 11:13:33) From: Tvrtko Ursulin We can convert engine stats from a spinlock to seqlock to ensure interrupt processing is never even a tiny bit delayed by parallel readers. There is a smidgen

Re: [Intel-gfx] [PATCH] drm/i915: Use seqlock in engine stats

2018-02-15 Thread Chris Wilson
Quoting Tvrtko Ursulin (2018-02-15 11:13:33) > From: Tvrtko Ursulin > > We can convert engine stats from a spinlock to seqlock to ensure interrupt > processing is never even a tiny bit delayed by parallel readers. > > There is a smidgen bit more cost on the write lock

[Intel-gfx] [PATCH] drm/i915: Use seqlock in engine stats

2018-02-15 Thread Tvrtko Ursulin
From: Tvrtko Ursulin We can convert engine stats from a spinlock to seqlock to ensure interrupt processing is never even a tiny bit delayed by parallel readers. There is a smidgen bit more cost on the write lock side, and an extremely unlikely chance that readers will