Re: [Intel-gfx] [PATCH] drm/i915: prevent gt fifo count underflow

2014-05-14 Thread Chris Wilson
On Wed, May 14, 2014 at 04:35:42PM +0300, Ville Syrjälä wrote: > On Wed, May 14, 2014 at 04:18:02PM +0300, Mika Kuoppala wrote: > > If we get the final value of zero as a count of free > > entries available, we will underflow our own fifo_count > > and then it will take a long time before we check

Re: [Intel-gfx] [PATCH] drm/i915: prevent gt fifo count underflow

2014-05-14 Thread Ville Syrjälä
On Wed, May 14, 2014 at 04:18:02PM +0300, Mika Kuoppala wrote: > If we get the final value of zero as a count of free > entries available, we will underflow our own fifo_count > and then it will take a long time before we check things again. > Admittedly we are in trouble already if we get into thi

[Intel-gfx] [PATCH] drm/i915: prevent gt fifo count underflow

2014-05-14 Thread Mika Kuoppala
If we get the final value of zero as a count of free entries available, we will underflow our own fifo_count and then it will take a long time before we check things again. Admittedly we are in trouble already if we get into this situation, but prevent the underflow by returning early. v2: Less co

[Intel-gfx] [PATCH] drm/i915: prevent gt fifo count underflow

2014-05-14 Thread Mika Kuoppala
If we get the final value of zero as a count of free entries available, we will underflow our own fifo_count and then it will take a long time before we check things again. Admittedly we are in trouble already if we get into this situation, but prevent the underflow by checking against zero before