diff options
| author | Chris Wilson <chris@chris-wilson.co.uk> | 2016-01-22 20:11:31 +0000 |
|---|---|---|
| committer | Chris Wilson <chris@chris-wilson.co.uk> | 2016-01-22 20:15:41 +0000 |
| commit | b48d4a7917ab793526be47559becc64aacd347ae (patch) | |
| tree | 84e635d713ebd174eedc695fadf3467493867ece /src/sna/sna_dri2.c | |
| parent | 26cc364d2eb078085ff4a6dd2968b426bf333b4f (diff) | |
sna/dri2: Tweak enginement placement on inherited render targets
The main motivation is try and isolate from an upcoming API change in
busy-ioctl, but is also based on the observation that we now only care
about write-synchronisation.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Diffstat (limited to 'src/sna/sna_dri2.c')
| -rw-r--r-- | src/sna/sna_dri2.c | 11 |
1 files changed, 5 insertions, 6 deletions
diff --git a/src/sna/sna_dri2.c b/src/sna/sna_dri2.c index 0e184567..54504346 100644 --- a/src/sna/sna_dri2.c +++ b/src/sna/sna_dri2.c @@ -1110,13 +1110,12 @@ static void sna_dri2_select_mode(struct sna *sna, struct kgem_bo *dst, struct kg * The ultimate question is whether preserving the ring outweighs * the cost of the query. */ - if (popcount(busy.busy >> 16) > 1) - mode = busy.busy & 0xffff ? KGEM_BLT : KGEM_RENDER; - else if (busy.busy & (0xfffe << 16)) + mode = KGEM_RENDER; + if ((busy.busy & 0xffff) == KGEM_BLT) mode = KGEM_BLT; - else - mode = KGEM_RENDER; - kgem_bo_mark_busy(&sna->kgem, busy.handle == src->handle ? src : dst, mode); + kgem_bo_mark_busy(&sna->kgem, + busy.handle == src->handle ? src : dst, + mode); _kgem_set_mode(&sna->kgem, mode); } |