Re: [Intel-gfx] [PATCH] igt: Test tiled bo for proper fence.

2014-12-19 Thread Daniel Vetter
On Thu, Dec 18, 2014 at 02:43:45PM -0800, Bob Paauwe wrote: > On Thu, 18 Dec 2014 21:31:43 +0100 > Daniel Vetter wrote: > > > On Thu, Dec 18, 2014 at 09:50:27AM -0800, Bob Paauwe wrote: > > > + > > > + /* Set the tiled buffer to all 0xff's */ > > > + memset(bo_tiled->virtual, 0xff, bo_tiled->size

Re: [Intel-gfx] [PATCH] igt: Test tiled bo for proper fence.

2014-12-18 Thread Bob Paauwe
On Thu, 18 Dec 2014 21:31:43 +0100 Daniel Vetter wrote: > On Thu, Dec 18, 2014 at 09:50:27AM -0800, Bob Paauwe wrote: > > When a tiled bo is allocated such that the actual size of the bo is > > larger than the region covered by the tiles. For example allocating > > a Y-tiled bo that is 1920 x 10

Re: [Intel-gfx] [PATCH] igt: Test tiled bo for proper fence.

2014-12-18 Thread Daniel Vetter
On Thu, Dec 18, 2014 at 09:50:27AM -0800, Bob Paauwe wrote: > When a tiled bo is allocated such that the actual size of the bo is > larger than the region covered by the tiles. For example allocating > a Y-tiled bo that is 1920 x 1088 (15 x 34 tiles) results in a bo > that is 0x20 bytes instea

[Intel-gfx] [PATCH] igt: Test tiled bo for proper fence.

2014-12-18 Thread Bob Paauwe
When a tiled bo is allocated such that the actual size of the bo is larger than the region covered by the tiles. For example allocating a Y-tiled bo that is 1920 x 1088 (15 x 34 tiles) results in a bo that is 0x20 bytes instead of the 0x1FE000 bytes covered by the tiles. The fence needs to cov