i965: Require pixel alignment for GPU copy blit

The blitter will start at a pixel's natural alignment. For PBOs, if the
provided offset if not aligned, bits will get dropped.

This change adds offset alignment check for src and dst, kicking back if
the requirements are not met.

The change is based on following verbiage from BSPEC:
 Color pixel sizes supported are 8, 16, and 32 bits per pixel (bpp).
 All pixels are naturally aligned.

Found in the following locations:
page 35 of intel-gfx-prm-osrc-hsw-blitter.pdf
page 29 of ivb_ihd_os_vol1_part4.pdf
page 29 of snb_ihd_os_vol1_part5.pdf

This behavior was observed with Steam Big Picture rendering incorrect
icon colors.  The fix has been tested on Ubuntu and SteamOS on Haswell.

Signed-off-by: Cody Northrop <cody@lunarg.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=83908
Reviewed-by: Neil Roberts <neil@linux.intel.com>
This commit is contained in:
Cody Northrop 2014-09-15 16:14:20 -06:00 committed by Ian Romanick
parent fc016bc0f3
commit 83e8bb5b1a
2 changed files with 6 additions and 4 deletions

View File

@ -271,9 +271,10 @@ intelEmitCopyBlit(struct intel_context *intel,
dst_buffer, dst_pitch, dst_offset, dst_x, dst_y, w, h);
/* Blit pitch must be dword-aligned. Otherwise, the hardware appears to drop
* the low bits.
* the low bits. Offsets must be naturally aligned.
*/
if (src_pitch % 4 != 0 || dst_pitch % 4 != 0)
if (src_pitch % 4 != 0 || src_offset % cpp != 0 ||
dst_pitch % 4 != 0 || dst_offset % cpp != 0)
return false;
/* For big formats (such as floating point), do the copy using 16 or 32bpp

View File

@ -344,9 +344,10 @@ intelEmitCopyBlit(struct brw_context *brw,
dst_buffer, dst_pitch, dst_offset, dst_x, dst_y, w, h);
/* Blit pitch must be dword-aligned. Otherwise, the hardware appears to drop
* the low bits.
* the low bits. Offsets must be naturally aligned.
*/
if (src_pitch % 4 != 0 || dst_pitch % 4 != 0)
if (src_pitch % 4 != 0 || src_offset % cpp != 0 ||
dst_pitch % 4 != 0 || dst_offset % cpp != 0)
return false;
/* For big formats (such as floating point), do the copy using 16 or 32bpp