[Intel-gfx] [PATCH] i965: fixes rendering distortion on 965GM/965G

[Eric Anholt]

On Sun, 2009-02-15 at 23:47 +0800, Xiang, Haihao wrote:
> 
> >-----Original Message-----
> >From: Xiang, Haihao
> >Sent: 2009年2月12日 10:02
> >To: intel-gfx at lists.freedesktop.org
> >Cc: Xiang, Haihao
> >Subject: [PATCH] i965: fixes rendering distortion on 965GM/965G
> >
> >clip_test optimization introduces rendering glitches in
> >Google Earth (see bug #19879), another simple case is xscreensaver/atunnel.
> >This fix can eliminate the rendering glitches for these cases, but I am not
> >sure it introduces rendering glitches for other cases or not.
> >
> >Maybe we should remove this clip_test optimization for RHW workaround.
> >---
> > src/mesa/drivers/dri/i965/brw_clip_tri.c |    6 +++---
> > 1 files changed, 3 insertions(+), 3 deletions(-)
> >
> >diff --git a/src/mesa/drivers/dri/i965/brw_clip_tri.c
> >b/src/mesa/drivers/dri/i965/brw_clip_tri.c
> >index 1dbba37..6cbe6e6 100644
> >--- a/src/mesa/drivers/dri/i965/brw_clip_tri.c
> >+++ b/src/mesa/drivers/dri/i965/brw_clip_tri.c
> >@@ -464,11 +464,11 @@ static void brw_clip_test( struct brw_clip_compile *c )
> >     brw_MOV(p, v2, deref_4f(vt2, c->offset[VERT_RESULT_HPOS]));
> >
> >     /* test nearz, xmin, ymin plane */
> >-    brw_CMP(p, t1, BRW_CONDITIONAL_LE, negate(v0), get_element(v0, 3));
> >+    brw_CMP(p, t1, BRW_CONDITIONAL_L, negate(v0), get_element(v0, 3));
> >     brw_set_predicate_control(p, BRW_PREDICATE_NONE);
> >-    brw_CMP(p, t2, BRW_CONDITIONAL_LE, negate(v1), get_element(v1, 3));
> >+    brw_CMP(p, t2, BRW_CONDITIONAL_L, negate(v1), get_element(v1, 3));
> >     brw_set_predicate_control(p, BRW_PREDICATE_NONE);
> >-    brw_CMP(p, t3, BRW_CONDITIONAL_LE, negate(v2), get_element(v2, 3));
> >+    brw_CMP(p, t3, BRW_CONDITIONAL_L, negate(v2), get_element(v2, 3));
> >     brw_set_predicate_control(p, BRW_PREDICATE_NONE);
> >     brw_XOR(p, t, t1, t2);
> >     brw_XOR(p, t1, t2, t3);
> >--
> >1.5.6.3
> 
> Hi, Eric
> 	Could you take a look at this patch?

I read the diff, but it didn't make any sense to me.  This would be
better if the commit explained not just that it fixed some testcases,
but why it fixed some testcases.

-- 
Eric Anholt
eric at anholt.net                         eric.anholt at intel.com


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 197 bytes
Desc: This is a digitally signed message part
URL: <http://lists.freedesktop.org/archives/intel-gfx/attachments/20090216/b50ea98d/attachment.sig>