Hey, I was just peeking through src/mesa/x86/3dnow.c and I found this
message:
/* There's a bug somewhere in the 3dnow_normal.S file that causes
* bad shading. Disable for now.
ASSIGN_NORM_GROUP( 3dnow );
*/
So I decided to check out src/mesa/x86/3dnow_normal.S in Mesa-6.5.3. I
found a some small logic bugs.
Basically, the author assumed that DEC sets the EFLAGS register and does a
conditional branch; however this isn't true: the ADD/SUB instructions do,
but INC/DEC do not. I believe that this might be causing the bug that is
mentioned in src/mesa/x86/3dnow.c. I'll check the other 3dnow code files to
see if the same assumption is being made later. For now, this patch should
fix the basic logic behind the loops -- but it may not fix the bug mentioned
in src/mesa/x86/3dnow.c. I guess someone will have to test that exact
configuration to see if in fact that resolves the bug(s) mentioned
Patrick Baggett
diff --git a/Makefile.DJ b/Makefile.DJ
diff --git a/Makefile.mgw b/Makefile.mgw
diff --git a/descrip.mms b/descrip.mms
diff --git a/mms-config. b/mms-config.
diff --git a/src/mesa/x86/3dnow_normal.S b/src/mesa/x86/3dnow_normal.S
index 4345c1b..f3bbcb2 100644
--- a/src/mesa/x86/3dnow_normal.S
+++ b/src/mesa/x86/3dnow_normal.S
@@ -129,8 +129,8 @@ LLBL (G3TN_transform):
PREFETCH ( REGIND(EDX) )
MOVD ( MM1, REGOFF(-8, EAX) ) /* write r2 */
- DEC_L ( EBP ) /* decrement normal counter */
- JA ( LLBL (G3TN_transform) )
+ SUB_L ( CONST(1), EBP ) /* decrement normal counter */
+ JNZ ( LLBL (G3TN_transform) )
POP_L ( EDX ) /* end of transform --- */
@@ -164,9 +164,9 @@ LLBL (G3TN_norm_w_lengths):
MOVD ( MM1, REGOFF(8, EAX) ) /* write new x2 */
ADD_L ( CONST(16), EAX ) /* next r */
- DEC_L ( EBP ) /* decrement normal counter */
+ SUB_L ( CONST(1), EBP ) /* decrement normal counter */
- JA ( LLBL (G3TN_norm_w_lengths) )
+ JNZ ( LLBL (G3TN_norm_w_lengths) )
JMP ( LLBL (G3TN_exit_3dnow) )
ALIGNTEXT32
@@ -192,7 +192,7 @@ LLBL (G3TN_norm):
MOVQ ( MM5, MM4 )
PUNPCKLDQ ( MM3, MM3 )
- DEC_L ( EBP ) /* decrement normal counter */
+ SUB_L ( CONST(1), EBP ) /* decrement normal counter */
PFMUL ( MM5, MM5 )
PFRSQIT1 ( MM3, MM5 )
@@ -204,7 +204,7 @@ LLBL (G3TN_norm):
PFMUL ( MM5, MM1 ) /* | x2 (normalize*/
MOVD ( MM1, REGOFF(-8, EAX) ) /* write new x2 */
- JA ( LLBL (G3TN_norm) )
+ JNZ ( LLBL (G3TN_norm) )
LLBL (G3TN_exit_3dnow):
FEMMS
@@ -289,13 +289,13 @@ LLBL (G3TNNR_norm_w_lengths): /* use precalculated lengths */
ADD_L ( CONST(4), EDI ) /* next length */
PFMUL ( MM3, MM6 ) /* x1 (normalized) | x0 (normalized) */
- DEC_L ( EBP ) /* decrement normal counter */
+ SUB_L ( CONST(1), EBP ) /* decrement normal counter */
MOVQ ( MM6, REGOFF(-16, EAX) ) /* write r0, r1 */
MOVD ( MM7, REGOFF(-8, EAX) ) /* write r2 */
MOVD ( REGIND(EDI), MM3 ) /* | length (x) */
- JA ( LLBL (G3TNNR_norm_w_lengths) )
+ JNZ ( LLBL (G3TNNR_norm_w_lengths) )
JMP ( LLBL (G3TNNR_exit_3dnow) )
ALIGNTEXT32
@@ -331,7 +331,7 @@ LLBL (G3TNNR_norm): /* need to calculate lengths */
PFMUL ( MM5, MM5 )
PFRSQIT1 ( MM3, MM5 )
- DEC_L ( EBP ) /* decrement normal counter */
+ SUB_L ( CONST(1), EBP ) /* decrement normal counter */
PFRCPIT2 ( MM4, MM5 )
PFMUL ( MM5, MM6 ) /* x1 (normalized) | x0 (normalized) */
@@ -340,7 +340,7 @@ LLBL (G3TNNR_norm): /* need to calculate lengths */
PFMUL ( MM5, MM7 ) /* | x2 (normalized) */
MOVD ( MM7, REGOFF(-8, EAX) ) /* write r2 */
- JA ( LLBL (G3TNNR_norm) )
+ JNZ ( LLBL (G3TNNR_norm) )
LLBL (G3TNNR_exit_3dnow):
@@ -411,11 +411,11 @@ LLBL (G3TRNR_rescale):
PFMUL ( MM2, MM5 ) /* | x2*m10 */
ADD_L ( CONST(16), EAX ) /* next r */
- DEC_L ( EBP ) /* decrement normal counter */
+ SUB_L ( CONST(1), EBP ) /* decrement normal counter */
MOVQ ( MM4, REGOFF(-16, EAX) ) /* write r0, r1 */
MOVD ( MM5, REGOFF(-8, EAX) ) /* write r2 */
- JA ( LLBL (G3TRNR_rescale) ) /* cnt > 0 ? -> process next normal */
+ JNZ ( LLBL (G3TRNR_rescale) ) /* cnt > 0 ? -> process next normal */
FEMMS
@@ -511,8 +511,8 @@ LLBL (G3TR_rescale):
PFADD ( MM2, MM1 ) /* *not used* | x0*m8+x1*m9+x2*m10 */
MOVD ( MM1, REGOFF(-8, EAX) ) /* write r2 */
- DEC_L ( EDI ) /* decrement normal counter */
- JA ( LLBL (G3TR_rescale) )
+ SUB_L ( CONST(1), EDI ) /* decrement normal counter */
+ JNZ ( LLBL (G3TR_rescale) )
FEMMS
@@ -574,11 +574,11 @@ LLBL (G3TNR_transform):
PFMUL ( MM2, MM5 ) /* | x2*m10 */
ADD_L ( CONST(16), EAX ) /* next r */
- DEC_L ( EDI ) /* decrement normal counter */
+ SUB_L ( CONST(1), EDI ) /* decrement normal counter */
MOVQ ( MM4, REGOFF(-16, EAX) ) /* write r0, r1 */
MOVD ( MM5, REGOFF(-8, EAX) ) /* write r2 */
- JA ( LLBL (G3TNR_transform) )
+ JNZ ( LLBL (G3TNR_transform) )
FEMMS
@@ -663,9 +663,9 @@ LLBL (G3T_transform):
PFADD ( MM2, MM1 ) /* *not used* | x0*m8+x1*m9+x2*m10 */
MOVD ( MM1, REGOFF(-8, EAX) ) /* write r2 */
- DEC_L ( EDI ) /* decrement normal counter */
+ SUB_L ( CONST(1), EDI ) /* decrement normal counter */
- JA ( LLBL (G3T_transform) )
+ JNZ ( LLBL (G3T_transform) )
FEMMS
@@ -730,9 +730,9 @@ LLBL (G3N_norm1): /* use precalculated lengths */
ADD_L ( CONST(16), EAX ) /* next r */
ADD_L ( CONST(4), EDX ) /* next length */
- DEC_L ( EBP ) /* decrement normal counter */
+ SUB_L ( CONST(1), EBP ) /* decrement normal counter */
- JA ( LLBL (G3N_norm1) )
+ JNZ ( LLBL (G3N_norm1) )
JMP ( LLBL (G3N_end1) )
@@ -765,7 +765,7 @@ LLBL (G3N_norm2): /* need to calculate lengths */
PFMUL ( MM5, MM5 )
PFRSQIT1 ( MM3, MM5 )
- DEC_L ( EBP ) /* decrement normal counter */
+ SUB_L ( CONST(1), EBP ) /* decrement normal counter */
PFRCPIT2 ( MM4, MM5 )
@@ -775,7 +775,7 @@ LLBL (G3N_norm2): /* need to calculate lengths */
PFMUL ( MM5, MM1 ) /* | x2 (normalized) */
MOVD ( MM1, REGOFF(-8, EAX) ) /* write new x2 */
- JA ( LLBL (G3N_norm2) )
+ JNZ ( LLBL (G3N_norm2) )
LLBL (G3N_end1):
FEMMS
@@ -835,8 +835,8 @@ LLBL (G3R_rescale):
MOVQ ( MM1, REGOFF(-16, EAX) ) /* write r0, r1 */
MOVD ( MM2, REGOFF(-8, EAX) ) /* write r2 */
- DEC_L ( EDX ) /* decrement normal counter */
- JA ( LLBL (G3R_rescale) )
+ SUB_L ( CONST(1), EDX ) /* decrement normal counter */
+ JNZ ( LLBL (G3R_rescale) )
FEMMS
-------------------------------------------------------------------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourceforge.net/powerbar/db2/
_______________________________________________
Mesa3d-dev mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/mesa3d-dev
