Added predictor stride argument(s) to subtract functions

Patch set 2: 64 bit build fix
Patch set 3: 64 bit crash fix

[Tero]
Patch set 4: Updated ARMv6 and NEON assembly.
             Added also minor NEON optimizations to subtract
             functions.

Patch set 5: x86 stride bug fix

Change-Id: I1fcca93e90c89b89ddc204e1c18f208682675c15

1 files changed, 35 insertions, 28 deletions

diff --git a/vp8/encoder/arm/armv6/vp8_subtract_armv6.asm b/vp8/encoder/arm/armv6/vp8_subtract_armv6.asm
index 0ca74387b..f329f8f73 100644
--- a/vp8/encoder/arm/armv6/vp8_subtract_armv6.asm
+++ b/vp8/encoder/arm/armv6/vp8_subtract_armv6.asm
@@ -72,22 +72,23 @@ loop_block
 ; r0    short *diff
 ; r1    unsigned char *usrc
 ; r2    unsigned char *vsrc
-; r3    unsigned char *pred
-; stack int stride
+; r3    int src_stride
+; sp    unsigned char *upred
+; sp    unsigned char *vpred
+; sp    int pred_stride
 |vp8_subtract_mbuv_armv6| PROC
 
-    stmfd   sp!, {r4-r12, lr}
+    stmfd   sp!, {r4-r11}
 
     add     r0, r0, #512        ; set *diff point to Cb
-    add     r3, r3, #256        ; set *pred point to Cb
-
     mov     r4, #8              ; loop count
-    ldr     r5, [sp, #40]       ; stride
+    ldr     r5, [sp, #32]       ; upred
+    ldr     r12, [sp, #40]      ; pred_stride
 
     ; Subtract U block
 loop_u
-    ldr     r6, [r1]            ; src       (A)
-    ldr     r7, [r3], #4        ; pred      (A)
+    ldr     r6, [r1]            ; usrc      (A)
+    ldr     r7, [r5]            ; upred     (A)
 
     uxtb16  r8, r6              ; [s2 | s0] (A)
     uxtb16  r9, r7              ; [p2 | p0] (A)
@@ -97,8 +98,8 @@ loop_u
     usub16  r6, r8, r9          ; [d2 | d0] (A)
     usub16  r7, r10, r11        ; [d3 | d1] (A)
 
-    ldr     r10, [r1, #4]       ; src       (B)
-    ldr     r11, [r3], #4       ; pred      (B)
+    ldr     r10, [r1, #4]       ; usrc      (B)
+    ldr     r11, [r5, #4]       ; upred     (B)
 
     pkhbt   r8, r6, r7, lsl #16 ; [d1 | d0] (A)
     pkhtb   r9, r7, r6, asr #16 ; [d3 | d2] (A)
@@ -114,7 +115,8 @@ loop_u
     usub16  r6, r8, r9          ; [d2 | d0] (B)
     usub16  r7, r10, r11        ; [d3 | d1] (B)
 
-    add     r1, r1, r5          ; update usrc pointer
+    add     r1, r1, r3          ; update usrc pointer
+    add     r5, r5, r12         ; update upred pointer
 
     pkhbt   r8, r6, r7, lsl #16 ; [d1 | d0] (B)
     pkhtb   r9, r7, r6, asr #16 ; [d3 | d2] (B)
@@ -125,12 +127,13 @@ loop_u
 
     bne     loop_u
 
+    ldr     r5, [sp, #36]       ; vpred
     mov     r4, #8              ; loop count
 
     ; Subtract V block
 loop_v
-    ldr     r6, [r2]            ; src       (A)
-    ldr     r7, [r3], #4        ; pred      (A)
+    ldr     r6, [r2]            ; vsrc      (A)
+    ldr     r7, [r5]            ; vpred     (A)
 
     uxtb16  r8, r6              ; [s2 | s0] (A)
     uxtb16  r9, r7              ; [p2 | p0] (A)
@@ -140,8 +143,8 @@ loop_v
     usub16  r6, r8, r9          ; [d2 | d0] (A)
     usub16  r7, r10, r11        ; [d3 | d1] (A)
 
-    ldr     r10, [r2, #4]       ; src       (B)
-    ldr     r11, [r3], #4       ; pred      (B)
+    ldr     r10, [r2, #4]       ; vsrc      (B)
+    ldr     r11, [r5, #4]       ; vpred     (B)
 
     pkhbt   r8, r6, r7, lsl #16 ; [d1 | d0] (A)
     pkhtb   r9, r7, r6, asr #16 ; [d3 | d2] (A)
@@ -157,7 +160,8 @@ loop_v
     usub16  r6, r8, r9          ; [d2 | d0] (B)
     usub16  r7, r10, r11        ; [d3 | d1] (B)
 
-    add     r2, r2, r5          ; update vsrc pointer
+    add     r2, r2, r3          ; update vsrc pointer
+    add     r5, r5, r12         ; update vpred pointer
 
     pkhbt   r8, r6, r7, lsl #16 ; [d1 | d0] (B)
     pkhtb   r9, r7, r6, asr #16 ; [d3 | d2] (B)
@@ -168,23 +172,25 @@ loop_v
 
     bne     loop_v
 
-    ldmfd   sp!, {r4-r12, pc}
+    ldmfd   sp!, {r4-r11}
+    bx      lr
 
     ENDP
 
 
 ; r0    short *diff
 ; r1    unsigned char *src
-; r2    unsigned char *pred
-; r3    int stride
+; r2    int src_stride
+; r3    unsigned char *pred
+; sp    int pred_stride
 |vp8_subtract_mby_armv6| PROC
 
     stmfd   sp!, {r4-r11}
-
+    ldr     r12, [sp, #32]      ; pred_stride
     mov     r4, #16
 loop
     ldr     r6, [r1]            ; src       (A)
-    ldr     r7, [r2], #4        ; pred      (A)
+    ldr     r7, [r3]            ; pred      (A)
 
     uxtb16  r8, r6              ; [s2 | s0] (A)
     uxtb16  r9, r7              ; [p2 | p0] (A)
@@ -195,7 +201,7 @@ loop
     usub16  r7, r10, r11        ; [d3 | d1] (A)
 
     ldr     r10, [r1, #4]       ; src       (B)
-    ldr     r11, [r2], #4       ; pred      (B)
+    ldr     r11, [r3, #4]       ; pred      (B)
 
     pkhbt   r8, r6, r7, lsl #16 ; [d1 | d0] (A)
     pkhtb   r9, r7, r6, asr #16 ; [d3 | d2] (A)
@@ -212,7 +218,7 @@ loop
     usub16  r7, r10, r11        ; [d3 | d1] (B)
 
     ldr     r10, [r1, #8]       ; src       (C)
-    ldr     r11, [r2], #4       ; pred      (C)
+    ldr     r11, [r3, #8]       ; pred      (C)
 
     pkhbt   r8, r6, r7, lsl #16 ; [d1 | d0] (B)
     pkhtb   r9, r7, r6, asr #16 ; [d3 | d2] (B)
@@ -229,10 +235,10 @@ loop
     usub16  r7, r10, r11        ; [d3 | d1] (C)
 
     ldr     r10, [r1, #12]      ; src       (D)
-    ldr     r11, [r2], #4       ; pred      (D)
+    ldr     r11, [r3, #12]      ; pred      (D)
 
-    pkhbt   r8, r6, r7, lsl #16  ; [d1 | d0] (C)
-    pkhtb   r9, r7, r6, asr #16  ; [d3 | d2] (C)
+    pkhbt   r8, r6, r7, lsl #16 ; [d1 | d0] (C)
+    pkhtb   r9, r7, r6, asr #16 ; [d3 | d2] (C)
 
     str     r8, [r0], #4        ; diff      (C)
     uxtb16  r8, r10             ; [s2 | s0] (D)
@@ -245,7 +251,8 @@ loop
     usub16  r6, r8, r9          ; [d2 | d0] (D)
     usub16  r7, r10, r11        ; [d3 | d1] (D)
 
-    add     r1, r1, r3          ; update src pointer
+    add     r1, r1, r2          ; update src pointer
+    add     r3, r3, r12         ; update pred pointer
 
     pkhbt   r8, r6, r7, lsl #16 ; [d1 | d0] (D)
     pkhtb   r9, r7, r6, asr #16 ; [d3 | d2] (D)
@@ -257,7 +264,7 @@ loop
     bne     loop
 
     ldmfd   sp!, {r4-r11}
-    mov     pc, lr
+    bx      lr
 
     ENDP