ref: dc004e8c17a1f330ed46eab092ca67e6f785fee1
parent: 02c30cdeeffcf8bf72b194925b91197f842a7fc1
parent: e3f850ee055a9238d17d6c503af27e287e6db28d
author: Johann <johannkoenig@google.com>
date: Tue Jun 28 12:34:10 EDT 2011
Merge "Avoid text relocations in ARM vp8 decoder"
--- a/vp8/common/arm/armv6/sixtappredict8x4_v6.asm
+++ b/vp8/common/arm/armv6/sixtappredict8x4_v6.asm
@@ -32,7 +32,7 @@
beq skip_firstpass_filter
;first-pass filter
- ldr r12, _filter8_coeff_
+ adr r12, filter8_coeff
sub r0, r0, r1, lsl #1
add r3, r1, #10 ; preload next low
@@ -127,7 +127,7 @@
cmp r3, #0
beq skip_secondpass_filter
- ldr r12, _filter8_coeff_
+ adr r12, filter8_coeff
add lr, r12, r3, lsl #4 ;calculate filter location
mov r2, #0x00080000
@@ -251,8 +251,6 @@
;-----------------
;One word each is reserved. Label filter_coeff can be used to access the data.
;Data address: filter_coeff, filter_coeff+4, filter_coeff+8 ...
-_filter8_coeff_
- DCD filter8_coeff
filter8_coeff
DCD 0x00000000, 0x00000080, 0x00000000, 0x00000000
DCD 0xfffa0000, 0x000c007b, 0x0000ffff, 0x00000000
--- a/vp8/common/arm/neon/bilinearpredict16x16_neon.asm
+++ b/vp8/common/arm/neon/bilinearpredict16x16_neon.asm
@@ -25,7 +25,7 @@
|vp8_bilinear_predict16x16_neon| PROC
push {r4-r5, lr}- ldr r12, _bifilter16_coeff_
+ adr r12, bifilter16_coeff
ldr r4, [sp, #12] ;load parameters from stack
ldr r5, [sp, #16] ;load parameters from stack
@@ -351,8 +351,6 @@
;-----------------
-_bifilter16_coeff_
- DCD bifilter16_coeff
bifilter16_coeff
DCD 128, 0, 112, 16, 96, 32, 80, 48, 64, 64, 48, 80, 32, 96, 16, 112
--- a/vp8/common/arm/neon/bilinearpredict4x4_neon.asm
+++ b/vp8/common/arm/neon/bilinearpredict4x4_neon.asm
@@ -25,7 +25,7 @@
|vp8_bilinear_predict4x4_neon| PROC
push {r4, lr}- ldr r12, _bifilter4_coeff_
+ adr r12, bifilter4_coeff
ldr r4, [sp, #8] ;load parameters from stack
ldr lr, [sp, #12] ;load parameters from stack
@@ -124,8 +124,6 @@
;-----------------
-_bifilter4_coeff_
- DCD bifilter4_coeff
bifilter4_coeff
DCD 128, 0, 112, 16, 96, 32, 80, 48, 64, 64, 48, 80, 32, 96, 16, 112
--- a/vp8/common/arm/neon/bilinearpredict8x4_neon.asm
+++ b/vp8/common/arm/neon/bilinearpredict8x4_neon.asm
@@ -25,7 +25,7 @@
|vp8_bilinear_predict8x4_neon| PROC
push {r4, lr}- ldr r12, _bifilter8x4_coeff_
+ adr r12, bifilter8x4_coeff
ldr r4, [sp, #8] ;load parameters from stack
ldr lr, [sp, #12] ;load parameters from stack
@@ -129,8 +129,6 @@
;-----------------
-_bifilter8x4_coeff_
- DCD bifilter8x4_coeff
bifilter8x4_coeff
DCD 128, 0, 112, 16, 96, 32, 80, 48, 64, 64, 48, 80, 32, 96, 16, 112
--- a/vp8/common/arm/neon/bilinearpredict8x8_neon.asm
+++ b/vp8/common/arm/neon/bilinearpredict8x8_neon.asm
@@ -25,7 +25,7 @@
|vp8_bilinear_predict8x8_neon| PROC
push {r4, lr}- ldr r12, _bifilter8_coeff_
+ adr r12, bifilter8_coeff
ldr r4, [sp, #8] ;load parameters from stack
ldr lr, [sp, #12] ;load parameters from stack
@@ -177,8 +177,6 @@
;-----------------
-_bifilter8_coeff_
- DCD bifilter8_coeff
bifilter8_coeff
DCD 128, 0, 112, 16, 96, 32, 80, 48, 64, 64, 48, 80, 32, 96, 16, 112
--- a/vp8/common/arm/neon/shortidct4x4llm_neon.asm
+++ b/vp8/common/arm/neon/shortidct4x4llm_neon.asm
@@ -31,7 +31,7 @@
;result of the multiplication that is needed in IDCT.
|vp8_short_idct4x4llm_neon| PROC
- ldr r12, _idct_coeff_
+ adr r12, idct_coeff
vld1.16 {q1, q2}, [r0] vld1.16 {d0}, [r12]@@ -114,8 +114,6 @@
;-----------------
-_idct_coeff_
- DCD idct_coeff
idct_coeff
DCD 0x4e7b4e7b, 0x8a8c8a8c
--- a/vp8/common/arm/neon/sixtappredict16x16_neon.asm
+++ b/vp8/common/arm/neon/sixtappredict16x16_neon.asm
@@ -15,6 +15,17 @@
PRESERVE8
AREA ||.text||, CODE, READONLY, ALIGN=2
+
+filter16_coeff
+ DCD 0, 0, 128, 0, 0, 0, 0, 0
+ DCD 0, -6, 123, 12, -1, 0, 0, 0
+ DCD 2, -11, 108, 36, -8, 1, 0, 0
+ DCD 0, -9, 93, 50, -6, 0, 0, 0
+ DCD 3, -16, 77, 77, -16, 3, 0, 0
+ DCD 0, -6, 50, 93, -9, 0, 0, 0
+ DCD 1, -8, 36, 108, -11, 2, 0, 0
+ DCD 0, -1, 12, 123, -6, 0, 0, 0
+
; r0 unsigned char *src_ptr,
; r1 int src_pixels_per_line,
; r2 int xoffset,
@@ -33,7 +44,7 @@
|vp8_sixtap_predict16x16_neon| PROC
push {r4-r5, lr}- ldr r12, _filter16_coeff_
+ adr r12, filter16_coeff
ldr r4, [sp, #12] ;load parameters from stack
ldr r5, [sp, #16] ;load parameters from stack
@@ -476,17 +487,4 @@
ENDP
;-----------------
-
-_filter16_coeff_
- DCD filter16_coeff
-filter16_coeff
- DCD 0, 0, 128, 0, 0, 0, 0, 0
- DCD 0, -6, 123, 12, -1, 0, 0, 0
- DCD 2, -11, 108, 36, -8, 1, 0, 0
- DCD 0, -9, 93, 50, -6, 0, 0, 0
- DCD 3, -16, 77, 77, -16, 3, 0, 0
- DCD 0, -6, 50, 93, -9, 0, 0, 0
- DCD 1, -8, 36, 108, -11, 2, 0, 0
- DCD 0, -1, 12, 123, -6, 0, 0, 0
-
END
--- a/vp8/common/arm/neon/sixtappredict4x4_neon.asm
+++ b/vp8/common/arm/neon/sixtappredict4x4_neon.asm
@@ -15,6 +15,17 @@
PRESERVE8
AREA ||.text||, CODE, READONLY, ALIGN=2
+
+filter4_coeff
+ DCD 0, 0, 128, 0, 0, 0, 0, 0
+ DCD 0, -6, 123, 12, -1, 0, 0, 0
+ DCD 2, -11, 108, 36, -8, 1, 0, 0
+ DCD 0, -9, 93, 50, -6, 0, 0, 0
+ DCD 3, -16, 77, 77, -16, 3, 0, 0
+ DCD 0, -6, 50, 93, -9, 0, 0, 0
+ DCD 1, -8, 36, 108, -11, 2, 0, 0
+ DCD 0, -1, 12, 123, -6, 0, 0, 0
+
; r0 unsigned char *src_ptr,
; r1 int src_pixels_per_line,
; r2 int xoffset,
@@ -25,7 +36,7 @@
|vp8_sixtap_predict_neon| PROC
push {r4, lr}- ldr r12, _filter4_coeff_
+ adr r12, filter4_coeff
ldr r4, [sp, #8] ;load parameters from stack
ldr lr, [sp, #12] ;load parameters from stack
@@ -407,17 +418,5 @@
ENDP
;-----------------
-
-_filter4_coeff_
- DCD filter4_coeff
-filter4_coeff
- DCD 0, 0, 128, 0, 0, 0, 0, 0
- DCD 0, -6, 123, 12, -1, 0, 0, 0
- DCD 2, -11, 108, 36, -8, 1, 0, 0
- DCD 0, -9, 93, 50, -6, 0, 0, 0
- DCD 3, -16, 77, 77, -16, 3, 0, 0
- DCD 0, -6, 50, 93, -9, 0, 0, 0
- DCD 1, -8, 36, 108, -11, 2, 0, 0
- DCD 0, -1, 12, 123, -6, 0, 0, 0
END
--- a/vp8/common/arm/neon/sixtappredict8x4_neon.asm
+++ b/vp8/common/arm/neon/sixtappredict8x4_neon.asm
@@ -15,6 +15,17 @@
PRESERVE8
AREA ||.text||, CODE, READONLY, ALIGN=2
+
+filter8_coeff
+ DCD 0, 0, 128, 0, 0, 0, 0, 0
+ DCD 0, -6, 123, 12, -1, 0, 0, 0
+ DCD 2, -11, 108, 36, -8, 1, 0, 0
+ DCD 0, -9, 93, 50, -6, 0, 0, 0
+ DCD 3, -16, 77, 77, -16, 3, 0, 0
+ DCD 0, -6, 50, 93, -9, 0, 0, 0
+ DCD 1, -8, 36, 108, -11, 2, 0, 0
+ DCD 0, -1, 12, 123, -6, 0, 0, 0
+
; r0 unsigned char *src_ptr,
; r1 int src_pixels_per_line,
; r2 int xoffset,
@@ -25,7 +36,7 @@
|vp8_sixtap_predict8x4_neon| PROC
push {r4-r5, lr}- ldr r12, _filter8_coeff_
+ adr r12, filter8_coeff
ldr r4, [sp, #12] ;load parameters from stack
ldr r5, [sp, #16] ;load parameters from stack
@@ -458,17 +469,5 @@
ENDP
;-----------------
-
-_filter8_coeff_
- DCD filter8_coeff
-filter8_coeff
- DCD 0, 0, 128, 0, 0, 0, 0, 0
- DCD 0, -6, 123, 12, -1, 0, 0, 0
- DCD 2, -11, 108, 36, -8, 1, 0, 0
- DCD 0, -9, 93, 50, -6, 0, 0, 0
- DCD 3, -16, 77, 77, -16, 3, 0, 0
- DCD 0, -6, 50, 93, -9, 0, 0, 0
- DCD 1, -8, 36, 108, -11, 2, 0, 0
- DCD 0, -1, 12, 123, -6, 0, 0, 0
END
--- a/vp8/common/arm/neon/sixtappredict8x8_neon.asm
+++ b/vp8/common/arm/neon/sixtappredict8x8_neon.asm
@@ -15,6 +15,17 @@
PRESERVE8
AREA ||.text||, CODE, READONLY, ALIGN=2
+
+filter8_coeff
+ DCD 0, 0, 128, 0, 0, 0, 0, 0
+ DCD 0, -6, 123, 12, -1, 0, 0, 0
+ DCD 2, -11, 108, 36, -8, 1, 0, 0
+ DCD 0, -9, 93, 50, -6, 0, 0, 0
+ DCD 3, -16, 77, 77, -16, 3, 0, 0
+ DCD 0, -6, 50, 93, -9, 0, 0, 0
+ DCD 1, -8, 36, 108, -11, 2, 0, 0
+ DCD 0, -1, 12, 123, -6, 0, 0, 0
+
; r0 unsigned char *src_ptr,
; r1 int src_pixels_per_line,
; r2 int xoffset,
@@ -25,7 +36,7 @@
|vp8_sixtap_predict8x8_neon| PROC
push {r4-r5, lr}- ldr r12, _filter8_coeff_
+ adr r12, filter8_coeff
ldr r4, [sp, #12] ;load parameters from stack
ldr r5, [sp, #16] ;load parameters from stack
@@ -509,17 +520,5 @@
ENDP
;-----------------
-
-_filter8_coeff_
- DCD filter8_coeff
-filter8_coeff
- DCD 0, 0, 128, 0, 0, 0, 0, 0
- DCD 0, -6, 123, 12, -1, 0, 0, 0
- DCD 2, -11, 108, 36, -8, 1, 0, 0
- DCD 0, -9, 93, 50, -6, 0, 0, 0
- DCD 3, -16, 77, 77, -16, 3, 0, 0
- DCD 0, -6, 50, 93, -9, 0, 0, 0
- DCD 1, -8, 36, 108, -11, 2, 0, 0
- DCD 0, -1, 12, 123, -6, 0, 0, 0
END
--- a/vp8/decoder/arm/neon/dequant_idct_neon.asm
+++ b/vp8/decoder/arm/neon/dequant_idct_neon.asm
@@ -35,7 +35,7 @@
ldr r1, [sp, #4] ; stride
- ldr r12, _CONSTANTS_
+ adr r12, cospi8sqrt2minus1 ; pointer to the first constant
vmul.i16 q1, q3, q5 ;input for short_idct4x4llm_neon
vmul.i16 q2, q4, q6
@@ -123,7 +123,6 @@
ENDP ; |vp8_dequant_idct_add_neon|
; Constant Pool
-_CONSTANTS_ DCD cospi8sqrt2minus1
cospi8sqrt2minus1 DCD 0x4e7b4e7b
sinpi8sqrt2 DCD 0x8a8c8a8c
--- a/vp8/decoder/arm/neon/idct_dequant_dc_full_2x_neon.asm
+++ b/vp8/decoder/arm/neon/idct_dequant_dc_full_2x_neon.asm
@@ -41,7 +41,7 @@
ldr r1, [sp, #4]
vld1.32 {d31[1]}, [r12]- ldr r2, _CONSTANTS_
+ adr r2, cospi8sqrt2minus1 ; pointer to the first constant
ldrh r12, [r1], #2 ; lo *dc
ldrh r1, [r1] ; hi *dc
@@ -198,7 +198,6 @@
ENDP ; |idct_dequant_dc_full_2x_neon|
; Constant Pool
-_CONSTANTS_ DCD cospi8sqrt2minus1
cospi8sqrt2minus1 DCD 0x4e7b
; because the lowest bit in 0x8a8c is 0, we can pre-shift this
sinpi8sqrt2 DCD 0x4546
--- a/vp8/decoder/arm/neon/idct_dequant_full_2x_neon.asm
+++ b/vp8/decoder/arm/neon/idct_dequant_full_2x_neon.asm
@@ -40,7 +40,7 @@
vld1.32 {d31[0]}, [r2] vld1.32 {d31[1]}, [r12]- ldr r2, _CONSTANTS_
+ adr r2, cospi8sqrt2minus1 ; pointer to the first constant
; dequant: q[i] = q[i] * dq[i]
vmul.i16 q2, q2, q0
@@ -190,7 +190,6 @@
ENDP ; |idct_dequant_full_2x_neon|
; Constant Pool
-_CONSTANTS_ DCD cospi8sqrt2minus1
cospi8sqrt2minus1 DCD 0x4e7b
; because the lowest bit in 0x8a8c is 0, we can pre-shift this
sinpi8sqrt2 DCD 0x4546
--
⑨