Chiebot-Mirror
/
opencv
8
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Remove alignment reqirements from NEON optimizations in libpng

Browse Source 
On Android memory is not guaranted to be aligned and applications often crash with BUSERROR
pull/35/merge
Andrey Kamaev 13 years ago
parent f017ad5943
commit 2885eb86b8
1 changed files with 14 additions and 14 deletions
Whitespace
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Split View
Diff Options
Show Stats Download Patch File Download Diff File
  1. 28
      3rdparty/libpng/arm/filter_neon.S

28
3rdparty/libpng/arm/filter_neon.S vendored
Unescape View File

@ -41,12 +41,12 @@ func png_read_filter_row_sub4_neon, export=1
ldr r3, [r0, #4] @ rowbytes
vmov.i8 d3, #0
1:
vld4.32 {d4[],d5[],d6[],d7[]}, [r1,:128]
vld4.32 {d4[],d5[],d6[],d7[]}, [r1]
vadd.u8 d0, d3, d4
vadd.u8 d1, d0, d5
vadd.u8 d2, d1, d6
vadd.u8 d3, d2, d7
vst4.32 {d0[0],d1[0],d2[0],d3[0]},[r1,:128]!
vst4.32 {d0[0],d1[0],d2[0],d3[0]},[r1]!
subs r3, r3, #16
bgt 1b
@ -67,7 +67,7 @@ func png_read_filter_row_sub3_neon, export=1
vadd.u8 d1, d0, d5
vext.8 d7, d23, d23, #1
vld1.8 {q11}, [r0], r12
vst1.32 {d0[0]}, [r1,:32], r2
vst1.32 {d0[0]}, [r1], r2
vadd.u8 d2, d1, d6
vst1.32 {d1[0]}, [r1], r2
vadd.u8 d3, d2, d7
@ -82,10 +82,10 @@ endfunc
func png_read_filter_row_up_neon, export=1
ldr r3, [r0, #4] @ rowbytes
1:
vld1.8 {q0}, [r1,:128]
vld1.8 {q1}, [r2,:128]!
vld1.8 {q0}, [r1]
vld1.8 {q1}, [r2]!
vadd.u8 q0, q0, q1
vst1.8 {q0}, [r1,:128]!
vst1.8 {q0}, [r1]!
subs r3, r3, #16
bgt 1b
@ -96,8 +96,8 @@ func png_read_filter_row_avg4_neon, export=1
ldr r12, [r0, #4] @ rowbytes
vmov.i8 d3, #0
1:
vld4.32 {d4[],d5[],d6[],d7[]}, [r1,:128]
vld4.32 {d16[],d17[],d18[],d19[]},[r2,:128]!
vld4.32 {d4[],d5[],d6[],d7[]}, [r1]
vld4.32 {d16[],d17[],d18[],d19[]},[r2]!
vhadd.u8 d0, d3, d16
vadd.u8 d0, d0, d4
vhadd.u8 d1, d0, d17
@ -106,7 +106,7 @@ func png_read_filter_row_avg4_neon, export=1
vadd.u8 d2, d2, d6
vhadd.u8 d3, d2, d19
vadd.u8 d3, d3, d7
vst4.32 {d0[0],d1[0],d2[0],d3[0]},[r1,:128]!
vst4.32 {d0[0],d1[0],d2[0],d3[0]},[r1]!
subs r12, r12, #16
bgt 1b
@ -133,7 +133,7 @@ func png_read_filter_row_avg3_neon, export=1
vadd.u8 d1, d1, d5
vext.8 d7, d23, d23, #1
vld1.8 {q11}, [r0], lr
vst1.32 {d0[0]}, [r1,:32], r4
vst1.32 {d0[0]}, [r1], r4
vhadd.u8 d2, d1, d18
vst1.32 {d1[0]}, [r1], r4
vext.8 d19, d21, d21, #1
@ -169,8 +169,8 @@ func png_read_filter_row_paeth4_neon, export=1
vmov.i8 d3, #0
vmov.i8 d20, #0
1:
vld4.32 {d4[],d5[],d6[],d7[]}, [r1,:128]
vld4.32 {d16[],d17[],d18[],d19[]},[r2,:128]!
vld4.32 {d4[],d5[],d6[],d7[]}, [r1]
vld4.32 {d16[],d17[],d18[],d19[]},[r2]!
paeth d0, d3, d16, d20
vadd.u8 d0, d0, d4
paeth d1, d0, d17, d16
@ -180,7 +180,7 @@ func png_read_filter_row_paeth4_neon, export=1
paeth d3, d2, d19, d18
vmov d20, d19
vadd.u8 d3, d3, d7
vst4.32 {d0[0],d1[0],d2[0],d3[0]},[r1,:128]!
vst4.32 {d0[0],d1[0],d2[0],d3[0]},[r1]!
subs r12, r12, #16
bgt 1b
@ -203,7 +203,7 @@ func png_read_filter_row_paeth3_neon, export=1
vadd.u8 d0, d0, d22
vext.8 d17, d20, d21, #3
paeth d1, d0, d17, d20
vst1.32 {d0[0]}, [r1,:32], r4
vst1.32 {d0[0]}, [r1], r4
vext.8 d6, d22, d23, #6
vadd.u8 d1, d1, d5
vext.8 d18, d20, d21, #6

Write Preview
Loading…
Cancel
Save