diff options
author | Chih-Hung Hsieh <chh@google.com> | 2015-08-19 11:44:57 -0700 |
---|---|---|
committer | Steve Kondik <steve@cyngn.com> | 2016-01-08 21:46:45 -0800 |
commit | 687730892c7bed143df27df76cfb673516a5b6a4 (patch) | |
tree | 4895ca2f379b8e6f2af57f0b45673963724b5f24 /media/libstagefright/codecs/on2/h264dec/source/arm_neon_asm_gcc/h264bsdWriteMacroblock.S | |
parent | 94f3841a5218c0e8d5dd19c3f92e38419f808725 (diff) | |
download | frameworks_av-687730892c7bed143df27df76cfb673516a5b6a4.zip frameworks_av-687730892c7bed143df27df76cfb673516a5b6a4.tar.gz frameworks_av-687730892c7bed143df27df76cfb673516a5b6a4.tar.bz2 |
Use unified syntax to compile with both llvm and gas.
* Remove useless stab debug info directives .func and .endfunc,
which are not recognized by clang/llvm integrated assembler.
* Replace require8 with REQUIRE8, preserve8 with PRESERVE8.
* Replace LDRNESB with LDRSBNE, LDREQSB with LDRSBEQ, etc.
* Replace VST1 {*.U8} with VST1.8 ..., VLD1 {*.U8} with VLD1.8,
and VMOV with VMOV.I8.
* New assembler output code is identical or equivalent to old gas output.
For example, the 3 references of
LDR r7, =0x80808080
in omxdl/arm_neon/vc/m4p10/src_gcc/armVCM4P10_Average_4x_Align_unsafe_s.S
produced 3 instances of
.word 0x80808080
by llvm integrated assembler but only one by gas.
Change-Id: Ifdcd5bee915a7534198d7f219bceab72045f6a22
Diffstat (limited to 'media/libstagefright/codecs/on2/h264dec/source/arm_neon_asm_gcc/h264bsdWriteMacroblock.S')
-rw-r--r-- | media/libstagefright/codecs/on2/h264dec/source/arm_neon_asm_gcc/h264bsdWriteMacroblock.S | 161 |
1 files changed, 80 insertions, 81 deletions
diff --git a/media/libstagefright/codecs/on2/h264dec/source/arm_neon_asm_gcc/h264bsdWriteMacroblock.S b/media/libstagefright/codecs/on2/h264dec/source/arm_neon_asm_gcc/h264bsdWriteMacroblock.S index 495d560..4093b92 100644 --- a/media/libstagefright/codecs/on2/h264dec/source/arm_neon_asm_gcc/h264bsdWriteMacroblock.S +++ b/media/libstagefright/codecs/on2/h264dec/source/arm_neon_asm_gcc/h264bsdWriteMacroblock.S @@ -16,8 +16,8 @@ #include "asm_common.S" - require8 - preserve8 + REQUIRE8 + PRESERVE8 .arm .fpu neon @@ -34,39 +34,39 @@ /* -- NEON registers -- */ -#define qRow0 Q0.U8 -#define qRow1 Q1.U8 -#define qRow2 Q2.U8 -#define qRow3 Q3.U8 -#define qRow4 Q4.U8 -#define qRow5 Q5.U8 -#define qRow6 Q6.U8 -#define qRow7 Q7.U8 -#define qRow8 Q8.U8 -#define qRow9 Q9.U8 -#define qRow10 Q10.U8 -#define qRow11 Q11.U8 -#define qRow12 Q12.U8 -#define qRow13 Q13.U8 -#define qRow14 Q14.U8 -#define qRow15 Q15.U8 - -#define dRow0 D0.U8 -#define dRow1 D1.U8 -#define dRow2 D2.U8 -#define dRow3 D3.U8 -#define dRow4 D4.U8 -#define dRow5 D5.U8 -#define dRow6 D6.U8 -#define dRow7 D7.U8 -#define dRow8 D8.U8 -#define dRow9 D9.U8 -#define dRow10 D10.U8 -#define dRow11 D11.U8 -#define dRow12 D12.U8 -#define dRow13 D13.U8 -#define dRow14 D14.U8 -#define dRow15 D15.U8 +#define qRow0 Q0 +#define qRow1 Q1 +#define qRow2 Q2 +#define qRow3 Q3 +#define qRow4 Q4 +#define qRow5 Q5 +#define qRow6 Q6 +#define qRow7 Q7 +#define qRow8 Q8 +#define qRow9 Q9 +#define qRow10 Q10 +#define qRow11 Q11 +#define qRow12 Q12 +#define qRow13 Q13 +#define qRow14 Q14 +#define qRow15 Q15 + +#define dRow0 D0 +#define dRow1 D1 +#define dRow2 D2 +#define dRow3 D3 +#define dRow4 D4 +#define dRow5 D5 +#define dRow6 D6 +#define dRow7 D7 +#define dRow8 D8 +#define dRow9 D9 +#define dRow10 D10 +#define dRow11 D11 +#define dRow12 D12 +#define dRow13 D13 +#define dRow14 D14 +#define dRow15 D15 /*------------------------------------------------------------------------------ @@ -99,59 +99,58 @@ function h264bsdWriteMacroblock, export=1 @ Write luma - VLD1 {qRow0, qRow1}, [data]! + VLD1.8 {qRow0, qRow1}, [data]! LSL width, width, #4 - VLD1 {qRow2, qRow3}, [data]! + VLD1.8 {qRow2, qRow3}, [data]! LSR cwidth, width, #1 - VST1 {qRow0}, [luma,:128], width - VLD1 {qRow4, qRow5}, [data]! - VST1 {qRow1}, [luma,:128], width - VLD1 {qRow6, qRow7}, [data]! - VST1 {qRow2}, [luma,:128], width - VLD1 {qRow8, qRow9}, [data]! - VST1 {qRow3}, [luma,:128], width - VLD1 {qRow10, qRow11}, [data]! - VST1 {qRow4}, [luma,:128], width - VLD1 {qRow12, qRow13}, [data]! - VST1 {qRow5}, [luma,:128], width - VLD1 {qRow14, qRow15}, [data]! - VST1 {qRow6}, [luma,:128], width - - VLD1 {qRow0, qRow1}, [data]! ;//cb rows 0,1,2,3 - VST1 {qRow7}, [luma,:128], width - VLD1 {qRow2, qRow3}, [data]! ;//cb rows 4,5,6,7 - VST1 {qRow8}, [luma,:128], width - VLD1 {qRow4, qRow5}, [data]! ;//cr rows 0,1,2,3 - VST1 {qRow9}, [luma,:128], width - VLD1 {qRow6, qRow7}, [data]! ;//cr rows 4,5,6,7 - VST1 {qRow10}, [luma,:128], width - VST1 {dRow0}, [cb,:64], cwidth - VST1 {dRow8}, [cr,:64], cwidth - VST1 {qRow11}, [luma,:128], width - VST1 {dRow1}, [cb,:64], cwidth - VST1 {dRow9}, [cr,:64], cwidth - VST1 {qRow12}, [luma,:128], width - VST1 {dRow2}, [cb,:64], cwidth - VST1 {dRow10}, [cr,:64], cwidth - VST1 {qRow13}, [luma,:128], width - VST1 {dRow3}, [cb,:64], cwidth - VST1 {dRow11}, [cr,:64], cwidth - VST1 {qRow14}, [luma,:128], width - VST1 {dRow4}, [cb,:64], cwidth - VST1 {dRow12}, [cr,:64], cwidth - VST1 {qRow15}, [luma] - VST1 {dRow5}, [cb,:64], cwidth - VST1 {dRow13}, [cr,:64], cwidth - VST1 {dRow6}, [cb,:64], cwidth - VST1 {dRow14}, [cr,:64], cwidth - VST1 {dRow7}, [cb,:64] - VST1 {dRow15}, [cr,:64] + VST1.8 {qRow0}, [luma,:128], width + VLD1.8 {qRow4, qRow5}, [data]! + VST1.8 {qRow1}, [luma,:128], width + VLD1.8 {qRow6, qRow7}, [data]! + VST1.8 {qRow2}, [luma,:128], width + VLD1.8 {qRow8, qRow9}, [data]! + VST1.8 {qRow3}, [luma,:128], width + VLD1.8 {qRow10, qRow11}, [data]! + VST1.8 {qRow4}, [luma,:128], width + VLD1.8 {qRow12, qRow13}, [data]! + VST1.8 {qRow5}, [luma,:128], width + VLD1.8 {qRow14, qRow15}, [data]! + VST1.8 {qRow6}, [luma,:128], width + + VLD1.8 {qRow0, qRow1}, [data]! ;//cb rows 0,1,2,3 + VST1.8 {qRow7}, [luma,:128], width + VLD1.8 {qRow2, qRow3}, [data]! ;//cb rows 4,5,6,7 + VST1.8 {qRow8}, [luma,:128], width + VLD1.8 {qRow4, qRow5}, [data]! ;//cr rows 0,1,2,3 + VST1.8 {qRow9}, [luma,:128], width + VLD1.8 {qRow6, qRow7}, [data]! ;//cr rows 4,5,6,7 + VST1.8 {qRow10}, [luma,:128], width + VST1.8 {dRow0}, [cb,:64], cwidth + VST1.8 {dRow8}, [cr,:64], cwidth + VST1.8 {qRow11}, [luma,:128], width + VST1.8 {dRow1}, [cb,:64], cwidth + VST1.8 {dRow9}, [cr,:64], cwidth + VST1.8 {qRow12}, [luma,:128], width + VST1.8 {dRow2}, [cb,:64], cwidth + VST1.8 {dRow10}, [cr,:64], cwidth + VST1.8 {qRow13}, [luma,:128], width + VST1.8 {dRow3}, [cb,:64], cwidth + VST1.8 {dRow11}, [cr,:64], cwidth + VST1.8 {qRow14}, [luma,:128], width + VST1.8 {dRow4}, [cb,:64], cwidth + VST1.8 {dRow12}, [cr,:64], cwidth + VST1.8 {qRow15}, [luma] + VST1.8 {dRow5}, [cb,:64], cwidth + VST1.8 {dRow13}, [cr,:64], cwidth + VST1.8 {dRow6}, [cb,:64], cwidth + VST1.8 {dRow14}, [cr,:64], cwidth + VST1.8 {dRow7}, [cb,:64] + VST1.8 {dRow15}, [cr,:64] VPOP {q4-q7} POP {r4-r6,pc} @ BX lr - .endfunc |