July 2021 - ffmpeg-cvslog - lists.ffmpeg.org

avutil/mips: Use $at as MMI macro temporary register
by Jiaxun Yang 28 Jul '21

28 Jul '21

ffmpeg | branch: master | Jiaxun Yang <jiaxun.yang(a)flygoat.com> | Fri Jul 23 13:53:43 2021 +0800| [a1cd62883f84072b7c9fa13e9a2512e13618399e] | committer: Michael Niedermayer avutil/mips: Use $at as MMI macro temporary register Some function had exceed 30 inline assembly register oprands limiation when using LOONGSON2 version of MMI macros. We can avoid that by take $at, which is register reserved for assembler, as temporary register. As none of instructions used in these macros is pseudo, it is safe to utilize $at here. Signed-off-by: Jiaxun Yang <jiaxun.yang(a)flygoat.com> Reviewed-by: Shiyou Yin <yinshiyou-hf(a)loongson.cn> Signed-off-by: Michael Niedermayer <michael(a)niedermayer.cc> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=a1cd62883f84072b7… --- libavutil/mips/mmiutils.h | 108 ++++++++++++++++++++++++++++------------------ 1 file changed, 66 insertions(+), 42 deletions(-) diff --git a/libavutil/mips/mmiutils.h b/libavutil/mips/mmiutils.h index 41715c6490..7991e14e84 100644 --- a/libavutil/mips/mmiutils.h +++ b/libavutil/mips/mmiutils.h @@ -29,74 +29,103 @@ #include "libavutil/mem_internal.h" #include "libavutil/mips/asmdefs.h" -#if HAVE_LOONGSON2 +/* + * These were used to define temporary registers for MMI marcos + * however now we're using $at. They're theoretically unnecessary + * but just leave them here to avoid mess. + */ +#define DECLARE_VAR_LOW32 +#define RESTRICT_ASM_LOW32 +#define DECLARE_VAR_ALL64 +#define RESTRICT_ASM_ALL64 +#define DECLARE_VAR_ADDRT +#define RESTRICT_ASM_ADDRT -#define DECLARE_VAR_LOW32 int32_t low32 -#define RESTRICT_ASM_LOW32 [low32]"=&r"(low32), -#define DECLARE_VAR_ALL64 int64_t all64 -#define RESTRICT_ASM_ALL64 [all64]"=&r"(all64), -#define DECLARE_VAR_ADDRT mips_reg addrt -#define RESTRICT_ASM_ADDRT [addrt]"=&r"(addrt), +#if HAVE_LOONGSON2 #define MMI_LWX(reg, addr, stride, bias) \ - PTR_ADDU "%[addrt], "#addr", "#stride" \n\t" \ - "lw "#reg", "#bias"(%[addrt]) \n\t" + ".set noat \n\t" \ + PTR_ADDU "$at, "#addr", "#stride" \n\t" \ + "lw "#reg", "#bias"($at) \n\t" \ + ".set at \n\t" #define MMI_SWX(reg, addr, stride, bias) \ - PTR_ADDU "%[addrt], "#addr", "#stride" \n\t" \ - "sw "#reg", "#bias"(%[addrt]) \n\t" + ".set noat \n\t" \ + PTR_ADDU "$at, "#addr", "#stride" \n\t" \ + "sw "#reg", "#bias"($at) \n\t" \ + ".set at \n\t" #define MMI_LDX(reg, addr, stride, bias) \ - PTR_ADDU "%[addrt], "#addr", "#stride" \n\t" \ - "ld "#reg", "#bias"(%[addrt]) \n\t" + ".set noat \n\t" \ + PTR_ADDU "$at, "#addr", "#stride" \n\t" \ + "ld "#reg", "#bias"($at) \n\t" \ + ".set at \n\t" #define MMI_SDX(reg, addr, stride, bias) \ - PTR_ADDU "%[addrt], "#addr", "#stride" \n\t" \ - "sd "#reg", "#bias"(%[addrt]) \n\t" + ".set noat \n\t" \ + PTR_ADDU "$at, "#addr", "#stride" \n\t" \ + "sd "#reg", "#bias"($at) \n\t" \ + ".set at \n\t" #define MMI_LWC1(fp, addr, bias) \ "lwc1 "#fp", "#bias"("#addr") \n\t" #define MMI_ULWC1(fp, addr, bias) \ - "ulw %[low32], "#bias"("#addr") \n\t" \ - "mtc1 %[low32], "#fp" \n\t" + ".set noat \n\t" \ + "ulw $at, "#bias"("#addr") \n\t" \ + "mtc1 $at, "#fp" \n\t" \ + ".set at \n\t" #define MMI_LWXC1(fp, addr, stride, bias) \ - PTR_ADDU "%[addrt], "#addr", "#stride" \n\t" \ - MMI_LWC1(fp, %[addrt], bias) + ".set noat \n\t" \ + PTR_ADDU "$at, "#addr", "#stride" \n\t" \ + MMI_LWC1(fp, $at, bias) \ + ".set at \n\t" #define MMI_SWC1(fp, addr, bias) \ "swc1 "#fp", "#bias"("#addr") \n\t" #define MMI_USWC1(fp, addr, bias) \ - "mfc1 %[low32], "#fp" \n\t" \ - "usw %[low32], "#bias"("#addr") \n\t" + ".set noat \n\t" \ + "mfc1 $at, "#fp" \n\t" \ + "usw $at, "#bias"("#addr") \n\t" \ + ".set at \n\t" #define MMI_SWXC1(fp, addr, stride, bias) \ - PTR_ADDU "%[addrt], "#addr", "#stride" \n\t" \ - MMI_SWC1(fp, %[addrt], bias) + ".set noat \n\t" \ + PTR_ADDU "$at, "#addr", "#stride" \n\t" \ + MMI_SWC1(fp, $at, bias) \ + ".set at \n\t" #define MMI_LDC1(fp, addr, bias) \ "ldc1 "#fp", "#bias"("#addr") \n\t" #define MMI_ULDC1(fp, addr, bias) \ - "uld %[all64], "#bias"("#addr") \n\t" \ - "dmtc1 %[all64], "#fp" \n\t" + ".set noat \n\t" \ + "uld $at, "#bias"("#addr") \n\t" \ + "dmtc1 $at, "#fp" \n\t" \ + ".set at \n\t" #define MMI_LDXC1(fp, addr, stride, bias) \ - PTR_ADDU "%[addrt], "#addr", "#stride" \n\t" \ - MMI_LDC1(fp, %[addrt], bias) + ".set noat \n\t" \ + PTR_ADDU "$at, "#addr", "#stride" \n\t" \ + MMI_LDC1(fp, $at, bias) \ + ".set at \n\t" #define MMI_SDC1(fp, addr, bias) \ "sdc1 "#fp", "#bias"("#addr") \n\t" #define MMI_USDC1(fp, addr, bias) \ - "dmfc1 %[all64], "#fp" \n\t" \ - "usd %[all64], "#bias"("#addr") \n\t" + ".set noat \n\t" \ + "dmfc1 $at, "#fp" \n\t" \ + "usd $at, "#bias"("#addr") \n\t" \ + ".set at \n\t" #define MMI_SDXC1(fp, addr, stride, bias) \ - PTR_ADDU "%[addrt], "#addr", "#stride" \n\t" \ - MMI_SDC1(fp, %[addrt], bias) + ".set noat \n\t" \ + PTR_ADDU "$at, "#addr", "#stride" \n\t" \ + MMI_SDC1(fp, $at, bias) \ + ".set at \n\t" #define MMI_LQ(reg1, reg2, addr, bias) \ "ld "#reg1", "#bias"("#addr") \n\t" \ @@ -116,11 +145,6 @@ #elif HAVE_LOONGSON3 /* !HAVE_LOONGSON2 */ -#define DECLARE_VAR_ALL64 -#define RESTRICT_ASM_ALL64 -#define DECLARE_VAR_ADDRT -#define RESTRICT_ASM_ADDRT - #define MMI_LWX(reg, addr, stride, bias) \ "gslwx "#reg", "#bias"("#addr", "#stride") \n\t" @@ -138,12 +162,12 @@ #if _MIPS_SIM == _ABIO32 /* workaround for 3A2000 gslwlc1 bug */ -#define DECLARE_VAR_LOW32 int32_t low32 -#define RESTRICT_ASM_LOW32 [low32]"=&r"(low32), - -#define MMI_ULWC1(fp, addr, bias) \ - "ulw %[low32], "#bias"("#addr") \n\t" \ - "mtc1 %[low32], "#fp" \n\t" +#define MMI_LWLRC1(fp, addr, bias, off) \ + ".set noat \n\t" \ + "lwl $at, "#bias"+"#off"("#addr") \n\t" \ + "lwr $at, "#bias"("#addr") \n\t" \ + "mtc1 $at, "#fp" \n\t" \ + ".set at \n\t" #else /* _MIPS_SIM != _ABIO32 */

1 0

avcodec/mips: Use MMI marcos to replace Loongson3 instructions
by Jiaxun Yang 28 Jul '21

28 Jul '21

ffmpeg | branch: master | Jiaxun Yang <jiaxun.yang(a)flygoat.com> | Fri Jul 23 13:53:42 2021 +0800| [1042039ccda7e9099f139b14981b48ee59900804] | committer: Michael Niedermayer avcodec/mips: Use MMI marcos to replace Loongson3 instructions Loongson3's extention instructions (prefixed with gs) are widely used in our MMI codebase. However, these instructions are not avilable on Loongson-2E/F while MMI code should work on these processors. Previously we introduced mmiutils marcos to provide backward compactbility but newly commited code didn't follow that. In this patch I revised the codebase and converted all these instructions into MMI marcos to get Loongson2 supproted again. Signed-off-by: Jiaxun Yang <jiaxun.yang(a)flygoat.com> Reviewed-by: Shiyou Yin <yinshiyou-hf(a)loongson.cn> Signed-off-by: Michael Niedermayer <michael(a)niedermayer.cc> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=1042039ccda7e9099… --- libavcodec/mips/h264chroma_mmi.c | 28 +++-- libavcodec/mips/h264dsp_mmi.c | 8 +- libavcodec/mips/hevcdsp_mmi.c | 251 ++++++++++++++++---------------------- libavcodec/mips/hpeldsp_mmi.c | 1 + libavcodec/mips/simple_idct_mmi.c | 49 ++++---- libavcodec/mips/vp3dsp_idct_mmi.c | 11 +- libavcodec/mips/vp8dsp_mmi.c | 100 ++++++--------- libavcodec/mips/vp9_mc_mmi.c | 128 ++++++++----------- 8 files changed, 247 insertions(+), 329 deletions(-) diff --git a/libavcodec/mips/h264chroma_mmi.c b/libavcodec/mips/h264chroma_mmi.c index cc2d7cb7e9..ec35c5a72e 100644 --- a/libavcodec/mips/h264chroma_mmi.c +++ b/libavcodec/mips/h264chroma_mmi.c @@ -31,6 +31,8 @@ void ff_put_h264_chroma_mc8_mmi(uint8_t *dst, uint8_t *src, ptrdiff_t stride, { double ftmp[12]; union mmi_intfloat64 A, B, C, D, E; + DECLARE_VAR_ALL64; + A.i = 64; if (!(x || y)) { @@ -57,7 +59,8 @@ void ff_put_h264_chroma_mc8_mmi(uint8_t *dst, uint8_t *src, ptrdiff_t stride, MMI_SDC1(%[ftmp3], %[dst], 0x00) PTR_ADDU "%[dst], %[dst], %[stride] \n\t" "bnez %[h], 1b \n\t" - : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), + : RESTRICT_ASM_ALL64 + [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]), [dst]"+&r"(dst), [src]"+&r"(src), [h]"+&r"(h) @@ -151,7 +154,8 @@ void ff_put_h264_chroma_mc8_mmi(uint8_t *dst, uint8_t *src, ptrdiff_t stride, MMI_SDC1(%[ftmp3], %[dst], 0x00) PTR_ADDU "%[dst], %[dst], %[stride] \n\t" "bnez %[h], 1b \n\t" - : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), + : RESTRICT_ASM_ALL64 + [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]), [ftmp4]"=&f"(ftmp[4]), [ftmp5]"=&f"(ftmp[5]), [ftmp6]"=&f"(ftmp[6]), [ftmp7]"=&f"(ftmp[7]), @@ -201,7 +205,8 @@ void ff_put_h264_chroma_mc8_mmi(uint8_t *dst, uint8_t *src, ptrdiff_t stride, MMI_SDC1(%[ftmp1], %[dst], 0x00) PTR_ADDU "%[dst], %[dst], %[stride] \n\t" "bnez %[h], 1b \n\t" - : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), + : RESTRICT_ASM_ALL64 + [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]), [ftmp4]"=&f"(ftmp[4]), [ftmp5]"=&f"(ftmp[5]), [ftmp6]"=&f"(ftmp[6]), [ftmp7]"=&f"(ftmp[7]), @@ -268,7 +273,8 @@ void ff_put_h264_chroma_mc8_mmi(uint8_t *dst, uint8_t *src, ptrdiff_t stride, MMI_SDC1(%[ftmp2], %[dst], 0x00) PTR_ADDU "%[dst], %[dst], %[stride] \n\t" "bnez %[h], 1b \n\t" - : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), + : RESTRICT_ASM_ALL64 + [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]), [ftmp4]"=&f"(ftmp[4]), [ftmp5]"=&f"(ftmp[5]), [ftmp6]"=&f"(ftmp[6]), [ftmp7]"=&f"(ftmp[7]), @@ -288,6 +294,8 @@ void ff_avg_h264_chroma_mc8_mmi(uint8_t *dst, uint8_t *src, ptrdiff_t stride, { double ftmp[10]; union mmi_intfloat64 A, B, C, D, E; + DECLARE_VAR_ALL64; + A.i = 64; if(!(x || y)){ @@ -310,7 +318,8 @@ void ff_avg_h264_chroma_mc8_mmi(uint8_t *dst, uint8_t *src, ptrdiff_t stride, PTR_ADDU "%[dst], %[dst], %[stride] \n\t" "addi %[h], %[h], -0x02 \n\t" "bnez %[h], 1b \n\t" - : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), + : RESTRICT_ASM_ALL64 + [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]), [dst]"+&r"(dst), [src]"+&r"(src), [h]"+&r"(h) @@ -373,7 +382,8 @@ void ff_avg_h264_chroma_mc8_mmi(uint8_t *dst, uint8_t *src, ptrdiff_t stride, MMI_SDC1(%[ftmp1], %[dst], 0x00) PTR_ADDU "%[dst], %[dst], %[stride] \n\t" "bnez %[h], 1b \n\t" - : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), + : RESTRICT_ASM_ALL64 + [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]), [ftmp4]"=&f"(ftmp[4]), [ftmp5]"=&f"(ftmp[5]), [ftmp6]"=&f"(ftmp[6]), [ftmp7]"=&f"(ftmp[7]), @@ -423,7 +433,8 @@ void ff_avg_h264_chroma_mc8_mmi(uint8_t *dst, uint8_t *src, ptrdiff_t stride, MMI_SDC1(%[ftmp1], %[dst], 0x00) PTR_ADDU "%[dst], %[dst], %[stride] \n\t" "bnez %[h], 1b \n\t" - : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), + : RESTRICT_ASM_ALL64 + [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]), [ftmp4]"=&f"(ftmp[4]), [ftmp5]"=&f"(ftmp[5]), [ftmp6]"=&f"(ftmp[6]), [ftmp7]"=&f"(ftmp[7]), @@ -471,7 +482,8 @@ void ff_avg_h264_chroma_mc8_mmi(uint8_t *dst, uint8_t *src, ptrdiff_t stride, MMI_SDC1(%[ftmp1], %[dst], 0x00) PTR_ADDU "%[dst], %[dst], %[stride] \n\t" "bnez %[h], 1b \n\t" - : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), + : RESTRICT_ASM_ALL64 + [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]), [ftmp4]"=&f"(ftmp[4]), [ftmp5]"=&f"(ftmp[5]), [ftmp6]"=&f"(ftmp[6]), [ftmp7]"=&f"(ftmp[7]), diff --git a/libavcodec/mips/h264dsp_mmi.c b/libavcodec/mips/h264dsp_mmi.c index 6e77995523..b5ab07c863 100644 --- a/libavcodec/mips/h264dsp_mmi.c +++ b/libavcodec/mips/h264dsp_mmi.c @@ -40,8 +40,8 @@ void ff_h264_add_pixels4_8_mmi(uint8_t *dst, int16_t *src, int stride) MMI_LDC1(%[ftmp3], %[src], 0x10) MMI_LDC1(%[ftmp4], %[src], 0x18) /* memset(src, 0, 32); */ - "gssqc1 %[ftmp0], %[ftmp0], 0x00(%[src]) \n\t" - "gssqc1 %[ftmp0], %[ftmp0], 0x10(%[src]) \n\t" + MMI_SQC1(%[ftmp0], %[ftmp0], %[src], 0x00) + MMI_SQC1(%[ftmp0], %[ftmp0], %[src], 0x10) MMI_ULWC1(%[ftmp5], %[dst0], 0x00) MMI_ULWC1(%[ftmp6], %[dst1], 0x00) MMI_ULWC1(%[ftmp7], %[dst2], 0x00) @@ -90,8 +90,8 @@ void ff_h264_idct_add_8_mmi(uint8_t *dst, int16_t *block, int stride) MMI_LDC1(%[ftmp3], %[block], 0x18) /* memset(block, 0, 32) */ "pxor %[ftmp4], %[ftmp4], %[ftmp4] \n\t" - "gssqc1 %[ftmp4], %[ftmp4], 0x00(%[block]) \n\t" - "gssqc1 %[ftmp4], %[ftmp4], 0x10(%[block]) \n\t" + MMI_SQC1(%[ftmp4], %[ftmp4], %[block], 0x00) + MMI_SQC1(%[ftmp4], %[ftmp4], %[block], 0x10) "dli %[tmp0], 0x01 \n\t" "mtc1 %[tmp0], %[ftmp8] \n\t" "dli %[tmp0], 0x06 \n\t" diff --git a/libavcodec/mips/hevcdsp_mmi.c b/libavcodec/mips/hevcdsp_mmi.c index 87fc2555a4..6583bef5da 100644 --- a/libavcodec/mips/hevcdsp_mmi.c +++ b/libavcodec/mips/hevcdsp_mmi.c @@ -35,6 +35,7 @@ void ff_hevc_put_hevc_qpel_h##w##_8_mmi(int16_t *dst, uint8_t *_src, \ double ftmp[15]; \ uint64_t rtmp[1]; \ const int8_t *filter = ff_hevc_qpel_filters[mx - 1]; \ + DECLARE_VAR_ALL64; \ \ x = x_step; \ y = height; \ @@ -50,14 +51,10 @@ void ff_hevc_put_hevc_qpel_h##w##_8_mmi(int16_t *dst, uint8_t *_src, \ \ "1: \n\t" \ "2: \n\t" \ - "gsldlc1 %[ftmp3], 0x07(%[src]) \n\t" \ - "gsldrc1 %[ftmp3], 0x00(%[src]) \n\t" \ - "gsldlc1 %[ftmp4], 0x08(%[src]) \n\t" \ - "gsldrc1 %[ftmp4], 0x01(%[src]) \n\t" \ - "gsldlc1 %[ftmp5], 0x09(%[src]) \n\t" \ - "gsldrc1 %[ftmp5], 0x02(%[src]) \n\t" \ - "gsldlc1 %[ftmp6], 0x0a(%[src]) \n\t" \ - "gsldrc1 %[ftmp6], 0x03(%[src]) \n\t" \ + MMI_ULDC1(%[ftmp3], %[src], 0x00) \ + MMI_ULDC1(%[ftmp4], %[src], 0x01) \ + MMI_ULDC1(%[ftmp5], %[src], 0x02) \ + MMI_ULDC1(%[ftmp6], %[src], 0x03) \ "punpcklbh %[ftmp7], %[ftmp3], %[ftmp0] \n\t" \ "punpckhbh %[ftmp8], %[ftmp3], %[ftmp0] \n\t" \ "pmullh %[ftmp7], %[ftmp7], %[ftmp1] \n\t" \ @@ -83,8 +80,7 @@ void ff_hevc_put_hevc_qpel_h##w##_8_mmi(int16_t *dst, uint8_t *_src, \ "paddh %[ftmp3], %[ftmp3], %[ftmp4] \n\t" \ "paddh %[ftmp5], %[ftmp5], %[ftmp6] \n\t" \ "paddh %[ftmp3], %[ftmp3], %[ftmp5] \n\t" \ - "gssdlc1 %[ftmp3], 0x07(%[dst]) \n\t" \ - "gssdrc1 %[ftmp3], 0x00(%[dst]) \n\t" \ + MMI_ULDC1(%[ftmp3], %[dst], 0x00) \ \ "daddi %[x], %[x], -0x01 \n\t" \ PTR_ADDIU "%[src], %[src], 0x04 \n\t" \ @@ -98,7 +94,8 @@ void ff_hevc_put_hevc_qpel_h##w##_8_mmi(int16_t *dst, uint8_t *_src, \ PTR_ADDU "%[src], %[src], %[stride] \n\t" \ PTR_ADDIU "%[dst], %[dst], 0x80 \n\t" \ "bnez %[y], 1b \n\t" \ - : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), \ + : RESTRICT_ASM_ALL64 \ + [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), \ [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]), \ [ftmp4]"=&f"(ftmp[4]), [ftmp5]"=&f"(ftmp[5]), \ [ftmp6]"=&f"(ftmp[6]), [ftmp7]"=&f"(ftmp[7]), \ @@ -134,6 +131,7 @@ void ff_hevc_put_hevc_qpel_hv##w##_8_mmi(int16_t *dst, uint8_t *_src, \ int16_t *tmp = tmp_array; \ double ftmp[15]; \ uint64_t rtmp[1]; \ + DECLARE_VAR_ALL64; \ \ src -= (QPEL_EXTRA_BEFORE * srcstride + 3); \ filter = ff_hevc_qpel_filters[mx - 1]; \ @@ -151,14 +149,10 @@ void ff_hevc_put_hevc_qpel_hv##w##_8_mmi(int16_t *dst, uint8_t *_src, \ \ "1: \n\t" \ "2: \n\t" \ - "gsldlc1 %[ftmp3], 0x07(%[src]) \n\t" \ - "gsldrc1 %[ftmp3], 0x00(%[src]) \n\t" \ - "gsldlc1 %[ftmp4], 0x08(%[src]) \n\t" \ - "gsldrc1 %[ftmp4], 0x01(%[src]) \n\t" \ - "gsldlc1 %[ftmp5], 0x09(%[src]) \n\t" \ - "gsldrc1 %[ftmp5], 0x02(%[src]) \n\t" \ - "gsldlc1 %[ftmp6], 0x0a(%[src]) \n\t" \ - "gsldrc1 %[ftmp6], 0x03(%[src]) \n\t" \ + MMI_ULDC1(%[ftmp3], %[src], 0x00) \ + MMI_ULDC1(%[ftmp4], %[src], 0x01) \ + MMI_ULDC1(%[ftmp5], %[src], 0x02) \ + MMI_ULDC1(%[ftmp6], %[src], 0x03) \ "punpcklbh %[ftmp7], %[ftmp3], %[ftmp0] \n\t" \ "punpckhbh %[ftmp8], %[ftmp3], %[ftmp0] \n\t" \ "pmullh %[ftmp7], %[ftmp7], %[ftmp1] \n\t" \ @@ -184,8 +178,7 @@ void ff_hevc_put_hevc_qpel_hv##w##_8_mmi(int16_t *dst, uint8_t *_src, \ "paddh %[ftmp3], %[ftmp3], %[ftmp4] \n\t" \ "paddh %[ftmp5], %[ftmp5], %[ftmp6] \n\t" \ "paddh %[ftmp3], %[ftmp3], %[ftmp5] \n\t" \ - "gssdlc1 %[ftmp3], 0x07(%[tmp]) \n\t" \ - "gssdrc1 %[ftmp3], 0x00(%[tmp]) \n\t" \ + MMI_ULDC1(%[ftmp3], %[tmp], 0x00) \ \ "daddi %[x], %[x], -0x01 \n\t" \ PTR_ADDIU "%[src], %[src], 0x04 \n\t" \ @@ -199,7 +192,8 @@ void ff_hevc_put_hevc_qpel_hv##w##_8_mmi(int16_t *dst, uint8_t *_src, \ PTR_ADDU "%[src], %[src], %[stride] \n\t" \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ "bnez %[y], 1b \n\t" \ - : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), \ + : RESTRICT_ASM_ALL64 \ + [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), \ [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]), \ [ftmp4]"=&f"(ftmp[4]), [ftmp5]"=&f"(ftmp[5]), \ [ftmp6]"=&f"(ftmp[6]), [ftmp7]"=&f"(ftmp[7]), \ @@ -228,29 +222,21 @@ void ff_hevc_put_hevc_qpel_hv##w##_8_mmi(int16_t *dst, uint8_t *_src, \ \ "1: \n\t" \ "2: \n\t" \ - "gsldlc1 %[ftmp3], 0x07(%[tmp]) \n\t" \ - "gsldrc1 %[ftmp3], 0x00(%[tmp]) \n\t" \ + MMI_ULDC1(%[ftmp3], %[tmp], 0x00) \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ - "gsldlc1 %[ftmp4], 0x07(%[tmp]) \n\t" \ - "gsldrc1 %[ftmp4], 0x00(%[tmp]) \n\t" \ + MMI_ULDC1(%[ftmp4], %[tmp], 0x00) \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ - "gsldlc1 %[ftmp5], 0x07(%[tmp]) \n\t" \ - "gsldrc1 %[ftmp5], 0x00(%[tmp]) \n\t" \ + MMI_ULDC1(%[ftmp5], %[tmp], 0x00) \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ - "gsldlc1 %[ftmp6], 0x07(%[tmp]) \n\t" \ - "gsldrc1 %[ftmp6], 0x00(%[tmp]) \n\t" \ + MMI_ULDC1(%[ftmp6], %[tmp], 0x00) \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ - "gsldlc1 %[ftmp7], 0x07(%[tmp]) \n\t" \ - "gsldrc1 %[ftmp7], 0x00(%[tmp]) \n\t" \ + MMI_ULDC1(%[ftmp7], %[tmp], 0x00) \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ - "gsldlc1 %[ftmp8], 0x07(%[tmp]) \n\t" \ - "gsldrc1 %[ftmp8], 0x00(%[tmp]) \n\t" \ + MMI_ULDC1(%[ftmp8], %[tmp], 0x00) \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ - "gsldlc1 %[ftmp9], 0x07(%[tmp]) \n\t" \ - "gsldrc1 %[ftmp9], 0x00(%[tmp]) \n\t" \ + MMI_ULDC1(%[ftmp9], %[tmp], 0x00) \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ - "gsldlc1 %[ftmp10], 0x07(%[tmp]) \n\t" \ - "gsldrc1 %[ftmp10], 0x00(%[tmp]) \n\t" \ + MMI_ULDC1(%[ftmp10], %[tmp], 0x00) \ PTR_ADDIU "%[tmp], %[tmp], -0x380 \n\t" \ TRANSPOSE_4H(%[ftmp3], %[ftmp4], %[ftmp5], %[ftmp6], \ %[ftmp11], %[ftmp12], %[ftmp13], %[ftmp14]) \ @@ -275,8 +261,7 @@ void ff_hevc_put_hevc_qpel_hv##w##_8_mmi(int16_t *dst, uint8_t *_src, \ "paddw %[ftmp5], %[ftmp5], %[ftmp6] \n\t" \ "psraw %[ftmp5], %[ftmp5], %[ftmp0] \n\t" \ "packsswh %[ftmp3], %[ftmp3], %[ftmp5] \n\t" \ - "gssdlc1 %[ftmp3], 0x07(%[dst]) \n\t" \ - "gssdrc1 %[ftmp3], 0x00(%[dst]) \n\t" \ + MMI_USDC1(%[ftmp3], %[dst], 0x00) \ \ "daddi %[x], %[x], -0x01 \n\t" \ PTR_ADDIU "%[dst], %[dst], 0x08 \n\t" \ @@ -290,7 +275,8 @@ void ff_hevc_put_hevc_qpel_hv##w##_8_mmi(int16_t *dst, uint8_t *_src, \ PTR_ADDIU "%[dst], %[dst], 0x80 \n\t" \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ "bnez %[y], 1b \n\t" \ - : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), \ + : RESTRICT_ASM_ALL64 \ + [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), \ [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]), \ [ftmp4]"=&f"(ftmp[4]), [ftmp5]"=&f"(ftmp[5]), \ [ftmp6]"=&f"(ftmp[6]), [ftmp7]"=&f"(ftmp[7]), \ @@ -333,6 +319,8 @@ void ff_hevc_put_hevc_qpel_bi_h##w##_8_mmi(uint8_t *_dst, \ uint64_t rtmp[1]; \ union av_intfloat64 shift; \ union av_intfloat64 offset; \ + DECLARE_VAR_ALL64; \ + DECLARE_VAR_LOW32; \ shift.i = 7; \ offset.i = 64; \ \ @@ -353,14 +341,10 @@ void ff_hevc_put_hevc_qpel_bi_h##w##_8_mmi(uint8_t *_dst, \ "1: \n\t" \ "li %[x], " #x_step " \n\t" \ "2: \n\t" \ - "gsldlc1 %[ftmp3], 0x07(%[src]) \n\t" \ - "gsldrc1 %[ftmp3], 0x00(%[src]) \n\t" \ - "gsldlc1 %[ftmp4], 0x08(%[src]) \n\t" \ - "gsldrc1 %[ftmp4], 0x01(%[src]) \n\t" \ - "gsldlc1 %[ftmp5], 0x09(%[src]) \n\t" \ - "gsldrc1 %[ftmp5], 0x02(%[src]) \n\t" \ - "gsldlc1 %[ftmp6], 0x0a(%[src]) \n\t" \ - "gsldrc1 %[ftmp6], 0x03(%[src]) \n\t" \ + MMI_ULDC1(%[ftmp3], %[src], 0x00) \ + MMI_ULDC1(%[ftmp4], %[src], 0x01) \ + MMI_ULDC1(%[ftmp5], %[src], 0x02) \ + MMI_ULDC1(%[ftmp6], %[src], 0x03) \ "punpcklbh %[ftmp7], %[ftmp3], %[ftmp0] \n\t" \ "punpckhbh %[ftmp8], %[ftmp3], %[ftmp0] \n\t" \ "pmullh %[ftmp7], %[ftmp7], %[ftmp1] \n\t" \ @@ -387,8 +371,7 @@ void ff_hevc_put_hevc_qpel_bi_h##w##_8_mmi(uint8_t *_dst, \ "paddh %[ftmp5], %[ftmp5], %[ftmp6] \n\t" \ "paddh %[ftmp3], %[ftmp3], %[ftmp5] \n\t" \ "paddh %[ftmp3], %[ftmp3], %[offset] \n\t" \ - "gsldlc1 %[ftmp4], 0x07(%[src2]) \n\t" \ - "gsldrc1 %[ftmp4], 0x00(%[src2]) \n\t" \ + MMI_ULDC1(%[ftmp4], %[src2], 0x00) \ "li %[rtmp0], 0x10 \n\t" \ "dmtc1 %[rtmp0], %[ftmp8] \n\t" \ "punpcklhw %[ftmp5], %[ftmp0], %[ftmp3] \n\t" \ @@ -407,8 +390,7 @@ void ff_hevc_put_hevc_qpel_bi_h##w##_8_mmi(uint8_t *_dst, \ "pcmpgth %[ftmp7], %[ftmp5], %[ftmp0] \n\t" \ "pand %[ftmp3], %[ftmp5], %[ftmp7] \n\t" \ "packushb %[ftmp3], %[ftmp3], %[ftmp3] \n\t" \ - "gsswlc1 %[ftmp3], 0x03(%[dst]) \n\t" \ - "gsswrc1 %[ftmp3], 0x00(%[dst]) \n\t" \ + MMI_USWC1(%[ftmp3], %[dst], 0x00) \ \ "daddi %[x], %[x], -0x01 \n\t" \ PTR_ADDIU "%[src], %[src], 0x04 \n\t" \ @@ -424,7 +406,8 @@ void ff_hevc_put_hevc_qpel_bi_h##w##_8_mmi(uint8_t *_dst, \ PTR_ADDU "%[dst], %[dst], %[dst_stride] \n\t" \ PTR_ADDIU "%[src2], %[src2], 0x80 \n\t" \ "bnez %[y], 1b \n\t" \ - : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), \ + : RESTRICT_ASM_ALL64 RESTRICT_ASM_LOW32 \ + [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), \ [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]), \ [ftmp4]"=&f"(ftmp[4]), [ftmp5]"=&f"(ftmp[5]), \ [ftmp6]"=&f"(ftmp[6]), [ftmp7]"=&f"(ftmp[7]), \ @@ -469,6 +452,8 @@ void ff_hevc_put_hevc_qpel_bi_hv##w##_8_mmi(uint8_t *_dst, \ uint64_t rtmp[1]; \ union av_intfloat64 shift; \ union av_intfloat64 offset; \ + DECLARE_VAR_ALL64; \ + DECLARE_VAR_LOW32; \ shift.i = 7; \ offset.i = 64; \ \ @@ -488,14 +473,10 @@ void ff_hevc_put_hevc_qpel_bi_hv##w##_8_mmi(uint8_t *_dst, \ \ "1: \n\t" \ "2: \n\t" \ - "gsldlc1 %[ftmp3], 0x07(%[src]) \n\t" \ - "gsldrc1 %[ftmp3], 0x00(%[src]) \n\t" \ - "gsldlc1 %[ftmp4], 0x08(%[src]) \n\t" \ - "gsldrc1 %[ftmp4], 0x01(%[src]) \n\t" \ - "gsldlc1 %[ftmp5], 0x09(%[src]) \n\t" \ - "gsldrc1 %[ftmp5], 0x02(%[src]) \n\t" \ - "gsldlc1 %[ftmp6], 0x0a(%[src]) \n\t" \ - "gsldrc1 %[ftmp6], 0x03(%[src]) \n\t" \ + MMI_ULDC1(%[ftmp3], %[src], 0x00) \ + MMI_ULDC1(%[ftmp4], %[src], 0x01) \ + MMI_ULDC1(%[ftmp5], %[src], 0x02) \ + MMI_ULDC1(%[ftmp6], %[src], 0x03) \ "punpcklbh %[ftmp7], %[ftmp3], %[ftmp0] \n\t" \ "punpckhbh %[ftmp8], %[ftmp3], %[ftmp0] \n\t" \ "pmullh %[ftmp7], %[ftmp7], %[ftmp1] \n\t" \ @@ -521,8 +502,7 @@ void ff_hevc_put_hevc_qpel_bi_hv##w##_8_mmi(uint8_t *_dst, \ "paddh %[ftmp3], %[ftmp3], %[ftmp4] \n\t" \ "paddh %[ftmp5], %[ftmp5], %[ftmp6] \n\t" \ "paddh %[ftmp3], %[ftmp3], %[ftmp5] \n\t" \ - "gssdlc1 %[ftmp3], 0x07(%[tmp]) \n\t" \ - "gssdrc1 %[ftmp3], 0x00(%[tmp]) \n\t" \ + MMI_USDC1(%[ftmp3], %[tmp], 0x00) \ \ "daddi %[x], %[x], -0x01 \n\t" \ PTR_ADDIU "%[src], %[src], 0x04 \n\t" \ @@ -536,7 +516,8 @@ void ff_hevc_put_hevc_qpel_bi_hv##w##_8_mmi(uint8_t *_dst, \ PTR_ADDU "%[src], %[src], %[stride] \n\t" \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ "bnez %[y], 1b \n\t" \ - : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), \ + : RESTRICT_ASM_ALL64 \ + [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), \ [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]), \ [ftmp4]"=&f"(ftmp[4]), [ftmp5]"=&f"(ftmp[5]), \ [ftmp6]"=&f"(ftmp[6]), [ftmp7]"=&f"(ftmp[7]), \ @@ -567,29 +548,21 @@ void ff_hevc_put_hevc_qpel_bi_hv##w##_8_mmi(uint8_t *_dst, \ "1: \n\t" \ "li %[x], " #x_step " \n\t" \ "2: \n\t" \ - "gsldlc1 %[ftmp3], 0x07(%[tmp]) \n\t" \ - "gsldrc1 %[ftmp3], 0x00(%[tmp]) \n\t" \ + MMI_ULDC1(%[ftmp3], %[tmp], 0x00) \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ - "gsldlc1 %[ftmp4], 0x07(%[tmp]) \n\t" \ - "gsldrc1 %[ftmp4], 0x00(%[tmp]) \n\t" \ + MMI_ULDC1(%[ftmp4], %[tmp], 0x00) \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ - "gsldlc1 %[ftmp5], 0x07(%[tmp]) \n\t" \ - "gsldrc1 %[ftmp5], 0x00(%[tmp]) \n\t" \ + MMI_ULDC1(%[ftmp5], %[tmp], 0x00) \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ - "gsldlc1 %[ftmp6], 0x07(%[tmp]) \n\t" \ - "gsldrc1 %[ftmp6], 0x00(%[tmp]) \n\t" \ + MMI_ULDC1(%[ftmp6], %[tmp], 0x00) \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ - "gsldlc1 %[ftmp7], 0x07(%[tmp]) \n\t" \ - "gsldrc1 %[ftmp7], 0x00(%[tmp]) \n\t" \ + MMI_ULDC1(%[ftmp7], %[tmp], 0x00) \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ - "gsldlc1 %[ftmp8], 0x07(%[tmp]) \n\t" \ - "gsldrc1 %[ftmp8], 0x00(%[tmp]) \n\t" \ + MMI_ULDC1(%[ftmp8], %[tmp], 0x00) \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ - "gsldlc1 %[ftmp9], 0x07(%[tmp]) \n\t" \ - "gsldrc1 %[ftmp9], 0x00(%[tmp]) \n\t" \ + MMI_ULDC1(%[ftmp9], %[tmp], 0x00) \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ - "gsldlc1 %[ftmp10], 0x07(%[tmp]) \n\t" \ - "gsldrc1 %[ftmp10], 0x00(%[tmp]) \n\t" \ + MMI_ULDC1(%[ftmp10], %[tmp], 0x00) \ PTR_ADDIU "%[tmp], %[tmp], -0x380 \n\t" \ TRANSPOSE_4H(%[ftmp3], %[ftmp4], %[ftmp5], %[ftmp6], \ %[ftmp11], %[ftmp12], %[ftmp13], %[ftmp14]) \ @@ -614,8 +587,7 @@ void ff_hevc_put_hevc_qpel_bi_hv##w##_8_mmi(uint8_t *_dst, \ "paddw %[ftmp5], %[ftmp5], %[ftmp6] \n\t" \ "psraw %[ftmp5], %[ftmp5], %[ftmp0] \n\t" \ "packsswh %[ftmp3], %[ftmp3], %[ftmp5] \n\t" \ - "gsldlc1 %[ftmp4], 0x07(%[src2]) \n\t" \ - "gsldrc1 %[ftmp4], 0x00(%[src2]) \n\t" \ + MMI_ULDC1(%[ftmp4], %[src2], 0x00) \ "pxor %[ftmp7], %[ftmp7], %[ftmp7] \n\t" \ "li %[rtmp0], 0x10 \n\t" \ "dmtc1 %[rtmp0], %[ftmp8] \n\t" \ @@ -637,8 +609,7 @@ void ff_hevc_put_hevc_qpel_bi_hv##w##_8_mmi(uint8_t *_dst, \ "pcmpgth %[ftmp7], %[ftmp5], %[ftmp7] \n\t" \ "pand %[ftmp3], %[ftmp5], %[ftmp7] \n\t" \ "packushb %[ftmp3], %[ftmp3], %[ftmp3] \n\t" \ - "gsswlc1 %[ftmp3], 0x03(%[dst]) \n\t" \ - "gsswrc1 %[ftmp3], 0x00(%[dst]) \n\t" \ + MMI_USWC1(%[ftmp3], %[dst], 0x00) \ \ "daddi %[x], %[x], -0x01 \n\t" \ PTR_ADDIU "%[src2], %[src2], 0x08 \n\t" \ @@ -654,7 +625,8 @@ void ff_hevc_put_hevc_qpel_bi_hv##w##_8_mmi(uint8_t *_dst, \ PTR_ADDU "%[dst], %[dst], %[stride] \n\t" \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ "bnez %[y], 1b \n\t" \ - : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), \ + : RESTRICT_ASM_ALL64 RESTRICT_ASM_LOW32 \ + [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), \ [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]), \ [ftmp4]"=&f"(ftmp[4]), [ftmp5]"=&f"(ftmp[5]), \ [ftmp6]"=&f"(ftmp[6]), [ftmp7]"=&f"(ftmp[7]), \ @@ -700,6 +672,8 @@ void ff_hevc_put_hevc_epel_bi_hv##w##_8_mmi(uint8_t *_dst, \ uint64_t rtmp[1]; \ union av_intfloat64 shift; \ union av_intfloat64 offset; \ + DECLARE_VAR_ALL64; \ + DECLARE_VAR_LOW32; \ shift.i = 7; \ offset.i = 64; \ \ @@ -716,14 +690,10 @@ void ff_hevc_put_hevc_epel_bi_hv##w##_8_mmi(uint8_t *_dst, \ \ "1: \n\t" \ "2: \n\t" \ - "gslwlc1 %[ftmp2], 0x03(%[src]) \n\t" \ - "gslwrc1 %[ftmp2], 0x00(%[src]) \n\t" \ - "gslwlc1 %[ftmp3], 0x04(%[src]) \n\t" \ - "gslwrc1 %[ftmp3], 0x01(%[src]) \n\t" \ - "gslwlc1 %[ftmp4], 0x05(%[src]) \n\t" \ - "gslwrc1 %[ftmp4], 0x02(%[src]) \n\t" \ - "gslwlc1 %[ftmp5], 0x06(%[src]) \n\t" \ - "gslwrc1 %[ftmp5], 0x03(%[src]) \n\t" \ + MMI_ULDC1(%[ftmp3], %[src], 0x00) \ + MMI_ULDC1(%[ftmp4], %[src], 0x01) \ + MMI_ULDC1(%[ftmp5], %[src], 0x02) \ + MMI_ULDC1(%[ftmp6], %[src], 0x03) \ "punpcklbh %[ftmp2], %[ftmp2], %[ftmp0] \n\t" \ "pmullh %[ftmp2], %[ftmp2], %[ftmp1] \n\t" \ "punpcklbh %[ftmp3], %[ftmp3], %[ftmp0] \n\t" \ @@ -737,8 +707,7 @@ void ff_hevc_put_hevc_epel_bi_hv##w##_8_mmi(uint8_t *_dst, \ "paddh %[ftmp2], %[ftmp2], %[ftmp3] \n\t" \ "paddh %[ftmp4], %[ftmp4], %[ftmp5] \n\t" \ "paddh %[ftmp2], %[ftmp2], %[ftmp4] \n\t" \ - "gssdlc1 %[ftmp2], 0x07(%[tmp]) \n\t" \ - "gssdrc1 %[ftmp2], 0x00(%[tmp]) \n\t" \ + MMI_ULDC1(%[ftmp2], %[tmp], 0x00) \ \ "daddi %[x], %[x], -0x01 \n\t" \ PTR_ADDIU "%[src], %[src], 0x04 \n\t" \ @@ -752,7 +721,8 @@ void ff_hevc_put_hevc_epel_bi_hv##w##_8_mmi(uint8_t *_dst, \ PTR_ADDU "%[src], %[src], %[stride] \n\t" \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ "bnez %[y], 1b \n\t" \ - : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), \ + : RESTRICT_ASM_ALL64 \ + [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), \ [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]), \ [ftmp4]"=&f"(ftmp[4]), [ftmp5]"=&f"(ftmp[5]), \ [ftmp6]"=&f"(ftmp[6]), [ftmp7]"=&f"(ftmp[7]), \ @@ -782,17 +752,13 @@ void ff_hevc_put_hevc_epel_bi_hv##w##_8_mmi(uint8_t *_dst, \ "1: \n\t" \ "li %[x], " #x_step " \n\t" \ "2: \n\t" \ - "gsldlc1 %[ftmp3], 0x07(%[tmp]) \n\t" \ - "gsldrc1 %[ftmp3], 0x00(%[tmp]) \n\t" \ + MMI_ULDC1(%[ftmp3], %[tmp], 0x00) \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ - "gsldlc1 %[ftmp4], 0x07(%[tmp]) \n\t" \ - "gsldrc1 %[ftmp4], 0x00(%[tmp]) \n\t" \ + MMI_ULDC1(%[ftmp4], %[tmp], 0x00) \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ - "gsldlc1 %[ftmp5], 0x07(%[tmp]) \n\t" \ - "gsldrc1 %[ftmp5], 0x00(%[tmp]) \n\t" \ + MMI_ULDC1(%[ftmp5], %[tmp], 0x00) \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ - "gsldlc1 %[ftmp6], 0x07(%[tmp]) \n\t" \ - "gsldrc1 %[ftmp6], 0x00(%[tmp]) \n\t" \ + MMI_ULDC1(%[ftmp6], %[tmp], 0x00) \ PTR_ADDIU "%[tmp], %[tmp], -0x180 \n\t" \ TRANSPOSE_4H(%[ftmp3], %[ftmp4], %[ftmp5], %[ftmp6], \ %[ftmp7], %[ftmp8], %[ftmp9], %[ftmp10]) \ @@ -807,8 +773,7 @@ void ff_hevc_put_hevc_epel_bi_hv##w##_8_mmi(uint8_t *_dst, \ "paddw %[ftmp5], %[ftmp5], %[ftmp6] \n\t" \ "psraw %[ftmp5], %[ftmp5], %[ftmp0] \n\t" \ "packsswh %[ftmp3], %[ftmp3], %[ftmp5] \n\t" \ - "gsldlc1 %[ftmp4], 0x07(%[src2]) \n\t" \ - "gsldrc1 %[ftmp4], 0x00(%[src2]) \n\t" \ + MMI_ULDC1(%[ftmp4], %[tmp], 0x02) \ "li %[rtmp0], 0x10 \n\t" \ "dmtc1 %[rtmp0], %[ftmp8] \n\t" \ "punpcklhw %[ftmp5], %[ftmp2], %[ftmp3] \n\t" \ @@ -829,8 +794,7 @@ void ff_hevc_put_hevc_epel_bi_hv##w##_8_mmi(uint8_t *_dst, \ "pcmpgth %[ftmp7], %[ftmp5], %[ftmp2] \n\t" \ "pand %[ftmp3], %[ftmp5], %[ftmp7] \n\t" \ "packushb %[ftmp3], %[ftmp3], %[ftmp3] \n\t" \ - "gsswlc1 %[ftmp3], 0x03(%[dst]) \n\t" \ - "gsswrc1 %[ftmp3], 0x00(%[dst]) \n\t" \ + MMI_USWC1(%[ftmp3], %[dst], 0x0) \ \ "daddi %[x], %[x], -0x01 \n\t" \ PTR_ADDIU "%[src2], %[src2], 0x08 \n\t" \ @@ -846,7 +810,8 @@ void ff_hevc_put_hevc_epel_bi_hv##w##_8_mmi(uint8_t *_dst, \ PTR_ADDU "%[dst], %[dst], %[stride] \n\t" \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ "bnez %[y], 1b \n\t" \ - : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), \ + : RESTRICT_ASM_LOW32 RESTRICT_ASM_ALL64 \ + [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), \ [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]), \ [ftmp4]"=&f"(ftmp[4]), [ftmp5]"=&f"(ftmp[5]), \ [ftmp6]"=&f"(ftmp[6]), [ftmp7]"=&f"(ftmp[7]), \ @@ -884,6 +849,7 @@ void ff_hevc_put_hevc_pel_bi_pixels##w##_8_mmi(uint8_t *_dst, \ double ftmp[12]; \ uint64_t rtmp[1]; \ union av_intfloat64 shift; \ + DECLARE_VAR_ALL64; \ shift.i = 7; \ \ y = height; \ @@ -901,12 +867,9 @@ void ff_hevc_put_hevc_pel_bi_pixels##w##_8_mmi(uint8_t *_dst, \ \ "1: \n\t" \ "2: \n\t" \ - "gsldlc1 %[ftmp5], 0x07(%[src]) \n\t" \ - "gsldrc1 %[ftmp5], 0x00(%[src]) \n\t" \ - "gsldlc1 %[ftmp2], 0x07(%[src2]) \n\t" \ - "gsldrc1 %[ftmp2], 0x00(%[src2]) \n\t" \ - "gsldlc1 %[ftmp3], 0x0f(%[src2]) \n\t" \ - "gsldrc1 %[ftmp3], 0x08(%[src2]) \n\t" \ + MMI_ULDC1(%[ftmp5], %[src], 0x00) \ + MMI_ULDC1(%[ftmp2], %[src2], 0x00) \ + MMI_ULDC1(%[ftmp3], %[src2], 0x08) \ "punpcklbh %[ftmp4], %[ftmp5], %[ftmp0] \n\t" \ "punpckhbh %[ftmp5], %[ftmp5], %[ftmp0] \n\t" \ "psllh %[ftmp4], %[ftmp4], %[ftmp1] \n\t" \ @@ -940,8 +903,7 @@ void ff_hevc_put_hevc_pel_bi_pixels##w##_8_mmi(uint8_t *_dst, \ "pand %[ftmp2], %[ftmp2], %[ftmp3] \n\t" \ "pand %[ftmp4], %[ftmp4], %[ftmp5] \n\t" \ "packushb %[ftmp2], %[ftmp2], %[ftmp4] \n\t" \ - "gssdlc1 %[ftmp2], 0x07(%[dst]) \n\t" \ - "gssdrc1 %[ftmp2], 0x00(%[dst]) \n\t" \ + MMI_USDC1(%[ftmp2], %[dst], 0x0) \ \ "daddi %[x], %[x], -0x01 \n\t" \ PTR_ADDIU "%[src], %[src], 0x08 \n\t" \ @@ -958,7 +920,8 @@ void ff_hevc_put_hevc_pel_bi_pixels##w##_8_mmi(uint8_t *_dst, \ PTR_ADDU "%[dst], %[dst], %[dststride] \n\t" \ PTR_ADDIU "%[src2], %[src2], 0x80 \n\t" \ "bnez %[y], 1b \n\t" \ - : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), \ + : RESTRICT_ASM_ALL64 \ + [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), \ [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]), \ [ftmp4]"=&f"(ftmp[4]), [ftmp5]"=&f"(ftmp[5]), \ [ftmp6]"=&f"(ftmp[6]), [ftmp7]"=&f"(ftmp[7]), \ @@ -1000,6 +963,8 @@ void ff_hevc_put_hevc_qpel_uni_hv##w##_8_mmi(uint8_t *_dst, \ uint64_t rtmp[1]; \ union av_intfloat64 shift; \ union av_intfloat64 offset; \ + DECLARE_VAR_ALL64; \ + DECLARE_VAR_LOW32; \ shift.i = 6; \ offset.i = 32; \ \ @@ -1019,14 +984,10 @@ void ff_hevc_put_hevc_qpel_uni_hv##w##_8_mmi(uint8_t *_dst, \ \ "1: \n\t" \ "2: \n\t" \ - "gsldlc1 %[ftmp3], 0x07(%[src]) \n\t" \ - "gsldrc1 %[ftmp3], 0x00(%[src]) \n\t" \ - "gsldlc1 %[ftmp4], 0x08(%[src]) \n\t" \ - "gsldrc1 %[ftmp4], 0x01(%[src]) \n\t" \ - "gsldlc1 %[ftmp5], 0x09(%[src]) \n\t" \ - "gsldrc1 %[ftmp5], 0x02(%[src]) \n\t" \ - "gsldlc1 %[ftmp6], 0x0a(%[src]) \n\t" \ - "gsldrc1 %[ftmp6], 0x03(%[src]) \n\t" \ + MMI_ULDC1(%[ftmp3], %[src], 0x00) \ + MMI_ULDC1(%[ftmp4], %[src], 0x01) \ + MMI_ULDC1(%[ftmp5], %[src], 0x02) \ + MMI_ULDC1(%[ftmp6], %[src], 0x03) \ "punpcklbh %[ftmp7], %[ftmp3], %[ftmp0] \n\t" \ "punpckhbh %[ftmp8], %[ftmp3], %[ftmp0] \n\t" \ "pmullh %[ftmp7], %[ftmp7], %[ftmp1] \n\t" \ @@ -1052,8 +1013,7 @@ void ff_hevc_put_hevc_qpel_uni_hv##w##_8_mmi(uint8_t *_dst, \ "paddh %[ftmp3], %[ftmp3], %[ftmp4] \n\t" \ "paddh %[ftmp5], %[ftmp5], %[ftmp6] \n\t" \ "paddh %[ftmp3], %[ftmp3], %[ftmp5] \n\t" \ - "gssdlc1 %[ftmp3], 0x07(%[tmp]) \n\t" \ - "gssdrc1 %[ftmp3], 0x00(%[tmp]) \n\t" \ + MMI_USDC1(%[ftmp3], %[tmp], 0x0) \ \ "daddi %[x], %[x], -0x01 \n\t" \ PTR_ADDIU "%[src], %[src], 0x04 \n\t" \ @@ -1067,7 +1027,8 @@ void ff_hevc_put_hevc_qpel_uni_hv##w##_8_mmi(uint8_t *_dst, \ PTR_ADDU "%[src], %[src], %[stride] \n\t" \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ "bnez %[y], 1b \n\t" \ - : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), \ + : RESTRICT_ASM_ALL64 \ + [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), \ [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]), \ [ftmp4]"=&f"(ftmp[4]), [ftmp5]"=&f"(ftmp[5]), \ [ftmp6]"=&f"(ftmp[6]), [ftmp7]"=&f"(ftmp[7]), \ @@ -1099,29 +1060,21 @@ void ff_hevc_put_hevc_qpel_uni_hv##w##_8_mmi(uint8_t *_dst, \ "1: \n\t" \ "li %[x], " #x_step " \n\t" \ "2: \n\t" \ - "gsldlc1 %[ftmp3], 0x07(%[tmp]) \n\t" \ - "gsldrc1 %[ftmp3], 0x00(%[tmp]) \n\t" \ + MMI_ULDC1(%[ftmp3], %[tmp], 0x00) \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ - "gsldlc1 %[ftmp4], 0x07(%[tmp]) \n\t" \ - "gsldrc1 %[ftmp4], 0x00(%[tmp]) \n\t" \ + MMI_ULDC1(%[ftmp4], %[tmp], 0x00) \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ - "gsldlc1 %[ftmp5], 0x07(%[tmp]) \n\t" \ - "gsldrc1 %[ftmp5], 0x00(%[tmp]) \n\t" \ + MMI_ULDC1(%[ftmp5], %[tmp], 0x00) \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ - "gsldlc1 %[ftmp6], 0x07(%[tmp]) \n\t" \ - "gsldrc1 %[ftmp6], 0x00(%[tmp]) \n\t" \ + MMI_ULDC1(%[ftmp6], %[tmp], 0x00) \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ - "gsldlc1 %[ftmp7], 0x07(%[tmp]) \n\t" \ - "gsldrc1 %[ftmp7], 0x00(%[tmp]) \n\t" \ + MMI_ULDC1(%[ftmp7], %[tmp], 0x00) \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ - "gsldlc1 %[ftmp8], 0x07(%[tmp]) \n\t" \ - "gsldrc1 %[ftmp8], 0x00(%[tmp]) \n\t" \ + MMI_ULDC1(%[ftmp8], %[tmp], 0x00) \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ - "gsldlc1 %[ftmp9], 0x07(%[tmp]) \n\t" \ - "gsldrc1 %[ftmp9], 0x00(%[tmp]) \n\t" \ + MMI_ULDC1(%[ftmp9], %[tmp], 0x00) \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ - "gsldlc1 %[ftmp10], 0x07(%[tmp]) \n\t" \ - "gsldrc1 %[ftmp10], 0x00(%[tmp]) \n\t" \ + MMI_ULDC1(%[ftmp10], %[tmp], 0x00) \ PTR_ADDIU "%[tmp], %[tmp], -0x380 \n\t" \ TRANSPOSE_4H(%[ftmp3], %[ftmp4], %[ftmp5], %[ftmp6], \ %[ftmp11], %[ftmp12], %[ftmp13], %[ftmp14]) \ @@ -1152,8 +1105,7 @@ void ff_hevc_put_hevc_qpel_uni_hv##w##_8_mmi(uint8_t *_dst, \ "pcmpgth %[ftmp7], %[ftmp3], %[ftmp7] \n\t" \ "pand %[ftmp3], %[ftmp3], %[ftmp7] \n\t" \ "packushb %[ftmp3], %[ftmp3], %[ftmp3] \n\t" \ - "gsswlc1 %[ftmp3], 0x03(%[dst]) \n\t" \ - "gsswrc1 %[ftmp3], 0x00(%[dst]) \n\t" \ + MMI_USWC1(%[ftmp3], %[dst], 0x00) \ \ "daddi %[x], %[x], -0x01 \n\t" \ PTR_ADDIU "%[tmp], %[tmp], 0x08 \n\t" \ @@ -1166,7 +1118,8 @@ void ff_hevc_put_hevc_qpel_uni_hv##w##_8_mmi(uint8_t *_dst, \ PTR_ADDU "%[dst], %[dst], %[stride] \n\t" \ PTR_ADDIU "%[tmp], %[tmp], 0x80 \n\t" \ "bnez %[y], 1b \n\t" \ - : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), \ + : RESTRICT_ASM_ALL64 RESTRICT_ASM_LOW32 \ + [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), \ [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]), \ [ftmp4]"=&f"(ftmp[4]), [ftmp5]"=&f"(ftmp[5]), \ [ftmp6]"=&f"(ftmp[6]), [ftmp7]"=&f"(ftmp[7]), \ diff --git a/libavcodec/mips/hpeldsp_mmi.c b/libavcodec/mips/hpeldsp_mmi.c index bf3e4636aa..8e9c0fa821 100644 --- a/libavcodec/mips/hpeldsp_mmi.c +++ b/libavcodec/mips/hpeldsp_mmi.c @@ -307,6 +307,7 @@ inline void ff_put_pixels4_l2_8_mmi(uint8_t *dst, const uint8_t *src1, double ftmp[4]; mips_reg addr[5]; DECLARE_VAR_LOW32; + DECLARE_VAR_ADDRT; __asm__ volatile ( "1: \n\t" diff --git a/libavcodec/mips/simple_idct_mmi.c b/libavcodec/mips/simple_idct_mmi.c index ad068a8251..4680520edc 100644 --- a/libavcodec/mips/simple_idct_mmi.c +++ b/libavcodec/mips/simple_idct_mmi.c @@ -56,6 +56,8 @@ DECLARE_ALIGNED(16, const int16_t, W_arr)[46] = { void ff_simple_idct_8_mmi(int16_t *block) { + DECLARE_VAR_ALL64; + BACKUP_REG __asm__ volatile ( @@ -142,20 +144,20 @@ void ff_simple_idct_8_mmi(int16_t *block) /* idctRowCondDC row0~8 */ /* load W */ - "gslqc1 $f19, $f18, 0x00(%[w_arr]) \n\t" - "gslqc1 $f21, $f20, 0x10(%[w_arr]) \n\t" - "gslqc1 $f23, $f22, 0x20(%[w_arr]) \n\t" - "gslqc1 $f25, $f24, 0x30(%[w_arr]) \n\t" - "gslqc1 $f17, $f16, 0x40(%[w_arr]) \n\t" + MMI_LQC1($f19, $f18, %[w_arr], 0x00) + MMI_LQC1($f21, $f20, %[w_arr], 0x10) + MMI_LQC1($f23, $f22, %[w_arr], 0x20) + MMI_LQC1($f25, $f24, %[w_arr], 0x30) + MMI_LQC1($f17, $f16, %[w_arr], 0x40) /* load source in block */ - "gslqc1 $f1, $f0, 0x00(%[block]) \n\t" - "gslqc1 $f3, $f2, 0x10(%[block]) \n\t" - "gslqc1 $f5, $f4, 0x20(%[block]) \n\t" - "gslqc1 $f7, $f6, 0x30(%[block]) \n\t" - "gslqc1 $f9, $f8, 0x40(%[block]) \n\t" - "gslqc1 $f11, $f10, 0x50(%[block]) \n\t" - "gslqc1 $f13, $f12, 0x60(%[block]) \n\t" - "gslqc1 $f15, $f14, 0x70(%[block]) \n\t" + MMI_LQC1($f1, $f0, %[block], 0x00) + MMI_LQC1($f3, $f2, %[block], 0x10) + MMI_LQC1($f5, $f4, %[block], 0x20) + MMI_LQC1($f7, $f6, %[block], 0x30) + MMI_LQC1($f9, $f8, %[block], 0x40) + MMI_LQC1($f11, $f10, %[block], 0x50) + MMI_LQC1($f13, $f12, %[block], 0x60) + MMI_LQC1($f15, $f14, %[block], 0x70) /* $9: mask ; $f17: ROW_SHIFT */ "dmfc1 $9, $f17 \n\t" @@ -253,8 +255,7 @@ void ff_simple_idct_8_mmi(int16_t *block) /* idctSparseCol col0~3 */ /* $f17: ff_p16_32; $f16: COL_SHIFT-16 */ - "gsldlc1 $f17, 0x57(%[w_arr]) \n\t" - "gsldrc1 $f17, 0x50(%[w_arr]) \n\t" + MMI_ULDC1($f17, %[w_arr], 0x50) "li $10, 4 \n\t" "dmtc1 $10, $f16 \n\t" "paddh $f0, $f0, $f17 \n\t" @@ -395,16 +396,16 @@ void ff_simple_idct_8_mmi(int16_t *block) "punpcklwd $f11, $f27, $f29 \n\t" "punpckhwd $f15, $f27, $f29 \n\t" /* Store */ - "gssqc1 $f1, $f0, 0x00(%[block]) \n\t" - "gssqc1 $f5, $f4, 0x10(%[block]) \n\t" - "gssqc1 $f9, $f8, 0x20(%[block]) \n\t" - "gssqc1 $f13, $f12, 0x30(%[block]) \n\t" - "gssqc1 $f3, $f2, 0x40(%[block]) \n\t" - "gssqc1 $f7, $f6, 0x50(%[block]) \n\t" - "gssqc1 $f11, $f10, 0x60(%[block]) \n\t" - "gssqc1 $f15, $f14, 0x70(%[block]) \n\t" + MMI_SQC1($f1, $f0, %[block], 0x00) + MMI_SQC1($f5, $f4, %[block], 0x10) + MMI_SQC1($f9, $f8, %[block], 0x20) + MMI_SQC1($f13, $f12, %[block], 0x30) + MMI_SQC1($f3, $f2, %[block], 0x40) + MMI_SQC1($f7, $f6, %[block], 0x50) + MMI_SQC1($f11, $f10, %[block], 0x60) + MMI_SQC1($f15, $f14, %[block], 0x70) - : [block]"+&r"(block) + : RESTRICT_ASM_ALL64 [block]"+&r"(block) : [w_arr]"r"(W_arr) : "memory" ); diff --git a/libavcodec/mips/vp3dsp_idct_mmi.c b/libavcodec/mips/vp3dsp_idct_mmi.c index 0d4cba19ec..d505fcb765 100644 --- a/libavcodec/mips/vp3dsp_idct_mmi.c +++ b/libavcodec/mips/vp3dsp_idct_mmi.c @@ -722,6 +722,8 @@ void ff_put_no_rnd_pixels_l2_mmi(uint8_t *dst, const uint8_t *src1, if (h == 8) { double ftmp[6]; uint64_t tmp[2]; + DECLARE_VAR_ALL64; + __asm__ volatile ( "li %[tmp0], 0x08 \n\t" "li %[tmp1], 0xfefefefe \n\t" @@ -730,10 +732,8 @@ void ff_put_no_rnd_pixels_l2_mmi(uint8_t *dst, const uint8_t *src1, "li %[tmp1], 0x01 \n\t" "dmtc1 %[tmp1], %[ftmp5] \n\t" "1: \n\t" - "gsldlc1 %[ftmp1], 0x07(%[src1]) \n\t" - "gsldrc1 %[ftmp1], 0x00(%[src1]) \n\t" - "gsldlc1 %[ftmp2], 0x07(%[src2]) \n\t" - "gsldrc1 %[ftmp2], 0x00(%[src2]) \n\t" + MMI_ULDC1(%[ftmp1], %[src1], 0x0) + MMI_ULDC1(%[ftmp2], %[src2], 0x0) "pxor %[ftmp3], %[ftmp1], %[ftmp2] \n\t" "pand %[ftmp3], %[ftmp3], %[ftmp4] \n\t" "psrlw %[ftmp3], %[ftmp3], %[ftmp5] \n\t" @@ -745,7 +745,8 @@ void ff_put_no_rnd_pixels_l2_mmi(uint8_t *dst, const uint8_t *src1, PTR_ADDU "%[dst], %[dst], %[stride] \n\t" PTR_ADDIU "%[tmp0], %[tmp0], -0x01 \n\t" "bnez %[tmp0], 1b \n\t" - : [dst]"+&r"(dst), [src1]"+&r"(src1), [src2]"+&r"(src2), + : RESTRICT_ASM_ALL64 + [dst]"+&r"(dst), [src1]"+&r"(src1), [src2]"+&r"(src2), [ftmp1]"=&f"(ftmp[0]), [ftmp2]"=&f"(ftmp[1]), [ftmp3]"=&f"(ftmp[2]), [ftmp4]"=&f"(ftmp[3]), [ftmp5]"=&f"(ftmp[4]), [ftmp6]"=&f"(ftmp[5]), [tmp0]"=&r"(tmp[0]), [tmp1]"=&r"(tmp[1]) diff --git a/libavcodec/mips/vp8dsp_mmi.c b/libavcodec/mips/vp8dsp_mmi.c index 327eaf561e..bc774aa365 100644 --- a/libavcodec/mips/vp8dsp_mmi.c +++ b/libavcodec/mips/vp8dsp_mmi.c @@ -791,51 +791,40 @@ static av_always_inline void vp8_v_loop_filter8_mmi(uint8_t *dst, DECLARE_DOUBLE_1; DECLARE_DOUBLE_2; DECLARE_UINT32_T; + DECLARE_VAR_ALL64; + __asm__ volatile( /* Get data from dst */ - "gsldlc1 %[q0], 0x07(%[dst]) \n\t" - "gsldrc1 %[q0], 0x00(%[dst]) \n\t" + MMI_ULDC1(%[q0], %[dst], 0x0) PTR_SUBU "%[tmp0], %[dst], %[stride] \n\t" - "gsldlc1 %[p0], 0x07(%[tmp0]) \n\t" - "gsldrc1 %[p0], 0x00(%[tmp0]) \n\t" + MMI_ULDC1(%[p0], %[tmp0], 0x0) PTR_SUBU "%[tmp0], %[tmp0], %[stride] \n\t" - "gsldlc1 %[p1], 0x07(%[tmp0]) \n\t" - "gsldrc1 %[p1], 0x00(%[tmp0]) \n\t" + MMI_ULDC1(%[p1], %[tmp0], 0x0) PTR_SUBU "%[tmp0], %[tmp0], %[stride] \n\t" - "gsldlc1 %[p2], 0x07(%[tmp0]) \n\t" - "gsldrc1 %[p2], 0x00(%[tmp0]) \n\t" + MMI_ULDC1(%[p2], %[tmp0], 0x0) PTR_SUBU "%[tmp0], %[tmp0], %[stride] \n\t" - "gsldlc1 %[p3], 0x07(%[tmp0]) \n\t" - "gsldrc1 %[p3], 0x00(%[tmp0]) \n\t" + MMI_ULDC1(%[p3], %[tmp0], 0x0) PTR_ADDU "%[tmp0], %[dst], %[stride] \n\t" - "gsldlc1 %[q1], 0x07(%[tmp0]) \n\t" - "gsldrc1 %[q1], 0x00(%[tmp0]) \n\t" + MMI_ULDC1(%[q1], %[tmp0], 0x0) PTR_ADDU "%[tmp0], %[tmp0], %[stride] \n\t" - "gsldlc1 %[q2], 0x07(%[tmp0]) \n\t" - "gsldrc1 %[q2], 0x00(%[tmp0]) \n\t" + MMI_ULDC1(%[q2], %[tmp0], 0x0) PTR_ADDU "%[tmp0], %[tmp0], %[stride] \n\t" - "gsldlc1 %[q3], 0x07(%[tmp0]) \n\t" - "gsldrc1 %[q3], 0x00(%[tmp0]) \n\t" + MMI_ULDC1(%[q3], %[tmp0], 0x0) MMI_VP8_LOOP_FILTER /* Move to dst */ - "gssdlc1 %[q0], 0x07(%[dst]) \n\t" - "gssdrc1 %[q0], 0x00(%[dst]) \n\t" + MMI_USDC1(%[q0], %[dst], 0x0) PTR_SUBU "%[tmp0], %[dst], %[stride] \n\t" - "gssdlc1 %[p0], 0x07(%[tmp0]) \n\t" - "gssdrc1 %[p0], 0x00(%[tmp0]) \n\t" + MMI_USDC1(%[p0], %[tmp0], 0x0) PTR_SUBU "%[tmp0], %[tmp0], %[stride] \n\t" - "gssdlc1 %[p1], 0x07(%[tmp0]) \n\t" - "gssdrc1 %[p1], 0x00(%[tmp0]) \n\t" + MMI_USDC1(%[p1], %[tmp0], 0x0) PTR_SUBU "%[tmp0], %[tmp0], %[stride] \n\t" - "gssdlc1 %[p2], 0x07(%[tmp0]) \n\t" - "gssdrc1 %[p2], 0x00(%[tmp0]) \n\t" + MMI_USDC1(%[p2], %[tmp0], 0x0) PTR_ADDU "%[tmp0], %[dst], %[stride] \n\t" - "gssdlc1 %[q1], 0x07(%[tmp0]) \n\t" - "gssdrc1 %[q1], 0x00(%[tmp0]) \n\t" + MMI_USDC1(%[q1], %[tmp0], 0x0) PTR_ADDU "%[tmp0], %[tmp0], %[stride] \n\t" - "gssdlc1 %[q2], 0x07(%[tmp0]) \n\t" - "gssdrc1 %[q2], 0x00(%[tmp0]) \n\t" - : [p3]"=&f"(ftmp[0]), [p2]"=&f"(ftmp[1]), + MMI_USDC1(%[q2], %[tmp0], 0x0) + : RESTRICT_ASM_ALL64 + [p3]"=&f"(ftmp[0]), [p2]"=&f"(ftmp[1]), [p1]"=&f"(ftmp[2]), [p0]"=&f"(ftmp[3]), [q0]"=&f"(ftmp[4]), [q1]"=&f"(ftmp[5]), [q2]"=&f"(ftmp[6]), [q3]"=&f"(ftmp[7]), @@ -876,31 +865,25 @@ static av_always_inline void vp8_h_loop_filter8_mmi(uint8_t *dst, DECLARE_DOUBLE_1; DECLARE_DOUBLE_2; DECLARE_UINT32_T; + DECLARE_VAR_ALL64; + __asm__ volatile( /* Get data from dst */ - "gsldlc1 %[p3], 0x03(%[dst]) \n\t" - "gsldrc1 %[p3], -0x04(%[dst]) \n\t" + MMI_ULDC1(%[p3], %[dst], -0x04) PTR_ADDU "%[tmp0], %[dst], %[stride] \n\t" - "gsldlc1 %[p2], 0x03(%[tmp0]) \n\t" - "gsldrc1 %[p2], -0x04(%[tmp0]) \n\t" + MMI_ULDC1(%[p2], %[tmp0], -0x04) PTR_ADDU "%[tmp0], %[tmp0], %[stride] \n\t" - "gsldlc1 %[p1], 0x03(%[tmp0]) \n\t" - "gsldrc1 %[p1], -0x04(%[tmp0]) \n\t" + MMI_ULDC1(%[p1], %[tmp0], -0x04) PTR_ADDU "%[tmp0], %[tmp0], %[stride] \n\t" - "gsldlc1 %[p0], 0x03(%[tmp0]) \n\t" - "gsldrc1 %[p0], -0x04(%[tmp0]) \n\t" + MMI_ULDC1(%[p0], %[tmp0], -0x04) PTR_ADDU "%[tmp0], %[tmp0], %[stride] \n\t" - "gsldlc1 %[q0], 0x03(%[tmp0]) \n\t" - "gsldrc1 %[q0], -0x04(%[tmp0]) \n\t" + MMI_ULDC1(%[q0], %[tmp0], -0x04) PTR_ADDU "%[tmp0], %[tmp0], %[stride] \n\t" - "gsldlc1 %[q1], 0x03(%[tmp0]) \n\t" - "gsldrc1 %[q1], -0x04(%[tmp0]) \n\t" + MMI_ULDC1(%[q1], %[tmp0], -0x04) PTR_ADDU "%[tmp0], %[tmp0], %[stride] \n\t" - "gsldlc1 %[q2], 0x03(%[tmp0]) \n\t" - "gsldrc1 %[q2], -0x04(%[tmp0]) \n\t" + MMI_ULDC1(%[q2], %[tmp0], -0x04) PTR_ADDU "%[tmp0], %[tmp0], %[stride] \n\t" - "gsldlc1 %[q3], 0x03(%[tmp0]) \n\t" - "gsldrc1 %[q3], -0x04(%[tmp0]) \n\t" + MMI_ULDC1(%[q3], %[tmp0], -0x04) /* Matrix transpose */ TRANSPOSE_8B(%[p3], %[p2], %[p1], %[p0], %[q0], %[q1], %[q2], %[q3], @@ -911,30 +894,23 @@ static av_always_inline void vp8_h_loop_filter8_mmi(uint8_t *dst, %[q0], %[q1], %[q2], %[q3], %[ftmp1], %[ftmp2], %[ftmp3], %[ftmp4]) /* Move to dst */ - "gssdlc1 %[p3], 0x03(%[dst]) \n\t" - "gssdrc1 %[p3], -0x04(%[dst]) \n\t" + MMI_USDC1(%[p3], %[dst], -0x04) PTR_ADDU "%[dst], %[dst], %[stride] \n\t" - "gssdlc1 %[p2], 0x03(%[dst]) \n\t" - "gssdrc1 %[p2], -0x04(%[dst]) \n\t" + MMI_USDC1(%[p2], %[dst], -0x04) PTR_ADDU "%[dst], %[dst], %[stride] \n\t" - "gssdlc1 %[p1], 0x03(%[dst]) \n\t" - "gssdrc1 %[p1], -0x04(%[dst]) \n\t" + MMI_USDC1(%[p1], %[dst], -0x04) PTR_ADDU "%[dst], %[dst], %[stride] \n\t" - "gssdlc1 %[p0], 0x03(%[dst]) \n\t" - "gssdrc1 %[p0], -0x04(%[dst]) \n\t" + MMI_USDC1(%[p0], %[dst], -0x04) PTR_ADDU "%[dst], %[dst], %[stride] \n\t" - "gssdlc1 %[q0], 0x03(%[dst]) \n\t" - "gssdrc1 %[q0], -0x04(%[dst]) \n\t" + MMI_USDC1(%[q0], %[dst], -0x04) PTR_ADDU "%[dst], %[dst], %[stride] \n\t" - "gssdlc1 %[q1], 0x03(%[dst]) \n\t" - "gssdrc1 %[q1], -0x04(%[dst]) \n\t" + MMI_USDC1(%[q1], %[dst], -0x04) PTR_ADDU "%[dst], %[dst], %[stride] \n\t" - "gssdlc1 %[q2], 0x03(%[dst]) \n\t" - "gssdrc1 %[q2], -0x04(%[dst]) \n\t" + MMI_USDC1(%[q2], %[dst], -0x04) PTR_ADDU "%[dst], %[dst], %[stride] \n\t" - "gssdlc1 %[q3], 0x03(%[dst]) \n\t" - "gssdrc1 %[q3], -0x04(%[dst]) \n\t" - : [p3]"=&f"(ftmp[0]), [p2]"=&f"(ftmp[1]), + MMI_USDC1(%[q3], %[dst], -0x04) + : RESTRICT_ASM_ALL64 + [p3]"=&f"(ftmp[0]), [p2]"=&f"(ftmp[1]), [p1]"=&f"(ftmp[2]), [p0]"=&f"(ftmp[3]), [q0]"=&f"(ftmp[4]), [q1]"=&f"(ftmp[5]), [q2]"=&f"(ftmp[6]), [q3]"=&f"(ftmp[7]), diff --git a/libavcodec/mips/vp9_mc_mmi.c b/libavcodec/mips/vp9_mc_mmi.c index 812f7a6994..495cac3d0b 100644 --- a/libavcodec/mips/vp9_mc_mmi.c +++ b/libavcodec/mips/vp9_mc_mmi.c @@ -77,29 +77,24 @@ static void convolve_horiz_mmi(const uint8_t *src, int32_t src_stride, { double ftmp[15]; uint32_t tmp[2]; + DECLARE_VAR_ALL64; src -= 3; src_stride -= w; dst_stride -= w; __asm__ volatile ( "move %[tmp1], %[width] \n\t" "pxor %[ftmp0], %[ftmp0], %[ftmp0] \n\t" - "gsldlc1 %[filter1], 0x07(%[filter]) \n\t" - "gsldrc1 %[filter1], 0x00(%[filter]) \n\t" - "gsldlc1 %[filter2], 0x0f(%[filter]) \n\t" - "gsldrc1 %[filter2], 0x08(%[filter]) \n\t" + MMI_ULDC1(%[filter1], %[filter], 0x00) + MMI_ULDC1(%[filter2], %[filter], 0x08) "li %[tmp0], 0x07 \n\t" "dmtc1 %[tmp0], %[ftmp13] \n\t" "punpcklwd %[ftmp13], %[ftmp13], %[ftmp13] \n\t" "1: \n\t" /* Get 8 data per row */ - "gsldlc1 %[ftmp5], 0x07(%[src]) \n\t" - "gsldrc1 %[ftmp5], 0x00(%[src]) \n\t" - "gsldlc1 %[ftmp7], 0x08(%[src]) \n\t" - "gsldrc1 %[ftmp7], 0x01(%[src]) \n\t" - "gsldlc1 %[ftmp9], 0x09(%[src]) \n\t" - "gsldrc1 %[ftmp9], 0x02(%[src]) \n\t" - "gsldlc1 %[ftmp11], 0x0A(%[src]) \n\t" - "gsldrc1 %[ftmp11], 0x03(%[src]) \n\t" + MMI_ULDC1(%[ftmp5], %[src], 0x00) + MMI_ULDC1(%[ftmp7], %[src], 0x01) + MMI_ULDC1(%[ftmp9], %[src], 0x02) + MMI_ULDC1(%[ftmp11], %[src], 0x03) "punpcklbh %[ftmp4], %[ftmp5], %[ftmp0] \n\t" "punpckhbh %[ftmp5], %[ftmp5], %[ftmp0] \n\t" "punpcklbh %[ftmp6], %[ftmp7], %[ftmp0] \n\t" @@ -127,7 +122,8 @@ static void convolve_horiz_mmi(const uint8_t *src, int32_t src_stride, PTR_ADDU "%[dst], %[dst], %[dst_stride] \n\t" PTR_ADDIU "%[height], %[height], -0x01 \n\t" "bnez %[height], 1b \n\t" - : [srcl]"=&f"(ftmp[0]), [srch]"=&f"(ftmp[1]), + : RESTRICT_ASM_ALL64 + [srcl]"=&f"(ftmp[0]), [srch]"=&f"(ftmp[1]), [filter1]"=&f"(ftmp[2]), [filter2]"=&f"(ftmp[3]), [ftmp0]"=&f"(ftmp[4]), [ftmp4]"=&f"(ftmp[5]), [ftmp5]"=&f"(ftmp[6]), [ftmp6]"=&f"(ftmp[7]), @@ -153,15 +149,14 @@ static void convolve_vert_mmi(const uint8_t *src, int32_t src_stride, double ftmp[17]; uint32_t tmp[1]; ptrdiff_t addr = src_stride; + DECLARE_VAR_ALL64; src_stride -= w; dst_stride -= w; __asm__ volatile ( "pxor %[ftmp0], %[ftmp0], %[ftmp0] \n\t" - "gsldlc1 %[ftmp4], 0x07(%[filter]) \n\t" - "gsldrc1 %[ftmp4], 0x00(%[filter]) \n\t" - "gsldlc1 %[ftmp5], 0x0f(%[filter]) \n\t" - "gsldrc1 %[ftmp5], 0x08(%[filter]) \n\t" + MMI_ULDC1(%[ftmp4], %[filter], 0x00) + MMI_ULDC1(%[ftmp5], %[filter], 0x08) "punpcklwd %[filter10], %[ftmp4], %[ftmp4] \n\t" "punpckhwd %[filter32], %[ftmp4], %[ftmp4] \n\t" "punpcklwd %[filter54], %[ftmp5], %[ftmp5] \n\t" @@ -171,29 +166,21 @@ static void convolve_vert_mmi(const uint8_t *src, int32_t src_stride, "punpcklwd %[ftmp13], %[ftmp13], %[ftmp13] \n\t" "1: \n\t" /* Get 8 data per column */ - "gsldlc1 %[ftmp4], 0x07(%[src]) \n\t" - "gsldrc1 %[ftmp4], 0x00(%[src]) \n\t" + MMI_ULDC1(%[ftmp4], %[src], 0x0) PTR_ADDU "%[tmp0], %[src], %[addr] \n\t" - "gsldlc1 %[ftmp5], 0x07(%[tmp0]) \n\t" - "gsldrc1 %[ftmp5], 0x00(%[tmp0]) \n\t" + MMI_ULDC1(%[ftmp5], %[tmp0], 0x0) PTR_ADDU "%[tmp0], %[tmp0], %[addr] \n\t" - "gsldlc1 %[ftmp6], 0x07(%[tmp0]) \n\t" - "gsldrc1 %[ftmp6], 0x00(%[tmp0]) \n\t" + MMI_ULDC1(%[ftmp6], %[tmp0], 0x0) PTR_ADDU "%[tmp0], %[tmp0], %[addr] \n\t" - "gsldlc1 %[ftmp7], 0x07(%[tmp0]) \n\t" - "gsldrc1 %[ftmp7], 0x00(%[tmp0]) \n\t" + MMI_ULDC1(%[ftmp7], %[tmp0], 0x0) PTR_ADDU "%[tmp0], %[tmp0], %[addr] \n\t" - "gsldlc1 %[ftmp8], 0x07(%[tmp0]) \n\t" - "gsldrc1 %[ftmp8], 0x00(%[tmp0]) \n\t" + MMI_ULDC1(%[ftmp8], %[tmp0], 0x0) PTR_ADDU "%[tmp0], %[tmp0], %[addr] \n\t" - "gsldlc1 %[ftmp9], 0x07(%[tmp0]) \n\t" - "gsldrc1 %[ftmp9], 0x00(%[tmp0]) \n\t" + MMI_ULDC1(%[ftmp9], %[tmp0], 0x0) PTR_ADDU "%[tmp0], %[tmp0], %[addr] \n\t" - "gsldlc1 %[ftmp10], 0x07(%[tmp0]) \n\t" - "gsldrc1 %[ftmp10], 0x00(%[tmp0]) \n\t" + MMI_ULDC1(%[ftmp10], %[tmp0], 0x0) PTR_ADDU "%[tmp0], %[tmp0], %[addr] \n\t" - "gsldlc1 %[ftmp11], 0x07(%[tmp0]) \n\t" - "gsldrc1 %[ftmp11], 0x00(%[tmp0]) \n\t" + MMI_ULDC1(%[ftmp11], %[tmp0], 0x0) "punpcklbh %[ftmp4], %[ftmp4], %[ftmp0] \n\t" "punpcklbh %[ftmp5], %[ftmp5], %[ftmp0] \n\t" "punpcklbh %[ftmp6], %[ftmp6], %[ftmp0] \n\t" @@ -221,7 +208,8 @@ static void convolve_vert_mmi(const uint8_t *src, int32_t src_stride, PTR_ADDU "%[dst], %[dst], %[dst_stride] \n\t" PTR_ADDIU "%[height], %[height], -0x01 \n\t" "bnez %[height], 1b \n\t" - : [srcl]"=&f"(ftmp[0]), [srch]"=&f"(ftmp[1]), + : RESTRICT_ASM_ALL64 + [srcl]"=&f"(ftmp[0]), [srch]"=&f"(ftmp[1]), [filter10]"=&f"(ftmp[2]), [filter32]"=&f"(ftmp[3]), [filter54]"=&f"(ftmp[4]), [filter76]"=&f"(ftmp[5]), [ftmp0]"=&f"(ftmp[6]), [ftmp4]"=&f"(ftmp[7]), @@ -247,6 +235,7 @@ static void convolve_avg_horiz_mmi(const uint8_t *src, int32_t src_stride, { double ftmp[15]; uint32_t tmp[2]; + DECLARE_VAR_ALL64; src -= 3; src_stride -= w; dst_stride -= w; @@ -254,23 +243,17 @@ static void convolve_avg_horiz_mmi(const uint8_t *src, int32_t src_stride, __asm__ volatile ( "move %[tmp1], %[width] \n\t" "pxor %[ftmp0], %[ftmp0], %[ftmp0] \n\t" - "gsldlc1 %[filter1], 0x07(%[filter]) \n\t" - "gsldrc1 %[filter1], 0x00(%[filter]) \n\t" - "gsldlc1 %[filter2], 0x0f(%[filter]) \n\t" - "gsldrc1 %[filter2], 0x08(%[filter]) \n\t" + MMI_ULDC1(%[filter1], %[filter], 0x00) + MMI_ULDC1(%[filter2], %[filter], 0x08) "li %[tmp0], 0x07 \n\t" "dmtc1 %[tmp0], %[ftmp13] \n\t" "punpcklwd %[ftmp13], %[ftmp13], %[ftmp13] \n\t" "1: \n\t" /* Get 8 data per row */ - "gsldlc1 %[ftmp5], 0x07(%[src]) \n\t" - "gsldrc1 %[ftmp5], 0x00(%[src]) \n\t" - "gsldlc1 %[ftmp7], 0x08(%[src]) \n\t" - "gsldrc1 %[ftmp7], 0x01(%[src]) \n\t" - "gsldlc1 %[ftmp9], 0x09(%[src]) \n\t" - "gsldrc1 %[ftmp9], 0x02(%[src]) \n\t" - "gsldlc1 %[ftmp11], 0x0A(%[src]) \n\t" - "gsldrc1 %[ftmp11], 0x03(%[src]) \n\t" + MMI_ULDC1(%[ftmp5], %[src], 0x00) + MMI_ULDC1(%[ftmp7], %[src], 0x01) + MMI_ULDC1(%[ftmp9], %[src], 0x02) + MMI_ULDC1(%[ftmp11], %[src], 0x03) "punpcklbh %[ftmp4], %[ftmp5], %[ftmp0] \n\t" "punpckhbh %[ftmp5], %[ftmp5], %[ftmp0] \n\t" "punpcklbh %[ftmp6], %[ftmp7], %[ftmp0] \n\t" @@ -289,8 +272,7 @@ static void convolve_avg_horiz_mmi(const uint8_t *src, int32_t src_stride, "packsswh %[srcl], %[srcl], %[srch] \n\t" "packushb %[ftmp12], %[srcl], %[ftmp0] \n\t" "punpcklbh %[ftmp12], %[ftmp12], %[ftmp0] \n\t" - "gsldlc1 %[ftmp4], 0x07(%[dst]) \n\t" - "gsldrc1 %[ftmp4], 0x00(%[dst]) \n\t" + MMI_ULDC1(%[ftmp4], %[dst], 0x0) "punpcklbh %[ftmp4], %[ftmp4], %[ftmp0] \n\t" "paddh %[ftmp12], %[ftmp12], %[ftmp4] \n\t" "li %[tmp0], 0x10001 \n\t" @@ -309,7 +291,8 @@ static void convolve_avg_horiz_mmi(const uint8_t *src, int32_t src_stride, PTR_ADDU "%[dst], %[dst], %[dst_stride] \n\t" PTR_ADDIU "%[height], %[height], -0x01 \n\t" "bnez %[height], 1b \n\t" - : [srcl]"=&f"(ftmp[0]), [srch]"=&f"(ftmp[1]), + : RESTRICT_ASM_ALL64 + [srcl]"=&f"(ftmp[0]), [srch]"=&f"(ftmp[1]), [filter1]"=&f"(ftmp[2]), [filter2]"=&f"(ftmp[3]), [ftmp0]"=&f"(ftmp[4]), [ftmp4]"=&f"(ftmp[5]), [ftmp5]"=&f"(ftmp[6]), [ftmp6]"=&f"(ftmp[7]), @@ -335,15 +318,14 @@ static void convolve_avg_vert_mmi(const uint8_t *src, int32_t src_stride, double ftmp[17]; uint32_t tmp[1]; ptrdiff_t addr = src_stride; + DECLARE_VAR_ALL64; src_stride -= w; dst_stride -= w; __asm__ volatile ( "pxor %[ftmp0], %[ftmp0], %[ftmp0] \n\t" - "gsldlc1 %[ftmp4], 0x07(%[filter]) \n\t" - "gsldrc1 %[ftmp4], 0x00(%[filter]) \n\t" - "gsldlc1 %[ftmp5], 0x0f(%[filter]) \n\t" - "gsldrc1 %[ftmp5], 0x08(%[filter]) \n\t" + MMI_ULDC1(%[ftmp4], %[filter], 0x00) + MMI_ULDC1(%[ftmp5], %[filter], 0x08) "punpcklwd %[filter10], %[ftmp4], %[ftmp4] \n\t" "punpckhwd %[filter32], %[ftmp4], %[ftmp4] \n\t" "punpcklwd %[filter54], %[ftmp5], %[ftmp5] \n\t" @@ -353,29 +335,21 @@ static void convolve_avg_vert_mmi(const uint8_t *src, int32_t src_stride, "punpcklwd %[ftmp13], %[ftmp13], %[ftmp13] \n\t" "1: \n\t" /* Get 8 data per column */ - "gsldlc1 %[ftmp4], 0x07(%[src]) \n\t" - "gsldrc1 %[ftmp4], 0x00(%[src]) \n\t" + MMI_ULDC1(%[ftmp4], %[src], 0x0) PTR_ADDU "%[tmp0], %[src], %[addr] \n\t" - "gsldlc1 %[ftmp5], 0x07(%[tmp0]) \n\t" - "gsldrc1 %[ftmp5], 0x00(%[tmp0]) \n\t" + MMI_ULDC1(%[ftmp5], %[tmp0], 0x0) PTR_ADDU "%[tmp0], %[tmp0], %[addr] \n\t" - "gsldlc1 %[ftmp6], 0x07(%[tmp0]) \n\t" - "gsldrc1 %[ftmp6], 0x00(%[tmp0]) \n\t" + MMI_ULDC1(%[ftmp6], %[tmp0], 0x0) PTR_ADDU "%[tmp0], %[tmp0], %[addr] \n\t" - "gsldlc1 %[ftmp7], 0x07(%[tmp0]) \n\t" - "gsldrc1 %[ftmp7], 0x00(%[tmp0]) \n\t" + MMI_ULDC1(%[ftmp7], %[tmp0], 0x0) PTR_ADDU "%[tmp0], %[tmp0], %[addr] \n\t" - "gsldlc1 %[ftmp8], 0x07(%[tmp0]) \n\t" - "gsldrc1 %[ftmp8], 0x00(%[tmp0]) \n\t" + MMI_ULDC1(%[ftmp8], %[tmp0], 0x0) PTR_ADDU "%[tmp0], %[tmp0], %[addr] \n\t" - "gsldlc1 %[ftmp9], 0x07(%[tmp0]) \n\t" - "gsldrc1 %[ftmp9], 0x00(%[tmp0]) \n\t" + MMI_ULDC1(%[ftmp9], %[tmp0], 0x0) PTR_ADDU "%[tmp0], %[tmp0], %[addr] \n\t" - "gsldlc1 %[ftmp10], 0x07(%[tmp0]) \n\t" - "gsldrc1 %[ftmp10], 0x00(%[tmp0]) \n\t" + MMI_ULDC1(%[ftmp10], %[tmp0], 0x0) PTR_ADDU "%[tmp0], %[tmp0], %[addr] \n\t" - "gsldlc1 %[ftmp11], 0x07(%[tmp0]) \n\t" - "gsldrc1 %[ftmp11], 0x00(%[tmp0]) \n\t" + MMI_ULDC1(%[ftmp11], %[tmp0], 0x0) "punpcklbh %[ftmp4], %[ftmp4], %[ftmp0] \n\t" "punpcklbh %[ftmp5], %[ftmp5], %[ftmp0] \n\t" "punpcklbh %[ftmp6], %[ftmp6], %[ftmp0] \n\t" @@ -394,8 +368,7 @@ static void convolve_avg_vert_mmi(const uint8_t *src, int32_t src_stride, "packsswh %[srcl], %[srcl], %[srch] \n\t" "packushb %[ftmp12], %[srcl], %[ftmp0] \n\t" "punpcklbh %[ftmp12], %[ftmp12], %[ftmp0] \n\t" - "gsldlc1 %[ftmp4], 0x07(%[dst]) \n\t" - "gsldrc1 %[ftmp4], 0x00(%[dst]) \n\t" + MMI_ULDC1(%[ftmp4], %[dst], 0x00) "punpcklbh %[ftmp4], %[ftmp4], %[ftmp0] \n\t" "paddh %[ftmp12], %[ftmp12], %[ftmp4] \n\t" "li %[tmp0], 0x10001 \n\t" @@ -414,7 +387,8 @@ static void convolve_avg_vert_mmi(const uint8_t *src, int32_t src_stride, PTR_ADDU "%[dst], %[dst], %[dst_stride] \n\t" PTR_ADDIU "%[height], %[height], -0x01 \n\t" "bnez %[height], 1b \n\t" - : [srcl]"=&f"(ftmp[0]), [srch]"=&f"(ftmp[1]), + : RESTRICT_ASM_ALL64 + [srcl]"=&f"(ftmp[0]), [srch]"=&f"(ftmp[1]), [filter10]"=&f"(ftmp[2]), [filter32]"=&f"(ftmp[3]), [filter54]"=&f"(ftmp[4]), [filter76]"=&f"(ftmp[5]), [ftmp0]"=&f"(ftmp[6]), [ftmp4]"=&f"(ftmp[7]), @@ -439,6 +413,7 @@ static void convolve_avg_mmi(const uint8_t *src, int32_t src_stride, { double ftmp[4]; uint32_t tmp[2]; + DECLARE_VAR_ALL64; src_stride -= w; dst_stride -= w; @@ -449,10 +424,8 @@ static void convolve_avg_mmi(const uint8_t *src, int32_t src_stride, "dmtc1 %[tmp0], %[ftmp3] \n\t" "punpcklhw %[ftmp3], %[ftmp3], %[ftmp3] \n\t" "1: \n\t" - "gslwlc1 %[ftmp1], 0x07(%[src]) \n\t" - "gslwrc1 %[ftmp1], 0x00(%[src]) \n\t" - "gslwlc1 %[ftmp2], 0x07(%[dst]) \n\t" - "gslwrc1 %[ftmp2], 0x00(%[dst]) \n\t" + MMI_ULDC1(%[ftmp1], %[src], 0x00) + MMI_ULDC1(%[ftmp2], %[dst], 0x00) "punpcklbh %[ftmp1], %[ftmp1], %[ftmp0] \n\t" "punpcklbh %[ftmp2], %[ftmp2], %[ftmp0] \n\t" "paddh %[ftmp1], %[ftmp1], %[ftmp2] \n\t" @@ -469,7 +442,8 @@ static void convolve_avg_mmi(const uint8_t *src, int32_t src_stride, PTR_ADDU "%[src], %[src], %[src_stride] \n\t" PTR_ADDIU "%[height], %[height], -0x01 \n\t" "bnez %[height], 1b \n\t" - : [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), + : RESTRICT_ASM_ALL64 + [ftmp0]"=&f"(ftmp[0]), [ftmp1]"=&f"(ftmp[1]), [ftmp2]"=&f"(ftmp[2]), [ftmp3]"=&f"(ftmp[3]), [tmp0]"=&r"(tmp[0]), [tmp1]"=&r"(tmp[1]), [src]"+&r"(src), [dst]"+&r"(dst),

1 0

avutil/mips: Use MMI_{L, S}QC1 macro in {SAVE, RECOVER}_REG
by Jiaxun Yang 28 Jul '21

28 Jul '21

ffmpeg | branch: master | Jiaxun Yang <jiaxun.yang(a)flygoat.com> | Fri Jul 23 13:53:41 2021 +0800| [b868272d7e9f7a50490abd856a5066b009779e07] | committer: Michael Niedermayer avutil/mips: Use MMI_{L, S}QC1 macro in {SAVE, RECOVER}_REG {SAVE,RECOVER}_REG will be available for Loongson2 again, also comment about the magic. Signed-off-by: Jiaxun Yang <jiaxun.yang(a)flygoat.com> Reviewed-by: Shiyou Yin <yinshiyou-hf(a)loongson.cn> Signed-off-by: Michael Niedermayer <michael(a)niedermayer.cc> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=b868272d7e9f7a504… --- libavutil/mips/mmiutils.h | 32 +++++++++++++++++--------------- 1 file changed, 17 insertions(+), 15 deletions(-) diff --git a/libavutil/mips/mmiutils.h b/libavutil/mips/mmiutils.h index 6a82caa908..41715c6490 100644 --- a/libavutil/mips/mmiutils.h +++ b/libavutil/mips/mmiutils.h @@ -204,25 +204,27 @@ #endif /* HAVE_LOONGSON2 */ /** - * backup register + * Backup saved registers + * We're not using compiler's clobber list as it's not smart enough + * to take advantage of quad word load/store. */ #define BACKUP_REG \ LOCAL_ALIGNED_16(double, temp_backup_reg, [8]); \ if (_MIPS_SIM == _ABI64) \ __asm__ volatile ( \ - "gssqc1 $f25, $f24, 0x00(%[temp]) \n\t" \ - "gssqc1 $f27, $f26, 0x10(%[temp]) \n\t" \ - "gssqc1 $f29, $f28, 0x20(%[temp]) \n\t" \ - "gssqc1 $f31, $f30, 0x30(%[temp]) \n\t" \ + MMI_SQC1($f25, $f24, %[temp], 0x00) \ + MMI_SQC1($f27, $f26, %[temp], 0x10) \ + MMI_SQC1($f29, $f28, %[temp], 0x20) \ + MMI_SQC1($f31, $f30, %[temp], 0x30) \ : \ : [temp]"r"(temp_backup_reg) \ : "memory" \ ); \ else \ __asm__ volatile ( \ - "gssqc1 $f22, $f20, 0x00(%[temp]) \n\t" \ - "gssqc1 $f26, $f24, 0x10(%[temp]) \n\t" \ - "gssqc1 $f30, $f28, 0x20(%[temp]) \n\t" \ + MMI_SQC1($f22, $f20, %[temp], 0x10) \ + MMI_SQC1($f26, $f24, %[temp], 0x10) \ + MMI_SQC1($f30, $f28, %[temp], 0x20) \ : \ : [temp]"r"(temp_backup_reg) \ : "memory" \ @@ -234,19 +236,19 @@ #define RECOVER_REG \ if (_MIPS_SIM == _ABI64) \ __asm__ volatile ( \ - "gslqc1 $f25, $f24, 0x00(%[temp]) \n\t" \ - "gslqc1 $f27, $f26, 0x10(%[temp]) \n\t" \ - "gslqc1 $f29, $f28, 0x20(%[temp]) \n\t" \ - "gslqc1 $f31, $f30, 0x30(%[temp]) \n\t" \ + MMI_LQC1($f25, $f24, %[temp], 0x00) \ + MMI_LQC1($f27, $f26, %[temp], 0x10) \ + MMI_LQC1($f29, $f28, %[temp], 0x20) \ + MMI_LQC1($f31, $f30, %[temp], 0x30) \ : \ : [temp]"r"(temp_backup_reg) \ : "memory" \ ); \ else \ __asm__ volatile ( \ - "gslqc1 $f22, $f20, 0x00(%[temp]) \n\t" \ - "gslqc1 $f26, $f24, 0x10(%[temp]) \n\t" \ - "gslqc1 $f30, $f28, 0x20(%[temp]) \n\t" \ + MMI_LQC1($f22, $f20, %[temp], 0x10) \ + MMI_LQC1($f26, $f24, %[temp], 0x10) \ + MMI_LQC1($f30, $f28, %[temp], 0x20) \ : \ : [temp]"r"(temp_backup_reg) \ : "memory" \

1 0

avformat/mov: Remove pointless EOF checks
by Andreas Rheinhardt 28 Jul '21

28 Jul '21

ffmpeg | branch: master | Andreas Rheinhardt <andreas.rheinhardt(a)outlook.com> | Sat Jul 24 06:14:13 2021 +0200| [65f9a8e4b817d4cca8cfdd4dcdaaf2817810df61] | committer: Andreas Rheinhardt avformat/mov: Remove pointless EOF checks 9888ffb1ce5e0a17f711b01933d504c72ea29d3b added checks for EOF in loops in the mov demuxer as a precaution against timeouts; yet there is no I/O in the loop when parsing the STSZ atom as the values are read from an already read buffer. So remove said checks. Reviewed-by: Michael Niedermayer <michael(a)niedermayer.cc> Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt(a)outlook.com> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=65f9a8e4b817d4cca… --- libavformat/mov.c | 7 +------ 1 file changed, 1 insertion(+), 6 deletions(-) diff --git a/libavformat/mov.c b/libavformat/mov.c index a847003dc2..e95d3d2a90 100644 --- a/libavformat/mov.c +++ b/libavformat/mov.c @@ -2884,7 +2884,7 @@ static int mov_read_stsz(MOVContext *c, AVIOContext *pb, MOVAtom atom) init_get_bits(&gb, buf, 8*num_bytes); - for (i = 0; i < entries && !pb->eof_reached; i++) { + for (i = 0; i < entries; i++) { sc->sample_sizes[i] = get_bits_long(&gb, field_size); if (sc->sample_sizes[i] < 0) { av_free(buf); @@ -2898,11 +2898,6 @@ static int mov_read_stsz(MOVContext *c, AVIOContext *pb, MOVAtom atom) av_free(buf); - if (pb->eof_reached) { - av_log(c->fc, AV_LOG_WARNING, "reached eof, corrupted STSZ atom\n"); - return AVERROR_EOF; - } - return 0; }

1 0

avformat/mov: Fix crash with too big STSZ atoms
by Andreas Rheinhardt 28 Jul '21

28 Jul '21

ffmpeg | branch: master | Andreas Rheinhardt <andreas.rheinhardt(a)outlook.com> | Sat Jul 24 05:43:12 2021 +0200| [c2d853c1aae22bbc7d9905c43a9f16cb2ba3ba33] | committer: Andreas Rheinhardt avformat/mov: Fix crash with too big STSZ atoms mov_read_stsz() did not ensure that every bit of a buffer is addressable by an int as is required by the get_bits API, leading to a crash in ticket #9344. Fix this by restricting the size more thoroughly. The file from said ticket will then be considered invalid; in the future, we might read and process the data in chunks to actually support such files. Fixes ticket #9344. Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt(a)outlook.com> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=c2d853c1aae22bbc7… --- libavformat/mov.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/libavformat/mov.c b/libavformat/mov.c index 139bcb4b5c..a847003dc2 100644 --- a/libavformat/mov.c +++ b/libavformat/mov.c @@ -2856,7 +2856,7 @@ static int mov_read_stsz(MOVContext *c, AVIOContext *pb, MOVAtom atom) if (!entries) return 0; - if (entries >= (UINT_MAX - 4) / field_size) + if (entries >= (INT_MAX - 4 - 8 * AV_INPUT_BUFFER_PADDING_SIZE) / field_size) return AVERROR_INVALIDDATA; if (sc->sample_sizes) av_log(c->fc, AV_LOG_WARNING, "Duplicated STSZ atom\n");

1 0

tools/target_dec_fuzzer: Adjust threshold for VP6F
by Michael Niedermayer 28 Jul '21

28 Jul '21

ffmpeg | branch: master | Michael Niedermayer <michael(a)niedermayer.cc> | Sun Jul 25 15:21:36 2021 +0200| [f0ed8de1d6a87648ba71ed05a338ab69aac0ac50] | committer: Michael Niedermayer tools/target_dec_fuzzer: Adjust threshold for VP6F Fixes: Timeout Fixes: 34791/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_VP6F_fuzzer-4571038838030336 Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg Signed-off-by: Michael Niedermayer <michael(a)niedermayer.cc> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=f0ed8de1d6a87648b… --- tools/target_dec_fuzzer.c | 1 + 1 file changed, 1 insertion(+) diff --git a/tools/target_dec_fuzzer.c b/tools/target_dec_fuzzer.c index 8bef96eb68..0195d8910e 100644 --- a/tools/target_dec_fuzzer.c +++ b/tools/target_dec_fuzzer.c @@ -205,6 +205,7 @@ int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size) { case AV_CODEC_ID_VP3: maxpixels /= 4096; break; case AV_CODEC_ID_VP4: maxpixels /= 4096; break; case AV_CODEC_ID_VP5: maxpixels /= 256; break; + case AV_CODEC_ID_VP6F: maxpixels /= 4096; break; case AV_CODEC_ID_VP7: maxpixels /= 256; break; case AV_CODEC_ID_VP9: maxpixels /= 4096; break; case AV_CODEC_ID_WAVPACK: maxsamples /= 1024; break;

1 0

tools/target_dec_fuzzer: Adjust the threshold for VP5
by Michael Niedermayer 28 Jul '21

28 Jul '21

ffmpeg | branch: master | Michael Niedermayer <michael(a)niedermayer.cc> | Tue Jul 20 17:51:47 2021 +0200| [46e8778300b7b9bd0f156a253170bd8f4ba08f24] | committer: Michael Niedermayer tools/target_dec_fuzzer: Adjust the threshold for VP5 Fixes: Timeout Fixes: 35793/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_VP5_fuzzer-6492854393372672 Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg Signed-off-by: Michael Niedermayer <michael(a)niedermayer.cc> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=46e8778300b7b9bd0… --- tools/target_dec_fuzzer.c | 1 + 1 file changed, 1 insertion(+) diff --git a/tools/target_dec_fuzzer.c b/tools/target_dec_fuzzer.c index 0a4bae3fa6..8bef96eb68 100644 --- a/tools/target_dec_fuzzer.c +++ b/tools/target_dec_fuzzer.c @@ -204,6 +204,7 @@ int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size) { case AV_CODEC_ID_VMNC: maxpixels /= 8192; break; case AV_CODEC_ID_VP3: maxpixels /= 4096; break; case AV_CODEC_ID_VP4: maxpixels /= 4096; break; + case AV_CODEC_ID_VP5: maxpixels /= 256; break; case AV_CODEC_ID_VP7: maxpixels /= 256; break; case AV_CODEC_ID_VP9: maxpixels /= 4096; break; case AV_CODEC_ID_WAVPACK: maxsamples /= 1024; break;

1 0

tools/target_dec_fuzzer: Adjust the threshold for VP3
by Michael Niedermayer 28 Jul '21

28 Jul '21

ffmpeg | branch: master | Michael Niedermayer <michael(a)niedermayer.cc> | Tue Jul 20 17:46:09 2021 +0200| [cce928e941e373aa3c197abe4452f69373319776] | committer: Michael Niedermayer tools/target_dec_fuzzer: Adjust the threshold for VP3 Fixes: Timeout Fixes: 35855/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_VP3_fuzzer-4851183540895744 Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg Signed-off-by: Michael Niedermayer <michael(a)niedermayer.cc> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=cce928e941e373aa3… --- tools/target_dec_fuzzer.c | 1 + 1 file changed, 1 insertion(+) diff --git a/tools/target_dec_fuzzer.c b/tools/target_dec_fuzzer.c index 843b447f83..0a4bae3fa6 100644 --- a/tools/target_dec_fuzzer.c +++ b/tools/target_dec_fuzzer.c @@ -202,6 +202,7 @@ int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size) { case AV_CODEC_ID_TSCC: maxpixels /= 1024; break; case AV_CODEC_ID_VC1IMAGE: maxpixels /= 8192; break; case AV_CODEC_ID_VMNC: maxpixels /= 8192; break; + case AV_CODEC_ID_VP3: maxpixels /= 4096; break; case AV_CODEC_ID_VP4: maxpixels /= 4096; break; case AV_CODEC_ID_VP7: maxpixels /= 256; break; case AV_CODEC_ID_VP9: maxpixels /= 4096; break;

1 0

avformat/mov: Fix incorrect overflow detection in mov_read_sidx()
by Michael Niedermayer 28 Jul '21

28 Jul '21

ffmpeg | branch: master | Michael Niedermayer <michael(a)niedermayer.cc> | Wed Apr 28 16:53:52 2021 +0200| [200406d930eff3202f3230f188f85f4ab9cf4525] | committer: Michael Niedermayer avformat/mov: Fix incorrect overflow detection in mov_read_sidx() Fixes: signed integer overflow: 9223372036854775807 + 1442840321 cannot be represented in type 'long' Fixes: 33670/clusterfuzz-testcase-minimized-ffmpeg_dem_MOV_fuzzer-6644379491106816 Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg Signed-off-by: Michael Niedermayer <michael(a)niedermayer.cc> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=200406d930eff3202… --- libavformat/mov.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/libavformat/mov.c b/libavformat/mov.c index 3fc5a1e8ab..139bcb4b5c 100644 --- a/libavformat/mov.c +++ b/libavformat/mov.c @@ -5098,7 +5098,7 @@ static int mov_read_sidx(MOVContext *c, AVIOContext *pb, MOVAtom atom) if (frag_stream_info) frag_stream_info->sidx_pts = timestamp; - if (av_sat_add64(offset, size) != offset + size || + if (av_sat_add64(offset, size) != offset + (uint64_t)size || av_sat_add64(pts, duration) != pts + (uint64_t)duration ) return AVERROR_INVALIDDATA;

1 0

tools/target_dec_fuzzer: Fix extradata duplication
by Michael Niedermayer 28 Jul '21

28 Jul '21

ffmpeg | branch: master | Michael Niedermayer <michael(a)niedermayer.cc> | Sat Jul 24 18:18:23 2021 +0200| [54b798638e68dcebed5c42a216b403004a97f73e] | committer: Michael Niedermayer tools/target_dec_fuzzer: Fix extradata duplication Fixes: out of array access Fixes: 36340/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_MPEG4_fuzzer-5872546875572224.fuzz Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg Signed-off-by: Michael Niedermayer <michael(a)niedermayer.cc> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=54b798638e68dcebe… --- tools/target_dec_fuzzer.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/tools/target_dec_fuzzer.c b/tools/target_dec_fuzzer.c index 3e7689b7ec..843b447f83 100644 --- a/tools/target_dec_fuzzer.c +++ b/tools/target_dec_fuzzer.c @@ -317,7 +317,7 @@ int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size) { } parser_avctx->codec_id = ctx->codec_id; parser_avctx->extradata_size = ctx->extradata_size; - parser_avctx->extradata = av_memdup(ctx->extradata, ctx->extradata_size); + parser_avctx->extradata = ctx->extradata ? av_memdup(ctx->extradata, ctx->extradata_size + AV_INPUT_BUFFER_PADDING_SIZE) : NULL; int got_frame;

1 0