July 2024 - ffmpeg-devel - lists.ffmpeg.org

[PATCH] avcodec/omx: Mark codec as hardware accelerated
by Cameron Gutman 04 Jul '24

04 Jul '24

This allows applications to preferentially select OpenMAX IL codecs over software-only codecs using the standard AV_CODEC_CAP_HARDWARE capability flag rather than requiring custom codec-specific logic. Signed-off-by: Cameron Gutman <aicommander(a)gmail.com> --- libavcodec/omx.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/libavcodec/omx.c b/libavcodec/omx.c index 86913237b2..87f6c4b5c8 100644 --- a/libavcodec/omx.c +++ b/libavcodec/omx.c @@ -948,7 +948,7 @@ const FFCodec ff_mpeg4_omx_encoder = { FF_CODEC_ENCODE_CB(omx_encode_frame), .close = omx_encode_end, .p.pix_fmts = omx_encoder_pix_fmts, - .p.capabilities = AV_CODEC_CAP_DELAY, + .p.capabilities = AV_CODEC_CAP_DELAY | AV_CODEC_CAP_HARDWARE, .caps_internal = FF_CODEC_CAP_INIT_CLEANUP, .p.priv_class = &omx_mpeg4enc_class, }; @@ -969,7 +969,7 @@ const FFCodec ff_h264_omx_encoder = { FF_CODEC_ENCODE_CB(omx_encode_frame), .close = omx_encode_end, .p.pix_fmts = omx_encoder_pix_fmts, - .p.capabilities = AV_CODEC_CAP_DELAY, + .p.capabilities = AV_CODEC_CAP_DELAY | AV_CODEC_CAP_HARDWARE, .caps_internal = FF_CODEC_CAP_INIT_CLEANUP, .p.priv_class = &omx_h264enc_class, }; -- 2.45.2

1 0

[PATCH v2 1/5] hlsenc: Fix the return value accumulation in append_single_file
by Martin Storsjö 04 Jul '24

04 Jul '24

Both the read_byte variable (which is accumulated into append_single_file) and the return value are int64_t; give the ret variable the right corresponding type too. --- libavformat/hlsenc.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/libavformat/hlsenc.c b/libavformat/hlsenc.c index f5c0243cf1..3d5eb47e84 100644 --- a/libavformat/hlsenc.c +++ b/libavformat/hlsenc.c @@ -2380,7 +2380,7 @@ static int hls_init_file_resend(AVFormatContext *s, VariantStream *vs) static int64_t append_single_file(AVFormatContext *s, VariantStream *vs) { - int ret = 0; + int64_t ret = 0; int64_t read_byte = 0; int64_t total_size = 0; char *filename = NULL; -- 2.39.3 (Apple Git-146)

2 7

[PATCH] tests: Add a missing dependency for the filter-atempo test
by Martin Storsjö 04 Jul '24

04 Jul '24

--- tests/fate/filter-audio.mak | 1 + 1 file changed, 1 insertion(+) diff --git a/tests/fate/filter-audio.mak b/tests/fate/filter-audio.mak index 8120620afe..cf6065b93f 100644 --- a/tests/fate/filter-audio.mak +++ b/tests/fate/filter-audio.mak @@ -413,6 +413,7 @@ fate-filter-hdcd-s32p: CMP = oneline fate-filter-hdcd-s32p: REF = 0c5513e83eedaa10ab6fac9ddc173cf5 FATE_AFILTER_SAMPLES-$(call FILTERDEMDECENCMUX, ATEMPO, WAV, PCM_S16LE, PCM_S16LE, WAV) += fate-filter-atempo +fate-filter-atempo: tests/data/asynth-44100-1.wav fate-filter-atempo: CMD = framecrc -i $(TARGET_PATH)/tests/data/asynth-44100-1.wav -af "atempo=2.0" -ar 44100 FATE_AFILTER-yes += fate-filter-formats -- 2.39.3 (Apple Git-146)

2 1

[PATCH 1/6] swscale: document SWS_FULL_CHR_H_* flags
by Niklas Haas 04 Jul '24

04 Jul '24

From: Niklas Haas <git(a)haasn.dev> Based on my best understanding of what they do, given the source code. --- libswscale/swscale.h | 28 ++++++++++++++++++++++++++-- 1 file changed, 26 insertions(+), 2 deletions(-) diff --git a/libswscale/swscale.h b/libswscale/swscale.h index 9d4612aaf3..e22931cab4 100644 --- a/libswscale/swscale.h +++ b/libswscale/swscale.h @@ -82,11 +82,35 @@ const char *swscale_license(void); #define SWS_PRINT_INFO 0x1000 //the following 3 flags are not completely implemented -//internal chrominance subsampling info + +/** + * Perform full chroma upsampling when converting to RGB as part of scaling. + * + * For example, when converting 50x50 yuv420p to 100x100 rgba, setting this flag + * will scale the chroma plane from 25x25 to 100x100 (4:4:4), and then convert + * the 100x100 yuv444p image to rgba in the final output step. + * + * Without this flag, the chroma plane is instead scaled to 50x100 (4:2:2), + * with a single chroma sample being re-used for both horizontally adjacent RGBA + * output pixels. + */ #define SWS_FULL_CHR_H_INT 0x2000 -//input subsampling info + +/** + * Perform full chroma interpolation when downscaling RGB sources. + * + * For example, when converting a 100x100 rgba source to 50x50 yuv444p, setting + * this flag will generate a 100x100 (4:4:4) chroma plane, which is then + * downscaled to the required 50x50. + * + * Without this flag, the chroma plane is instead generated at 50x100 (dropping + * every other pixel), before then being downscaled to the required 50x50 + * resolution. + */ #define SWS_FULL_CHR_H_INP 0x4000 + #define SWS_DIRECT_BGR 0x8000 + #define SWS_ACCURATE_RND 0x40000 #define SWS_BITEXACT 0x80000 #define SWS_ERROR_DIFFUSION 0x800000 -- 2.45.2

2 7

[PATCH] doc/filter: fix grammar in tiltandshift filter
by Gyan Doshi 04 Jul '24

04 Jul '24

--- doc/filters.texi | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/doc/filters.texi b/doc/filters.texi index 0ff7c142b6..c9c4f7cf6b 100644 --- a/doc/filters.texi +++ b/doc/filters.texi @@ -23546,13 +23546,13 @@ tile=3x2:nb_frames=5:padding=7:margin=2 @end itemize @section tiltandshift -Apply tilt-and-shift effet. +Apply tilt-and-shift effect. What happens when you invert time and space? Normally a video is composed of several frames that represent a different instant of time and shows a scene that evolves in the space captured by the -frame. This filter is the antipode of that concept, taking inspiration by +frame. This filter is the antipode of that concept, taking inspiration from tilt and shift photography. A filtered frame contains the whole timeline of events composing the sequence, @@ -23591,7 +23591,7 @@ How many columns should be inserted before end of filtering. @end table -Normally the filter shifts and tils from the very first frame, and stops when +Normally the filter shifts and tilts from the very first frame, and stops when the last one is received. However, before filtering starts, normal video may be preseved, so that the effect is slowly shifted in its place. Similarly, the last video frame may be reconstructed at the end. Alternatively it is @@ -23599,7 +23599,7 @@ possible to just start and end with black. @table @samp @item none -Filtering is starts immediately and ends when the last frame is received. +Filtering starts immediately and ends when the last frame is received. @item frame The first frames or the very last frame are kept intact during processing. -- 2.44.0

2 3

[PATCH 1/3] lavc/h264dsp: R-V V 8-bit h264_idct_add16
by Rémi Denis-Courmont 03 Jul '24

03 Jul '24

While this *tends* to be faster than plain C, the performance numbers are all over the place, presuambly due to the conditional character of the main loop. Some additional micro-optimisations should be feasible after the underlying h264_idct_add and h264_idct_dc_add functions are also implemented. Then it will no longer be necesseray to stricly abide by the C ABI. --- libavcodec/riscv/Makefile | 2 +- libavcodec/riscv/h264dsp_init.c | 6 ++ libavcodec/riscv/h264idct_rvv.S | 106 ++++++++++++++++++++++++++++++++ 3 files changed, 113 insertions(+), 1 deletion(-) create mode 100644 libavcodec/riscv/h264idct_rvv.S diff --git a/libavcodec/riscv/Makefile b/libavcodec/riscv/Makefile index a1510e8c6e..92e1544e76 100644 --- a/libavcodec/riscv/Makefile +++ b/libavcodec/riscv/Makefile @@ -31,7 +31,7 @@ RVV-OBJS-$(CONFIG_H263DSP) += riscv/h263dsp_rvv.o OBJS-$(CONFIG_H264CHROMA) += riscv/h264_chroma_init_riscv.o RVV-OBJS-$(CONFIG_H264CHROMA) += riscv/h264_mc_chroma.o OBJS-$(CONFIG_H264DSP) += riscv/h264dsp_init.o -RVV-OBJS-$(CONFIG_H264DSP) += riscv/h264dsp_rvv.o +RVV-OBJS-$(CONFIG_H264DSP) += riscv/h264dsp_rvv.o riscv/h264idct_rvv.o OBJS-$(CONFIG_HUFFYUV_DECODER) += riscv/huffyuvdsp_init.o RVV-OBJS-$(CONFIG_HUFFYUV_DECODER) += riscv/huffyuvdsp_rvv.o OBJS-$(CONFIG_IDCTDSP) += riscv/idctdsp_init.o diff --git a/libavcodec/riscv/h264dsp_init.c b/libavcodec/riscv/h264dsp_init.c index ab412a9924..5f318e3f2b 100644 --- a/libavcodec/riscv/h264dsp_init.c +++ b/libavcodec/riscv/h264dsp_init.c @@ -34,6 +34,10 @@ void ff_h264_h_loop_filter_luma_8_rvv(uint8_t *pix, ptrdiff_t stride, void ff_h264_h_loop_filter_luma_mbaff_8_rvv(uint8_t *pix, ptrdiff_t stride, int alpha, int beta, int8_t *tc0); +void ff_h264_idct_add16_8_rvv(uint8_t *dst, const int *blockoffset, + int16_t *block, int stride, + const uint8_t nnzc[5 * 8]); + extern int ff_startcode_find_candidate_rvb(const uint8_t *, int); extern int ff_startcode_find_candidate_rvv(const uint8_t *, int); @@ -52,6 +56,8 @@ av_cold void ff_h264dsp_init_riscv(H264DSPContext *dsp, const int bit_depth, dsp->h264_h_loop_filter_luma = ff_h264_h_loop_filter_luma_8_rvv; dsp->h264_h_loop_filter_luma_mbaff = ff_h264_h_loop_filter_luma_mbaff_8_rvv; + + dsp->h264_idct_add16 = ff_h264_idct_add16_8_rvv; } dsp->startcode_find_candidate = ff_startcode_find_candidate_rvv; } diff --git a/libavcodec/riscv/h264idct_rvv.S b/libavcodec/riscv/h264idct_rvv.S new file mode 100644 index 0000000000..74083f8221 --- /dev/null +++ b/libavcodec/riscv/h264idct_rvv.S @@ -0,0 +1,106 @@ +/* + * Copyright © 2024 Rémi Denis-Courmont. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions are met: + * + * 1. Redistributions of source code must retain the above copyright notice, + * this list of conditions and the following disclaimer. + * + * 2. Redistributions in binary form must reproduce the above copyright notice, + * this list of conditions and the following disclaimer in the documentation + * and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" + * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE + * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR + * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF + * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS + * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN + * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) + * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE + * POSSIBILITY OF SUCH DAMAGE. + */ + +#include "libavutil/riscv/asm.S" + +const ff_h264_scan8 + .byte 014, 015, 024, 025, 016, 017, 026, 027 + .byte 034, 035, 044, 045, 036, 037, 046, 047 +endconst + +#if (__riscv_xlen == 64) +.irp depth, 8 +func ff_h264_idct_add16_\depth\()_rvv, zve32x + addi sp, sp, -80 + lla t0, ff_h264_scan8 + sd s0, (sp) + li t1, 32 << (\depth > 8) + mv s0, sp + sd ra, 8(sp) + sd s1, 16(sp) + sd s2, 24(sp) + sd s3, 32(sp) + sd s4, 40(sp) + sd s5, 48(sp) + sd s6, 56(sp) + sd s7, 64(sp) + vsetivli zero, 16, e8, m1, ta, ma + vle8.v v8, (t0) + vlse16.v v16, (a2), t1 + vluxei8.v v12, (a4), v8 +.if \depth == 8 + vsetvli zero, zero, e16, m2, ta, ma +.else + vsetvli zero, zero, e32, m4, ta, ma +.endif + vmsne.vi v1, v16, 0 + vsetvli zero, zero, e8, m1, ta, ma + vmseq.vi v2, v12, 1 + vmsne.vi v0, v12, 0 + vmand.mm v1, v1, v2 + vsetvli zero, zero, e16, m2, ta, ma + vmv.x.s s2, v0 + vmv.x.s s3, v1 + li s1, 16 + mv s4, a0 + mv s5, a1 + mv s6, a2 + mv s7, a3 +1: + andi t0, s2, 1 + addi s1, s1, -1 + srli s2, s2, 1 + beqz t0, 3f # if (nnz) + lw t2, (s5) # block_offset[i] + andi t1, s3, 1 + mv a1, s6 + mv a2, s7 + add a0, s4, t2 + beqz t1, 2f # if (nnz == 1 && block[i * 16]) + call ff_h264_idct_dc_add_\depth\()_c + j 3f +2: + call ff_h264_idct_add_\depth\()_c +3: + srli s3, s3, 1 + addi s5, s5, 4 + addi s6, s6, 16 * 2 << (\depth > 8) + bnez s1, 1b + + ld s7, 64(sp) + ld s6, 56(sp) + ld s5, 48(sp) + ld s4, 40(sp) + ld s3, 32(sp) + ld s2, 24(sp) + ld s1, 16(sp) + ld ra, 8(sp) + ld s0, 0(sp) + addi sp, sp, 80 + ret +endfunc +.endr +#endif -- 2.45.2

1 5

[PATCH v2] avcodec/riscv: add h264 dc idct rvv
by J. Dekker 03 Jul '24

03 Jul '24

checkasm: bench runs 131072 (1 << 17) h264_idct4_add_dc_8bpp_c: 1.5 h264_idct4_add_dc_8bpp_rvv_i64: 0.7 h264_idct4_add_dc_9bpp_c: 1.5 h264_idct4_add_dc_9bpp_rvv_i64: 0.7 h264_idct4_add_dc_10bpp_c: 1.5 h264_idct4_add_dc_10bpp_rvv_i64: 0.7 h264_idct4_add_dc_12bpp_c: 1.2 h264_idct4_add_dc_12bpp_rvv_i64: 0.7 h264_idct4_add_dc_14bpp_c: 1.2 h264_idct4_add_dc_14bpp_rvv_i64: 0.7 h264_idct8_add_dc_8bpp_c: 5.2 h264_idct8_add_dc_8bpp_rvv_i64: 1.5 h264_idct8_add_dc_9bpp_c: 5.5 h264_idct8_add_dc_9bpp_rvv_i64: 1.2 h264_idct8_add_dc_10bpp_c: 5.5 h264_idct8_add_dc_10bpp_rvv_i64: 1.2 h264_idct8_add_dc_12bpp_c: 4.2 h264_idct8_add_dc_12bpp_rvv_i64: 1.2 h264_idct8_add_dc_14bpp_c: 4.2 h264_idct8_add_dc_14bpp_rvv_i64: 1.2 Signed-off-by: J. Dekker <jdek(a)itanimul.li> --- rdcycle always returns 0 on my board, clock_gettime() seems as noisy as rdtime (just with bigger numbers). libavcodec/riscv/Makefile | 1 + libavcodec/riscv/h264dsp_init.c | 42 +++++++- libavcodec/riscv/h264dsp_rvv.S | 176 ++++++++++++++++++++++++++++++++ 3 files changed, 216 insertions(+), 3 deletions(-) create mode 100644 libavcodec/riscv/h264dsp_rvv.S diff --git a/libavcodec/riscv/Makefile b/libavcodec/riscv/Makefile index c180223141..a1510e8c6e 100644 --- a/libavcodec/riscv/Makefile +++ b/libavcodec/riscv/Makefile @@ -31,6 +31,7 @@ RVV-OBJS-$(CONFIG_H263DSP) += riscv/h263dsp_rvv.o OBJS-$(CONFIG_H264CHROMA) += riscv/h264_chroma_init_riscv.o RVV-OBJS-$(CONFIG_H264CHROMA) += riscv/h264_mc_chroma.o OBJS-$(CONFIG_H264DSP) += riscv/h264dsp_init.o +RVV-OBJS-$(CONFIG_H264DSP) += riscv/h264dsp_rvv.o OBJS-$(CONFIG_HUFFYUV_DECODER) += riscv/huffyuvdsp_init.o RVV-OBJS-$(CONFIG_HUFFYUV_DECODER) += riscv/huffyuvdsp_rvv.o OBJS-$(CONFIG_IDCTDSP) += riscv/idctdsp_init.o diff --git a/libavcodec/riscv/h264dsp_init.c b/libavcodec/riscv/h264dsp_init.c index dbbf3db400..8c77303ec6 100644 --- a/libavcodec/riscv/h264dsp_init.c +++ b/libavcodec/riscv/h264dsp_init.c @@ -1,4 +1,5 @@ /* + * Copyright (c) 2024 J. Dekker <jdek(a)itanimul.li> * Copyright © 2024 Rémi Denis-Courmont. * * This file is part of FFmpeg. @@ -24,22 +25,57 @@ #include "libavutil/attributes.h" #include "libavutil/cpu.h" +#include "libavutil/riscv/cpu.h" #include "libavcodec/h264dsp.h" extern int ff_startcode_find_candidate_rvb(const uint8_t *, int); extern int ff_startcode_find_candidate_rvv(const uint8_t *, int); +void ff_h264_idct4_dc_add_8_rvv(uint8_t *dst, int16_t *block, int stride); +void ff_h264_idct8_dc_add_8_rvv(uint8_t *dst, int16_t *block, int stride); +void ff_h264_idct4_dc_add_9_rvv(uint8_t *dst, int16_t *block, int stride); +void ff_h264_idct8_dc_add_9_rvv(uint8_t *dst, int16_t *block, int stride); +void ff_h264_idct4_dc_add_10_rvv(uint8_t *dst, int16_t *block, int stride); +void ff_h264_idct8_dc_add_10_rvv(uint8_t *dst, int16_t *block, int stride); +void ff_h264_idct4_dc_add_12_rvv(uint8_t *dst, int16_t *block, int stride); +void ff_h264_idct8_dc_add_12_rvv(uint8_t *dst, int16_t *block, int stride); +void ff_h264_idct4_dc_add_14_rvv(uint8_t *dst, int16_t *block, int stride); +void ff_h264_idct8_dc_add_14_rvv(uint8_t *dst, int16_t *block, int stride); -av_cold void ff_h264dsp_init_riscv(H264DSPContext *dsp, const int bit_depth, +av_cold void ff_h264dsp_init_riscv(H264DSPContext *c, const int bit_depth, const int chroma_format_idc) { #if HAVE_RV int flags = av_get_cpu_flags(); if (flags & AV_CPU_FLAG_RVB_BASIC) - dsp->startcode_find_candidate = ff_startcode_find_candidate_rvb; + c->startcode_find_candidate = ff_startcode_find_candidate_rvb; # if HAVE_RVV if (flags & AV_CPU_FLAG_RVV_I32) - dsp->startcode_find_candidate = ff_startcode_find_candidate_rvv; + c->startcode_find_candidate = ff_startcode_find_candidate_rvv; # endif + if ((flags & AV_CPU_FLAG_RVV_I64) && ff_rv_vlen_least(16)) { + switch(bit_depth) { + case 8: + c->h264_idct_dc_add = ff_h264_idct4_dc_add_8_rvv; + c->h264_idct8_dc_add = ff_h264_idct8_dc_add_8_rvv; + break; + case 9: + c->h264_idct_dc_add = ff_h264_idct4_dc_add_9_rvv; + c->h264_idct8_dc_add = ff_h264_idct8_dc_add_9_rvv; + break; + case 10: + c->h264_idct_dc_add = ff_h264_idct4_dc_add_10_rvv; + c->h264_idct8_dc_add = ff_h264_idct8_dc_add_10_rvv; + break; + case 12: + c->h264_idct_dc_add = ff_h264_idct4_dc_add_12_rvv; + c->h264_idct8_dc_add = ff_h264_idct8_dc_add_12_rvv; + break; + case 14: + c->h264_idct_dc_add = ff_h264_idct4_dc_add_14_rvv; + c->h264_idct8_dc_add = ff_h264_idct8_dc_add_14_rvv; + break; + } + } #endif } diff --git a/libavcodec/riscv/h264dsp_rvv.S b/libavcodec/riscv/h264dsp_rvv.S new file mode 100644 index 0000000000..57f0433f7c --- /dev/null +++ b/libavcodec/riscv/h264dsp_rvv.S @@ -0,0 +1,176 @@ +/* + * SPDX-License-Identifier: BSD-2-Clause + * + * Copyright (c) 2024 J. Dekker <jdek(a)itanimul.li> + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" + * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE + * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR + * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF + * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS + * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN + * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) + * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE + * POSSIBILITY OF SUCH DAMAGE. + */ + +#include "libavutil/riscv/asm.S" + +.macro idct_dc_add8 width +func ff_h264_idct\width\()_dc_add_8_rvv, zve64x, zba + vsetivli zero, \width, e16, m1, ta, ma + lh a3, 0(a1) + addi a3, a3, 32 + srai a3, a3, 6 + sh zero, 0(a1) +.if \width == 8 + vlsseg8e8.v v24, (a0), a2 +.else + vlsseg4e8.v v24, (a0), a2 +.endif + vzext.vf2 v0, v24 + vzext.vf2 v2, v25 + vzext.vf2 v4, v26 + vzext.vf2 v6, v27 +.if \width == 8 + vzext.vf2 v10, v28 + vzext.vf2 v12, v29 + vzext.vf2 v14, v30 + vzext.vf2 v16, v31 +.endif + vadd.vx v0, v0, a3 + vadd.vx v2, v2, a3 + vadd.vx v4, v4, a3 + vadd.vx v6, v6, a3 +.if \width == 8 + vadd.vx v10, v10, a3 + vadd.vx v12, v12, a3 + vadd.vx v14, v14, a3 + vadd.vx v16, v16, a3 +.endif + vmax.vx v0, v0, zero + vmax.vx v2, v2, zero + vmax.vx v4, v4, zero + vmax.vx v6, v6, zero +.if \width == 8 + vmax.vx v10, v10, zero + vmax.vx v12, v12, zero + vmax.vx v14, v14, zero + vmax.vx v16, v16, zero +.endif + vsetvli zero, zero, e8, mf2, ta, ma + vnclipu.wi v24, v0, 0 + vnclipu.wi v25, v2, 0 + vnclipu.wi v26, v4, 0 + vnclipu.wi v27, v6, 0 +.if \width == 8 + vnclipu.wi v28, v10, 0 + vnclipu.wi v29, v12, 0 + vnclipu.wi v30, v14, 0 + vnclipu.wi v31, v16, 0 + vssseg8e8.v v24, (a0), a2 +.else + vssseg4e8.v v24, (a0), a2 +.endif + ret +endfunc +.endm + +idct_dc_add8 4 +idct_dc_add8 8 + +.macro idct_dc_add width +func ff_h264_idct\width\()_dc_add_16_rvv, zve64x, zba + vsetivli zero, \width, e16, m1, ta, ma + lw a3, 0(a1) + addi a3, a3, 32 + srai a3, a3, 6 + sw zero, 0(a1) + add t4, a0, a2 + sh1add t5, a2, a0 + sh1add t6, a2, t4 +.if \width == 8 + sh2add t0, a2, a0 + sh2add t1, a2, t4 + sh2add t2, a2, t5 + sh2add t3, a2, t6 +.endif + vle16.v v0, (a0) + vle16.v v2, (t4) + vle16.v v4, (t5) + vle16.v v6, (t6) +.if \width == 8 + vle16.v v10, (t0) + vle16.v v12, (t1) + vle16.v v14, (t2) + vle16.v v16, (t3) +.endif + vadd.vx v0, v0, a3 + vadd.vx v2, v2, a3 + vadd.vx v4, v4, a3 + vadd.vx v6, v6, a3 +.if \width == 8 + vadd.vx v10, v10, a3 + vadd.vx v12, v12, a3 + vadd.vx v14, v14, a3 + vadd.vx v16, v16, a3 +.endif + vmax.vx v0, v0, zero + vmax.vx v2, v2, zero + vmax.vx v4, v4, zero + vmax.vx v6, v6, zero +.if \width == 8 + vmax.vx v10, v10, zero + vmax.vx v12, v12, zero + vmax.vx v14, v14, zero + vmax.vx v16, v16, zero +.endif + vmin.vx v0, v0, a5 + vmin.vx v2, v2, a5 + vmin.vx v4, v4, a5 + vmin.vx v6, v6, a5 +.if \width == 8 + vmin.vx v10, v10, a5 + vmin.vx v12, v12, a5 + vmin.vx v14, v14, a5 + vmin.vx v16, v16, a5 +.endif + vse16.v v0, (a0) + vse16.v v2, (t4) + vse16.v v4, (t5) + vse16.v v6, (t6) +.if \width == 8 + vse16.v v10, (t0) + vse16.v v12, (t1) + vse16.v v14, (t2) + vse16.v v16, (t3) +.endif + ret +endfunc +.endm + +idct_dc_add 4 +idct_dc_add 8 + +.irp depth,9,10,12,14 +func ff_h264_idct4_dc_add_\depth\()_rvv, zve64x + li a5, (1 << \depth) - 1 + j ff_h264_idct4_dc_add_16_rvv +endfunc + +func ff_h264_idct8_dc_add_\depth\()_rvv, zve64x + li a5, (1 << \depth) - 1 + j ff_h264_idct8_dc_add_16_rvv +endfunc +.endr -- 2.44.1

2 1

[PATCH 2/2] fate/atrac1: add test for 4-track file
by asivery 03 Jul '24

03 Jul '24

Empty Message

1 0

[PATCH 1/2] avcodec/pngdec: fix mDCv typo
by Leo Izen 03 Jul '24

03 Jul '24

When mDCv support was added, there was a typo in both variable names and also the MKTAG itself, incorrectly listing it as mDVc. The tag name stands for Mastering Display Color Volume so mDCv is correct. See other files such as av1dec.c which uses mdcv. Typo originally introduced in c7a57b0f70f8d1574aa0f0dbe98db85d8ac91c76. Signed-off-by: Leo Izen <leo.izen(a)gmail.com> Reported-by: Ramiro Polla <ramiro.polla(a)gmail.com> --- libavcodec/pngdec.c | 53 +++++++++++++++++++++++---------------------- 1 file changed, 27 insertions(+), 26 deletions(-) diff --git a/libavcodec/pngdec.c b/libavcodec/pngdec.c index 1c910e6a5b..180806e5e1 100644 --- a/libavcodec/pngdec.c +++ b/libavcodec/pngdec.c @@ -86,11 +86,12 @@ typedef struct PNGDecContext { int have_clli; uint32_t clli_max; uint32_t clli_avg; - int have_mdvc; - uint16_t mdvc_primaries[3][2]; - uint16_t mdvc_white_point[2]; - uint32_t mdvc_max_lum; - uint32_t mdvc_min_lum; + /* Mastering Display Color Volume */ + int have_mdcv; + uint16_t mdcv_primaries[3][2]; + uint16_t mdcv_white_point[2]; + uint32_t mdcv_max_lum; + uint32_t mdcv_min_lum; enum PNGHeaderState hdr_state; enum PNGImageState pic_state; @@ -763,24 +764,24 @@ static int populate_avctx_color_fields(AVCodecContext *avctx, AVFrame *frame) } } - if (s->have_mdvc) { - AVMasteringDisplayMetadata *mdvc; + if (s->have_mdcv) { + AVMasteringDisplayMetadata *mdcv; - ret = ff_decode_mastering_display_new(avctx, frame, &mdvc); + ret = ff_decode_mastering_display_new(avctx, frame, &mdcv); if (ret < 0) return ret; - if (mdvc) { - mdvc->has_primaries = 1; + if (mdcv) { + mdcv->has_primaries = 1; for (int i = 0; i < 3; i++) { - mdvc->display_primaries[i][0] = av_make_q(s->mdvc_primaries[i][0], 50000); - mdvc->display_primaries[i][1] = av_make_q(s->mdvc_primaries[i][1], 50000); + mdcv->display_primaries[i][0] = av_make_q(s->mdcv_primaries[i][0], 50000); + mdcv->display_primaries[i][1] = av_make_q(s->mdcv_primaries[i][1], 50000); } - mdvc->white_point[0] = av_make_q(s->mdvc_white_point[0], 50000); - mdvc->white_point[1] = av_make_q(s->mdvc_white_point[1], 50000); - mdvc->has_luminance = 1; - mdvc->max_luminance = av_make_q(s->mdvc_max_lum, 10000); - mdvc->min_luminance = av_make_q(s->mdvc_min_lum, 10000); + mdcv->white_point[0] = av_make_q(s->mdcv_white_point[0], 50000); + mdcv->white_point[1] = av_make_q(s->mdcv_white_point[1], 50000); + mdcv->has_luminance = 1; + mdcv->max_luminance = av_make_q(s->mdcv_max_lum, 10000); + mdcv->min_luminance = av_make_q(s->mdcv_min_lum, 10000); } } @@ -1571,20 +1572,20 @@ static int decode_frame_common(AVCodecContext *avctx, PNGDecContext *s, s->clli_max = bytestream2_get_be32u(&gb_chunk); s->clli_avg = bytestream2_get_be32u(&gb_chunk); break; - case MKTAG('m', 'D', 'V', 'c'): + case MKTAG('m', 'D', 'C', 'v'): if (bytestream2_get_bytes_left(&gb_chunk) != 24) { - av_log(avctx, AV_LOG_WARNING, "Invalid mDVc chunk size: %d\n", bytestream2_get_bytes_left(&gb_chunk)); + av_log(avctx, AV_LOG_WARNING, "Invalid mDCv chunk size: %d\n", bytestream2_get_bytes_left(&gb_chunk)); break; } - s->have_mdvc = 1; + s->have_mdcv = 1; for (int i = 0; i < 3; i++) { - s->mdvc_primaries[i][0] = bytestream2_get_be16u(&gb_chunk); - s->mdvc_primaries[i][1] = bytestream2_get_be16u(&gb_chunk); + s->mdcv_primaries[i][0] = bytestream2_get_be16u(&gb_chunk); + s->mdcv_primaries[i][1] = bytestream2_get_be16u(&gb_chunk); } - s->mdvc_white_point[0] = bytestream2_get_be16u(&gb_chunk); - s->mdvc_white_point[1] = bytestream2_get_be16u(&gb_chunk); - s->mdvc_max_lum = bytestream2_get_be32u(&gb_chunk); - s->mdvc_min_lum = bytestream2_get_be32u(&gb_chunk); + s->mdcv_white_point[0] = bytestream2_get_be16u(&gb_chunk); + s->mdcv_white_point[1] = bytestream2_get_be16u(&gb_chunk); + s->mdcv_max_lum = bytestream2_get_be32u(&gb_chunk); + s->mdcv_min_lum = bytestream2_get_be32u(&gb_chunk); break; case MKTAG('I', 'E', 'N', 'D'): if (!(s->pic_state & PNG_ALLIMAGE)) -- 2.45.2

3 6

[PATCH] avcodec/hw_base_encode: Add missing include
by Andreas Rheinhardt 03 Jul '24

03 Jul '24

Fixes checkheaders. Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt(a)outlook.com> --- libavcodec/hw_base_encode.h | 1 + 1 file changed, 1 insertion(+) diff --git a/libavcodec/hw_base_encode.h b/libavcodec/hw_base_encode.h index e528f2013b..ac0cc7b80c 100644 --- a/libavcodec/hw_base_encode.h +++ b/libavcodec/hw_base_encode.h @@ -19,6 +19,7 @@ #ifndef AVCODEC_HW_BASE_ENCODE_H #define AVCODEC_HW_BASE_ENCODE_H +#include "avcodec.h" #include "libavutil/hwcontext.h" #include "libavutil/fifo.h" -- 2.40.1

3 2