November 2025 - ffmpeg-cvslog

[ffmpeg] branch master updated. 81362b319e avcodec/x86/me_cmp: Avoid call on UNIX64
by ffmpeg-git＠ffmpeg.org 26 Nov '25

26 Nov '25

The branch, master has been updated via 81362b319ea7244d8d17110adfa59f10c7e78268 (commit) via 23720df371d7da9ea8074b01753497f7194973f4 (commit) via 6588bf368613a7fdc1970c77030a94f31fcfa970 (commit) from f3346ca6f7d8249738ae0d9544b62be5c51ccfcd (commit) - Log ----------------------------------------------------------------- commit 81362b319ea7244d8d17110adfa59f10c7e78268 Author: Andreas Rheinhardt <andreas.rheinhardt(a)outlook.com> AuthorDate: Tue Nov 25 16:28:51 2025 +0100 Commit: James Almer <jamrial(a)gmail.com> CommitDate: Wed Nov 26 00:01:09 2025 +0000 avcodec/x86/me_cmp: Avoid call on UNIX64 The internal functions for calculating the hadamard difference of two 8x8 blocks have no epilogue on UNIX64, so one can avoid the call altogether by placing the 8x8 function so that it directly falls into the internal function. Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt(a)outlook.com> diff --git a/libavcodec/x86/me_cmp.asm b/libavcodec/x86/me_cmp.asm index 2a196d03bb..3ac8acee2c 100644 --- a/libavcodec/x86/me_cmp.asm +++ b/libavcodec/x86/me_cmp.asm @@ -121,12 +121,8 @@ SECTION .text movd %3, %1 %endmacro -%macro hadamard8_16_wrapper 2 -cglobal hadamard8_diff, 4, 4, %1, %2*mmsize - call hadamard8x8_diff %+ SUFFIX - RET - -cglobal hadamard8_diff16, 5, 6, %1, %2*mmsize +%macro HADAMARD8_DIFF 1 +cglobal hadamard8_diff16, 5, 6, %1, 2*mmsize*ARCH_X86_32 call hadamard8x8_diff %+ SUFFIX mov r5d, eax @@ -151,9 +147,10 @@ cglobal hadamard8_diff16, 5, 6, %1, %2*mmsize .done: mov eax, r5d RET -%endmacro -%macro HADAMARD8_DIFF 1 +cglobal hadamard8_diff, 4, 4, %1, 2*mmsize*ARCH_X86_32 + TAIL_CALL hadamard8x8_diff %+ SUFFIX, 0 + ; r1, r2 and r3 are not clobbered in this function, so 16x16 can ; simply call this 2x2x (and that's why we access rsp+gprsize ; everywhere, which is rsp of calling function) @@ -171,8 +168,6 @@ hadamard8x8_diff %+ SUFFIX: HSUM m0, m1, eax and eax, 0xFFFF ret - -hadamard8_16_wrapper %1, 2*ARCH_X86_32 %endmacro INIT_XMM sse2 commit 23720df371d7da9ea8074b01753497f7194973f4 Author: Andreas Rheinhardt <andreas.rheinhardt(a)outlook.com> AuthorDate: Tue Nov 25 15:45:44 2025 +0100 Commit: James Almer <jamrial(a)gmail.com> CommitDate: Wed Nov 26 00:01:09 2025 +0000 avcodec/me_cmp: Remove MMXEXT hadamard diff functions The SSE2 and SSSE3 functions are now available everywhere, making the MMXEXT functions irrelevant. Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt(a)outlook.com> diff --git a/libavcodec/x86/me_cmp.asm b/libavcodec/x86/me_cmp.asm index e123089ba3..2a196d03bb 100644 --- a/libavcodec/x86/me_cmp.asm +++ b/libavcodec/x86/me_cmp.asm @@ -112,7 +112,6 @@ SECTION .text ; about 100k on extreme inputs. But that's very unlikely to occur in natural video, ; and it's even more unlikely to not have any alternative mvs/modes with lower cost. %macro HSUM 3 -%if cpuflag(sse2) movhlps %2, %1 paddusw %1, %2 pshuflw %2, %1, 0xE @@ -120,35 +119,6 @@ SECTION .text pshuflw %2, %1, 0x1 paddusw %1, %2 movd %3, %1 -%elif cpuflag(mmxext) - pshufw %2, %1, 0xE - paddusw %1, %2 - pshufw %2, %1, 0x1 - paddusw %1, %2 - movd %3, %1 -%elif cpuflag(mmx) - mova %2, %1 - psrlq %1, 32 - paddusw %1, %2 - mova %2, %1 - psrlq %1, 16 - paddusw %1, %2 - movd %3, %1 -%endif -%endmacro - -%macro STORE4 5 - mova [%1+mmsize*0], %2 - mova [%1+mmsize*1], %3 - mova [%1+mmsize*2], %4 - mova [%1+mmsize*3], %5 -%endmacro - -%macro LOAD4 5 - mova %2, [%1+mmsize*0] - mova %3, [%1+mmsize*1] - mova %4, [%1+mmsize*2] - mova %5, [%1+mmsize*3] %endmacro %macro hadamard8_16_wrapper 2 @@ -183,8 +153,10 @@ cglobal hadamard8_diff16, 5, 6, %1, %2*mmsize RET %endmacro -%macro HADAMARD8_DIFF 0-1 -%if cpuflag(sse2) +%macro HADAMARD8_DIFF 1 +; r1, r2 and r3 are not clobbered in this function, so 16x16 can +; simply call this 2x2x (and that's why we access rsp+gprsize +; everywhere, which is rsp of calling function) hadamard8x8_diff %+ SUFFIX: lea r0, [r3*3] DIFF_PIXELS_8 r1, r2, 0, r3, r0, rsp+gprsize @@ -201,60 +173,8 @@ hadamard8x8_diff %+ SUFFIX: ret hadamard8_16_wrapper %1, 2*ARCH_X86_32 -%elif cpuflag(mmx) -ALIGN 16 -; int ff_hadamard8_diff_ ## cpu(MPVEncContext *s, const uint8_t *src1, -; const uint8_t *src2, ptrdiff_t stride, int h) -; r0 = void *s = unused, int h = unused (always 8) -; note how r1, r2 and r3 are not clobbered in this function, so 16x16 -; can simply call this 2x2x (and that's why we access rsp+gprsize -; everywhere, which is rsp of calling func -hadamard8x8_diff %+ SUFFIX: - lea r0, [r3*3] - - ; first 4x8 pixels - DIFF_PIXELS_8 r1, r2, 0, r3, r0, rsp+gprsize+0x60 - HADAMARD8 - mova [rsp+gprsize+0x60], m7 - TRANSPOSE4x4W 0, 1, 2, 3, 7 - STORE4 rsp+gprsize, m0, m1, m2, m3 - mova m7, [rsp+gprsize+0x60] - TRANSPOSE4x4W 4, 5, 6, 7, 0 - STORE4 rsp+gprsize+0x40, m4, m5, m6, m7 - - ; second 4x8 pixels - DIFF_PIXELS_8 r1, r2, 4, r3, r0, rsp+gprsize+0x60 - HADAMARD8 - mova [rsp+gprsize+0x60], m7 - TRANSPOSE4x4W 0, 1, 2, 3, 7 - STORE4 rsp+gprsize+0x20, m0, m1, m2, m3 - mova m7, [rsp+gprsize+0x60] - TRANSPOSE4x4W 4, 5, 6, 7, 0 - - LOAD4 rsp+gprsize+0x40, m0, m1, m2, m3 - HADAMARD8 - ABS_SUM_8x8_32 rsp+gprsize+0x60 - mova [rsp+gprsize+0x60], m0 - - LOAD4 rsp+gprsize , m0, m1, m2, m3 - LOAD4 rsp+gprsize+0x20, m4, m5, m6, m7 - HADAMARD8 - ABS_SUM_8x8_32 rsp+gprsize - paddusw m0, [rsp+gprsize+0x60] - - HSUM m0, m1, eax - and rax, 0xFFFF - ret - -hadamard8_16_wrapper 0, 13 -%endif %endmacro -%if HAVE_ALIGNED_STACK == 0 -INIT_MMX mmxext -HADAMARD8_DIFF -%endif - INIT_XMM sse2 %if ARCH_X86_64 %define ABS_SUM_8x8 ABS_SUM_8x8_64 diff --git a/libavcodec/x86/me_cmp_init.c b/libavcodec/x86/me_cmp_init.c index 35abbbf7f5..3a8b46f4e1 100644 --- a/libavcodec/x86/me_cmp_init.c +++ b/libavcodec/x86/me_cmp_init.c @@ -77,7 +77,6 @@ int ff_vsad16u_approx_sse2(MPVEncContext *v, const uint8_t *pix1, const uint8_t int ff_hadamard8_diff16_ ## cpu(MPVEncContext *s, const uint8_t *src1, \ const uint8_t *src2, ptrdiff_t stride, int h); -hadamard_func(mmxext) hadamard_func(sse2) hadamard_func(ssse3) @@ -116,11 +115,6 @@ av_cold void ff_me_cmp_init_x86(MECmpContext *c, AVCodecContext *avctx) int cpu_flags = av_get_cpu_flags(); if (EXTERNAL_MMXEXT(cpu_flags)) { -#if !HAVE_ALIGNED_STACK - c->hadamard8_diff[0] = ff_hadamard8_diff16_mmxext; - c->hadamard8_diff[1] = ff_hadamard8_diff_mmxext; -#endif - c->sad[1] = ff_sad8_mmxext; c->pix_abs[1][0] = ff_sad8_mmxext; commit 6588bf368613a7fdc1970c77030a94f31fcfa970 Author: Andreas Rheinhardt <andreas.rheinhardt(a)outlook.com> AuthorDate: Tue Nov 25 15:29:19 2025 +0100 Commit: James Almer <jamrial(a)gmail.com> CommitDate: Wed Nov 26 00:01:09 2025 +0000 avcodec/x86/me_cmp: Avoid manual stack handling Use x86inc's stack alignment feature instead of allocating the stack manually*; this means that this code now also automatically supports unaligned stacks, so that the SSE2 and SSSE3 functions will now be available everywhere. *: The code for this was also buggy: It resulted in the stack pointer to be 4 mod 8 for x64 for the mmxext version before it was disabled in 542765ce3eccbca587d54262a512cbdb1407230d, because it hardcode 4 instead of using gprsize. Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt(a)outlook.com> diff --git a/libavcodec/x86/me_cmp.asm b/libavcodec/x86/me_cmp.asm index 4545eae276..e123089ba3 100644 --- a/libavcodec/x86/me_cmp.asm +++ b/libavcodec/x86/me_cmp.asm @@ -152,23 +152,11 @@ SECTION .text %endmacro %macro hadamard8_16_wrapper 2 -cglobal hadamard8_diff, 4, 4, %1 -%ifndef m8 - %assign pad %2*mmsize-(4+stack_offset&(mmsize-1)) - SUB rsp, pad -%endif +cglobal hadamard8_diff, 4, 4, %1, %2*mmsize call hadamard8x8_diff %+ SUFFIX -%ifndef m8 - ADD rsp, pad -%endif RET -cglobal hadamard8_diff16, 5, 6, %1 -%ifndef m8 - %assign pad %2*mmsize-(4+stack_offset&(mmsize-1)) - SUB rsp, pad -%endif - +cglobal hadamard8_diff16, 5, 6, %1, %2*mmsize call hadamard8x8_diff %+ SUFFIX mov r5d, eax @@ -192,9 +180,6 @@ cglobal hadamard8_diff16, 5, 6, %1 .done: mov eax, r5d -%ifndef m8 - ADD rsp, pad -%endif RET %endmacro @@ -215,7 +200,7 @@ hadamard8x8_diff %+ SUFFIX: and eax, 0xFFFF ret -hadamard8_16_wrapper %1, 3 +hadamard8_16_wrapper %1, 2*ARCH_X86_32 %elif cpuflag(mmx) ALIGN 16 ; int ff_hadamard8_diff_ ## cpu(MPVEncContext *s, const uint8_t *src1, @@ -261,7 +246,7 @@ hadamard8x8_diff %+ SUFFIX: and rax, 0xFFFF ret -hadamard8_16_wrapper 0, 14 +hadamard8_16_wrapper 0, 13 %endif %endmacro diff --git a/libavcodec/x86/me_cmp_init.c b/libavcodec/x86/me_cmp_init.c index d4503eef3b..35abbbf7f5 100644 --- a/libavcodec/x86/me_cmp_init.c +++ b/libavcodec/x86/me_cmp_init.c @@ -146,10 +146,8 @@ av_cold void ff_me_cmp_init_x86(MECmpContext *c, AVCodecContext *avctx) c->pix_abs[0][2] = ff_sad16_y2_sse2; c->pix_abs[0][3] = ff_sad16_xy2_sse2; -#if HAVE_ALIGNED_STACK c->hadamard8_diff[0] = ff_hadamard8_diff16_sse2; c->hadamard8_diff[1] = ff_hadamard8_diff_sse2; -#endif if (avctx->codec_id != AV_CODEC_ID_SNOW) { c->sad[0] = ff_sad16_sse2; @@ -179,10 +177,8 @@ av_cold void ff_me_cmp_init_x86(MECmpContext *c, AVCodecContext *avctx) c->nsse[1] = nsse8_ssse3; c->sum_abs_dctelem = ff_sum_abs_dctelem_ssse3; -#if HAVE_ALIGNED_STACK c->hadamard8_diff[0] = ff_hadamard8_diff16_ssse3; c->hadamard8_diff[1] = ff_hadamard8_diff_ssse3; -#endif } #endif } ----------------------------------------------------------------------- Summary of changes: libavcodec/x86/me_cmp.asm | 116 +++---------------------------------------- libavcodec/x86/me_cmp_init.c | 10 ---- 2 files changed, 8 insertions(+), 118 deletions(-) hooks/post-receive --

1 0

[ffmpeg] branch master updated. f3346ca6f7 avfilter/x86/f_ebur128: only use filter_channels_avx for >= 2 channels
by ffmpeg-git＠ffmpeg.org 25 Nov '25

25 Nov '25

The branch, master has been updated via f3346ca6f7d8249738ae0d9544b62be5c51ccfcd (commit) from a75b15a4ab88ae26427b7dd518983cdc13a4b4b5 (commit) - Log ----------------------------------------------------------------- commit f3346ca6f7d8249738ae0d9544b62be5c51ccfcd Author: Niklas Haas <git(a)haasn.dev> AuthorDate: Mon Nov 17 10:50:54 2025 +0100 Commit: Niklas Haas <ffmpeg(a)haasn.dev> CommitDate: Tue Nov 25 22:13:57 2025 +0000 avfilter/x86/f_ebur128: only use filter_channels_avx for >= 2 channels The approach of this ASM routine is to process two channels at a time using AVX instructions. Obviously, there is no point in doing this if there is only a single channel; in which case the scalar loop would be better. Fixes a performance regression when filtering mono audio on certain CPUs, notably e.g. the Intel N100. diff --git a/libavfilter/x86/f_ebur128_init.c b/libavfilter/x86/f_ebur128_init.c index 4a7109746d..6d43b4887f 100644 --- a/libavfilter/x86/f_ebur128_init.c +++ b/libavfilter/x86/f_ebur128_init.c @@ -33,7 +33,8 @@ av_cold void ff_ebur128_init_x86(EBUR128DSPContext *dsp, int nb_channels) int cpu_flags = av_get_cpu_flags(); if (ARCH_X86_64 && EXTERNAL_AVX(cpu_flags)) { - dsp->filter_channels = ff_ebur128_filter_channels_avx; + if (nb_channels >= 2) + dsp->filter_channels = ff_ebur128_filter_channels_avx; if (nb_channels == 2) dsp->find_peak = ff_ebur128_find_peak_2ch_avx; } ----------------------------------------------------------------------- Summary of changes: libavfilter/x86/f_ebur128_init.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) hooks/post-receive --

1 0

[ffmpeg] branch master updated. a75b15a4ab avfilter/vf_drawvg: round color values to avoid differences on some platforms
by ffmpeg-git＠ffmpeg.org 25 Nov '25

25 Nov '25

The branch, master has been updated via a75b15a4ab88ae26427b7dd518983cdc13a4b4b5 (commit) from 1d84ab331cb75a1c21d2705b24448b9971380697 (commit) - Log ----------------------------------------------------------------- commit a75b15a4ab88ae26427b7dd518983cdc13a4b4b5 Author: Kacper Michajłow <kasper93(a)gmail.com> AuthorDate: Tue Nov 25 18:06:33 2025 +0100 Commit: Kacper Michajłow <kasper93(a)gmail.com> CommitDate: Tue Nov 25 22:32:50 2025 +0100 avfilter/vf_drawvg: round color values to avoid differences on some platforms This ensures consistent color conversion between double and u8 and guarantees that values remain consistent across different platforms, especially when x87 math is used. Note that libcairo also performs rounding internally when converting doubles to integers, see _cairo_color_double_to_short(). Fixes: fate-filter-drawvg-interpreter Signed-off-by: Kacper Michajłow <kasper93(a)gmail.com> diff --git a/libavfilter/tests/drawvg.c b/libavfilter/tests/drawvg.c index 2d151bf4d2..9fb233d969 100644 --- a/libavfilter/tests/drawvg.c +++ b/libavfilter/tests/drawvg.c @@ -167,7 +167,7 @@ void cairo_set_source(cairo_t *cr, cairo_pattern_t *source) { printf("%s", __func__); #define PRINT_COLOR(prefix) \ - printf(prefix "#%02x%02x%02x%02x", (int)(r*255), (int)(g*255), (int)(b*255), (int)(a*255)) + printf(prefix "#%02lx%02lx%02lx%02lx", lround(r*255), lround(g*255), lround(b*255), lround(a*255)) switch (cairo_pattern_get_type(source)) { case CAIRO_PATTERN_TYPE_SOLID: diff --git a/libavfilter/vf_drawvg.c b/libavfilter/vf_drawvg.c index fd9270ee13..5d3008084c 100644 --- a/libavfilter/vf_drawvg.c +++ b/libavfilter/vf_drawvg.c @@ -1981,7 +1981,7 @@ static int vgs_eval( b = numerics[3]; } - #define C(v, o) ((uint32_t)(av_clipd(v, 0, 1) * 255) << o) + #define C(v, o) ((uint32_t)lround(av_clipd(v, 0, 1) * 255) << o) state->vars[user_var] = (double)( C(r, 24) diff --git a/tests/ref/fate/filter-drawvg-interpreter b/tests/ref/fate/filter-drawvg-interpreter index 21c6ccd848..3fc33e9c07 100644 --- a/tests/ref/fate/filter-drawvg-interpreter +++ b/tests/ref/fate/filter-drawvg-interpreter @@ -64,16 +64,16 @@ cairo_set_dash [ -1.0 ] 4.0 cairo_set_dash [ ] 0.0 cairo_move_to 1.0 2.0 cairo_rel_line_to -1.0 -2.0 -cairo_set_source #19334c66 +cairo_set_source #1a334d66 cairo_set_fill_rule 0 cairo_fill -cairo_set_source #475b3d66 +cairo_set_source #475c3d66 cairo_set_fill_rule 0 cairo_fill -cairo_set_source #7f99b2cc +cairo_set_source #8099b3cc cairo_set_fill_rule 0 cairo_fill -cairo_set_source #a8d7efe5 +cairo_set_source #a8d8f0e6 cairo_set_fill_rule 0 cairo_fill cairo_rel_line_to 1.0 3.0 ----------------------------------------------------------------------- Summary of changes: libavfilter/tests/drawvg.c | 2 +- libavfilter/vf_drawvg.c | 2 +- tests/ref/fate/filter-drawvg-interpreter | 8 ++++---- 3 files changed, 6 insertions(+), 6 deletions(-) hooks/post-receive --

1 0

[ffmpeg] branch master updated. 1d84ab331c vulkan/prores: Adopt the same IDCT routine as the prores-raw hwaccel
by ffmpeg-git＠ffmpeg.org 25 Nov '25

25 Nov '25

The branch, master has been updated via 1d84ab331cb75a1c21d2705b24448b9971380697 (commit) from 2b221fdb4a1bdc556225eb15688dc112b81f2246 (commit) - Log ----------------------------------------------------------------- commit 1d84ab331cb75a1c21d2705b24448b9971380697 Author: averne <averne381(a)gmail.com> AuthorDate: Sun Nov 2 20:23:28 2025 +0100 Commit: Lynne <dev(a)lynne.ee> CommitDate: Tue Nov 25 17:54:56 2025 +0000 vulkan/prores: Adopt the same IDCT routine as the prores-raw hwaccel The added rounding at the final output conforms to the SMPTE document and reduces the deviation against the software decoder. diff --git a/libavcodec/vulkan/prores_idct.comp b/libavcodec/vulkan/prores_idct.comp index 4b39b3d8ae..05ba8e4967 100644 --- a/libavcodec/vulkan/prores_idct.comp +++ b/libavcodec/vulkan/prores_idct.comp @@ -37,47 +37,77 @@ void put_px(uint tex_idx, ivec2 pos, uint v) #endif } +const float idct_8x8_scales[] = { + 0.353553390593274f, // cos(4 * pi/16) / 2 + 0.490392640201615f, // cos(1 * pi/16) / 2 + 0.461939766255643f, // cos(2 * pi/16) / 2 + 0.415734806151273f, // cos(3 * pi/16) / 2 + 0.353553390593274f, // cos(4 * pi/16) / 2 + 0.277785116509801f, // cos(5 * pi/16) / 2 + 0.191341716182545f, // cos(6 * pi/16) / 2 + 0.097545161008064f, // cos(7 * pi/16) / 2 +}; + /* 7.4 Inverse Transform */ void idct(uint block, uint offset, uint stride) { - float c0 = blocks[block][0*stride + offset]; - float c1 = blocks[block][1*stride + offset]; - float c2 = blocks[block][2*stride + offset]; - float c3 = blocks[block][3*stride + offset]; - float c4 = blocks[block][4*stride + offset]; - float c5 = blocks[block][5*stride + offset]; - float c6 = blocks[block][6*stride + offset]; - float c7 = blocks[block][7*stride + offset]; - - float tmp1 = c6 * 1.4142134189605712891 + (c2 - c6); - float tmp2 = c6 * 1.4142134189605712891 - (c2 - c6); - - float a1 = (c0 + c4) * 0.35355341434478759766 + tmp1 * 0.46193981170654296875; - float a4 = (c0 + c4) * 0.35355341434478759766 - tmp1 * 0.46193981170654296875; - - float a3 = (c0 - c4) * 0.35355341434478759766 + tmp2 * 0.19134169816970825195; - float a2 = (c0 - c4) * 0.35355341434478759766 - tmp2 * 0.19134169816970825195; - - float tmp3 = (c3 - c5) * 0.70710682868957519531 + c7; - float tmp4 = (c3 - c5) * 0.70710682868957519531 - c7; - - float tmp5 = (c5 - c7) * 1.4142134189605712891 + (c5 - c7) + (c1 - c3); - float tmp6 = (c5 - c7) * -1.4142134189605712891 + (c5 - c7) + (c1 - c3); - - float m1 = tmp3 * 2.6131260395050048828 + tmp5; - float m4 = tmp3 * -2.6131260395050048828 + tmp5; - - float m2 = tmp4 * 1.0823919773101806641 + tmp6; - float m3 = tmp4 * -1.0823919773101806641 + tmp6; - - blocks[block][0*stride + offset] = m1 * 0.49039259552955627441 + a1; - blocks[block][7*stride + offset] = m1 * -0.49039259552955627441 + a1; - blocks[block][1*stride + offset] = m2 * 0.41573479771614074707 + a2; - blocks[block][6*stride + offset] = m2 * -0.41573479771614074707 + a2; - blocks[block][2*stride + offset] = m3 * 0.27778509259223937988 + a3; - blocks[block][5*stride + offset] = m3 * -0.27778509259223937988 + a3; - blocks[block][3*stride + offset] = m4 * 0.097545139491558074951 + a4; - blocks[block][4*stride + offset] = m4 * -0.097545139491558074951 + a4; + float t0, t1, t2, t3, t4, t5, t6, t7, u8; + float u0, u1, u2, u3, u4, u5, u6, u7; + + /* Input */ + t0 = blocks[block][0*stride + offset]; + u4 = blocks[block][1*stride + offset]; + t2 = blocks[block][2*stride + offset]; + u6 = blocks[block][3*stride + offset]; + t1 = blocks[block][4*stride + offset]; + u5 = blocks[block][5*stride + offset]; + t3 = blocks[block][6*stride + offset]; + u7 = blocks[block][7*stride + offset]; + + /* Embedded scaled inverse 4-point Type-II DCT */ + u0 = t0 + t1; + u1 = t0 - t1; + u3 = t2 + t3; + u2 = (t2 - t3)*(1.4142135623730950488016887242097f) - u3; + t0 = u0 + u3; + t3 = u0 - u3; + t1 = u1 + u2; + t2 = u1 - u2; + + /* Embedded scaled inverse 4-point Type-IV DST */ + t5 = u5 + u6; + t6 = u5 - u6; + t7 = u4 + u7; + t4 = u4 - u7; + u7 = t7 + t5; + u5 = (t7 - t5)*(1.4142135623730950488016887242097f); + u8 = (t4 + t6)*(1.8477590650225735122563663787936f); + u4 = u8 - t4*(1.0823922002923939687994464107328f); + u6 = u8 - t6*(2.6131259297527530557132863468544f); + t7 = u7; + t6 = t7 - u6; + t5 = t6 + u5; + t4 = t5 - u4; + + /* Butterflies */ + u0 = t0 + t7; + u7 = t0 - t7; + u6 = t1 + t6; + u1 = t1 - t6; + u2 = t2 + t5; + u5 = t2 - t5; + u4 = t3 + t4; + u3 = t3 - t4; + + /* Output */ + blocks[block][0*stride + offset] = u0; + blocks[block][1*stride + offset] = u1; + blocks[block][2*stride + offset] = u2; + blocks[block][3*stride + offset] = u3; + blocks[block][4*stride + offset] = u4; + blocks[block][5*stride + offset] = u5; + blocks[block][6*stride + offset] = u6; + blocks[block][7*stride + offset] = u7; } void main(void) @@ -90,7 +120,7 @@ void main(void) /* Coalesced load of DCT coeffs in shared memory, inverse quantization */ if (act) { /** - * According to spec indexing an array in push constant memory with + * According to the VK spec indexing an array in push constant memory with * a non-dynamically uniform value is illegal ($15.9.1 in v1.4.326), * so copy the whole matrix locally. */ @@ -101,8 +131,9 @@ void main(void) int qscale = qidx > 128 ? (qidx - 96) << 2 : qidx; [[unroll]] for (uint i = 0; i < 8; ++i) { - int v = sign_extend(int(get_px(comp, ivec2(gid.x, (gid.y << 3) + i))), 16); - blocks[block][i * 9 + idx] = float(v * qscale * int(qmat[(i << 3) + idx])); + int c = sign_extend(int(get_px(comp, ivec2(gid.x, (gid.y << 3) + i))), 16); + float v = float(c * qscale * int(qmat[(i << 3) + idx])); + blocks[block][i * 9 + idx] = v * idct_8x8_scales[idx] * idct_8x8_scales[i]; } } @@ -121,7 +152,7 @@ void main(void) barrier(); if (act) { [[unroll]] for (uint i = 0; i < 8; ++i) { - float v = blocks[block][i * 9 + idx] * fact + off; + float v = round(blocks[block][i * 9 + idx] * fact + off); put_px(comp, ivec2(gid.x, (gid.y << 3) + i), clamp(int(v), 0, maxv)); } } ----------------------------------------------------------------------- Summary of changes: libavcodec/vulkan/prores_idct.comp | 115 +++++++++++++++++++++++-------------- 1 file changed, 73 insertions(+), 42 deletions(-) hooks/post-receive --

1 0

[ffmpeg] branch master updated. 2b221fdb4a avfilter/zscale: add support for resize filter spline64
by ffmpeg-git＠ffmpeg.org 25 Nov '25

25 Nov '25

The branch, master has been updated via 2b221fdb4a1bdc556225eb15688dc112b81f2246 (commit) from 0486ad61ceb71b7a45976e6e45a4d2a2813aef75 (commit) - Log ----------------------------------------------------------------- commit 2b221fdb4a1bdc556225eb15688dc112b81f2246 Author: Gyan Doshi <ffmpeg(a)gyani.pro> AuthorDate: Tue Nov 25 12:41:37 2025 +0530 Commit: Gyan Doshi <ffmpeg(a)gyani.pro> CommitDate: Tue Nov 25 12:42:41 2025 +0530 avfilter/zscale: add support for resize filter spline64 Fixes #20928 diff --git a/doc/filters.texi b/doc/filters.texi index 7605812428..168ea0d2da 100644 --- a/doc/filters.texi +++ b/doc/filters.texi @@ -26504,6 +26504,7 @@ Possible values are: @item bicubic @item spline16 @item spline36 +@item spline64 @item lanczos @end table diff --git a/libavfilter/vf_zscale.c b/libavfilter/vf_zscale.c index 9ea50133c6..96cbd3c4b2 100644 --- a/libavfilter/vf_zscale.c +++ b/libavfilter/vf_zscale.c @@ -940,13 +940,14 @@ static const AVOption zscale_options[] = { { "ordered", 0, 0, AV_OPT_TYPE_CONST, {.i64 = ZIMG_DITHER_ORDERED}, 0, 0, FLAGS, .unit = "dither" }, { "random", 0, 0, AV_OPT_TYPE_CONST, {.i64 = ZIMG_DITHER_RANDOM}, 0, 0, FLAGS, .unit = "dither" }, { "error_diffusion", 0, 0, AV_OPT_TYPE_CONST, {.i64 = ZIMG_DITHER_ERROR_DIFFUSION}, 0, 0, FLAGS, .unit = "dither" }, - { "filter", "set filter type", OFFSET(filter), AV_OPT_TYPE_INT, {.i64 = ZIMG_RESIZE_BILINEAR}, 0, ZIMG_RESIZE_LANCZOS, FLAGS, .unit = "filter" }, - { "f", "set filter type", OFFSET(filter), AV_OPT_TYPE_INT, {.i64 = ZIMG_RESIZE_BILINEAR}, 0, ZIMG_RESIZE_LANCZOS, FLAGS, .unit = "filter" }, + { "filter", "set filter type", OFFSET(filter), AV_OPT_TYPE_INT, {.i64 = ZIMG_RESIZE_BILINEAR}, 0, ZIMG_RESIZE_SPLINE64, FLAGS, .unit = "filter" }, + { "f", "set filter type", OFFSET(filter), AV_OPT_TYPE_INT, {.i64 = ZIMG_RESIZE_BILINEAR}, 0, ZIMG_RESIZE_SPLINE64, FLAGS, .unit = "filter" }, { "point", 0, 0, AV_OPT_TYPE_CONST, {.i64 = ZIMG_RESIZE_POINT}, 0, 0, FLAGS, .unit = "filter" }, { "bilinear", 0, 0, AV_OPT_TYPE_CONST, {.i64 = ZIMG_RESIZE_BILINEAR}, 0, 0, FLAGS, .unit = "filter" }, { "bicubic", 0, 0, AV_OPT_TYPE_CONST, {.i64 = ZIMG_RESIZE_BICUBIC}, 0, 0, FLAGS, .unit = "filter" }, { "spline16", 0, 0, AV_OPT_TYPE_CONST, {.i64 = ZIMG_RESIZE_SPLINE16}, 0, 0, FLAGS, .unit = "filter" }, { "spline36", 0, 0, AV_OPT_TYPE_CONST, {.i64 = ZIMG_RESIZE_SPLINE36}, 0, 0, FLAGS, .unit = "filter" }, + { "spline64", 0, 0, AV_OPT_TYPE_CONST, {.i64 = ZIMG_RESIZE_SPLINE64}, 0, 0, FLAGS, .unit = "filter" }, { "lanczos", 0, 0, AV_OPT_TYPE_CONST, {.i64 = ZIMG_RESIZE_LANCZOS}, 0, 0, FLAGS, .unit = "filter" }, { "out_range", "set color range", OFFSET(range), AV_OPT_TYPE_INT, {.i64 = -1}, -1, ZIMG_RANGE_FULL, FLAGS, .unit = "range" }, { "range", "set color range", OFFSET(range), AV_OPT_TYPE_INT, {.i64 = -1}, -1, ZIMG_RANGE_FULL, FLAGS, .unit = "range" }, ----------------------------------------------------------------------- Summary of changes: doc/filters.texi | 1 + libavfilter/vf_zscale.c | 5 +++-- 2 files changed, 4 insertions(+), 2 deletions(-) hooks/post-receive --

1 0

[ffmpeg] branch master updated. 0486ad61ce configure: replace openssl header check with 1.1.1 API
by ffmpeg-git＠ffmpeg.org 24 Nov '25

24 Nov '25

The branch, master has been updated via 0486ad61ceb71b7a45976e6e45a4d2a2813aef75 (commit) from f790de2a878861a34820d7d8ab64badd192331f5 (commit) - Log ----------------------------------------------------------------- commit 0486ad61ceb71b7a45976e6e45a4d2a2813aef75 Author: Jack Lau <jacklau1222gm(a)gmail.com> AuthorDate: Mon Nov 17 13:32:05 2025 +0800 Commit: James Almer <jamrial(a)gmail.com> CommitDate: Mon Nov 24 21:29:45 2025 +0000 configure: replace openssl header check with 1.1.1 API Fix #20571 Avoid build errors with openssl forks (like libressl) that lack some APIs. This patch replace header check for OPENSSL_init_ssl (was added in 1.1.0) with the OpenSSL 1.1.1 new API DTLS_get_data_mtu. Signed-off-by: Jack Lau <jacklau1222gm(a)gmail.com> diff --git a/configure b/configure index 7d6061b55c..99734e9d03 100755 --- a/configure +++ b/configure @@ -7353,12 +7353,12 @@ enabled omx_rpi && { test_code cc OMX_Core.h OMX_IndexConfigBrcmVideoR enabled omx && require_headers OMX_Core.h && \ warn "The OpenMAX encoders are deprecated and will be removed in future versions" -enabled openssl && { { check_pkg_config openssl "openssl >= 3.0.0" openssl/ssl.h OPENSSL_init_ssl && +enabled openssl && { { check_pkg_config openssl "openssl >= 3.0.0" openssl/ssl.h DTLS_get_data_mtu && { enabled gplv3 || ! enabled gpl || enabled nonfree || die "ERROR: OpenSSL >=3.0.0 requires --enable-version3"; }; } || { enabled gpl && ! enabled nonfree && die "ERROR: OpenSSL <3.0.0 is incompatible with the gpl"; } || - check_pkg_config openssl "openssl >= 1.1.1" openssl/ssl.h OPENSSL_init_ssl || - check_lib openssl openssl/ssl.h OPENSSL_init_ssl -lssl -lcrypto || - check_lib openssl openssl/ssl.h OPENSSL_init_ssl -lssl -lcrypto -lws2_32 -lgdi32 || + check_pkg_config openssl "openssl >= 1.1.1" openssl/ssl.h DTLS_get_data_mtu || + check_lib openssl openssl/ssl.h DTLS_get_data_mtu -lssl -lcrypto || + check_lib openssl openssl/ssl.h DTLS_get_data_mtu -lssl -lcrypto -lws2_32 -lgdi32 || die "ERROR: openssl (>= 1.1.1) not found"; } enabled pocketsphinx && require_pkg_config pocketsphinx pocketsphinx pocketsphinx/pocketsphinx.h ps_init enabled rkmpp && { require_pkg_config rkmpp rockchip_mpp rockchip/rk_mpi.h mpp_create && ----------------------------------------------------------------------- Summary of changes: configure | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) hooks/post-receive --

1 0

[ffmpeg] branch master updated. f790de2a87 aarch64/vvc: Optimisations of put_luma_h() functions for 10/12-bit
by ffmpeg-git＠ffmpeg.org 24 Nov '25

24 Nov '25

The branch, master has been updated via f790de2a878861a34820d7d8ab64badd192331f5 (commit) from 8b71eeb72e74545d10d6e69dc6be612e81148791 (commit) - Log ----------------------------------------------------------------- commit f790de2a878861a34820d7d8ab64badd192331f5 Author: Georgii Zagoruiko <george.zaguri(a)gmail.com> AuthorDate: Mon Nov 24 19:51:26 2025 +0000 Commit: Martin Storsjö <martin(a)martin.st> CommitDate: Mon Nov 24 21:22:55 2025 +0000 aarch64/vvc: Optimisations of put_luma_h() functions for 10/12-bit RPi4 (auto-vectorisation is turned on) put_luma_h_10_4x4_c: 282.8 ( 1.00x) put_luma_h_10_8x8_c: 1069.5 ( 1.00x) put_luma_h_10_8x8_neon: 207.5 ( 5.15x) put_luma_h_10_16x16_c: 1999.6 ( 1.00x) put_luma_h_10_16x16_neon: 777.5 ( 2.57x) put_luma_h_10_32x32_c: 6612.9 ( 1.00x) put_luma_h_10_32x32_neon: 3201.6 ( 2.07x) put_luma_h_10_64x64_c: 25059.0 ( 1.00x) put_luma_h_10_64x64_neon: 13623.5 ( 1.84x) put_luma_h_10_128x128_c: 91310.1 ( 1.00x) put_luma_h_10_128x128_neon: 50358.3 ( 1.81x) put_luma_h_12_4x4_c: 282.1 ( 1.00x) put_luma_h_12_8x8_c: 1068.4 ( 1.00x) put_luma_h_12_8x8_neon: 207.7 ( 5.14x) put_luma_h_12_16x16_c: 1998.0 ( 1.00x) put_luma_h_12_16x16_neon: 777.5 ( 2.57x) put_luma_h_12_32x32_c: 6612.0 ( 1.00x) put_luma_h_12_32x32_neon: 3201.6 ( 2.07x) put_luma_h_12_64x64_c: 25036.8 ( 1.00x) put_luma_h_12_64x64_neon: 13595.1 ( 1.84x) put_luma_h_12_128x128_c: 91305.8 ( 1.00x) put_luma_h_12_128x128_neon: 50359.7 ( 1.81x) Apple M2 Air (auto-vectorisation is turned on) put_luma_h_10_4x4_c: 0.3 ( 1.00x) put_luma_h_10_8x8_c: 1.0 ( 1.00x) put_luma_h_10_8x8_neon: 0.4 ( 2.59x) put_luma_h_10_16x16_c: 2.9 ( 1.00x) put_luma_h_10_16x16_neon: 1.4 ( 2.01x) put_luma_h_10_32x32_c: 9.4 ( 1.00x) put_luma_h_10_32x32_neon: 5.8 ( 1.62x) put_luma_h_10_64x64_c: 35.6 ( 1.00x) put_luma_h_10_64x64_neon: 23.6 ( 1.51x) put_luma_h_10_128x128_c: 131.1 ( 1.00x) put_luma_h_10_128x128_neon: 92.6 ( 1.42x) put_luma_h_12_4x4_c: 0.3 ( 1.00x) put_luma_h_12_8x8_c: 1.0 ( 1.00x) put_luma_h_12_8x8_neon: 0.4 ( 2.58x) put_luma_h_12_16x16_c: 2.9 ( 1.00x) put_luma_h_12_16x16_neon: 1.4 ( 2.00x) put_luma_h_12_32x32_c: 9.4 ( 1.00x) put_luma_h_12_32x32_neon: 5.8 ( 1.61x) put_luma_h_12_64x64_c: 35.3 ( 1.00x) put_luma_h_12_64x64_neon: 23.3 ( 1.52x) put_luma_h_12_128x128_c: 131.2 ( 1.00x) put_luma_h_12_128x128_neon: 92.4 ( 1.42x) diff --git a/libavcodec/aarch64/vvc/dsp_init.c b/libavcodec/aarch64/vvc/dsp_init.c index b7dc1d89f8..aa75d22b78 100644 --- a/libavcodec/aarch64/vvc/dsp_init.c +++ b/libavcodec/aarch64/vvc/dsp_init.c @@ -30,6 +30,18 @@ #define BDOF_BLOCK_SIZE 16 #define BDOF_MIN_BLOCK_SIZE 4 +void ff_vvc_put_luma_h8_10_neon(int16_t *dst, const uint8_t *_src, const ptrdiff_t _src_stride, + const int height, const int8_t *hf, const int8_t *vf, const int width); +void ff_vvc_put_luma_h16_10_neon(int16_t *dst, const uint8_t *_src, const ptrdiff_t _src_stride, + const int height, const int8_t *hf, const int8_t *vf, const int width); +void ff_vvc_put_luma_h_x16_10_neon(int16_t *dst, const uint8_t *_src, const ptrdiff_t _src_stride, + const int height, const int8_t *hf, const int8_t *vf, const int width); +void ff_vvc_put_luma_h8_12_neon(int16_t *dst, const uint8_t *_src, const ptrdiff_t _src_stride, + const int height, const int8_t *hf, const int8_t *vf, const int width); +void ff_vvc_put_luma_h16_12_neon(int16_t *dst, const uint8_t *_src, const ptrdiff_t _src_stride, + const int height, const int8_t *hf, const int8_t *vf, const int width); +void ff_vvc_put_luma_h_x16_12_neon(int16_t *dst, const uint8_t *_src, const ptrdiff_t _src_stride, + const int height, const int8_t *hf, const int8_t *vf, const int width); void ff_alf_classify_sum_neon(int *sum0, int *sum1, int16_t *grad, uint32_t gshift, uint32_t steps); @@ -245,6 +257,11 @@ void ff_vvc_dsp_init_aarch64(VVCDSPContext *const c, const int bd) c->inter.dmvr[0][1] = ff_vvc_dmvr_h_10_neon; c->inter.dmvr[1][1] = ff_vvc_dmvr_hv_10_neon; c->inter.apply_bdof = ff_vvc_apply_bdof_10_neon; + c->inter.put[0][2][0][1] = ff_vvc_put_luma_h8_10_neon; + c->inter.put[0][3][0][1] = ff_vvc_put_luma_h16_10_neon; + c->inter.put[0][4][0][1] = + c->inter.put[0][5][0][1] = + c->inter.put[0][6][0][1] = ff_vvc_put_luma_h_x16_10_neon; c->alf.filter[LUMA] = alf_filter_luma_10_neon; c->alf.filter[CHROMA] = alf_filter_chroma_10_neon; @@ -256,6 +273,11 @@ void ff_vvc_dsp_init_aarch64(VVCDSPContext *const c, const int bd) c->inter.dmvr[0][1] = ff_vvc_dmvr_h_12_neon; c->inter.dmvr[1][1] = ff_vvc_dmvr_hv_12_neon; c->inter.apply_bdof = ff_vvc_apply_bdof_12_neon; + c->inter.put[0][2][0][1] = ff_vvc_put_luma_h8_12_neon; + c->inter.put[0][3][0][1] = ff_vvc_put_luma_h16_12_neon; + c->inter.put[0][4][0][1] = + c->inter.put[0][5][0][1] = + c->inter.put[0][6][0][1] = ff_vvc_put_luma_h_x16_12_neon; c->alf.filter[LUMA] = alf_filter_luma_12_neon; c->alf.filter[CHROMA] = alf_filter_chroma_12_neon; diff --git a/libavcodec/aarch64/vvc/inter.S b/libavcodec/aarch64/vvc/inter.S index a874edf889..41444ec44c 100644 --- a/libavcodec/aarch64/vvc/inter.S +++ b/libavcodec/aarch64/vvc/inter.S @@ -1713,3 +1713,122 @@ endfunc #undef GRADIENT_V1_OFFSET #undef VX_OFFSET #undef VY_OFFSET + +#define VVC_MAX_PB_SIZE 128 + +.macro put_luma_h_x8_vector_filter shift + // 8 bytes from hf loaded to v0.8h + // 32 bytes from _src loaded to v20.8h & v21.8h where v21.8h is loaded for shift to v1.8h,..,v6.8h,v17.8h + // v24.4h & v25.4h are output vectors to store + ext v1.16b, v20.16b, v21.16b, #2 + ext v2.16b, v20.16b, v21.16b, #4 + ext v3.16b, v20.16b, v21.16b, #6 + ext v4.16b, v20.16b, v21.16b, #8 + ext v5.16b, v20.16b, v21.16b, #10 + ext v6.16b, v20.16b, v21.16b, #12 + ext v17.16b, v20.16b, v21.16b, #14 + smull v24.4s, v20.4h, v0.h[0] + smull2 v25.4s, v20.8h, v0.h[0] + smlal v24.4s, v1.4h, v0.h[1] + smlal2 v25.4s, v1.8h, v0.h[1] + smlal v24.4s, v2.4h, v0.h[2] + smlal2 v25.4s, v2.8h, v0.h[2] + smlal v24.4s, v3.4h, v0.h[3] + smlal2 v25.4s, v3.8h, v0.h[3] + smlal v24.4s, v4.4h, v0.h[4] + smlal2 v25.4s, v4.8h, v0.h[4] + smlal v24.4s, v5.4h, v0.h[5] + smlal2 v25.4s, v5.8h, v0.h[5] + smlal v24.4s, v6.4h, v0.h[6] + smlal2 v25.4s, v6.8h, v0.h[6] + smlal v24.4s, v17.4h, v0.h[7] + smlal2 v25.4s, v17.8h, v0.h[7] + sqshrn v24.4h, v24.4s, #(\shift) + sqshrn v25.4h, v25.4s, #(\shift) +.endm + +.macro put_luma_h8_xx_neon shift + mov x9, #(VVC_MAX_PB_SIZE * 2) + ld1 {v0.8b}, [x4] + sub x1, x1, #6 + sxtl v0.8h, v0.8b +1: + ld1 {v20.8h, v21.8h}, [x1], x2 + put_luma_h_x8_vector_filter \shift + subs w3, w3, #1 + st1 {v24.4h, v25.4h}, [x0], x9 + b.gt 1b + ret +.endm + +.macro put_luma_h16_xx_neon shift + mov x9, #(VVC_MAX_PB_SIZE * 2) + ld1 {v0.8b}, [x4] + sub x9, x9, #16 + sub x1, x1, #6 + sxtl v0.8h, v0.8b +1: + ld1 {v20.8h, v21.8h, v22.8h}, [x1], x2 + put_luma_h_x8_vector_filter \shift + mov v20.16b, v21.16b + mov v21.16b, v22.16b + st1 {v24.4h, v25.4h}, [x0], #16 + put_luma_h_x8_vector_filter \shift + subs w3, w3, #1 + st1 {v24.4h, v25.4h}, [x0], x9 + b.gt 1b + ret +.endm + +.macro put_luma_h_x16_xx_neon shift + mov x9, #(VVC_MAX_PB_SIZE * 2) + ld1 {v0.8b}, [x4] + sub x9, x9, w6, uxtw #1 + sub x2, x2, w6, uxtw #1 + sxtl v0.8h, v0.8b + sub x1, x1, #6 + sub x2, x2, #16 +1: + ld1 {v20.8h}, [x1], #16 + mov w8, w6 +2: + ld1 {v21.8h, v22.8h}, [x1], #32 + put_luma_h_x8_vector_filter \shift + mov v20.16b, v21.16b + mov v21.16b, v22.16b + st1 {v24.4h, v25.4h}, [x0], #16 + put_luma_h_x8_vector_filter \shift + mov v20.16b, v21.16b + subs w8, w8, #16 + st1 {v24.4h, v25.4h}, [x0], #16 + b.gt 2b + subs w3, w3, #1 + add x0, x0, x9 + add x1, x1, x2 + b.gt 1b + ret +.endm + +function ff_vvc_put_luma_h8_10_neon, export=1 + put_luma_h8_xx_neon 2 +endfunc + +function ff_vvc_put_luma_h8_12_neon, export=1 + put_luma_h8_xx_neon 4 +endfunc + +function ff_vvc_put_luma_h16_10_neon, export=1 + put_luma_h16_xx_neon 2 +endfunc + +function ff_vvc_put_luma_h16_12_neon, export=1 + put_luma_h16_xx_neon 4 +endfunc + +function ff_vvc_put_luma_h_x16_10_neon, export=1 + put_luma_h_x16_xx_neon 2 +endfunc + +function ff_vvc_put_luma_h_x16_12_neon, export=1 + put_luma_h_x16_xx_neon 4 +endfunc ----------------------------------------------------------------------- Summary of changes: libavcodec/aarch64/vvc/dsp_init.c | 22 +++++++ libavcodec/aarch64/vvc/inter.S | 119 ++++++++++++++++++++++++++++++++++++++ 2 files changed, 141 insertions(+) hooks/post-receive --

1 0

[ffmpeg] branch master updated. 8b71eeb72e tools: Make indent_arm_assembly.pl able to reformat a file in place
by ffmpeg-git＠ffmpeg.org 24 Nov '25

24 Nov '25

The branch, master has been updated via 8b71eeb72e74545d10d6e69dc6be612e81148791 (commit) from 140b4f28c303abc531daeac4eb42dd014cdb5afb (commit) - Log ----------------------------------------------------------------- commit 8b71eeb72e74545d10d6e69dc6be612e81148791 Author: Martin Storsjö <martin(a)martin.st> AuthorDate: Tue Nov 18 12:13:37 2025 +0200 Commit: Martin Storsjö <martin(a)martin.st> CommitDate: Mon Nov 24 23:02:05 2025 +0200 tools: Make indent_arm_assembly.pl able to reformat a file in place This allows using the tool for one-off reindentations without needing the check_arm_indent.sh script (e.g. for use outside of ffmpeg), without having to pipe the file through stdin/stdout. diff --git a/tools/indent_arm_assembly.pl b/tools/indent_arm_assembly.pl index 7d5d1ecef2..359c2bcf4f 100755 --- a/tools/indent_arm_assembly.pl +++ b/tools/indent_arm_assembly.pl @@ -32,8 +32,11 @@ # - Optionally align operand columns vertically according to their # maximum width (accommodating for e.g. x0 vs x10, or v0.8b vs v16.16b). # -# The input code is passed to stdin, and the reformatted code is written -# on stdout. +# The script can be executed as "indent_arm_assembly.pl file [outfile]". +# If no outfile is specified, the given file is overwritten in place. +# +# Alternatively, the if no file parameters are given, the script reads input +# code on stdin, and outputs the reformatted code on stdout. use strict; @@ -41,6 +44,8 @@ my $indent_operands = 0; my $instr_indent = 8; my $operand_indent = 24; my $match_indent = 0; +my $file; +my $outfile; while (@ARGV) { my $opt = shift; @@ -54,7 +59,13 @@ while (@ARGV) { } elsif ($opt eq "-match-indent") { $match_indent = 1; } else { - die "Unrecognized parameter $opt\n"; + if (!$file) { + $file = $opt; + } elsif (!$outfile) { + $outfile = $opt; + } else { + die "Unrecognized parameter $opt\n"; + } } } @@ -130,7 +141,26 @@ sub columns { return indentcolumns($rest, 3); } -while (<STDIN>) { +my $in; +my $out; +my $tempfile; + +if ($file) { + open(INPUT, "$file") or die "Unable to open $file: $!"; + $in = *INPUT; + if ($outfile) { + open(OUTPUT, ">$outfile") or die "Unable to open $outfile: $!"; + } else { + $tempfile = "$file.tmp"; + open(OUTPUT, ">$tempfile") or die "Unable to open $tempfile: $!"; + } + $out = *OUTPUT; +} else { + $in = *STDIN; + $out = *STDOUT; +} + +while (<$in>) { # Trim off trailing whitespace. chomp; if (/^([\.\w\d]+:)?(\s+)([\w\\][\w\\\.]*)(?:(\s+)(.*)|$)/) { @@ -201,5 +231,13 @@ while (<STDIN>) { $_ = $label . $indent . $instr . $operand_space . $rest; } } - print $_ . "\n"; + print $out $_ . "\n"; +} + +if ($file) { + close(INPUT); + close(OUTPUT); +} +if ($tempfile) { + rename($tempfile, $file); } ----------------------------------------------------------------------- Summary of changes: tools/indent_arm_assembly.pl | 48 +++++++++++++++++++++++++++++++++++++++----- 1 file changed, 43 insertions(+), 5 deletions(-) hooks/post-receive --

1 0

[ffmpeg] branch master updated. 140b4f28c3 avutil/hwcontext_amf: move AVMutex to internal context
by ffmpeg-git＠ffmpeg.org 24 Nov '25

24 Nov '25

The branch, master has been updated via 140b4f28c303abc531daeac4eb42dd014cdb5afb (commit) from efd6e85abbf8cbea5992554823b27e2e32c8f2c3 (commit) - Log ----------------------------------------------------------------- commit 140b4f28c303abc531daeac4eb42dd014cdb5afb Author: Dmitrii Ovchinnikov <ovchinnikov.dmitrii(a)gmail.com> AuthorDate: Thu Nov 13 17:40:26 2025 +0100 Commit: Dmitrii Ovchinnikov <ovchinnikov.dmitrii(a)gmail.com> CommitDate: Mon Nov 24 20:06:24 2025 +0000 avutil/hwcontext_amf: move AVMutex to internal context diff --git a/doc/APIchanges b/doc/APIchanges index bfea83c73a..93c6f92704 100644 --- a/doc/APIchanges +++ b/doc/APIchanges @@ -2,6 +2,9 @@ The last version increases of all libraries were on 2025-03-28 API changes, most recent first: +2025-11-18 - xxxxxxxxxx - lavu 60.19.100 - hwcontext_amf.h + avutil/hwcontext_amf: add lock and unlock for AVAMFDeviceContext. + 2025-11-16 - xxxxxxxxxx - lavu 60.18.100 - cpu.h Deprecate AV_CPU_FLAG_FORCE without replacement. diff --git a/libavcodec/amfenc.c b/libavcodec/amfenc.c index 2174c5bdb2..b07da236c7 100644 --- a/libavcodec/amfenc.c +++ b/libavcodec/amfenc.c @@ -564,9 +564,11 @@ static int amf_submit_frame_locked(AVCodecContext *avctx, AVFrame *frame, AMFSur AVHWDeviceContext *hw_device_ctx = (AVHWDeviceContext*)ctx->device_ctx_ref->data; AVAMFDeviceContext *amf_device_ctx = (AVAMFDeviceContext *)hw_device_ctx->hwctx; - ff_mutex_lock(&amf_device_ctx->mutex); + if (amf_device_ctx->lock) + amf_device_ctx->lock(amf_device_ctx->lock_ctx); ret = amf_submit_frame(avctx, frame, surface_resubmit); - ff_mutex_unlock(&amf_device_ctx->mutex); + if (amf_device_ctx->unlock) + amf_device_ctx->unlock(amf_device_ctx->lock_ctx); return ret; } diff --git a/libavutil/hwcontext_amf.c b/libavutil/hwcontext_amf.c index acd9627c68..c754dc4ee5 100644 --- a/libavutil/hwcontext_amf.c +++ b/libavutil/hwcontext_amf.c @@ -39,6 +39,7 @@ #include "pixdesc.h" #include "pixfmt.h" #include "imgutils.h" +#include "thread.h" #include "libavutil/avassert.h" #include <AMF/core/Surface.h> #include <AMF/core/Trace.h> @@ -49,6 +50,15 @@ #endif #define FFMPEG_AMF_WRITER_ID L"ffmpeg_amf" +static void amf_lock_default(void *opaque) +{ + ff_mutex_lock((AVMutex*)opaque); +} + +static void amf_unlock_default(void *opaque) +{ + ff_mutex_unlock((AVMutex*)opaque); +} typedef struct AmfTraceWriter { AMFTraceWriterVtbl *vtblp; @@ -352,7 +362,7 @@ static int amf_transfer_data_from(AVHWFramesContext *ctx, AVFrame *dst, static void amf_device_uninit(AVHWDeviceContext *device_ctx) { - AVAMFDeviceContext *amf_ctx = device_ctx->hwctx; + AVAMFDeviceContext *amf_ctx = device_ctx->hwctx; AMF_RESULT res = AMF_NOT_INITIALIZED; AMFTrace *trace; @@ -377,8 +387,14 @@ static void amf_device_uninit(AVHWDeviceContext *device_ctx) amf_writer_free(amf_ctx->trace_writer); } + if (amf_ctx->lock_ctx == amf_lock_default) { + ff_mutex_destroy((AVMutex*)amf_ctx->lock_ctx); + av_freep(&amf_ctx->lock_ctx); + amf_ctx->lock = NULL; + amf_ctx->unlock = NULL; + } + amf_ctx->version = 0; - ff_mutex_destroy(&amf_ctx->mutex); } static int amf_device_init(AVHWDeviceContext *ctx) @@ -387,6 +403,16 @@ static int amf_device_init(AVHWDeviceContext *ctx) AMFContext1 *context1 = NULL; AMF_RESULT res; + if (!amf_ctx->lock) { + amf_ctx->lock_ctx = av_mallocz(sizeof(AVMutex)); + if (!amf_ctx->lock_ctx) { + return AVERROR(ENOMEM); + } + ff_mutex_init((AVMutex*)amf_ctx->lock_ctx, NULL); + amf_ctx->lock = amf_lock_default; + amf_ctx->unlock = amf_unlock_default; + } + #ifdef _WIN32 res = amf_ctx->context->pVtbl->InitDX11(amf_ctx->context, NULL, AMF_DX11_1); if (res == AMF_OK || res == AMF_ALREADY_INITIALIZED) { @@ -415,7 +441,7 @@ static int amf_device_init(AVHWDeviceContext *ctx) } } #endif - ff_mutex_init(&amf_ctx->mutex, NULL); + return 0; } diff --git a/libavutil/hwcontext_amf.h b/libavutil/hwcontext_amf.h index 5b726e3b9e..6f2cabc878 100644 --- a/libavutil/hwcontext_amf.h +++ b/libavutil/hwcontext_amf.h @@ -26,7 +26,6 @@ #include <AMF/core/Context.h> #include <AMF/core/Trace.h> #include <AMF/core/Debug.h> -#include "thread.h" /** * This struct is allocated as AVHWDeviceContext.hwctx @@ -39,7 +38,10 @@ typedef struct AVAMFDeviceContext { int64_t version; ///< version of AMF runtime AMFContext *context; AMF_MEMORY_TYPE memory_type; - AVMutex mutex; + + void (*lock)(void *lock_ctx); + void (*unlock)(void *lock_ctx); + void *lock_ctx; } AVAMFDeviceContext; enum AMF_SURFACE_FORMAT av_av_to_amf_format(enum AVPixelFormat fmt); diff --git a/libavutil/version.h b/libavutil/version.h index d1385a8829..db250d5c9e 100644 --- a/libavutil/version.h +++ b/libavutil/version.h @@ -79,7 +79,7 @@ */ #define LIBAVUTIL_VERSION_MAJOR 60 -#define LIBAVUTIL_VERSION_MINOR 18 +#define LIBAVUTIL_VERSION_MINOR 19 #define LIBAVUTIL_VERSION_MICRO 100 #define LIBAVUTIL_VERSION_INT AV_VERSION_INT(LIBAVUTIL_VERSION_MAJOR, \ ----------------------------------------------------------------------- Summary of changes: doc/APIchanges | 3 +++ libavcodec/amfenc.c | 6 ++++-- libavutil/hwcontext_amf.c | 32 +++++++++++++++++++++++++++++--- libavutil/hwcontext_amf.h | 6 ++++-- libavutil/version.h | 2 +- 5 files changed, 41 insertions(+), 8 deletions(-) hooks/post-receive --

1 0

[ffmpeg] branch release/8.0 updated. 565d14e4ff avfilter/f_select: Added activate for aselect
by ffmpeg-git＠ffmpeg.org 24 Nov '25

24 Nov '25

The branch, release/8.0 has been updated via 565d14e4ffa2e43a7299bd20841e241c23c5847d (commit) from 45455e13153876dbe1ad2be45947c7fb8242e703 (commit) - Log ----------------------------------------------------------------- commit 565d14e4ffa2e43a7299bd20841e241c23c5847d Author: Anders Rein <anders(a)onemimir.com> AuthorDate: Mon Nov 17 23:52:49 2025 +0100 Commit: Marton Balint <cus(a)passwd.hu> CommitDate: Mon Nov 24 21:05:04 2025 +0100 avfilter/f_select: Added activate for aselect During migration to the activation filter API the aselect filter was accidentally turned into a no-op filter. (cherry picked from commit 7411e902da5bb1983e54772b3d16bc2d71297ed4) diff --git a/libavfilter/f_select.c b/libavfilter/f_select.c index 46208a7b76..3781e50814 100644 --- a/libavfilter/f_select.c +++ b/libavfilter/f_select.c @@ -504,6 +504,7 @@ const FFFilter ff_af_aselect = { .p.flags = AVFILTER_FLAG_DYNAMIC_OUTPUTS, .init = aselect_init, .uninit = uninit, + .activate = activate, .priv_size = sizeof(SelectContext), FILTER_INPUTS(avfilter_af_aselect_inputs), }; ----------------------------------------------------------------------- Summary of changes: libavfilter/f_select.c | 1 + 1 file changed, 1 insertion(+) hooks/post-receive --

1 0