PR #22279 opened by mkver URL: https://code.ffmpeg.org/FFmpeg/FFmpeg/pulls/22279 Patch URL: https://code.ffmpeg.org/FFmpeg/FFmpeg/pulls/22279.patch
>From 6f10bf5149f30695d5c761776d0fe02be4f15042 Mon Sep 17 00:00:00 2001 From: Andreas Rheinhardt <[email protected]> Date: Tue, 24 Feb 2026 18:33:44 +0100 Subject: [PATCH 1/3] tests/checkasm/h264chroma: Fix initialization, range of values This commit fixes four related bugs: 1. The >8bpp codepath only initializes half the buffer. The remaining half used leftover samples from the 8bpp codepath which initialized the complete buffer. 2. The 8bpp codepath tests only 2 bit inputs (&3). This means that the second half of the buffer only uses 10 bits (in fact, only values of the form 000000xx000000xxb) when treated as uint16_t in the >8bpp test. Due to 1., using more bits in the 8bpp test would make the >8bpp tests fail (the intermediates would no longer fit into 16bits). 3. For the >8bpp tests, the first half of the buffer would only be tested with 8bpp input. 4. The 8bpp codepath initializes the whole buffer, but only uses half of it. Signed-off-by: Andreas Rheinhardt <[email protected]> --- tests/checkasm/h264chroma.c | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) diff --git a/tests/checkasm/h264chroma.c b/tests/checkasm/h264chroma.c index 52aa220152..b5b8d64855 100644 --- a/tests/checkasm/h264chroma.c +++ b/tests/checkasm/h264chroma.c @@ -30,11 +30,12 @@ #define randomize_buffers(bit_depth) \ do { \ if (bit_depth == 8) { \ - for (int i = 0; i < 16*18*2; i++) \ - src[i] = rnd() & 0x3; \ + for (int i = 0; i < 16*18; i++) \ + src[i] = rnd(); \ } else { \ - for (int i = 0; i < 16*18; i += 2) \ - AV_WN16(&src[i], rnd() & 0xFF); \ + unsigned mask = (1 << bit_depth) - 1;\ + for (int i = 0; i < 16*18*2; i += 2) \ + AV_WN16A(&src[i], rnd() & mask); \ } \ } while (0) -- 2.52.0 >From cb5bb4e66bd37b73908fc5fb6d89a138aaa34b46 Mon Sep 17 00:00:00 2001 From: Andreas Rheinhardt <[email protected]> Date: Tue, 24 Feb 2026 18:53:22 +0100 Subject: [PATCH 2/3] tests/checkasm/h264chroma: Add bit depth to error output Signed-off-by: Andreas Rheinhardt <[email protected]> --- tests/checkasm/h264chroma.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/tests/checkasm/h264chroma.c b/tests/checkasm/h264chroma.c index b5b8d64855..8a96d1b973 100644 --- a/tests/checkasm/h264chroma.c +++ b/tests/checkasm/h264chroma.c @@ -64,7 +64,7 @@ static void check_chroma_mc(void) call_ref(dst0, src, 16 * SIZEOF_PIXEL, 16, x, y); \ call_new(dst1, src, 16 * SIZEOF_PIXEL, 16, x, y); \ if (memcmp(dst0, dst1, 16 * 16 * SIZEOF_PIXEL)) { \ - fprintf(stderr, #name ": x:%i, y:%i\n", x, y); \ + fprintf(stderr, #name "_%d: x:%i, y:%i\n", bit_depth, x, y); \ fail(); \ } \ bench_new(dst1, src, 16 * SIZEOF_PIXEL, 16, x, y); \ -- 2.52.0 >From 4c1e72ac273fddc0846304db728d593688eb95d4 Mon Sep 17 00:00:00 2001 From: Andreas Rheinhardt <[email protected]> Date: Tue, 24 Feb 2026 18:57:27 +0100 Subject: [PATCH 3/3] tests/checkasm/h264chroma: Don't overalign The input only need to have natural alignment (i.e. 1 in the 8bpp case, 2 otherwise), the output need only have 16 (namely in the width 8 >8bpp case). Signed-off-by: Andreas Rheinhardt <[email protected]> --- tests/checkasm/h264chroma.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/tests/checkasm/h264chroma.c b/tests/checkasm/h264chroma.c index 8a96d1b973..732e992983 100644 --- a/tests/checkasm/h264chroma.c +++ b/tests/checkasm/h264chroma.c @@ -42,9 +42,9 @@ static void check_chroma_mc(void) { H264ChromaContext h; - LOCAL_ALIGNED_32(uint8_t, src, [16 * 18 * 2]); - LOCAL_ALIGNED_32(uint8_t, dst0, [16 * 18 * 2]); - LOCAL_ALIGNED_32(uint8_t, dst1, [16 * 18 * 2]); + DECLARE_ALIGNED_4(uint8_t, src) [16 * 18 * 2]; + DECLARE_ALIGNED_16(uint8_t, dst0)[16 * 18 * 2]; + DECLARE_ALIGNED_16(uint8_t, dst1)[16 * 18 * 2]; declare_func_emms(AV_CPU_FLAG_MMX, void, uint8_t *dst, const uint8_t *src, ptrdiff_t stride, int h, int x, int y); -- 2.52.0 _______________________________________________ ffmpeg-devel mailing list -- [email protected] To unsubscribe send an email to [email protected]
