July 2025 - ffmpeg-devel - lists.ffmpeg.org

[PATCH] vulkan_encode_h264/5: Fix uninitialized return value in write_extra_headers
by David Rosca 28 Jul '25

28 Jul '25

--- libavcodec/vulkan_encode_h264.c | 1 + libavcodec/vulkan_encode_h265.c | 1 + 2 files changed, 2 insertions(+) diff --git a/libavcodec/vulkan_encode_h264.c b/libavcodec/vulkan_encode_h264.c index cdc87fb4ca..fdd3cc8a55 100644 --- a/libavcodec/vulkan_encode_h264.c +++ b/libavcodec/vulkan_encode_h264.c @@ -1311,6 +1311,7 @@ static int write_extra_headers(AVCodecContext *avctx, if (err < 0) goto fail; } else { + err = 0; *data_len = 0; } diff --git a/libavcodec/vulkan_encode_h265.c b/libavcodec/vulkan_encode_h265.c index cd50f2f756..8c25cacded 100644 --- a/libavcodec/vulkan_encode_h265.c +++ b/libavcodec/vulkan_encode_h265.c @@ -1471,6 +1471,7 @@ static int write_extra_headers(AVCodecContext *avctx, if (err < 0) goto fail; } else { + err = 0; *data_len = 0; } -- 2.47.0

1 1

Re: [FFmpeg-devel] [PATCH] checkasm/h264dsp: Fix stack-buffer-overflow, effective-type violations
by Marvin Scholz 28 Jul '25

28 Jul '25

On 25 Jun 2025, at 23:39, Tristan Matthews via ffmpeg-devel wrote: > On Monday, June 16th, 2025 at 2:22 PM, Andreas Rheinhardt <andreas.rheinhardt(a)outlook.com> wrote: > > Tristan Matthews via ffmpeg-devel: > > On Monday, June 16th, 2025 at 6:53 AM, Andreas Rheinhardt andreas.rheinhardt(a)outlook.com wrote: > > Patch attached. > > LGTM, and sorry about that, for future reference did you do anything in particular (instrumentation wise) to catch the stack overflow? I tried make fate on a few different machines and didn't hit that. > > Best, > Tristan > > > I used (Clang) ASAN. Presumably Zhao Zhili (who first reported the stack > buffer overflow) did so, too. > > - Andreas > > > It'd be good to merge either this or Zhao Zhili's patch to get checkasm back to being ASAN clean. > > Best, > Tristan If there are no objections, I would like to merge this patch soon. > _______________________________________________ > ffmpeg-devel mailing list > ffmpeg-devel(a)ffmpeg.org > https://ffmpeg.org/mailman/listinfo/ffmpeg-devel > > To unsubscribe, visit link above, or email > ffmpeg-devel-request(a)ffmpeg.org with subject "unsubscribe".

2 1

[PATCH 1/1] libavformat/img2enc.c: rm warning ‘.tmp’ directive output may be truncated
by Cai Fan 28 Jul '25

28 Jul '25

Signed-off-by: Cai Fan <caifan0425(a)163.com> --- libavformat/img2enc.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/libavformat/img2enc.c b/libavformat/img2enc.c index 41638d92b8..672cfe18fd 100644 --- a/libavformat/img2enc.c +++ b/libavformat/img2enc.c @@ -180,6 +180,10 @@ static int write_packet(AVFormatContext *s, AVPacket *pkt) } for (i = 0; i < 4; i++) { av_dict_copy(&options, img->protocol_opts, 0); + if (strlen(filename) + 5 >= sizeof(img->tmp[i])) { + av_log(s, AV_LOG_ERROR, "Filename too long: %s\n", filename); + return AVERROR(EINVAL); + } snprintf(img->tmp[i], sizeof(img->tmp[i]), "%s.tmp", filename); av_strlcpy(img->target[i], filename, sizeof(img->target[i])); if (s->io_open(s, &pb[i], img->use_rename ? img->tmp[i] : filename, AVIO_FLAG_WRITE, &options) < 0) { -- 2.34.1

2 1

[PATCH v3] avformat/avisynth: fix segfault when also compiled with libvmaf
by Daniel Bermond 28 Jul '25

28 Jul '25

When FFmpeg is compiled with support for both avisynth and libvmaf, a segmentation fault occurs when using avisynth (.avs) input. This happens because both avisynthplus and vmaf have the exactly same C++ symbol 'Cache::~Cache()'[1][2], which is a C++ destructor. When using avisynth input, this destructor is also called in the vmaf side, although no vmaf object was created, thus causing a segmentation fault. Without changing this conflicting symbol in avisynthplus and/or vmaf upstream code, a solution is to switch the 'RTLD_LOCAL'[3] flag for 'RTLD_DEEPBIND'[4] when calling 'dlopen()' on the avisynthplus library. This will make the avisynthplus symbols loaded by 'dlopen()' to be ahead of the symbols loaded in the global scope, and consequently avoid the segmentation fault by preventing the conflicting vmaf C++ destructor to be called. More details can be seen in the discussion of the Arch Linux bug report[5] about this issue. [1] https://github.com/AviSynth/AviSynthPlus/blob/v3.7.5/avs_core/core/cache.h#… [2] https://github.com/Netflix/vmaf/blob/v3.0.0/libvmaf/src/svm.cpp#L75 [3] https://man.archlinux.org/man/dlopen.3#RTLD_LOCAL [4] https://man.archlinux.org/man/dlopen.3#RTLD_DEEPBIND [5] https://gitlab.archlinux.org/archlinux/packaging/packages/ffmpeg/-/issues/19 Signed-off-by: Daniel Bermond <danielbermond(a)gmail.com> --- libavformat/avisynth.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/libavformat/avisynth.c b/libavformat/avisynth.c index cb2be10925..bef61b4a60 100644 --- a/libavformat/avisynth.c +++ b/libavformat/avisynth.c @@ -120,7 +120,7 @@ typedef struct AviSynthContext { static av_cold int avisynth_load_library(AviSynthContext *avs) { - avs->avs_library.library = dlopen(AVISYNTH_LIB, RTLD_NOW | RTLD_LOCAL); + avs->avs_library.library = dlopen(AVISYNTH_LIB, RTLD_NOW | RTLD_DEEPBIND); if (!avs->avs_library.library) return AVERROR_UNKNOWN; -- 2.50.1

3 4

[PATCH v2] avformat/avisynth: fix segfault when also compiled with libvmaf
by Daniel Bermond 27 Jul '25

27 Jul '25

When FFmpeg is compiled with support for both avisynth and libvmaf, a segmentation fault occurs when using avisynth (.avs) input. This happens because both avisynthplus and vmaf have the exactly same C++ symbol 'Cache::~Cache()'[1][2], which is a C++ destructor. When using avisynth input, this destructor is also called in the vmaf side, although no vmaf object was created, thus causing a segmentation fault. Without changing this conflicting symbol in avisynthplus and/or vmaf upstream code, a solution is to switch the 'RTLD_LOCAL'[3] flag for 'RTLD_DEEPBIND'[4] when calling 'dlopen()' on the avisynthplus library. This will make the avisynthplus symbols loaded by 'dlopen()' to be ahead of the symbols loaded in the global scope, and consequently avoid the segmentation fault by preventing the conflicting vmaf C++ destructor to be called. More details can be seen in the discussion of the Arch Linux bug report[5] about this issue. [1] https://github.com/AviSynth/AviSynthPlus/blob/v3.7.5/avs_core/core/cache.h#… [2] https://github.com/Netflix/vmaf/blob/v3.0.0/libvmaf/src/svm.cpp#L75 [3] https://man.archlinux.org/man/dlopen.3#RTLD_LOCAL [4] https://man.archlinux.org/man/dlopen.3#RTLD_DEEPBIND [5] https://gitlab.archlinux.org/archlinux/packaging/packages/ffmpeg/-/issues/19 Signed-off-by: Daniel Bermond <danielbermond(a)gmail.com> --- libavformat/avisynth.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/libavformat/avisynth.c b/libavformat/avisynth.c index cb2be10925..bef61b4a60 100644 --- a/libavformat/avisynth.c +++ b/libavformat/avisynth.c @@ -120,7 +120,7 @@ typedef struct AviSynthContext { static av_cold int avisynth_load_library(AviSynthContext *avs) { - avs->avs_library.library = dlopen(AVISYNTH_LIB, RTLD_NOW | RTLD_LOCAL); + avs->avs_library.library = dlopen(AVISYNTH_LIB, RTLD_NOW | RTLD_DEEPBIND); if (!avs->avs_library.library) return AVERROR_UNKNOWN; -- 2.50.1

1 0

[PATCH] avformat/avisynth: fix segfault when also compiled with libvmaf
by Daniel Bermond 27 Jul '25

27 Jul '25

When FFmpeg is compiled with support for both avisynth and libvmaf, a segmentation fault occurs when using avisynth (.avs) input. This happens because both avisynthplus and vmaf have the exactly same C++ symbol 'Cache::~Cache()'[1][2], which is a C++ destructor. When using avisynth input, this destructor is also called in the vmaf side, although no vmaf object was created, thus causing a segmentation fault. Without changing this conflicting symbol in avisynthplus and/or vmaf upstream code, a solution is to switch the 'RTLD_LOCAL'[3] flag for 'RTLD_DEEPBIND'[4] when calling 'dlopen()' on the avisynthplus library. This will make the avisynthplus symbols loaded by 'dlopen()' to be ahead of the symbols loaded in the global scope, and consequently avoid the segmentation fault by preventing the conflicting vmaf C++ destructor to be called. More details can be seen in the discussion of the Arch Linux bug report[5] about this issue. [1] https://github.com/AviSynth/AviSynthPlus/blob/v3.7.5/avs_core/core/cache.h#… [2] https://github.com/Netflix/vmaf/blob/v3.0.0/libvmaf/src/svm.cpp#L75 [3] https://man.archlinux.org/man/dlopen.3#RTLD_LOCAL [4] https://man.archlinux.org/man/dlopen.3#RTLD_DEEPBIND [5] https://gitlab.archlinux.org/archlinux/packaging/packages/ffmpeg/-/issues/19 Signed-off-by: Daniel Bermond <danielbermond(a)gmail.com> --- libavformat/avisynth.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/libavformat/avisynth.c b/libavformat/avisynth.c index cb2be10925..bef61b4a60 100644 --- a/libavformat/avisynth.c +++ b/libavformat/avisynth.c @@ -120,7 +120,7 @@ typedef struct AviSynthContext { static av_cold int avisynth_load_library(AviSynthContext *avs) { - avs->avs_library.library = dlopen(AVISYNTH_LIB, RTLD_NOW | RTLD_LOCAL); + avs->avs_library.library = dlopen(AVISYNTH_LIB, RTLD_NOW | RTLD_DEEPBIND); if (!avs->avs_library.library) return AVERROR_UNKNOWN; -- 2.50.1

1 0

[FFFjo] Re: [FFmpeg/FFmpeg] avcodec/dvbsubenc: add a min_bpp option to work around some decoders. (PR #20023)
by waider 26 Jul '25

26 Jul '25

Requested to review #20023 ( https://code.ffmpeg.org/FFmpeg/FFmpeg/pulls/20023 ). --- View it on FFmpeg Forgejo ( https://code.ffmpeg.org/FFmpeg/FFmpeg/pulls/20023#issuecomment-499 ) or reply to this email directly.

1 0

Re: [FFmpeg-devel] [PATCH] [RFC] global/clang-format: Add .clang-format configuration for consistent formatting
by Jacob Lifshay 26 Jul '25

26 Jul '25

Replying to softworkz from the mailing list archive: I tried out the .clang-format file you proposed and I ended up using: AlignConsecutiveDeclarations: Consecutive instead of: > +AlignConsecutiveDeclarations: AcrossEmptyLinesAndComments since otherwise clang-format will indent the function names for most of a header file like so, which imo just wastes horizontal space: > /** <docs> */ > int av_foo(); > > /** <docs> */ > int av_bar(); > > /** <docs> */ > LongTypeName av_baz(); Jacob

1 0

Authorize BtbN
by Derek Buitenhuis 25 Jul '25

25 Jul '25

https://0x0.st/85bn.png I know who BtbN is, random contributor does not. Appears sketchy. - Derek

3 5

[PATCH 01/10] avformat/os_support: check invalid socket value correctly on Windows
by Kacper Michajłow 25 Jul '25

25 Jul '25

SOCKET defined in winsock2.h is unsigned and invalid value is defined as INVALID_SOCKET. Check this explicity to avoid compiler warnings. See: https://learn.microsoft.com/en-us/windows/win32/winsock/socket-data-type-2 Signed-off-by: Kacper Michajłow <kasper93(a)gmail.com> --- libavformat/os_support.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/libavformat/os_support.c b/libavformat/os_support.c index 4d6eb8a74c..17d77a01d5 100644 --- a/libavformat/os_support.c +++ b/libavformat/os_support.c @@ -244,13 +244,16 @@ int ff_poll(struct pollfd *fds, nfds_t numfds, int timeout) n = 0; for (i = 0; i < numfds; i++) { +#if !HAVE_WINSOCK2_H if (fds[i].fd < 0) continue; -#if !HAVE_WINSOCK2_H if (fds[i].fd >= FD_SETSIZE) { errno = EINVAL; return -1; } +#else + if (fds[i].fd == INVALID_SOCKET) + continue; #endif /* !HAVE_WINSOCK2_H */ if (fds[i].events & POLLIN) -- 2.50.1

2 11