[FFmpeg-devel] [PATCH v2 1/2] avcodec: Fix warnings with signed/unsigned compare in bitstream.h
Andreas Rheinhardt
andreas.rheinhardt at outlook.com
Sat Mar 25 06:47:01 EET 2023
Devin Heitmueller:
> When including the header in decklink_enc.cpp it would be fed
> through the C++ compiler rather than the C compiler, which has
> more strict warnings when comparing signed/unsigned values.
>
> Make the local variables unsigned to match the arguments they are
> being passed for those functions.
>
> Signed-off-by: Devin Heitmueller <dheitmueller at ltnglobal.com>
> ---
> libavcodec/bytestream.h | 10 +++++-----
> 1 file changed, 5 insertions(+), 5 deletions(-)
>
> diff --git a/libavcodec/bytestream.h b/libavcodec/bytestream.h
> index d0033f14f3..67080604b9 100644
> --- a/libavcodec/bytestream.h
> +++ b/libavcodec/bytestream.h
> @@ -180,7 +180,7 @@ static av_always_inline void bytestream2_skipu(GetByteContext *g,
> static av_always_inline void bytestream2_skip_p(PutByteContext *p,
> unsigned int size)
> {
> - int size2;
> + unsigned int size2;
> if (p->eof)
> return;
> size2 = FFMIN(p->buffer_end - p->buffer, size);
> @@ -268,7 +268,7 @@ static av_always_inline unsigned int bytestream2_get_buffer(GetByteContext *g,
> uint8_t *dst,
> unsigned int size)
> {
> - int size2 = FFMIN(g->buffer_end - g->buffer, size);
> + unsigned int size2 = FFMIN(g->buffer_end - g->buffer, size);
> memcpy(dst, g->buffer, size2);
> g->buffer += size2;
> return size2;
> @@ -287,7 +287,7 @@ static av_always_inline unsigned int bytestream2_put_buffer(PutByteContext *p,
> const uint8_t *src,
> unsigned int size)
> {
> - int size2;
> + unsigned int size2;
> if (p->eof)
> return 0;
> size2 = FFMIN(p->buffer_end - p->buffer, size);
> @@ -311,7 +311,7 @@ static av_always_inline void bytestream2_set_buffer(PutByteContext *p,
> const uint8_t c,
> unsigned int size)
> {
> - int size2;
> + unsigned int size2;
> if (p->eof)
> return;
> size2 = FFMIN(p->buffer_end - p->buffer, size);
> @@ -348,7 +348,7 @@ static av_always_inline unsigned int bytestream2_copy_buffer(PutByteContext *p,
> GetByteContext *g,
> unsigned int size)
> {
> - int size2;
> + unsigned int size2;
>
> if (p->eof)
> return 0;
The bytestream APIs are allowed to overread if the buffer is padded and
the user manages this himself. So you are not allowed to presume that
g->buffer_end - g->buffer is positive.
- Andreas
More information about the ffmpeg-devel
mailing list