On date Sunday 2015-03-15 15:15:34 +0100, Clément Bœsch encoded: > On Sun, Mar 15, 2015 at 03:11:14PM +0100, Clément Bœsch wrote: > > On Sun, Mar 15, 2015 at 03:07:16PM +0100, Stefano Sabatini wrote: > > > On date Sunday 2015-03-15 14:24:29 +0100, Clément Bœsch encoded: > > > > --- > > > > libavfilter/formats.c | 45 > > > > ++++++++++++++++++++++++++++++++------------- > > > > libavfilter/formats.h | 10 +++++----- > > > > 2 files changed, 37 insertions(+), 18 deletions(-) > > > > > > > > diff --git a/libavfilter/formats.c b/libavfilter/formats.c > > > > index 6393416..4f9773b 100644 > > > > --- a/libavfilter/formats.c > > > > +++ b/libavfilter/formats.c > > > > @@ -401,7 +401,12 @@ AVFilterChannelLayouts *ff_all_channel_counts(void) > > > > } > > > > > > > > #define FORMATS_REF(f, ref) > > > > \ > > > > - void *tmp = av_realloc_array(f->refs, sizeof(*f->refs), > > > > f->refcount + 1); \ > > > > + void *tmp; > > > > \ > > > > + > > > > \ > > > > + if (!ref) > > > > \ > > > > + return AVERROR_BUG; > > > > \ > > > > > > I'd prefer to crash or assert here, assuming the function doesn't > > > assume NULL, same below. > > > > > > > In the current state, these functions could be called with a NULL > > parameter. Random examples: > > > > libavfilter/src_movie.c: ff_formats_ref(ff_make_format_list(list), > > &outlink->in_formats); > > libavfilter/src_movie.c: ff_formats_ref(ff_make_format_list(list), > > &outlink->in_samplerates); > > libavfilter/vf_extractplanes.c: > > ff_formats_ref(ff_make_format_list(in_pixfmts), > > &ctx->inputs[0]->out_formats); > > libavfilter/vf_extractplanes.c: > > ff_formats_ref(ff_make_format_list(out_pixfmts), > > &ctx->outputs[i]->in_formats); > > > > So I'd better not do that. > > > > > (Unrelated note: "bug" is a silly term, "defect" is more proper - I'm > > > with Dijkstra here). > > > > > To elaborate on this, the bug here is referring to an allocation check not > done in the caller (there are many currently since I'm just introducing > the error handling).
I won't block this patch, but getting a crash in a specified point of the program is more useful than failing with this: A bug occurred somewhere -- FFmpeg = Foolish and Frenzy Muttering Perennial Evil Gorilla _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
