On 3/20/12, Maans Rullgaard <[email protected]> wrote: > Paul B Mahol <[email protected]> writes: > >> On 3/20/12, Maans Rullgaard <[email protected]> wrote: >>> Paul B Mahol <[email protected]> writes: >>> >>>> On 3/20/12, Maans Rullgaard <[email protected]> wrote: >>>>> Paul B Mahol <[email protected]> writes: >>>>> >>>>>> On 3/20/12, Maans Rullgaard <[email protected]> wrote: >>>>>>> Paul B Mahol <[email protected]> writes: >>>>>>> >>>>>>>> On 3/20/12, Maans Rullgaard <[email protected]> wrote: >>>>>>>>> Paul B Mahol <[email protected]> writes: >>>>>>>>> >>>>>>>>>> static av_always_inline void bytestream2_skip(GetByteContext *g, >>>>>>>>>> - unsigned int size) >>>>>>>>>> + int size) >>>>>>>>>> { >>>>>>>>>> g->buffer += FFMIN(g->buffer_end - g->buffer, size); >>>>>>>>>> } >>>>>>>>> >>>>>>>>> Think about what happens with a large negative offset. >>>>>>>> >>>>>>>> Nothing better than with very big positive offset. >>>>>>> >>>>>>> But a lot worse. With the signed argument type, it allows setting >>>>>>> the >>>>>>> position before the start of the buffer. >>>>>> >>>>>> Exactly, take look at bytestream2_init*. >>>>> >>>>> Why do you refuse to understand the problem? >>>> >>>> There are multiple problems. What problem do you think I refuse to >>>> understand? >>> >>> The one I pointed out and which is caused by this patch. >> >> Your point is irrelevant because bytestream2_init use signed for size. > > You completely and utterly fail to understand the problem.
I'm leaving this problem to be resolved by experts. _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
