I took a crack at the more general change (it will be arriving shortly).
It's split into two patches, the first is a simplified version of this
change and the second patch more strongly defines the behavior on failure
of both the user supplied function and av_lockmgr_register. Please
consider this
On Mon, Sep 29, 2014 at 09:57:02PM -0700, Manfred Georg wrote:
> Answers inline.
>
> On Mon, Sep 29, 2014 at 7:07 PM, Michael Niedermayer
> wrote:
>
> > On Mon, Sep 29, 2014 at 02:41:38PM -0700, Manfred Georg wrote:
> > > A badly behaving user provided mutex manager (such as that in OpenCV)
> >
On Mon, Sep 29, 2014 at 02:41:38PM -0700, Manfred Georg wrote:
> A badly behaving user provided mutex manager (such as that in OpenCV) may not
> reset the mutex to NULL on destruction. This can cause a problem for a later
> mutex manager (which may assert that the mutex is NULL before creating).
A badly behaving user provided mutex manager (such as that in OpenCV) may not
reset the mutex to NULL on destruction. This can cause a problem for a later
mutex manager (which may assert that the mutex is NULL before creating).
---
libavcodec/utils.c | 15 +--
1 file changed, 9 inse
