[FFmpeg-devel] [PATCH] libavcodec/lpc.c: Fix warning about uninitialized variable

[Patrik Kullman]

On Tue, 2009-02-10 at 21:05 -0500, Justin Ruggles wrote:
> Jai Menon wrote:
> > On Tue, Feb 10, 2009 at 2:38 PM, Patrik Kullman <patrik at yes.nu> wrote:
> >>
> >> On Tue, 2009-02-10 at 14:29 +0530, Jai Menon wrote:
> >>> On Tue, Feb 10, 2009 at 2:14 PM, Jai Menon <jmenon86 at gmail.com> wrote:
> >>>> On Tue, Feb 10, 2009 at 2:08 PM, Patrik Kullman <patrik at yes.nu> wrote:
> >>>>> If use_lpc < 1 and max_order > 0, weight would be uninitialized.
> >>>> since a use_lpc of 0 was never intended, maybe an assert could be used....
> >>> to clarify, i meant an assert on use_lpc == 0
> >> Well, a negative value in use_lpc wouldn't initialize weight either.
> >> Are negative values intended?
> > 
> > Justin could probably comment but semantically it makes no sense.
> > also, this is supposed  to be used internally so asserting should be ok.
> > anyhow, please wait for others to comment/approve/disapprove.
> 
> Adding "assert(use_lpc > 0);" sounds like a good idea. Perhaps that
> parameter should have better documentation as well.
> 0  = LPC with fixed pre-defined coeffs
> 1  = LPC with coeffs determined by Levinson-Durbin recursion
> 2+ = LPC with coeffs determined by Cholesky factorization using
> (use_lpc-1) passes.
> 
> Therefore, ff_lpc_calc_coefs() should not be called with use_lpc<=0.
> 
> -Justin

So how about I add the documentation in the patch?
-------------- next part --------------
A non-text attachment was scrubbed...
Name: ffmpeg-lpc-uninit-weight.patch
Type: text/x-patch
Size: 1189 bytes
Desc: not available
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20090211/52121299/attachment.bin>