[vlc-devel] [patch] Adding AMT module as a network access source. AMT allows IP multicast streams to be tunnelled over a non-multicast enabled network

Thomas Guillem thomas at gllm.fr
Mon Apr 8 09:36:39 CEST 2019


Nitpicking: you git commit title is too  long, and you should use the imperative mood.

cf. https://chris.beams.io/posts/git-commit/

On Sat, Apr 6, 2019, at 01:22, Wayne BRASSEM wrote:
> No worries.  I liked the strdup() approach too for that reason.
> 
> What happened to MODULES_LIST file in the modules folder?  In my latest 
> git pull it’s gone.  That’s why I didn’t add it to the patch.  I only 
> modified the existing Makefile.am so it was included in the build.
> 
> I also spotted a potential issue in the previous BlockUDP routine with 
> respect to packet overflow.  So that’s (hopefully) better in this 
> version too.
> 
> I trust that the block_ChainGather() does two things:
> 	1. Allocates a new, contiguous block of memory for the formerly 
> fragmented one
> 	2. Releases the memory blocks pointed to by the chained block argument 
> passed in
> 
> In my testing it seemed this way it behaves, but I could find no 
> documentation to prove it.
> W.
> 
> > On Apr 5, 2019, at 2:59 PM, Tristan Matthews <tmatth at videolan.org> wrote:
> > 
> > On Fri, Apr 5, 2019 at 2:40 PM Wayne Brassem <wbrassem at rogers.com> wrote:
> >> 
> >> Yeah, I know but strcpy() doesn’t allocate any memory, merely uses the existing buffer.  Anyways in the new patch I standardized on using strdup() which allocates so I know I have to free each time.
> > 
> > Yeah bad call (twice!) on my part. On the plus I'd still say your new
> > patch is better since sys->srcAddr is now exactly sized to hold
> > "0.0.0.0", whereas before it was assumed that the destination buffer
> > was large enough.
> > 
> > Best,
> > -t
> > 
> >>> On Apr 5, 2019, at 11:13 AM, Tristan Matthews <tmatth at videolan.org> wrote:
> >>> 
> >>> On Fri, Apr 5, 2019 at 11:03 AM Wayne Brassem <wbrassem at rogers.com> wrote:
> >>>> 
> >>>> I’ve checked the code over and can’t see how any of the referenced sys->srcAddr statements below gets leaked.  This buffer is freed in Close(), but is in use until then.
> >>>> What am I missing?  I don’t want to resubmit if it’s still wrong.
> >>> 
> >>> In the previous patch submission, you were overwriting sys->srcAddr
> >>> with a strcpy without freeing it first.
> >>> The patch you just sent does not have this issue.
> >>> 
> >>> I'll take a closer look at the latest patch soon.
> >>> 
> >>> Best,
> >>> Tristan
> >>> _______________________________________________
> >>> vlc-devel mailing list
> >>> To unsubscribe or modify your subscription options:
> >>> https://mailman.videolan.org/listinfo/vlc-devel
> >> 
> >> _______________________________________________
> >> vlc-devel mailing list
> >> To unsubscribe or modify your subscription options:
> >> https://mailman.videolan.org/listinfo/vlc-devel
> > _______________________________________________
> > vlc-devel mailing list
> > To unsubscribe or modify your subscription options:
> > https://mailman.videolan.org/listinfo/vlc-devel
> 
> _______________________________________________
> vlc-devel mailing list
> To unsubscribe or modify your subscription options:
> https://mailman.videolan.org/listinfo/vlc-devel


More information about the vlc-devel mailing list