Hi Jean-Paul,<div> it seems that your patch is the right one. I tested it and it seems to correctly fix the problem.</div><div>Regards</div><div>Roberto<br><br><div class="gmail_quote">On 4 September 2012 13:34, Jean-Paul Saman <span dir="ltr"><<a href="mailto:jpsaman@videolan.org" target="_blank">jpsaman@videolan.org</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Robert,<br>
<div><div class="h5"><br>
On Mon, Sep 3, 2012 at 3:07 PM, Roberto Corno <<a href="mailto:corno.roberto@gmail.com">corno.roberto@gmail.com</a>> wrote:<br>
> SEGV happens within dvbpsi_decoder_psi_section_add when p->i_number ==<br>
> p_section->i_number and p_prev == NULL.<br>
> It's the first branch of the while(p) loop.<br>
> I got it with a heavily degraded transport stream, where a lot of CC error<br>
> occur.<br>
> I've added the check for p_prev==NULL<br>
> Patch attached.<br>
<br>
</div></div>Your patch ignores the new section instead of replacing the section<br>
with the same number.<br>
The caller could therefor leak the section pointer. Please try this<br>
patch instead.<br>
<br>
Kind regards,<br>
<br>
Jean-Paul Saman<br>
</blockquote></div><br></div>