<<< Date Index >>>     <<< Thread Index >>>

Re: Bug#260623: incomplete mbyte sequences + bogus libc's regexec(): segfault



* Christoph Berg [Wed, 04 Aug 2004 13:22:29 +0200]:
> Re: Adeodato Simó in <20040724023504.GA22492@xxxxxxxxxxxxxxx>
> >   there is a patch [attached] for this by Anders Helmersson
> >   <anders.helmersson@xxxxxxxxx>, which it may be arguable if should be
> >   included in mutt or not: the current behavior is "garbage in, garbage
> >   out" (GIGO), the "GO" part being already fixed in libc's cvs, the "GI"
> >   part being still there (in mutt).

> >   quality of this patch. i understand what it does but i have two main
> >   concerns: first, if it's really necessary to start checking at the
> >   beginning of the buffer instead of at the end (patch from myself
> >   trying to do so also attached; however, i don't know whether a
> >   subsequence of a valid mbyte sequence is always invalid); second,
> >   whether a call to fseek would be necessary.

> I see that both patches you've send are different from
> patch-1.5.4.helmersson.incomplete_multibyte which was included in
> Debian's Mutt until 1.5.5.1-200401. The patch was removed afterwards.

> Has the issue in #244549 been dealt with in either patch?

  yes, Anders Helmersson modified his patch to address #244549. this
  patch-1.5.6.helmersson.incomplete_multibyte.2 is the one that debian's
  mutt currently ships.

  mine will probably have some bug hidden, although i'm using it without
  trouble.

> #244549: mutt hangs when doing a search in a message that contains a
> <Nul> character due to patch-1.5.4.helmersson.incomplete_multibyte

> Christoph


-- 
Adeodato Simó
    EM: asp16 [ykwim] alu.ua.es | PK: DA6AE621
    Listening to: Presuntos Implicados - Gente
 
Let us not be ashamed to speak what we shame not to think.
                -- Michel de Montaigne