[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Patch to improve behaviour in gnutls_x509_crl_set_version() when `ch
Re: Patch to improve behaviour in gnutls_x509_crl_set_version() when `char' is unsigned
Mon, 26 May 2008 12:53:25 +0200
Gnus/5.110009 (No Gnus v0.9) Emacs/22.2 (gnu/linux)
Laurence Withers <address@hidden> writes:
> On some platforms, char can be unsigned by default (in my case I'm using gcc
> 3.4.5 on arm). When that's the case, I noticed a compilation warning in
> lib/x509/crl_write.c .
> The attached patch fixes the warning, but also slightly changes the way the
> test is carried out.
> The documentation for the function states that the supplied version parameter
> must be 1 or 2, and implies it can be higher but not 0. The patch explicitly
> checks that version >= 1 without performing any arithmetic in a char type. If
> the check fails, the function returns with GNUTLS_E_INVALID_REQUEST, which I
> guess is the right thing to do. If the check succeeds, the previous behaviour
> is used. However, a char value of 0 is never decremented, so there is never
> any possibility of an overflow.
> If this behaviour isn't really wanted, and the exact previous semantics of
> function should be preserved (version 0 -> 0, version 1 -> 0, version 2 -> 1
> etc.) then let me know and I'll rework the patch.
Hi! Thanks for the report. I think we need to preserve the old
behaviour for 0, so I reworked the patch slightly.
Did you notice any similar compilation warnings? Which compilation
flags did you use?