Steve Langasek <[EMAIL PROTECTED]> writes: > On Tue, Jan 09, 2007 at 09:47:53AM +0100, Simon Josefsson wrote: >> Simon Josefsson <[EMAIL PROTECTED]> writes: > >> >> Has this problem been seen on any archs other than amd64, where it >> >> fails the testsuite and therefore generates no binaries? > >> > No. > >> I just realized that I were wrong here, and that this may affect the >> choice of severity level. > >> The problem applies to all architectures where sizeof(int) < >> sizeof(size_t), which includes ia64 and maybe other 64-bit platforms. > > Ah, yes; assuming this code is liable to break under other circumstances at > runtime then,
The buggy function is part of the external API, so yes, there are no guarantees for what it will be used for, and I think it will break for most typical uses, unless the compiler leaves extra room on the stack (which it seems to have happened to do for the 64-bit and non-amd64 platforms, otherwise the self test would have crashed on those platforms as well). > it may well be a release critical bug. I'll let Russ decide. A fixed package should be ready to upload anyway. /Simon _______________________________________________ Help-gss mailing list [email protected] http://lists.gnu.org/mailman/listinfo/help-gss
