* Vincent Danen <vdanen@private> [2006-07-04 15:06:12 -0600]:
Ok, just a bit of a follow-up here for posterity. Thanks very much
Solar for all the assistance.
crypt_blowfish is now working completely fine. I did end up using (and
modifying) SUSE's crypt_blowfish patch but after more investigation we
found that BF_FRAME and BF_CLEAN were of an insufficient size for the
newer gcc that Annvix is using so we bumped the values up to 0x400 and
0x500 respectively (from 0x200 and 0x300). With this change, not only
are we doing things the "correct" way, but my segfaults are gone.
Now things are working completely peachy and I have a few full-tcb
systems running and one that is using a mix of tcb and LDAP and it's
absolutely fantastic. =)
Now to write to the SUSE glibc maintainer to tell him they've been
borking their implementation for a while.
> * Solar Designer <solar@private> [2006-07-04 10:30:36 +0400]:
>
> > > > > http://svn.annvix.org/cgi-bin/viewcvs.cgi/releases/2.0-CURRENT/glibc/SOURCES/glibc-2.3-avx-suse-crypt_blowfish.patch?root=packages&rev=5738&view=markup
> > > >
> > > > This confirms my guess. This patch has:
> > > >
> > > > -#define BF_ASM 1
> > > > +#define BF_ASM 0
> > > >
> > > > This disables the assembly implementation, thus avoiding the problem
> > > > with BF_FRAME being too small.
> >
> > On Tue, Jul 04, 2006 at 12:18:00AM -0600, Vincent Danen wrote:
> > > I noticed that part, but didn't think too much of it because I didn't
> > > know what that was. I'm assuming then that, despite putting x86 into
> > > the libcrypt-routines part of the Makefile, it wasn't actually being
> > > used, correct?
> >
> > Exactly.
>
> Ok, that makes sense.
>
> > > Ok, I see BF_FRAME is set to 0x200, but I have no idea what would be
> > > sufficient to increase it too. Care to give me something I can replace
> > > that with and recompile?
> >
> > As I wrote to you in a private e-mail -
> >
> > I suggest that you first try to set BF_FRAME really high - say, to 0x4000 -
> > to confirm that this does indeed resolve the problem. Then decrease it
> > to a value that is 0x100 to 0x200 bytes higher than the required minimum
> > (such that it doesn't fail on another build with a similar version of gcc).
>
> I'll be trying that tonight... unfortunately, weekend is over and there
> are other things to patch.
--
{FEE30AD4 : 7F6C A60C 06C2 4811 FA1C A2BC 2EBC 5E32 FEE3 0AD4}
mysql> SELECT * FROM users WHERE clue > 0;
Empty set (0.00sec)
This archive was generated by hypermail 2.1.3 : Sat Jul 08 2006 - 09:17:02 PDT