[Cryptominisat-devel] I need a bit of help debuggingCryptoMiniSat

Mate Soos soos.mate at gmail.com
Jeu 20 Jan 01:02:42 CET 2011


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Dear Vegard,

On 01/20/2011 12:25 AM, Vegard Nossum wrote:
> Here are some suggestions, some of them might require a bit of work, though:
> 
> 2. Test for the bug (using the test-case) in the configure script. If
> it triggers, give a warning and/or compile with lower -O level.

Good idea. "-O2 -fno-tree-pre" works fine, and according to the bug
tracker, the problem seems to be with "-ftree-pre", so the workaround
will be a check and if it fails, then addition of "-fno-tree-pre" as
extra compilation option, plus a warning to the user.

> In any case, it might be an idea to include the gcc version that was
> used to compile CryptoMiniSat somewhere in the solver output? There is
> a very simple way to do this -- the __VERSION__ macro is a string with
> the gcc version.

Good idea. Will do that.

> PS: I wonder if this is the same bug that hit me initially with
> --nthreads, because that too triggered using the -ftree-pre flag
> (which is enabled in -O3) and it was also gcc 4.5.1...

I am now certain that that's the case. I/we have received bug reports
with every 4.5.x version from 4.5.0 to 4.5.3. I was actually very
surprised that you have managed to point it at "-ftree-pre" at the time.

Finally, CryptoMS 2.9.0 will be released tomorrow :)

Bests,

Mate

- -- 
Mate Soos
Security Research Labs
http://www.srlabs.de
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAk03e6IACgkQsTOOstKb0jmPDACdGEXPrdzcNaO51BZY+4MuCSLQ
LEYAn3NiCIe6FzJtFs2GyKWKfwVZrp+l
=TR9D
-----END PGP SIGNATURE-----



More information about the Cryptominisat-devel mailing list