Helmut, Thanks for the sample.
> > > It is. Anyway, the last thing I updated was the p5-URI port on Jan > > > 14th and 30 minutes later the problems began. > > > BTW, only one single mail (from ~ 5.000 a day) is affectd by now... > > Finally, I was able to get the core file. And (surpise, surprise ;) ) > it is libc: > > Core was generated by `perl5.8.9'. > Program terminated with signal 11, Segmentation fault. > (gdb) bt > #0 0x2827e3ca in malloc () from /lib/libc.so.7 > #1 0x2827fb42 in realloc () from /lib/libc.so.7 > #2 0x280fba37 in Perl_safesysrealloc () from > /usr/local/lib/perl5/5.8.9/mach/CORE/libperl.so > #3 0x281447e8 in Perl_savestack_grow_cnt () from > /usr/local/lib/perl5/5.8.9/mach/CORE/libperl.so > #4 0x2816e02b in S_regcppush () from > /usr/local/lib/perl5/5.8.9/mach/CORE/libperl.so > #5 0x28172ba1 in S_regmatch () from > /usr/local/lib/perl5/5.8.9/mach/CORE/libperl.so This backtrace just corresponds to Perl running out of stack during regexp evaluation. > This GDB was configured as "i386-marcel-freebsd" > The perl is from ports I presume. Which version? > 5.8.9 multiThreaded Running a multithreaded perl with an application that is heavy on regexp evaluations (like SpamAssassin) is asking for trouble, especially on FreeBSD, where by default a per-thread stack size is rather small. Regular expressions in perl 5.8.* can require lots of stack space on degenerated input strings. In your case the crash happens in code produced by do_uri_tests() in the Mail::SpamAssassin::Plugin::Check - both the SA 3.2.5 and 3.3.0 are alike in this respect. Just a plain regexp evaluation in a complex regular expression with a degenerate input data. Reconfiguring perl 5.8.9 with disabled multithreading will provide a huge stack compared to yours, reducing the chance of a crash happening. I believe the 5.10.* now uses heap for regexp evaluations, so this would be another choice - perl 5.10.1 is in ports too, and works very well with SpamAssassin. Mark ------------------------------------------------------------------------------ Throughout its 18-year history, RSA Conference consistently attracts the world's best and brightest in the field, creating opportunities for Conference attendees to learn about information security's most important issues through interactions with peers, luminaries and emerging and established companies. http://p.sf.net/sfu/rsaconf-dev2dev _______________________________________________ AMaViS-user mailing list AMaViS-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/amavis-user AMaViS-FAQ:http://www.amavis.org/amavis-faq.php3 AMaViS-HowTos:http://www.amavis.org/howto/
