Segfault in XML::LibXML
Bruce Miller <[email protected]>
| Newsgroups | gmane.comp.lang.perl.xml |
|---|---|
| Organization | NIST |
| Message-ID | <[email protected]> |
Hi all; I've just started getting one of those lovely errors we all enjoy so much. You know the ones: they sometimes give segfaults, but not every time, and only happens on long jobs... So, I ran valgrind (with PERL_DESTRUCT_LEVEL=2), which then takes _even_ longer, and end up with some kind of reference counting error, I presume? (see below). This is on fedora 10, perl-5.10.0-56 libxml2-2.6.3.1 perl-XML-LibXML-1.66-2 Undoubtedly, this isn't enough to proceed with, but as this is relatively unfamiliar territory for me: Does anyone have suggestions how to localize the problem better? > ==4053== Invalid read of size 8 > ==4053== at 0xAF9F838: PmmREFCNT_dec (in /usr/lib64/perl5/vendor_perl/5.10.0/x86_64-linux-thread-multi/auto/XML/LibXML/LibXML.so) > ==4053== by 0xAF32D1A: XS_XML__LibXML__Node_DESTROY (in /usr/lib64/perl5/vendor_perl/5.10.0/x86_64-linux-thread-multi/auto/XML/LibXML/LibXML.so) > ==4053== by 0x38EA4BF6B4: Perl_pp_entersub (in /usr/lib64/perl5/5.10.0/x86_64-linux-thread-multi/CORE/libperl.so) > ==4053== by 0x38EA4BA90B: Perl_call_sv (in /usr/lib64/perl5/5.10.0/x86_64-linux-thread-multi/CORE/libperl.so) > ==4053== by 0x38EA4D63DC: Perl_sv_clear (in /usr/lib64/perl5/5.10.0/x86_64-linux-thread-multi/CORE/libperl.so) > ==4053== by 0x38EA4D6D40: Perl_sv_free2 (in /usr/lib64/perl5/5.10.0/x86_64-linux-thread-multi/CORE/libperl.so) > ==4053== by 0x38EA4D66F0: Perl_sv_clear (in /usr/lib64/perl5/5.10.0/x86_64-linux-thread-multi/CORE/libperl.so) > ==4053== by 0x38EA4D6D40: Perl_sv_free2 (in /usr/lib64/perl5/5.10.0/x86_64-linux-thread-multi/CORE/libperl.so) > ==4053== by 0x38EA4ADAF0: Perl_hv_free_ent (in /usr/lib64/perl5/5.10.0/x86_64-linux-thread-multi/CORE/libperl.so) > ==4053== by 0x38EA4AF020: (within /usr/lib64/perl5/5.10.0/x86_64-linux-thread-multi/CORE/libperl.so) > ==4053== by 0x38EA4B1020: Perl_hv_undef (in /usr/lib64/perl5/5.10.0/x86_64-linux-thread-multi/CORE/libperl.so) > ==4053== by 0x38EA4D6805: Perl_sv_clear (in /usr/lib64/perl5/5.10.0/x86_64-linux-thread-multi/CORE/libperl.so) > ==4053== Address 0x2711ff30 is 0 bytes inside a block of size 34 free'd > ==4053== at 0x4A0609F: free (vg_replace_malloc.c:323) > ==4053== by 0x38EA4B05C1: Perl_hv_common (in /usr/lib64/perl5/5.10.0/x86_64-linux-thread-multi/CORE/libperl.so) > ==4053== by 0x38EA4C3B09: Perl_pp_helem (in /usr/lib64/perl5/5.10.0/x86_64-linux-thread-multi/CORE/libperl.so) > ==4053== by 0x38EA494AC6: Perl_runops_debug (in /usr/lib64/perl5/5.10.0/x86_64-linux-thread-multi/CORE/libperl.so) > ==4053== by 0x38EA4BB5D6: perl_run (in /usr/lib64/perl5/5.10.0/x86_64-linux-thread-multi/CORE/libperl.so) > ==4053== by 0x400CF3: main (in /usr/bin/perl) > > ==4053== > ==4053== ERROR SUMMARY: 2 errors from 1 contexts (suppressed: 4 from 1) > ==4053== malloc/free: in use at exit: 56,246 bytes in 521 blocks. > ==4053== malloc/free: 323,176,837 allocs, 323,176,316 frees, 21,619,182,773 bytes allocated. > ==4053== For counts of detected errors, rerun with: -v > ==4053== searching for pointers to 521 not-freed blocks. > ==4053== checked 803,936 bytes. > ==4053== > ==4053== LEAK SUMMARY: > ==4053== definitely lost: 0 bytes in 0 blocks. > ==4053== possibly lost: 0 bytes in 0 blocks. > ==4053== still reachable: 56,246 bytes in 521 blocks. > ==4053== suppressed: 0 bytes in 0 blocks. > ==4053== Rerun with --leak-check=full to see details of leaked memory. -- [email protected] http://math.nist.gov/~BMiller/ _______________________________________________ Perl-XML mailing list [email protected] To unsubscribe: http://listserv.ActiveState.com/mailman/mysubs