Random Segfaults on FreeBSD 9.0-RELEASE and 9.1-RC1

irssibot commented 11 years ago

- - Issue #875 opened by @rfrederick

Irssi 0.8.15 built from FreeBSD Ports segfaults at random times on both FreeBSD 9.0-RELEASE and 9.1-RC1 (arch is i386). The backtrace for each segfault is somewhat similar.

Backtrace #1:

#0  0x0806acf9 in textbuffer_remove ()
#1  0x0805be4c in gui_printtext_deinit ()
#2  0x080dabce in signal_stop ()
#3  0x080db113 in signal_emit_id ()
#4  0x08096d5c in printtext_deinit ()
#5  0x080dabce in signal_stop ()
#6  0x080db113 in signal_emit_id ()
#7  0x08096ff0 in printtext_init ()
#8  0x0809725a in printformat_module_dest_charargs ()
#9  0x08097325 in printformat_module_dest_args ()
#10 0x0809742c in printformat_module_args ()
#11 0x08097477 in printformat_module ()
#12 0x0808cf52 in expand_emphasis ()
#13 0x080dabce in signal_stop ()
#14 0x080db1bf in signal_emit ()
#15 0x08075922 in fe_events_deinit ()
#16 0x080dabce in signal_stop ()
#17 0x080db1bf in signal_emit ()
#18 0x080a2513 in irc_irc_deinit ()
#19 0x080dabce in signal_stop ()
#20 0x080db113 in signal_emit_id ()
#21 0x080a21c3 in irc_irc_deinit ()
#22 0x080dabce in signal_stop ()
#23 0x080db113 in signal_emit_id ()
#24 0x080a23db in irc_irc_deinit ()
#25 0x080cce1e in mask_match ()
#26 0x2836cabd in g_io_channel_unix_get_fd () from /usr/local/lib/libglib-2.0.so.0
#27 0x28326247 in g_main_context_dispatch () from /usr/local/lib/libglib-2.0.so.0
#28 0x2832a2ce in g_main_context_prepare () from /usr/local/lib/libglib-2.0.so.0
#29 0x2832a8f5 in g_main_context_iteration () from /usr/local/lib/libglib-2.0.so.0
#30 0x0806e9a4 in main ()

Backtrace #2:

#0  0x0806acf9 in textbuffer_remove ()
#1  0x0805be4c in gui_printtext_deinit ()
#2  0x080dabce in signal_stop ()
#3  0x080db113 in signal_emit_id ()
#4  0x08096d5c in printtext_deinit ()
#5  0x080dabce in signal_stop ()
#6  0x080db113 in signal_emit_id ()
#7  0x08096ff0 in printtext_init ()
#8  0x080976d2 in printtext_string_window ()
#9  0x0809f550 in window_find_name ()
#10 0x0809f5d7 in window_find_name ()
#11 0x28300c46 in g_source_get_time () from /usr/local/lib/libglib-2.0.so.0
#12 0x28300247 in g_main_context_dispatch () from /usr/local/lib/libglib-2.0.so.0
#13 0x283042ce in g_main_context_prepare () from /usr/local/lib/libglib-2.0.so.0
#14 0x283048f5 in g_main_context_iteration () from /usr/local/lib/libglib-2.0.so.0
#15 0x0806e9a4 in main ()

At one point I also received a perl core dump along with the irssi core dump.

Perl Backtrace:

#0  0x28139d11 in Perl_sv_2mortal () from /usr/local/lib/perl5/5.14.2/mach/CORE/libperl.so
#1  0x2815dbba in Perl_leave_scope () from /usr/local/lib/perl5/5.14.2/mach/CORE/libperl.so
#2  0x2818b5d0 in S_regtry () from /usr/local/lib/perl5/5.14.2/mach/CORE/libperl.so
#3  0x28198aa2 in Perl_regexec_flags () from /usr/local/lib/perl5/5.14.2/mach/CORE/libperl.so
#4  0x28132c7f in Perl_pp_subst () from /usr/local/lib/perl5/5.14.2/mach/CORE/libperl.so
#5  0x2812eefb in Perl_runops_standard () from /usr/local/lib/perl5/5.14.2/mach/CORE/libperl.so
#6  0x280c7704 in perl_run () from /usr/local/lib/perl5/5.14.2/mach/CORE/libperl.so
#7  0x08048b11 in main ()

irssibot commented 11 years ago

- - Comment 1565 by @rfrederick

The latest segafault (backtrace below) looks to have occurred when the screen_away.pl script was marking me as away after detachment of the GNU screen instance irssi was running in.

#0  0x0806acf9 in textbuffer_remove ()
#1  0x0805be4c in gui_printtext_deinit ()
#2  0x080dabce in signal_stop ()
#3  0x080db113 in signal_emit_id ()
#4  0x08096d5c in printtext_deinit ()
#5  0x080dabce in signal_stop ()
#6  0x080db113 in signal_emit_id ()
#7  0x08096ff0 in printtext_init ()
#8  0x0809725a in printformat_module_dest_charargs ()
#9  0x08097325 in printformat_module_dest_args ()
#10 0x0809742c in printformat_module_args ()
#11 0x08097477 in printformat_module ()
#12 0x08076cc1 in fe_events_numeric_deinit ()
#13 0x080dabce in signal_stop ()
#14 0x080db1bf in signal_emit ()
#15 0x080a2513 in irc_irc_deinit ()
#16 0x080dabce in signal_stop ()
#17 0x080db113 in signal_emit_id ()
#18 0x080a21c3 in irc_irc_deinit ()
#19 0x080dabce in signal_stop ()
#20 0x080db113 in signal_emit_id ()
#21 0x080a23db in irc_irc_deinit ()
#22 0x080cce1e in mask_match ()
#23 0x28346abd in g_io_channel_unix_get_fd () from /usr/local/lib/libglib-2.0.so.0
#24 0x28300247 in g_main_context_dispatch () from /usr/local/lib/libglib-2.0.so.0
#25 0x283042ce in g_main_context_prepare () from /usr/local/lib/libglib-2.0.so.0
#26 0x283048f5 in g_main_context_iteration () from /usr/local/lib/libglib-2.0.so.0
#27 0x0806e9a4 in main ()

irssibot commented 11 years ago

- - Comment 1566 by @bazerka

Can you please provide a list of your loaded scripts?

TIA.

irssibot commented 11 years ago

- - Comment 1567 by @bazerka

Also, are you using irssi from ports? If so, what version? ( pkg_info | grep irssi )

irssibot commented 11 years ago

- - Comment 1568 by @bazerka
If you're using 0.8.15_4 from ports, can you please try deleting /usr/ports/irc/irssi/files/patch-irssi-utf8 and re-build/re-install the port and see if that alleviates the segfaulting for you?

[Edit: portversion typo]

irssibot commented 11 years ago

- - Comment 1569 by @rfrederick

Yes, I'm running irssi from ports, version 0.8.15_4. The scripts I normally have loaded are autorejoin.pl, cron.pl, nickcolor.pl, and screen_away.pl.

I've recompiled the port without the utf8 patch and will see if it has any effect.

irssibot commented 11 years ago

- - Comment 1570 by @rfrederick

I've run both 0.8.15_4 without the utf8 patch and 0.8.15_5 (in which the utf8 patch was removed) for a week now without any segfaults.

irssibot commented 11 years ago

- - Comment 1571 by @bazerka

That's good news, thanks for checking.

irssibot commented 11 years ago

- - Comment 1637 by @kcwu

Hi Ryan, What encoding do you use?

irssibot commented 11 years ago

- - Comment 1639 by @rfrederick

I'm currently using UTF-8 as my encoding.

irssibot commented 11 years ago

- - Comment 1640 by @kcwu

Hi Ryan, Could you help try this http://kcwu.csie.org/~kcwu/tmp/diff-irssi-utf8-2.txt ?

irssibot commented 11 years ago

- - Comment 1641 by @rfrederick

I've rebuilt irssi with the patch. I'll give it a week or so to see if any segfaults arise.

irssibot commented 11 years ago

- - Comment 1644 by @rfrederick

I finally had a chance to run irssi for an extended period of time with the patch, and I have yet to experience any segfaults after around 11 days of uptime.

irssibot commented 11 years ago

- - Closed by @bazerka

Problem was caused by a downstream patch.

irssi-import / bugs.irssi.org

Random Segfaults on FreeBSD 9.0-RELEASE and 9.1-RC1 #875