Summary: | ASTERISK-24561: Core dump on res_fax_spandsp | ||
Reporter: | Jakub Włudarczyk (jwludarczyk@emakom.pl) | Labels: | |
Date Opened: | 2014-11-26 09:43:46.000-0600 | Date Closed: | 2015-03-11 17:30:04 |
Priority: | Major | Regression? | |
Status: | Closed/Complete | Components: | Resources/res_fax_spandsp |
Versions: | 12.6.1 12.7.0 | Frequency of Occurrence | Frequent |
Related Issues: | |||
Environment: | Debian Jessie spandsp-0.0.6pre21 build from source | Attachments: | ( 0) backtrace.txt ( 1) backtrace.txt ( 2) full.7z |
Description: | Crash when using res_fax_spandsp T.38 Gateway. Unfortunately its impossible to use valgrind, as this is on production, | ||
Comments: | By: Rusty Newton (rnewton) 2014-11-26 15:23:11.713-0600 Please remove the trace and reattach it without marking it as contribution (it doesn't show up as you don't have a license, and regardless, it isn't a contribution). Can you reproduce the crash? If so, can you please attach an Asterisk full log showing a few minutes up until the crash, with DEBUG messages enabled, set to 5 or above? By: Jakub Włudarczyk (jwludarczyk@emakom.pl) 2014-11-27 03:12:18.221-0600 backtrace By: Jakub Włudarczyk (jwludarczyk@emakom.pl) 2014-11-27 03:26:51.933-0600 backtrace reatached, i`ve turned debug on, will get back with logs when it happens(almost everyday) By: Jakub Włudarczyk (jwludarczyk@emakom.pl) 2014-12-01 04:38:47.023-0600 full log(debug=5) of todays crash By: Jakub Włudarczyk (jwludarczyk@emakom.pl) 2014-12-03 03:18:11.564-0600 full log attached By: Rusty Newton (rnewton) 2014-12-09 17:05:39.330-0600 Looks like there might be memory corruption. You already said you can't get Valgrind output, but you can you get output with the MALLOC_DEBUG compiler flag? Without either of those, there may be nothing further a developer can do. Instructions: https://wiki.asterisk.org/wiki/display/AST/MALLOC_DEBUG+Compiler+Flag By: Jakub Włudarczyk (jwludarczyk@emakom.pl) 2014-12-17 07:53:12.007-0600 unfortunately ive been waiting for 7 days, and it hasn`t crashed, will wait a bit more, mmlog contains some warnings: WARNING: High fence violation of 0x7f1f60078060 allocated at res_fax_spandsp.c spandsp_fax_new() line 526 WARNING: High fence violation of 0x7f1fa80cdaf0 allocated at res_fax_spandsp.c spandsp_fax_new() line 526 By: Rusty Newton (rnewton) 2014-12-17 09:45:28.491-0600 Okay we'll give it another few weeks. By: Richard Mudgett (rmudgett) 2014-12-17 13:30:56.180-0600 MALLOC_DEBUG has likely prevented the crash because critical memory wasn't corrupted by the extra memory allocated for the high fence pattern. By: Rusty Newton (rnewton) 2014-12-17 16:01:37.147-0600 Now knowing what Richard mentioned - can you provide some additional debug? While running with MALLOC_DEBUG... * A full debug log, similar to what you already have, but additionally "fax set debug on" enabled, and specifically captured during the time of the WARNING's observed in the mmlog. * What version of SpanDSP are you using? * PCAP of the call that was happening during the time of the mmlog warning messages if possible? By: Jakub Włudarczyk (jwludarczyk@emakom.pl) 2015-01-08 02:12:08.713-0600 Unfortunately warnings dont occur in the mmlog any more. I will turn debug on and wait for new warnings. Im using spandsp-0.0.6pre21 build from source By: Rusty Newton (rnewton) 2015-01-29 17:58:41.809-0600 Still no more warnings? By: Rusty Newton (rnewton) 2015-02-20 19:17:12.035-0600 If you stop using mmlog, does the crash occur any longer? I'm not sure where we can go from here without the additional information requested. By: Rusty Newton (rnewton) 2015-03-11 17:30:15.332-0500 Suspended due to lack of activity. Please request a bug marshal in #asterisk-bugs on the IRC network irc.freenode.net to reopen the issue should you have the additional information requested. Further information on issue tracker usage can be found in the Asterisk Issue Guidlines [1]. [1] https://wiki.asterisk.org/wiki/display/AST/Asterisk+Issue+Guidelines |