Red Hat Bugzilla – Bug 984229
[abrt] icedax-1.1.11-14.fc18: register_index_position: Process /usr/bin/icedax was killed by signal 11 (SIGSEGV)
Last modified: 2016-11-24 10:53:28 EST
Description of problem:
icedax dev=/dev/sr1 -vall cddb=0 -info-only
it crashed with the message:
index scan: 12...Memory fault(coredump)
Version-Release number of selected component:
cmdline: cdda2wav dev=/dev/sr1 -vall -B cddb=1 -info-only
runlevel: N 5
var_log_messages: Jul 13 13:57:11 localhost abrt: Saved core dump of pid 4857 (/usr/bin/icedax) to /var/tmp/abrt/ccpp-2013-07-13-13:57:11-4857 (2686976 bytes)
Thread no. 1 (2 frames)
#2 register_index_position at /usr/src/debug/cdrkit-1.1.11/icedax/toc.c:3093
#3 ScanIndices at /usr/src/debug/cdrkit-1.1.11/icedax/toc.c:3199
Created attachment 773150 [details]
Created attachment 773151 [details]
Created attachment 773152 [details]
Created attachment 773153 [details]
Created attachment 773154 [details]
Created attachment 773155 [details]
Created attachment 773156 [details]
Created attachment 773157 [details]
I can not reproduce the crash. Works for me.
Maybe it will be interesting for glibc people, since the crash was inside a call to malloc(16).
99.999% of the time if you're getting an error in the allocator it's actually the fault of the user program or some library used by the user program, not glibc.
The failure location is not particularly informative other than it's in the use of fast small bins which are acceptable to corruption since their information is encoded right beside the allocation on the same page of memory. Therefore any overwrites by the application can corrupt the fast bins. The only way to harden the allocator is to disable fast bins, and add checking (which you can do via the API).
Marking CLOSED/WORKSFORME based on Frantisek's comments.