Created attachment 357133 [details] x86_64 bzip2 compressed core dump of checkout as described above Description of problem: I'm getting an occasional segfault doing an upload. I've captured a core (attached) which shows it dying in an apparently harmless malloc. This presumably points to heap corruption somewhere. Version-Release number of selected component (if applicable): 0aaf061c2f73f8852c59afaa9706f42bc578caae with the attached tiny patch applied, compiled with CFLAGS=-g. How reproducible: Highly variable. Was about 50%, now much less. I'm guessing it's a race condition affected by disk caches. Steps to Reproduce: 1. LIBGUESTFS_DEBUG=1 ./fish/guestfish -a /media/passport/RHEL53PV64-20090213.img.qcow2 -m /dev/VolGroup00/LogVol00:/ -m /dev/sda1:/boot 2. upload ~/Download/strace-4.5.18-2.el5_3.3.x86_64.rpm /tmp/strace-4.5.18-2.el5_3.3.x86_64.rpm
Created attachment 357134 [details] The patch applied against 0aaf061c2f73f8852c59afaa9706f42bc578caae when core was produced
Created attachment 357135 [details] Some gdb output from attached core
Setting product to Virtualization Tools. This needs to be retested against a more recent libguestfs since that part of the code has been substantially modified.
Unknown, but unlikely this still affects the latest versions of libguestfs.