On 2015-11-17 03:08, Scotty Edmonds wrote:
> Sorry, I'm not at all familiar with backtrace.
>
General procedure to get a backtrace on a system without core files:
1. Make sure you have debugging symbols installed for the program you 
want the backtrace for, and ideally any libraries it uses.  (These 
usually are in packages with names ending in -dbg or -debug.  In the 
case of btrfs, you'll probably need symbols for glibc, lzo, and zlib).
2. Start the program under gdb ('gdb /sbin/btrfs')
3. When gdb is done loading, type in 'start' followed by the arguments 
for the program (in this case, 'rescue chunk-recover -y /dev/sdg')
4. If gdb then says something about stopping at a breakpoint, enter the 
command 'continue'.
5. When the program being debugged crashes, you should get a backtrace 
from gdb, if not, you can enter the command 'bt' to tell it to dump a 
backtrace.
6. After copying the backtrace somewhere you can put it in an e-mail, 
you can exit gdb with the 'exit' command, it may complain about 
something still running, it's safe to tell it to kill this process.

If instead you have your system configured for core dumps:
1. Make sure you have debugging symbols installed for the program you 
want the backtrace for, and ideally any libraries it uses.
2. Run gdb with the program and core file ('gdb /sbin/btrfs 
/path/to/core/file').
3. Enter the command 'bt to get a backtrace.