[Bug 10925] non-atomic xattr replacement in btrfs => rsync --read-batch random errors

samba-bugs at samba.org samba-bugs at samba.org
Wed Nov 19 09:21:56 MST 2014


--- Comment #1 from roland <devzero at web.de> ---
interesting find, if btrfs has xattr races, but the question is how to produce
an appropriate repro case.

on a tiny btrfs here on my debian wheezy system, i did some massive parallel
run of 

"setfattr -h -n user.myattr -v attrval dir" in one window, 
whereas doing
getfattr -h -n user.myattr dir  in another window, 

and would have expected to seem some errors. i didn`t. as this is the default
kernel, maybe the race is being introduced in a later version.

furthermore - what specific problem (security or whatever) do you see here with
rsync, if there is a race window between setting and reading an xattr (which
seems is not to be expected behaviour of a filesystem) ?

also see:
If you have any bug, problems, performance issues or questions while using
Btrfs, please email the Btrfs mailing list (no subscription required). Please
report bugs also on Bugzilla.

You are receiving this mail because:
You are the QA Contact for the bug.

More information about the rsync mailing list