Oh, well... Still nothing useful on that livelock (I had something
that looked kinda-sorta like a non-invasive solution, but it deadlocks), so
it's just Miklos' vmsplice fix for now. Please, pull from the usual place -
git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs.git for-linus
Shortlog:
Miklos Szeredi (1):
vfs: fix vmplice_to_user()
Diffstat:
fs/splice.c | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)
PS: livelock looks triggerable without check_submounts_and_drop() anywhere
in sight - vfs_rmdir() does shrink_dcache_parent() on prospective victim
right before calling ->rmdir(). So rmdir a vs rmdir a/b/c will get you
shrink_dcache_parent() on a and a/b/c at the same time - no locks in common
there. So we'll probably need to shift those calls after successful calls
of ->rmdir() (and similar in overwriting ->rename()), in addition to whatever
we cook in check_submounts_and_drop()...
I still would like to see detailed traces on what gets triggered by udev -
hopefully the patch I've sent to Mika won't disrupt the livelock by printks
it's doing ;-/
--
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
