[10:21:39] --- abo has left
[10:22:22] --- abo has become available
[10:51:38] --- jaltman has left: Disconnected
[11:11:15] --- jaltman has become available
[11:33:48] --- Russ has become available
[11:52:47] --- jaltman has left: Replaced by new connection
[11:53:36] --- jaltman has become available
[12:08:25] --- deason has become available
[12:32:33] --- jaltman has left: Disconnected
[12:32:42] --- jaltman has become available
[15:12:12] <kaduk@mit.edu/barnowl> Hm, lock order reversal with a vnode lock after the glock while
symlinking.
[15:14:49] <jaltman> where?
[15:16:54] <kaduk@mit.edu/barnowl> FBSD/osi_vnodeops:555
afs_vcache.c:2461
FBSD/osi_vnodeops.c:1374
(my sources are slightly tweaked, so the line numbers may be a bit
off.)
The backtrace isn't terribly enlightening.
[15:17:37] <kaduk@mit.edu/barnowl> Pretty blatant, looking at the afs_vop_symlink() source.
[15:38:56] <kaduk@mit.edu/barnowl> I suppose I should note that the actual operation was a fs mkmount,
not a manual symlink.
[17:11:59] <kaduk@mit.edu/barnowl> ... and now I'm getting spurious ENOENTs.
[17:18:07] --- jaltman has left: Replaced by new connection
[17:18:08] --- jaltman has become available
[17:35:29] --- jaltman has left: Disconnected
[17:35:50] --- jaltman has become available
[18:37:05] --- jaltman has left: Disconnected
[18:40:07] --- jaltman has become available
[21:20:10] <kaduk@mit.edu/barnowl> Hm, it seems that this is a reclaim failure cropping up ...
[22:00:54] --- deason has left
[22:32:06] --- Russ has left: Disconnected