Re: ls -al .zfs/snapshots dir reboots machine

From: Hidetoshi Shimokawa <simokawa_at_FreeBSD.ORG>
Date: Wed, 18 Apr 2007 00:02:06 +0900
On 4/17/07, Pawel Jakub Dawidek <pjd_at_freebsd.org> wrote:
> On Tue, Apr 17, 2007 at 10:51:37AM +0900, Hidetoshi Shimokawa wrote:
> > I can repoduce this hang.
> >
> > # cd /pool/test/.zfs/snapshot
> > # /bin/ls
> > snap
> > # /bin/ls -F
> > ls: snap: No such file or directory
> > # /bin/ls -F
> > (hangs)
>
> Are you able to break into debugger? I was able to reproduce it once
> (but with panic in crdup()), but I can no longer do it...

In my case, only 'ls' hangs and other part of the system is normal.

# ps alxg | grep ls
    0  2129  2118   0  -4  0  6100  1512 zfs    D+    p0    0:00.01 /bin/ls -F

Once, I got "panic: lockmgr: locking against myself"  10 or 20 min. later
of "ls -F".

> > It looks like another name cache problem for me.
> > I noticed 'cache_purge(vp)' is commented out in domount() in opensolaris_vfs.c.
> > Shouldn't we need 'dnlc_purge_vp(vp)' here?
>
> There is dnlc_purge_vfsp() at the begining of zfs_umount(), which should
> be sufficient.

domount() of opensolaris also has dnlc_purge_vp() to invalidate cached
entry for the mount point. Do you really think it's unnecessary?

http://src.opensolaris.org/source/xref/onnv/onnv-gate/usr/src/uts/common/fs/vfs.c#1383

-- 
/\ Hidetoshi Shimokawa
\/  simokawa_at_FreeBSD.ORG
Received on Tue Apr 17 2007 - 13:02:09 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:08 UTC