Re: 2.2.15pre12: inode leak or something, maybe smbfs related

From: Ville Herva (
Date: Mon Apr 24 2000 - 09:32:49 EST

On Mon, Apr 24, 2000 at 01:25:17PM +0100, you [Alan Cox] claimed:
> > Apr 24 10:05:25 babbage kernel: Call Trace: [reiserfs_iget+59/116] [reiserfs_lookup+79/184] [reiserfs_lookup+124/184] [do_anonymous_page+110/128] [real_lookup+92/180] [do_no_page+75/248] [lookup_dentry+300/496]
> > Apr 24 10:05:25 babbage kernel: [__namei+47/100] [do_page_fault+271/952] [sys_access+163/308] [error_code+45/64] [system_call+52/64]
> > Apr 24 10:05:25 babbage kernel: Code: 8b 13 8b 01 39 c2 73 08 b8 ff ff ff ff eb 1a 90 39 c2 76 08
> >
> > I guess that's not a reisefs bug... Dir cache corruption or something?
> >
> > I tried lsof -p:ing many processes, but none seems to use more files than
> > it should.

One thing I just thought of is that I messed up with a FAT partition
before that. I copied stuff to there while vmware was running and
simultaneously accessing the partition. Then I deleted the stuff. All this
was clearly not too clever, and I don't blaim fatfs for not too happy:

Apr 23 11:43:16 babbage kernel: Filesystem panic (dev 03:01).
Apr 23 11:43:16 babbage kernel: fat_free: deleting beyond EOF
Apr 23 11:43:16 babbage kernel: File system has been set read-only

I just tought that is it possible, that it could have corrupted the dir
cache in the first place?

The night before, updatedb was messing with the smbfs mount as well, and
it produced this stuff,

Apr 23 04:45:15 babbage kernel: smb_proc_readdir_long: name=\home\vherva\Sharity.1.08\*, entries=0, rcls=1, err=5
Apr 23 04:45:15 babbage kernel: smb_refill_dircache: readdir failed, result=-13

but not this

Apr 23 05:56:57 babbage kernel: smb_setup_header: Aieee, xmit len > packet! len=8196, size=8192

as it did after the fat accident. If that's possible, we can perhaps
forget about the smbfs oops.
> Thats almost certainly a reiserfs bug. It may well be reiserfs failing to
> handle the fact smbfs stole all the inodes or similar however.

Sounds reasonable.
> There is a 2.2.16pre1 (v 2.2.15pre19 since Linus hasnt replied to the requests
> to put out 2.2.15 final yet) on That has some changes in it
> from Tridge to clean up a pile of smbfs bits and bobs. I'd be interested
> to know if that works (but without reiserfs also patched in)

Ok. I'll give it a try.

-- v --

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to
Please read the FAQ at

This archive was generated by hypermail 2b29 : Sun Apr 30 2000 - 21:00:07 EST