Re: kern/56189: System panicked: /: bad dir ino 5045785 at offset 0: Bad dir (too big)

[nia <nia%NetBSD.org@localhost>]

The following reply was made to PR kern/56189; it has been noted by GNATS.

From: nia <nia%NetBSD.org@localhost>
To: gnats-bugs%netbsd.org@localhost
Cc: 
Subject: Re: kern/56189: System panicked: /: bad dir ino 5045785 at offset 0:
 Bad dir (too big)
Date: Fri, 4 Jun 2021 21:12:33 +0000

 On Fri, Jun 04, 2021 at 07:20:01PM +0000, David Holland wrote:
 >  On Wed, May 19, 2021 at 10:20:00AM +0000, nia%pkgsrc.org@localhost wrote:
 >   > System panicked: /: bad dir ino 5045785 at offset 0: Bad dir (too
 >   > big), reclen=0x4918, namlen=61, dirsiz=72 <= reclen=18712 <=
 >   > maxsize=512, flags=0x46005000, entryoffsetinblock=0, dirblksiz=512
 >  
 >  That's a thoroughly corrupt directory block; most likely it isn't a
 >  directory block at all.
 >  
 >  My first recommendation would be: track down the directory whose inode
 >  is 5045785 and see if it's garbaged. Also it might be interesting to
 >  print the inode structure (from fsdb or maybe ddb). Then (whether or
 >  not it's overtly borked) it's probably a good idea to fsck.
 
 Hmm, this didn't occur to me for some reason.
 
 The directory is /usr/share/man/man8/cobalt (???) and appears normal.
 
 After this panic occurred, I was advised to force a fsck - but it
 didn't find anything. As mentioned, the problem went away on reboot
 and hasn't appered again.
 
 >  I see from the fstab that it's wapbl; what kind of storage is it? (In
 >  particular, is it a consumer-grade SSD that has been subjected to a
 >  power failure recently?)
 
 Consumer grade SSD; SATA; 1TB.
 It gets subjected to power failure a lot, unfortunately.