Bug 32574

Summary:	ufsdirhash_findfree: bad stats
Product:	Base System	Reporter:	Martin Kammerhofer <dada>
Component:	kern	Assignee:	iedowse
Status:	Closed FIXED
Severity:	Affects Only Me	CC:	mkamm
Priority:	Normal
Version:	4.4-STABLE
Hardware:	Any
OS:	Any

Description Martin Kammerhofer 2001-12-07 10:10:00 UTC

PANIC during "make -j8 buildworld"

Transscript of panic message:
	panic: ufs_dirhash_findfree: bad stats

Stack trace (function names only):
	Debugger
	panic
	ufs_dirhash_findfree
	ufs_lookup
	ufs_vnoperate
	ufs_cache_lookup
	ufs_vnoperate
	lookup
	namei
	vn_open
	open
	syscall2
	Xint0x80_syscall

Fix: 

Removing "options UFS_DIRHASH" from
	the kernel configuration will help.
How-To-Repeat: 	Not known.

Comment 1 Sheldon Hearn freebsd_committer

2001-12-07 11:06:44 UTC

Responsible Changed
From-To: freebsd-bugs->iedowse

Ian's the UFS_DIRHASH man.

Comment 2 iedowse 2001-12-07 11:09:56 UTC

In message <Pine.LNX.4.33.0112071103380.785-100000@pluto.tu-graz.ac.at>, Martin
 Kammerhofer writes:

>PANIC during "make -j8 buildworld"
>
>Transscript of panic message:
>	panic: ufs_dirhash_findfree: bad stats
>
>Stack trace (function names only):
>	Debugger
>	panic
>	ufs_dirhash_findfree
>	ufs_lookup

Thanks for the problem report - is this repeatable? If so, could
you try to get a crash dump and a corresponding debug kernel? I'd
be particularly interested in the values of the various variables
used in the ufsdirhash_findfree() function (e.g *dh, dh->dh_blkfree[],
dh->dh_firstfree[]).

The panic is a dirhash-internal consistency check, so either there
is a dirhash bug, or something has corrupted memory.

>options 	ENABLE_VFS_IOOPT

Have you enabled "vfs.ioopt"? It was mentioned on a mailing list
recently that there are problems with this optimisation. It's
unlikely to be the cause of the panic, but worth knowing.

Ian

Comment 3 iedowse 2001-12-11 18:47:57 UTC

In message <Pine.BSF.4.21.0112111929540.57263-100000@homebox.kammerhofer.org>, 
Martin Kammerhofer writes:
>Em - I'm really sorry, but further investigation has revealed, that this
>panic was caused by a memory module. It coincidently went dodgy shortly
>after I compiled a kernel with UFS_DIRHASH.
>
>Sorry again about the false alarm. Please close this PR.

Phew! Thanks for the update. I'll close the PR, but if it happens
again or you have any reason to suspect that there is a real dirahsh
bug, then please let me know.

Thanks,

Ian

Comment 4 iedowse freebsd_committer

2001-12-11 19:33:05 UTC

State Changed
From-To: open->closed


Closed at the submitter's request; the problem was apparently caused 
by bad memory.