Re: mutex lock assertion panic in vm_page_free() on sparc64

From: Ruslan Ermilov <ru_at_freebsd.org>
Date: Wed, 23 Feb 2005 10:45:52 +0200
Hi Alan,

On Tue, Feb 22, 2005 at 03:29:41PM -0600, Alan Cox wrote:
> On Tue, Feb 22, 2005 at 02:22:56PM -0500, Jung-uk Kim wrote:
> > 
> > FYI, anholt says 'backing vm/uma_core.c to r1.114 has fixed the 
> > issue':
> > 
> > http://docs.freebsd.org/cgi/mid.cgi?1109089389.4267.7.camel
> 
> Ok, thanks.  I'd still like to see the contents of the offending
> page.  When UMA overwrites the page's object field with a slab
> pointer, it sets a flag within the page.  I'd very much like
> to know if this flag is set.
> 
I had trouble saving a kernel core -- 6 or 7 times I tried this,
it resulted in the same panic.  Finally, I gave up and reverted
to vm/uma_core.c,v 1.114, this game me a working savecore.
Unfortunately, either kdgb(1) doesn't work on sparc64 at the
moment, or this core is troublesome -- it would endlessly print
this:

kgdb: kvm_read: invalid address (203d2025)
kgdb: kvm_read: invalid address (6e666f20)
kgdb: kvm_read: invalid address (203d2025)
kgdb: kvm_read: invalid address (6e666f20)
kgdb: kvm_read: invalid address (203d2025)
kgdb: kvm_read: invalid address (6e666f20)
kgdb: kvm_read: invalid address (203d2025)
kgdb: kvm_read: invalid address (6e666f20)
...

At least I can confirm that reverting to rev. 1.114 fixes a
problem.


Cheers,
-- 
Ruslan Ermilov
ru_at_FreeBSD.org
FreeBSD committer

Received on Wed Feb 23 2005 - 07:45:51 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:28 UTC