Re: BUG: soft lockup detected on CPU#0! in sys_close and ext3

["Dennis J.A. Bijwaard" <dennis@h8922032063.dsl.speedlinq.nl>]

Hi Andrew,

Thanks for your reply. The machine has 512MB and some more swap:

Mem:    510960k total,   504876k used,     6084k free,     1868k buffers
Swap:   674640k total,     2652k used,   671988k free,   354832k cached

Machine may be slow for current standards, it has 2 * 500Mhz

Kind regards,
                Dennis

* Andrew Morton <akpm@osdl.org> [061015 21:13]:
> On Sun, 15 Oct 2006 19:56:40 +0200
> "Dennis J.A. Bijwaard" <dennis@h8922032063.dsl.speedlinq.nl> wrote:
> 
> > I got two soft lockups on one of the CPUs just now. I'm unsure if this
> > problem is in ext3, sys_close, or general kernel, so I've CC'd the
> > kernel list.
> > 
> > [1.] One line summary of the problem:
> > 
> > BUG: soft lockup detected on CPU#0! in sys_close/fput and ext3 journaling
> 
> Both warnings occurred when the kernel was tearing down large amounts of
> pagecache via invalidate_inode_pages().  One instances was a blockdev
> (probably the final close on the dvd) and the other was a regular file
> (perhaps a large dvd image?)
> 
> The CPU is slow: 500MHz pIII.  How much memory does it have?
> 
> So the kernel was doing a lot of work, on a slow CPU.  Perhaps that simply
> exceeded the softlockup timeout.  If that's true then the machine should
> have recovered.  Once it did, and once it didn't.  I don't know why it
> didn't.
>