Current-Users archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: crash in tmpfs_itimes since vmlocking2 merge



On Mon, Jan 21, 2008 at 11:50:20PM +0200, Antti Kantee wrote:
> On Mon Jan 21 2008 at 15:33:41 -0600, David Young wrote:
> > The following crash during shutdown is fairly repeatable.  It began
> > to happen after I updated to -current from pre-vmlocking2 sources.
> > It may be necessary to reproduce my thicket of null mounts in order to
> > reproduce the crash,
> > 
> > /dev/wd0a on / type ffs (read-only, local)
> > tmpfs on /dev type tmpfs (union, local)
> >
> > [snip]
> > 
> > unmounting /dev (tmpfs)...
> > 
> > unmounting / (/dev/wd0a)...uvm_fault(0xc0398760, 0xc6984000, 1) -> 0xe
> > 
> > kernel: supervisor trap page fault, code=0
> 
> You are trying to modify the vnode for /dev/wd0a to update the access time
> info, but tmpfs has been garbage collected already when unmounting /dev.
> 
> This is the problem of forcibly unmounting a file system with device
> vnodes on it.  There was a small discussion about splitting specfs
> into two layers: one for the "host" file system (tmpfs in this case)
> and one for the actual device info (specfs).  Then you could rip the
> "host" part out without affecting the device part before the device goes
> inactive and can be safely put to rest.

I'd actually say this is a bug in deadfs or tmpfs. Since we know the
vnode's still in use, we should leave one around that doesn't cause the 
kernel to crash.

Take care,

Bill

Attachment: pgpuh6nBj25i4.pgp
Description: PGP signature



Home | Main Index | Thread Index | Old Index