Re: [BUG 2.6.26-rc1] scsi (or sysfs or vm?) oops in sr_probe()

[James Bottomley <James.Bottomley@HansenPartnership.com>]

On Sun, 2008-05-04 at 10:39 -0500, James Bottomley wrote:
> On Sun, 2008-05-04 at 10:25 -0500, James Bottomley wrote:
> > On Sun, 2008-05-04 at 07:22 -0400, Jeff Garzik wrote:
> > > See attached files for additional config detail.
> > > 
> > > Platform summary:  AMD64 2x2 (== 4 cores), Fedora 8/x86-64
> > > 
> > > Driver 'sr' needs updating - please use bus_type methods
> > > BUG: unable to handle kernel NULL pointer dereference at 00000000000005d8
> > > IP: [<ffffffff8025e10c>] get_page_from_freelist+0x43/0x573
> > > PGD 7e90f067 PUD 7edbc067 PMD 0
> > > Oops: 0000 [1] SMP
> > 
> > Is this with CONFIG_SLAB or CONFIG_SLUB?
> > 
> > I don't see it in my test systems, but they're CONFIG_SLAB
> 
> OK, should have looked at the attachments:
> 
> ...
> CONFIG_SLAB=y
> # CONFIG_SLUB is not set
> # CONFIG_SLOB is not set
> ...
> 
> Since you also have this:
> 
> CONFIG_DEBUG_KERNEL=y
> 
> Can you tell me what source line corresponds to 
> 
> :sr_mod:sr_probe+0x17b/0x4a9

Ping on this, please.  I suspect it will tell us the faulting call is
kmalloc and that it's definitely a vm issue, but we need this data as a
first step.

James