Re: 2.6.15-rc5-mm2: ehci_hcd crashes on load sometimes

[Andrew Morton <akpm@osdl.org>]

"Rafael J. Wysocki" <rjw@sisk.pl> wrote:
>
> On Monday, 12 December 2005 22:09, Andrew Morton wrote:
> > "Rafael J. Wysocki" <rjw@sisk.pl> wrote:
> > >
> > >  > It's best to actually send a copy of line 620 - kernels vary a lot, and
> > >  > many developers won't have that particualr -mm tree handy.
> > >  > 
> > >  > The way I normally do this is to do `gdb vmlinux' and then `l
> > >  > *0xffffffff880ad9d0'.
> > > 
> > >  Does it work for modules too?
> > 
> > Ah.  There are certainly ways of doing this - see the kgdb documentation. 
> > Or you can work out the module load address, gdb the module and do the
> > appropriate arithmetic I guess.
> > 
> > Generally I just statically link anything which I want to play with.
> 
> Still, the oops is from a module.  I could link it statically for debugging,
> but then the address would be different to the one in the oops.
> 
> Anyway, please tell me if my reasoning was correct: I thought I couldn't
> figure it out based on the absolute address, but I could use the
> displacements.  Namely, it followed from the oops that the problem
> occured at the address {:ehci_hcd:ehci_irq+224}, which is at the
> offset 224 wrt ehci_irq, so I did:
> 
> gdb drivers/usb/host/ehci-hcd.o
> 
> In gdb I did:
> 
> info line ehci_irq
> 
> and it told me the address the line started at, so I added 224 to it and
> got the line 620.

That's a good way of doing it.