Re: v2.6.26-rc7: BUG: unable to handle kernel NULL pointer dereference

["Vegard Nossum" <vegard.nossum@gmail.com>]

On Tue, Jun 24, 2008 at 3:36 AM, Rusty Russell <rusty@rustcorp.com.au> wrote:
> Vegard's analysis is flawed: just because cpu is offline, it still must be <
> nr_cpu_ids, which is based on possible cpus.  Unless something crazy is
> happening, but a quick grep doesn't reveal anyone manipulating nr_cpu_ids.

Hm, you are right and I was wrong. I'm sorry, it just seemed too
obvious to be any other way, and I made some assumptions about
nr_cpu_ids. (IIRC, nr_node_ids changes dynamically as nodes are
added/removed, so I assumed it was the same for CPUs.)

This doesn't change the fact that get_cpu_sysdev(cpu) returns NULL,
however. This variable, the per-cpu cpu_sys_device, is only ever
changed in two places, register_cpu() and unregister_cpu(); in
register_cpu(), it is set to

                per_cpu(cpu_sys_devices, num) = &cpu->sysdev;,

and in unregister_cpu(), it is set to

        per_cpu(cpu_sys_devices, logical_cpu) = NULL;.

So it seems *likely* that register_cpu() was never called (after the
previous unregister_cpu(), which we know happened successfully).

register_cpu() is called from arch_register_cpu(), which is called
from toplogy_init() and acpi_processor_hotadd_init(). Now, the
topology_init() call-chain is uninteresting, since it only happens at
boot. The question is whether acpi_processor_hotadd_init() will be
called if the arch-specific __cpu_up() fails...

But I am not able to follow that code.

Thanks for looking at this.


Vegard

PS: I'll withdraw the statement that this is probably a regression. It
seems more likely that nobody ever hit the "cpu failed to init" case
before.

-- 
"The animistic metaphor of the bug that maliciously sneaked in while
the programmer was not looking is intellectually dishonest as it
disguises that the error is the programmer's own creation."
	-- E. W. Dijkstra, EWD1036