[lkml]   [2001]   [Sep]   [23]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
SubjectRe: PATCH: linux-2.4.10-pre14/drivers/sound/maestro.c ignored pci_module_init results
>> = Adam Richter
> = Marcus Meissner

>> The initialization routine in
>> linux-2.4.10-pre14/drivers/sound/maestro.c ignores the return value
>> from pci_module_init, and allows module initialization to succeed
>> even if pci_module_init failed. pci_module_init fails and unloads
>> the driver if the caller is a module and there is no matching hardware.
>> Because maestro.c ignored this failure, loading maestro.o on a system
>> where the corresponding alsa driver was already loaded or on a system
>> without matchin hardware would result in a kernel null pointer dereference
>> in pci_unregister_driver when the module is unloaded or when one
>> attempts to reboot the system (i.e., when the module attempt to
>> unregister a PCI driver that is not registered).

>Why and where does it Oops?

On a machine that has no maestro hardware or that already has
the alsa drivers bound to the maestro PCI hardware, either of the
following will cause a null pointer dereference when
pci_unregister_driver tries to unregister a driver that is not registered:

modprobe maestro
rmmod maestro

(as cleanup_maestro incorrectly calls pci_unregister_driver
on a PCI driver that is not loaded.)


modprobe meastro

(as maestro_notifier incorrectly calls pci_unregister_driver
on a PCI driver that is not loaded.)

I experimentally verified both of these on a machine that already
had the maestro ALSA drivers loaded. I assume the null pointer dereference
would be from the list_del(&drv->node) in pci_unregister_driver, since
list_del calls __list_del, which assumes that drv->node->{prev,next} are
not NULL, but they would have been set to NULL by the first
pci_remove_module, which pci_module_init called when the driver failed
to bind to anything, but which the maestro driver incorrectly ignored.

>The code for pci_unregister_driver in
>drivers/pci/pci.c looks correct and should not Oops.
>The reboot notifier might be problematic, but I have not checked it.

There is nothing wrong with pci_unregister_driver. The bug
is where init_maestro in maestro.c ignored the results of
pci_module_init, which is what my patch fixes.

Adam J. Richter __ ______________ 4880 Stevens Creek Blvd, Suite 104 \ / San Jose, California 95129-1034
+1 408 261-6630 | g g d r a s i l United States of America
fax +1 408 261-6631 "Free Software For The Rest Of Us."
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to
More majordomo info at
Please read the FAQ at

 \ /
  Last update: 2005-03-22 13:03    [W:0.029 / U:0.932 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site