Re: [PATCH] s390 (8/10): zfcp fixes

[Heiko Carstens]

Hi Greg,

>> I'll ask because the zfcp patches are still pending and I want to get this
>> issue resolved before the next try to get them integrated.
>I think you need to talk to the scsi people, as kfree() should _never_
>need to be set as the release function. There's something just wrong
>with the design if that is necessary.

Actually this issue is not SCSI related, but a generic problem that one
can run into when fiddling around with modules / module unloading.

This was already discussed elsewhere. Please have a look here:

http://lwn.net/Articles/67421/
(paragraph "Module unloading in a reference counted world")

and here for the whole thread:

http://www.ussg.iu.edu/hypermail/linux/kernel/0401.2/1832.html
(just saw that you participated there :)

I would be more than happy to have a nice release function for the zfcp
generated objects, but I don't think this is currently possible without
having the potential problem to run into an Oops after module unloading.

Heiko

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/