Re: module_put_and_exit() and free_module()

[Rusty Russell <rusty@rustcorp.com.au>]

Aleksa Sarai <cyphar@cyphar.com> writes:
>>From my understanding, module_put_and_exit() can be used inside a
> module to (from within the module) kill itself. However, it doesn't
> seem to properly free the modules references (and internal
> bookkeeping) since module_put_and_exit() doesn't call free_module().
> And attempting to remove the module after loading it and it running
> module_put_and_exit() causes any attempt to remove the module to fail
> with EBUSY.

It's not a general mechanism!  It's for kernel threads which want to
decrement module use counts as they exit: if they did this in the module
there would be a moment where they are still running but the module
could be unloaded.

> Am I missing something here? Is the purpose of module_put_and_exit()
> different, or does module_put() cause the module references to get
> reaped later? If that's the case, why do you get EBUSY when trying to
> remove the module (surely you should get an ENOENT)? Is it even safe
> to attempt to remove a module from within itself?

module_put() simply controls the reference count.  Module removal only
succeeds if the reference count is 0.

I would guess that you called module_put_and_exit() without a thread,
and oopsed.

Cheers,
Rusty.