Re: [PATCH] ohci1394: dma_pool_destroy while in_atomic() && irqs_disabled()

[Andrew Morton <akpm@osdl.org>]

Parag Warudkar <kernel-stuff@comcast.net> wrote:
>
> Andrew Morton wrote:
> 
> ...
> >- We'll need a flush_workqueue() in the teardown function for that data
> >  structure to ensure that any pending callbacks have completed before we
> >  free the storage.
> >  
> >
> By saying flush_workqueue did you intend to suggest using separate work 
> queue for ohci1394?

No.  Using keventd and flush_scheduled_work() should be OK in this
application.  rmmod isn't very common.

> >  Care needs to be taken to ensure that the work_struct is suitably
> >  initialised so that the flush_workqueue() will work OK even if the
> >  callback has never been scheduled.
> >  
> >
> Didn't understand this one  - Is this about properly NULL'ing elements 
> in work queue so flush_workqueue doesn't touch them? Can you elaborate 
> please?

Well, it's just that the shutdown code needs to run flush_scheduled_work()
or flush_workqueue() on a work_struct which may or may not have been used. 
So we need to ensure that it has sane contents.  Just run INIT_WORK() on it
in the setup code.