Re: [PATCH 7/7] driver-core : convert semaphore to mutex in structclass

[Dave Young]

On Sat, Jan 19, 2008 at 10:39:33AM +0100, Jarek Poplawski wrote:
> Dave Young wrote, On 01/18/2008 10:07 AM:
> 
> > On Jan 18, 2008 4:23 PM, Jarek Poplawski <jarkao2@xxxxxxxxx> wrote:
> 
> >> On Fri, Jan 18, 2008 at 03:48:02PM +0800, Dave Young wrote:
> 
> ...
> 
> >>> 1) Using CLASS_NORMAL/CLASS_PARENT/CLASS_CHILD will be enough.
> >>> or
> >>> 2) Simply add SINGLE_LEVEL_NESTING in class_device_add and other
> >>> class_device functions because it is the only possible nest-lock place
> >>> as I know.
> 
> 
> Dave, after looking a bit at this it seems you could be "mostly" right
> with this 2). Maybe I've missed something (I didn't verify this yet), but
> it looks like +1 level (SINGLE_LEVEL_NESTING) could be needed in:
> class_device_add() (as you did), but probably also class_device_del() and
> class_device_destroy().

Yes, I think so too.

> 
> ...But, there seems to be "little" problem, if there is used this recursion
> with: class_intf->add()/remove() in class_device_add()/del()?! Then Kay
> is right about possibility of deeper nesting. If this path is really used,
> and any of these class_device_* functions with locking are called, then
> this patch couldn't work like this. So, there is a question: how deep
> nesting is currently used here?

Currently I couldn't find such use in kernel source. IMO, drivers would not use it like this in the future because class_device will going away soon.

> 
> Regards,
> Jarek P.
--
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/