Re: [PATCH] mtd: nand: do FIFO processing in nand_get_device()

[Brian Norris <computersforpeace@gmail.com>]

On Wed, Nov 25, 2015 at 06:35:43PM +0100, Sebastian Andrzej Siewior wrote:
> I have here a live lock in UBI doing
>   ensure_wear_leveling() -> wear_leveling_worker() -> ubi_eba_copy_leb()
>   MOVE_RETRY -> schedule_erase() -> ensure_wear_leveling()
> 
> on the same PEB over and over again. The reason for MOVE_RETRY is that
> the LEB-Lock owner is stucked in nand_get_device() and does not get the
> device lock. The PEB-lock owner is only scheduled on the CPU while the UBI
> thread is idle during erase or read while (again) owning the device-lock
> so the LEB-lock owner makes no progress.
> 
> To fix this live lock I ensure that there FIFO processing in
> nand_get_device(). On release the first waiter is marked as the new
> owner. If someone asks for the device and is not the waiter to which
> nand device has been handed over then it will put itself on the
> waitqueue.
> The FIFO processing was suggested by Peter Zijlstra.
> 
> As a small optimization I use add_wait_queue_exclusive() instead
> add_wait_queue() to make sure that only _one_ waiter is woken up and not
> all of them.
> 
> Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
> ---
> Would a stable be considered as reasonable?

Your threading is a bit confusing. Is this patch still needed, or did
you fix everything by fixing UBI with these?

http://lists.infradead.org/pipermail/linux-mtd/2015-November/063745.html
http://lists.infradead.org/pipermail/linux-mtd/2015-November/063744.html
http://lists.infradead.org/pipermail/linux-mtd/2015-November/063746.html

Brian