[PATCH 3/6] UBI: Fastmap: Ensure that all fastmap work is done upon WL shutdown

[Artem Bityutskiy]

On Mon, 2014-11-24 at 14:20 +0100, Richard Weinberger wrote:
> ...otherwise the deferred work might run after datastructures
> got freed and corrupt memory.
> 
> Signed-off-by: Richard Weinberger <richard at nod.at>
> ---
>  drivers/mtd/ubi/wl.c | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/drivers/mtd/ubi/wl.c b/drivers/mtd/ubi/wl.c
> index 7f135df..cb2e571 100644
> --- a/drivers/mtd/ubi/wl.c
> +++ b/drivers/mtd/ubi/wl.c
> @@ -2041,6 +2041,9 @@ static void protection_queue_destroy(struct ubi_device *ubi)
>  void ubi_wl_close(struct ubi_device *ubi)
>  {
>  	dbg_wl("close the WL sub-system");
> +#ifdef CONFIG_MTD_UBI_FASTMAP
> +	flush_work(&ubi->fm_work);
> +#endif

If you are using the work infrastructure implemented in wl.c, then
fastmap work should be no different to any other work. And we do flush
all works in 'shutdown_work()'. The fastmap work should be flushed there
too.

I think we discussed this already - there should be one single queue of
works, managed by the same set of functions, all flushed in the same
place, one-by-one...

Obviously, there is some misunderstanding. This looks like lack of
separation and misuse of layering. I am missing explanations why I am
wrong...