mtdpart: propagate _get/put_device()

If the master device has callbacks for _get/put_device()
and this MTD has slaves a get_mtd_device() call on paritions
will never issue the registered callbacks.
Fix this by propergating _get/put_device() down.

Signed-off-by: Richard Weinberger <richard@nod.at>
---
Hi!

I've found this issue while heavily reworking nandsim.
With my changes applied you can add/remove MTD's during
runtime. I figured that nandsim's refcounting goes nuts.
After tracking down the root cause I figured that
_get/put_device() has no effect on MTD slaves.

Thanks,
//richard

---
 drivers/mtd/mtdpart.c | 19 +++++++++++++++++++
 1 file changed, 19 insertions(+)

On Wed, 29 Jun 2016 22:14:23 +0200
Richard Weinberger <richard@nod.at> wrote:

> If the master device has callbacks for _get/put_device()
> and this MTD has slaves a get_mtd_device() call on paritions
> will never issue the registered callbacks.
> Fix this by propergating _get/put_device() down.
> 
> Signed-off-by: Richard Weinberger <richard@nod.at>

Nit below, otherwise it looks good to me:

Reviewed-by: Boris Brezillon <boris.brezillon@free-electrons.com>

> ---
> Hi!
> 
> I've found this issue while heavily reworking nandsim.
> With my changes applied you can add/remove MTD's during
> runtime. I figured that nandsim's refcounting goes nuts.
> After tracking down the root cause I figured that
> _get/put_device() has no effect on MTD slaves.
> 
> Thanks,
> //richard
> 
> ---
>  drivers/mtd/mtdpart.c | 19 +++++++++++++++++++
>  1 file changed, 19 insertions(+)
> 
> diff --git a/drivers/mtd/mtdpart.c b/drivers/mtd/mtdpart.c
> index 1f13e32..938dbeb 100644
> --- a/drivers/mtd/mtdpart.c
> +++ b/drivers/mtd/mtdpart.c
> @@ -317,6 +317,19 @@ static int part_block_markbad(struct mtd_info *mtd, loff_t ofs)
>  	return res;
>  }
>  
> +static int part_get_device(struct mtd_info *mtd)
> +{
> +	struct mtd_part *part = mtd_to_part(mtd);
> +	return part->master->_get_device(part->master);
> +}
> +
> +static void part_put_device(struct mtd_info *mtd)
> +{
> +	struct mtd_part *part = mtd_to_part(mtd);
> +	part->master->_put_device(part->master);
> +}
> +
> +

Extra blank line here.

>  static int part_ooblayout_ecc(struct mtd_info *mtd, int section,
>  			      struct mtd_oob_region *oobregion)
>  {
> @@ -463,6 +476,12 @@ static struct mtd_part *allocate_partition(struct mtd_info *master,
>  		slave->mtd._block_isbad = part_block_isbad;
>  	if (master->_block_markbad)
>  		slave->mtd._block_markbad = part_block_markbad;
> +
> +	if (master->_get_device)
> +		slave->mtd._get_device = part_get_device;
> +	if (master->_put_device)
> +		slave->mtd._put_device = part_put_device;
> +
>  	slave->mtd._erase = part_erase;
>  	slave->master = master;
>  	slave->offset = part->offset;

Boris,

Am 04.07.2016 um 11:35 schrieb Boris Brezillon:
> On Wed, 29 Jun 2016 22:14:23 +0200
> Richard Weinberger <richard@nod.at> wrote:
> 
>> If the master device has callbacks for _get/put_device()
>> and this MTD has slaves a get_mtd_device() call on paritions
>> will never issue the registered callbacks.
>> Fix this by propergating _get/put_device() down.
>>
>> Signed-off-by: Richard Weinberger <richard@nod.at>
> 
> Nit below, otherwise it looks good to me:
> 
> Reviewed-by: Boris Brezillon <boris.brezillon@free-electrons.com>

Will send a v2 without the extra blank line and the typo
in the commit message. :)

Thanks,
//richard