Re: [PATCH 01/27] mtd: nand: introduce function to fix a common bug in most nand-drivers not showing a device in sysfs

[Frans Klaver <fransklaver@gmail.com>]

On Wed, Nov 5, 2014 at 10:34 AM, Brian Norris
<computersforpeace@gmail.com> wrote:
> On Sun, Nov 02, 2014 at 10:03:53PM +0100, Frans Klaver wrote:
>> On Wed, May 28, 2014 at 01:43:44AM -0700, Brian Norris wrote:
>> > And in fact, if any drivers are missing mtd->name, perhaps it's best to
>> > just modify the MTD registration to give them a default:
>> >
>> >     if (!mtd->name)
>> >             mtd->name = dev_name(&pdev->dev);
>> >
>>
>> ...
>>
>> > How about we rethink the "helper" approach, and instead just do
>> > validation in the core code? This would cover most of the important
>> > parts of your helper, I think:
>> >
>> > diff --git a/drivers/mtd/mtdcore.c b/drivers/mtd/mtdcore.c
>> > index d201feeb3ca6..39ba5812a5a3 100644
>> > --- a/drivers/mtd/mtdcore.c
>> > +++ b/drivers/mtd/mtdcore.c
>> > @@ -397,6 +397,11 @@ int add_mtd_device(struct mtd_info *mtd)
>> >     if (device_register(&mtd->dev) != 0)
>> >             goto fail_added;
>> >
>> > +   if (mtd->dev.parent)
>> > +           mtd->owner = mtd->dev.parent->driver->owner;
>> > +   else
>> > +           WARN_ON(1);
>> > +
>>
>> So I've picked this up now. I do largely agree with the suggested
>> approach where the validation and default settings are done in the core
>> code. There is a problem with this, though. There are MTD devices that
>> call mtd_device_parse_register() in the _init() function (such as the
>> maps drivers). These don't have a device ready to be used as parent, and
>> they would always be throwing this warning.
>
> Yeah, I came across the same thing. I think gluebi is another example.
>
>> So either not having a parent device is bad, or it isn't. The comment
>> suggests it is, the existing code suggests it isn't. So we'll need to
>> make a decision about who's right.
>
> I think not having a parent is not really bad. It's helpful for tracking
> the device hierarchy in sysfs, but it's not strictly necessary. So we
> should probably not do anything drastic like WARN_ON() yet.

OK. I'll probably add some sane defaults to add_mtd_device() and have
the mtd drivers make use of that fact. Since this isn't really
critical, I guess fixing the sysfs entry won't need back-porting.


>> >     if (MTD_DEVT(i))
>> >             device_create(&mtd_class, mtd->dev.parent,
>> >                           MTD_DEVT(i) + 1,
>> > diff --git a/drivers/mtd/mtdpart.c b/drivers/mtd/mtdpart.c
>> > index 1ca9aec141ff..9869bbef50cf 100644
>> > --- a/drivers/mtd/mtdpart.c
>> > +++ b/drivers/mtd/mtdpart.c
>> > @@ -370,7 +370,6 @@ static struct mtd_part *allocate_partition(struct mtd_info *master,
>> >     slave->mtd.subpage_sft = master->subpage_sft;
>> >
>> >     slave->mtd.name = name;
>> > -   slave->mtd.owner = master->owner;
>>
>> What would be the purpose of removing this line? Owner is already set?
>> Can we rely on that?
>
> I'm not completely sure why I wrote that, but I think the only call site
> for alloc_partition() is in mtd_add_partition(), which calls
> add_mtd_device().

Alright, I'll have a brief look into that again, then.

Thanks,
Frans