Re: [PATCH v2] mtd: spi-nor: add debugfs nodes for querying the flash name and id

[Zhuohao Lee <zhuohao@xxxxxxxxxxxx>]

On Tue, May 7, 2019 at 11:10 PM Boris Brezillon
<boris.brezillon@xxxxxxxxxxxxx> wrote:
>
> On Tue, 7 May 2019 23:06:32 +0800
> Zhuohao Lee <zhuohao@xxxxxxxxxxxx> wrote:
>
>
> > > > > > @@ -231,8 +231,11 @@ static int m25p_probe(struct spi_mem *spimem)
> > > > >
> > > > > Can we add this to function that is generic to all spi-nor devices,
> > > > > instead of making this specific to m25p?
> > > > I can't find a better way to insert the spi_nor_debugfs_create()
> > > > inside spi_nor.c.
> > > > Another way is adding spi_nor_debugfs_create() to all of the caller.
> > > > What do you think? Any other suggestion?
> > >
> > > That, or maybe create a new spi_nor_device_register that does both
> > > mtd_device_register and that spi_nor_debugfs_create call?
> > Thanks for suggestion. I feel that putting the mtd_device_register
> > (high level api) inside the spi-nor (low level api)
> > isn't perfect. This also will limit the caller to call this api to
> > register mtd device with debugfs and lost the flexibility.
> > I'll keep the original idea that adding spi_nor_debugfs_create() to
> > all of the caller.
>
> Why don't you move that to the MTD layer? If you add partname/partid
> fields to mtd_info you'll have everything you need to make that generic.
> It's then up to the upper layer to fill those fields before calling
> mtd_device_register().
>
Ah... i took the wrong way. I removed the partname/partid from mtd.h.
So, i can't use it inside the mtd_core.c. Instead, i created
spi_nor_debugfs_create()
for creating debugfs.
i'll submit a patch to add back the partname/partid to mtd.h
>
> > > > >
> > > > > > +               return;
> > > > > > +       }
> > > > > > +       debugfs_create_file("flashid", S_IRUSR, root, nor,
> > > > > > +                       &flashid_dbg_fops);
> > > > > > +       debugfs_create_file("flashname", S_IRUSR, root, nor,
> > > > > > +                       &flashname_dbg_fops);
>
> I thought we agreed on partname/partid. Any reason for switching back
> to flashname/flashid?
Sorry, per reply above, i took the wrong approach. So, i used old name
because i put the debugfs stuff in spi-nor.c.

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/