Re: [PATCH v1] mtd: core: add sysfs nodes for querying the flash name and id

[Miquel Raynal <miquel.raynal@bootlin.com>]

Hi Zhuohao,

Zhuohao Lee <zhuohao@chromium.org> wrote on Tue, 2 Apr 2019 21:03:28
+0800:

> Boris, Thanks for your comment.
> But I can't find the related stuff (or i got the wrong way). Is that
> feature merged?
> Could you please point out the commit? I can make the change base on it.

I think Boris just told you that the debugfs MTD directory is suitable
to receive an information coming from different type of devices like
spi-nor, nor, nand, etc.

The feature of having the device name there does not exist yet, you
have to do it. For NAND I already wrote something like that which has
not been merged but could be adapted to export the data in debugfs
instead of sysfs, see:

Raw NAND infrastructure to retrieve the ID:
http://patchwork.ozlabs.org/patch/837901/

Adding a sysfs entry (which was already rejected at this time):
http://patchwork.ozlabs.org/patch/837903/

Thanks,
Miquèl

> 
> On Tue, Apr 2, 2019 at 8:01 PM Boris Brezillon
> <boris.brezillon@collabora.com> wrote:
> >
> > On Tue, 2 Apr 2019 19:06:37 +0800
> > Zhuohao Lee <zhuohao@chromium.org> wrote:
> >  
> > > On Tue, Apr 2, 2019 at 4:26 PM Vignesh Raghavendra <vigneshr@ti.com> wrote:  
> > > >
> > > >
> > > >
> > > > On 02/04/19 1:26 PM, Boris Brezillon wrote:  
> > > > > On Tue, 2 Apr 2019 15:39:54 +0800
> > > > > Zhuohao Lee <zhuohao@chromium.org> wrote:
> > > > >  
> > > > >> Thanks Boris for the comment. Please take a look the reply at below.
> > > > >>
> > > > >> On Mon, Apr 1, 2019 at 5:27 PM Boris Brezillon
> > > > >> <boris.brezillon@collabora.com> wrote:  
> > > > >>>
> > > > >>> On Thu, 28 Mar 2019 12:59:10 +0800
> > > > >>> Zhuohao Lee <zhuohao@chromium.org> wrote:
> > > > >>>  
> > > > >>>> Currently, we don't have sysfs nodes for querying the underlying flash
> > > > >>>> name and flash id. This information is important especially when we
> > > > >>>> want to know the flash detail of the defective system. In order to
> > > > >>>> support the query, we add two pointers (*flashname, *id) into the
> > > > >>>> mtd_info structure and create two sysfs nodes (flashname, id). This
> > > > >>>> patch is modified based on the SPI-NOR flash system as we only have
> > > > >>>> that system now. But the idea should be applied to the other flash
> > > > >>>> driver like NAND flash.
> > > > >>>>
> > > > >>>> The output of new sysfs nodes on my device are:
> > > > >>>> cat /sys/devices/platform/soc/11010000.spi/spi_master/spi1/spi1.0/mtd/mtd0/flashname
> > > > >>>> w25q64dw
> > > > >>>> cat /sys/devices/platform/soc/11010000.spi/spi_master/spi1/spi1.0/mtd/mtd0/id
> > > > >>>> ef6017  
> > > > >>>
> > > > >>> I'm not sure I like the idea of exposing this kind of info through
> > > > >>> sysfs as it then makes part of the ABI. Did you consider exposing that
> > > > >>> through debugfs?  
> > > > >>
> > > > >> Yes, i did consider the debugfs. I think the debugfs is depended on
> > > > >> CONFIG_DEBUG_FS.
> > > > >> If removing that config, the partname and partid will be lost. So, i
> > > > >> proposed to use
> > > > >> sysfs.  
> > > > >
> > > > > Then just enable debugfs if you need this information :P.  
> > > My original intention is adding the new sysfs nodes (i.e partname and
> > > partid) into the common place like mtdcore.c
> > > so that the userspace program can just read the common sysfs nodes.
> > > So far, what i can contribute is for the spi-nor but if the other
> > > flash drivers, like nand-flash, can also support
> > > the partname and partid query, then, we can have common sysfs nodes to
> > > query the underlying device info.
> > > Compare to the debugfs, i think different drivers may have different
> > > name for the partname and partid.  
> >
> > We've recently unified how MTD related stuff are exposed through
> > debugfs and you can now have generic MTD fields exposed there.  
> 
> ______________________________________________________
> Linux MTD discussion mailing list
> http://lists.infradead.org/mailman/listinfo/linux-mtd/


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