by Boris Brezillon

[permalink] [raw]

Subject: Re: [PATCH v2 06/29] mtd: Add support for reading MTD devices via the nvmem API

Hi Bartosz,

On Fri, 10 Aug 2018 10:05:03 +0200
Bartosz Golaszewski <[email protected]> wrote:

> From: Alban Bedel <[email protected]>
>
> Allow drivers that use the nvmem API to read data stored on MTD devices.
> For this the mtd devices are registered as read-only NVMEM providers.
>
> Signed-off-by: Alban Bedel <[email protected]>
> [Bartosz:
> - use the managed variant of nvmem_register(),
> - set the nvmem name]
> Signed-off-by: Bartosz Golaszewski <[email protected]>

What happened to the 2 other patches of Alban's series? I'd really like
the DT case to be handled/agreed on in the same patchset, but IIRC,
Alban and Srinivas disagreed on how this should be represented. I
hope this time we'll come to an agreement, because the MTD <-> NVMEM
glue has been floating around for quite some time...

Regards,

Boris

> ---
> drivers/mtd/Kconfig | 1 +
> drivers/mtd/mtdcore.c | 50 +++++++++++++++++++++++++++++++++++++++++
> include/linux/mtd/mtd.h | 2 ++
> 3 files changed, 53 insertions(+)
>
> diff --git a/drivers/mtd/Kconfig b/drivers/mtd/Kconfig
> index 46ab7feec6b6..f5549482d0df 100644
> --- a/drivers/mtd/Kconfig
> +++ b/drivers/mtd/Kconfig
> @@ -1,5 +1,6 @@
> menuconfig MTD
> tristate "Memory Technology Device (MTD) support"
> + imply NVMEM
> help
> Memory Technology Devices are flash, RAM and similar chips, often
> used for solid state file systems on embedded devices. This option
> diff --git a/drivers/mtd/mtdcore.c b/drivers/mtd/mtdcore.c
> index 42395df06be9..a57302eaceb5 100644
> --- a/drivers/mtd/mtdcore.c
> +++ b/drivers/mtd/mtdcore.c
> @@ -488,6 +488,49 @@ int mtd_pairing_groups(struct mtd_info *mtd)
> }
> EXPORT_SYMBOL_GPL(mtd_pairing_groups);
>
> +static int mtd_nvmem_reg_read(void *priv, unsigned int offset,
> + void *val, size_t bytes)
> +{
> + struct mtd_info *mtd = priv;
> + size_t retlen;
> + int err;
> +
> + err = mtd_read(mtd, offset, bytes, &retlen, val);
> + if (err && err != -EUCLEAN)
> + return err;
> +
> + return retlen == bytes ? 0 : -EIO;
> +}
> +
> +static int mtd_nvmem_add(struct mtd_info *mtd)
> +{
> + struct nvmem_config config = { };
> +
> + config.dev = &mtd->dev;
> + config.owner = THIS_MODULE;
> + config.name = mtd->name;
> + config.reg_read = mtd_nvmem_reg_read;
> + config.size = mtd->size;
> + config.word_size = 1;
> + config.stride = 1;
> + config.read_only = true;
> + config.root_only = true;
> + config.priv = mtd;
> +
> + mtd->nvmem = devm_nvmem_register(&mtd->dev, &config);
> + if (IS_ERR(mtd->nvmem)) {
> + /* Just ignore if there is no NVMEM support in the kernel */
> + if (PTR_ERR(mtd->nvmem) == -ENOSYS) {
> + mtd->nvmem = NULL;
> + } else {
> + dev_err(&mtd->dev, "Failed to register NVMEM device\n");
> + return PTR_ERR(mtd->nvmem);
> + }
> + }
> +
> + return 0;
> +}
> +
> static struct dentry *dfs_dir_mtd;
>
> /**
> @@ -570,6 +613,11 @@ int add_mtd_device(struct mtd_info *mtd)
> if (error)
> goto fail_added;
>
> + /* Add the nvmem provider */
> + error = mtd_nvmem_add(mtd);
> + if (error)
> + goto fail_nvmem_add;
> +
> if (!IS_ERR_OR_NULL(dfs_dir_mtd)) {
> mtd->dbg.dfs_dir = debugfs_create_dir(dev_name(&mtd->dev), dfs_dir_mtd);
> if (IS_ERR_OR_NULL(mtd->dbg.dfs_dir)) {
> @@ -595,6 +643,8 @@ int add_mtd_device(struct mtd_info *mtd)
> __module_get(THIS_MODULE);
> return 0;
>
> +fail_nvmem_add:
> + device_unregister(&mtd->dev);
> fail_added:
> of_node_put(mtd_get_of_node(mtd));
> idr_remove(&mtd_idr, i);
> diff --git a/include/linux/mtd/mtd.h b/include/linux/mtd/mtd.h
> index a86c4fa93115..8121c6582285 100644
> --- a/include/linux/mtd/mtd.h
> +++ b/include/linux/mtd/mtd.h
> @@ -25,6 +25,7 @@
> #include <linux/notifier.h>
> #include <linux/device.h>
> #include <linux/of.h>
> +#include <linux/nvmem-provider.h>
>
> #include <mtd/mtd-abi.h>
>
> @@ -339,6 +340,7 @@ struct mtd_info {
> struct device dev;
> int usecount;
> struct mtd_debug_info dbg;
> + struct nvmem_device *nvmem;
> };
>
> int mtd_ooblayout_ecc(struct mtd_info *mtd, int section,

2018-08-19 11:34:26

On Mon, 20 Aug 2018 20:20:38 +0200
Boris Brezillon <[email protected]> wrote:

> On Mon, 20 Aug 2018 11:43:34 +0100
> Srinivas Kandagatla <[email protected]> wrote:
>
> >
> > Overall am still not able to clear visualize on how MTD bindings with
> > nvmem cells would look in both partition and un-partition usecases?
> > An example DT would be nice here!!
>
> Something along those lines:

We must also have a compatible string on the nvmem-cells node to make
sure we don't clash with the old style MTD partitions, or some other
device specific binding.

>
> mtdnode {
> nvmem-cells {
compatible = "nvmem-cells";
> #address-cells = <1>;
> #size-cells = <1>;
>
> cell@0 {
> reg = <0x0 0x14>;
> };
> };
>
> partitions {
> compatible = "fixed-partitions";
> #address-cells = <1>;
> #size-cells = <1>;
>
> partition@0 {
> reg = <0x0 0x20000>;
>
> nvmem-cells {
compatible = "nvmem-cells";
> #address-cells = <1>;
> #size-cells = <1>;
>
> cell@0 {
> reg = <0x0 0x10>;
> };
> };
> };
> };
> };

Alban

Attachments:

(No filename) (836.00 B)
OpenPGP digital signature

2018-08-20 22:56:15

On Tue, 21 Aug 2018 10:50:07 +0100
Srinivas Kandagatla <[email protected]> wrote:

> Just curious...Is there a reason why we can't do it like this?:
> Is this because of issue of #address-cells and #size-cells Or mtd
> bindings always prefer subnodes?
>
> mtdnode {
> reg = <0x0123000 0x40000>;
> #address-cells = <1>;
> #size-cells = <1>;
> cell@0 {
> compatible = "nvmem-cell";
> reg = <0x0 0x14>;
> };
>
> partitions {
> compatible = "fixed-partitions";
> #address-cells = <1>;
> #size-cells = <1>;
>
> partition@0 {
> reg = <0x0 0x20000>;
> cell@0 {
> compatible = "nvmem-cell";
> reg = <0x0 0x10>;
> };
> };
> };
> };

That would work, the MTD partitions parser ignore child nodes with a
compatible string when looking for "old style" partitions, see [1].
However we still have the a potential address space clash between the
nvmem cells and the main device binding.

Alban

[1]: https://elixir.bootlin.com/linux/latest/source/drivers/mtd/ofpart.c#L28

Attachments:

(No filename) (849.00 B)
OpenPGP digital signature

2018-08-21 12:01:34

by Srinivas Kandagatla

[permalink] [raw]

Subject: Re: [PATCH v2 06/29] mtd: Add support for reading MTD devices via the nvmem API

On 21/08/18 12:39, Alban wrote:
> However we still have the a potential address space clash between the
> nvmem cells and the main device binding.
Can you elaborate?

--srini

2018-08-21 12:31:28

On Thu, Oct 4, 2018 at 4:36 PM Sowmini Varadhan
<[email protected]> wrote:
>
> Just catching up on this thread, so please excuse any unintentional
> misquotes here.
>
> > > > > David: I couldn't find a place in sparc code where any ethernet device
> > > > > would be registered, so is there a chance that nobody is using it?
> > > >
> > > > SPARC uses a true Open Firmware implementation, so it would register
> > > > drivers through the CONFIG_OF infrastructure.
>
> correct
>
> > The calls are not even that old, and clearly added intentionally for sparc,
> > see commit ba94272d08a7 ("i40e: use eth_platform_get_mac_address()")
> > which added the first one.
>
> Yes, correct again. Wouldn't PPC also end up doing the same thing?
>
> See also commit c762dff24c06 (for ixgbe) - without this fix sparc systems
> will come up with a bogus mac address and then you end up having to
> manually fix this in ugly ways.

The of_get_mac_address() portion is not controversial at all I think,
the question was whether we need the fallback to
arch_get_platform_mac_address() in any of the drivers that
call eth_platform_get_mac_address().

Arnd