2023-03-06 12:58:35

by Michael Walle

[permalink] [raw]
Subject: [PATCH 1/4] mtd: core: provide unique name for nvmem device, take two

Commit c048b60d39e1 ("mtd: core: provide unique name for nvmem device")
tries to give the nvmem device a unique name, but fails badly if the mtd
device doesn't have a "struct device" associated with it, i.e. if
CONFIG_MTD_PARTITIONED_MASTER is not set. This will result in the name
"(null)-user-otp", which is not unique. It seems the best we can do is
to use the compatible name together with a unique identifier added by
the nvmem subsystem by using NVMEM_DEVID_AUTO.

Fixes: c048b60d39e1 ("mtd: core: provide unique name for nvmem device")
Cc: [email protected]
Signed-off-by: Michael Walle <[email protected]>
---
drivers/mtd/mtdcore.c | 4 +---
1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/drivers/mtd/mtdcore.c b/drivers/mtd/mtdcore.c
index 0feacb9fbdac..3fe2825a85a1 100644
--- a/drivers/mtd/mtdcore.c
+++ b/drivers/mtd/mtdcore.c
@@ -888,8 +888,7 @@ static struct nvmem_device *mtd_otp_nvmem_register(struct mtd_info *mtd,

/* OTP nvmem will be registered on the physical device */
config.dev = mtd->dev.parent;
- config.name = kasprintf(GFP_KERNEL, "%s-%s", dev_name(&mtd->dev), compatible);
- config.id = NVMEM_DEVID_NONE;
+ config.name = compatible;
config.owner = THIS_MODULE;
config.type = NVMEM_TYPE_OTP;
config.root_only = true;
@@ -905,7 +904,6 @@ static struct nvmem_device *mtd_otp_nvmem_register(struct mtd_info *mtd,
nvmem = NULL;

of_node_put(np);
- kfree(config.name);

return nvmem;
}
--
2.30.2



2023-03-06 12:58:37

by Michael Walle

[permalink] [raw]
Subject: [PATCH 3/4] mtd: core: fix error path for nvmem provider

If mtd_otp_nvmem_add() fails, the partitions won't be removed
because there is simply no call to del_mtd_partitions().
Unfortunately, add_mtd_partitions() will print all partitions to
the kernel console. If mtd_otp_nvmem_add() returns -EPROBE_DEFER
this would print the partitions multiple times to the kernel
console. Instead move mtd_otp_nvmem_add() to the beginning of the
function.

Fixes: 4b361cfa8624 ("mtd: core: add OTP nvmem provider support")
Cc: [email protected]
Signed-off-by: Michael Walle <[email protected]>
---
drivers/mtd/mtdcore.c | 13 ++++++++++---
1 file changed, 10 insertions(+), 3 deletions(-)

diff --git a/drivers/mtd/mtdcore.c b/drivers/mtd/mtdcore.c
index 4a19e69fb77b..d6675bf03557 100644
--- a/drivers/mtd/mtdcore.c
+++ b/drivers/mtd/mtdcore.c
@@ -1022,10 +1022,14 @@ int mtd_device_parse_register(struct mtd_info *mtd, const char * const *types,

mtd_set_dev_defaults(mtd);

+ ret = mtd_otp_nvmem_add(mtd);
+ if (ret)
+ goto out;
+
if (IS_ENABLED(CONFIG_MTD_PARTITIONED_MASTER)) {
ret = add_mtd_device(mtd);
if (ret)
- return ret;
+ goto out;
}

/* Prefer parsed partitions over driver-provided fallback */
@@ -1060,9 +1064,12 @@ int mtd_device_parse_register(struct mtd_info *mtd, const char * const *types,
register_reboot_notifier(&mtd->reboot_notifier);
}

- ret = mtd_otp_nvmem_add(mtd);
-
out:
+ if (ret) {
+ nvmem_unregister(mtd->otp_user_nvmem);
+ nvmem_unregister(mtd->otp_factory_nvmem);
+ }
+
if (ret && device_is_registered(&mtd->dev))
del_mtd_device(mtd);

--
2.30.2


2023-03-06 12:58:40

by Michael Walle

[permalink] [raw]
Subject: [PATCH 2/4] mtd: core: fix nvmem error reporting

The master MTD will only have an associated device if
CONFIG_MTD_PARTITIONED_MASTER is set, thus we cannot use dev_err() on
mtd->dev. Instead use the parent device which is the physical flash
memory.

Fixes: 4b361cfa8624 ("mtd: core: add OTP nvmem provider support")
Cc: [email protected]
Signed-off-by: Michael Walle <[email protected]>
---
drivers/mtd/mtdcore.c | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/drivers/mtd/mtdcore.c b/drivers/mtd/mtdcore.c
index 3fe2825a85a1..4a19e69fb77b 100644
--- a/drivers/mtd/mtdcore.c
+++ b/drivers/mtd/mtdcore.c
@@ -938,6 +938,7 @@ static int mtd_nvmem_fact_otp_reg_read(void *priv, unsigned int offset,

static int mtd_otp_nvmem_add(struct mtd_info *mtd)
{
+ struct device *dev = mtd->dev.parent;
struct nvmem_device *nvmem;
ssize_t size;
int err;
@@ -951,7 +952,7 @@ static int mtd_otp_nvmem_add(struct mtd_info *mtd)
nvmem = mtd_otp_nvmem_register(mtd, "user-otp", size,
mtd_nvmem_user_otp_reg_read);
if (IS_ERR(nvmem)) {
- dev_err(&mtd->dev, "Failed to register OTP NVMEM device\n");
+ dev_err(dev, "Failed to register OTP NVMEM device\n");
return PTR_ERR(nvmem);
}
mtd->otp_user_nvmem = nvmem;
@@ -969,7 +970,7 @@ static int mtd_otp_nvmem_add(struct mtd_info *mtd)
nvmem = mtd_otp_nvmem_register(mtd, "factory-otp", size,
mtd_nvmem_fact_otp_reg_read);
if (IS_ERR(nvmem)) {
- dev_err(&mtd->dev, "Failed to register OTP NVMEM device\n");
+ dev_err(dev, "Failed to register OTP NVMEM device\n");
err = PTR_ERR(nvmem);
goto err;
}
--
2.30.2


2023-03-07 19:36:53

by Miquel Raynal

[permalink] [raw]
Subject: Re: [PATCH 3/4] mtd: core: fix error path for nvmem provider

On Mon, 2023-03-06 at 12:58:04 UTC, Michael Walle wrote:
> If mtd_otp_nvmem_add() fails, the partitions won't be removed
> because there is simply no call to del_mtd_partitions().
> Unfortunately, add_mtd_partitions() will print all partitions to
> the kernel console. If mtd_otp_nvmem_add() returns -EPROBE_DEFER
> this would print the partitions multiple times to the kernel
> console. Instead move mtd_otp_nvmem_add() to the beginning of the
> function.
>
> Fixes: 4b361cfa8624 ("mtd: core: add OTP nvmem provider support")
> Cc: [email protected]
> Signed-off-by: Michael Walle <[email protected]>

Applied to https://git.kernel.org/pub/scm/linux/kernel/git/mtd/linux.git mtd/next, thanks.

Miquel

2023-03-07 19:36:58

by Miquel Raynal

[permalink] [raw]
Subject: Re: [PATCH 2/4] mtd: core: fix nvmem error reporting

On Mon, 2023-03-06 at 12:58:03 UTC, Michael Walle wrote:
> The master MTD will only have an associated device if
> CONFIG_MTD_PARTITIONED_MASTER is set, thus we cannot use dev_err() on
> mtd->dev. Instead use the parent device which is the physical flash
> memory.
>
> Fixes: 4b361cfa8624 ("mtd: core: add OTP nvmem provider support")
> Cc: [email protected]
> Signed-off-by: Michael Walle <[email protected]>

Applied to https://git.kernel.org/pub/scm/linux/kernel/git/mtd/linux.git mtd/next, thanks.

Miquel

2023-03-07 19:37:03

by Miquel Raynal

[permalink] [raw]
Subject: Re: [PATCH 1/4] mtd: core: provide unique name for nvmem device, take two

On Mon, 2023-03-06 at 12:58:02 UTC, Michael Walle wrote:
> Commit c048b60d39e1 ("mtd: core: provide unique name for nvmem device")
> tries to give the nvmem device a unique name, but fails badly if the mtd
> device doesn't have a "struct device" associated with it, i.e. if
> CONFIG_MTD_PARTITIONED_MASTER is not set. This will result in the name
> "(null)-user-otp", which is not unique. It seems the best we can do is
> to use the compatible name together with a unique identifier added by
> the nvmem subsystem by using NVMEM_DEVID_AUTO.
>
> Fixes: c048b60d39e1 ("mtd: core: provide unique name for nvmem device")
> Cc: [email protected]
> Signed-off-by: Michael Walle <[email protected]>

Applied to https://git.kernel.org/pub/scm/linux/kernel/git/mtd/linux.git mtd/next, thanks.

Miquel

2023-03-07 20:50:03

by Michael Walle

[permalink] [raw]
Subject: Re: [PATCH 1/4] mtd: core: provide unique name for nvmem device, take two

Hi Miquel,

Am 2023-03-06 13:58, schrieb Michael Walle:
> Commit c048b60d39e1 ("mtd: core: provide unique name for nvmem device")
> tries to give the nvmem device a unique name, but fails badly if the
> mtd
> device doesn't have a "struct device" associated with it, i.e. if
> CONFIG_MTD_PARTITIONED_MASTER is not set. This will result in the name
> "(null)-user-otp", which is not unique. It seems the best we can do is
> to use the compatible name together with a unique identifier added by
> the nvmem subsystem by using NVMEM_DEVID_AUTO.
>
> Fixes: c048b60d39e1 ("mtd: core: provide unique name for nvmem device")
> Cc: [email protected]
> Signed-off-by: Michael Walle <[email protected]>
> ---
> drivers/mtd/mtdcore.c | 4 +---
> 1 file changed, 1 insertion(+), 3 deletions(-)
>
> diff --git a/drivers/mtd/mtdcore.c b/drivers/mtd/mtdcore.c
> index 0feacb9fbdac..3fe2825a85a1 100644
> --- a/drivers/mtd/mtdcore.c
> +++ b/drivers/mtd/mtdcore.c
> @@ -888,8 +888,7 @@ static struct nvmem_device
> *mtd_otp_nvmem_register(struct mtd_info *mtd,
>
> /* OTP nvmem will be registered on the physical device */
> config.dev = mtd->dev.parent;
> - config.name = kasprintf(GFP_KERNEL, "%s-%s", dev_name(&mtd->dev),
> compatible);
> - config.id = NVMEM_DEVID_NONE;
> + config.name = compatible;

Damn.. although the commit message says it using NVMEM_DEVID_AUTO,
this is missing here. :/

Since you already picked it, should I send a follow up patch or can
you just drop this patch and I'll resend it?

-michael

> config.owner = THIS_MODULE;
> config.type = NVMEM_TYPE_OTP;
> config.root_only = true;
> @@ -905,7 +904,6 @@ static struct nvmem_device
> *mtd_otp_nvmem_register(struct mtd_info *mtd,
> nvmem = NULL;
>
> of_node_put(np);
> - kfree(config.name);
>
> return nvmem;
> }