[PATCH] mtd: leave the device type unset when naming fails

[PATCH] mtd: leave the device type unset when naming fails

[James Hilliard]

add_mtd_device() uses dev.type to reject a second registration of the
same MTD. It sets that field before dev_set_name(), but the failure path
for the name allocation leaves it set even though no device was
registered. Retrying registration then warns and fails with -EEXIST.

Set the type only after naming succeeds. The later registration error
paths already clear it before releasing the device.

Fixes: fa06052d637b ("mtd: Convert to dynamically allocated bdi infrastructure")
Cc: stable@vger.kernel.org
Signed-off-by: James Hilliard <james.hilliard1@gmail.com>
---
 drivers/mtd/mtdcore.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/mtd/mtdcore.c b/drivers/mtd/mtdcore.c
index 491a27d0315f..4f62a13dc481 100644
--- a/drivers/mtd/mtdcore.c
+++ b/drivers/mtd/mtdcore.c
@@ -798,12 +798,12 @@ int add_mtd_device(struct mtd_info *mtd)
 	/* Caller should have set dev.parent to match the
 	 * physical device, if appropriate.
 	 */
-	mtd->dev.type = &mtd_devtype;
 	mtd->dev.class = &mtd_class;
 	mtd->dev.devt = MTD_DEVT(i);
 	error = dev_set_name(&mtd->dev, "mtd%d", i);
 	if (error)
 		goto fail_devname;
+	mtd->dev.type = &mtd_devtype;
 	dev_set_drvdata(&mtd->dev, mtd);
 	mtd_check_of_node(mtd);
 	of_node_get(mtd_get_of_node(mtd));

---
base-commit: 1c1a342aceec79528b3ba51f376eca2be5928fe7
change-id: 20261108-submit-mtd-devname-retry-v1-3f7cb2ed8083

Best regards,
--  
James Hilliard <james.hilliard1@gmail.com>


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

Re: [PATCH] mtd: leave the device type unset when naming fails

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 1 potential
issue to consider:
- [Low] mtd: inaccurate Fixes tag in commit message for add_mtd_device()
--

commit e76bbebb3c54ac45bcccfcf9ff4a9021c3ac0eac
Author: James Hilliard <james.hilliard1@gmail.com>

> mtd: leave the device type unset when naming fails
>
> add_mtd_device() uses dev.type to reject a second registration of the
> same MTD. It sets that field before dev_set_name(), but the failure path
> for the name allocation leaves it set even though no device was
> registered. Retrying registration then warns and fails with -EEXIST.
>
> Set the type only after naming succeeds. The later registration error
> paths already clear it before releasing the device.
>
> Fixes: fa06052d637b ("mtd: Convert to dynamically allocated bdi infrastructure")
> Cc: stable@vger.kernel.org
> Signed-off-by: James Hilliard <james.hilliard1@gmail.com>

[Severity: Low]
Should the Fixes tag reference 6697dae1e2da ("mtd: capture device name
setting failure when adding mtd") instead?

While fa06052d637b ("mtd: Convert to dynamically allocated bdi
infrastructure") changed the duplicate registration guard in
add_mtd_device() to check mtd->dev.type, dev_set_name() was not checked
for errors at that point.  The fail_devname error path that returns before
device_register() with mtd->dev.type still set was introduced in
6697dae1e2da ("mtd: capture device name setting failure when adding mtd"),
and clearing mtd->dev.type on the fail_added path was added in
9d4af746af8c ("mtd: fix double free and WARN_ON in add_mtd_device() error
paths").

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261108-submit-mtd-devname-retry-v1-v1-1-a67f332c54dd@gmail.com?part=1

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