[PATCH 1/2] cdev: pass master cdev to cdev ops
Ahmad Fatoum
a.fatoum at pengutronix.de
Mon Jun 2 03:20:32 PDT 2025
Hi Sascha,
On 6/2/25 12:12, Sascha Hauer wrote:
> On Mon, Jun 02, 2025 at 11:03:56AM +0200, Ahmad Fatoum wrote:
>> Hello Sascha,
>>
>> On 6/2/25 10:51, Sascha Hauer wrote:
>>> cdev partitions are cdevs themselves, but they are handled internally in
>>> the devfs-core. cdev partitions should not be passed as context pointers
>>> to the cdev ops though, because the drivers might do a container_of() on
>>> them to retrieve their driver data. Pass the original cdev the driver
>>> has registered as context to the ops. With this we can drop some quirks
>>> in the nvmem core in the next step.
>>
>> Given that read/write and so on take absolute offsets, I agree that this
>> is a sensible change.
>>
>>> Signed-off-by: Sascha Hauer <s.hauer at pengutronix.de>
>>
>> With below nit addressed:
>>
>> Reviewed-by: Ahmad Fatoum <a.fatoum at pengutronix.de>
>>
>>> ---
>>> fs/devfs-core.c | 53 +++++++++++++++++++++++++++++++++++++++++------------
>>> 1 file changed, 41 insertions(+), 12 deletions(-)
>>>
>>> diff --git a/fs/devfs-core.c b/fs/devfs-core.c
>>> index 5fafcaecc70d63287e0b716cc0133198c28e79e9..73a13be7336df1eb4e24e8a936986b6069f3783c 100644
>>> --- a/fs/devfs-core.c
>>> +++ b/fs/devfs-core.c
>>> @@ -224,12 +224,21 @@ int cdev_find_free_index(const char *basename)
>>> return -EBUSY; /* all indexes are used */
>>> }
>>>
>>> +static struct cdev *cdev_get_master(struct cdev *cdev)
>>> +{
>>> + while (cdev && cdev_is_partition(cdev))
>>> + cdev = cdev->master;
>>
>> This can be made into a simple if condition as we do not support links
>> to links, see cdev_readlink.
>
> This patch is not about links, it's about partitions and we can indeed
> partitiion a partition.
>
> You can try with
>
> addpart /dev/imx-ocotp 128 at 0x10(foo)
> addpart /dev/imx-ocotp.foo 64 at 0x10(foo1)
>
> reading from the latter fails with a core dump without this patch.
>
> As far as I can tell links are not affected by this problem as we always
> resolve links by doing a cdev = cdev_readlink(cdev), so the link cdevs
> are never passed down to drivers.
Thanks for clarification. You can collect my R-B.
Cheers,
Ahmad
>
> Sascha
>
--
Pengutronix e.K. | |
Steuerwalder Str. 21 | http://www.pengutronix.de/ |
31137 Hildesheim, Germany | Phone: +49-5121-206917-0 |
Amtsgericht Hildesheim, HRA 2686 | Fax: +49-5121-206917-5555 |
More information about the barebox
mailing list