Re: [PATCH v15 3/5] rpmsg: core: match rpmsg device IDs by prefix

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [High] Global change to prefix matching breaks exact match semantics, causing namespace collisions and incorrect driver binding.
- [High] Prefix matching breaks module auto-loading via udev because exported module aliases remain exact matches.
--

--- Patch [3]: [PATCH v15 3/5] rpmsg: core: match rpmsg device IDs by prefix ---
commit e561ac26959a610ff64bc4d231b963dec9f0caf9
Author: Shenwei Wang <shenwei.wang@nxp.com>

rpmsg: core: match rpmsg device IDs by prefix

This patch updates rpmsg_id_match to compare only the length of the driver's
ID string instead of the full RPMSG_NAME_SIZE. This allows id_table entries
to act as prefixes, enabling drivers to bind to dynamically suffixed device
names.

> diff --git a/drivers/rpmsg/rpmsg_core.c b/drivers/rpmsg/rpmsg_core.c
> index 04bfcc75f05b4..97a4db52b0401 100644
> --- a/drivers/rpmsg/rpmsg_core.c
> +++ b/drivers/rpmsg/rpmsg_core.c
> @@ -386,7 +386,9 @@ ATTRIBUTE_GROUPS(rpmsg_dev);
>  static inline int rpmsg_id_match(const struct rpmsg_device *rpdev,
>  				  const struct rpmsg_device_id *id)
>  {
> -	return strncmp(id->name, rpdev->id.name, RPMSG_NAME_SIZE) == 0;
> +	size_t len = strnlen(id->name, RPMSG_NAME_SIZE);
> +
> +	return strncmp(id->name, rpdev->id.name, len) == 0;

[Severity: High]
Does changing this to a strict prefix match prevent rpmsg drivers from
enforcing an exact name match?

If there is a driver for "sensor" and another for "sensor-temp", wouldn't
the "sensor" driver incorrectly bind to the "sensor-temp" device if it
happens to be evaluated first?

It appears the correct architectural way to multiplex multiple identical
services in the rpmsg framework is to use the exact same service name
(channel name) and distinguish instances using unique source and destination
endpoint addresses, rather than dynamically suffixed names.

[Severity: High]
Will this prefix matching logic break module auto-loading via udev?

When the firmware creates a device with a dynamically suffixed name,
rpmsg_uevent() emits a uevent containing the full suffixed name:

drivers/rpmsg/rpmsg_core.c:rpmsg_uevent() {
    ...
    return add_uevent_var(env, "MODALIAS=" RPMSG_DEVICE_MODALIAS_FMT,
                          rpdev->id.name);
}

However, scripts/mod/file2alias.c generates module aliases based on the
exact string in the rpmsg_device_id table, without a trailing wildcard:

scripts/mod/file2alias.c:do_rpmsg_entry() {
    ...
    module_alias_printf(mod, false, RPMSG_DEVICE_MODALIAS_FMT, *name);
    ...
}

Since the module alias lacks a wildcard, userspace tools like modprobe
will attempt an exact string match and fail to load the driver for the
suffixed device name.

>  }
>  
>  /* match rpmsg channel and rpmsg driver */

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260721204704.400781-1-shenwei.wang@oss.nxp.com?part=3