On Fri, Oct 17, 2025 at 5:57 PM Frank Li [off-list ref] wrote:

On Fri, Oct 17, 2025 at 04:20:23AM -0700, Laurentiu Mihalcea wrote:

quoted

From: Laurentiu Mihalcea <redacted>

Some NXP SoCs integrate one or more, per-subsystem, block control IPs,
which allow users to control the assertion and de-assertion of the
reset lines tied to some peripherals present in said subsystems. Some
examples of such SoCs include i.MX8MP with AUDIOMIX block control and
i.MX8ULP with SIM LPAV.

Some of the aformentioned block control IPs exhibit a common pattern with
respect to the assertion and de-assertion of the reset lines. Namely, the
user is able to control the state of the reset line by toggling a bit from
one of the IP's registers.

Linux can take advantage of this pattern and, instead of having one driver
for each block control IP, a single, more generic driver could be used.

To allow this to happen, the previous approach, in which a single reset
map is used, is replaced by a per-driver approach, in which each auxiliary
device driver holds a reference to a certain reset map.

Can you shorter your commit message?, basically, you just add
imx8mp_reset_info for difference auxiliary_device_id.

It is always preferred to have commit messages trying to explain the context.
I have never heard anyone complaining that the commit message is too long :D.

Daniel.