dts: imx8ml_m7: fix mailbox node compatible for NXP i.MX8M Plus SoC

[emfend]

Commit 7d1f435 ("drivers: ipc: Enable messaging unit driver for iMX.RT multicore SOCs") changed the device tree compatible for the NXP i.MX rev2 MU from 'nxp,imx-mu' to 'nxp,imx-mu-rev2'.
To make the mailbox usable again on this SoC, the compatible needs to be adjusted accordingly.

[danieldegrasse]

To make the mailbox usable again on this SoC, the compatible needs to be adjusted accordingly.

Slightly confused as to why? When I made that commit, I believe that the reason I used a separate compatible was because the RT series MCUs used a different SDK, and needed a different header for the same IP block (looking back on it, I question the wisdom of that choice)

I guess the iMX8M SOCs have now migrated to MCUX SDK (@JiafeiPan can you confirm?), so now both platforms can use the fsl_mu.h header?

If so, we might want to consider fixing my mistake here- we could probably use the same compatible, and make the header inclusion dependent on CONFIG_HAS_MCUX rather than the compatible

[iuliana-prodan]

@danieldegrasse just FYI we are also using the fsl_mu for iMX8M DSP - see https://github.com/zephyrproject-rtos/zephyr/blob/main/dts/xtensa/nxp/nxp_imx8m.dtsi#L72
Recently was added mbox_nxp_imx_mu.c which is also using fsl_mu from mcux_sdk.

So do we use mu_imx somewhere? Do we still need it?

If yes, maybe we can switch all to fsl_mu and remove this and stop the confusion of mu vs mu2.
If no, then we can remove it faster :)

[MaureenHelm]

@danieldegrasse ping

[danieldegrasse]

@emfend can you look at #64662 to see if it resolves your issue? I'd prefer to remove nxp,imx-mu-rev2, since using a DT compatible to determine which HAL header to include was a bad decision to start with.

[emfend]

@emfend can you look at #64662 to see if it resolves your issue? I'd prefer to remove nxp,imx-mu-rev2, since using a DT compatible to determine which HAL header to include was a bad decision to start with.

Yes, that looks good. This means I can at least compile it again. Unfortunately, I can't test this on real hardware at the moment.

[dleach02]

@emfend, if you are good with @danieldegrasse direction with his PR, can you close this one?