Re: [PATCH v3 13/23] arm64: dts: qcom: x1e80100: Fix ADSP memory base and length

[Johan Hovold <johan@kernel.org>]

On Fri, Dec 13, 2024 at 04:45:30PM +0100, Krzysztof Kozlowski wrote:
> On 13/12/2024 16:35, Johan Hovold wrote:
> > On Fri, Dec 13, 2024 at 03:54:02PM +0100, Krzysztof Kozlowski wrote:
> >> The address space in ADSP PAS (Peripheral Authentication Service)
> >> remoteproc node should point to the QDSP PUB address space
> >> (QDSP6...SS_PUB): 0x0680_0000 with length of 0x10000.
> >>
> >> 0x3000_0000, value used so far, is the main region of CDSP and was
> >> simply copied from other/older DTS.
> >>
> >> Correct the base address and length, which also moves the node to
> >> different place to keep things sorted by unit address.  The diff looks
> >> big, but only the unit address and "reg" property were changed.  This
> >> should have no functional impact on Linux users, because PAS loader does
> >> not use this address space at all.
> >>
> >> Fixes: 5f2a9cd4b104 ("arm64: dts: qcom: x1e80100: Add ADSP/CDSP remoteproc nodes")
> >> Cc: stable@vger.kernel.org
> > 
> > Why bother with backporting any of these when there is no functional
> > impact?
> 
> Not sure, I assumed someone might be using kernel DTS from stable
> branches in other projects. Kernel is the source of DTS and stable
> kernel has the DTS in both stable and fixed way. If 3rd party project
> keeps pulling always latest DTS from latest kernel, they will see so
> many ABI breaks and so many incompatibilities (we discussed it in
> Vienna) that they will probably curse their approach and say "never
> again". Using stable branch DTS could be a solution.

That makes some sense.

> Such 3rd party project might actually use above device nodes in their
> drivers. It's just some of Linux kernel drivers which do not use them
> (other like PIL seems to use addresses).

But this is more questionable given that the current addresses are
completely off in this case.

> Plus DTS is used by 3rd party Linux kernels (out of tree), which while
> we do not care in a way of driving our development, but we do consider
> them possible users. They also might be relying on stable kernel branch
> for this.

Same here.

I realise this is a bit of a grey area, but given the size of the diffs
and the no functional impact this could be an opportunity to try to
uphold the stable kernel rules:

	- It cannot be bigger than 100 lines, with context.
	- It must either fix a real bug that bothers people or just add
	  a device ID.

Johan