Re: [PATCH v3 1/2] arm64: topology: move store_cpu_topology() to shared code

From: Conor.Dooley
Date: Mon Jul 11 2022 - 12:39:15 EST

On 11/07/2022 16:24, Sudeep Holla wrote:
> EXTERNAL EMAIL: Do not click links or open attachments unless you know the content is safe
> On Mon, Jul 11, 2022 at 04:50:38PM +0200, Greg Kroah-Hartman wrote:
>> On Mon, Jul 11, 2022 at 03:35:42PM +0100, Sudeep Holla wrote:
>>> On Sat, Jul 09, 2022 at 04:23:54PM +0100, Conor Dooley wrote:
>>>> From: Conor Dooley <conor.dooley@xxxxxxxxxxxxx>
>>>> arm64's method of defining a default cpu topology requires only minimal
>>>> changes to apply to RISC-V also. The current arm64 implementation exits
>>>> early in a uniprocessor configuration by reading MPIDR & claiming that
>>>> uniprocessor can rely on the default values.
>>>> This is appears to be a hangover from prior to '3102bc0e6ac7 ("arm64:
>>>> topology: Stop using MPIDR for topology information")', because the
>>>> current code just assigns default values for multiprocessor systems.
>>>> With the MPIDR references removed, store_cpu_topolgy() can be moved to
>>>> the common arch_topology code.
>>> Looks good. FWIW,
>>> Reviewed-by: Sudeep Holla <sudeep.holla@xxxxxxx>
>>>> CC: stable@xxxxxxxxxxxxxxx
>>> However, while I understand the reason why this is needed in stable trees
>>> for RISC-V, I am not sure if we want this for stable tree at-least on arm64.
>>> I leave that part to Greg and Will.
>> Why would it be good for one arch but bad for another?
> Not really bad as such. Just needs testing and must not change much ideally,
> but it really depends on which stable trees we will target and what is the
> original state there. As mentioned in the commit, this changed a bit around
> v5.8/9 on arm64 and not sure what kernels RISC-V needs this. There could
> be some surprises on some Andriod platforms but that is something we can
> look at when if and when there are complaints.
> I am in general not sure what is the -stable tree rules is such situation and
> hence made the noise so that we are aware that we may need more work than just
> backporting this patch. Also this is just my opinion. If we decide to backport
> esp. to kernels older than the one containing 3102bc0e6ac7, then arm64 may need
> more changes or probably we can pull that commit if that makes it easier. Based
> on what is decided and what are the targeted -stable trees, we can dig deeper.

There's always the option of, for the older kernels, not migrating arm64 at all
and just wrap store_cpu_topo with "if RISCV" rather than "if RISCV || ARM64".