linux-kernel - Re: [PATCH 1/1] firmware: arm_scmi: Increase MAX

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-Id: <DEVFS1MAP8J6.2263USIPE4Y74@linaro.org>
Date: Thu, 11 Dec 2025 13:54:01 +0000
From: "Alexey Klimov" <alexey.klimov@...aro.org>
To: "Sudeep Holla" <sudeep.holla@....com>
Cc: "Vivek Aknurwar" <vivek.aknurwar@....qualcomm.com>,
 <cristian.marussi@....com>, <linux-kernel@...r.kernel.org>,
 <linux-arm-kernel@...ts.infradead.org>, <linux-arm-msm@...r.kernel.org>,
 <mike.tipton@....qualcomm.com>
Subject: Re: [PATCH 1/1] firmware: arm_scmi: Increase MAX_OPPS to 64

On Thu Dec 11, 2025 at 1:48 PM GMT, Sudeep Holla wrote:
> On Thu, Dec 11, 2025 at 01:14:06PM +0000, Alexey Klimov wrote:
>> > On 10/14/2025 12:34 AM, Vivek Aknurwar wrote:
>> >> Some upcoming SoCs define more than 32 operating performance points (OPPs),
>> >> exceeding the current SCMI protocol limit. Increase MAX_OPPS to 64
>> >> (next power of 2) to support these configurations.
>> 
>> Didn't touch for a while. The way it is stated confuses me a bit.
>> Should the value defined by protocol be updated out of the blue?
>> Should the protocol (defined by spec) be changed first?
>> 
>
> Ah good point on confusing commit message. I just assumed it is limitation
> of the implementation. I can update the log when applying. It is not spec
> or protocol limitation for sure.
>
> How about this ?
>
>   | firmware: arm_scmi: Increase performance MAX_OPPS limit to 64
>   |
>   | Some platforms expose more than 32 operating performance points (OPPs)
>   | per performance domain via the SCMI performance protocol, but the
>   | driver currently limits the number of OPPs it can handle to 32 via
>   | MAX_OPPS.
>   |
>   | Bump MAX_OPPS to 64 so that these platforms can register all their
>   | performance levels. This is an internal limit in the driver only and
>   | does not affect the SCMI protocol ABI.
>   |
>   | 64 is chosen as the next power of two above the existing limit.

Yeah, that sounds better :)

I also thought that this was a driver limitation, not the protocol/spec one
as stated in the original patch.

I don't mind updating the commit message like this (but I am not the author
of the original patch).

Best regards,
Alexey