[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <aDmhYZydu5UlTxF1@pluto>
Date: Fri, 30 May 2025 13:15:29 +0100
From: Cristian Marussi <cristian.marussi@....com>
To: Dhruva Gole <d-gole@...com>
Cc: Cristian Marussi <cristian.marussi@....com>, sudeep.holla@....com,
Ulf Hansson <ulf.hansson@...aro.org>, arm-scmi@...r.kernel.org,
linux-arm-kernel@...ts.infradead.org, linux-pm@...r.kernel.org,
linux-kernel@...r.kernel.org
Subject: Re: [RFC PATCH] pmdomain: arm: scmi_pm_domain: Do lazy init as part
of probe
On Fri, May 30, 2025 at 04:05:27PM +0530, Dhruva Gole wrote:
> Optimize the SCMI power domain driver to only initialize domains that are
> actually referenced in the device tree. Previously, the driver would
> initialize all possible domains up to the maximum ID, which could lead to
> unnecessary firmware calls and longer probe times.
Hi Dhruva,
this in general mostly fine to me.
>
> Key changes:
> - Scan device tree to identify which power domains are actually referenced
> - Use bitmap to track needed domains instead of initializing all
> - Only perform state queries and initialization for referenced domains
> - Maintain proper array sizing for power domain framework compatibility
> - Keep full provider structure to support late binding
>
> This optimization reduces probe time and unnecessary firmware interactions
> by only initializing power domains that are actually used in the system.
> For example, in a system with 100 possible domains but only 3 referenced
> in the device tree, we now only initialize those 3 domains instead of
> all 100.
>
> Signed-off-by: Dhruva Gole <d-gole@...com>
> ---
>
> Hi all,
> The approach of doing a lazy init was briefly proposed in this [1] 2024
> Embedded Open Source talk. It was also brought up in the monthly ARM
> SCMI meetings that take place and it didn't recieve too much opposition.
>
> This greatly helps to improve the boot time, and I have some data to
> back this up as well. This[2] experiment was done on a TI AM62L SoC
> (which is yet to make it upstream) to measure the time taken in the scmi
> pm domain probe function when it does a full 0..N scmi pd init vs just
> the ones being used in the device tree.
>
> If you have any feedback on this, please let me know. If not, I will go
> ahead and post a "non-RFC" patch assuming everyone is mostly on board
> with this. Also request other SCMI consumers to test this out as much as
> possible to see if it breaks in any situations.
I'll post a few minor remarks on the upcoming non-RFC.
Thanks,
Cristian
Powered by blists - more mailing lists