Re: [PATCH V1 RESEND 1/3] perf/imx_ddr: Add system PMU identifier for userspace

From: John Garry
Date: Wed May 27 2020 - 10:35:57 EST

I also really dislike this. What's the preferred way to identify the SoC
from userspace?

/proc/cpuinfo? ;)

The *SoC*!

For an non-firmware specific case, I'd say soc_device should be. I'd
guess ACPI systems don't use it and for them it's dmidecode typically.
The other problem I have with soc_device is it is optional.

Hi Will,

John -- what do you think about using soc_device to expose this information,
with ACPI systems using DMI data instead?

Generally I don't think that DMI is reliable, and I saw this as the least preferred choice. I'm looking at the sysfs DMI info for my dev board, and I don't even see anything like a SoC identifier.

As for the event_source device sysfs identifier file, it would not always contain effectively the same as the SoC ID.

Certain PMUs which I'm interested in plan to have probe-able identification info available in future.

BTW, Shaokun now tells me that the HiSi uncore PMU HW have such registers to identify the implementation. I didn't know.

So we could add that identifier file for those PMUs as proof-of-concept, exposing that register.

As for other PMUs which I'm interested in, again, future versions should have such registers to self-identify.

So using something derived from the DT compat string would hopefully be the uncommon case.