On 1/28/22 6:30 PM, Bjorn Andersson wrote:
On Fri 28 Jan 02:39 PST 2022, Lukasz Luba wrote:
On 1/28/22 3:25 AM, Bjorn Andersson wrote:
In the event that the SoC is under thermal pressure while booting it's
possible for the dcvs notification to happen inbetween the cpufreq
framework calling init and it actually updating the policy's
related_cpus cpumask.
Prior to the introduction of the thermal pressure update helper an empty
cpumask would simply result in the thermal pressure of no cpus being
updated, but the new code will attempt to dereference an invalid per_cpu
variable.
Just to confirm, is that per-cpu var the 'policy->related_cpus' in this
driver?
Correct, we boot under thermal pressure, so the interrupt fires before
we return from "init", which means that related_cpus is still 0.
Avoid this problem by using the newly reintroduced "ready" callback, to
postpone enabling the IRQ until the related_cpus cpumask is filled in.
Fixes: 0258cb19c77d ("cpufreq: qcom-cpufreq-hw: Use new thermal pressure update function")
You have 'Fixes' tagging here, which might be picked by the stable tree.
The code uses the reverted callback .ready(), which might be missing
there (since patch 1/2 doesn't have tagging). This patch looks like a
proper fix for the root cause.
Yes, the pair would need to be picked up.
Anyway, I'm going to send a patch, which adds a check for null cpumask
in the topology_update_thermal_pressure()
It was removed after the review comments:
https://lore.kernel.org/linux-pm/20211028054459.dve6s2my2tq7odem@vireshk-i7/
I attempted that in v1:
https://lore.kernel.org/all/20220118185612.2067031-2-bjorn.andersson@xxxxxxxxxx/
And while patch 1 is broken, I think Greg and Sudeep made it clear that
they didn't want a condition to guard against the caller passing cpus of
0.
Thanks for the link, I missed that conversation.
That's why I in v2 reverted to postpone the thermal pressure IRQ until
cpufreq is "ready".
Which is fixing the root cause, but involves this backporting
of the new API callback into stable.
Sorry to hear that you had to fight with this tricky mem fault.
There is a 'good' outcome from this: we know that the platform
instantly has thermal issues during boot.
Regards,
Lukasz
