Re: Kernel crash on boot, arm64 VM

[Qu Wenruo <wqu@suse.com>]

在 2025/6/2 21:42, Sudeep Holla 写道:
> On Mon, Jun 02, 2025 at 11:35:22AM +0100, Will Deacon wrote:
>> [+Sudeep]
>>
> 
> Thanks for adding me.
> 
>> On Mon, Jun 02, 2025 at 10:00:38AM +0930, Qu Wenruo wrote:
>>> It looks like there is a regression related to the device tree/acpi parsing
>>> in the latest upstream kernel branch.
>>
>> I've kept the crash log below, but I suspect this is due to the __free()
>> cleanup path in dt_idle_state_present(), introduced recently by
>> 5836ebeb4a2b ("cpuidle: psci: Avoid initializing faux device if no DT
>> idle states are present").

Reverting works, thanks a lot!

>>
> 
> Hi Qu,
> 
> Do you also have this commit ?
> 
> 39cdf87a97fd ("cpuidle: psci: Fix uninitialized variable in dt_idle_state_present()")

Not yet, the branch I'm using is from btrfs' development branch, which 
is slightly older than the upstream.
(The base commit is a56baa225308 ("Merge tag 'for-6.16-tag' of 
git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux"))


Just tried upstream, which also fixed the problem.

So I guess it's really the same problem.
Sorry for the noise.

I'll just notify the btrfs community to rebase the development branch

And thank you both for pointing out the proper fix.

Thanks,
Qu

> 
> Just trying to see if it is same issue or if this something else even
> with the above commit included.
>