Re: [RFC PATCH v2 1/2] platform: arm64: qcom-hamoa-ec: monitor Slim 7x legacy sensors

[Anvesh Jain P <anvesh.p@oss.qualcomm.com>]

On 9/8/2026 8:06 PM, Birk Skyum wrote:
> Hi Anvesh,
> 
> Sashiko flagged the pre-existing SCI cleanup gap on the reference-board
> probe path. I reproduced it in KUnit with simulated I2C and IRQ devices:
> when the thermal-capability read fails, SCI remains enabled after the
> driver core releases the interrupt.
> 
> I noticed you already proposed the devm_add_action_or_reset() conversion
> and agreed to split it out after Konrad's review:
> https://lkml.iu.edu/hypermail/linux/kernel/2607.3/10628.html
> 
> With that conversion applied locally, the failure-path and removal tests
> pass. Is there a standalone version available to test? Happy to share the
> reproducer if useful. I'll keep this separate from the Yoga sensor RFC.
> 
> For the other v2 findings, I've queued the named I2C initializers for the
> next revision. Guenter left the driver split to me, and v2 already moved
> the documentation out of Documentation/hwmon as requested.
> 
> Thanks,
> Birk

Hi Birk,

Yes, we already have that fix, SCI disable is now registered as a devm
action right after it's enabled in probe, so it unwinds on any probe
failure, not just on remove(). Same for the i2c_device_id
named-initializer thing bot flagged.

I'm working on v2 right now. Once I get consensus on all the review
comments, I'll post it with these fixes included very soon, once it's
out so you can test against your reproducer.

-- 
Best Regards,
Anvesh