Re: [PATCH] platform/x86: serial-multi-instantiate: Do not request IRQ for AW88399

From: Marco Giunta

Date: Thu Nov 12 2026 - 10:02:36 EST


Hi all,

chiming in as the AW88399 HDA side-codec driver maintainer.
I agree with Ilpo that more context is needed. What's the issue with
IRQ_RESOURCE_AUTO exactly? For all supported Lenovo laptops, so far no one
has reported problems with SMI i2c device instantiation (as far as I'm
aware). On my own Legion Pro 7 16AFR10H with debugging enabled, the SMI
driver simply prints "Using gpio irq".

Based on this:
https://lore.kernel.org/linux-sound/20261111043601.306427-1-guanlan@xxxxxxxxxx/
I know awinic is adding support for a new Lenovo device. Does this laptop
have issues in its ACPI firmware, that cause SMI to error out before
instantiating the I2C devices?

That said, given that we currently aren't really using anything
IRQ-related in the HDA driver, this change doesn't look particularly
problematic if this new Lenovo laptop lacks a GPIO assigned for IRQ.
However, I think IRQ_RESOURCE_AUTO | IRQ_RESOURCE_OPT would be more
appropriate than IRQ_RESOURCE_NONE, as we can keep IRQ resolution on the
laptops where ACPI firmware provides one, while avoiding SMI erroring out
on those where it's missing.
This is the same approach taken for tas2781 in commit 1d1b8b0734af
("platform/x86: serial-multi-instantiate: Add IRQ_RESOURCE_OPT for IRQ
missing projects"), which addresses the same situation.
The corresponding HDA driver also doesn't make use of IRQ, so I think
the tas2781 precedent applies here cleanly.

Tao, can you confirm if this is the case (no IRQ assignment in the ACPI
tables for this new device)? And could you share the ACPI DSDT excerpt
for the AWDZ8399 node of the device in question? For reference, my
Legion's one can be found here:
https://github.com/marco-giunta/lenovo-acpi-dump
If the above is correct, then please submit a v2 with this change (with my
Suggested-by); alternatively, I can submit it myself if preferred (e.g.
with your Reported-by).

Also, I know the get_maintainer script won't print my name for changes
like this, but please CC me anyway for anything impacting the HDA AW88399
driver; I'd like to catch potential regressions early. Or at the very
least, I can provide Tested-by tags since I actually own one of the
affected laptops.

Best regards,
Marco