Re: [PATCH] platform/x86: bitland-mifs-wmi: Add Xiaomi Book Pro 14 platform profile values

[Chris Taraszka]

Hi,

Following up with test results. I tested v12 on the Xiaomi Book Pro
14 (TM2424, BIOS XMAPT4B0P0909), applied on pdx86 for-next and loaded
as a module on v7.3-rc5, with a local TM2424 entry: 0x02, 0x0a
(no_set), 0x03, 0x09 and 0x04, plus skip_ac_type_check and
accept_zero_function. All four profiles apply on AC and PL1 follows,
and pm_test=devices suspend/resume is clean on AC and on battery.
Without the entry, GET fails with -EINVAL and every SET with -ENOMSG,
as expected.

Mingyou, two more things:

1. In 6/6 only -EOPNOTSUPP falls back to balanced. Without the entry
above, restoring the profile fails with -ENOMSG and the resume
callback returns it:

    bitland_mifs_wmi_resume [bitland_mifs_wmi] returns -42
    PM: failed to resume: error -42

A failed restore leaves the machine usable, so I think resume should
warn and return 0 for any error after trying the fallback. The same
goes for suspend: bitland_mifs_wmi_suspend() still returns the
laptop_profile_get() error, which aborts system suspend on any model
whose firmware reports a value missing from its table. That is what my
earlier "Don't abort suspend when platform profile access fails" patch
was about, and it could be folded into 6/6.

2. With skip_ac_type_check, bitland_pp_check_capability() still
refuses balanced-performance and performance without a power supply.
On BIOS 0909 the firmware runs 0x04 on battery (PL1 58 W), so this
blocks a mode that works here. Kento reported that BIOS 0707 NACKs
0x04 on battery, which bitland_mifs_wmi_call() already turns into an
error. Could skip_ac_type_check skip the power supply check as well
and leave the decision to the firmware? Kento, would that be a problem
on 0A0A?

Thanks,
Chris