Received-SPF: pass (google.com: domain of linux-kernel+bounces-128831-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:4601:e00::3 as permitted sender) client-ip=2604:1380:4601:e00::3;
Feedback-ID: i5ec1447f:Fastmail
From: Luke Jones <luke@ljones.dev>
To: Ilpo =?ISO-8859-1?Q?J=E4rvinen?= <ilpo.jarvinen@linux.intel.com>
Cc: Hans de Goede <hdegoede@redhat.com>, corentin.chary@gmail.com,
 platform-driver-x86@vger.kernel.org, LKML <linux-kernel@vger.kernel.org>
Subject:
 Re: [PATCH v2 8/9] platform/x86: asus-wmi: Add support for MCU powersave
Date: Wed, 03 Apr 2024 12:30:05 +1300
Message-ID: <24071185.4csPzL39Zc@fedora>
In-Reply-To: <71426f4c-a44a-cf87-7045-b3f2b3fc240e@linux.intel.com>
References:
 <20240402022607.34625-1-luke@ljones.dev>
 <20240402022607.34625-9-luke@ljones.dev>
 <71426f4c-a44a-cf87-7045-b3f2b3fc240e@linux.intel.com>
Precedence: bulk
MIME-Version: 1.0
Content-Transfer-Encoding: quoted-printable
Content-Type: text/plain; charset="utf-8"

On Wednesday, 3 April 2024 12:01:15=E2=80=AFAM NZDT Ilpo J=C3=A4rvinen wrot=
e:
> On Tue, 2 Apr 2024, Luke D. Jones wrote:
> > Add support for an MCU powersave WMI call. This is intended to set the
> > MCU in to a low-power mode when sleeping. This mode can cut sleep power
> > use by around half.
> >=20
> > Signed-off-by: Luke D. Jones <luke@ljones.dev>
>=20
> Hi,
>=20
> I fail to follow the logic of the patch here. This patch makes it
> configurable which is not bad in itself but what is the reason why a user
> would not always want to cut sleep power usage down? So this sounds like a
> feature that the user wants always enabled if available.
>=20
> So what are the downsides of enabling it if it's supported?

Honestly, I'm not sure. Previously it was a source of issues but with recen=
t=20
bios and more work in the various gaming-handheld distros it's much less of=
 a=20
problem. The issue before was that the MCU would appear every second suspen=
d=20
due to the suspend sequence being more parallel compared to windows being=20
sequential - the acpi calls related to this would "unplug" the USB devices=
=20
that are related to the n-key (keyboard, same internal dev as laptops) and =
not=20
complete it before suspend. And then on resume it was unreliable.

I worked around this by calling the unplug very early, and trying to "plug"=
=20
super early also so that subsystems wouldn't notice the absence of the devi=
ce=20
and create new devices + paths on add. Most of the requirement for that cam=
e=20
from some userspace programs unable to handle the unplug/plug part, but als=
o=20
bad device state occurring.

But now with the forced wait for the device to finish its task, and then th=
e=20
forced wait before letting it add itself back everything is fine - although=
 it=20
does mean everything sees a device properly unplugged and plugged.

All of the above is to say that the "powersave" function was also part of t=
he=20
issue as delayed things further and we couldn't add the device back before=
=20
subsystems noticed.

Distros like bazzite and chimeraOS are now using this patch, and I'd like t=
o=20
leave it to them to set a default for now. If it turns out everything is=20
indeed safe as houses then we can adjust the kernel default.

A side-note I think is that because there is a forced wait time due to unab=
le=20
to use the right acpi path, the old excuse of "but users might want the dev=
ice=20
to wake faster by turning off powersave" doesn't really apply now.

I will discuss with the main stakeholders, but for now can we accept as is?=
 If=20
changes are required we'll get them done before the merge window.

>=20
> > ---
> >=20
> >  .../ABI/testing/sysfs-platform-asus-wmi       | 11 +++-
> >  drivers/platform/x86/asus-wmi.c               | 50 +++++++++++++++++++
> >  2 files changed, 60 insertions(+), 1 deletion(-)
> >=20
> > diff --git a/Documentation/ABI/testing/sysfs-platform-asus-wmi
> > b/Documentation/ABI/testing/sysfs-platform-asus-wmi index
> > 41b92e53e88a..28144371a0f1 100644
> > --- a/Documentation/ABI/testing/sysfs-platform-asus-wmi
> > +++ b/Documentation/ABI/testing/sysfs-platform-asus-wmi
> > @@ -202,4 +202,13 @@ Contact:	"Luke Jones" <luke@ljones.dev>
> >=20
> >  Description:
> >  		Set if the BIOS POST sound is played on boot.
> >  	=09
> >  			* 0 - False,
> >=20
> > -			* 1 - True
> > \ No newline at end of file
> > +			* 1 - True
> > +
> > +What:		/sys/devices/platform/<platform>/mcu_powersave
> > +Date:		Apr 2024
> > +KernelVersion:	6.10
> > +Contact:	"Luke Jones" <luke@ljones.dev>
> > +Description:
> > +		Set if the MCU can go in to low-power mode on system=20
sleep
> > +			* 0 - False,
> > +			* 1 - True
> > diff --git a/drivers/platform/x86/asus-wmi.c
> > b/drivers/platform/x86/asus-wmi.c index ddf568ef8c5e..cf872eed0986 1006=
44
> > --- a/drivers/platform/x86/asus-wmi.c
> > +++ b/drivers/platform/x86/asus-wmi.c
> > @@ -1292,6 +1292,53 @@ static ssize_t nv_temp_target_show(struct device
> > *dev,>=20
> >  }
> >  static DEVICE_ATTR_RW(nv_temp_target);
> >=20
> > +/* Ally MCU Powersave
> > ********************************************************/ +static ssize=
_t
> > mcu_powersave_show(struct device *dev,
> > +				   struct device_attribute *attr,=20
char *buf)
> > +{
> > +	struct asus_wmi *asus =3D dev_get_drvdata(dev);
> > +	int result;
> > +
> > +	result =3D asus_wmi_get_devstate_simple(asus,
> > ASUS_WMI_DEVID_MCU_POWERSAVE); +	if (result < 0)
> > +		return result;
> > +
> > +	return sysfs_emit(buf, "%d\n", result);
> > +}
> > +
> > +static ssize_t mcu_powersave_store(struct device *dev,
> > +				    struct device_attribute *attr,
> > +				    const char *buf, size_t count)
> > +{
> > +	int result, err;
> > +	u32 enable;
> > +
> > +	struct asus_wmi *asus =3D dev_get_drvdata(dev);
> > +
> > +	result =3D kstrtou32(buf, 10, &enable);
> > +	if (result)
> > +		return result;
> > +
> > +	if (enable > 1)
> > +		return -EINVAL;
> > +
> > +	err =3D asus_wmi_set_devstate(ASUS_WMI_DEVID_MCU_POWERSAVE, enable,
> > &result); +	if (err) {
> > +		pr_warn("Failed to set MCU powersave: %d\n", err);
> > +		return err;
> > +	}
> > +
> > +	if (result > 1) {
> > +		pr_warn("Failed to set MCU powersave (result): 0x%x\n",=20
result);
> > +		return -EIO;
> > +	}
> > +
> > +	sysfs_notify(&asus->platform_device->dev.kobj, NULL,=20
"mcu_powersave");
> > +
> > +	return count;
> > +}
> > +static DEVICE_ATTR_RW(mcu_powersave);
> > +
> >=20
> >  /* Battery
> >  ********************************************************************/
> > =20
> >  /* The battery maximum charging percentage */
> >=20
> > @@ -4299,6 +4346,7 @@ static struct attribute *platform_attributes[] =
=3D {
> >=20
> >  	&dev_attr_ppt_platform_sppt.attr,
> >  	&dev_attr_nv_dynamic_boost.attr,
> >  	&dev_attr_nv_temp_target.attr,
> >=20
> > +	&dev_attr_mcu_powersave.attr,
> >=20
> >  	&dev_attr_boot_sound.attr,
> >  	&dev_attr_panel_od.attr,
> >  	&dev_attr_mini_led_mode.attr,
> >=20
> > @@ -4352,6 +4400,8 @@ static umode_t asus_sysfs_is_visible(struct kobje=
ct
> > *kobj,>=20
> >  		devid =3D ASUS_WMI_DEVID_NV_DYN_BOOST;
> >  =09
> >  	else if (attr =3D=3D &dev_attr_nv_temp_target.attr)
> >  =09
> >  		devid =3D ASUS_WMI_DEVID_NV_THERM_TARGET;
> >=20
> > +	else if (attr =3D=3D &dev_attr_mcu_powersave.attr)
> > +		devid =3D ASUS_WMI_DEVID_MCU_POWERSAVE;
> >=20
> >  	else if (attr =3D=3D &dev_attr_boot_sound.attr)
> >  =09
> >  		devid =3D ASUS_WMI_DEVID_BOOT_SOUND;
> >  =09
> >  	else if (attr =3D=3D &dev_attr_panel_od.attr)