Re: [PATCH 7/9] platform/x86: asus-wmi: add enable/disable CPU cores
From: Luke Jones
Date: Tue May 28 2024 - 17:38:04 EST
On Tue, 28 May 2024, at 9:27 PM, Ilpo Järvinen wrote:
> Hi,
>
> Hans, please check my question below.
>
> On Tue, 28 May 2024, Luke D. Jones wrote:
>
> > Exposes the WMI functions for enable/disable of performance and
> > efficiency cores on some laptop models (largely Intel only).
> >
> > Signed-off-by: Luke D. Jones <luke@xxxxxxxxxx>
> > ---
>
> > diff --git a/Documentation/ABI/testing/sysfs-platform-asus-wmi b/Documentation/ABI/testing/sysfs-platform-asus-wmi
> > index 3b4eeea75b7b..ac881e72e374 100644
> > --- a/Documentation/ABI/testing/sysfs-platform-asus-wmi
> > +++ b/Documentation/ABI/testing/sysfs-platform-asus-wmi
> > @@ -226,3 +226,22 @@ Description:
> > Set panel to UHD or FHD mode
> > * 0 - UHD,
> > * 1 - FHD
> > +
> > +What: /sys/devices/platform/<platform>/cores_enabled
> > +Date: Jun 2024
> > +KernelVersion: 6.11
> > +Contact: "Luke Jones" <luke@xxxxxxxxxx>
> > +Description:
> > + Enable/disable efficiency and performance cores. The format is
> > + 0x[E][P] where [E] is the efficiency core count, and [P] is
> > + the perfromance core count. If the core count is a single digit
>
> performance
>
> > + it is preceded by a 0 such as 0x0406; E=4, P=6, 0x1006; E=10, P=6
> > +
> > +What: /sys/devices/platform/<platform>/cores_max
> > +Date: Jun 2024
> > +KernelVersion: 6.11
> > +Contact: "Luke Jones" <luke@xxxxxxxxxx>
> > +Description:
> > + Show the maximum performance and efficiency core countin format
> > + 0x[E][P] where [E] is the efficiency core count, and [P] is
> > + the perfromance core count.
>
> performance
>
> > diff --git a/drivers/platform/x86/asus-wmi.c b/drivers/platform/x86/asus-wmi.c
> > index 4b045f1828f1..f62a36dfcd4b 100644
> > --- a/drivers/platform/x86/asus-wmi.c
> > +++ b/drivers/platform/x86/asus-wmi.c
> > @@ -815,6 +815,46 @@ static ssize_t panel_fhd_store(struct device *dev,
> > WMI_SIMPLE_SHOW(panel_fhd, "%d\n", ASUS_WMI_DEVID_PANEL_FHD);
> > static DEVICE_ATTR_RW(panel_fhd);
> >
> > +/* Efficiency and Performance core control **********************************/
> > +static ssize_t cores_enabled_store(struct device *dev,
> > + struct device_attribute *attr,
> > + const char *buf, size_t count)
> > +{
> > + struct asus_wmi *asus = dev_get_drvdata(dev);
> > + int result, err;
> > + u32 cores, max;
> > +
> > + result = kstrtou32(buf, 16, &cores);
> > + if (result)
> > + return result;
> > +
> > + err = asus_wmi_get_devstate(asus, ASUS_WMI_DEVID_CORES_MAX, &max);
> > + if (err < 0)
> > + return err;
> > +
> > + if (cores > max) {
>
> This only checks one part of it and the P part can contain whatever
> garbage as long as E is small enough?
>
> But I'm not sure if it's good idea to have these two changed through the
> same sysfs file, I'm leaning more on that it would be better to split the
> interface for P and E.
>
> Hans, what you think about this?
I'm inclined to agree. It's no issue to change it.
Ack all reviews. I'll work through them over the week but with a lower priority while I await both yours and Hans response in other replies with Mario regarding the firmware_attributes change/use.
>
> > + pr_warn("Core count 0x%x exceeds max: 0x%x\n", cores, max);
> > + return -EIO;
> > + }
> > +
> > + err = asus_wmi_set_devstate(ASUS_WMI_DEVID_CORES_SET, cores, &result);
> > + if (err) {
> > + pr_warn("Failed to set cores_enabled: %d\n", err);
> > + return err;
> > + }
> > +
> > + pr_info("Enabled core count changed, reboot required\n");
> > + sysfs_notify(&asus->platform_device->dev.kobj, NULL, "cores_enabled");
> > +
> > + return count;
> > +}
>
> > @@ -4131,6 +4173,9 @@ static umode_t asus_sysfs_is_visible(struct kobject *kobj,
> > devid = ASUS_WMI_DEVID_PANEL_OD;
> > else if (attr == &dev_attr_panel_fhd.attr)
> > devid = ASUS_WMI_DEVID_PANEL_FHD;
> > + else if (attr == &dev_attr_cores_enabled.attr
> > + || attr == &dev_attr_cores_max.attr)
>
> Wrong alignment.
>
> --
> i.
>
>