Re: [PATCH] asus-wmi: Support the hardware GPU MUX on some laptops

From: Luke Jones <luke@ljones.dev>
To: Lukas Wunner <lukas@wunner.de>
Cc: hdegoede@redhat.com, markgross@kernel.org,
	platform-driver-x86@vger.kernel.org,
	linux-kernel@vger.kernel.org
Subject: Re: [PATCH] asus-wmi: Support the hardware GPU MUX on some laptops
Date: Mon, 15 Aug 2022 09:44:35 +1200	[thread overview]
Message-ID: <BEKMGR.6HATWYKRRTJQ1@ljones.dev> (raw)
In-Reply-To: <20220814104210.GA22519@wunner.de>

On Sun, Aug 14 2022 at 12:42:10 +0200, Lukas Wunner <lukas@wunner.de> 
wrote:
> On Sun, Aug 14, 2022 at 09:09:43PM +1200, Luke Jones wrote:
>>  On Sat, 2022-08-13 at 12:55 +0200, Lukas Wunner wrote:
>>  > On Sat, Aug 13, 2022 at 09:26:24PM +1200, Luke D. Jones wrote:
>>  > > Support the hardware GPU MUX switch available on some models. 
>> This
>>  > > switch can toggle the MUX between:
>>  > >
>>  > > - 0, Dedicated mode
>>  > > - 1, Optimus mode
>>  > >
>>  > > Optimus mode is the regular iGPU + dGPU available, while 
>> dedicated
>>  > > mode switches the system to have only the dGPU available.
>>  >
>>  > Could you please integrate this with the framework provided by:
>>  >
>>  >   include/linux/vga_switcheroo.h
>>  >   drivers/gpu/vga/vga_switcheroo.c
>>  >
>>  > vga_switcheroo will then automatically expose a sysfs interface.
>> 
>>  I did investigate this first before submitting. The way asus does 
>> it is
>>  not standard at all. On switch you must reboot, and the change isn't
>>  reflected by the ACPI get method until reboot. It's very reflective 
>> of
>>  how they used dgpu_disable to work around windows issues that we do 
>> in
>>  Linux by removing the device from the device tree.
>> 
>>  The key thing is a reboot is required. This is not done on-the-fly. 
>> I
>>  have a two year old GX502 which has the same method as exposed here,
>>  and also a 2022 TUF laptop with same method. My understanding of 
>> this
>>  pariicular method is that it isn't the same one as what Nvidia is
>>  advertising, and ASUS is perhaps misadvertising it - the suspision 
>> is
>>  raised by the fact that my GX502 machine predates what Nvidia is
>>  advertising.
> 
> I see, thanks for the explanation.  You may want to add that
> background information to the commit message if/when respinning.
> Indeed vga_switcheroo facilitates GPU switching at runtime,
> not upon next reboot.
> 
> Thanks,
> 
> Lukas

Good point, yes I will do that after reviews.