Hi, On 8/23/21 1:26 PM, Luke Jones wrote: > > > On Mon, Aug 23 2021 at 12:28:21 +0200, Bastien Nocera <hadess@xxxxxxxxxx> wrote: >> On Sat, 2021-08-21 at 09:30 +1200, Luke Jones wrote: >>> >>> >>> On Fri, Aug 20 2021 at 13:39:02 +0200, Bastien Nocera >>> <hadess@xxxxxxxxxx> wrote: >>> > On Fri, 2021-08-20 at 23:00 +1200, Luke Jones wrote: >>> > > >>> > > >>> > > On Fri, Aug 20 2021 at 12:51:08 +0200, Bastien Nocera >>> > > <hadess@xxxxxxxxxx> wrote: >>> > > > On Fri, 2021-08-20 at 12:43 +0200, Bastien Nocera wrote: >>> > > > > On Fri, 2021-08-20 at 22:33 +1200, Luke Jones wrote: >>> > > > > > > Am I going to get bug reports from Asus users that will >>> > > > > complain >>> > > > > > > that >>> > > > > > > power-profiles-daemon doesn't work correctly, where I >>> > > will >>> > > > > have >>> > > > > > > to >>> > > > > > > wearily ask if they're using an Asus Rog laptop? >>> > > > > > >>> > > > > > No. Definitely not. The changes to fan curves per-profile >>> > > need >>> > > > > to >>> > > > > > be >>> > > > > > explicitly enabled and set. So a new user will be unaware >>> > > that >>> > > > > this >>> > > > > > control exists (until they look for it) and their laptop >>> > > will >>> > > > > > behave >>> > > > > > exactly as default. >>> > > > > >>> > > > > "The user will need to change the fan curves manually so >>> > > will >>> > > > > definitely remember to mention it in bug reports" is a very >>> > > > > different >>> > > > > thing to "the user can't change the fan curves to be >>> > > nonsensical >>> > > > > and >>> > > > > mean opposite things". >>> > > > > >>> > > > > I can assure you that I will eventually get bug reports >>> > > from >>> > > > > "power >>> > > > > users" who break their setup and wonder why things don't >>> > > work >>> > > > > properly, >>> > > > > without ever mentioning the changes they made changes to >>> > > the >>> > > fan >>> > > > > curves, or anything else they might have changed. >>> > > > >>> > > > A way to taint the settings that power-profiles-daemon could >>> > > catch >>> > > > would be fine by me. I absolutely don't want to have to >>> > > support >>> > > > somebody's tweaks until they undo them. >>> > > >>> > > Definitely understood. Do you have something in mind? >>> > >>> > A sysfs attribute with boolean data that shows whether custom fan >>> > curves are used would be enough. >>> >>> The path /sys/devices/platform/asus-nb-wmi/active_fan_curve_profiles >>> should be usable like this? I added this as the method for >>> controlling >>> which fan curves for which profiles are active. >>> >>> If empty, then no custom fan curves are active at all. If it contains >>> any combination of strings "quiet, balanced, performance" then those >>> associated (named) platform_profiles have an active fan curve and you >>> can throw up a general warning, maybe add the contents of that file >>> too? >> >> That works for me, although I would probably have preferred a way that >> wasn't specific to the asus-wmi module, I'm sure I can made do with >> that. > > Oh I see, you were looking to get a more general solution implemented? Maybe something like "/sys/devices/platform/asus-nb-wmi/platform_profile_tainted"? This could be an opportunity to maybe make a standardised naming scheme for it. That would standardize the name, but not the location (path to the name); so I'm not sure how helpful that would be. I think that for now going with /sys/devices/platform/asus-nb-wmi/active_fan_curve_profiles is fine and if we hit the same situation with a 2nd driver then maybe do something under the /sys/firmware/acpi/platform_profile* namespace. Maybe something like: /sys/firmware/acpi/platform_profile_flags or something which can communicate a bunch of 0 (keyword not present) / 1 values by containing a list of space separated keywords like: "custom-fan-profiles", where "custom-fan-profiles" would only show up when they are activated ? Regards, Hans
