linux-kernel - Re: [PATCH v9 1/7] ACPI: scan: Obtain device's desired enumeration power state

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20210129212211.GK32460@paasikivi.fi.intel.com>
Date:   Fri, 29 Jan 2021 23:22:11 +0200
From:   Sakari Ailus <sakari.ailus@...ux.intel.com>
To:     "Rafael J. Wysocki" <rafael@...nel.org>
Cc:     linux-i2c <linux-i2c@...r.kernel.org>,
        Wolfram Sang <wsa@...-dreams.de>,
        ACPI Devel Maling List <linux-acpi@...r.kernel.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        "Mani, Rajmohan" <rajmohan.mani@...el.com>,
        Tomasz Figa <tfiga@...omium.org>,
        Bartosz Golaszewski <bgolaszewski@...libre.com>,
        Bingbu Cao <bingbu.cao@...el.com>,
        Chiranjeevi Rapolu <chiranjeevi.rapolu@...el.com>,
        Hyungwoo Yang <hyungwoo.yang@...el.com>,
        Linux Media Mailing List <linux-media@...r.kernel.org>
Subject: Re: [PATCH v9 1/7] ACPI: scan: Obtain device's desired enumeration
 power state

On Fri, Jan 29, 2021 at 05:57:17PM +0100, Rafael J. Wysocki wrote:
> On Fri, Jan 29, 2021 at 5:45 PM Sakari Ailus
> <sakari.ailus@...ux.intel.com> wrote:
> >
> > Hi Rafael,
> >
> > Thanks for the comments.
> >
> > On Fri, Jan 29, 2021 at 03:07:57PM +0100, Rafael J. Wysocki wrote:
> > > On Fri, Jan 29, 2021 at 12:27 AM Sakari Ailus
> > > <sakari.ailus@...ux.intel.com> wrote:
> > > >
> > > > Store a device's desired enumeration power state in struct
> > > > acpi_device_power_flags during acpi_device object's initialisation.
> > > >
> > > > Signed-off-by: Sakari Ailus <sakari.ailus@...ux.intel.com>
> > > > ---
> > > >  drivers/acpi/scan.c     | 6 ++++++
> > > >  include/acpi/acpi_bus.h | 3 ++-
> > > >  2 files changed, 8 insertions(+), 1 deletion(-)
> > > >
> > > > diff --git a/drivers/acpi/scan.c b/drivers/acpi/scan.c
> > > > index 1d7a02ee45e05..b077c645c9845 100644
> > > > --- a/drivers/acpi/scan.c
> > > > +++ b/drivers/acpi/scan.c
> > > > @@ -987,6 +987,8 @@ static void acpi_bus_init_power_state(struct acpi_device *device, int state)
> > > >
> > > >  static void acpi_bus_get_power_flags(struct acpi_device *device)
> > > >  {
> > > > +       unsigned long long pre;
> > > > +       acpi_status status;
> > > >         u32 i;
> > > >
> > > >         /* Presence of _PS0|_PR0 indicates 'power manageable' */
> > > > @@ -1008,6 +1010,10 @@ static void acpi_bus_get_power_flags(struct acpi_device *device)
> > > >         if (acpi_has_method(device->handle, "_DSW"))
> > > >                 device->power.flags.dsw_present = 1;
> > > >
> > > > +       status = acpi_evaluate_integer(device->handle, "_PRE", NULL, &pre);
> > > > +       if (ACPI_SUCCESS(status) && !pre)
> > > > +               device->power.flags.allow_low_power_probe = 1;
> > >
> > > While this is what has been discussed and thanks for taking it into
> > > account, I'm now thinking that it may be cleaner to introduce a new
> > > object to return the deepest power state of the device in which it can
> > > be enumerated, say _DSE (Device State for Enumeration) such that 4
> > > means D3cold, 3 - D3hot and so on, so the above check can be replaced
> > > with something like
> > >
> > > status = acpi_evaluate_integer(device->handle, "_PRE", NULL, &dse);
> >
> > s/_PRE/_DSE/
> >
> > ?
> 
> Yes, sorry.
> 
> >
> > > if (ACPI_FAILURE(status))
> >
> > ACPI_SUCCESS?
> 
> Yup.
> 
> > >         device->power.state_for_enumeratin = dse;
> > >
> > > And then, it is a matter of comparing ->power.state_for_enumeratin
> > > with ->power.state and putting the device into D0 if the former is
> > > shallower than the latter.
> > >
> > > What do you think?
> >
> > Sounds good. How about calling the function e.g.
> > acpi_device_resume_for_probe(), so runtime PM could be used to resume the
> > device if the function returns true?
> 
> I'd rather try to power it up before enabling runtime PM, because in
> order to do the latter properly, you need to know if the device is
> active or suspended to start with.
> 
> So you need something like (pseudo-code)
> 
> if (this_device_needs_to_be_on(ACPI_COMPANION(dev))) {
>    acpi_device_set_power(ACPI_COMPANION(dev), ACPI_STATE_D0);
>    pm_runtime_set_active(dev);
> } else {
>    pm_runtime_set_suspended(dev);

I guess the else branch isn't needed? The device remains suspended if its
state hasn't been changed.

> }
> 
> and then you can enable PM-runtime.

Yes, agreed, this is what drivers should do. The I²C framework would use
the function and conditionally power the device on before enabling runtime
PM.

This is how it's implemented by the set already but I think the change in
semantics requires a little more still.

-- 
Sakari Ailus