RE: [RFC v4 3/5] platform/x86/intel/pmc: core: Drop check_counters

[Limonciello, Mario]

[Public]



> -----Original Message-----
> From: Sven van Ashbrook <svenva@xxxxxxxxxxxx>
> Sent: Wednesday, November 23, 2022 11:47
> To: Limonciello, Mario <Mario.Limonciello@xxxxxxx>
> Cc: Rafael J . Wysocki <rafael@xxxxxxxxxx>; Rajneesh Bhardwaj
> <irenic.rajneesh@xxxxxxxxx>; David E Box <david.e.box@xxxxxxxxx>; Raul
> Rangel <rrangel@xxxxxxxxxxxx>; linux-pm@xxxxxxxxxxxxxxx; platform-
> driver-x86@xxxxxxxxxxxxxxx; Pavel Machek <pavel@xxxxxx>; Len Brown
> <len.brown@xxxxxxxxx>; John Stultz <jstultz@xxxxxxxxxx>; Thomas
> Gleixner <tglx@xxxxxxxxxxxxx>; Stephen Boyd <sboyd@xxxxxxxxxx>; S-k,
> Shyam-sundar <Shyam-sundar.S-k@xxxxxxx>; Rajat Jain
> <rajatja@xxxxxxxxxx>; Hans de Goede <hdegoede@xxxxxxxxxx>; linux-
> kernel@xxxxxxxxxxxxxxx; Mark Gross <markgross@xxxxxxxxxx>
> Subject: Re: [RFC v4 3/5] platform/x86/intel/pmc: core: Drop check_counters
> 
> Hi Mario, comments below.
> 
> On Thu, Nov 17, 2022 at 5:58 PM Mario Limonciello
> <mario.limonciello@xxxxxxx> wrote:
> >
> > `check_counters` is a stateful variable for indicating whether or
> > not to be checking if counters incremented on resume from s2idle.
> >
> > As the module already has code to gate whether to check the counters
> > that will fail the suspend when this is enabled, use that instead.
> >
> > Signed-off-by: Mario Limonciello <mario.limonciello@xxxxxxx>
> > ---
> > RFC v3->v4:
> >  * No changes
> > ---
> >  drivers/platform/x86/intel/pmc/core.c | 7 ++-----
> >  drivers/platform/x86/intel/pmc/core.h | 1 -
> >  2 files changed, 2 insertions(+), 6 deletions(-)
> >
> > diff --git a/drivers/platform/x86/intel/pmc/core.c
> b/drivers/platform/x86/intel/pmc/core.c
> > index 17ec5825d13d..adc2cae4db28 100644
> > --- a/drivers/platform/x86/intel/pmc/core.c
> > +++ b/drivers/platform/x86/intel/pmc/core.c
> > @@ -2059,8 +2059,6 @@ static __maybe_unused int
> pmc_core_suspend(struct device *dev)
> >  {
> >         struct pmc_dev *pmcdev = dev_get_drvdata(dev);
> >
> > -       pmcdev->check_counters = false;
> > -
> >         /* No warnings on S0ix failures */
> >         if (!warn_on_s0ix_failures)
> >                 return 0;
> > @@ -2077,7 +2075,6 @@ static __maybe_unused int
> pmc_core_suspend(struct device *dev)
> >         if (pmc_core_dev_state_get(pmcdev, &pmcdev->s0ix_counter))
> >                 return -EIO;
> >
> > -       pmcdev->check_counters = true;
> >         return 0;
> >  }
> >
> > @@ -2113,10 +2110,10 @@ static __maybe_unused int
> pmc_core_resume(struct device *dev)
> >         const struct pmc_bit_map **maps = pmcdev->map->lpm_sts;
> >         int offset = pmcdev->map->lpm_status_offset;
> >
> > -       if (!pmcdev->check_counters)
> > +       if (!pmc_core_is_s0ix_failed(pmcdev))
> 
> Will this break the "CPU did not enter SLP_S0!!!" warning?
> 
> As far as I can tell,
> If an Intel system uses S3 instead of S0ix, pmcdev->s0ix_counter will
> not get updated in the
> suspend callback. In the resume callback, the counter check in
> pmc_core_is_s0ix_failed()
> no longer makes any sense. It either fails all the time (if
> pmcdev->s0ix_counter was inited with a non-
> zero value) or succeeds all the time (if pmcdev->s0ix_counter was zero-
> inited).

Ah yeah;  So this patch probably doesn't make sense as is.  It would mean
either needing to check pm_suspend_via_firmware() in the resume callback
or just skipping it.  I'll drop it in the next revision of this.

> 
> >                 return 0;
> >
> > -       if (!pmc_core_is_s0ix_failed(pmcdev))
> > +       if (!warn_on_s0ix_failures)
> >                 return 0;
> >
> >         if (pmc_core_is_pc10_failed(pmcdev)) {
> > diff --git a/drivers/platform/x86/intel/pmc/core.h
> b/drivers/platform/x86/intel/pmc/core.h
> > index 7a059e02c265..5687e91e884c 100644
> > --- a/drivers/platform/x86/intel/pmc/core.h
> > +++ b/drivers/platform/x86/intel/pmc/core.h
> > @@ -316,7 +316,6 @@ struct pmc_reg_map {
> >   * @pmc_xram_read_bit: flag to indicate whether PMC XRAM shadow
> registers
> >   *                     used to read MPHY PG and PLL status are available
> >   * @mutex_lock:                mutex to complete one transcation
> > - * @check_counters:    On resume, check if counters are getting
> incremented
> >   * @pc10_counter:      PC10 residency counter
> >   * @s0ix_counter:      S0ix residency (step adjusted)
> >   * @num_lpm_modes:     Count of enabled modes
> > --
> > 2.34.1
> >

<<attachment: winmail.dat>>