mbox series

[0/4] ACPI/ACPICA: Only flush caches on S1/S2/S3 and C3

Message ID 20211206122952.74139-1-kirill.shutemov@linux.intel.com
Headers show
Series ACPI/ACPICA: Only flush caches on S1/S2/S3 and C3 | expand

Message

Kirill A. Shutemov Dec. 6, 2021, 12:29 p.m. UTC 
Does it look like you want?

Kirill A. Shutemov (4):
  ACPICA: Do not flush cache for on entering S4 and S5
  ACPI: PM: Remove redundant cache flushing
  ACPI: processor idle: Only flush cache on entering C3
  ACPI: PM: Avoid cache flush on entering S4

 drivers/acpi/acpica/hwesleep.c  | 3 ++-
 drivers/acpi/acpica/hwsleep.c   | 3 ++-
 drivers/acpi/acpica/hwxfsleep.c | 2 --
 drivers/acpi/processor_idle.c   | 3 ++-
 drivers/acpi/sleep.c            | 9 +++------
 5 files changed, 9 insertions(+), 11 deletions(-)

Comments

Peter Zijlstra Dec. 6, 2021, 3:03 p.m. UTC | #1 
On Mon, Dec 06, 2021 at 03:29:51PM +0300, Kirill A. Shutemov wrote:
> According to the ACPI spec v6.4, section 8.2, cache flushing required
> on entering C3 power state.
> 
> Avoid flushing cache on entering other power states.
> 
> Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
> ---
>  drivers/acpi/processor_idle.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/acpi/processor_idle.c b/drivers/acpi/processor_idle.c
> index 76ef1bcc8848..01495aca850e 100644
> --- a/drivers/acpi/processor_idle.c
> +++ b/drivers/acpi/processor_idle.c
> @@ -567,7 +567,8 @@ static int acpi_idle_play_dead(struct cpuidle_device *dev, int index)
>  {
>  	struct acpi_processor_cx *cx = per_cpu(acpi_cstate[index], dev->cpu);
>  
> -	ACPI_FLUSH_CPU_CACHE();
> +	if (cx->type == ACPI_STATE_C3)
> +		ACPI_FLUSH_CPU_CACHE();
>  

acpi_idle_enter() already does this, acpi_idle_enter_s2idle() has it
confused again,

Also, I think acpi_idle_enter() does it too late; consider
acpi_idle_enter_mb(). Either that or the BM crud needs more comments.
Rafael J. Wysocki Dec. 7, 2021, 4:35 p.m. UTC | #2 
On Mon, Dec 6, 2021 at 1:30 PM Kirill A. Shutemov
<kirill.shutemov@linux.intel.com> wrote:
>
> ACPICA code takes care about cache flushing on S1/S2/S3 in
> acpi_hw_extended_sleep() and acpi_hw_legacy_sleep().
>
> acpi_suspend_enter() calls into ACPICA code via acpi_enter_sleep_state()
> for S1 or x86_acpi_suspend_lowlevel() for S3. It only need to flush
> cache for S2 (not sure if this call path is ever used for S2).
>
> acpi_sleep_prepare() call tree:
>   __acpi_pm_prepare()
>     acpi_pm_prepare()
>       acpi_suspend_ops::prepare_late()
>       acpi_hibernation_ops::pre_snapshot()
>       acpi_hibernation_ops::prepare()
>     acpi_suspend_begin_old()
>       acpi_suspend_begin_old::begin()
>   acpi_hibernation_begin_old()
>     acpi_hibernation_ops_old::acpi_hibernation_begin_old()
>   acpi_power_off_prepare()
>     pm_power_off_prepare()
>
> Hibernation (S4) and Power Off (S5) don't require cache flushing. So,
> the only interesting callsites are acpi_suspend_ops::prepare_late() and
> acpi_suspend_begin_old::begin(). Both of them have cache flush on
> ->enter() operation in acpi_suspend_enter().
>
> Remove redundant ACPI_FLUSH_CPU_CACHE() in acpi_sleep_prepare() and
> acpi_suspend_enter().
>
> Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
> ---
>  drivers/acpi/sleep.c | 7 +++----
>  1 file changed, 3 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/acpi/sleep.c b/drivers/acpi/sleep.c
> index eaa47753b758..14e8df0ac762 100644
> --- a/drivers/acpi/sleep.c
> +++ b/drivers/acpi/sleep.c
> @@ -73,7 +73,6 @@ static int acpi_sleep_prepare(u32 acpi_state)
>                 acpi_set_waking_vector(acpi_wakeup_address);
>
>         }
> -       ACPI_FLUSH_CPU_CACHE();
>  #endif
>         pr_info("Preparing to enter system sleep state S%d\n", acpi_state);
>         acpi_enable_wakeup_devices(acpi_state);
> @@ -566,15 +565,15 @@ static int acpi_suspend_enter(suspend_state_t pm_state)
>         u32 acpi_state = acpi_target_sleep_state;
>         int error;
>
> -       ACPI_FLUSH_CPU_CACHE();
> -
>         trace_suspend_resume(TPS("acpi_suspend"), acpi_state, true);
>         switch (acpi_state) {
>         case ACPI_STATE_S1:
>                 barrier();
>                 status = acpi_enter_sleep_state(acpi_state);
>                 break;
> -
> +       case ACPI_STATE_S2:
> +               ACPI_FLUSH_CPU_CACHE();
> +               break;

I don't think this is needed for S2, because the function doesn't do
anything low-level in that case and simply returns (IOW, S2 isn't
really supported).

>         case ACPI_STATE_S3:
>                 if (!acpi_suspend_lowlevel)
>                         return -ENOSYS;
> --
> 2.32.0
>
Rafael J. Wysocki Dec. 8, 2021, 4:26 p.m. UTC | #3 
On Mon, Dec 6, 2021 at 4:03 PM Peter Zijlstra <peterz@infradead.org> wrote:
>
> On Mon, Dec 06, 2021 at 03:29:51PM +0300, Kirill A. Shutemov wrote:
> > According to the ACPI spec v6.4, section 8.2, cache flushing required
> > on entering C3 power state.
> >
> > Avoid flushing cache on entering other power states.
> >
> > Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
> > ---
> >  drivers/acpi/processor_idle.c | 3 ++-
> >  1 file changed, 2 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/acpi/processor_idle.c b/drivers/acpi/processor_idle.c
> > index 76ef1bcc8848..01495aca850e 100644
> > --- a/drivers/acpi/processor_idle.c
> > +++ b/drivers/acpi/processor_idle.c
> > @@ -567,7 +567,8 @@ static int acpi_idle_play_dead(struct cpuidle_device *dev, int index)
> >  {
> >       struct acpi_processor_cx *cx = per_cpu(acpi_cstate[index], dev->cpu);
> >
> > -     ACPI_FLUSH_CPU_CACHE();
> > +     if (cx->type == ACPI_STATE_C3)
> > +             ACPI_FLUSH_CPU_CACHE();
> >
>
> acpi_idle_enter() already does this, acpi_idle_enter_s2idle() has it
> confused again,

No, they do the same thing: acpi_idle_enter_bm() if flags.bm_check is set.

> Also, I think acpi_idle_enter() does it too late; consider
> acpi_idle_enter_mb(). Either that or the BM crud needs more comments.

I think the latter.

Evidently, acpi_idle_play_dead(() doesn't support FFH and the BM
thing, so it is only necessary to flush the cache when using
ACPI_CSTATE_SYSTEMIO and when cx->type is C3.
Kirill A. Shutemov Dec. 9, 2021, 1:32 p.m. UTC | #4 
On Tue, Dec 07, 2021 at 05:35:38PM +0100, Rafael J. Wysocki wrote:
> I don't think this is needed for S2, because the function doesn't do
> anything low-level in that case and simply returns (IOW, S2 isn't
> really supported).

Updated patch is below. Does it look good?
Kirill A. Shutemov Dec. 9, 2021, 1:33 p.m. UTC | #5 
On Wed, Dec 08, 2021 at 05:26:12PM +0100, Rafael J. Wysocki wrote:
> On Mon, Dec 6, 2021 at 4:03 PM Peter Zijlstra <peterz@infradead.org> wrote:
> >
> > On Mon, Dec 06, 2021 at 03:29:51PM +0300, Kirill A. Shutemov wrote:
> > > According to the ACPI spec v6.4, section 8.2, cache flushing required
> > > on entering C3 power state.
> > >
> > > Avoid flushing cache on entering other power states.
> > >
> > > Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
> > > ---
> > >  drivers/acpi/processor_idle.c | 3 ++-
> > >  1 file changed, 2 insertions(+), 1 deletion(-)
> > >
> > > diff --git a/drivers/acpi/processor_idle.c b/drivers/acpi/processor_idle.c
> > > index 76ef1bcc8848..01495aca850e 100644
> > > --- a/drivers/acpi/processor_idle.c
> > > +++ b/drivers/acpi/processor_idle.c
> > > @@ -567,7 +567,8 @@ static int acpi_idle_play_dead(struct cpuidle_device *dev, int index)
> > >  {
> > >       struct acpi_processor_cx *cx = per_cpu(acpi_cstate[index], dev->cpu);
> > >
> > > -     ACPI_FLUSH_CPU_CACHE();
> > > +     if (cx->type == ACPI_STATE_C3)
> > > +             ACPI_FLUSH_CPU_CACHE();
> > >
> >
> > acpi_idle_enter() already does this, acpi_idle_enter_s2idle() has it
> > confused again,
> 
> No, they do the same thing: acpi_idle_enter_bm() if flags.bm_check is set.
> 
> > Also, I think acpi_idle_enter() does it too late; consider
> > acpi_idle_enter_mb(). Either that or the BM crud needs more comments.
> 
> I think the latter.
> 
> Evidently, acpi_idle_play_dead(() doesn't support FFH and the BM
> thing, so it is only necessary to flush the cache when using
> ACPI_CSTATE_SYSTEMIO and when cx->type is C3.

I'm new to this and not completely follow what I need to change.

Does it look correct?
Rafael J. Wysocki Dec. 17, 2021, 5:58 p.m. UTC | #6 
On Thu, Dec 9, 2021 at 2:33 PM Kirill A. Shutemov
<kirill.shutemov@linux.intel.com> wrote:
>
> On Wed, Dec 08, 2021 at 05:26:12PM +0100, Rafael J. Wysocki wrote:
> > On Mon, Dec 6, 2021 at 4:03 PM Peter Zijlstra <peterz@infradead.org> wrote:
> > >
> > > On Mon, Dec 06, 2021 at 03:29:51PM +0300, Kirill A. Shutemov wrote:
> > > > According to the ACPI spec v6.4, section 8.2, cache flushing required
> > > > on entering C3 power state.
> > > >
> > > > Avoid flushing cache on entering other power states.
> > > >
> > > > Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
> > > > ---
> > > >  drivers/acpi/processor_idle.c | 3 ++-
> > > >  1 file changed, 2 insertions(+), 1 deletion(-)
> > > >
> > > > diff --git a/drivers/acpi/processor_idle.c b/drivers/acpi/processor_idle.c
> > > > index 76ef1bcc8848..01495aca850e 100644
> > > > --- a/drivers/acpi/processor_idle.c
> > > > +++ b/drivers/acpi/processor_idle.c
> > > > @@ -567,7 +567,8 @@ static int acpi_idle_play_dead(struct cpuidle_device *dev, int index)
> > > >  {
> > > >       struct acpi_processor_cx *cx = per_cpu(acpi_cstate[index], dev->cpu);
> > > >
> > > > -     ACPI_FLUSH_CPU_CACHE();
> > > > +     if (cx->type == ACPI_STATE_C3)
> > > > +             ACPI_FLUSH_CPU_CACHE();
> > > >
> > >
> > > acpi_idle_enter() already does this, acpi_idle_enter_s2idle() has it
> > > confused again,
> >
> > No, they do the same thing: acpi_idle_enter_bm() if flags.bm_check is set.
> >
> > > Also, I think acpi_idle_enter() does it too late; consider
> > > acpi_idle_enter_mb(). Either that or the BM crud needs more comments.
> >
> > I think the latter.
> >
> > Evidently, acpi_idle_play_dead(() doesn't support FFH and the BM
> > thing, so it is only necessary to flush the cache when using
> > ACPI_CSTATE_SYSTEMIO and when cx->type is C3.
>
> I'm new to this and not completely follow what I need to change.
>
> Does it look correct?

It does, but I liked the original one more and so that one has been
applied as 5.17 material (with some edits in the changelog).

Thanks!

> From 3c544bc95a16d6a23dcb0aa50ee905d5e97c9ce5 Mon Sep 17 00:00:00 2001
> From: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
> Date: Thu, 9 Dec 2021 16:24:44 +0300
> Subject: [PATCH] ACPI: processor idle: Only flush cache on entering C3
>
> According to the ACPI spec v6.4, section 8.2, cache flushing required
> on entering C3 power state.
>
> Avoid flushing cache on entering other power states.
>
> Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
> ---
>  drivers/acpi/processor_idle.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/acpi/processor_idle.c b/drivers/acpi/processor_idle.c
> index 76ef1bcc8848..d2a4d4446eff 100644
> --- a/drivers/acpi/processor_idle.c
> +++ b/drivers/acpi/processor_idle.c
> @@ -567,7 +567,9 @@ static int acpi_idle_play_dead(struct cpuidle_device *dev, int index)
>  {
>         struct acpi_processor_cx *cx = per_cpu(acpi_cstate[index], dev->cpu);
>
> -       ACPI_FLUSH_CPU_CACHE();
> +       if (cx->entry_method == ACPI_CSTATE_SYSTEMIO &&
> +           cx->type == ACPI_STATE_C3)
> +               ACPI_FLUSH_CPU_CACHE();
>
>         while (1) {
>
> --
>  Kirill A. Shutemov
Rafael J. Wysocki Dec. 17, 2021, 6:04 p.m. UTC | #7 
On Thu, Dec 9, 2021 at 2:32 PM Kirill A. Shutemov
<kirill.shutemov@linux.intel.com> wrote:
>
> On Tue, Dec 07, 2021 at 05:35:38PM +0100, Rafael J. Wysocki wrote:
> > I don't think this is needed for S2, because the function doesn't do
> > anything low-level in that case and simply returns (IOW, S2 isn't
> > really supported).
>
> Updated patch is below. Does it look good?

It does, and so applied as 5.17 material with some minor edits in the changelog.

Thanks!

> From 5eb4ec7d8dd463ba186b779dcef2a802d999c59c Mon Sep 17 00:00:00 2001
> From: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
> Date: Thu, 9 Dec 2021 16:08:02 +0300
> Subject: [PATCH 1/2] ACPI: PM: Remove redundant cache flushing
>
> ACPICA code takes care about cache flushing on S1/S2/S3 in
> acpi_hw_extended_sleep() and acpi_hw_legacy_sleep().
>
> acpi_suspend_enter() calls into ACPICA code via acpi_enter_sleep_state()
> for S1 or x86_acpi_suspend_lowlevel() for S3.
>
> acpi_sleep_prepare() call tree:
>   __acpi_pm_prepare()
>     acpi_pm_prepare()
>       acpi_suspend_ops::prepare_late()
>       acpi_hibernation_ops::pre_snapshot()
>       acpi_hibernation_ops::prepare()
>     acpi_suspend_begin_old()
>       acpi_suspend_begin_old::begin()
>   acpi_hibernation_begin_old()
>     acpi_hibernation_ops_old::acpi_hibernation_begin_old()
>   acpi_power_off_prepare()
>     pm_power_off_prepare()
>
> Hibernation (S4) and Power Off (S5) don't require cache flushing. So,
> the only interesting callsites are acpi_suspend_ops::prepare_late() and
> acpi_suspend_begin_old::begin(). Both of them have cache flush on
> ->enter() operation in acpi_suspend_enter().
>
> Remove redundant ACPI_FLUSH_CPU_CACHE() in acpi_sleep_prepare() and
> acpi_suspend_enter().
>
> Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
> ---
>  drivers/acpi/sleep.c | 3 ---
>  1 file changed, 3 deletions(-)
>
> diff --git a/drivers/acpi/sleep.c b/drivers/acpi/sleep.c
> index eaa47753b758..5ca6c223ba3d 100644
> --- a/drivers/acpi/sleep.c
> +++ b/drivers/acpi/sleep.c
> @@ -73,7 +73,6 @@ static int acpi_sleep_prepare(u32 acpi_state)
>                 acpi_set_waking_vector(acpi_wakeup_address);
>
>         }
> -       ACPI_FLUSH_CPU_CACHE();
>  #endif
>         pr_info("Preparing to enter system sleep state S%d\n", acpi_state);
>         acpi_enable_wakeup_devices(acpi_state);
> @@ -566,8 +565,6 @@ static int acpi_suspend_enter(suspend_state_t pm_state)
>         u32 acpi_state = acpi_target_sleep_state;
>         int error;
>
> -       ACPI_FLUSH_CPU_CACHE();
> -
>         trace_suspend_resume(TPS("acpi_suspend"), acpi_state, true);
>         switch (acpi_state) {
>         case ACPI_STATE_S1:
> --
>  Kirill A. Shutemov