Hi,
I have an Intel X5-Z8350 laptop which used to work fine on s2idle
enter/exit with kernel 5.3. After upgrading to kernel 5.4 and later,
the system can still exit s2idle by power button. However, if I try to
wake it up from a keystroke, the system will freeze and then no longer
respond even to the power button. I can only shut it down and power on
again.
I tried to 'git bisect' to find out which commit causes the
difference, it shows me the commit "ACPI: PM: s2idle: Prevent spurious
SCIs from waking up the system". My laptop can work as usual to exit
s2idle by a keystroke with reverting it on kernel 5.4.
Then I tried to check whether I can reproduce it on the latest
mainline kernel, the same problem still there. But I can not fix it by
simply reverting. I found the acpi_s2idle_wake() has been refined on
the latest mainline kernel, and I have to make modifications as
follows to make my laptop work.
@@ -1024,7 +1024,7 @@ static bool acpi_s2idle_wake(void)
* regarded as a spurious one.
*/
if (!acpi_ec_dispatch_gpe())
- return false;
+ return true;
/*
* Cancel the wakeup and process all pending events in case
Maybe there's something special on my laptop so a keystroke is
considered as a spurious event on this machine? Don't know if the DSDT
helps or not.
https://gist.github.com/mschiu77/de8af6da78be12cf442853e7747f76ed
Please let me know if there's anything I can help to address this
problem. Thanks
Chris
On Mon, May 4, 2020 at 7:09 PM Chris Chiu <[email protected]> wrote:
>
> Hi,
> I have an Intel X5-Z8350 laptop which used to work fine on s2idle
> enter/exit with kernel 5.3. After upgrading to kernel 5.4 and later,
> the system can still exit s2idle by power button. However, if I try to
> wake it up from a keystroke, the system will freeze and then no longer
> respond even to the power button. I can only shut it down and power on
> again.
>
> I tried to 'git bisect' to find out which commit causes the
> difference, it shows me the commit "ACPI: PM: s2idle: Prevent spurious
> SCIs from waking up the system". My laptop can work as usual to exit
> s2idle by a keystroke with reverting it on kernel 5.4.
>
> Then I tried to check whether I can reproduce it on the latest
> mainline kernel, the same problem still there. But I can not fix it by
> simply reverting. I found the acpi_s2idle_wake() has been refined on
> the latest mainline kernel, and I have to make modifications as
> follows to make my laptop work.
Thanks for the report, the issue evidently is EC-related.
> @@ -1024,7 +1024,7 @@ static bool acpi_s2idle_wake(void)
> * regarded as a spurious one.
> */
> if (!acpi_ec_dispatch_gpe())
> - return false;
> + return true;
Have you tried commenting out simply removing the if () check and the
following return statement?
If not, can you please try this and see what happens?
>
> /*
> * Cancel the wakeup and process all pending events in case
>
On Wed, May 6, 2020 at 11:32 AM Rafael J. Wysocki <[email protected]> wrote:
>
> On Mon, May 4, 2020 at 7:09 PM Chris Chiu <[email protected]> wrote:
> >
> > Hi,
> > I have an Intel X5-Z8350 laptop which used to work fine on s2idle
> > enter/exit with kernel 5.3. After upgrading to kernel 5.4 and later,
> > the system can still exit s2idle by power button. However, if I try to
> > wake it up from a keystroke, the system will freeze and then no longer
> > respond even to the power button. I can only shut it down and power on
> > again.
> >
> > I tried to 'git bisect' to find out which commit causes the
> > difference, it shows me the commit "ACPI: PM: s2idle: Prevent spurious
> > SCIs from waking up the system". My laptop can work as usual to exit
> > s2idle by a keystroke with reverting it on kernel 5.4.
> >
> > Then I tried to check whether I can reproduce it on the latest
> > mainline kernel, the same problem still there. But I can not fix it by
> > simply reverting. I found the acpi_s2idle_wake() has been refined on
> > the latest mainline kernel, and I have to make modifications as
> > follows to make my laptop work.
>
> Thanks for the report, the issue evidently is EC-related.
>
> > @@ -1024,7 +1024,7 @@ static bool acpi_s2idle_wake(void)
> > * regarded as a spurious one.
> > */
> > if (!acpi_ec_dispatch_gpe())
> > - return false;
> > + return true;
>
> Have you tried commenting out simply removing the if () check and the
> following return statement?
Scratch that.
Instead, please try doing
acpi_ec_dispatch_gpe()
instead of the if () and the following return statement.
On Wed, May 6, 2020 at 6:19 PM Rafael J. Wysocki <[email protected]> wrote:
>
> On Wed, May 6, 2020 at 11:32 AM Rafael J. Wysocki <[email protected]> wrote:
> >
> >
> > Thanks for the report, the issue evidently is EC-related.
> >
> > > @@ -1024,7 +1024,7 @@ static bool acpi_s2idle_wake(void)
> > > * regarded as a spurious one.
> > > */
> > > if (!acpi_ec_dispatch_gpe())
> > > - return false;
> > > + return true;
> >
> > Have you tried commenting out simply removing the if () check and the
> > following return statement?
>
> Scratch that.
>
> Instead, please try doing
>
> acpi_ec_dispatch_gpe()
>
> instead of the if () and the following return statement.
Yes. I verified with the modification you suggested on my laptop. It's
working OK.
I can wake from a keystroke w/o problem.
@ -1024,8 +1024,7 @@ static bool acpi_s2idle_wake(void)
* If the EC GPE status bit has not been set, the wakeup is
* regarded as a spurious one.
*/
- if (!acpi_ec_dispatch_gpe())
- return false;
+ acpi_ec_dispatch_gpe();
/*
* Cancel the wakeup and process all pending events in case
On Thursday, May 7, 2020 5:38:11 AM CEST Chris Chiu wrote:
> On Wed, May 6, 2020 at 6:19 PM Rafael J. Wysocki <[email protected]> wrote:
> >
> > On Wed, May 6, 2020 at 11:32 AM Rafael J. Wysocki <[email protected]> wrote:
> > >
> > >
> > > Thanks for the report, the issue evidently is EC-related.
> > >
> > > > @@ -1024,7 +1024,7 @@ static bool acpi_s2idle_wake(void)
> > > > * regarded as a spurious one.
> > > > */
> > > > if (!acpi_ec_dispatch_gpe())
> > > > - return false;
> > > > + return true;
> > >
> > > Have you tried commenting out simply removing the if () check and the
> > > following return statement?
> >
> > Scratch that.
> >
> > Instead, please try doing
> >
> > acpi_ec_dispatch_gpe()
> >
> > instead of the if () and the following return statement.
>
> Yes. I verified with the modification you suggested on my laptop. It's
> working OK.
> I can wake from a keystroke w/o problem.
>
> @ -1024,8 +1024,7 @@ static bool acpi_s2idle_wake(void)
> * If the EC GPE status bit has not been set, the wakeup is
> * regarded as a spurious one.
> */
> - if (!acpi_ec_dispatch_gpe())
> - return false;
> + acpi_ec_dispatch_gpe();
>
> /*
> * Cancel the wakeup and process all pending events in case
>
OK, great, thanks for the confirmation!
Does the appended patch work for you then?
It should be functionally equivalent to the above change if I didn't mess it up.
---
drivers/acpi/ec.c | 23 ++++++++++++++---------
drivers/acpi/internal.h | 1 -
drivers/acpi/sleep.c | 14 ++------------
3 files changed, 16 insertions(+), 22 deletions(-)
Index: linux-pm/drivers/acpi/sleep.c
===================================================================
--- linux-pm.orig/drivers/acpi/sleep.c
+++ linux-pm/drivers/acpi/sleep.c
@@ -1013,21 +1013,11 @@ static bool acpi_s2idle_wake(void)
if (acpi_check_wakeup_handlers())
return true;
- /*
- * If the status bit is set for any enabled GPE other than the
- * EC one, the wakeup is regarded as a genuine one.
- */
- if (acpi_ec_other_gpes_active())
+ /* Check non-EC GPE wakeups and dispatch the EC GPE. */
+ if (acpi_ec_dispatch_gpe())
return true;
/*
- * If the EC GPE status bit has not been set, the wakeup is
- * regarded as a spurious one.
- */
- if (!acpi_ec_dispatch_gpe())
- return false;
-
- /*
* Cancel the wakeup and process all pending events in case
* there are any wakeup ones in there.
*
Index: linux-pm/drivers/acpi/ec.c
===================================================================
--- linux-pm.orig/drivers/acpi/ec.c
+++ linux-pm/drivers/acpi/ec.c
@@ -1994,23 +1994,28 @@ void acpi_ec_set_gpe_wake_mask(u8 action
acpi_set_gpe_wake_mask(NULL, first_ec->gpe, action);
}
-bool acpi_ec_other_gpes_active(void)
-{
- return acpi_any_gpe_status_set(first_ec ? first_ec->gpe : U32_MAX);
-}
-
bool acpi_ec_dispatch_gpe(void)
{
u32 ret;
if (!first_ec)
- return false;
+ return acpi_any_gpe_status_set(U32_MAX);
+ /*
+ * Report wakeup if the status bit is set for any enabled GPE other
+ * than the EC one.
+ */
+ if (acpi_any_gpe_status_set(first_ec->gpe))
+ return true;
+
+ /*
+ * Dispatch the EC GPE in-band, but do not report wakeup in any case
+ * to allow the caller to process events properly after that.
+ */
ret = acpi_dispatch_gpe(NULL, first_ec->gpe);
- if (ret == ACPI_INTERRUPT_HANDLED) {
+ if (ret == ACPI_INTERRUPT_HANDLED)
pm_pr_dbg("EC GPE dispatched\n");
- return true;
- }
+
return false;
}
#endif /* CONFIG_PM_SLEEP */
Index: linux-pm/drivers/acpi/internal.h
===================================================================
--- linux-pm.orig/drivers/acpi/internal.h
+++ linux-pm/drivers/acpi/internal.h
@@ -202,7 +202,6 @@ void acpi_ec_remove_query_handler(struct
#ifdef CONFIG_PM_SLEEP
void acpi_ec_flush_work(void);
-bool acpi_ec_other_gpes_active(void);
bool acpi_ec_dispatch_gpe(void);
#endif
On Fri, May 8, 2020 at 2:05 AM Rafael J. Wysocki <[email protected]> wrote:
>
> On Thursday, May 7, 2020 5:38:11 AM CEST Chris Chiu wrote:
> > On Wed, May 6, 2020 at 6:19 PM Rafael J. Wysocki <[email protected]> wrote:
> > >
> > > On Wed, May 6, 2020 at 11:32 AM Rafael J. Wysocki <[email protected]> wrote:
> > > >
> > > >
> > > > Thanks for the report, the issue evidently is EC-related.
> > > >
> > > > > @@ -1024,7 +1024,7 @@ static bool acpi_s2idle_wake(void)
> > > > > * regarded as a spurious one.
> > > > > */
> > > > > if (!acpi_ec_dispatch_gpe())
> > > > > - return false;
> > > > > + return true;
> > > >
> > > > Have you tried commenting out simply removing the if () check and the
> > > > following return statement?
> > >
> > > Scratch that.
> > >
> > > Instead, please try doing
> > >
> > > acpi_ec_dispatch_gpe()
> > >
> > > instead of the if () and the following return statement.
> >
> > Yes. I verified with the modification you suggested on my laptop. It's
> > working OK.
> > I can wake from a keystroke w/o problem.
> >
> > @ -1024,8 +1024,7 @@ static bool acpi_s2idle_wake(void)
> > * If the EC GPE status bit has not been set, the wakeup is
> > * regarded as a spurious one.
> > */
> > - if (!acpi_ec_dispatch_gpe())
> > - return false;
> > + acpi_ec_dispatch_gpe();
> >
> > /*
> > * Cancel the wakeup and process all pending events in case
> >
>
> OK, great, thanks for the confirmation!
>
> Does the appended patch work for you then?
>
> It should be functionally equivalent to the above change if I didn't mess it up.
>
>
> ---
> drivers/acpi/ec.c | 23 ++++++++++++++---------
> drivers/acpi/internal.h | 1 -
> drivers/acpi/sleep.c | 14 ++------------
> 3 files changed, 16 insertions(+), 22 deletions(-)
>
> Index: linux-pm/drivers/acpi/sleep.c
> ===================================================================
> --- linux-pm.orig/drivers/acpi/sleep.c
> +++ linux-pm/drivers/acpi/sleep.c
> @@ -1013,21 +1013,11 @@ static bool acpi_s2idle_wake(void)
> if (acpi_check_wakeup_handlers())
> return true;
>
> - /*
> - * If the status bit is set for any enabled GPE other than the
> - * EC one, the wakeup is regarded as a genuine one.
> - */
> - if (acpi_ec_other_gpes_active())
> + /* Check non-EC GPE wakeups and dispatch the EC GPE. */
> + if (acpi_ec_dispatch_gpe())
> return true;
>
> /*
> - * If the EC GPE status bit has not been set, the wakeup is
> - * regarded as a spurious one.
> - */
> - if (!acpi_ec_dispatch_gpe())
> - return false;
> -
> - /*
> * Cancel the wakeup and process all pending events in case
> * there are any wakeup ones in there.
> *
> Index: linux-pm/drivers/acpi/ec.c
> ===================================================================
> --- linux-pm.orig/drivers/acpi/ec.c
> +++ linux-pm/drivers/acpi/ec.c
> @@ -1994,23 +1994,28 @@ void acpi_ec_set_gpe_wake_mask(u8 action
> acpi_set_gpe_wake_mask(NULL, first_ec->gpe, action);
> }
>
> -bool acpi_ec_other_gpes_active(void)
> -{
> - return acpi_any_gpe_status_set(first_ec ? first_ec->gpe : U32_MAX);
> -}
> -
> bool acpi_ec_dispatch_gpe(void)
> {
> u32 ret;
>
> if (!first_ec)
> - return false;
> + return acpi_any_gpe_status_set(U32_MAX);
>
> + /*
> + * Report wakeup if the status bit is set for any enabled GPE other
> + * than the EC one.
> + */
> + if (acpi_any_gpe_status_set(first_ec->gpe))
> + return true;
> +
> + /*
> + * Dispatch the EC GPE in-band, but do not report wakeup in any case
> + * to allow the caller to process events properly after that.
> + */
> ret = acpi_dispatch_gpe(NULL, first_ec->gpe);
> - if (ret == ACPI_INTERRUPT_HANDLED) {
> + if (ret == ACPI_INTERRUPT_HANDLED)
> pm_pr_dbg("EC GPE dispatched\n");
> - return true;
> - }
> +
> return false;
> }
> #endif /* CONFIG_PM_SLEEP */
> Index: linux-pm/drivers/acpi/internal.h
> ===================================================================
> --- linux-pm.orig/drivers/acpi/internal.h
> +++ linux-pm/drivers/acpi/internal.h
> @@ -202,7 +202,6 @@ void acpi_ec_remove_query_handler(struct
>
> #ifdef CONFIG_PM_SLEEP
> void acpi_ec_flush_work(void);
> -bool acpi_ec_other_gpes_active(void);
> bool acpi_ec_dispatch_gpe(void);
> #endif
>
>
Ｈi Rafael,
Thanks for your patch. I tried it on my laptop and it's also
working fine. Wake up by keystroke/power button/lid open all work as
expected. Thanks.
Chris
On Friday, May 8, 2020 10:22:09 AM CEST Chris Chiu wrote:
> On Fri, May 8, 2020 at 2:05 AM Rafael J. Wysocki <[email protected]> wrote:
> >
> > On Thursday, May 7, 2020 5:38:11 AM CEST Chris Chiu wrote:
> > > On Wed, May 6, 2020 at 6:19 PM Rafael J. Wysocki <[email protected]> wrote:
> > > >
> > > > On Wed, May 6, 2020 at 11:32 AM Rafael J. Wysocki <[email protected]> wrote:
> > > > >
> > > > >
> > > > > Thanks for the report, the issue evidently is EC-related.
> > > > >
> > > > > > @@ -1024,7 +1024,7 @@ static bool acpi_s2idle_wake(void)
> > > > > > * regarded as a spurious one.
> > > > > > */
> > > > > > if (!acpi_ec_dispatch_gpe())
> > > > > > - return false;
> > > > > > + return true;
> > > > >
> > > > > Have you tried commenting out simply removing the if () check and the
> > > > > following return statement?
> > > >
> > > > Scratch that.
> > > >
> > > > Instead, please try doing
> > > >
> > > > acpi_ec_dispatch_gpe()
> > > >
> > > > instead of the if () and the following return statement.
> > >
> > > Yes. I verified with the modification you suggested on my laptop. It's
> > > working OK.
> > > I can wake from a keystroke w/o problem.
> > >
> > > @ -1024,8 +1024,7 @@ static bool acpi_s2idle_wake(void)
> > > * If the EC GPE status bit has not been set, the wakeup is
> > > * regarded as a spurious one.
> > > */
> > > - if (!acpi_ec_dispatch_gpe())
> > > - return false;
> > > + acpi_ec_dispatch_gpe();
> > >
> > > /*
> > > * Cancel the wakeup and process all pending events in case
> > >
> >
> > OK, great, thanks for the confirmation!
> >
> > Does the appended patch work for you then?
> >
> > It should be functionally equivalent to the above change if I didn't mess it up.
> >
> >
> > ---
> > drivers/acpi/ec.c | 23 ++++++++++++++---------
> > drivers/acpi/internal.h | 1 -
> > drivers/acpi/sleep.c | 14 ++------------
> > 3 files changed, 16 insertions(+), 22 deletions(-)
> >
> > Index: linux-pm/drivers/acpi/sleep.c
> > ===================================================================
> > --- linux-pm.orig/drivers/acpi/sleep.c
> > +++ linux-pm/drivers/acpi/sleep.c
> > @@ -1013,21 +1013,11 @@ static bool acpi_s2idle_wake(void)
> > if (acpi_check_wakeup_handlers())
> > return true;
> >
> > - /*
> > - * If the status bit is set for any enabled GPE other than the
> > - * EC one, the wakeup is regarded as a genuine one.
> > - */
> > - if (acpi_ec_other_gpes_active())
> > + /* Check non-EC GPE wakeups and dispatch the EC GPE. */
> > + if (acpi_ec_dispatch_gpe())
> > return true;
> >
> > /*
> > - * If the EC GPE status bit has not been set, the wakeup is
> > - * regarded as a spurious one.
> > - */
> > - if (!acpi_ec_dispatch_gpe())
> > - return false;
> > -
> > - /*
> > * Cancel the wakeup and process all pending events in case
> > * there are any wakeup ones in there.
> > *
> > Index: linux-pm/drivers/acpi/ec.c
> > ===================================================================
> > --- linux-pm.orig/drivers/acpi/ec.c
> > +++ linux-pm/drivers/acpi/ec.c
> > @@ -1994,23 +1994,28 @@ void acpi_ec_set_gpe_wake_mask(u8 action
> > acpi_set_gpe_wake_mask(NULL, first_ec->gpe, action);
> > }
> >
> > -bool acpi_ec_other_gpes_active(void)
> > -{
> > - return acpi_any_gpe_status_set(first_ec ? first_ec->gpe : U32_MAX);
> > -}
> > -
> > bool acpi_ec_dispatch_gpe(void)
> > {
> > u32 ret;
> >
> > if (!first_ec)
> > - return false;
> > + return acpi_any_gpe_status_set(U32_MAX);
> >
> > + /*
> > + * Report wakeup if the status bit is set for any enabled GPE other
> > + * than the EC one.
> > + */
> > + if (acpi_any_gpe_status_set(first_ec->gpe))
> > + return true;
> > +
> > + /*
> > + * Dispatch the EC GPE in-band, but do not report wakeup in any case
> > + * to allow the caller to process events properly after that.
> > + */
> > ret = acpi_dispatch_gpe(NULL, first_ec->gpe);
> > - if (ret == ACPI_INTERRUPT_HANDLED) {
> > + if (ret == ACPI_INTERRUPT_HANDLED)
> > pm_pr_dbg("EC GPE dispatched\n");
> > - return true;
> > - }
> > +
> > return false;
> > }
> > #endif /* CONFIG_PM_SLEEP */
> > Index: linux-pm/drivers/acpi/internal.h
> > ===================================================================
> > --- linux-pm.orig/drivers/acpi/internal.h
> > +++ linux-pm/drivers/acpi/internal.h
> > @@ -202,7 +202,6 @@ void acpi_ec_remove_query_handler(struct
> >
> > #ifdef CONFIG_PM_SLEEP
> > void acpi_ec_flush_work(void);
> > -bool acpi_ec_other_gpes_active(void);
> > bool acpi_ec_dispatch_gpe(void);
> > #endif
> >
> >
>
> Ｈi Rafael,
> Thanks for your patch. I tried it on my laptop and it's also
> working fine. Wake up by keystroke/power button/lid open all work as
> expected. Thanks.
Thanks for the confirmation!
Please also verify the final version of the patch available from
https://patchwork.kernel.org/patch/11537215/
Of course, it will only make a difference if the ec_no_wakeup switch is set
on your system (either as a result of blacklisting or via the kernel command
line).
Thanks!
On Sat, May 9, 2020 at 12:50 AM Rafael J. Wysocki <[email protected]> wrote:
>
> On Friday, May 8, 2020 10:22:09 AM CEST Chris Chiu wrote:
> > On Fri, May 8, 2020 at 2:05 AM Rafael J. Wysocki <[email protected]> wrote:
> >
> > Ｈi Rafael,
> > Thanks for your patch. I tried it on my laptop and it's also
> > working fine. Wake up by keystroke/power button/lid open all work as
> > expected. Thanks.
>
> Thanks for the confirmation!
>
> Please also verify the final version of the patch available from
>
> https://patchwork.kernel.org/patch/11537215/
>
> Of course, it will only make a difference if the ec_no_wakeup switch is set
> on your system (either as a result of blacklisting or via the kernel command
> line).
>
> Thanks!
>
>
>
Thanks. Rafael. I've tested it on my laptop and it's still working well.
Chris
On Mon, May 11, 2020 at 8:54 AM Chris Chiu <[email protected]> wrote:
>
> On Sat, May 9, 2020 at 12:50 AM Rafael J. Wysocki <[email protected]> wrote:
> >
> > On Friday, May 8, 2020 10:22:09 AM CEST Chris Chiu wrote:
> > > On Fri, May 8, 2020 at 2:05 AM Rafael J. Wysocki <[email protected]> wrote:
> > >
> > > Ｈi Rafael,
> > > Thanks for your patch. I tried it on my laptop and it's also
> > > working fine. Wake up by keystroke/power button/lid open all work as
> > > expected. Thanks.
> >
> > Thanks for the confirmation!
> >
> > Please also verify the final version of the patch available from
> >
> > https://patchwork.kernel.org/patch/11537215/
> >
> > Of course, it will only make a difference if the ec_no_wakeup switch is set
> > on your system (either as a result of blacklisting or via the kernel command
> > line).
> >
> > Thanks!
> >
> >
> >
>
> Thanks. Rafael. I've tested it on my laptop and it's still working well.
Thanks!