* [RFT][PATCH 1/2] ACPI: EC / PM: Disable non-wakeup GPEs for suspend-to-idle
2018-12-17 11:19 [RFT][PATCH 0/2] ACPI / PM: Avoid spurious wakeups from non-wakeup GPEs in suspend-to-idle Rafael J. Wysocki
@ 2018-12-17 11:21 ` Rafael J. Wysocki
2018-12-17 11:22 ` [RFT][PATCH 2/2] ACPI: PM: Loop in full LPS0 mode only Rafael J. Wysocki
2018-12-17 16:50 ` [RFT][PATCH 0/2] ACPI / PM: Avoid spurious wakeups from non-wakeup GPEs in suspend-to-idle Mika Westerberg
2 siblings, 0 replies; 5+ messages in thread
From: Rafael J. Wysocki @ 2018-12-17 11:21 UTC (permalink / raw)
To: Linux ACPI; +Cc: Linux PM, LKML, Zhang Rui, Mika Westerberg
From: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
There are systems in which non-wakeup GPEs fire during the "noirq"
suspend stage of suspending devices and that effectively prevents the
system that tries to suspend to idle from entering any low-power
state at all. If the offending GPE fires regularly and often enough,
the system appears to be suspended, but in fact it is in a tight loop
over "noirq" suspend and "noirq" resume of devices all the time.
To prevent that from happening, disable all non-wakeup GPEs except
for the EC GPE for suspend-to-idle (the EC GPE is special, because
on some systems it has to be enabled for power button wakeup events
to be generated as expected).
Fixes: 147a7d9d25ca (ACPI / PM: Do not reconfigure GPEs for suspend-to-idle)
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
---
drivers/acpi/ec.c | 12 ++++++++++++
drivers/acpi/internal.h | 2 ++
drivers/acpi/sleep.c | 11 +++++++++++
3 files changed, 25 insertions(+)
Index: linux-pm/drivers/acpi/ec.c
===================================================================
--- linux-pm.orig/drivers/acpi/ec.c
+++ linux-pm/drivers/acpi/ec.c
@@ -1034,6 +1034,18 @@ void acpi_ec_unblock_transactions(void)
acpi_ec_start(first_ec, true);
}
+void acpi_ec_mark_gpe_for_wake(void)
+{
+ if (first_ec && !ec_no_wakeup)
+ acpi_mark_gpe_for_wake(NULL, first_ec->gpe);
+}
+
+void acpi_ec_set_gpe_wake_mask(u8 action)
+{
+ if (first_ec && !ec_no_wakeup)
+ acpi_set_gpe_wake_mask(NULL, first_ec->gpe, action);
+}
+
void acpi_ec_dispatch_gpe(void)
{
if (first_ec)
Index: linux-pm/drivers/acpi/internal.h
===================================================================
--- linux-pm.orig/drivers/acpi/internal.h
+++ linux-pm/drivers/acpi/internal.h
@@ -187,6 +187,8 @@ int acpi_ec_ecdt_probe(void);
int acpi_ec_dsdt_probe(void);
void acpi_ec_block_transactions(void);
void acpi_ec_unblock_transactions(void);
+void acpi_ec_mark_gpe_for_wake(void);
+void acpi_ec_set_gpe_wake_mask(u8 action);
void acpi_ec_dispatch_gpe(void);
int acpi_ec_add_query_handler(struct acpi_ec *ec, u8 query_bit,
acpi_handle handle, acpi_ec_query_func func,
Index: linux-pm/drivers/acpi/sleep.c
===================================================================
--- linux-pm.orig/drivers/acpi/sleep.c
+++ linux-pm/drivers/acpi/sleep.c
@@ -940,6 +940,8 @@ static int lps0_device_attach(struct acp
acpi_handle_debug(adev->handle, "_DSM function mask: 0x%x\n",
bitmask);
+
+ acpi_ec_mark_gpe_for_wake();
} else {
acpi_handle_debug(adev->handle,
"_DSM function 0 evaluation failed\n");
@@ -968,11 +970,16 @@ static int acpi_s2idle_prepare(void)
if (lps0_device_handle) {
acpi_sleep_run_lps0_dsm(ACPI_LPS0_SCREEN_OFF);
acpi_sleep_run_lps0_dsm(ACPI_LPS0_ENTRY);
+
+ acpi_ec_set_gpe_wake_mask(ACPI_GPE_ENABLE);
}
if (acpi_sci_irq_valid())
enable_irq_wake(acpi_sci_irq);
+ /* Change the configuration of GPEs to avoid spurious wakeup. */
+ acpi_enable_all_wakeup_gpes();
+ acpi_os_wait_events_complete();
return 0;
}
@@ -1017,10 +1024,14 @@ static void acpi_s2idle_sync(void)
static void acpi_s2idle_restore(void)
{
+ acpi_enable_all_runtime_gpes();
+
if (acpi_sci_irq_valid())
disable_irq_wake(acpi_sci_irq);
if (lps0_device_handle) {
+ acpi_ec_set_gpe_wake_mask(ACPI_GPE_DISABLE);
+
acpi_sleep_run_lps0_dsm(ACPI_LPS0_EXIT);
acpi_sleep_run_lps0_dsm(ACPI_LPS0_SCREEN_ON);
}
^ permalink raw reply [flat|nested] 5+ messages in thread* [RFT][PATCH 2/2] ACPI: PM: Loop in full LPS0 mode only
2018-12-17 11:19 [RFT][PATCH 0/2] ACPI / PM: Avoid spurious wakeups from non-wakeup GPEs in suspend-to-idle Rafael J. Wysocki
2018-12-17 11:21 ` [RFT][PATCH 1/2] ACPI: EC / PM: Disable non-wakeup GPEs for suspend-to-idle Rafael J. Wysocki
@ 2018-12-17 11:22 ` Rafael J. Wysocki
2018-12-17 16:50 ` [RFT][PATCH 0/2] ACPI / PM: Avoid spurious wakeups from non-wakeup GPEs in suspend-to-idle Mika Westerberg
2 siblings, 0 replies; 5+ messages in thread
From: Rafael J. Wysocki @ 2018-12-17 11:22 UTC (permalink / raw)
To: Linux ACPI; +Cc: Linux PM, LKML, Zhang Rui, Mika Westerberg
From: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
After a previous change, all non-wakeup GPEs are disabled for
suspend-to-idle unless full Low-Power S0 (LPS0) mode is in use, so
it is not necessary to do anything in acpi_s2idle_wake() unless in
full LPS0 mode, which is when lps0_device_handle is set only.
Modify the code accordingly.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
---
drivers/acpi/sleep.c | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
Index: linux-pm/drivers/acpi/sleep.c
===================================================================
--- linux-pm.orig/drivers/acpi/sleep.c
+++ linux-pm/drivers/acpi/sleep.c
@@ -985,6 +985,8 @@ static int acpi_s2idle_prepare(void)
static void acpi_s2idle_wake(void)
{
+ if (!lps0_device_handle)
+ return;
if (pm_debug_messages_on)
lpi_check_constraints();
@@ -1003,8 +1005,7 @@ static void acpi_s2idle_wake(void)
* takes too much time for EC wakeup events to survive, so look
* for them now.
*/
- if (lps0_device_handle)
- acpi_ec_dispatch_gpe();
+ acpi_ec_dispatch_gpe();
}
}
^ permalink raw reply [flat|nested] 5+ messages in thread