PM: hibernate: Avoid missing wakeup events during hibernation
Subsystems (3)
PATCH v4 1/1 PM: hibernate: Avoid missing wakeup events during hibernation
Wakeup events that occur in the hibernation process's hibernation_platform_enter() cannot wake up the system. Although the current hibernation framework will execute part of the recovery process after a wakeup event occurs, it ultimately performs a shutdown operation because the system does not check the return value of hibernation_platform_enter(). In short, if a wakeup event occurs before putting the system into the final low-power state, it will be missed.
To solve this problem, check the return value of hibernation_platform_enter(). When it returns -EAGAIN or -EBUSY (indicate the occurrence of a wakeup event), execute the hibernation recovery process, discard the previously saved image, and ultimately return to the working state.
---
[PATCH v2]:
- Optimize the "if" condition logic.
- Link to v1: https://lore.kernel.org/all/[email protected]
[PATCH v3]:
- Use pr_info instead of pr_err.
- Fix undeclared function 'swsusp_unmark' build error.
- Refine commit and printing message.
- Change the subject.
- Link to v2: https://lore.kernel.org/all/[email protected]
[PATCH v4]:
- Define an empty swsusp_unmark() function for scenarios where CONFIG_SUSPEND is not defined.
- Link to v3: https://lore.kernel.org/all/[email protected]
--- kernel/power/hibernate.c | 10 ++++++++-- kernel/power/power.h | 2 ++ 2 files changed, 10 insertions(+), 2 deletions(-)
kernel/power/hibernate.c +8 -2
@@ -642,9 +642,9 @@ int hibernation_platform_enter(void)*/ static void power_down (void ) { -#ifdef CONFIG_SUSPEND int error; +#ifdef CONFIG_SUSPEND if (hibernation_mode ==HIBERNATION_SUSPEND ) { error =suspend_devices_and_enter (mem_sleep_current);if (error) {
@@ -667,7 +667,13 @@ static void power_down(void)kernel_restart (NULL );break ;case HIBERNATION_PLATFORM : -hibernation_platform_enter (); + error =hibernation_platform_enter (); +if (error == -EAGAIN || error == -EBUSY ) { +swsusp_unmark (); + events_check_enabled = false; +pr_info ("Hibernation process aborted due to detected wakeup event.\n" ); +return ; + } fallthrough;case HIBERNATION_SHUTDOWN :if (kernel_can_power_off ())
kernel/power/power.h +2 -0
@@ -175,6 +175,8 @@ extern int swsusp_write(unsigned int flags);void swsusp_close (void );#ifdef CONFIG_SUSPEND extern int swsusp_unmark (void ); +#else +static inline int swsusp_unmark (void ) {return 0 ; }#endif struct __kernel_old_timeval ;
Signature · 1 line
2.17.0
Quoted 70 lines · Chris Feng
Wakeup events that occur in the hibernation process's
hibernation_platform_enter()cannot wake up the system. Although the current hibernation framework will execute part of the recovery process after a wakeup event occurs, it ultimately performs a shutdown operation because the system does not check the return value ofhibernation_platform_enter(). In short, if a wakeup event occurs before putting the system into the final low-power state, it will be missed.To solve this problem, check the return value of
hibernation_platform_enter(). When it returns-EAGAINor-EBUSY(indicate the occurrence of a wakeup event), execute the hibernation recovery process, discard the previously saved image, and ultimately return to the working state.Signed-off-by Chris Feng <[email protected]>---
[PATCH v2]:
- Optimize the "if" condition logic.
- Link to v1: https://lore.kernel.org/all/[email protected]
[PATCH v3]:
- Usepr_infoinstead ofpr_err.
- Fix undeclared function 'swsusp_unmark' build error.
- Refine commit and printing message.
- Change the subject.
- Link to v2: https://lore.kernel.org/all/[email protected]
[PATCH v4]:
- Define an emptyswsusp_unmark()function for scenarios whereCONFIG_SUSPENDis not defined.
- Link to v3: https://lore.kernel.org/all/[email protected]--- kernel/power/hibernate.c | 10 ++++++++-- kernel/power/power.h | 2 ++ 2 files changed, 10 insertions(+), 2 deletions(-)kernel/power/hibernate.c +8
-2@@ -642,9 +642,9 @@ int hibernation_platform_enter(void)*/ static void power_down (void ) { -#ifdef CONFIG_SUSPEND int error; +#ifdef CONFIG_SUSPEND if (hibernation_mode ==HIBERNATION_SUSPEND ) { error =suspend_devices_and_enter (mem_sleep_current);if (error) {@@ -667,7 +667,13 @@ static void power_down(void)kernel_restart (NULL );break ;case HIBERNATION_PLATFORM : -hibernation_platform_enter (); + error =hibernation_platform_enter (); +if (error == -EAGAIN || error == -EBUSY ) { +swsusp_unmark (); + events_check_enabled = false; +pr_info ("Hibernation process aborted due to detected wakeup event.\n" ); +return ; + } fallthrough;case HIBERNATION_SHUTDOWN :if (kernel_can_power_off ())kernel/power/power.h +0
-0
kernel/power/power.h extern int swsusp_write(unsigned int flags);
Quoted 9 lines
kernel/power/power.h +2
-1@@ -175,6 +175,8 @@ extern int swsusp_write(unsigned int flags);void swsusp_close (void );#ifdef CONFIG_SUSPEND extern int swsusp_unmark (void ); +#else +static inline int swsusp_unmark (void ) {return 0 ; }#endif struct __kernel_old_timeval ; --
Applied as 6.8 material, but I have rephrased the pr_info() message added by this patch.
Thanks!
[REGRESSION][BISECTED] "Wakeup event detected during hibernation" on Dell Precision 7780 (was "[PATCH v4] PM: hibernate: Avoid missing wakeup events during hibernation")
In reply to Chris FengChris Feng <[email protected]>:
Quoted 8 lines
Wakeup events that occur in the hibernation process's
hibernation_platform_enter()cannot wake up the system. Although the current hibernation framework will execute part of the recovery process after a wakeup event occurs, it ultimately performs a shutdown operation because the system does not check the return value ofhibernation_platform_enter(). In short, if a wakeup event occurs before putting the system into the final low-power state, it will be missed.
Quoted 5 lines
To solve this problem, check the return value of
hibernation_platform_enter(). When it returns-EAGAINor-EBUSY(indicate the occurrence of a wakeup event), execute the hibernation recovery process, discard the previously saved image, and ultimately return to the working state.
#regzbot introduced: 0c4cae1bc00d31c78858c184ede351baea232bdb
Hibernation doesn't work on my laptop.
My laptop is Dell Precision 7780.
Hibernation starts, then aborts. dmesg contains:
[ 28.283320] PM: hibernation: Wakeup event detected during hibernation, rolling back.
I did bisect. The bug reproduces starting with 0c4cae1bc00d.
The bug still reproduces on v6.18-rc2.
Note: there is another problem with PM on my laptop, which is tracked here:
https://lore.kernel.org/all/[email protected]/ .
So I always revert 1796f808e4bb in my kernels.
But this doesn't prevent this hibernation bug.
In other words, this hibernation bug still happens on v6.18-rc2 even with 1796f808e4bb reverted.
Steps to reproduce:
1. Power off laptop, then power on
2. Hibernate ("sudo systemctl hibernate")
Hibernate will not work.
Note that to reproduce the bug you need power off laptop, and then power on it in step 1. Merely reboot doesn't cause bug. I. e. reproducibilty depends on whether last boot was reboot or cold power on.
Also: suspend works normally (assuming I reverted 1796f808e4bb), but hibernate doesn't.
This is "dmesg --level=debug+" on v6.18-rc2-with-1796f808e4bb-reverted:
https://zerobin.net/?0459f6411446622d#8i0Ifo6o68By3+UlYUr2t2KL7YLXsKEXrkfszpE77Rw=
This is config of this kernel:
https://zerobin.net/?04e89ceab8284c1d#9bpaZKqVXFSaeav/WW6GOCwD4i3SozzQ8pBEJ6LWwVM=
Signature · 1 line
Askar Safin
Re: [REGRESSION][BISECTED] "Wakeup event detected during hibernation" on Dell Precision 7780 (was "[PATCH v4] PM: hibernate: Avoid missing wakeup events during hibernation")
Quoted 52 lines · Askar Safin
Chris Feng <[email protected]>:
Quoted 8 lines
Wakeup events that occur in the hibernation process's
hibernation_platform_enter()cannot wake up the system. Although the current hibernation framework will execute part of the recovery process after a wakeup event occurs, it ultimately performs a shutdown operation because the system does not check the return value ofhibernation_platform_enter(). In short, if a wakeup event occurs before putting the system into the final low-power state, it will be missed.Quoted 5 lines
To solve this problem, check the return value of
hibernation_platform_enter(). When it returns-EAGAINor-EBUSY(indicate the occurrence of a wakeup event), execute the hibernation recovery process, discard the previously saved image, and ultimately return to the working state.#regzbot introduced:
0c4cae1bc00d31c78858c184ede351baea232bdbHibernation doesn't work on my laptop.
My laptop is Dell Precision 7780.
Hibernation starts, then aborts. dmesg contains:
[ 28.283320] PM: hibernation: Wakeup event detected during hibernation, rolling back.I did bisect. The bug reproduces starting with
0c4cae1bc00d.The bug still reproduces on v6.18-rc2.
Note: there is another problem with PM on my laptop, which is tracked here:
https://lore.kernel.org/all/[email protected]/ .So I always revert
1796f808e4bbin my kernels.But this doesn't prevent this hibernation bug.
In other words, this hibernation bug still happens on v6.18-rc2 even with
1796f808e4bbreverted.Steps to reproduce:
1. Power off laptop, then power on
2. Hibernate ("sudo systemctl hibernate")Hibernate will not work.
Note that to reproduce the bug you need power off laptop, and then power on it in step 1. Merely reboot doesn't cause bug. I. e. reproducibilty depends on whether last boot was reboot or cold power on.
Also: suspend works normally (assuming I reverted
1796f808e4bb), but hibernate doesn't.
Quoted 8 lines
This is "dmesg --level=debug+" on v6.18-rc2-with-1796f808e4bb-reverted:
https://zerobin.net/?0459f6411446622d#8i0Ifo6o68By3+UlYUr2t2KL7YLXsKEXrkfszpE77Rw=
This is config of this kernel:
https://zerobin.net/?04e89ceab8284c1d#9bpaZKqVXFSaeav/WW6GOCwD4i3SozzQ8pBEJ6LWwVM=
To me it sounds like it's potentially the same root cause for the early suspend wakeup as well as the hibernate wakeup.
Assuming the GPIO pointed out in that other thread (https://lore.kernel.org/all/[email protected]/) is indeed the touchpad GPIO I had a suspicion. Are you closing the lid and putting the laptop in a bag or putting anything on top of it?
I wonder if the touchpad is not disabled by the hardware and getting pressure through the lid and physically clicking/activating.
I've seen this exact issue occur on a Framework 16 as well. The workaround is to disable wakeup from touchpad before suspend/hibernate.
Re: [REGRESSION][BISECTED] "Wakeup event detected during hibernation" on Dell Precision 7780 (was "[PATCH v4] PM: hibernate: Avoid missing wakeup events during hibernation")
Quoted 2 lines · Mario Limonciello
To me it sounds like it's potentially the same root cause for the early suspend wakeup as well as the hibernate wakeup.
Adding "gpiolib_acpi.ignore_wake=VEN_0488:00@355" or "gpiolib_acpi.ignore_wake=INTC1085:00@355" doesn't fix this bug.
Quoted 2 lines
Are you closing the lid
and putting the laptop in a bag or putting anything on top of it?
No. I just click "Hibernate" button using touchpad in GUI, and hibernate fails.
I just did some printf-debugging. Results:
We call
https://elixir.bootlin.com/linux/v6.18/source/kernel/power/hibernate.c#L870 , it calls
https://elixir.bootlin.com/linux/v6.18/source/kernel/power/hibernate.c#L723 , it calls
https://elixir.bootlin.com/linux/v6.18/source/kernel/power/hibernate.c#L636 , it calls dpm_suspend
https://elixir.bootlin.com/linux/v6.18/source/drivers/base/power/main.c#L2265 , that dpm_suspend returns error, and this error leads to
"Wakeup event detected during hibernation" message.
Any further pointers? How to debug?
Signature · 1 line
Askar Safin
Re: [REGRESSION][BISECTED] "Wakeup event detected during hibernation" on Dell Precision 7780 (was "[PATCH v4] PM: hibernate: Avoid missing wakeup events during hibernation")
Quoted 5 lines · Askar Safin
Quoted 2 lines · Mario Limonciello
To me it sounds like it's potentially the same root cause for the early suspend wakeup as well as the hibernate wakeup.
Quoted 9 lines
Adding "gpiolib_acpi.ignore_wake=VEN_0488:00@355" or "gpiolib_acpi.ignore_wake=INTC1085:00@355" doesn't fix this bug.
Quoted 2 lines
Are you closing the lid
and putting the laptop in a bag or putting anything on top of it?No. I just click "Hibernate" button using touchpad in GUI, and hibernate fails.
Can you try using command line instead in case it's actually the touchpad click still being registered?
Quoted 13 lines
I just did some printf-debugging. Results:
We call
https://elixir.bootlin.com/linux/v6.18/source/kernel/power/hibernate.c#L870 , it calls
https://elixir.bootlin.com/linux/v6.18/source/kernel/power/hibernate.c#L723 , it calls
https://elixir.bootlin.com/linux/v6.18/source/kernel/power/hibernate.c#L636 , it callsdpm_suspend
https://elixir.bootlin.com/linux/v6.18/source/drivers/base/power/main.c#L2265 , thatdpm_suspendreturns error, and this error leads to
"Wakeup event detected during hibernation" message.
Quoted 2 lines
Any further pointers? How to debug?
If you haven't already; turn on /sys/power/pm_debug_messages and /sys/power/pm_print_times.
This should help clarify the last functions that are run and any debug level PM related messages that are emitted that lead to the failure.
You can also check /sys/power/pm_wakeup_irq after the failure to confirm which interrupt woke the system (assuming that was the source of the early wakeup event).
Re: [REGRESSION][BISECTED] "Wakeup event detected during hibernation" on Dell Precision 7780 (was "[PATCH v4] PM: hibernate: Avoid missing wakeup events during hibernation")
Mario Limonciello <[email protected]>:
Quoted 2 lines
Can you try using command line instead in case it's actually the touchpad click still being registered?
Hi. I found another hibernation bug and I want to fix it first, because I know how to fix it:
https://lore.kernel.org/all/[email protected]/
But I need help to proceed writing patch.
Signature · 1 line
Askar Safin
Re: [REGRESSION][BISECTED] "Wakeup event detected during hibernation" on Dell Precision 7780 (was "[PATCH v4] PM: hibernate: Avoid missing wakeup events during hibernation")
In reply to Askar SafinAskar Safin <[email protected]>:
Quoted 3 lines
This is "dmesg --level=debug+" on v6.18-rc2-with-1796f808e4bb-reverted:
https://zerobin.net/?0459f6411446622d#8i0Ifo6o68By3+UlYUr2t2KL7YLXsKEXrkfszpE77Rw=
This log contains WARNING. It is caused by unrelated bug, which is reproducible both on real hardware and in Qemu. I reported it here:
https://lore.kernel.org/regressions/[email protected]/
Signature · 1 line
Askar Safin