[PATCH 0/3] Fixes for hybrid sleep

[PATCH 0/3] Fixes for hybrid sleep

[Mario Limonciello (AMD)]

Ionut Nechita reported recently a hibernate failure, but in debugging
the issue it's actually not a hibernate failure; but a hybrid sleep
failure.

Multiple changes related to the change of when swap is disabled in
the suspend sequence contribute to the failure.  See the individual
patches for details.

Link: https://gitlab.freedesktop.org/drm/amd/-/issues/4573

NOTE: I realize this is super late in the cycle, so sorry about that,
but I debugged it as fast as I could as soon as I heard about it.
If it needs to push out to the next cycle it is what it is.

As it touches two subsystems it either needs to go through linux-pm
or drm.  I would suggest linux-pm with an Ack from Alex on the 3rd
patch.

Cc: Ionut Nechita <ionut_n2001@yahoo.com>
Mario Limonciello (3):
  PM: hibernate: Fix hybrid-sleep
  PM: hibernate: Add pm_hibernation_mode_is_suspend()
  drm/amd: Fix hybrid sleep

 drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c |  2 +-
 include/linux/suspend.h                 |  2 ++
 kernel/power/hibernate.c                | 13 ++++++++++++-
 3 files changed, 15 insertions(+), 2 deletions(-)

-- 
2.51.0

[PATCH 1/3] PM: hibernate: Fix hybrid-sleep

[Mario Limonciello (AMD)]

Hybrid sleep will hibernate the system followed by running through
the suspend routine.  Since both the hibernate and the suspend routine
will call pm_restrict_gfp_mask(), pm_restore_gfp_mask() must be called
before starting the suspend sequence.

Add an explicit call to pm_restore_gfp_mask() to power_down() before
the suspend sequence starts. Don't call pm_restore_gfp_mask() when
exiting suspend sequence it is already called:

```
power_down()
->suspend_devices_and_enter()
-->dpm_resume_end()
```

Reported-by: Ionut Nechita <ionut_n2001@yahoo.com>
Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4573
Tested-by: Ionut Nechita <ionut_n2001@yahoo.com>
Fixes: 12ffc3b1513eb ("PM: Restrict swap use to later in the suspend sequence")
Signed-off-by: Mario Limonciello (AMD) <superm1@kernel.org>
---
 kernel/power/hibernate.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/kernel/power/hibernate.c b/kernel/power/hibernate.c
index 2f66ab453823..1defe8081e64 100644
--- a/kernel/power/hibernate.c
+++ b/kernel/power/hibernate.c
@@ -695,6 +695,7 @@ static void power_down(void)
 
 #ifdef CONFIG_SUSPEND
 	if (hibernation_mode == HIBERNATION_SUSPEND) {
+		pm_restore_gfp_mask();
 		error = suspend_devices_and_enter(mem_sleep_current);
 		if (error) {
 			hibernation_mode = hibernation_ops ?
@@ -862,7 +863,8 @@ int hibernate(void)
 				power_down();
 		}
 		in_suspend = 0;
-		pm_restore_gfp_mask();
+		if (hibernation_mode != HIBERNATION_SUSPEND)
+			pm_restore_gfp_mask();
 	} else {
 		pm_pr_dbg("Hibernation image restored successfully.\n");
 	}
-- 
2.51.0

Re: [PATCH 1/3] PM: hibernate: Fix hybrid-sleep

[kernel test robot]

Hi Mario,

kernel test robot noticed the following build errors:

[auto build test ERROR on amd-pstate/linux-next]
[also build test ERROR on amd-pstate/bleeding-edge linus/master v6.17-rc7 next-20250924]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]

url:    https://github.com/intel-lab-lkp/linux/commits/Mario-Limonciello-AMD/PM-hibernate-Fix-hybrid-sleep/20250925-045432
base:   https://git.kernel.org/pub/scm/linux/kernel/git/superm1/linux.git linux-next
patch link:    https://lore.kernel.org/r/20250924205211.1059571-2-superm1%40kernel.org
patch subject: [PATCH 1/3] PM: hibernate: Fix hybrid-sleep
config: i386-randconfig-014-20250925 (https://download.01.org/0day-ci/archive/20250925/202509252323.KEdz98a4-lkp@intel.com/config)
compiler: clang version 20.1.8 (https://github.com/llvm/llvm-project 87f0227cb60147a26a1eeb4fb06e3b505e9c7261)
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20250925/202509252323.KEdz98a4-lkp@intel.com/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202509252323.KEdz98a4-lkp@intel.com/

All errors (new ones prefixed by >>):

>> kernel/power/hibernate.c:866:27: error: use of undeclared identifier 'HIBERNATION_SUSPEND'
     866 |                 if (hibernation_mode != HIBERNATION_SUSPEND)
         |                                         ^
   1 error generated.


vim +/HIBERNATION_SUSPEND +866 kernel/power/hibernate.c

   776	
   777	/**
   778	 * hibernate - Carry out system hibernation, including saving the image.
   779	 */
   780	int hibernate(void)
   781	{
   782		bool snapshot_test = false;
   783		unsigned int sleep_flags;
   784		int error;
   785	
   786		if (!hibernation_available()) {
   787			pm_pr_dbg("Hibernation not available.\n");
   788			return -EPERM;
   789		}
   790	
   791		/*
   792		 * Query for the compression algorithm support if compression is enabled.
   793		 */
   794		if (!nocompress) {
   795			strscpy(hib_comp_algo, hibernate_compressor);
   796			if (!crypto_has_acomp(hib_comp_algo, 0, CRYPTO_ALG_ASYNC)) {
   797				pr_err("%s compression is not available\n", hib_comp_algo);
   798				return -EOPNOTSUPP;
   799			}
   800		}
   801	
   802		sleep_flags = lock_system_sleep();
   803		/* The snapshot device should not be opened while we're running */
   804		if (!hibernate_acquire()) {
   805			error = -EBUSY;
   806			goto Unlock;
   807		}
   808	
   809		pr_info("hibernation entry\n");
   810		pm_prepare_console();
   811		error = pm_notifier_call_chain_robust(PM_HIBERNATION_PREPARE, PM_POST_HIBERNATION);
   812		if (error)
   813			goto Restore;
   814	
   815		ksys_sync_helper();
   816		if (filesystem_freeze_enabled)
   817			filesystems_freeze();
   818	
   819		error = freeze_processes();
   820		if (error)
   821			goto Exit;
   822	
   823		lock_device_hotplug();
   824		/* Allocate memory management structures */
   825		error = create_basic_memory_bitmaps();
   826		if (error)
   827			goto Thaw;
   828	
   829		error = hibernation_snapshot(hibernation_mode == HIBERNATION_PLATFORM);
   830		if (error || freezer_test_done)
   831			goto Free_bitmaps;
   832	
   833		if (in_suspend) {
   834			unsigned int flags = 0;
   835	
   836			if (hibernation_mode == HIBERNATION_PLATFORM)
   837				flags |= SF_PLATFORM_MODE;
   838			if (nocompress) {
   839				flags |= SF_NOCOMPRESS_MODE;
   840			} else {
   841			        flags |= SF_CRC32_MODE;
   842	
   843				/*
   844				 * By default, LZO compression is enabled. Use SF_COMPRESSION_ALG_LZ4
   845				 * to override this behaviour and use LZ4.
   846				 *
   847				 * Refer kernel/power/power.h for more details
   848				 */
   849	
   850				if (!strcmp(hib_comp_algo, COMPRESSION_ALGO_LZ4))
   851					flags |= SF_COMPRESSION_ALG_LZ4;
   852				else
   853					flags |= SF_COMPRESSION_ALG_LZO;
   854			}
   855	
   856			pm_pr_dbg("Writing hibernation image.\n");
   857			error = swsusp_write(flags);
   858			swsusp_free();
   859			if (!error) {
   860				if (hibernation_mode == HIBERNATION_TEST_RESUME)
   861					snapshot_test = true;
   862				else
   863					power_down();
   864			}
   865			in_suspend = 0;
 > 866			if (hibernation_mode != HIBERNATION_SUSPEND)
   867				pm_restore_gfp_mask();
   868		} else {
   869			pm_pr_dbg("Hibernation image restored successfully.\n");
   870		}
   871	
   872	 Free_bitmaps:
   873		free_basic_memory_bitmaps();
   874	 Thaw:
   875		unlock_device_hotplug();
   876		if (snapshot_test) {
   877			pm_pr_dbg("Checking hibernation image\n");
   878			error = swsusp_check(false);
   879			if (!error)
   880				error = load_image_and_restore();
   881		}
   882		thaw_processes();
   883	
   884		/* Don't bother checking whether freezer_test_done is true */
   885		freezer_test_done = false;
   886	 Exit:
   887		filesystems_thaw();
   888		pm_notifier_call_chain(PM_POST_HIBERNATION);
   889	 Restore:
   890		pm_restore_console();
   891		hibernate_release();
   892	 Unlock:
   893		unlock_system_sleep(sleep_flags);
   894		pr_info("hibernation exit\n");
   895	
   896		return error;
   897	}
   898	

-- 
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki

[PATCH 2/3] PM: hibernate: Add pm_hibernation_mode_is_suspend()

[Mario Limonciello (AMD)]

Some drivers have different flows for hibernation and suspend. If
the driver opportunistically will skip thaw() then it needs a hint
to know what is happening after the hibernate.

Introduce a new symbol pm_hibernation_mode_is_suspend() that drivers
can call to determine if suspending the system for this purpose.

Tested-by: Ionut Nechita <ionut_n2001@yahoo.com>
Signed-off-by: Mario Limonciello (AMD) <superm1@kernel.org>
---
 include/linux/suspend.h  | 2 ++
 kernel/power/hibernate.c | 9 +++++++++
 2 files changed, 11 insertions(+)

diff --git a/include/linux/suspend.h b/include/linux/suspend.h
index 317ae31e89b3..4b11efd68729 100644
--- a/include/linux/suspend.h
+++ b/include/linux/suspend.h
@@ -479,6 +479,7 @@ extern void unlock_system_sleep(unsigned int);
 
 extern bool pm_sleep_transition_in_progress(void);
 bool pm_hibernate_is_recovering(void);
+bool pm_hibernation_mode_is_suspend(void);
 
 #else /* !CONFIG_PM_SLEEP */
 
@@ -513,6 +514,7 @@ static inline void unlock_system_sleep(unsigned int flags) {}
 
 static inline bool pm_sleep_transition_in_progress(void) { return false; }
 static inline bool pm_hibernate_is_recovering(void) { return false; }
+static inline bool pm_hibernation_mode_is_suspend(void) { return false; }
 
 #endif /* !CONFIG_PM_SLEEP */
 
diff --git a/kernel/power/hibernate.c b/kernel/power/hibernate.c
index 1defe8081e64..a991a1362d77 100644
--- a/kernel/power/hibernate.c
+++ b/kernel/power/hibernate.c
@@ -80,6 +80,15 @@ static const struct platform_hibernation_ops *hibernation_ops;
 
 static atomic_t hibernate_atomic = ATOMIC_INIT(1);
 
+/**
+ * pm_hibernation_mode_is_suspend - Check if hibernation has been set to suspend
+ */
+bool pm_hibernation_mode_is_suspend(void)
+{
+	return hibernation_mode == HIBERNATION_SUSPEND;
+}
+EXPORT_SYMBOL_GPL(pm_hibernation_mode_is_suspend);
+
 bool hibernate_acquire(void)
 {
 	return atomic_add_unless(&hibernate_atomic, -1, 0);
-- 
2.51.0

Re: [PATCH 2/3] PM: hibernate: Add pm_hibernation_mode_is_suspend()

[kernel test robot]

Hi Mario,

kernel test robot noticed the following build warnings:

[auto build test WARNING on amd-pstate/linux-next]
[also build test WARNING on amd-pstate/bleeding-edge linus/master v6.17-rc7 next-20250924]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]

url:    https://github.com/intel-lab-lkp/linux/commits/Mario-Limonciello-AMD/PM-hibernate-Fix-hybrid-sleep/20250925-045432
base:   https://git.kernel.org/pub/scm/linux/kernel/git/superm1/linux.git linux-next
patch link:    https://lore.kernel.org/r/20250924205211.1059571-3-superm1%40kernel.org
patch subject: [PATCH 2/3] PM: hibernate: Add pm_hibernation_mode_is_suspend()
config: i386-randconfig-004-20250925 (https://download.01.org/0day-ci/archive/20250925/202509252350.szp2l5YT-lkp@intel.com/config)
compiler: gcc-14 (Debian 14.2.0-19) 14.2.0
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20250925/202509252350.szp2l5YT-lkp@intel.com/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202509252350.szp2l5YT-lkp@intel.com/

All warnings (new ones prefixed by >>):

   kernel/power/hibernate.c: In function 'pm_hibernation_mode_is_suspend':
   kernel/power/hibernate.c:88:36: error: 'HIBERNATION_SUSPEND' undeclared (first use in this function); did you mean 'HIBERNATION_SHUTDOWN'?
      88 |         return hibernation_mode == HIBERNATION_SUSPEND;
         |                                    ^~~~~~~~~~~~~~~~~~~
         |                                    HIBERNATION_SHUTDOWN
   kernel/power/hibernate.c:88:36: note: each undeclared identifier is reported only once for each function it appears in
   kernel/power/hibernate.c: In function 'hibernate':
   kernel/power/hibernate.c:875:41: error: 'HIBERNATION_SUSPEND' undeclared (first use in this function); did you mean 'HIBERNATION_SHUTDOWN'?
     875 |                 if (hibernation_mode != HIBERNATION_SUSPEND)
         |                                         ^~~~~~~~~~~~~~~~~~~
         |                                         HIBERNATION_SHUTDOWN
   kernel/power/hibernate.c: In function 'pm_hibernation_mode_is_suspend':
>> kernel/power/hibernate.c:89:1: warning: control reaches end of non-void function [-Wreturn-type]
      89 | }
         | ^


vim +89 kernel/power/hibernate.c

    82	
    83	/**
    84	 * pm_hibernation_mode_is_suspend - Check if hibernation has been set to suspend
    85	 */
    86	bool pm_hibernation_mode_is_suspend(void)
    87	{
  > 88		return hibernation_mode == HIBERNATION_SUSPEND;
  > 89	}
    90	EXPORT_SYMBOL_GPL(pm_hibernation_mode_is_suspend);
    91	

-- 
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki

[PATCH 3/3] drm/amd: Fix hybrid sleep

[Mario Limonciello (AMD)]

[Why]
commit 530694f54dd5e ("drm/amdgpu: do not resume device in thaw for
normal hibernation") optimized the flow for systems that are going
into S4 where the power would be turned off.  Basically the thaw()
callback wouldn't resume the device if the hibernation image was
successfully created since the system would be powered off.

This however isn't the correct flow for a system entering into
s0i3 after the hibernation image is created.  Some of the amdgpu
callbacks have different behavior depending upon the intended
state of the suspend.

[How]
Use pm_hibernation_mode_is_suspend() as an input to decide whether
to run resume during thaw() callback.

Reported-by: Ionut Nechita <ionut_n2001@yahoo.com>
Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4573
Tested-by: Ionut Nechita <ionut_n2001@yahoo.com>
Fixes: 530694f54dd5e ("drm/amdgpu: do not resume device in thaw for normal hibernation")
Signed-off-by: Mario Limonciello (AMD) <superm1@kernel.org>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
index 395c6be901ce..dcea66aadfa3 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
@@ -2665,7 +2665,7 @@ static int amdgpu_pmops_thaw(struct device *dev)
 	struct drm_device *drm_dev = dev_get_drvdata(dev);
 
 	/* do not resume device if it's normal hibernation */
-	if (!pm_hibernate_is_recovering())
+	if (!pm_hibernate_is_recovering() && !pm_hibernation_mode_is_suspend())
 		return 0;
 
 	return amdgpu_device_resume(drm_dev, true);
-- 
2.51.0

RE: [PATCH 3/3] drm/amd: Fix hybrid sleep

[Deucher, Alexander]

[Public]

> -----Original Message-----
> From: amd-gfx <amd-gfx-bounces@lists.freedesktop.org> On Behalf Of Mario
> Limonciello (AMD)
> Sent: Wednesday, September 24, 2025 4:52 PM
> To: Deucher, Alexander <Alexander.Deucher@amd.com>; Rafael J . Wysocki
> <rafael@kernel.org>
> Cc: Zhang, GuoQing (Sam) <GuoQing.Zhang@amd.com>; open list:RADEON and
> AMDGPU DRM DRIVERS <amd-gfx@lists.freedesktop.org>; open
> list:HIBERNATION (aka Software Suspend, aka swsusp) <linux-
> pm@vger.kernel.org>; Mario Limonciello <superm1@kernel.org>; Ionut Nechita
> <ionut_n2001@yahoo.com>
> Subject: [PATCH 3/3] drm/amd: Fix hybrid sleep
>
> [Why]
> commit 530694f54dd5e ("drm/amdgpu: do not resume device in thaw for normal
> hibernation") optimized the flow for systems that are going into S4 where the power
> would be turned off.  Basically the thaw() callback wouldn't resume the device if the
> hibernation image was successfully created since the system would be powered
> off.
>
> This however isn't the correct flow for a system entering into
> s0i3 after the hibernation image is created.  Some of the amdgpu callbacks have
> different behavior depending upon the intended state of the suspend.
>
> [How]
> Use pm_hibernation_mode_is_suspend() as an input to decide whether to run
> resume during thaw() callback.
>
> Reported-by: Ionut Nechita <ionut_n2001@yahoo.com>
> Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4573
> Tested-by: Ionut Nechita <ionut_n2001@yahoo.com>
> Fixes: 530694f54dd5e ("drm/amdgpu: do not resume device in thaw for normal
> hibernation")
> Signed-off-by: Mario Limonciello (AMD) <superm1@kernel.org>

Series is:
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Feel free to take it through the Linux pm tree.

Alex

> ---
>  drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
> b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
> index 395c6be901ce..dcea66aadfa3 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
> @@ -2665,7 +2665,7 @@ static int amdgpu_pmops_thaw(struct device *dev)
>       struct drm_device *drm_dev = dev_get_drvdata(dev);
>
>       /* do not resume device if it's normal hibernation */
> -     if (!pm_hibernate_is_recovering())
> +     if (!pm_hibernate_is_recovering() &&
> +!pm_hibernation_mode_is_suspend())
>               return 0;
>
>       return amdgpu_device_resume(drm_dev, true);
> --
> 2.51.0

Re: [PATCH 0/3] Fixes for hybrid sleep

[Kenneth Crudup]

(Thanks Mario for the hint re: replying via Lore!)

On 9/24/25 13:52, Mario Limonciello (AMD) wrote:
> Ionut Nechita reported recently a hibernate failure, but in debugging
> the issue it's actually not a hibernate failure; but a hybrid sleep
> failure.
> 
> Multiple changes related to the change of when swap is disabled in
> the suspend sequence contribute to the failure.  See the individual
> patches for details.
> 
> Link: https://gitlab.freedesktop.org/drm/amd/-/issues/4573

I just wanted to say that the non-AMD-display bits of this patch were 
tested by me and seem to have fixed a problem I'd had where if I had an 
NFS share mounted on suspend/hibernate, that ~50% of the time when 
resuming the machine would be up, but disk I/O would be deadlocked (so I 
really couldn't do anything except SysRq-S/U/B).

Still going to test a few more suspend/resume/hibernate cycles, but it 
does seem to have fixed that issue for me. There was a corresponding 
patch in linux-pm earlier along the same lines, but that one only had a 
limited success rate.

I have "hybrid sleep" enabled, where systemd will go into full 
hibernation after 4 hrs, BTW.

>   drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c |  2 +-
>   include/linux/suspend.h                 |  2 ++
>   kernel/power/hibernate.c                | 13 ++++++++++++-
>   3 files changed, 15 insertions(+), 2 deletions(-)
> 

Tested-By: Kenneth Crudup <kenny@panix.com>

-Kenny

-- 
Kenneth R. Crudup / Sr. SW Engineer, Scott County Consulting, Orange 
County CA