5 (C) 2007 Rafael J. Wysocki <rjw@sisk.pl>, GPL
7 I. What is the freezing of tasks?
8 =================================
10 The freezing of tasks is a mechanism by which user space processes and some
11 kernel threads are controlled during hibernation or system-wide suspend (on some
17 There are three per-task flags used for that, PF_NOFREEZE, PF_FROZEN
18 and PF_FREEZER_SKIP (the last one is auxiliary). The tasks that have
19 PF_NOFREEZE unset (all user space processes and some kernel threads) are
20 regarded as 'freezable' and treated in a special way before the system enters a
21 suspend state as well as before a hibernation image is created (in what follows
22 we only consider hibernation, but the description also applies to suspend).
24 Namely, as the first step of the hibernation procedure the function
25 freeze_processes() (defined in kernel/power/process.c) is called. A system-wide
26 variable system_freezing_cnt (as opposed to a per-task flag) is used to indicate
27 whether the system is to undergo a freezing operation. And freeze_processes()
28 sets this variable. After this, it executes try_to_freeze_tasks() that sends a
29 fake signal to all user space processes, and wakes up all the kernel threads.
30 All freezable tasks must react to that by calling try_to_freeze(), which
31 results in a call to __refrigerator() (defined in kernel/freezer.c), which sets
32 the task's PF_FROZEN flag, changes its state to TASK_UNINTERRUPTIBLE and makes
33 it loop until PF_FROZEN is cleared for it. Then, we say that the task is
34 'frozen' and therefore the set of functions handling this mechanism is referred
35 to as 'the freezer' (these functions are defined in kernel/power/process.c,
36 kernel/freezer.c & include/linux/freezer.h). User space processes are generally
37 frozen before kernel threads.
39 __refrigerator() must not be called directly. Instead, use the
40 try_to_freeze() function (defined in include/linux/freezer.h), that checks
41 if the task is to be frozen and makes the task enter __refrigerator().
43 For user space processes try_to_freeze() is called automatically from the
44 signal-handling code, but the freezable kernel threads need to call it
45 explicitly in suitable places or use the wait_event_freezable() or
46 wait_event_freezable_timeout() macros (defined in include/linux/freezer.h)
47 that combine interruptible sleep with checking if the task is to be frozen and
48 calling try_to_freeze(). The main loop of a freezable kernel thread may look
49 like the following one::
54 wait_event_freezable(khubd_wait,
55 !list_empty(&hub_event_list) ||
56 kthread_should_stop());
57 } while (!kthread_should_stop() || !list_empty(&hub_event_list));
59 (from drivers/usb/core/hub.c::hub_thread()).
61 If a freezable kernel thread fails to call try_to_freeze() after the freezer has
62 initiated a freezing operation, the freezing of tasks will fail and the entire
63 hibernation operation will be cancelled. For this reason, freezable kernel
64 threads must call try_to_freeze() somewhere or use one of the
65 wait_event_freezable() and wait_event_freezable_timeout() macros.
67 After the system memory state has been restored from a hibernation image and
68 devices have been reinitialized, the function thaw_processes() is called in
69 order to clear the PF_FROZEN flag for each frozen task. Then, the tasks that
70 have been frozen leave __refrigerator() and continue running.
73 Rationale behind the functions dealing with freezing and thawing of tasks
74 -------------------------------------------------------------------------
77 - freezes only userspace tasks
79 freeze_kernel_threads():
80 - freezes all tasks (including kernel threads) because we can't freeze
81 kernel threads without freezing userspace tasks
83 thaw_kernel_threads():
84 - thaws only kernel threads; this is particularly useful if we need to do
85 anything special in between thawing of kernel threads and thawing of
86 userspace tasks, or if we want to postpone the thawing of userspace tasks
89 - thaws all tasks (including kernel threads) because we can't thaw userspace
90 tasks without thawing kernel threads
93 III. Which kernel threads are freezable?
94 ========================================
96 Kernel threads are not freezable by default. However, a kernel thread may clear
97 PF_NOFREEZE for itself by calling set_freezable() (the resetting of PF_NOFREEZE
98 directly is not allowed). From this point it is regarded as freezable
99 and must call try_to_freeze() in a suitable place.
101 IV. Why do we do that?
102 ======================
104 Generally speaking, there is a couple of reasons to use the freezing of tasks:
106 1. The principal reason is to prevent filesystems from being damaged after
107 hibernation. At the moment we have no simple means of checkpointing
108 filesystems, so if there are any modifications made to filesystem data and/or
109 metadata on disks, we cannot bring them back to the state from before the
110 modifications. At the same time each hibernation image contains some
111 filesystem-related information that must be consistent with the state of the
112 on-disk data and metadata after the system memory state has been restored
113 from the image (otherwise the filesystems will be damaged in a nasty way,
114 usually making them almost impossible to repair). We therefore freeze
115 tasks that might cause the on-disk filesystems' data and metadata to be
116 modified after the hibernation image has been created and before the
117 system is finally powered off. The majority of these are user space
118 processes, but if any of the kernel threads may cause something like this
119 to happen, they have to be freezable.
121 2. Next, to create the hibernation image we need to free a sufficient amount of
122 memory (approximately 50% of available RAM) and we need to do that before
123 devices are deactivated, because we generally need them for swapping out.
124 Then, after the memory for the image has been freed, we don't want tasks
125 to allocate additional memory and we prevent them from doing that by
126 freezing them earlier. [Of course, this also means that device drivers
127 should not allocate substantial amounts of memory from their .suspend()
128 callbacks before hibernation, but this is a separate issue.]
130 3. The third reason is to prevent user space processes and some kernel threads
131 from interfering with the suspending and resuming of devices. A user space
132 process running on a second CPU while we are suspending devices may, for
133 example, be troublesome and without the freezing of tasks we would need some
134 safeguards against race conditions that might occur in such a case.
136 Although Linus Torvalds doesn't like the freezing of tasks, he said this in one
137 of the discussions on LKML (https://lore.kernel.org/r/alpine.LFD.0.98.0704271801020.9964@woody.linux-foundation.org):
139 "RJW:> Why we freeze tasks at all or why we freeze kernel threads?
141 Linus: In many ways, 'at all'.
143 I **do** realize the IO request queue issues, and that we cannot actually do
144 s2ram with some devices in the middle of a DMA. So we want to be able to
145 avoid *that*, there's no question about that. And I suspect that stopping
146 user threads and then waiting for a sync is practically one of the easier
149 So in practice, the 'at all' may become a 'why freeze kernel threads?' and
150 freezing user threads I don't find really objectionable."
152 Still, there are kernel threads that may want to be freezable. For example, if
153 a kernel thread that belongs to a device driver accesses the device directly, it
154 in principle needs to know when the device is suspended, so that it doesn't try
155 to access it at that time. However, if the kernel thread is freezable, it will
156 be frozen before the driver's .suspend() callback is executed and it will be
157 thawed after the driver's .resume() callback has run, so it won't be accessing
158 the device while it's suspended.
160 4. Another reason for freezing tasks is to prevent user space processes from
161 realizing that hibernation (or suspend) operation takes place. Ideally, user
162 space processes should not notice that such a system-wide operation has
163 occurred and should continue running without any problems after the restore
164 (or resume from suspend). Unfortunately, in the most general case this
165 is quite difficult to achieve without the freezing of tasks. Consider,
166 for example, a process that depends on all CPUs being online while it's
167 running. Since we need to disable nonboot CPUs during the hibernation,
168 if this process is not frozen, it may notice that the number of CPUs has
169 changed and may start to work incorrectly because of that.
171 V. Are there any problems related to the freezing of tasks?
172 ===========================================================
176 First of all, the freezing of kernel threads may be tricky if they depend one
177 on another. For example, if kernel thread A waits for a completion (in the
178 TASK_UNINTERRUPTIBLE state) that needs to be done by freezable kernel thread B
179 and B is frozen in the meantime, then A will be blocked until B is thawed, which
180 may be undesirable. That's why kernel threads are not freezable by default.
182 Second, there are the following two problems related to the freezing of user
185 1. Putting processes into an uninterruptible sleep distorts the load average.
186 2. Now that we have FUSE, plus the framework for doing device drivers in
187 userspace, it gets even more complicated because some userspace processes are
188 now doing the sorts of things that kernel threads do
189 (https://lists.linux-foundation.org/pipermail/linux-pm/2007-May/012309.html).
191 The problem 1. seems to be fixable, although it hasn't been fixed so far. The
192 other one is more serious, but it seems that we can work around it by using
193 hibernation (and suspend) notifiers (in that case, though, we won't be able to
194 avoid the realization by the user space processes that the hibernation is taking
197 There are also problems that the freezing of tasks tends to expose, although
198 they are not directly related to it. For example, if request_firmware() is
199 called from a device driver's .resume() routine, it will timeout and eventually
200 fail, because the user land process that should respond to the request is frozen
201 at this point. So, seemingly, the failure is due to the freezing of tasks.
202 Suppose, however, that the firmware file is located on a filesystem accessible
203 only through another device that hasn't been resumed yet. In that case,
204 request_firmware() will fail regardless of whether or not the freezing of tasks
205 is used. Consequently, the problem is not really related to the freezing of
206 tasks, since it generally exists anyway.
208 A driver must have all firmwares it may need in RAM before suspend() is called.
209 If keeping them is not practical, for example due to their size, they must be
210 requested early enough using the suspend notifier API described in
211 Documentation/driver-api/pm/notifiers.rst.
213 VI. Are there any precautions to be taken to prevent freezing failures?
214 =======================================================================
218 First of all, grabbing the 'system_transition_mutex' lock to mutually exclude a
219 piece of code from system-wide sleep such as suspend/hibernation is not
220 encouraged. If possible, that piece of code must instead hook onto the
221 suspend/hibernation notifiers to achieve mutual exclusion. Look at the
222 CPU-Hotplug code (kernel/cpu.c) for an example.
224 However, if that is not feasible, and grabbing 'system_transition_mutex' is
225 deemed necessary, it is strongly discouraged to directly call
226 mutex_[un]lock(&system_transition_mutex) since that could lead to freezing
227 failures, because if the suspend/hibernate code successfully acquired the
228 'system_transition_mutex' lock, and hence that other entity failed to acquire
229 the lock, then that task would get blocked in TASK_UNINTERRUPTIBLE state. As a
230 consequence, the freezer would not be able to freeze that task, leading to
233 However, the [un]lock_system_sleep() APIs are safe to use in this scenario,
234 since they ask the freezer to skip freezing this task, since it is anyway
235 "frozen enough" as it is blocked on 'system_transition_mutex', which will be
236 released only after the entire suspend/hibernation sequence is complete. So, to
237 summarize, use [un]lock_system_sleep() instead of directly using
238 mutex_[un]lock(&system_transition_mutex). That would prevent freezing failures.
243 /sys/power/pm_freeze_timeout controls how long it will cost at most to freeze
244 all user space processes or all freezable kernel threads, in unit of
245 millisecond. The default value is 20000, with range of unsigned integer.