xref: /linux/Documentation/power/suspend-and-interrupts.rst (revision 72bea132f3680ee51e7ed2cee62892b6f5121909)
1====================================
2System Suspend and Device Interrupts
3====================================
4
5Copyright (C) 2014 Intel Corp.
6Author: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
7
8
9Suspending and Resuming Device IRQs
10-----------------------------------
11
12Device interrupt request lines (IRQs) are generally disabled during system
13suspend after the "late" phase of suspending devices (that is, after all of the
14->prepare, ->suspend and ->suspend_late callbacks have been executed for all
15devices).  That is done by suspend_device_irqs().
16
17The rationale for doing so is that after the "late" phase of device suspend
18there is no legitimate reason why any interrupts from suspended devices should
19trigger and if any devices have not been suspended properly yet, it is better to
20block interrupts from them anyway.  Also, in the past we had problems with
21interrupt handlers for shared IRQs that device drivers implementing them were
22not prepared for interrupts triggering after their devices had been suspended.
23In some cases they would attempt to access, for example, memory address spaces
24of suspended devices and cause unpredictable behavior to ensue as a result.
25Unfortunately, such problems are very difficult to debug and the introduction
26of suspend_device_irqs(), along with the "noirq" phase of device suspend and
27resume, was the only practical way to mitigate them.
28
29Device IRQs are re-enabled during system resume, right before the "early" phase
30of resuming devices (that is, before starting to execute ->resume_early
31callbacks for devices).  The function doing that is resume_device_irqs().
32
33
34The IRQF_NO_SUSPEND Flag
35------------------------
36
37There are interrupts that can legitimately trigger during the entire system
38suspend-resume cycle, including the "noirq" phases of suspending and resuming
39devices as well as during the time when nonboot CPUs are taken offline and
40brought back online.  That applies to timer interrupts in the first place,
41but also to IPIs and to some other special-purpose interrupts.
42
43The IRQF_NO_SUSPEND flag is used to indicate that to the IRQ subsystem when
44requesting a special-purpose interrupt.  It causes suspend_device_irqs() to
45leave the corresponding IRQ enabled so as to allow the interrupt to work as
46expected during the suspend-resume cycle, but does not guarantee that the
47interrupt will wake the system from a suspended state -- for such cases it is
48necessary to use enable_irq_wake().
49
50Note that the IRQF_NO_SUSPEND flag affects the entire IRQ and not just one
51user of it.  Thus, if the IRQ is shared, all of the interrupt handlers installed
52for it will be executed as usual after suspend_device_irqs(), even if the
53IRQF_NO_SUSPEND flag was not passed to request_irq() (or equivalent) by some of
54the IRQ's users.  For this reason, using IRQF_NO_SUSPEND and IRQF_SHARED at the
55same time should be avoided.
56
57
58System Wakeup Interrupts, enable_irq_wake() and disable_irq_wake()
59------------------------------------------------------------------
60
61System wakeup interrupts generally need to be configured to wake up the system
62from sleep states, especially if they are used for different purposes (e.g. as
63I/O interrupts) in the working state.
64
65That may involve turning on a special signal handling logic within the platform
66(such as an SoC) so that signals from a given line are routed in a different way
67during system sleep so as to trigger a system wakeup when needed.  For example,
68the platform may include a dedicated interrupt controller used specifically for
69handling system wakeup events.  Then, if a given interrupt line is supposed to
70wake up the system from sleep states, the corresponding input of that interrupt
71controller needs to be enabled to receive signals from the line in question.
72After wakeup, it generally is better to disable that input to prevent the
73dedicated controller from triggering interrupts unnecessarily.
74
75The IRQ subsystem provides two helper functions to be used by device drivers for
76those purposes.  Namely, enable_irq_wake() turns on the platform's logic for
77handling the given IRQ as a system wakeup interrupt line and disable_irq_wake()
78turns that logic off.
79
80Calling enable_irq_wake() causes suspend_device_irqs() to treat the given IRQ
81in a special way.  Namely, the IRQ remains enabled, but on the first interrupt
82it will be disabled, marked as pending and "suspended" so that it will be
83re-enabled by resume_device_irqs() during the subsequent system resume.  Also
84the PM core is notified about the event which causes the system suspend in
85progress to be aborted (that doesn't have to happen immediately, but at one
86of the points where the suspend thread looks for pending wakeup events).
87
88This way every interrupt from a wakeup interrupt source will either cause the
89system suspend currently in progress to be aborted or wake up the system if
90already suspended.  However, after suspend_device_irqs() interrupt handlers are
91not executed for system wakeup IRQs.  They are only executed for IRQF_NO_SUSPEND
92IRQs at that time, but those IRQs should not be configured for system wakeup
93using enable_irq_wake().
94
95
96Interrupts and Suspend-to-Idle
97------------------------------
98
99Suspend-to-idle (also known as the "freeze" sleep state) is a relatively new
100system sleep state that works by idling all of the processors and waiting for
101interrupts right after the "noirq" phase of suspending devices.
102
103Of course, this means that all of the interrupts with the IRQF_NO_SUSPEND flag
104set will bring CPUs out of idle while in that state, but they will not cause the
105IRQ subsystem to trigger a system wakeup.
106
107System wakeup interrupts, in turn, will trigger wakeup from suspend-to-idle in
108analogy with what they do in the full system suspend case.  The only difference
109is that the wakeup from suspend-to-idle is signaled using the usual working
110state interrupt delivery mechanisms and doesn't require the platform to use
111any special interrupt handling logic for it to work.
112
113
114IRQF_NO_SUSPEND and enable_irq_wake()
115-------------------------------------
116
117There are very few valid reasons to use both enable_irq_wake() and the
118IRQF_NO_SUSPEND flag on the same IRQ, and it is never valid to use both for the
119same device.
120
121First of all, if the IRQ is not shared, the rules for handling IRQF_NO_SUSPEND
122interrupts (interrupt handlers are invoked after suspend_device_irqs()) are
123directly at odds with the rules for handling system wakeup interrupts (interrupt
124handlers are not invoked after suspend_device_irqs()).
125
126Second, both enable_irq_wake() and IRQF_NO_SUSPEND apply to entire IRQs and not
127to individual interrupt handlers, so sharing an IRQ between a system wakeup
128interrupt source and an IRQF_NO_SUSPEND interrupt source does not generally
129make sense.
130
131In rare cases an IRQ can be shared between a wakeup device driver and an
132IRQF_NO_SUSPEND user. In order for this to be safe, the wakeup device driver
133must be able to discern spurious IRQs from genuine wakeup events (signalling
134the latter to the core with pm_system_wakeup()), must use enable_irq_wake() to
135ensure that the IRQ will function as a wakeup source, and must request the IRQ
136with IRQF_COND_SUSPEND to tell the core that it meets these requirements. If
137these requirements are not met, it is not valid to use IRQF_COND_SUSPEND.
138