1 .. _NMI_rcu_doc:
2
3 Using RCU to Protect Dynamic NMI Handlers
4 =========================================
5
6
7 Although RCU is usually used to protect read-m
8 it is possible to use RCU to provide dynamic n
9 handlers, as well as dynamic irq handlers. Th
10 how to do this, drawing loosely from Zwane Mwa
11 work in an old version of "arch/x86/kernel/tra
12
13 The relevant pieces of code are listed below,
14 brief explanation::
15
16 static int dummy_nmi_callback(struct p
17 {
18 return 0;
19 }
20
21 The dummy_nmi_callback() function is a "dummy"
22 nothing, but returns zero, thus saying that it
23 the NMI handler to take the default machine-sp
24
25 static nmi_callback_t nmi_callback = d
26
27 This nmi_callback variable is a global functio
28 NMI handler::
29
30 void do_nmi(struct pt_regs * regs, lon
31 {
32 int cpu;
33
34 nmi_enter();
35
36 cpu = smp_processor_id();
37 ++nmi_count(cpu);
38
39 if (!rcu_dereference_sched(nmi
40 default_do_nmi(regs);
41
42 nmi_exit();
43 }
44
45 The do_nmi() function processes each NMI. It
46 in the same way that a hardware irq would, the
47 count of NMIs. It then invokes the NMI handle
48 function pointer. If this handler returns zer
49 default_do_nmi() function to handle a machine-
50 preemption is restored.
51
52 In theory, rcu_dereference_sched() is not need
53 only on i386, which in theory does not need rc
54 anyway. However, in practice it is a good doc
55 for anyone attempting to do something similar
56 with aggressive optimizing compilers.
57
58 Quick Quiz:
59 Why might the rcu_dereference_
60
61 :ref:`Answer to Quick Quiz <answer_quick_quiz_
62
63 Back to the discussion of NMI and RCU::
64
65 void set_nmi_callback(nmi_callback_t c
66 {
67 rcu_assign_pointer(nmi_callbac
68 }
69
70 The set_nmi_callback() function registers an N
71 data that is to be used by the callback must b
72 the call to set_nmi_callback(). On architectu
73 writes, the rcu_assign_pointer() ensures that
74 initialized values::
75
76 void unset_nmi_callback(void)
77 {
78 rcu_assign_pointer(nmi_callbac
79 }
80
81 This function unregisters an NMI handler, rest
82 dummy_nmi_handler(). However, there may well
83 currently executing on some other CPU. We the
84 up any data structures used by the old NMI han
85 of it completes on all other CPUs.
86
87 One way to accomplish this is via synchronize_
88 follows::
89
90 unset_nmi_callback();
91 synchronize_rcu();
92 kfree(my_nmi_data);
93
94 This works because (as of v4.20) synchronize_r
95 CPUs complete any preemption-disabled segments
96 executing.
97 Since NMI handlers disable preemption, synchro
98 not to return until all ongoing NMI handlers e
99 to free up the handler's data as soon as synch
100
101 Important note: for this to work, the architec
102 invoke nmi_enter() and nmi_exit() on NMI entry
103
104 .. _answer_quick_quiz_NMI:
105
106 Answer to Quick Quiz:
107 Why might the rcu_dereference_sched()
108
109 The caller to set_nmi_callback() might
110 initialized some data that is to be us
111 handler. In this case, the rcu_derefe
112 be needed, because otherwise a CPU tha
113 just after the new handler was set mig
114 to the new NMI handler, but the old pr
115 version of the handler's data.
116
117 This same sad story can happen on othe
118 a compiler with aggressive pointer-val
119 optimizations. (But please don't!)
120
121 More important, the rcu_dereference_sc
122 clear to someone reading the code that
123 being protected by RCU-sched.
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.