]>
Commit | Line | Data |
---|---|---|
215e262f KO |
1 | /* |
2 | * Percpu refcounts: | |
3 | * (C) 2012 Google, Inc. | |
4 | * Author: Kent Overstreet <koverstreet@google.com> | |
5 | * | |
6 | * This implements a refcount with similar semantics to atomic_t - atomic_inc(), | |
7 | * atomic_dec_and_test() - but percpu. | |
8 | * | |
9 | * There's one important difference between percpu refs and normal atomic_t | |
10 | * refcounts; you have to keep track of your initial refcount, and then when you | |
11 | * start shutting down you call percpu_ref_kill() _before_ dropping the initial | |
12 | * refcount. | |
13 | * | |
14 | * The refcount will have a range of 0 to ((1U << 31) - 1), i.e. one bit less | |
15 | * than an atomic_t - this is because of the way shutdown works, see | |
eecc16ba | 16 | * percpu_ref_kill()/PERCPU_COUNT_BIAS. |
215e262f KO |
17 | * |
18 | * Before you call percpu_ref_kill(), percpu_ref_put() does not check for the | |
19 | * refcount hitting 0 - it can't, if it was in percpu mode. percpu_ref_kill() | |
20 | * puts the ref back in single atomic_t mode, collecting the per cpu refs and | |
21 | * issuing the appropriate barriers, and then marks the ref as shutting down so | |
22 | * that percpu_ref_put() will check for the ref hitting 0. After it returns, | |
23 | * it's safe to drop the initial ref. | |
24 | * | |
25 | * USAGE: | |
26 | * | |
27 | * See fs/aio.c for some example usage; it's used there for struct kioctx, which | |
28 | * is created when userspaces calls io_setup(), and destroyed when userspace | |
29 | * calls io_destroy() or the process exits. | |
30 | * | |
31 | * In the aio code, kill_ioctx() is called when we wish to destroy a kioctx; it | |
32 | * calls percpu_ref_kill(), then hlist_del_rcu() and sychronize_rcu() to remove | |
33 | * the kioctx from the proccess's list of kioctxs - after that, there can't be | |
34 | * any new users of the kioctx (from lookup_ioctx()) and it's then safe to drop | |
35 | * the initial ref with percpu_ref_put(). | |
36 | * | |
37 | * Code that does a two stage shutdown like this often needs some kind of | |
38 | * explicit synchronization to ensure the initial refcount can only be dropped | |
39 | * once - percpu_ref_kill() does this for you, it returns true once and false if | |
40 | * someone else already called it. The aio code uses it this way, but it's not | |
41 | * necessary if the code has some other mechanism to synchronize teardown. | |
42 | * around. | |
43 | */ | |
44 | ||
45 | #ifndef _LINUX_PERCPU_REFCOUNT_H | |
46 | #define _LINUX_PERCPU_REFCOUNT_H | |
47 | ||
48 | #include <linux/atomic.h> | |
49 | #include <linux/kernel.h> | |
50 | #include <linux/percpu.h> | |
51 | #include <linux/rcupdate.h> | |
a34375ef | 52 | #include <linux/gfp.h> |
215e262f KO |
53 | |
54 | struct percpu_ref; | |
ac899061 | 55 | typedef void (percpu_ref_func_t)(struct percpu_ref *); |
215e262f | 56 | |
9e804d1f TH |
57 | /* flags set in the lower bits of percpu_ref->percpu_count_ptr */ |
58 | enum { | |
59 | __PERCPU_REF_ATOMIC = 1LU << 0, /* operating in atomic mode */ | |
60 | }; | |
61 | ||
215e262f | 62 | struct percpu_ref { |
e625305b | 63 | atomic_long_t count; |
215e262f KO |
64 | /* |
65 | * The low bit of the pointer indicates whether the ref is in percpu | |
9a1049da | 66 | * mode; if set, then get/put will manipulate the atomic_t. |
215e262f | 67 | */ |
eecc16ba | 68 | unsigned long percpu_count_ptr; |
ac899061 | 69 | percpu_ref_func_t *release; |
9e804d1f | 70 | percpu_ref_func_t *confirm_switch; |
215e262f KO |
71 | struct rcu_head rcu; |
72 | }; | |
73 | ||
acac7883 | 74 | int __must_check percpu_ref_init(struct percpu_ref *ref, |
a34375ef | 75 | percpu_ref_func_t *release, gfp_t gfp); |
9a1049da | 76 | void percpu_ref_exit(struct percpu_ref *ref); |
dbece3a0 TH |
77 | void percpu_ref_kill_and_confirm(struct percpu_ref *ref, |
78 | percpu_ref_func_t *confirm_kill); | |
a2237370 | 79 | void percpu_ref_reinit(struct percpu_ref *ref); |
dbece3a0 TH |
80 | |
81 | /** | |
82 | * percpu_ref_kill - drop the initial ref | |
83 | * @ref: percpu_ref to kill | |
84 | * | |
85 | * Must be used to drop the initial ref on a percpu refcount; must be called | |
86 | * precisely once before shutdown. | |
87 | * | |
88 | * Puts @ref in non percpu mode, then does a call_rcu() before gathering up the | |
89 | * percpu counters and dropping the initial ref. | |
90 | */ | |
91 | static inline void percpu_ref_kill(struct percpu_ref *ref) | |
92 | { | |
93 | return percpu_ref_kill_and_confirm(ref, NULL); | |
94 | } | |
215e262f | 95 | |
eae7975d TH |
96 | /* |
97 | * Internal helper. Don't use outside percpu-refcount proper. The | |
98 | * function doesn't return the pointer and let the caller test it for NULL | |
99 | * because doing so forces the compiler to generate two conditional | |
eecc16ba | 100 | * branches as it can't assume that @ref->percpu_count is not NULL. |
eae7975d | 101 | */ |
9e804d1f TH |
102 | static inline bool __ref_is_percpu(struct percpu_ref *ref, |
103 | unsigned long __percpu **percpu_countp) | |
eae7975d | 104 | { |
eecc16ba | 105 | unsigned long percpu_ptr = ACCESS_ONCE(ref->percpu_count_ptr); |
eae7975d | 106 | |
2d722782 TH |
107 | /* paired with smp_store_release() in percpu_ref_reinit() */ |
108 | smp_read_barrier_depends(); | |
109 | ||
9e804d1f | 110 | if (unlikely(percpu_ptr & __PERCPU_REF_ATOMIC)) |
eae7975d TH |
111 | return false; |
112 | ||
eecc16ba | 113 | *percpu_countp = (unsigned long __percpu *)percpu_ptr; |
eae7975d TH |
114 | return true; |
115 | } | |
215e262f KO |
116 | |
117 | /** | |
118 | * percpu_ref_get - increment a percpu refcount | |
ac899061 | 119 | * @ref: percpu_ref to get |
215e262f | 120 | * |
6251f997 TH |
121 | * Analagous to atomic_long_inc(). |
122 | * | |
123 | * This function is safe to call as long as @ref is between init and exit. | |
124 | */ | |
215e262f KO |
125 | static inline void percpu_ref_get(struct percpu_ref *ref) |
126 | { | |
eecc16ba | 127 | unsigned long __percpu *percpu_count; |
215e262f | 128 | |
a4244454 | 129 | rcu_read_lock_sched(); |
215e262f | 130 | |
9e804d1f | 131 | if (__ref_is_percpu(ref, &percpu_count)) |
eecc16ba | 132 | this_cpu_inc(*percpu_count); |
215e262f | 133 | else |
e625305b | 134 | atomic_long_inc(&ref->count); |
215e262f | 135 | |
a4244454 | 136 | rcu_read_unlock_sched(); |
215e262f KO |
137 | } |
138 | ||
4fb6e250 TH |
139 | /** |
140 | * percpu_ref_tryget - try to increment a percpu refcount | |
141 | * @ref: percpu_ref to try-get | |
142 | * | |
143 | * Increment a percpu refcount unless its count already reached zero. | |
144 | * Returns %true on success; %false on failure. | |
145 | * | |
6251f997 | 146 | * This function is safe to call as long as @ref is between init and exit. |
4fb6e250 TH |
147 | */ |
148 | static inline bool percpu_ref_tryget(struct percpu_ref *ref) | |
149 | { | |
eecc16ba | 150 | unsigned long __percpu *percpu_count; |
6251f997 | 151 | int ret; |
4fb6e250 TH |
152 | |
153 | rcu_read_lock_sched(); | |
154 | ||
9e804d1f | 155 | if (__ref_is_percpu(ref, &percpu_count)) { |
eecc16ba | 156 | this_cpu_inc(*percpu_count); |
4fb6e250 TH |
157 | ret = true; |
158 | } else { | |
e625305b | 159 | ret = atomic_long_inc_not_zero(&ref->count); |
4fb6e250 TH |
160 | } |
161 | ||
162 | rcu_read_unlock_sched(); | |
163 | ||
164 | return ret; | |
165 | } | |
166 | ||
dbece3a0 | 167 | /** |
2070d50e | 168 | * percpu_ref_tryget_live - try to increment a live percpu refcount |
dbece3a0 TH |
169 | * @ref: percpu_ref to try-get |
170 | * | |
171 | * Increment a percpu refcount unless it has already been killed. Returns | |
172 | * %true on success; %false on failure. | |
173 | * | |
6251f997 TH |
174 | * Completion of percpu_ref_kill() in itself doesn't guarantee that this |
175 | * function will fail. For such guarantee, percpu_ref_kill_and_confirm() | |
176 | * should be used. After the confirm_kill callback is invoked, it's | |
177 | * guaranteed that no new reference will be given out by | |
178 | * percpu_ref_tryget_live(). | |
4fb6e250 | 179 | * |
6251f997 | 180 | * This function is safe to call as long as @ref is between init and exit. |
dbece3a0 | 181 | */ |
2070d50e | 182 | static inline bool percpu_ref_tryget_live(struct percpu_ref *ref) |
dbece3a0 | 183 | { |
eecc16ba | 184 | unsigned long __percpu *percpu_count; |
dbece3a0 TH |
185 | int ret = false; |
186 | ||
a4244454 | 187 | rcu_read_lock_sched(); |
dbece3a0 | 188 | |
9e804d1f | 189 | if (__ref_is_percpu(ref, &percpu_count)) { |
eecc16ba | 190 | this_cpu_inc(*percpu_count); |
dbece3a0 TH |
191 | ret = true; |
192 | } | |
193 | ||
a4244454 | 194 | rcu_read_unlock_sched(); |
dbece3a0 TH |
195 | |
196 | return ret; | |
197 | } | |
198 | ||
215e262f KO |
199 | /** |
200 | * percpu_ref_put - decrement a percpu refcount | |
ac899061 | 201 | * @ref: percpu_ref to put |
215e262f KO |
202 | * |
203 | * Decrement the refcount, and if 0, call the release function (which was passed | |
204 | * to percpu_ref_init()) | |
6251f997 TH |
205 | * |
206 | * This function is safe to call as long as @ref is between init and exit. | |
215e262f KO |
207 | */ |
208 | static inline void percpu_ref_put(struct percpu_ref *ref) | |
209 | { | |
eecc16ba | 210 | unsigned long __percpu *percpu_count; |
215e262f | 211 | |
a4244454 | 212 | rcu_read_lock_sched(); |
215e262f | 213 | |
9e804d1f | 214 | if (__ref_is_percpu(ref, &percpu_count)) |
eecc16ba | 215 | this_cpu_dec(*percpu_count); |
e625305b | 216 | else if (unlikely(atomic_long_dec_and_test(&ref->count))) |
215e262f KO |
217 | ref->release(ref); |
218 | ||
a4244454 | 219 | rcu_read_unlock_sched(); |
215e262f KO |
220 | } |
221 | ||
2d722782 TH |
222 | /** |
223 | * percpu_ref_is_zero - test whether a percpu refcount reached zero | |
224 | * @ref: percpu_ref to test | |
225 | * | |
226 | * Returns %true if @ref reached zero. | |
6251f997 TH |
227 | * |
228 | * This function is safe to call as long as @ref is between init and exit. | |
2d722782 TH |
229 | */ |
230 | static inline bool percpu_ref_is_zero(struct percpu_ref *ref) | |
231 | { | |
eecc16ba | 232 | unsigned long __percpu *percpu_count; |
2d722782 | 233 | |
9e804d1f | 234 | if (__ref_is_percpu(ref, &percpu_count)) |
2d722782 | 235 | return false; |
e625305b | 236 | return !atomic_long_read(&ref->count); |
2d722782 TH |
237 | } |
238 | ||
215e262f | 239 | #endif |