Blame view
kernel/locking/qspinlock.c
16.6 KB
c942fddf8 treewide: Replace... |
1 |
// SPDX-License-Identifier: GPL-2.0-or-later |
a33fda35e locking/qspinlock... |
2 3 4 |
/* * Queued spinlock * |
a33fda35e locking/qspinlock... |
5 |
* (C) Copyright 2013-2015 Hewlett-Packard Development Company, L.P. |
81d3dc9a3 locking/qspinlock... |
6 |
* (C) Copyright 2013-2014,2018 Red Hat, Inc. |
a33fda35e locking/qspinlock... |
7 |
* (C) Copyright 2015 Intel Corp. |
64d816cba locking/qspinlock... |
8 |
* (C) Copyright 2015 Hewlett-Packard Enterprise Development LP |
a33fda35e locking/qspinlock... |
9 |
* |
81d3dc9a3 locking/qspinlock... |
10 |
* Authors: Waiman Long <longman@redhat.com> |
a33fda35e locking/qspinlock... |
11 12 |
* Peter Zijlstra <peterz@infradead.org> */ |
a23db284f locking/pvqspinlo... |
13 14 |
#ifndef _GEN_PV_LOCK_SLOWPATH |
a33fda35e locking/qspinlock... |
15 16 17 18 19 20 |
#include <linux/smp.h> #include <linux/bug.h> #include <linux/cpumask.h> #include <linux/percpu.h> #include <linux/hardirq.h> #include <linux/mutex.h> |
5671360f2 locking/qspinlock... |
21 |
#include <linux/prefetch.h> |
69f9cae90 locking/qspinlock... |
22 |
#include <asm/byteorder.h> |
a33fda35e locking/qspinlock... |
23 24 25 |
#include <asm/qspinlock.h> /* |
81d3dc9a3 locking/qspinlock... |
26 27 28 29 30 |
* Include queued spinlock statistics code */ #include "qspinlock_stat.h" /* |
a33fda35e locking/qspinlock... |
31 32 |
* The basic principle of a queue-based spinlock can best be understood * by studying a classic queue-based spinlock implementation called the |
57097124c locking/qspinlock... |
33 34 35 |
* MCS lock. A copy of the original MCS lock paper ("Algorithms for Scalable * Synchronization on Shared-Memory Multiprocessors by Mellor-Crummey and * Scott") is available at |
a33fda35e locking/qspinlock... |
36 |
* |
57097124c locking/qspinlock... |
37 |
* https://bugzilla.kernel.org/show_bug.cgi?id=206115 |
a33fda35e locking/qspinlock... |
38 |
* |
57097124c locking/qspinlock... |
39 40 41 |
* This queued spinlock implementation is based on the MCS lock, however to * make it fit the 4 bytes we assume spinlock_t to be, and preserve its * existing API, we must modify it somehow. |
a33fda35e locking/qspinlock... |
42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 |
* * In particular; where the traditional MCS lock consists of a tail pointer * (8 bytes) and needs the next pointer (another 8 bytes) of its own node to * unlock the next pending (next->locked), we compress both these: {tail, * next->locked} into a single u32 value. * * Since a spinlock disables recursion of its own context and there is a limit * to the contexts that can nest; namely: task, softirq, hardirq, nmi. As there * are at most 4 nesting levels, it can be encoded by a 2-bit number. Now * we can encode the tail by combining the 2-bit nesting level with the cpu * number. With one byte for the lock value and 3 bytes for the tail, only a * 32-bit word is now needed. Even though we only need 1 bit for the lock, * we extend it to a full byte to achieve better performance for architectures * that support atomic byte write. * * We also change the first spinner to spin on the lock bit instead of its * node; whereby avoiding the need to carry a node from lock to unlock, and * preserving existing lock API. This also makes the unlock code simpler and * faster. |
69f9cae90 locking/qspinlock... |
61 62 63 64 |
* * N.B. The current implementation only supports architectures that allow * atomic operations on smaller 8-bit and 16-bit data types. * |
a33fda35e locking/qspinlock... |
65 66 67 |
*/ #include "mcs_spinlock.h" |
0fa809ca7 locking/pvqspinlo... |
68 |
#define MAX_NODES 4 |
a33fda35e locking/qspinlock... |
69 |
|
0fa809ca7 locking/pvqspinlo... |
70 71 72 73 74 75 76 77 78 79 80 81 |
/* * On 64-bit architectures, the mcs_spinlock structure will be 16 bytes in * size and four of them will fit nicely in one 64-byte cacheline. For * pvqspinlock, however, we need more space for extra data. To accommodate * that, we insert two more long words to pad it up to 32 bytes. IOW, only * two of them can fit in a cacheline in this case. That is OK as it is rare * to have more than 2 levels of slowpath nesting in actual use. We don't * want to penalize pvqspinlocks to optimize for a rare case in native * qspinlocks. */ struct qnode { struct mcs_spinlock mcs; |
a23db284f locking/pvqspinlo... |
82 |
#ifdef CONFIG_PARAVIRT_SPINLOCKS |
0fa809ca7 locking/pvqspinlo... |
83 |
long reserved[2]; |
a23db284f locking/pvqspinlo... |
84 |
#endif |
0fa809ca7 locking/pvqspinlo... |
85 |
}; |
a23db284f locking/pvqspinlo... |
86 |
|
a33fda35e locking/qspinlock... |
87 |
/* |
6512276d9 locking/qspinlock... |
88 89 90 91 92 93 94 95 96 97 98 99 |
* The pending bit spinning loop count. * This heuristic is used to limit the number of lockword accesses * made by atomic_cond_read_relaxed when waiting for the lock to * transition out of the "== _Q_PENDING_VAL" state. We don't spin * indefinitely because there's no guarantee that we'll make forward * progress. */ #ifndef _Q_PENDING_LOOPS #define _Q_PENDING_LOOPS 1 #endif /* |
a33fda35e locking/qspinlock... |
100 101 102 103 |
* Per-CPU queue node structures; we can never have more than 4 nested * contexts: task, softirq, hardirq, nmi. * * Exactly fits one 64-byte cacheline on a 64-bit architecture. |
a23db284f locking/pvqspinlo... |
104 105 |
* * PV doubles the storage and uses the second cacheline for PV state. |
a33fda35e locking/qspinlock... |
106 |
*/ |
0fa809ca7 locking/pvqspinlo... |
107 |
static DEFINE_PER_CPU_ALIGNED(struct qnode, qnodes[MAX_NODES]); |
a33fda35e locking/qspinlock... |
108 109 110 111 112 |
/* * We must be able to distinguish between no-tail and the tail at 0:0, * therefore increment the cpu number by one. */ |
8d53fa190 locking/qspinlock... |
113 |
static inline __pure u32 encode_tail(int cpu, int idx) |
a33fda35e locking/qspinlock... |
114 115 |
{ u32 tail; |
a33fda35e locking/qspinlock... |
116 117 118 119 120 |
tail = (cpu + 1) << _Q_TAIL_CPU_OFFSET; tail |= idx << _Q_TAIL_IDX_OFFSET; /* assume < 4 */ return tail; } |
8d53fa190 locking/qspinlock... |
121 |
static inline __pure struct mcs_spinlock *decode_tail(u32 tail) |
a33fda35e locking/qspinlock... |
122 123 124 |
{ int cpu = (tail >> _Q_TAIL_CPU_OFFSET) - 1; int idx = (tail & _Q_TAIL_IDX_MASK) >> _Q_TAIL_IDX_OFFSET; |
0fa809ca7 locking/pvqspinlo... |
125 126 127 128 129 130 131 |
return per_cpu_ptr(&qnodes[idx].mcs, cpu); } static inline __pure struct mcs_spinlock *grab_mcs_node(struct mcs_spinlock *base, int idx) { return &((struct qnode *)base + idx)->mcs; |
a33fda35e locking/qspinlock... |
132 |
} |
c1fb159db locking/qspinlock... |
133 |
#define _Q_LOCKED_PENDING_MASK (_Q_LOCKED_MASK | _Q_PENDING_MASK) |
2c83e8e94 locking/qspinlock... |
134 |
#if _Q_PENDING_BITS == 8 |
69f9cae90 locking/qspinlock... |
135 |
/** |
59fb586b4 locking/qspinlock... |
136 137 138 139 140 141 142 143 144 145 146 |
* clear_pending - clear the pending bit. * @lock: Pointer to queued spinlock structure * * *,1,* -> *,0,* */ static __always_inline void clear_pending(struct qspinlock *lock) { WRITE_ONCE(lock->pending, 0); } /** |
69f9cae90 locking/qspinlock... |
147 148 149 150 151 152 153 154 155 |
* clear_pending_set_locked - take ownership and clear the pending bit. * @lock: Pointer to queued spinlock structure * * *,1,0 -> *,0,1 * * Lock stealing is not allowed if this function is used. */ static __always_inline void clear_pending_set_locked(struct qspinlock *lock) { |
625e88be1 locking/qspinlock... |
156 |
WRITE_ONCE(lock->locked_pending, _Q_LOCKED_VAL); |
69f9cae90 locking/qspinlock... |
157 158 159 160 161 162 163 164 |
} /* * xchg_tail - Put in the new queue tail code word & retrieve previous one * @lock : Pointer to queued spinlock structure * @tail : The new queue tail code word * Return: The previous queue tail code word * |
548095dea locking: Remove s... |
165 |
* xchg(lock, tail), which heads an address dependency |
69f9cae90 locking/qspinlock... |
166 167 168 169 170 |
* * p,*,* -> n,*,* ; prev = xchg(lock, node) */ static __always_inline u32 xchg_tail(struct qspinlock *lock, u32 tail) { |
64d816cba locking/qspinlock... |
171 |
/* |
9d4646d14 locking/qspinlock... |
172 173 |
* We can use relaxed semantics since the caller ensures that the * MCS node is properly initialized before updating the tail. |
64d816cba locking/qspinlock... |
174 |
*/ |
9d4646d14 locking/qspinlock... |
175 |
return (u32)xchg_relaxed(&lock->tail, |
64d816cba locking/qspinlock... |
176 |
tail >> _Q_TAIL_OFFSET) << _Q_TAIL_OFFSET; |
69f9cae90 locking/qspinlock... |
177 178 179 |
} #else /* _Q_PENDING_BITS == 8 */ |
a33fda35e locking/qspinlock... |
180 |
/** |
59fb586b4 locking/qspinlock... |
181 182 183 184 185 186 187 188 189 190 191 |
* clear_pending - clear the pending bit. * @lock: Pointer to queued spinlock structure * * *,1,* -> *,0,* */ static __always_inline void clear_pending(struct qspinlock *lock) { atomic_andnot(_Q_PENDING_VAL, &lock->val); } /** |
6403bd7d0 locking/qspinlock... |
192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 |
* clear_pending_set_locked - take ownership and clear the pending bit. * @lock: Pointer to queued spinlock structure * * *,1,0 -> *,0,1 */ static __always_inline void clear_pending_set_locked(struct qspinlock *lock) { atomic_add(-_Q_PENDING_VAL + _Q_LOCKED_VAL, &lock->val); } /** * xchg_tail - Put in the new queue tail code word & retrieve previous one * @lock : Pointer to queued spinlock structure * @tail : The new queue tail code word * Return: The previous queue tail code word * * xchg(lock, tail) * * p,*,* -> n,*,* ; prev = xchg(lock, node) */ static __always_inline u32 xchg_tail(struct qspinlock *lock, u32 tail) { u32 old, new, val = atomic_read(&lock->val); for (;;) { new = (val & _Q_LOCKED_PENDING_MASK) | tail; |
64d816cba locking/qspinlock... |
218 |
/* |
9d4646d14 locking/qspinlock... |
219 220 221 |
* We can use relaxed semantics since the caller ensures that * the MCS node is properly initialized before updating the * tail. |
64d816cba locking/qspinlock... |
222 |
*/ |
9d4646d14 locking/qspinlock... |
223 |
old = atomic_cmpxchg_relaxed(&lock->val, val, new); |
6403bd7d0 locking/qspinlock... |
224 225 226 227 228 229 230 |
if (old == val) break; val = old; } return old; } |
69f9cae90 locking/qspinlock... |
231 |
#endif /* _Q_PENDING_BITS == 8 */ |
6403bd7d0 locking/qspinlock... |
232 233 |
/** |
7aa54be29 locking/qspinlock... |
234 235 236 237 238 239 240 241 242 243 244 245 246 247 |
* queued_fetch_set_pending_acquire - fetch the whole lock value and set pending * @lock : Pointer to queued spinlock structure * Return: The previous lock value * * *,*,* -> *,1,* */ #ifndef queued_fetch_set_pending_acquire static __always_inline u32 queued_fetch_set_pending_acquire(struct qspinlock *lock) { return atomic_fetch_or_acquire(_Q_PENDING_VAL, &lock->val); } #endif /** |
2c83e8e94 locking/qspinlock... |
248 249 250 251 252 253 254 |
* set_locked - Set the lock bit and own the lock * @lock: Pointer to queued spinlock structure * * *,*,0 -> *,0,1 */ static __always_inline void set_locked(struct qspinlock *lock) { |
625e88be1 locking/qspinlock... |
255 |
WRITE_ONCE(lock->locked, _Q_LOCKED_VAL); |
2c83e8e94 locking/qspinlock... |
256 |
} |
a23db284f locking/pvqspinlo... |
257 258 259 260 261 262 263 |
/* * Generate the native code for queued_spin_unlock_slowpath(); provide NOPs for * all the PV callbacks. */ static __always_inline void __pv_init_node(struct mcs_spinlock *node) { } |
cd0272fab locking/pvqspinlo... |
264 265 |
static __always_inline void __pv_wait_node(struct mcs_spinlock *node, struct mcs_spinlock *prev) { } |
75d227028 locking/pvqspinlo... |
266 267 |
static __always_inline void __pv_kick_node(struct qspinlock *lock, struct mcs_spinlock *node) { } |
1c4941fd5 locking/pvqspinlo... |
268 269 270 |
static __always_inline u32 __pv_wait_head_or_lock(struct qspinlock *lock, struct mcs_spinlock *node) { return 0; } |
a23db284f locking/pvqspinlo... |
271 272 273 274 275 276 |
#define pv_enabled() false #define pv_init_node __pv_init_node #define pv_wait_node __pv_wait_node #define pv_kick_node __pv_kick_node |
1c4941fd5 locking/pvqspinlo... |
277 |
#define pv_wait_head_or_lock __pv_wait_head_or_lock |
a23db284f locking/pvqspinlo... |
278 279 280 281 282 283 |
#ifdef CONFIG_PARAVIRT_SPINLOCKS #define queued_spin_lock_slowpath native_queued_spin_lock_slowpath #endif #endif /* _GEN_PV_LOCK_SLOWPATH */ |
2c83e8e94 locking/qspinlock... |
284 |
/** |
a33fda35e locking/qspinlock... |
285 286 287 288 |
* queued_spin_lock_slowpath - acquire the queued spinlock * @lock: Pointer to queued spinlock structure * @val: Current value of the queued spinlock 32-bit word * |
c1fb159db locking/qspinlock... |
289 |
* (queue tail, pending bit, lock value) |
a33fda35e locking/qspinlock... |
290 |
* |
c1fb159db locking/qspinlock... |
291 292 293 294 295 296 297 298 299 300 301 302 303 |
* fast : slow : unlock * : : * uncontended (0,0,0) -:--> (0,0,1) ------------------------------:--> (*,*,0) * : | ^--------.------. / : * : v \ \ | : * pending : (0,1,1) +--> (0,1,0) \ | : * : | ^--' | | : * : v | | : * uncontended : (n,x,y) +--> (n,0,0) --' | : * queue : | ^--' | : * : v | : * contended : (*,x,y) +--> (*,0,0) ---> (*,0,1) -' : * queue : ^--' : |
a33fda35e locking/qspinlock... |
304 305 306 307 |
*/ void queued_spin_lock_slowpath(struct qspinlock *lock, u32 val) { struct mcs_spinlock *prev, *next, *node; |
59fb586b4 locking/qspinlock... |
308 |
u32 old, tail; |
a33fda35e locking/qspinlock... |
309 310 311 |
int idx; BUILD_BUG_ON(CONFIG_NR_CPUS >= (1U << _Q_TAIL_CPU_BITS)); |
a23db284f locking/pvqspinlo... |
312 |
if (pv_enabled()) |
81d3dc9a3 locking/qspinlock... |
313 |
goto pv_queue; |
a23db284f locking/pvqspinlo... |
314 |
|
43b3f0289 locking/qspinlock... |
315 |
if (virt_spin_lock(lock)) |
2aa79af64 locking/qspinlock... |
316 |
return; |
c1fb159db locking/qspinlock... |
317 |
/* |
6512276d9 locking/qspinlock... |
318 319 |
* Wait for in-progress pending->locked hand-overs with a bounded * number of spins so that we guarantee forward progress. |
c1fb159db locking/qspinlock... |
320 321 322 323 |
* * 0,1,0 -> 0,0,1 */ if (val == _Q_PENDING_VAL) { |
6512276d9 locking/qspinlock... |
324 325 326 |
int cnt = _Q_PENDING_LOOPS; val = atomic_cond_read_relaxed(&lock->val, (VAL != _Q_PENDING_VAL) || !cnt--); |
c1fb159db locking/qspinlock... |
327 328 329 |
} /* |
59fb586b4 locking/qspinlock... |
330 331 332 333 334 335 |
* If we observe any contention; queue. */ if (val & ~_Q_LOCKED_MASK) goto queue; /* |
c1fb159db locking/qspinlock... |
336 337 |
* trylock || pending * |
756b1df4c locking/qspinlock... |
338 |
* 0,0,* -> 0,1,* -> 0,0,1 pending, trylock |
c1fb159db locking/qspinlock... |
339 |
*/ |
7aa54be29 locking/qspinlock... |
340 |
val = queued_fetch_set_pending_acquire(lock); |
756b1df4c locking/qspinlock... |
341 |
|
53bf57fab locking/qspinlock... |
342 |
/* |
756b1df4c locking/qspinlock... |
343 344 345 346 347 |
* If we observe contention, there is a concurrent locker. * * Undo and queue; our setting of PENDING might have made the * n,0,0 -> 0,0,0 transition fail and it will now be waiting * on @next to become !NULL. |
53bf57fab locking/qspinlock... |
348 349 |
*/ if (unlikely(val & ~_Q_LOCKED_MASK)) { |
756b1df4c locking/qspinlock... |
350 351 |
/* Undo PENDING if we set it. */ |
53bf57fab locking/qspinlock... |
352 353 |
if (!(val & _Q_PENDING_MASK)) clear_pending(lock); |
756b1df4c locking/qspinlock... |
354 |
|
53bf57fab locking/qspinlock... |
355 |
goto queue; |
59fb586b4 locking/qspinlock... |
356 |
} |
c1fb159db locking/qspinlock... |
357 358 |
/* |
53bf57fab locking/qspinlock... |
359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 |
* We're pending, wait for the owner to go away. * * 0,1,1 -> 0,1,0 * * this wait loop must be a load-acquire such that we match the * store-release that clears the locked bit and create lock * sequentiality; this is because not all * clear_pending_set_locked() implementations imply full * barriers. */ if (val & _Q_LOCKED_MASK) atomic_cond_read_acquire(&lock->val, !(VAL & _Q_LOCKED_MASK)); /* * take ownership and clear the pending bit. * * 0,1,0 -> 0,0,1 |
c1fb159db locking/qspinlock... |
376 |
*/ |
53bf57fab locking/qspinlock... |
377 |
clear_pending_set_locked(lock); |
ad53fa10f locking/qspinlock... |
378 |
lockevent_inc(lock_pending); |
53bf57fab locking/qspinlock... |
379 |
return; |
c1fb159db locking/qspinlock... |
380 381 382 383 384 385 |
/* * End of pending bit optimistic spinning and beginning of MCS * queuing. */ queue: |
ad53fa10f locking/qspinlock... |
386 |
lockevent_inc(lock_slowpath); |
81d3dc9a3 locking/qspinlock... |
387 |
pv_queue: |
0fa809ca7 locking/pvqspinlo... |
388 |
node = this_cpu_ptr(&qnodes[0].mcs); |
a33fda35e locking/qspinlock... |
389 390 |
idx = node->count++; tail = encode_tail(smp_processor_id(), idx); |
d682b596d locking/qspinlock... |
391 392 393 394 395 396 397 398 399 400 |
/* * 4 nodes are allocated based on the assumption that there will * not be nested NMIs taking spinlocks. That may not be true in * some architectures even though the chance of needing more than * 4 nodes will still be extremely unlikely. When that happens, * we fall back to spinning on the lock directly without using * any MCS node. This is not the most elegant solution, but is * simple enough. */ if (unlikely(idx >= MAX_NODES)) { |
ad53fa10f locking/qspinlock... |
401 |
lockevent_inc(lock_no_node); |
d682b596d locking/qspinlock... |
402 403 404 405 |
while (!queued_spin_trylock(lock)) cpu_relax(); goto release; } |
0fa809ca7 locking/pvqspinlo... |
406 |
node = grab_mcs_node(node, idx); |
11dc13224 locking/qspinlock... |
407 408 |
/* |
1222109a5 locking/qspinlock... |
409 410 |
* Keep counts of non-zero index values: */ |
ad53fa10f locking/qspinlock... |
411 |
lockevent_cond_inc(lock_use_node2 + idx - 1, idx); |
1222109a5 locking/qspinlock... |
412 413 |
/* |
11dc13224 locking/qspinlock... |
414 415 416 417 418 |
* Ensure that we increment the head node->count before initialising * the actual node. If the compiler is kind enough to reorder these * stores, then an IRQ could overwrite our assignments. */ barrier(); |
a33fda35e locking/qspinlock... |
419 420 |
node->locked = 0; node->next = NULL; |
a23db284f locking/pvqspinlo... |
421 |
pv_init_node(node); |
a33fda35e locking/qspinlock... |
422 423 |
/* |
6403bd7d0 locking/qspinlock... |
424 425 426 |
* We touched a (possibly) cold cacheline in the per-cpu queue node; * attempt the trylock once more in the hope someone let go while we * weren't watching. |
a33fda35e locking/qspinlock... |
427 |
*/ |
6403bd7d0 locking/qspinlock... |
428 429 |
if (queued_spin_trylock(lock)) goto release; |
a33fda35e locking/qspinlock... |
430 431 |
/* |
9d4646d14 locking/qspinlock... |
432 433 434 435 436 437 438 439 |
* Ensure that the initialisation of @node is complete before we * publish the updated tail via xchg_tail() and potentially link * @node into the waitqueue via WRITE_ONCE(prev->next, node) below. */ smp_wmb(); /* * Publish the updated tail. |
6403bd7d0 locking/qspinlock... |
440 441 442 443 |
* We have already touched the queueing cacheline; don't bother with * pending stuff. * * p,*,* -> n,*,* |
a33fda35e locking/qspinlock... |
444 |
*/ |
6403bd7d0 locking/qspinlock... |
445 |
old = xchg_tail(lock, tail); |
aa68744f8 locking/qspinlock... |
446 |
next = NULL; |
a33fda35e locking/qspinlock... |
447 448 449 450 451 |
/* * if there was a previous node; link it and wait until reaching the * head of the waitqueue. */ |
6403bd7d0 locking/qspinlock... |
452 |
if (old & _Q_TAIL_MASK) { |
a33fda35e locking/qspinlock... |
453 |
prev = decode_tail(old); |
95bcade33 locking/qspinlock... |
454 |
|
9d4646d14 locking/qspinlock... |
455 456 |
/* Link @node into the waitqueue. */ WRITE_ONCE(prev->next, node); |
a33fda35e locking/qspinlock... |
457 |
|
cd0272fab locking/pvqspinlo... |
458 |
pv_wait_node(node, prev); |
a33fda35e locking/qspinlock... |
459 |
arch_mcs_spin_lock_contended(&node->locked); |
81b559866 locking/qspinlock... |
460 461 462 463 464 465 466 467 468 469 |
/* * While waiting for the MCS lock, the next pointer may have * been set by another lock waiter. We optimistically load * the next pointer & prefetch the cacheline for writing * to reduce latency in the upcoming MCS unlock operation. */ next = READ_ONCE(node->next); if (next) prefetchw(next); |
a33fda35e locking/qspinlock... |
470 471 472 |
} /* |
c1fb159db locking/qspinlock... |
473 474 |
* we're at the head of the waitqueue, wait for the owner & pending to * go away. |
a33fda35e locking/qspinlock... |
475 |
* |
c1fb159db locking/qspinlock... |
476 |
* *,x,y -> *,0,0 |
2c83e8e94 locking/qspinlock... |
477 478 479 480 481 482 |
* * this wait loop must use a load-acquire such that we match the * store-release that clears the locked bit and create lock * sequentiality; this is because the set_locked() function below * does not imply a full barrier. * |
1c4941fd5 locking/pvqspinlo... |
483 484 |
* The PV pv_wait_head_or_lock function, if active, will acquire * the lock and return a non-zero value. So we have to skip the |
f9c811fac locking/qspinlock... |
485 486 |
* atomic_cond_read_acquire() call. As the next PV queue head hasn't * been designated yet, there is no way for the locked value to become |
1c4941fd5 locking/pvqspinlo... |
487 488 489 490 491 |
* _Q_SLOW_VAL. So both the set_locked() and the * atomic_cmpxchg_relaxed() calls will be safe. * * If PV isn't active, 0 will be returned instead. * |
a33fda35e locking/qspinlock... |
492 |
*/ |
1c4941fd5 locking/pvqspinlo... |
493 494 |
if ((val = pv_wait_head_or_lock(lock, node))) goto locked; |
f9c811fac locking/qspinlock... |
495 |
val = atomic_cond_read_acquire(&lock->val, !(VAL & _Q_LOCKED_PENDING_MASK)); |
a33fda35e locking/qspinlock... |
496 |
|
1c4941fd5 locking/pvqspinlo... |
497 |
locked: |
a33fda35e locking/qspinlock... |
498 499 500 |
/* * claim the lock: * |
c1fb159db locking/qspinlock... |
501 |
* n,0,0 -> 0,0,1 : lock, uncontended |
59fb586b4 locking/qspinlock... |
502 |
* *,*,0 -> *,*,1 : lock, contended |
2c83e8e94 locking/qspinlock... |
503 |
* |
59fb586b4 locking/qspinlock... |
504 505 506 |
* If the queue head is the only one in the queue (lock value == tail) * and nobody is pending, clear the tail code and grab the lock. * Otherwise, we only need to grab the lock. |
a33fda35e locking/qspinlock... |
507 |
*/ |
c61da58d8 locking/qspinlock... |
508 |
|
ae75d9089 locking/qspinlock... |
509 |
/* |
756b1df4c locking/qspinlock... |
510 511 |
* In the PV case we might already have _Q_LOCKED_VAL set, because * of lock stealing; therefore we must also allow: |
ae75d9089 locking/qspinlock... |
512 |
* |
756b1df4c locking/qspinlock... |
513 514 515 516 517 |
* n,0,1 -> 0,0,1 * * Note: at this point: (val & _Q_PENDING_MASK) == 0, because of the * above wait condition, therefore any concurrent setting of * PENDING will make the uncontended transition fail. |
ae75d9089 locking/qspinlock... |
518 |
*/ |
756b1df4c locking/qspinlock... |
519 520 521 522 |
if ((val & _Q_TAIL_MASK) == tail) { if (atomic_try_cmpxchg_relaxed(&lock->val, &val, _Q_LOCKED_VAL)) goto release; /* No contention */ } |
a33fda35e locking/qspinlock... |
523 |
|
756b1df4c locking/qspinlock... |
524 525 526 527 528 |
/* * Either somebody is queued behind us or _Q_PENDING_VAL got set * which will then detect the remaining tail and queue behind us * ensuring we'll see a @next. */ |
c61da58d8 locking/qspinlock... |
529 |
set_locked(lock); |
a33fda35e locking/qspinlock... |
530 |
/* |
aa68744f8 locking/qspinlock... |
531 |
* contended path; wait for next if not observed yet, release. |
a33fda35e locking/qspinlock... |
532 |
*/ |
c131a198c locking/qspinlock... |
533 534 |
if (!next) next = smp_cond_load_relaxed(&node->next, (VAL)); |
a33fda35e locking/qspinlock... |
535 |
|
2c83e8e94 locking/qspinlock... |
536 |
arch_mcs_spin_unlock_contended(&next->locked); |
75d227028 locking/pvqspinlo... |
537 |
pv_kick_node(lock, next); |
a33fda35e locking/qspinlock... |
538 539 540 541 542 |
release: /* * release the node */ |
0fa809ca7 locking/pvqspinlo... |
543 |
__this_cpu_dec(qnodes[0].mcs.count); |
a33fda35e locking/qspinlock... |
544 545 |
} EXPORT_SYMBOL(queued_spin_lock_slowpath); |
a23db284f locking/pvqspinlo... |
546 547 548 549 550 551 552 553 554 555 556 557 558 |
/* * Generate the paravirt code for queued_spin_unlock_slowpath(). */ #if !defined(_GEN_PV_LOCK_SLOWPATH) && defined(CONFIG_PARAVIRT_SPINLOCKS) #define _GEN_PV_LOCK_SLOWPATH #undef pv_enabled #define pv_enabled() true #undef pv_init_node #undef pv_wait_node #undef pv_kick_node |
1c4941fd5 locking/pvqspinlo... |
559 |
#undef pv_wait_head_or_lock |
a23db284f locking/pvqspinlo... |
560 561 562 563 564 565 |
#undef queued_spin_lock_slowpath #define queued_spin_lock_slowpath __pv_queued_spin_lock_slowpath #include "qspinlock_paravirt.h" #include "qspinlock.c" |
05eee619e x86/kvm: Add "nop... |
566 567 568 569 570 571 572 |
bool nopvspin __initdata; static __init int parse_nopvspin(char *arg) { nopvspin = true; return 0; } early_param("nopvspin", parse_nopvspin); |
a23db284f locking/pvqspinlo... |
573 |
#endif |