Michael S. Tsirkin | 2e0ab8c | 2016-06-13 23:54:31 +0300 | [diff] [blame] | 1 | /* |
| 2 | * Definitions for the 'struct ptr_ring' datastructure. |
| 3 | * |
| 4 | * Author: |
| 5 | * Michael S. Tsirkin <mst@redhat.com> |
| 6 | * |
| 7 | * Copyright (C) 2016 Red Hat, Inc. |
| 8 | * |
| 9 | * This program is free software; you can redistribute it and/or modify it |
| 10 | * under the terms of the GNU General Public License as published by the |
| 11 | * Free Software Foundation; either version 2 of the License, or (at your |
| 12 | * option) any later version. |
| 13 | * |
| 14 | * This is a limited-size FIFO maintaining pointers in FIFO order, with |
| 15 | * one CPU producing entries and another consuming entries from a FIFO. |
| 16 | * |
| 17 | * This implementation tries to minimize cache-contention when there is a |
| 18 | * single producer and a single consumer CPU. |
| 19 | */ |
| 20 | |
| 21 | #ifndef _LINUX_PTR_RING_H |
| 22 | #define _LINUX_PTR_RING_H 1 |
| 23 | |
| 24 | #ifdef __KERNEL__ |
| 25 | #include <linux/spinlock.h> |
| 26 | #include <linux/cache.h> |
| 27 | #include <linux/types.h> |
| 28 | #include <linux/compiler.h> |
| 29 | #include <linux/cache.h> |
| 30 | #include <linux/slab.h> |
| 31 | #include <asm/errno.h> |
| 32 | #endif |
| 33 | |
| 34 | struct ptr_ring { |
| 35 | int producer ____cacheline_aligned_in_smp; |
| 36 | spinlock_t producer_lock; |
| 37 | int consumer ____cacheline_aligned_in_smp; |
| 38 | spinlock_t consumer_lock; |
| 39 | /* Shared consumer/producer data */ |
| 40 | /* Read-only by both the producer and the consumer */ |
| 41 | int size ____cacheline_aligned_in_smp; /* max entries in queue */ |
| 42 | void **queue; |
| 43 | }; |
| 44 | |
| 45 | /* Note: callers invoking this in a loop must use a compiler barrier, |
Michael S. Tsirkin | 5d49de5 | 2016-06-13 23:54:45 +0300 | [diff] [blame^] | 46 | * for example cpu_relax(). If ring is ever resized, callers must hold |
| 47 | * producer_lock - see e.g. ptr_ring_full. Otherwise, if callers don't hold |
| 48 | * producer_lock, the next call to __ptr_ring_produce may fail. |
Michael S. Tsirkin | 2e0ab8c | 2016-06-13 23:54:31 +0300 | [diff] [blame] | 49 | */ |
| 50 | static inline bool __ptr_ring_full(struct ptr_ring *r) |
| 51 | { |
| 52 | return r->queue[r->producer]; |
| 53 | } |
| 54 | |
| 55 | static inline bool ptr_ring_full(struct ptr_ring *r) |
| 56 | { |
Michael S. Tsirkin | 5d49de5 | 2016-06-13 23:54:45 +0300 | [diff] [blame^] | 57 | bool ret; |
| 58 | |
| 59 | spin_lock(&r->producer_lock); |
| 60 | ret = __ptr_ring_full(r); |
| 61 | spin_unlock(&r->producer_lock); |
| 62 | |
| 63 | return ret; |
| 64 | } |
| 65 | |
| 66 | static inline bool ptr_ring_full_irq(struct ptr_ring *r) |
| 67 | { |
| 68 | bool ret; |
| 69 | |
| 70 | spin_lock_irq(&r->producer_lock); |
| 71 | ret = __ptr_ring_full(r); |
| 72 | spin_unlock_irq(&r->producer_lock); |
| 73 | |
| 74 | return ret; |
| 75 | } |
| 76 | |
| 77 | static inline bool ptr_ring_full_any(struct ptr_ring *r) |
| 78 | { |
| 79 | unsigned long flags; |
| 80 | bool ret; |
| 81 | |
| 82 | spin_lock_irqsave(&r->producer_lock, flags); |
| 83 | ret = __ptr_ring_full(r); |
| 84 | spin_unlock_irqrestore(&r->producer_lock, flags); |
| 85 | |
| 86 | return ret; |
| 87 | } |
| 88 | |
| 89 | static inline bool ptr_ring_full_bh(struct ptr_ring *r) |
| 90 | { |
| 91 | bool ret; |
| 92 | |
| 93 | spin_lock_bh(&r->producer_lock); |
| 94 | ret = __ptr_ring_full(r); |
| 95 | spin_unlock_bh(&r->producer_lock); |
| 96 | |
| 97 | return ret; |
Michael S. Tsirkin | 2e0ab8c | 2016-06-13 23:54:31 +0300 | [diff] [blame] | 98 | } |
| 99 | |
| 100 | /* Note: callers invoking this in a loop must use a compiler barrier, |
Michael S. Tsirkin | 5d49de5 | 2016-06-13 23:54:45 +0300 | [diff] [blame^] | 101 | * for example cpu_relax(). Callers must hold producer_lock. |
Michael S. Tsirkin | 2e0ab8c | 2016-06-13 23:54:31 +0300 | [diff] [blame] | 102 | */ |
| 103 | static inline int __ptr_ring_produce(struct ptr_ring *r, void *ptr) |
| 104 | { |
Michael S. Tsirkin | 5d49de5 | 2016-06-13 23:54:45 +0300 | [diff] [blame^] | 105 | if (r->queue[r->producer]) |
Michael S. Tsirkin | 2e0ab8c | 2016-06-13 23:54:31 +0300 | [diff] [blame] | 106 | return -ENOSPC; |
| 107 | |
| 108 | r->queue[r->producer++] = ptr; |
| 109 | if (unlikely(r->producer >= r->size)) |
| 110 | r->producer = 0; |
| 111 | return 0; |
| 112 | } |
| 113 | |
| 114 | static inline int ptr_ring_produce(struct ptr_ring *r, void *ptr) |
| 115 | { |
| 116 | int ret; |
| 117 | |
| 118 | spin_lock(&r->producer_lock); |
| 119 | ret = __ptr_ring_produce(r, ptr); |
| 120 | spin_unlock(&r->producer_lock); |
| 121 | |
| 122 | return ret; |
| 123 | } |
| 124 | |
| 125 | static inline int ptr_ring_produce_irq(struct ptr_ring *r, void *ptr) |
| 126 | { |
| 127 | int ret; |
| 128 | |
| 129 | spin_lock_irq(&r->producer_lock); |
| 130 | ret = __ptr_ring_produce(r, ptr); |
| 131 | spin_unlock_irq(&r->producer_lock); |
| 132 | |
| 133 | return ret; |
| 134 | } |
| 135 | |
| 136 | static inline int ptr_ring_produce_any(struct ptr_ring *r, void *ptr) |
| 137 | { |
| 138 | unsigned long flags; |
| 139 | int ret; |
| 140 | |
| 141 | spin_lock_irqsave(&r->producer_lock, flags); |
| 142 | ret = __ptr_ring_produce(r, ptr); |
| 143 | spin_unlock_irqrestore(&r->producer_lock, flags); |
| 144 | |
| 145 | return ret; |
| 146 | } |
| 147 | |
| 148 | static inline int ptr_ring_produce_bh(struct ptr_ring *r, void *ptr) |
| 149 | { |
| 150 | int ret; |
| 151 | |
| 152 | spin_lock_bh(&r->producer_lock); |
| 153 | ret = __ptr_ring_produce(r, ptr); |
| 154 | spin_unlock_bh(&r->producer_lock); |
| 155 | |
| 156 | return ret; |
| 157 | } |
| 158 | |
| 159 | /* Note: callers invoking this in a loop must use a compiler barrier, |
| 160 | * for example cpu_relax(). Callers must take consumer_lock |
| 161 | * if they dereference the pointer - see e.g. PTR_RING_PEEK_CALL. |
Michael S. Tsirkin | 5d49de5 | 2016-06-13 23:54:45 +0300 | [diff] [blame^] | 162 | * If ring is never resized, and if the pointer is merely |
| 163 | * tested, there's no need to take the lock - see e.g. __ptr_ring_empty. |
Michael S. Tsirkin | 2e0ab8c | 2016-06-13 23:54:31 +0300 | [diff] [blame] | 164 | */ |
| 165 | static inline void *__ptr_ring_peek(struct ptr_ring *r) |
| 166 | { |
| 167 | return r->queue[r->consumer]; |
| 168 | } |
| 169 | |
Michael S. Tsirkin | 5d49de5 | 2016-06-13 23:54:45 +0300 | [diff] [blame^] | 170 | /* Note: callers invoking this in a loop must use a compiler barrier, |
| 171 | * for example cpu_relax(). Callers must take consumer_lock |
| 172 | * if the ring is ever resized - see e.g. ptr_ring_empty. |
| 173 | */ |
| 174 | static inline bool __ptr_ring_empty(struct ptr_ring *r) |
| 175 | { |
| 176 | return !__ptr_ring_peek(r); |
| 177 | } |
| 178 | |
Michael S. Tsirkin | 2e0ab8c | 2016-06-13 23:54:31 +0300 | [diff] [blame] | 179 | static inline bool ptr_ring_empty(struct ptr_ring *r) |
| 180 | { |
Michael S. Tsirkin | 5d49de5 | 2016-06-13 23:54:45 +0300 | [diff] [blame^] | 181 | bool ret; |
| 182 | |
| 183 | spin_lock(&r->consumer_lock); |
| 184 | ret = __ptr_ring_empty(r); |
| 185 | spin_unlock(&r->consumer_lock); |
| 186 | |
| 187 | return ret; |
| 188 | } |
| 189 | |
| 190 | static inline bool ptr_ring_empty_irq(struct ptr_ring *r) |
| 191 | { |
| 192 | bool ret; |
| 193 | |
| 194 | spin_lock_irq(&r->consumer_lock); |
| 195 | ret = __ptr_ring_empty(r); |
| 196 | spin_unlock_irq(&r->consumer_lock); |
| 197 | |
| 198 | return ret; |
| 199 | } |
| 200 | |
| 201 | static inline bool ptr_ring_empty_any(struct ptr_ring *r) |
| 202 | { |
| 203 | unsigned long flags; |
| 204 | bool ret; |
| 205 | |
| 206 | spin_lock_irqsave(&r->consumer_lock, flags); |
| 207 | ret = __ptr_ring_empty(r); |
| 208 | spin_unlock_irqrestore(&r->consumer_lock, flags); |
| 209 | |
| 210 | return ret; |
| 211 | } |
| 212 | |
| 213 | static inline bool ptr_ring_empty_bh(struct ptr_ring *r) |
| 214 | { |
| 215 | bool ret; |
| 216 | |
| 217 | spin_lock_bh(&r->consumer_lock); |
| 218 | ret = __ptr_ring_empty(r); |
| 219 | spin_unlock_bh(&r->consumer_lock); |
| 220 | |
| 221 | return ret; |
Michael S. Tsirkin | 2e0ab8c | 2016-06-13 23:54:31 +0300 | [diff] [blame] | 222 | } |
| 223 | |
| 224 | /* Must only be called after __ptr_ring_peek returned !NULL */ |
| 225 | static inline void __ptr_ring_discard_one(struct ptr_ring *r) |
| 226 | { |
| 227 | r->queue[r->consumer++] = NULL; |
| 228 | if (unlikely(r->consumer >= r->size)) |
| 229 | r->consumer = 0; |
| 230 | } |
| 231 | |
| 232 | static inline void *__ptr_ring_consume(struct ptr_ring *r) |
| 233 | { |
| 234 | void *ptr; |
| 235 | |
| 236 | ptr = __ptr_ring_peek(r); |
| 237 | if (ptr) |
| 238 | __ptr_ring_discard_one(r); |
| 239 | |
| 240 | return ptr; |
| 241 | } |
| 242 | |
| 243 | static inline void *ptr_ring_consume(struct ptr_ring *r) |
| 244 | { |
| 245 | void *ptr; |
| 246 | |
| 247 | spin_lock(&r->consumer_lock); |
| 248 | ptr = __ptr_ring_consume(r); |
| 249 | spin_unlock(&r->consumer_lock); |
| 250 | |
| 251 | return ptr; |
| 252 | } |
| 253 | |
| 254 | static inline void *ptr_ring_consume_irq(struct ptr_ring *r) |
| 255 | { |
| 256 | void *ptr; |
| 257 | |
| 258 | spin_lock_irq(&r->consumer_lock); |
| 259 | ptr = __ptr_ring_consume(r); |
| 260 | spin_unlock_irq(&r->consumer_lock); |
| 261 | |
| 262 | return ptr; |
| 263 | } |
| 264 | |
| 265 | static inline void *ptr_ring_consume_any(struct ptr_ring *r) |
| 266 | { |
| 267 | unsigned long flags; |
| 268 | void *ptr; |
| 269 | |
| 270 | spin_lock_irqsave(&r->consumer_lock, flags); |
| 271 | ptr = __ptr_ring_consume(r); |
| 272 | spin_unlock_irqrestore(&r->consumer_lock, flags); |
| 273 | |
| 274 | return ptr; |
| 275 | } |
| 276 | |
| 277 | static inline void *ptr_ring_consume_bh(struct ptr_ring *r) |
| 278 | { |
| 279 | void *ptr; |
| 280 | |
| 281 | spin_lock_bh(&r->consumer_lock); |
| 282 | ptr = __ptr_ring_consume(r); |
| 283 | spin_unlock_bh(&r->consumer_lock); |
| 284 | |
| 285 | return ptr; |
| 286 | } |
| 287 | |
| 288 | /* Cast to structure type and call a function without discarding from FIFO. |
| 289 | * Function must return a value. |
| 290 | * Callers must take consumer_lock. |
| 291 | */ |
| 292 | #define __PTR_RING_PEEK_CALL(r, f) ((f)(__ptr_ring_peek(r))) |
| 293 | |
| 294 | #define PTR_RING_PEEK_CALL(r, f) ({ \ |
| 295 | typeof((f)(NULL)) __PTR_RING_PEEK_CALL_v; \ |
| 296 | \ |
| 297 | spin_lock(&(r)->consumer_lock); \ |
| 298 | __PTR_RING_PEEK_CALL_v = __PTR_RING_PEEK_CALL(r, f); \ |
| 299 | spin_unlock(&(r)->consumer_lock); \ |
| 300 | __PTR_RING_PEEK_CALL_v; \ |
| 301 | }) |
| 302 | |
| 303 | #define PTR_RING_PEEK_CALL_IRQ(r, f) ({ \ |
| 304 | typeof((f)(NULL)) __PTR_RING_PEEK_CALL_v; \ |
| 305 | \ |
| 306 | spin_lock_irq(&(r)->consumer_lock); \ |
| 307 | __PTR_RING_PEEK_CALL_v = __PTR_RING_PEEK_CALL(r, f); \ |
| 308 | spin_unlock_irq(&(r)->consumer_lock); \ |
| 309 | __PTR_RING_PEEK_CALL_v; \ |
| 310 | }) |
| 311 | |
| 312 | #define PTR_RING_PEEK_CALL_BH(r, f) ({ \ |
| 313 | typeof((f)(NULL)) __PTR_RING_PEEK_CALL_v; \ |
| 314 | \ |
| 315 | spin_lock_bh(&(r)->consumer_lock); \ |
| 316 | __PTR_RING_PEEK_CALL_v = __PTR_RING_PEEK_CALL(r, f); \ |
| 317 | spin_unlock_bh(&(r)->consumer_lock); \ |
| 318 | __PTR_RING_PEEK_CALL_v; \ |
| 319 | }) |
| 320 | |
| 321 | #define PTR_RING_PEEK_CALL_ANY(r, f) ({ \ |
| 322 | typeof((f)(NULL)) __PTR_RING_PEEK_CALL_v; \ |
| 323 | unsigned long __PTR_RING_PEEK_CALL_f;\ |
| 324 | \ |
| 325 | spin_lock_irqsave(&(r)->consumer_lock, __PTR_RING_PEEK_CALL_f); \ |
| 326 | __PTR_RING_PEEK_CALL_v = __PTR_RING_PEEK_CALL(r, f); \ |
| 327 | spin_unlock_irqrestore(&(r)->consumer_lock, __PTR_RING_PEEK_CALL_f); \ |
| 328 | __PTR_RING_PEEK_CALL_v; \ |
| 329 | }) |
| 330 | |
Michael S. Tsirkin | 5d49de5 | 2016-06-13 23:54:45 +0300 | [diff] [blame^] | 331 | static inline void **__ptr_ring_init_queue_alloc(int size, gfp_t gfp) |
| 332 | { |
| 333 | return kzalloc(ALIGN(size * sizeof(void *), SMP_CACHE_BYTES), gfp); |
| 334 | } |
| 335 | |
Michael S. Tsirkin | 2e0ab8c | 2016-06-13 23:54:31 +0300 | [diff] [blame] | 336 | static inline int ptr_ring_init(struct ptr_ring *r, int size, gfp_t gfp) |
| 337 | { |
Michael S. Tsirkin | 5d49de5 | 2016-06-13 23:54:45 +0300 | [diff] [blame^] | 338 | r->queue = __ptr_ring_init_queue_alloc(size, gfp); |
Michael S. Tsirkin | 2e0ab8c | 2016-06-13 23:54:31 +0300 | [diff] [blame] | 339 | if (!r->queue) |
| 340 | return -ENOMEM; |
| 341 | |
| 342 | r->size = size; |
| 343 | r->producer = r->consumer = 0; |
| 344 | spin_lock_init(&r->producer_lock); |
| 345 | spin_lock_init(&r->consumer_lock); |
| 346 | |
| 347 | return 0; |
| 348 | } |
| 349 | |
Michael S. Tsirkin | 5d49de5 | 2016-06-13 23:54:45 +0300 | [diff] [blame^] | 350 | static inline int ptr_ring_resize(struct ptr_ring *r, int size, gfp_t gfp, |
| 351 | void (*destroy)(void *)) |
Michael S. Tsirkin | 2e0ab8c | 2016-06-13 23:54:31 +0300 | [diff] [blame] | 352 | { |
Michael S. Tsirkin | 5d49de5 | 2016-06-13 23:54:45 +0300 | [diff] [blame^] | 353 | unsigned long flags; |
| 354 | int producer = 0; |
| 355 | void **queue = __ptr_ring_init_queue_alloc(size, gfp); |
| 356 | void **old; |
| 357 | void *ptr; |
| 358 | |
| 359 | if (!queue) |
| 360 | return -ENOMEM; |
| 361 | |
| 362 | spin_lock_irqsave(&(r)->producer_lock, flags); |
| 363 | |
| 364 | while ((ptr = ptr_ring_consume(r))) |
| 365 | if (producer < size) |
| 366 | queue[producer++] = ptr; |
| 367 | else if (destroy) |
| 368 | destroy(ptr); |
| 369 | |
| 370 | r->size = size; |
| 371 | r->producer = producer; |
| 372 | r->consumer = 0; |
| 373 | old = r->queue; |
| 374 | r->queue = queue; |
| 375 | |
| 376 | spin_unlock_irqrestore(&(r)->producer_lock, flags); |
| 377 | |
| 378 | kfree(old); |
| 379 | |
| 380 | return 0; |
| 381 | } |
| 382 | |
| 383 | static inline void ptr_ring_cleanup(struct ptr_ring *r, void (*destroy)(void *)) |
| 384 | { |
| 385 | void *ptr; |
| 386 | |
| 387 | if (destroy) |
| 388 | while ((ptr = ptr_ring_consume(r))) |
| 389 | destroy(ptr); |
Michael S. Tsirkin | 2e0ab8c | 2016-06-13 23:54:31 +0300 | [diff] [blame] | 390 | kfree(r->queue); |
| 391 | } |
| 392 | |
| 393 | #endif /* _LINUX_PTR_RING_H */ |