Mauro Carvalho Chehab | e548cde | 2017-05-11 09:55:30 -0300 | [diff] [blame] | 1 | =========================== |
| 2 | Unreliable Guide To Locking |
| 3 | =========================== |
| 4 | |
| 5 | :Author: Rusty Russell |
| 6 | |
| 7 | Introduction |
| 8 | ============ |
| 9 | |
| 10 | Welcome, to Rusty's Remarkably Unreliable Guide to Kernel Locking |
| 11 | issues. This document describes the locking systems in the Linux Kernel |
| 12 | in 2.6. |
| 13 | |
| 14 | With the wide availability of HyperThreading, and preemption in the |
| 15 | Linux Kernel, everyone hacking on the kernel needs to know the |
| 16 | fundamentals of concurrency and locking for SMP. |
| 17 | |
| 18 | The Problem With Concurrency |
| 19 | ============================ |
| 20 | |
| 21 | (Skip this if you know what a Race Condition is). |
| 22 | |
| 23 | In a normal program, you can increment a counter like so: |
| 24 | |
| 25 | :: |
| 26 | |
| 27 | very_important_count++; |
| 28 | |
| 29 | |
| 30 | This is what they would expect to happen: |
| 31 | |
Mauro Carvalho Chehab | e548cde | 2017-05-11 09:55:30 -0300 | [diff] [blame] | 32 | |
Mauro Carvalho Chehab | 475c5ef | 2017-05-11 16:15:07 -0300 | [diff] [blame] | 33 | .. table:: Expected Results |
| 34 | |
| 35 | +------------------------------------+------------------------------------+ |
| 36 | | Instance 1 | Instance 2 | |
| 37 | +====================================+====================================+ |
| 38 | | read very_important_count (5) | | |
| 39 | +------------------------------------+------------------------------------+ |
| 40 | | add 1 (6) | | |
| 41 | +------------------------------------+------------------------------------+ |
| 42 | | write very_important_count (6) | | |
| 43 | +------------------------------------+------------------------------------+ |
| 44 | | | read very_important_count (6) | |
| 45 | +------------------------------------+------------------------------------+ |
| 46 | | | add 1 (7) | |
| 47 | +------------------------------------+------------------------------------+ |
| 48 | | | write very_important_count (7) | |
| 49 | +------------------------------------+------------------------------------+ |
Mauro Carvalho Chehab | e548cde | 2017-05-11 09:55:30 -0300 | [diff] [blame] | 50 | |
| 51 | This is what might happen: |
| 52 | |
Mauro Carvalho Chehab | 475c5ef | 2017-05-11 16:15:07 -0300 | [diff] [blame] | 53 | .. table:: Possible Results |
Mauro Carvalho Chehab | e548cde | 2017-05-11 09:55:30 -0300 | [diff] [blame] | 54 | |
Mauro Carvalho Chehab | 475c5ef | 2017-05-11 16:15:07 -0300 | [diff] [blame] | 55 | +------------------------------------+------------------------------------+ |
| 56 | | Instance 1 | Instance 2 | |
| 57 | +====================================+====================================+ |
| 58 | | read very_important_count (5) | | |
| 59 | +------------------------------------+------------------------------------+ |
| 60 | | | read very_important_count (5) | |
| 61 | +------------------------------------+------------------------------------+ |
| 62 | | add 1 (6) | | |
| 63 | +------------------------------------+------------------------------------+ |
| 64 | | | add 1 (6) | |
| 65 | +------------------------------------+------------------------------------+ |
| 66 | | write very_important_count (6) | | |
| 67 | +------------------------------------+------------------------------------+ |
| 68 | | | write very_important_count (6) | |
| 69 | +------------------------------------+------------------------------------+ |
| 70 | |
Mauro Carvalho Chehab | e548cde | 2017-05-11 09:55:30 -0300 | [diff] [blame] | 71 | |
| 72 | Race Conditions and Critical Regions |
| 73 | ------------------------------------ |
| 74 | |
| 75 | This overlap, where the result depends on the relative timing of |
| 76 | multiple tasks, is called a race condition. The piece of code containing |
| 77 | the concurrency issue is called a critical region. And especially since |
| 78 | Linux starting running on SMP machines, they became one of the major |
| 79 | issues in kernel design and implementation. |
| 80 | |
| 81 | Preemption can have the same effect, even if there is only one CPU: by |
| 82 | preempting one task during the critical region, we have exactly the same |
| 83 | race condition. In this case the thread which preempts might run the |
| 84 | critical region itself. |
| 85 | |
| 86 | The solution is to recognize when these simultaneous accesses occur, and |
| 87 | use locks to make sure that only one instance can enter the critical |
| 88 | region at any time. There are many friendly primitives in the Linux |
| 89 | kernel to help you do this. And then there are the unfriendly |
| 90 | primitives, but I'll pretend they don't exist. |
| 91 | |
| 92 | Locking in the Linux Kernel |
| 93 | =========================== |
| 94 | |
| 95 | If I could give you one piece of advice: never sleep with anyone crazier |
Mauro Carvalho Chehab | dc89fca | 2017-05-11 16:15:16 -0300 | [diff] [blame] | 96 | than yourself. But if I had to give you advice on locking: **keep it |
| 97 | simple**. |
Mauro Carvalho Chehab | e548cde | 2017-05-11 09:55:30 -0300 | [diff] [blame] | 98 | |
| 99 | Be reluctant to introduce new locks. |
| 100 | |
| 101 | Strangely enough, this last one is the exact reverse of my advice when |
Mauro Carvalho Chehab | dc89fca | 2017-05-11 16:15:16 -0300 | [diff] [blame] | 102 | you **have** slept with someone crazier than yourself. And you should |
Mauro Carvalho Chehab | e548cde | 2017-05-11 09:55:30 -0300 | [diff] [blame] | 103 | think about getting a big dog. |
| 104 | |
| 105 | Two Main Types of Kernel Locks: Spinlocks and Mutexes |
| 106 | ----------------------------------------------------- |
| 107 | |
| 108 | There are two main types of kernel locks. The fundamental type is the |
| 109 | spinlock (``include/asm/spinlock.h``), which is a very simple |
| 110 | single-holder lock: if you can't get the spinlock, you keep trying |
| 111 | (spinning) until you can. Spinlocks are very small and fast, and can be |
| 112 | used anywhere. |
| 113 | |
| 114 | The second type is a mutex (``include/linux/mutex.h``): it is like a |
| 115 | spinlock, but you may block holding a mutex. If you can't lock a mutex, |
| 116 | your task will suspend itself, and be woken up when the mutex is |
| 117 | released. This means the CPU can do something else while you are |
| 118 | waiting. There are many cases when you simply can't sleep (see |
| 119 | `What Functions Are Safe To Call From Interrupts? <#sleeping-things>`__), |
| 120 | and so have to use a spinlock instead. |
| 121 | |
| 122 | Neither type of lock is recursive: see |
| 123 | `Deadlock: Simple and Advanced <#deadlock>`__. |
| 124 | |
| 125 | Locks and Uniprocessor Kernels |
| 126 | ------------------------------ |
| 127 | |
| 128 | For kernels compiled without ``CONFIG_SMP``, and without |
| 129 | ``CONFIG_PREEMPT`` spinlocks do not exist at all. This is an excellent |
| 130 | design decision: when no-one else can run at the same time, there is no |
| 131 | reason to have a lock. |
| 132 | |
| 133 | If the kernel is compiled without ``CONFIG_SMP``, but ``CONFIG_PREEMPT`` |
| 134 | is set, then spinlocks simply disable preemption, which is sufficient to |
| 135 | prevent any races. For most purposes, we can think of preemption as |
| 136 | equivalent to SMP, and not worry about it separately. |
| 137 | |
| 138 | You should always test your locking code with ``CONFIG_SMP`` and |
| 139 | ``CONFIG_PREEMPT`` enabled, even if you don't have an SMP test box, |
| 140 | because it will still catch some kinds of locking bugs. |
| 141 | |
| 142 | Mutexes still exist, because they are required for synchronization |
| 143 | between user contexts, as we will see below. |
| 144 | |
| 145 | Locking Only In User Context |
| 146 | ---------------------------- |
| 147 | |
| 148 | If you have a data structure which is only ever accessed from user |
| 149 | context, then you can use a simple mutex (``include/linux/mutex.h``) to |
| 150 | protect it. This is the most trivial case: you initialize the mutex. |
| 151 | Then you can call :c:func:`mutex_lock_interruptible()` to grab the |
| 152 | mutex, and :c:func:`mutex_unlock()` to release it. There is also a |
| 153 | :c:func:`mutex_lock()`, which should be avoided, because it will |
| 154 | not return if a signal is received. |
| 155 | |
| 156 | Example: ``net/netfilter/nf_sockopt.c`` allows registration of new |
| 157 | :c:func:`setsockopt()` and :c:func:`getsockopt()` calls, with |
| 158 | :c:func:`nf_register_sockopt()`. Registration and de-registration |
| 159 | are only done on module load and unload (and boot time, where there is |
| 160 | no concurrency), and the list of registrations is only consulted for an |
| 161 | unknown :c:func:`setsockopt()` or :c:func:`getsockopt()` system |
| 162 | call. The ``nf_sockopt_mutex`` is perfect to protect this, especially |
| 163 | since the setsockopt and getsockopt calls may well sleep. |
| 164 | |
| 165 | Locking Between User Context and Softirqs |
| 166 | ----------------------------------------- |
| 167 | |
| 168 | If a softirq shares data with user context, you have two problems. |
| 169 | Firstly, the current user context can be interrupted by a softirq, and |
| 170 | secondly, the critical region could be entered from another CPU. This is |
| 171 | where :c:func:`spin_lock_bh()` (``include/linux/spinlock.h``) is |
| 172 | used. It disables softirqs on that CPU, then grabs the lock. |
| 173 | :c:func:`spin_unlock_bh()` does the reverse. (The '_bh' suffix is |
| 174 | a historical reference to "Bottom Halves", the old name for software |
| 175 | interrupts. It should really be called spin_lock_softirq()' in a |
| 176 | perfect world). |
| 177 | |
| 178 | Note that you can also use :c:func:`spin_lock_irq()` or |
| 179 | :c:func:`spin_lock_irqsave()` here, which stop hardware interrupts |
| 180 | as well: see `Hard IRQ Context <#hardirq-context>`__. |
| 181 | |
| 182 | This works perfectly for UP as well: the spin lock vanishes, and this |
| 183 | macro simply becomes :c:func:`local_bh_disable()` |
| 184 | (``include/linux/interrupt.h``), which protects you from the softirq |
| 185 | being run. |
| 186 | |
| 187 | Locking Between User Context and Tasklets |
| 188 | ----------------------------------------- |
| 189 | |
| 190 | This is exactly the same as above, because tasklets are actually run |
| 191 | from a softirq. |
| 192 | |
| 193 | Locking Between User Context and Timers |
| 194 | --------------------------------------- |
| 195 | |
| 196 | This, too, is exactly the same as above, because timers are actually run |
| 197 | from a softirq. From a locking point of view, tasklets and timers are |
| 198 | identical. |
| 199 | |
| 200 | Locking Between Tasklets/Timers |
| 201 | ------------------------------- |
| 202 | |
| 203 | Sometimes a tasklet or timer might want to share data with another |
| 204 | tasklet or timer. |
| 205 | |
| 206 | The Same Tasklet/Timer |
| 207 | ~~~~~~~~~~~~~~~~~~~~~~ |
| 208 | |
| 209 | Since a tasklet is never run on two CPUs at once, you don't need to |
| 210 | worry about your tasklet being reentrant (running twice at once), even |
| 211 | on SMP. |
| 212 | |
| 213 | Different Tasklets/Timers |
| 214 | ~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 215 | |
| 216 | If another tasklet/timer wants to share data with your tasklet or timer |
| 217 | , you will both need to use :c:func:`spin_lock()` and |
| 218 | :c:func:`spin_unlock()` calls. :c:func:`spin_lock_bh()` is |
| 219 | unnecessary here, as you are already in a tasklet, and none will be run |
| 220 | on the same CPU. |
| 221 | |
| 222 | Locking Between Softirqs |
| 223 | ------------------------ |
| 224 | |
| 225 | Often a softirq might want to share data with itself or a tasklet/timer. |
| 226 | |
| 227 | The Same Softirq |
| 228 | ~~~~~~~~~~~~~~~~ |
| 229 | |
| 230 | The same softirq can run on the other CPUs: you can use a per-CPU array |
| 231 | (see `Per-CPU Data <#per-cpu>`__) for better performance. If you're |
| 232 | going so far as to use a softirq, you probably care about scalable |
| 233 | performance enough to justify the extra complexity. |
| 234 | |
| 235 | You'll need to use :c:func:`spin_lock()` and |
| 236 | :c:func:`spin_unlock()` for shared data. |
| 237 | |
| 238 | Different Softirqs |
| 239 | ~~~~~~~~~~~~~~~~~~ |
| 240 | |
| 241 | You'll need to use :c:func:`spin_lock()` and |
| 242 | :c:func:`spin_unlock()` for shared data, whether it be a timer, |
| 243 | tasklet, different softirq or the same or another softirq: any of them |
| 244 | could be running on a different CPU. |
| 245 | |
| 246 | Hard IRQ Context |
| 247 | ================ |
| 248 | |
| 249 | Hardware interrupts usually communicate with a tasklet or softirq. |
| 250 | Frequently this involves putting work in a queue, which the softirq will |
| 251 | take out. |
| 252 | |
| 253 | Locking Between Hard IRQ and Softirqs/Tasklets |
| 254 | ---------------------------------------------- |
| 255 | |
| 256 | If a hardware irq handler shares data with a softirq, you have two |
| 257 | concerns. Firstly, the softirq processing can be interrupted by a |
| 258 | hardware interrupt, and secondly, the critical region could be entered |
| 259 | by a hardware interrupt on another CPU. This is where |
| 260 | :c:func:`spin_lock_irq()` is used. It is defined to disable |
| 261 | interrupts on that cpu, then grab the lock. |
| 262 | :c:func:`spin_unlock_irq()` does the reverse. |
| 263 | |
| 264 | The irq handler does not to use :c:func:`spin_lock_irq()`, because |
| 265 | the softirq cannot run while the irq handler is running: it can use |
| 266 | :c:func:`spin_lock()`, which is slightly faster. The only exception |
| 267 | would be if a different hardware irq handler uses the same lock: |
| 268 | :c:func:`spin_lock_irq()` will stop that from interrupting us. |
| 269 | |
| 270 | This works perfectly for UP as well: the spin lock vanishes, and this |
| 271 | macro simply becomes :c:func:`local_irq_disable()` |
| 272 | (``include/asm/smp.h``), which protects you from the softirq/tasklet/BH |
| 273 | being run. |
| 274 | |
| 275 | :c:func:`spin_lock_irqsave()` (``include/linux/spinlock.h``) is a |
| 276 | variant which saves whether interrupts were on or off in a flags word, |
| 277 | which is passed to :c:func:`spin_unlock_irqrestore()`. This means |
| 278 | that the same code can be used inside an hard irq handler (where |
| 279 | interrupts are already off) and in softirqs (where the irq disabling is |
| 280 | required). |
| 281 | |
| 282 | Note that softirqs (and hence tasklets and timers) are run on return |
| 283 | from hardware interrupts, so :c:func:`spin_lock_irq()` also stops |
| 284 | these. In that sense, :c:func:`spin_lock_irqsave()` is the most |
| 285 | general and powerful locking function. |
| 286 | |
| 287 | Locking Between Two Hard IRQ Handlers |
| 288 | ------------------------------------- |
| 289 | |
| 290 | It is rare to have to share data between two IRQ handlers, but if you |
| 291 | do, :c:func:`spin_lock_irqsave()` should be used: it is |
| 292 | architecture-specific whether all interrupts are disabled inside irq |
| 293 | handlers themselves. |
| 294 | |
| 295 | Cheat Sheet For Locking |
| 296 | ======================= |
| 297 | |
| 298 | Pete Zaitcev gives the following summary: |
| 299 | |
| 300 | - If you are in a process context (any syscall) and want to lock other |
| 301 | process out, use a mutex. You can take a mutex and sleep |
| 302 | (``copy_from_user*(`` or ``kmalloc(x,GFP_KERNEL)``). |
| 303 | |
| 304 | - Otherwise (== data can be touched in an interrupt), use |
| 305 | :c:func:`spin_lock_irqsave()` and |
| 306 | :c:func:`spin_unlock_irqrestore()`. |
| 307 | |
| 308 | - Avoid holding spinlock for more than 5 lines of code and across any |
| 309 | function call (except accessors like :c:func:`readb()`). |
| 310 | |
| 311 | Table of Minimum Requirements |
| 312 | ----------------------------- |
| 313 | |
Mauro Carvalho Chehab | dc89fca | 2017-05-11 16:15:16 -0300 | [diff] [blame] | 314 | The following table lists the **minimum** locking requirements between |
Mauro Carvalho Chehab | e548cde | 2017-05-11 09:55:30 -0300 | [diff] [blame] | 315 | various contexts. In some cases, the same context can only be running on |
| 316 | one CPU at a time, so no locking is required for that context (eg. a |
| 317 | particular thread can only run on one CPU at a time, but if it needs |
| 318 | shares data with another thread, locking is required). |
| 319 | |
| 320 | Remember the advice above: you can always use |
| 321 | :c:func:`spin_lock_irqsave()`, which is a superset of all other |
| 322 | spinlock primitives. |
| 323 | |
Mauro Carvalho Chehab | 5b9fd1d | 2017-05-11 10:45:47 -0300 | [diff] [blame] | 324 | ============== ============= ============= ========= ========= ========= ========= ======= ======= ============== ============== |
| 325 | . IRQ Handler A IRQ Handler B Softirq A Softirq B Tasklet A Tasklet B Timer A Timer B User Context A User Context B |
| 326 | ============== ============= ============= ========= ========= ========= ========= ======= ======= ============== ============== |
| 327 | IRQ Handler A None |
| 328 | IRQ Handler B SLIS None |
| 329 | Softirq A SLI SLI SL |
| 330 | Softirq B SLI SLI SL SL |
| 331 | Tasklet A SLI SLI SL SL None |
| 332 | Tasklet B SLI SLI SL SL SL None |
| 333 | Timer A SLI SLI SL SL SL SL None |
| 334 | Timer B SLI SLI SL SL SL SL SL None |
| 335 | User Context A SLI SLI SLBH SLBH SLBH SLBH SLBH SLBH None |
| 336 | User Context B SLI SLI SLBH SLBH SLBH SLBH SLBH SLBH MLI None |
| 337 | ============== ============= ============= ========= ========= ========= ========= ======= ======= ============== ============== |
Mauro Carvalho Chehab | e548cde | 2017-05-11 09:55:30 -0300 | [diff] [blame] | 338 | |
| 339 | Table: Table of Locking Requirements |
| 340 | |
| 341 | +--------+----------------------------+ |
| 342 | | SLIS | spin_lock_irqsave | |
| 343 | +--------+----------------------------+ |
| 344 | | SLI | spin_lock_irq | |
| 345 | +--------+----------------------------+ |
| 346 | | SL | spin_lock | |
| 347 | +--------+----------------------------+ |
| 348 | | SLBH | spin_lock_bh | |
| 349 | +--------+----------------------------+ |
| 350 | | MLI | mutex_lock_interruptible | |
| 351 | +--------+----------------------------+ |
| 352 | |
| 353 | Table: Legend for Locking Requirements Table |
| 354 | |
| 355 | The trylock Functions |
| 356 | ===================== |
| 357 | |
| 358 | There are functions that try to acquire a lock only once and immediately |
| 359 | return a value telling about success or failure to acquire the lock. |
| 360 | They can be used if you need no access to the data protected with the |
| 361 | lock when some other thread is holding the lock. You should acquire the |
| 362 | lock later if you then need access to the data protected with the lock. |
| 363 | |
| 364 | :c:func:`spin_trylock()` does not spin but returns non-zero if it |
| 365 | acquires the spinlock on the first try or 0 if not. This function can be |
| 366 | used in all contexts like :c:func:`spin_lock()`: you must have |
| 367 | disabled the contexts that might interrupt you and acquire the spin |
| 368 | lock. |
| 369 | |
| 370 | :c:func:`mutex_trylock()` does not suspend your task but returns |
| 371 | non-zero if it could lock the mutex on the first try or 0 if not. This |
| 372 | function cannot be safely used in hardware or software interrupt |
| 373 | contexts despite not sleeping. |
| 374 | |
| 375 | Common Examples |
| 376 | =============== |
| 377 | |
| 378 | Let's step through a simple example: a cache of number to name mappings. |
| 379 | The cache keeps a count of how often each of the objects is used, and |
| 380 | when it gets full, throws out the least used one. |
| 381 | |
| 382 | All In User Context |
| 383 | ------------------- |
| 384 | |
| 385 | For our first example, we assume that all operations are in user context |
| 386 | (ie. from system calls), so we can sleep. This means we can use a mutex |
| 387 | to protect the cache and all the objects within it. Here's the code:: |
| 388 | |
| 389 | #include <linux/list.h> |
| 390 | #include <linux/slab.h> |
| 391 | #include <linux/string.h> |
| 392 | #include <linux/mutex.h> |
| 393 | #include <asm/errno.h> |
| 394 | |
| 395 | struct object |
| 396 | { |
| 397 | struct list_head list; |
| 398 | int id; |
| 399 | char name[32]; |
| 400 | int popularity; |
| 401 | }; |
| 402 | |
| 403 | /* Protects the cache, cache_num, and the objects within it */ |
| 404 | static DEFINE_MUTEX(cache_lock); |
| 405 | static LIST_HEAD(cache); |
| 406 | static unsigned int cache_num = 0; |
| 407 | #define MAX_CACHE_SIZE 10 |
| 408 | |
| 409 | /* Must be holding cache_lock */ |
| 410 | static struct object *__cache_find(int id) |
| 411 | { |
| 412 | struct object *i; |
| 413 | |
| 414 | list_for_each_entry(i, &cache, list) |
| 415 | if (i->id == id) { |
| 416 | i->popularity++; |
| 417 | return i; |
| 418 | } |
| 419 | return NULL; |
| 420 | } |
| 421 | |
| 422 | /* Must be holding cache_lock */ |
| 423 | static void __cache_delete(struct object *obj) |
| 424 | { |
| 425 | BUG_ON(!obj); |
| 426 | list_del(&obj->list); |
| 427 | kfree(obj); |
| 428 | cache_num--; |
| 429 | } |
| 430 | |
| 431 | /* Must be holding cache_lock */ |
| 432 | static void __cache_add(struct object *obj) |
| 433 | { |
| 434 | list_add(&obj->list, &cache); |
| 435 | if (++cache_num > MAX_CACHE_SIZE) { |
| 436 | struct object *i, *outcast = NULL; |
| 437 | list_for_each_entry(i, &cache, list) { |
| 438 | if (!outcast || i->popularity < outcast->popularity) |
| 439 | outcast = i; |
| 440 | } |
| 441 | __cache_delete(outcast); |
| 442 | } |
| 443 | } |
| 444 | |
| 445 | int cache_add(int id, const char *name) |
| 446 | { |
| 447 | struct object *obj; |
| 448 | |
| 449 | if ((obj = kmalloc(sizeof(*obj), GFP_KERNEL)) == NULL) |
| 450 | return -ENOMEM; |
| 451 | |
| 452 | strlcpy(obj->name, name, sizeof(obj->name)); |
| 453 | obj->id = id; |
| 454 | obj->popularity = 0; |
| 455 | |
| 456 | mutex_lock(&cache_lock); |
| 457 | __cache_add(obj); |
| 458 | mutex_unlock(&cache_lock); |
| 459 | return 0; |
| 460 | } |
| 461 | |
| 462 | void cache_delete(int id) |
| 463 | { |
| 464 | mutex_lock(&cache_lock); |
| 465 | __cache_delete(__cache_find(id)); |
| 466 | mutex_unlock(&cache_lock); |
| 467 | } |
| 468 | |
| 469 | int cache_find(int id, char *name) |
| 470 | { |
| 471 | struct object *obj; |
| 472 | int ret = -ENOENT; |
| 473 | |
| 474 | mutex_lock(&cache_lock); |
| 475 | obj = __cache_find(id); |
| 476 | if (obj) { |
| 477 | ret = 0; |
| 478 | strcpy(name, obj->name); |
| 479 | } |
| 480 | mutex_unlock(&cache_lock); |
| 481 | return ret; |
| 482 | } |
| 483 | |
| 484 | Note that we always make sure we have the cache_lock when we add, |
| 485 | delete, or look up the cache: both the cache infrastructure itself and |
| 486 | the contents of the objects are protected by the lock. In this case it's |
| 487 | easy, since we copy the data for the user, and never let them access the |
| 488 | objects directly. |
| 489 | |
| 490 | There is a slight (and common) optimization here: in |
| 491 | :c:func:`cache_add()` we set up the fields of the object before |
| 492 | grabbing the lock. This is safe, as no-one else can access it until we |
| 493 | put it in cache. |
| 494 | |
| 495 | Accessing From Interrupt Context |
| 496 | -------------------------------- |
| 497 | |
| 498 | Now consider the case where :c:func:`cache_find()` can be called |
| 499 | from interrupt context: either a hardware interrupt or a softirq. An |
| 500 | example would be a timer which deletes object from the cache. |
| 501 | |
| 502 | The change is shown below, in standard patch format: the ``-`` are lines |
| 503 | which are taken away, and the ``+`` are lines which are added. |
| 504 | |
| 505 | :: |
| 506 | |
| 507 | --- cache.c.usercontext 2003-12-09 13:58:54.000000000 +1100 |
| 508 | +++ cache.c.interrupt 2003-12-09 14:07:49.000000000 +1100 |
| 509 | @@ -12,7 +12,7 @@ |
| 510 | int popularity; |
| 511 | }; |
| 512 | |
| 513 | -static DEFINE_MUTEX(cache_lock); |
| 514 | +static DEFINE_SPINLOCK(cache_lock); |
| 515 | static LIST_HEAD(cache); |
| 516 | static unsigned int cache_num = 0; |
| 517 | #define MAX_CACHE_SIZE 10 |
| 518 | @@ -55,6 +55,7 @@ |
| 519 | int cache_add(int id, const char *name) |
| 520 | { |
| 521 | struct object *obj; |
| 522 | + unsigned long flags; |
| 523 | |
| 524 | if ((obj = kmalloc(sizeof(*obj), GFP_KERNEL)) == NULL) |
| 525 | return -ENOMEM; |
| 526 | @@ -63,30 +64,33 @@ |
| 527 | obj->id = id; |
| 528 | obj->popularity = 0; |
| 529 | |
| 530 | - mutex_lock(&cache_lock); |
| 531 | + spin_lock_irqsave(&cache_lock, flags); |
| 532 | __cache_add(obj); |
| 533 | - mutex_unlock(&cache_lock); |
| 534 | + spin_unlock_irqrestore(&cache_lock, flags); |
| 535 | return 0; |
| 536 | } |
| 537 | |
| 538 | void cache_delete(int id) |
| 539 | { |
| 540 | - mutex_lock(&cache_lock); |
| 541 | + unsigned long flags; |
| 542 | + |
| 543 | + spin_lock_irqsave(&cache_lock, flags); |
| 544 | __cache_delete(__cache_find(id)); |
| 545 | - mutex_unlock(&cache_lock); |
| 546 | + spin_unlock_irqrestore(&cache_lock, flags); |
| 547 | } |
| 548 | |
| 549 | int cache_find(int id, char *name) |
| 550 | { |
| 551 | struct object *obj; |
| 552 | int ret = -ENOENT; |
| 553 | + unsigned long flags; |
| 554 | |
| 555 | - mutex_lock(&cache_lock); |
| 556 | + spin_lock_irqsave(&cache_lock, flags); |
| 557 | obj = __cache_find(id); |
| 558 | if (obj) { |
| 559 | ret = 0; |
| 560 | strcpy(name, obj->name); |
| 561 | } |
| 562 | - mutex_unlock(&cache_lock); |
| 563 | + spin_unlock_irqrestore(&cache_lock, flags); |
| 564 | return ret; |
| 565 | } |
| 566 | |
| 567 | Note that the :c:func:`spin_lock_irqsave()` will turn off |
| 568 | interrupts if they are on, otherwise does nothing (if we are already in |
| 569 | an interrupt handler), hence these functions are safe to call from any |
| 570 | context. |
| 571 | |
| 572 | Unfortunately, :c:func:`cache_add()` calls :c:func:`kmalloc()` |
| 573 | with the ``GFP_KERNEL`` flag, which is only legal in user context. I |
| 574 | have assumed that :c:func:`cache_add()` is still only called in |
| 575 | user context, otherwise this should become a parameter to |
| 576 | :c:func:`cache_add()`. |
| 577 | |
| 578 | Exposing Objects Outside This File |
| 579 | ---------------------------------- |
| 580 | |
| 581 | If our objects contained more information, it might not be sufficient to |
| 582 | copy the information in and out: other parts of the code might want to |
| 583 | keep pointers to these objects, for example, rather than looking up the |
| 584 | id every time. This produces two problems. |
| 585 | |
| 586 | The first problem is that we use the ``cache_lock`` to protect objects: |
| 587 | we'd need to make this non-static so the rest of the code can use it. |
| 588 | This makes locking trickier, as it is no longer all in one place. |
| 589 | |
| 590 | The second problem is the lifetime problem: if another structure keeps a |
| 591 | pointer to an object, it presumably expects that pointer to remain |
| 592 | valid. Unfortunately, this is only guaranteed while you hold the lock, |
| 593 | otherwise someone might call :c:func:`cache_delete()` and even |
| 594 | worse, add another object, re-using the same address. |
| 595 | |
| 596 | As there is only one lock, you can't hold it forever: no-one else would |
| 597 | get any work done. |
| 598 | |
| 599 | The solution to this problem is to use a reference count: everyone who |
| 600 | has a pointer to the object increases it when they first get the object, |
| 601 | and drops the reference count when they're finished with it. Whoever |
| 602 | drops it to zero knows it is unused, and can actually delete it. |
| 603 | |
| 604 | Here is the code:: |
| 605 | |
| 606 | --- cache.c.interrupt 2003-12-09 14:25:43.000000000 +1100 |
| 607 | +++ cache.c.refcnt 2003-12-09 14:33:05.000000000 +1100 |
| 608 | @@ -7,6 +7,7 @@ |
| 609 | struct object |
| 610 | { |
| 611 | struct list_head list; |
| 612 | + unsigned int refcnt; |
| 613 | int id; |
| 614 | char name[32]; |
| 615 | int popularity; |
| 616 | @@ -17,6 +18,35 @@ |
| 617 | static unsigned int cache_num = 0; |
| 618 | #define MAX_CACHE_SIZE 10 |
| 619 | |
| 620 | +static void __object_put(struct object *obj) |
| 621 | +{ |
| 622 | + if (--obj->refcnt == 0) |
| 623 | + kfree(obj); |
| 624 | +} |
| 625 | + |
| 626 | +static void __object_get(struct object *obj) |
| 627 | +{ |
| 628 | + obj->refcnt++; |
| 629 | +} |
| 630 | + |
| 631 | +void object_put(struct object *obj) |
| 632 | +{ |
| 633 | + unsigned long flags; |
| 634 | + |
| 635 | + spin_lock_irqsave(&cache_lock, flags); |
| 636 | + __object_put(obj); |
| 637 | + spin_unlock_irqrestore(&cache_lock, flags); |
| 638 | +} |
| 639 | + |
| 640 | +void object_get(struct object *obj) |
| 641 | +{ |
| 642 | + unsigned long flags; |
| 643 | + |
| 644 | + spin_lock_irqsave(&cache_lock, flags); |
| 645 | + __object_get(obj); |
| 646 | + spin_unlock_irqrestore(&cache_lock, flags); |
| 647 | +} |
| 648 | + |
| 649 | /* Must be holding cache_lock */ |
| 650 | static struct object *__cache_find(int id) |
| 651 | { |
| 652 | @@ -35,6 +65,7 @@ |
| 653 | { |
| 654 | BUG_ON(!obj); |
| 655 | list_del(&obj->list); |
| 656 | + __object_put(obj); |
| 657 | cache_num--; |
| 658 | } |
| 659 | |
| 660 | @@ -63,6 +94,7 @@ |
| 661 | strlcpy(obj->name, name, sizeof(obj->name)); |
| 662 | obj->id = id; |
| 663 | obj->popularity = 0; |
| 664 | + obj->refcnt = 1; /* The cache holds a reference */ |
| 665 | |
| 666 | spin_lock_irqsave(&cache_lock, flags); |
| 667 | __cache_add(obj); |
| 668 | @@ -79,18 +111,15 @@ |
| 669 | spin_unlock_irqrestore(&cache_lock, flags); |
| 670 | } |
| 671 | |
| 672 | -int cache_find(int id, char *name) |
| 673 | +struct object *cache_find(int id) |
| 674 | { |
| 675 | struct object *obj; |
| 676 | - int ret = -ENOENT; |
| 677 | unsigned long flags; |
| 678 | |
| 679 | spin_lock_irqsave(&cache_lock, flags); |
| 680 | obj = __cache_find(id); |
| 681 | - if (obj) { |
| 682 | - ret = 0; |
| 683 | - strcpy(name, obj->name); |
| 684 | - } |
| 685 | + if (obj) |
| 686 | + __object_get(obj); |
| 687 | spin_unlock_irqrestore(&cache_lock, flags); |
| 688 | - return ret; |
| 689 | + return obj; |
| 690 | } |
| 691 | |
| 692 | We encapsulate the reference counting in the standard 'get' and 'put' |
| 693 | functions. Now we can return the object itself from |
| 694 | :c:func:`cache_find()` which has the advantage that the user can |
| 695 | now sleep holding the object (eg. to :c:func:`copy_to_user()` to |
| 696 | name to userspace). |
| 697 | |
| 698 | The other point to note is that I said a reference should be held for |
| 699 | every pointer to the object: thus the reference count is 1 when first |
| 700 | inserted into the cache. In some versions the framework does not hold a |
| 701 | reference count, but they are more complicated. |
| 702 | |
| 703 | Using Atomic Operations For The Reference Count |
| 704 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 705 | |
Mauro Carvalho Chehab | dc89fca | 2017-05-11 16:15:16 -0300 | [diff] [blame] | 706 | In practice, :c:type:`atomic_t` would usually be used for refcnt. There are a |
Mauro Carvalho Chehab | e548cde | 2017-05-11 09:55:30 -0300 | [diff] [blame] | 707 | number of atomic operations defined in ``include/asm/atomic.h``: these |
| 708 | are guaranteed to be seen atomically from all CPUs in the system, so no |
| 709 | lock is required. In this case, it is simpler than using spinlocks, |
| 710 | although for anything non-trivial using spinlocks is clearer. The |
| 711 | :c:func:`atomic_inc()` and :c:func:`atomic_dec_and_test()` |
| 712 | are used instead of the standard increment and decrement operators, and |
| 713 | the lock is no longer used to protect the reference count itself. |
| 714 | |
| 715 | :: |
| 716 | |
| 717 | --- cache.c.refcnt 2003-12-09 15:00:35.000000000 +1100 |
| 718 | +++ cache.c.refcnt-atomic 2003-12-11 15:49:42.000000000 +1100 |
| 719 | @@ -7,7 +7,7 @@ |
| 720 | struct object |
| 721 | { |
| 722 | struct list_head list; |
| 723 | - unsigned int refcnt; |
| 724 | + atomic_t refcnt; |
| 725 | int id; |
| 726 | char name[32]; |
| 727 | int popularity; |
| 728 | @@ -18,33 +18,15 @@ |
| 729 | static unsigned int cache_num = 0; |
| 730 | #define MAX_CACHE_SIZE 10 |
| 731 | |
| 732 | -static void __object_put(struct object *obj) |
| 733 | -{ |
| 734 | - if (--obj->refcnt == 0) |
| 735 | - kfree(obj); |
| 736 | -} |
| 737 | - |
| 738 | -static void __object_get(struct object *obj) |
| 739 | -{ |
| 740 | - obj->refcnt++; |
| 741 | -} |
| 742 | - |
| 743 | void object_put(struct object *obj) |
| 744 | { |
| 745 | - unsigned long flags; |
| 746 | - |
| 747 | - spin_lock_irqsave(&cache_lock, flags); |
| 748 | - __object_put(obj); |
| 749 | - spin_unlock_irqrestore(&cache_lock, flags); |
| 750 | + if (atomic_dec_and_test(&obj->refcnt)) |
| 751 | + kfree(obj); |
| 752 | } |
| 753 | |
| 754 | void object_get(struct object *obj) |
| 755 | { |
| 756 | - unsigned long flags; |
| 757 | - |
| 758 | - spin_lock_irqsave(&cache_lock, flags); |
| 759 | - __object_get(obj); |
| 760 | - spin_unlock_irqrestore(&cache_lock, flags); |
| 761 | + atomic_inc(&obj->refcnt); |
| 762 | } |
| 763 | |
| 764 | /* Must be holding cache_lock */ |
| 765 | @@ -65,7 +47,7 @@ |
| 766 | { |
| 767 | BUG_ON(!obj); |
| 768 | list_del(&obj->list); |
| 769 | - __object_put(obj); |
| 770 | + object_put(obj); |
| 771 | cache_num--; |
| 772 | } |
| 773 | |
| 774 | @@ -94,7 +76,7 @@ |
| 775 | strlcpy(obj->name, name, sizeof(obj->name)); |
| 776 | obj->id = id; |
| 777 | obj->popularity = 0; |
| 778 | - obj->refcnt = 1; /* The cache holds a reference */ |
| 779 | + atomic_set(&obj->refcnt, 1); /* The cache holds a reference */ |
| 780 | |
| 781 | spin_lock_irqsave(&cache_lock, flags); |
| 782 | __cache_add(obj); |
| 783 | @@ -119,7 +101,7 @@ |
| 784 | spin_lock_irqsave(&cache_lock, flags); |
| 785 | obj = __cache_find(id); |
| 786 | if (obj) |
| 787 | - __object_get(obj); |
| 788 | + object_get(obj); |
| 789 | spin_unlock_irqrestore(&cache_lock, flags); |
| 790 | return obj; |
| 791 | } |
| 792 | |
| 793 | Protecting The Objects Themselves |
| 794 | --------------------------------- |
| 795 | |
| 796 | In these examples, we assumed that the objects (except the reference |
| 797 | counts) never changed once they are created. If we wanted to allow the |
| 798 | name to change, there are three possibilities: |
| 799 | |
| 800 | - You can make ``cache_lock`` non-static, and tell people to grab that |
| 801 | lock before changing the name in any object. |
| 802 | |
| 803 | - You can provide a :c:func:`cache_obj_rename()` which grabs this |
| 804 | lock and changes the name for the caller, and tell everyone to use |
| 805 | that function. |
| 806 | |
| 807 | - You can make the ``cache_lock`` protect only the cache itself, and |
| 808 | use another lock to protect the name. |
| 809 | |
| 810 | Theoretically, you can make the locks as fine-grained as one lock for |
| 811 | every field, for every object. In practice, the most common variants |
| 812 | are: |
| 813 | |
| 814 | - One lock which protects the infrastructure (the ``cache`` list in |
| 815 | this example) and all the objects. This is what we have done so far. |
| 816 | |
| 817 | - One lock which protects the infrastructure (including the list |
| 818 | pointers inside the objects), and one lock inside the object which |
| 819 | protects the rest of that object. |
| 820 | |
| 821 | - Multiple locks to protect the infrastructure (eg. one lock per hash |
| 822 | chain), possibly with a separate per-object lock. |
| 823 | |
| 824 | Here is the "lock-per-object" implementation: |
| 825 | |
| 826 | :: |
| 827 | |
| 828 | --- cache.c.refcnt-atomic 2003-12-11 15:50:54.000000000 +1100 |
| 829 | +++ cache.c.perobjectlock 2003-12-11 17:15:03.000000000 +1100 |
| 830 | @@ -6,11 +6,17 @@ |
| 831 | |
| 832 | struct object |
| 833 | { |
| 834 | + /* These two protected by cache_lock. */ |
| 835 | struct list_head list; |
| 836 | + int popularity; |
| 837 | + |
| 838 | atomic_t refcnt; |
| 839 | + |
| 840 | + /* Doesn't change once created. */ |
| 841 | int id; |
| 842 | + |
| 843 | + spinlock_t lock; /* Protects the name */ |
| 844 | char name[32]; |
| 845 | - int popularity; |
| 846 | }; |
| 847 | |
| 848 | static DEFINE_SPINLOCK(cache_lock); |
| 849 | @@ -77,6 +84,7 @@ |
| 850 | obj->id = id; |
| 851 | obj->popularity = 0; |
| 852 | atomic_set(&obj->refcnt, 1); /* The cache holds a reference */ |
| 853 | + spin_lock_init(&obj->lock); |
| 854 | |
| 855 | spin_lock_irqsave(&cache_lock, flags); |
| 856 | __cache_add(obj); |
| 857 | |
| 858 | Note that I decide that the popularity count should be protected by the |
| 859 | ``cache_lock`` rather than the per-object lock: this is because it (like |
| 860 | the :c:type:`struct list_head <list_head>` inside the object) |
| 861 | is logically part of the infrastructure. This way, I don't need to grab |
| 862 | the lock of every object in :c:func:`__cache_add()` when seeking |
| 863 | the least popular. |
| 864 | |
| 865 | I also decided that the id member is unchangeable, so I don't need to |
| 866 | grab each object lock in :c:func:`__cache_find()` to examine the |
| 867 | id: the object lock is only used by a caller who wants to read or write |
| 868 | the name field. |
| 869 | |
| 870 | Note also that I added a comment describing what data was protected by |
| 871 | which locks. This is extremely important, as it describes the runtime |
| 872 | behavior of the code, and can be hard to gain from just reading. And as |
| 873 | Alan Cox says, “Lock data, not code”. |
| 874 | |
| 875 | Common Problems |
| 876 | =============== |
| 877 | |
| 878 | Deadlock: Simple and Advanced |
| 879 | ----------------------------- |
| 880 | |
| 881 | There is a coding bug where a piece of code tries to grab a spinlock |
| 882 | twice: it will spin forever, waiting for the lock to be released |
| 883 | (spinlocks, rwlocks and mutexes are not recursive in Linux). This is |
| 884 | trivial to diagnose: not a |
| 885 | stay-up-five-nights-talk-to-fluffy-code-bunnies kind of problem. |
| 886 | |
| 887 | For a slightly more complex case, imagine you have a region shared by a |
| 888 | softirq and user context. If you use a :c:func:`spin_lock()` call |
| 889 | to protect it, it is possible that the user context will be interrupted |
| 890 | by the softirq while it holds the lock, and the softirq will then spin |
| 891 | forever trying to get the same lock. |
| 892 | |
| 893 | Both of these are called deadlock, and as shown above, it can occur even |
| 894 | with a single CPU (although not on UP compiles, since spinlocks vanish |
| 895 | on kernel compiles with ``CONFIG_SMP``\ =n. You'll still get data |
| 896 | corruption in the second example). |
| 897 | |
| 898 | This complete lockup is easy to diagnose: on SMP boxes the watchdog |
| 899 | timer or compiling with ``DEBUG_SPINLOCK`` set |
| 900 | (``include/linux/spinlock.h``) will show this up immediately when it |
| 901 | happens. |
| 902 | |
| 903 | A more complex problem is the so-called 'deadly embrace', involving two |
| 904 | or more locks. Say you have a hash table: each entry in the table is a |
| 905 | spinlock, and a chain of hashed objects. Inside a softirq handler, you |
| 906 | sometimes want to alter an object from one place in the hash to another: |
| 907 | you grab the spinlock of the old hash chain and the spinlock of the new |
| 908 | hash chain, and delete the object from the old one, and insert it in the |
| 909 | new one. |
| 910 | |
| 911 | There are two problems here. First, if your code ever tries to move the |
| 912 | object to the same chain, it will deadlock with itself as it tries to |
| 913 | lock it twice. Secondly, if the same softirq on another CPU is trying to |
| 914 | move another object in the reverse direction, the following could |
| 915 | happen: |
| 916 | |
| 917 | +-----------------------+-----------------------+ |
| 918 | | CPU 1 | CPU 2 | |
| 919 | +=======================+=======================+ |
| 920 | | Grab lock A -> OK | Grab lock B -> OK | |
| 921 | +-----------------------+-----------------------+ |
| 922 | | Grab lock B -> spin | Grab lock A -> spin | |
| 923 | +-----------------------+-----------------------+ |
| 924 | |
| 925 | Table: Consequences |
| 926 | |
| 927 | The two CPUs will spin forever, waiting for the other to give up their |
| 928 | lock. It will look, smell, and feel like a crash. |
| 929 | |
| 930 | Preventing Deadlock |
| 931 | ------------------- |
| 932 | |
| 933 | Textbooks will tell you that if you always lock in the same order, you |
| 934 | will never get this kind of deadlock. Practice will tell you that this |
| 935 | approach doesn't scale: when I create a new lock, I don't understand |
| 936 | enough of the kernel to figure out where in the 5000 lock hierarchy it |
| 937 | will fit. |
| 938 | |
| 939 | The best locks are encapsulated: they never get exposed in headers, and |
| 940 | are never held around calls to non-trivial functions outside the same |
| 941 | file. You can read through this code and see that it will never |
| 942 | deadlock, because it never tries to grab another lock while it has that |
| 943 | one. People using your code don't even need to know you are using a |
| 944 | lock. |
| 945 | |
| 946 | A classic problem here is when you provide callbacks or hooks: if you |
| 947 | call these with the lock held, you risk simple deadlock, or a deadly |
| 948 | embrace (who knows what the callback will do?). Remember, the other |
| 949 | programmers are out to get you, so don't do this. |
| 950 | |
| 951 | Overzealous Prevention Of Deadlocks |
| 952 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 953 | |
| 954 | Deadlocks are problematic, but not as bad as data corruption. Code which |
| 955 | grabs a read lock, searches a list, fails to find what it wants, drops |
| 956 | the read lock, grabs a write lock and inserts the object has a race |
| 957 | condition. |
| 958 | |
| 959 | If you don't see why, please stay the fuck away from my code. |
| 960 | |
| 961 | Racing Timers: A Kernel Pastime |
| 962 | ------------------------------- |
| 963 | |
| 964 | Timers can produce their own special problems with races. Consider a |
| 965 | collection of objects (list, hash, etc) where each object has a timer |
| 966 | which is due to destroy it. |
| 967 | |
| 968 | If you want to destroy the entire collection (say on module removal), |
| 969 | you might do the following:: |
| 970 | |
| 971 | /* THIS CODE BAD BAD BAD BAD: IF IT WAS ANY WORSE IT WOULD USE |
| 972 | HUNGARIAN NOTATION */ |
| 973 | spin_lock_bh(&list_lock); |
| 974 | |
| 975 | while (list) { |
| 976 | struct foo *next = list->next; |
| 977 | del_timer(&list->timer); |
| 978 | kfree(list); |
| 979 | list = next; |
| 980 | } |
| 981 | |
| 982 | spin_unlock_bh(&list_lock); |
| 983 | |
| 984 | |
| 985 | Sooner or later, this will crash on SMP, because a timer can have just |
| 986 | gone off before the :c:func:`spin_lock_bh()`, and it will only get |
| 987 | the lock after we :c:func:`spin_unlock_bh()`, and then try to free |
| 988 | the element (which has already been freed!). |
| 989 | |
| 990 | This can be avoided by checking the result of |
| 991 | :c:func:`del_timer()`: if it returns 1, the timer has been deleted. |
| 992 | If 0, it means (in this case) that it is currently running, so we can |
| 993 | do:: |
| 994 | |
| 995 | retry: |
| 996 | spin_lock_bh(&list_lock); |
| 997 | |
| 998 | while (list) { |
| 999 | struct foo *next = list->next; |
| 1000 | if (!del_timer(&list->timer)) { |
| 1001 | /* Give timer a chance to delete this */ |
| 1002 | spin_unlock_bh(&list_lock); |
| 1003 | goto retry; |
| 1004 | } |
| 1005 | kfree(list); |
| 1006 | list = next; |
| 1007 | } |
| 1008 | |
| 1009 | spin_unlock_bh(&list_lock); |
| 1010 | |
| 1011 | |
| 1012 | Another common problem is deleting timers which restart themselves (by |
| 1013 | calling :c:func:`add_timer()` at the end of their timer function). |
| 1014 | Because this is a fairly common case which is prone to races, you should |
| 1015 | use :c:func:`del_timer_sync()` (``include/linux/timer.h``) to |
| 1016 | handle this case. It returns the number of times the timer had to be |
| 1017 | deleted before we finally stopped it from adding itself back in. |
| 1018 | |
| 1019 | Locking Speed |
| 1020 | ============= |
| 1021 | |
| 1022 | There are three main things to worry about when considering speed of |
| 1023 | some code which does locking. First is concurrency: how many things are |
| 1024 | going to be waiting while someone else is holding a lock. Second is the |
| 1025 | time taken to actually acquire and release an uncontended lock. Third is |
| 1026 | using fewer, or smarter locks. I'm assuming that the lock is used fairly |
| 1027 | often: otherwise, you wouldn't be concerned about efficiency. |
| 1028 | |
| 1029 | Concurrency depends on how long the lock is usually held: you should |
| 1030 | hold the lock for as long as needed, but no longer. In the cache |
| 1031 | example, we always create the object without the lock held, and then |
| 1032 | grab the lock only when we are ready to insert it in the list. |
| 1033 | |
| 1034 | Acquisition times depend on how much damage the lock operations do to |
| 1035 | the pipeline (pipeline stalls) and how likely it is that this CPU was |
| 1036 | the last one to grab the lock (ie. is the lock cache-hot for this CPU): |
| 1037 | on a machine with more CPUs, this likelihood drops fast. Consider a |
| 1038 | 700MHz Intel Pentium III: an instruction takes about 0.7ns, an atomic |
| 1039 | increment takes about 58ns, a lock which is cache-hot on this CPU takes |
| 1040 | 160ns, and a cacheline transfer from another CPU takes an additional 170 |
| 1041 | to 360ns. (These figures from Paul McKenney's `Linux Journal RCU |
| 1042 | article <http://www.linuxjournal.com/article.php?sid=6993>`__). |
| 1043 | |
| 1044 | These two aims conflict: holding a lock for a short time might be done |
| 1045 | by splitting locks into parts (such as in our final per-object-lock |
| 1046 | example), but this increases the number of lock acquisitions, and the |
| 1047 | results are often slower than having a single lock. This is another |
| 1048 | reason to advocate locking simplicity. |
| 1049 | |
| 1050 | The third concern is addressed below: there are some methods to reduce |
| 1051 | the amount of locking which needs to be done. |
| 1052 | |
| 1053 | Read/Write Lock Variants |
| 1054 | ------------------------ |
| 1055 | |
| 1056 | Both spinlocks and mutexes have read/write variants: ``rwlock_t`` and |
| 1057 | :c:type:`struct rw_semaphore <rw_semaphore>`. These divide |
| 1058 | users into two classes: the readers and the writers. If you are only |
| 1059 | reading the data, you can get a read lock, but to write to the data you |
| 1060 | need the write lock. Many people can hold a read lock, but a writer must |
| 1061 | be sole holder. |
| 1062 | |
| 1063 | If your code divides neatly along reader/writer lines (as our cache code |
| 1064 | does), and the lock is held by readers for significant lengths of time, |
| 1065 | using these locks can help. They are slightly slower than the normal |
| 1066 | locks though, so in practice ``rwlock_t`` is not usually worthwhile. |
| 1067 | |
| 1068 | Avoiding Locks: Read Copy Update |
| 1069 | -------------------------------- |
| 1070 | |
| 1071 | There is a special method of read/write locking called Read Copy Update. |
| 1072 | Using RCU, the readers can avoid taking a lock altogether: as we expect |
| 1073 | our cache to be read more often than updated (otherwise the cache is a |
| 1074 | waste of time), it is a candidate for this optimization. |
| 1075 | |
| 1076 | How do we get rid of read locks? Getting rid of read locks means that |
| 1077 | writers may be changing the list underneath the readers. That is |
| 1078 | actually quite simple: we can read a linked list while an element is |
| 1079 | being added if the writer adds the element very carefully. For example, |
| 1080 | adding ``new`` to a single linked list called ``list``:: |
| 1081 | |
| 1082 | new->next = list->next; |
| 1083 | wmb(); |
| 1084 | list->next = new; |
| 1085 | |
| 1086 | |
| 1087 | The :c:func:`wmb()` is a write memory barrier. It ensures that the |
| 1088 | first operation (setting the new element's ``next`` pointer) is complete |
| 1089 | and will be seen by all CPUs, before the second operation is (putting |
| 1090 | the new element into the list). This is important, since modern |
| 1091 | compilers and modern CPUs can both reorder instructions unless told |
| 1092 | otherwise: we want a reader to either not see the new element at all, or |
| 1093 | see the new element with the ``next`` pointer correctly pointing at the |
| 1094 | rest of the list. |
| 1095 | |
| 1096 | Fortunately, there is a function to do this for standard |
| 1097 | :c:type:`struct list_head <list_head>` lists: |
| 1098 | :c:func:`list_add_rcu()` (``include/linux/list.h``). |
| 1099 | |
| 1100 | Removing an element from the list is even simpler: we replace the |
| 1101 | pointer to the old element with a pointer to its successor, and readers |
| 1102 | will either see it, or skip over it. |
| 1103 | |
| 1104 | :: |
| 1105 | |
| 1106 | list->next = old->next; |
| 1107 | |
| 1108 | |
| 1109 | There is :c:func:`list_del_rcu()` (``include/linux/list.h``) which |
| 1110 | does this (the normal version poisons the old object, which we don't |
| 1111 | want). |
| 1112 | |
| 1113 | The reader must also be careful: some CPUs can look through the ``next`` |
| 1114 | pointer to start reading the contents of the next element early, but |
| 1115 | don't realize that the pre-fetched contents is wrong when the ``next`` |
| 1116 | pointer changes underneath them. Once again, there is a |
| 1117 | :c:func:`list_for_each_entry_rcu()` (``include/linux/list.h``) |
| 1118 | to help you. Of course, writers can just use |
| 1119 | :c:func:`list_for_each_entry()`, since there cannot be two |
| 1120 | simultaneous writers. |
| 1121 | |
| 1122 | Our final dilemma is this: when can we actually destroy the removed |
| 1123 | element? Remember, a reader might be stepping through this element in |
| 1124 | the list right now: if we free this element and the ``next`` pointer |
| 1125 | changes, the reader will jump off into garbage and crash. We need to |
| 1126 | wait until we know that all the readers who were traversing the list |
| 1127 | when we deleted the element are finished. We use |
| 1128 | :c:func:`call_rcu()` to register a callback which will actually |
| 1129 | destroy the object once all pre-existing readers are finished. |
| 1130 | Alternatively, :c:func:`synchronize_rcu()` may be used to block |
| 1131 | until all pre-existing are finished. |
| 1132 | |
| 1133 | But how does Read Copy Update know when the readers are finished? The |
| 1134 | method is this: firstly, the readers always traverse the list inside |
| 1135 | :c:func:`rcu_read_lock()`/:c:func:`rcu_read_unlock()` pairs: |
| 1136 | these simply disable preemption so the reader won't go to sleep while |
| 1137 | reading the list. |
| 1138 | |
| 1139 | RCU then waits until every other CPU has slept at least once: since |
| 1140 | readers cannot sleep, we know that any readers which were traversing the |
| 1141 | list during the deletion are finished, and the callback is triggered. |
| 1142 | The real Read Copy Update code is a little more optimized than this, but |
| 1143 | this is the fundamental idea. |
| 1144 | |
| 1145 | :: |
| 1146 | |
| 1147 | --- cache.c.perobjectlock 2003-12-11 17:15:03.000000000 +1100 |
| 1148 | +++ cache.c.rcupdate 2003-12-11 17:55:14.000000000 +1100 |
| 1149 | @@ -1,15 +1,18 @@ |
| 1150 | #include <linux/list.h> |
| 1151 | #include <linux/slab.h> |
| 1152 | #include <linux/string.h> |
| 1153 | +#include <linux/rcupdate.h> |
| 1154 | #include <linux/mutex.h> |
| 1155 | #include <asm/errno.h> |
| 1156 | |
| 1157 | struct object |
| 1158 | { |
| 1159 | - /* These two protected by cache_lock. */ |
| 1160 | + /* This is protected by RCU */ |
| 1161 | struct list_head list; |
| 1162 | int popularity; |
| 1163 | |
| 1164 | + struct rcu_head rcu; |
| 1165 | + |
| 1166 | atomic_t refcnt; |
| 1167 | |
| 1168 | /* Doesn't change once created. */ |
| 1169 | @@ -40,7 +43,7 @@ |
| 1170 | { |
| 1171 | struct object *i; |
| 1172 | |
| 1173 | - list_for_each_entry(i, &cache, list) { |
| 1174 | + list_for_each_entry_rcu(i, &cache, list) { |
| 1175 | if (i->id == id) { |
| 1176 | i->popularity++; |
| 1177 | return i; |
| 1178 | @@ -49,19 +52,25 @@ |
| 1179 | return NULL; |
| 1180 | } |
| 1181 | |
| 1182 | +/* Final discard done once we know no readers are looking. */ |
| 1183 | +static void cache_delete_rcu(void *arg) |
| 1184 | +{ |
| 1185 | + object_put(arg); |
| 1186 | +} |
| 1187 | + |
| 1188 | /* Must be holding cache_lock */ |
| 1189 | static void __cache_delete(struct object *obj) |
| 1190 | { |
| 1191 | BUG_ON(!obj); |
| 1192 | - list_del(&obj->list); |
| 1193 | - object_put(obj); |
| 1194 | + list_del_rcu(&obj->list); |
| 1195 | cache_num--; |
| 1196 | + call_rcu(&obj->rcu, cache_delete_rcu); |
| 1197 | } |
| 1198 | |
| 1199 | /* Must be holding cache_lock */ |
| 1200 | static void __cache_add(struct object *obj) |
| 1201 | { |
| 1202 | - list_add(&obj->list, &cache); |
| 1203 | + list_add_rcu(&obj->list, &cache); |
| 1204 | if (++cache_num > MAX_CACHE_SIZE) { |
| 1205 | struct object *i, *outcast = NULL; |
| 1206 | list_for_each_entry(i, &cache, list) { |
| 1207 | @@ -104,12 +114,11 @@ |
| 1208 | struct object *cache_find(int id) |
| 1209 | { |
| 1210 | struct object *obj; |
| 1211 | - unsigned long flags; |
| 1212 | |
| 1213 | - spin_lock_irqsave(&cache_lock, flags); |
| 1214 | + rcu_read_lock(); |
| 1215 | obj = __cache_find(id); |
| 1216 | if (obj) |
| 1217 | object_get(obj); |
| 1218 | - spin_unlock_irqrestore(&cache_lock, flags); |
| 1219 | + rcu_read_unlock(); |
| 1220 | return obj; |
| 1221 | } |
| 1222 | |
| 1223 | Note that the reader will alter the popularity member in |
| 1224 | :c:func:`__cache_find()`, and now it doesn't hold a lock. One |
| 1225 | solution would be to make it an ``atomic_t``, but for this usage, we |
| 1226 | don't really care about races: an approximate result is good enough, so |
| 1227 | I didn't change it. |
| 1228 | |
| 1229 | The result is that :c:func:`cache_find()` requires no |
| 1230 | synchronization with any other functions, so is almost as fast on SMP as |
| 1231 | it would be on UP. |
| 1232 | |
| 1233 | There is a further optimization possible here: remember our original |
| 1234 | cache code, where there were no reference counts and the caller simply |
| 1235 | held the lock whenever using the object? This is still possible: if you |
| 1236 | hold the lock, no one can delete the object, so you don't need to get |
| 1237 | and put the reference count. |
| 1238 | |
| 1239 | Now, because the 'read lock' in RCU is simply disabling preemption, a |
| 1240 | caller which always has preemption disabled between calling |
| 1241 | :c:func:`cache_find()` and :c:func:`object_put()` does not |
| 1242 | need to actually get and put the reference count: we could expose |
| 1243 | :c:func:`__cache_find()` by making it non-static, and such |
| 1244 | callers could simply call that. |
| 1245 | |
| 1246 | The benefit here is that the reference count is not written to: the |
| 1247 | object is not altered in any way, which is much faster on SMP machines |
| 1248 | due to caching. |
| 1249 | |
| 1250 | Per-CPU Data |
| 1251 | ------------ |
| 1252 | |
| 1253 | Another technique for avoiding locking which is used fairly widely is to |
| 1254 | duplicate information for each CPU. For example, if you wanted to keep a |
| 1255 | count of a common condition, you could use a spin lock and a single |
| 1256 | counter. Nice and simple. |
| 1257 | |
| 1258 | If that was too slow (it's usually not, but if you've got a really big |
| 1259 | machine to test on and can show that it is), you could instead use a |
| 1260 | counter for each CPU, then none of them need an exclusive lock. See |
| 1261 | :c:func:`DEFINE_PER_CPU()`, :c:func:`get_cpu_var()` and |
| 1262 | :c:func:`put_cpu_var()` (``include/linux/percpu.h``). |
| 1263 | |
| 1264 | Of particular use for simple per-cpu counters is the ``local_t`` type, |
| 1265 | and the :c:func:`cpu_local_inc()` and related functions, which are |
| 1266 | more efficient than simple code on some architectures |
| 1267 | (``include/asm/local.h``). |
| 1268 | |
| 1269 | Note that there is no simple, reliable way of getting an exact value of |
| 1270 | such a counter, without introducing more locks. This is not a problem |
| 1271 | for some uses. |
| 1272 | |
| 1273 | Data Which Mostly Used By An IRQ Handler |
| 1274 | ---------------------------------------- |
| 1275 | |
| 1276 | If data is always accessed from within the same IRQ handler, you don't |
| 1277 | need a lock at all: the kernel already guarantees that the irq handler |
| 1278 | will not run simultaneously on multiple CPUs. |
| 1279 | |
| 1280 | Manfred Spraul points out that you can still do this, even if the data |
| 1281 | is very occasionally accessed in user context or softirqs/tasklets. The |
| 1282 | irq handler doesn't use a lock, and all other accesses are done as so:: |
| 1283 | |
| 1284 | spin_lock(&lock); |
| 1285 | disable_irq(irq); |
| 1286 | ... |
| 1287 | enable_irq(irq); |
| 1288 | spin_unlock(&lock); |
| 1289 | |
| 1290 | The :c:func:`disable_irq()` prevents the irq handler from running |
| 1291 | (and waits for it to finish if it's currently running on other CPUs). |
| 1292 | The spinlock prevents any other accesses happening at the same time. |
| 1293 | Naturally, this is slower than just a :c:func:`spin_lock_irq()` |
| 1294 | call, so it only makes sense if this type of access happens extremely |
| 1295 | rarely. |
| 1296 | |
| 1297 | What Functions Are Safe To Call From Interrupts? |
| 1298 | ================================================ |
| 1299 | |
| 1300 | Many functions in the kernel sleep (ie. call schedule()) directly or |
| 1301 | indirectly: you can never call them while holding a spinlock, or with |
| 1302 | preemption disabled. This also means you need to be in user context: |
| 1303 | calling them from an interrupt is illegal. |
| 1304 | |
| 1305 | Some Functions Which Sleep |
| 1306 | -------------------------- |
| 1307 | |
| 1308 | The most common ones are listed below, but you usually have to read the |
| 1309 | code to find out if other calls are safe. If everyone else who calls it |
| 1310 | can sleep, you probably need to be able to sleep, too. In particular, |
| 1311 | registration and deregistration functions usually expect to be called |
| 1312 | from user context, and can sleep. |
| 1313 | |
| 1314 | - Accesses to userspace: |
| 1315 | |
| 1316 | - :c:func:`copy_from_user()` |
| 1317 | |
| 1318 | - :c:func:`copy_to_user()` |
| 1319 | |
| 1320 | - :c:func:`get_user()` |
| 1321 | |
| 1322 | - :c:func:`put_user()` |
| 1323 | |
Mauro Carvalho Chehab | dc89fca | 2017-05-11 16:15:16 -0300 | [diff] [blame] | 1324 | - :c:func:`kmalloc(GFP_KERNEL) <kmalloc>` |
Mauro Carvalho Chehab | e548cde | 2017-05-11 09:55:30 -0300 | [diff] [blame] | 1325 | |
| 1326 | - :c:func:`mutex_lock_interruptible()` and |
| 1327 | :c:func:`mutex_lock()` |
| 1328 | |
| 1329 | There is a :c:func:`mutex_trylock()` which does not sleep. |
| 1330 | Still, it must not be used inside interrupt context since its |
| 1331 | implementation is not safe for that. :c:func:`mutex_unlock()` |
| 1332 | will also never sleep. It cannot be used in interrupt context either |
| 1333 | since a mutex must be released by the same task that acquired it. |
| 1334 | |
| 1335 | Some Functions Which Don't Sleep |
| 1336 | -------------------------------- |
| 1337 | |
| 1338 | Some functions are safe to call from any context, or holding almost any |
| 1339 | lock. |
| 1340 | |
| 1341 | - :c:func:`printk()` |
| 1342 | |
| 1343 | - :c:func:`kfree()` |
| 1344 | |
| 1345 | - :c:func:`add_timer()` and :c:func:`del_timer()` |
| 1346 | |
| 1347 | Mutex API reference |
| 1348 | =================== |
| 1349 | |
| 1350 | .. kernel-doc:: include/linux/mutex.h |
| 1351 | :internal: |
| 1352 | |
| 1353 | .. kernel-doc:: kernel/locking/mutex.c |
| 1354 | :export: |
| 1355 | |
| 1356 | Futex API reference |
| 1357 | =================== |
| 1358 | |
| 1359 | .. kernel-doc:: kernel/futex.c |
| 1360 | :internal: |
| 1361 | |
| 1362 | Further reading |
| 1363 | =============== |
| 1364 | |
| 1365 | - ``Documentation/locking/spinlocks.txt``: Linus Torvalds' spinlocking |
| 1366 | tutorial in the kernel sources. |
| 1367 | |
| 1368 | - Unix Systems for Modern Architectures: Symmetric Multiprocessing and |
| 1369 | Caching for Kernel Programmers: |
| 1370 | |
| 1371 | Curt Schimmel's very good introduction to kernel level locking (not |
| 1372 | written for Linux, but nearly everything applies). The book is |
| 1373 | expensive, but really worth every penny to understand SMP locking. |
| 1374 | [ISBN: 0201633388] |
| 1375 | |
| 1376 | Thanks |
| 1377 | ====== |
| 1378 | |
| 1379 | Thanks to Telsa Gwynne for DocBooking, neatening and adding style. |
| 1380 | |
| 1381 | Thanks to Martin Pool, Philipp Rumpf, Stephen Rothwell, Paul Mackerras, |
| 1382 | Ruedi Aschwanden, Alan Cox, Manfred Spraul, Tim Waugh, Pete Zaitcev, |
| 1383 | James Morris, Robert Love, Paul McKenney, John Ashby for proofreading, |
| 1384 | correcting, flaming, commenting. |
| 1385 | |
| 1386 | Thanks to the cabal for having no influence on this document. |
| 1387 | |
| 1388 | Glossary |
| 1389 | ======== |
| 1390 | |
| 1391 | preemption |
| 1392 | Prior to 2.5, or when ``CONFIG_PREEMPT`` is unset, processes in user |
| 1393 | context inside the kernel would not preempt each other (ie. you had that |
| 1394 | CPU until you gave it up, except for interrupts). With the addition of |
| 1395 | ``CONFIG_PREEMPT`` in 2.5.4, this changed: when in user context, higher |
| 1396 | priority tasks can "cut in": spinlocks were changed to disable |
| 1397 | preemption, even on UP. |
| 1398 | |
| 1399 | bh |
| 1400 | Bottom Half: for historical reasons, functions with '_bh' in them often |
| 1401 | now refer to any software interrupt, e.g. :c:func:`spin_lock_bh()` |
| 1402 | blocks any software interrupt on the current CPU. Bottom halves are |
| 1403 | deprecated, and will eventually be replaced by tasklets. Only one bottom |
| 1404 | half will be running at any time. |
| 1405 | |
| 1406 | Hardware Interrupt / Hardware IRQ |
| 1407 | Hardware interrupt request. :c:func:`in_irq()` returns true in a |
| 1408 | hardware interrupt handler. |
| 1409 | |
| 1410 | Interrupt Context |
| 1411 | Not user context: processing a hardware irq or software irq. Indicated |
| 1412 | by the :c:func:`in_interrupt()` macro returning true. |
| 1413 | |
| 1414 | SMP |
| 1415 | Symmetric Multi-Processor: kernels compiled for multiple-CPU machines. |
| 1416 | (``CONFIG_SMP=y``). |
| 1417 | |
| 1418 | Software Interrupt / softirq |
| 1419 | Software interrupt handler. :c:func:`in_irq()` returns false; |
| 1420 | :c:func:`in_softirq()` returns true. Tasklets and softirqs both |
| 1421 | fall into the category of 'software interrupts'. |
| 1422 | |
| 1423 | Strictly speaking a softirq is one of up to 32 enumerated software |
| 1424 | interrupts which can run on multiple CPUs at once. Sometimes used to |
| 1425 | refer to tasklets as well (ie. all software interrupts). |
| 1426 | |
| 1427 | tasklet |
| 1428 | A dynamically-registrable software interrupt, which is guaranteed to |
| 1429 | only run on one CPU at a time. |
| 1430 | |
| 1431 | timer |
| 1432 | A dynamically-registrable software interrupt, which is run at (or close |
| 1433 | to) a given time. When running, it is just like a tasklet (in fact, they |
Mauro Carvalho Chehab | dc89fca | 2017-05-11 16:15:16 -0300 | [diff] [blame] | 1434 | are called from the ``TIMER_SOFTIRQ``). |
Mauro Carvalho Chehab | e548cde | 2017-05-11 09:55:30 -0300 | [diff] [blame] | 1435 | |
| 1436 | UP |
Mauro Carvalho Chehab | dc89fca | 2017-05-11 16:15:16 -0300 | [diff] [blame] | 1437 | Uni-Processor: Non-SMP. (``CONFIG_SMP=n``). |
Mauro Carvalho Chehab | e548cde | 2017-05-11 09:55:30 -0300 | [diff] [blame] | 1438 | |
| 1439 | User Context |
| 1440 | The kernel executing on behalf of a particular process (ie. a system |
| 1441 | call or trap) or kernel thread. You can tell which process with the |
| 1442 | ``current`` macro.) Not to be confused with userspace. Can be |
| 1443 | interrupted by software or hardware interrupts. |
| 1444 | |
| 1445 | Userspace |
| 1446 | A process executing its own code outside the kernel. |