Root/Documentation/hwspinlock.txt

Source at commit ec7cab4cbb721bff91ec924ec691efd8daf36579 created 12 years 8 months ago.
By Maarten ter Huurne, MIPS: JZ4740: A320: Updated quickstart documentation.
1Hardware Spinlock Framework
2
31. Introduction
4
5Hardware spinlock modules provide hardware assistance for synchronization
6and mutual exclusion between heterogeneous processors and those not operating
7under a single, shared operating system.
8
9For example, OMAP4 has dual Cortex-A9, dual Cortex-M3 and a C64x+ DSP,
10each of which is running a different Operating System (the master, A9,
11is usually running Linux and the slave processors, the M3 and the DSP,
12are running some flavor of RTOS).
13
14A generic hwspinlock framework allows platform-independent drivers to use
15the hwspinlock device in order to access data structures that are shared
16between remote processors, that otherwise have no alternative mechanism
17to accomplish synchronization and mutual exclusion operations.
18
19This is necessary, for example, for Inter-processor communications:
20on OMAP4, cpu-intensive multimedia tasks are offloaded by the host to the
21remote M3 and/or C64x+ slave processors (by an IPC subsystem called Syslink).
22
23To achieve fast message-based communications, a minimal kernel support
24is needed to deliver messages arriving from a remote processor to the
25appropriate user process.
26
27This communication is based on simple data structures that is shared between
28the remote processors, and access to it is synchronized using the hwspinlock
29module (remote processor directly places new messages in this shared data
30structure).
31
32A common hwspinlock interface makes it possible to have generic, platform-
33independent, drivers.
34
352. User API
36
37  struct hwspinlock *hwspin_lock_request(void);
38   - dynamically assign an hwspinlock and return its address, or NULL
39     in case an unused hwspinlock isn't available. Users of this
40     API will usually want to communicate the lock's id to the remote core
41     before it can be used to achieve synchronization.
42     Can be called from an atomic context (this function will not sleep) but
43     not from within interrupt context.
44
45  struct hwspinlock *hwspin_lock_request_specific(unsigned int id);
46   - assign a specific hwspinlock id and return its address, or NULL
47     if that hwspinlock is already in use. Usually board code will
48     be calling this function in order to reserve specific hwspinlock
49     ids for predefined purposes.
50     Can be called from an atomic context (this function will not sleep) but
51     not from within interrupt context.
52
53  int hwspin_lock_free(struct hwspinlock *hwlock);
54   - free a previously-assigned hwspinlock; returns 0 on success, or an
55     appropriate error code on failure (e.g. -EINVAL if the hwspinlock
56     is already free).
57     Can be called from an atomic context (this function will not sleep) but
58     not from within interrupt context.
59
60  int hwspin_lock_timeout(struct hwspinlock *hwlock, unsigned int timeout);
61   - lock a previously-assigned hwspinlock with a timeout limit (specified in
62     msecs). If the hwspinlock is already taken, the function will busy loop
63     waiting for it to be released, but give up when the timeout elapses.
64     Upon a successful return from this function, preemption is disabled so
65     the caller must not sleep, and is advised to release the hwspinlock as
66     soon as possible, in order to minimize remote cores polling on the
67     hardware interconnect.
68     Returns 0 when successful and an appropriate error code otherwise (most
69     notably -ETIMEDOUT if the hwspinlock is still busy after timeout msecs).
70     The function will never sleep.
71
72  int hwspin_lock_timeout_irq(struct hwspinlock *hwlock, unsigned int timeout);
73   - lock a previously-assigned hwspinlock with a timeout limit (specified in
74     msecs). If the hwspinlock is already taken, the function will busy loop
75     waiting for it to be released, but give up when the timeout elapses.
76     Upon a successful return from this function, preemption and the local
77     interrupts are disabled, so the caller must not sleep, and is advised to
78     release the hwspinlock as soon as possible.
79     Returns 0 when successful and an appropriate error code otherwise (most
80     notably -ETIMEDOUT if the hwspinlock is still busy after timeout msecs).
81     The function will never sleep.
82
83  int hwspin_lock_timeout_irqsave(struct hwspinlock *hwlock, unsigned int to,
84                            unsigned long *flags);
85   - lock a previously-assigned hwspinlock with a timeout limit (specified in
86     msecs). If the hwspinlock is already taken, the function will busy loop
87     waiting for it to be released, but give up when the timeout elapses.
88     Upon a successful return from this function, preemption is disabled,
89     local interrupts are disabled and their previous state is saved at the
90     given flags placeholder. The caller must not sleep, and is advised to
91     release the hwspinlock as soon as possible.
92     Returns 0 when successful and an appropriate error code otherwise (most
93     notably -ETIMEDOUT if the hwspinlock is still busy after timeout msecs).
94     The function will never sleep.
95
96  int hwspin_trylock(struct hwspinlock *hwlock);
97   - attempt to lock a previously-assigned hwspinlock, but immediately fail if
98     it is already taken.
99     Upon a successful return from this function, preemption is disabled so
100     caller must not sleep, and is advised to release the hwspinlock as soon as
101     possible, in order to minimize remote cores polling on the hardware
102     interconnect.
103     Returns 0 on success and an appropriate error code otherwise (most
104     notably -EBUSY if the hwspinlock was already taken).
105     The function will never sleep.
106
107  int hwspin_trylock_irq(struct hwspinlock *hwlock);
108   - attempt to lock a previously-assigned hwspinlock, but immediately fail if
109     it is already taken.
110     Upon a successful return from this function, preemption and the local
111     interrupts are disabled so caller must not sleep, and is advised to
112     release the hwspinlock as soon as possible.
113     Returns 0 on success and an appropriate error code otherwise (most
114     notably -EBUSY if the hwspinlock was already taken).
115     The function will never sleep.
116
117  int hwspin_trylock_irqsave(struct hwspinlock *hwlock, unsigned long *flags);
118   - attempt to lock a previously-assigned hwspinlock, but immediately fail if
119     it is already taken.
120     Upon a successful return from this function, preemption is disabled,
121     the local interrupts are disabled and their previous state is saved
122     at the given flags placeholder. The caller must not sleep, and is advised
123     to release the hwspinlock as soon as possible.
124     Returns 0 on success and an appropriate error code otherwise (most
125     notably -EBUSY if the hwspinlock was already taken).
126     The function will never sleep.
127
128  void hwspin_unlock(struct hwspinlock *hwlock);
129   - unlock a previously-locked hwspinlock. Always succeed, and can be called
130     from any context (the function never sleeps). Note: code should _never_
131     unlock an hwspinlock which is already unlocked (there is no protection
132     against this).
133
134  void hwspin_unlock_irq(struct hwspinlock *hwlock);
135   - unlock a previously-locked hwspinlock and enable local interrupts.
136     The caller should _never_ unlock an hwspinlock which is already unlocked.
137     Doing so is considered a bug (there is no protection against this).
138     Upon a successful return from this function, preemption and local
139     interrupts are enabled. This function will never sleep.
140
141  void
142  hwspin_unlock_irqrestore(struct hwspinlock *hwlock, unsigned long *flags);
143   - unlock a previously-locked hwspinlock.
144     The caller should _never_ unlock an hwspinlock which is already unlocked.
145     Doing so is considered a bug (there is no protection against this).
146     Upon a successful return from this function, preemption is reenabled,
147     and the state of the local interrupts is restored to the state saved at
148     the given flags. This function will never sleep.
149
150  int hwspin_lock_get_id(struct hwspinlock *hwlock);
151   - retrieve id number of a given hwspinlock. This is needed when an
152     hwspinlock is dynamically assigned: before it can be used to achieve
153     mutual exclusion with a remote cpu, the id number should be communicated
154     to the remote task with which we want to synchronize.
155     Returns the hwspinlock id number, or -EINVAL if hwlock is null.
156
1573. Typical usage
158
159#include <linux/hwspinlock.h>
160#include <linux/err.h>
161
162int hwspinlock_example1(void)
163{
164    struct hwspinlock *hwlock;
165    int ret;
166
167    /* dynamically assign a hwspinlock */
168    hwlock = hwspin_lock_request();
169    if (!hwlock)
170        ...
171
172    id = hwspin_lock_get_id(hwlock);
173    /* probably need to communicate id to a remote processor now */
174
175    /* take the lock, spin for 1 sec if it's already taken */
176    ret = hwspin_lock_timeout(hwlock, 1000);
177    if (ret)
178        ...
179
180    /*
181     * we took the lock, do our thing now, but do NOT sleep
182     */
183
184    /* release the lock */
185    hwspin_unlock(hwlock);
186
187    /* free the lock */
188    ret = hwspin_lock_free(hwlock);
189    if (ret)
190        ...
191
192    return ret;
193}
194
195int hwspinlock_example2(void)
196{
197    struct hwspinlock *hwlock;
198    int ret;
199
200    /*
201     * assign a specific hwspinlock id - this should be called early
202     * by board init code.
203     */
204    hwlock = hwspin_lock_request_specific(PREDEFINED_LOCK_ID);
205    if (!hwlock)
206        ...
207
208    /* try to take it, but don't spin on it */
209    ret = hwspin_trylock(hwlock);
210    if (!ret) {
211        pr_info("lock is already taken\n");
212        return -EBUSY;
213    }
214
215    /*
216     * we took the lock, do our thing now, but do NOT sleep
217     */
218
219    /* release the lock */
220    hwspin_unlock(hwlock);
221
222    /* free the lock */
223    ret = hwspin_lock_free(hwlock);
224    if (ret)
225        ...
226
227    return ret;
228}
229
230
2314. API for implementors
232
233  int hwspin_lock_register(struct hwspinlock *hwlock);
234   - to be called from the underlying platform-specific implementation, in
235     order to register a new hwspinlock instance. Can be called from an atomic
236     context (this function will not sleep) but not from within interrupt
237     context. Returns 0 on success, or appropriate error code on failure.
238
239  struct hwspinlock *hwspin_lock_unregister(unsigned int id);
240   - to be called from the underlying vendor-specific implementation, in order
241     to unregister an existing (and unused) hwspinlock instance.
242     Can be called from an atomic context (will not sleep) but not from
243     within interrupt context.
244     Returns the address of hwspinlock on success, or NULL on error (e.g.
245     if the hwspinlock is sill in use).
246
2475. struct hwspinlock
248
249This struct represents an hwspinlock instance. It is registered by the
250underlying hwspinlock implementation using the hwspin_lock_register() API.
251
252/**
253 * struct hwspinlock - vendor-specific hwspinlock implementation
254 *
255 * @dev: underlying device, will be used with runtime PM api
256 * @ops: vendor-specific hwspinlock handlers
257 * @id: a global, unique, system-wide, index of the lock.
258 * @lock: initialized and used by hwspinlock core
259 * @owner: underlying implementation module, used to maintain module ref count
260 */
261struct hwspinlock {
262    struct device *dev;
263    const struct hwspinlock_ops *ops;
264    int id;
265    spinlock_t lock;
266    struct module *owner;
267};
268
269The underlying implementation is responsible to assign the dev, ops, id and
270owner members. The lock member, OTOH, is initialized and used by the hwspinlock
271core.
272
2736. Implementation callbacks
274
275There are three possible callbacks defined in 'struct hwspinlock_ops':
276
277struct hwspinlock_ops {
278    int (*trylock)(struct hwspinlock *lock);
279    void (*unlock)(struct hwspinlock *lock);
280    void (*relax)(struct hwspinlock *lock);
281};
282
283The first two callbacks are mandatory:
284
285The ->trylock() callback should make a single attempt to take the lock, and
286return 0 on failure and 1 on success. This callback may _not_ sleep.
287
288The ->unlock() callback releases the lock. It always succeed, and it, too,
289may _not_ sleep.
290
291The ->relax() callback is optional. It is called by hwspinlock core while
292spinning on a lock, and can be used by the underlying implementation to force
293a delay between two successive invocations of ->trylock(). It may _not_ sleep.
294

Archive Download this file



interactive