4 * Copyright (C) 2008 Steven Rostedt <srostedt@redhat.com>
6 #include <linux/trace_events.h>
7 #include <linux/ring_buffer.h>
8 #include <linux/trace_clock.h>
9 #include <linux/sched/clock.h>
10 #include <linux/trace_seq.h>
11 #include <linux/spinlock.h>
12 #include <linux/irq_work.h>
13 #include <linux/uaccess.h>
14 #include <linux/hardirq.h>
15 #include <linux/kthread.h> /* for self test */
16 #include <linux/kmemcheck.h>
17 #include <linux/module.h>
18 #include <linux/percpu.h>
19 #include <linux/mutex.h>
20 #include <linux/delay.h>
21 #include <linux/slab.h>
22 #include <linux/init.h>
23 #include <linux/hash.h>
24 #include <linux/list.h>
25 #include <linux/cpu.h>
27 #include <asm/local.h>
29 static void update_pages_handler(struct work_struct
*work
);
32 * The ring buffer header is special. We must manually up keep it.
34 int ring_buffer_print_entry_header(struct trace_seq
*s
)
36 trace_seq_puts(s
, "# compressed entry header\n");
37 trace_seq_puts(s
, "\ttype_len : 5 bits\n");
38 trace_seq_puts(s
, "\ttime_delta : 27 bits\n");
39 trace_seq_puts(s
, "\tarray : 32 bits\n");
40 trace_seq_putc(s
, '\n');
41 trace_seq_printf(s
, "\tpadding : type == %d\n",
42 RINGBUF_TYPE_PADDING
);
43 trace_seq_printf(s
, "\ttime_extend : type == %d\n",
44 RINGBUF_TYPE_TIME_EXTEND
);
45 trace_seq_printf(s
, "\tdata max type_len == %d\n",
46 RINGBUF_TYPE_DATA_TYPE_LEN_MAX
);
48 return !trace_seq_has_overflowed(s
);
52 * The ring buffer is made up of a list of pages. A separate list of pages is
53 * allocated for each CPU. A writer may only write to a buffer that is
54 * associated with the CPU it is currently executing on. A reader may read
55 * from any per cpu buffer.
57 * The reader is special. For each per cpu buffer, the reader has its own
58 * reader page. When a reader has read the entire reader page, this reader
59 * page is swapped with another page in the ring buffer.
61 * Now, as long as the writer is off the reader page, the reader can do what
62 * ever it wants with that page. The writer will never write to that page
63 * again (as long as it is out of the ring buffer).
65 * Here's some silly ASCII art.
68 * |reader| RING BUFFER
70 * +------+ +---+ +---+ +---+
79 * |reader| RING BUFFER
80 * |page |------------------v
81 * +------+ +---+ +---+ +---+
90 * |reader| RING BUFFER
91 * |page |------------------v
92 * +------+ +---+ +---+ +---+
97 * +------------------------------+
101 * |buffer| RING BUFFER
102 * |page |------------------v
103 * +------+ +---+ +---+ +---+
105 * | New +---+ +---+ +---+
108 * +------------------------------+
111 * After we make this swap, the reader can hand this page off to the splice
112 * code and be done with it. It can even allocate a new page if it needs to
113 * and swap that into the ring buffer.
115 * We will be using cmpxchg soon to make all this lockless.
119 /* Used for individual buffers (after the counter) */
120 #define RB_BUFFER_OFF (1 << 20)
122 #define BUF_PAGE_HDR_SIZE offsetof(struct buffer_data_page, data)
124 #define RB_EVNT_HDR_SIZE (offsetof(struct ring_buffer_event, array))
125 #define RB_ALIGNMENT 4U
126 #define RB_MAX_SMALL_DATA (RB_ALIGNMENT * RINGBUF_TYPE_DATA_TYPE_LEN_MAX)
127 #define RB_EVNT_MIN_SIZE 8U /* two 32bit words */
129 #ifndef CONFIG_HAVE_64BIT_ALIGNED_ACCESS
130 # define RB_FORCE_8BYTE_ALIGNMENT 0
131 # define RB_ARCH_ALIGNMENT RB_ALIGNMENT
133 # define RB_FORCE_8BYTE_ALIGNMENT 1
134 # define RB_ARCH_ALIGNMENT 8U
137 #define RB_ALIGN_DATA __aligned(RB_ARCH_ALIGNMENT)
139 /* define RINGBUF_TYPE_DATA for 'case RINGBUF_TYPE_DATA:' */
140 #define RINGBUF_TYPE_DATA 0 ... RINGBUF_TYPE_DATA_TYPE_LEN_MAX
143 RB_LEN_TIME_EXTEND
= 8,
144 RB_LEN_TIME_STAMP
= 16,
147 #define skip_time_extend(event) \
148 ((struct ring_buffer_event *)((char *)event + RB_LEN_TIME_EXTEND))
150 static inline int rb_null_event(struct ring_buffer_event
*event
)
152 return event
->type_len
== RINGBUF_TYPE_PADDING
&& !event
->time_delta
;
155 static void rb_event_set_padding(struct ring_buffer_event
*event
)
157 /* padding has a NULL time_delta */
158 event
->type_len
= RINGBUF_TYPE_PADDING
;
159 event
->time_delta
= 0;
163 rb_event_data_length(struct ring_buffer_event
*event
)
168 length
= event
->type_len
* RB_ALIGNMENT
;
170 length
= event
->array
[0];
171 return length
+ RB_EVNT_HDR_SIZE
;
175 * Return the length of the given event. Will return
176 * the length of the time extend if the event is a
179 static inline unsigned
180 rb_event_length(struct ring_buffer_event
*event
)
182 switch (event
->type_len
) {
183 case RINGBUF_TYPE_PADDING
:
184 if (rb_null_event(event
))
187 return event
->array
[0] + RB_EVNT_HDR_SIZE
;
189 case RINGBUF_TYPE_TIME_EXTEND
:
190 return RB_LEN_TIME_EXTEND
;
192 case RINGBUF_TYPE_TIME_STAMP
:
193 return RB_LEN_TIME_STAMP
;
195 case RINGBUF_TYPE_DATA
:
196 return rb_event_data_length(event
);
205 * Return total length of time extend and data,
206 * or just the event length for all other events.
208 static inline unsigned
209 rb_event_ts_length(struct ring_buffer_event
*event
)
213 if (event
->type_len
== RINGBUF_TYPE_TIME_EXTEND
) {
214 /* time extends include the data event after it */
215 len
= RB_LEN_TIME_EXTEND
;
216 event
= skip_time_extend(event
);
218 return len
+ rb_event_length(event
);
222 * ring_buffer_event_length - return the length of the event
223 * @event: the event to get the length of
225 * Returns the size of the data load of a data event.
226 * If the event is something other than a data event, it
227 * returns the size of the event itself. With the exception
228 * of a TIME EXTEND, where it still returns the size of the
229 * data load of the data event after it.
231 unsigned ring_buffer_event_length(struct ring_buffer_event
*event
)
235 if (event
->type_len
== RINGBUF_TYPE_TIME_EXTEND
)
236 event
= skip_time_extend(event
);
238 length
= rb_event_length(event
);
239 if (event
->type_len
> RINGBUF_TYPE_DATA_TYPE_LEN_MAX
)
241 length
-= RB_EVNT_HDR_SIZE
;
242 if (length
> RB_MAX_SMALL_DATA
+ sizeof(event
->array
[0]))
243 length
-= sizeof(event
->array
[0]);
246 EXPORT_SYMBOL_GPL(ring_buffer_event_length
);
248 /* inline for ring buffer fast paths */
249 static __always_inline
void *
250 rb_event_data(struct ring_buffer_event
*event
)
252 if (event
->type_len
== RINGBUF_TYPE_TIME_EXTEND
)
253 event
= skip_time_extend(event
);
254 BUG_ON(event
->type_len
> RINGBUF_TYPE_DATA_TYPE_LEN_MAX
);
255 /* If length is in len field, then array[0] has the data */
257 return (void *)&event
->array
[0];
258 /* Otherwise length is in array[0] and array[1] has the data */
259 return (void *)&event
->array
[1];
263 * ring_buffer_event_data - return the data of the event
264 * @event: the event to get the data from
266 void *ring_buffer_event_data(struct ring_buffer_event
*event
)
268 return rb_event_data(event
);
270 EXPORT_SYMBOL_GPL(ring_buffer_event_data
);
272 #define for_each_buffer_cpu(buffer, cpu) \
273 for_each_cpu(cpu, buffer->cpumask)
276 #define TS_MASK ((1ULL << TS_SHIFT) - 1)
277 #define TS_DELTA_TEST (~TS_MASK)
279 /* Flag when events were overwritten */
280 #define RB_MISSED_EVENTS (1 << 31)
281 /* Missed count stored at end */
282 #define RB_MISSED_STORED (1 << 30)
284 struct buffer_data_page
{
285 u64 time_stamp
; /* page time stamp */
286 local_t commit
; /* write committed index */
287 unsigned char data
[] RB_ALIGN_DATA
; /* data of buffer page */
291 * Note, the buffer_page list must be first. The buffer pages
292 * are allocated in cache lines, which means that each buffer
293 * page will be at the beginning of a cache line, and thus
294 * the least significant bits will be zero. We use this to
295 * add flags in the list struct pointers, to make the ring buffer
299 struct list_head list
; /* list of buffer pages */
300 local_t write
; /* index for next write */
301 unsigned read
; /* index for next read */
302 local_t entries
; /* entries on this page */
303 unsigned long real_end
; /* real end of data */
304 struct buffer_data_page
*page
; /* Actual data page */
308 * The buffer page counters, write and entries, must be reset
309 * atomically when crossing page boundaries. To synchronize this
310 * update, two counters are inserted into the number. One is
311 * the actual counter for the write position or count on the page.
313 * The other is a counter of updaters. Before an update happens
314 * the update partition of the counter is incremented. This will
315 * allow the updater to update the counter atomically.
317 * The counter is 20 bits, and the state data is 12.
319 #define RB_WRITE_MASK 0xfffff
320 #define RB_WRITE_INTCNT (1 << 20)
322 static void rb_init_page(struct buffer_data_page
*bpage
)
324 local_set(&bpage
->commit
, 0);
328 * ring_buffer_page_len - the size of data on the page.
329 * @page: The page to read
331 * Returns the amount of data on the page, including buffer page header.
333 size_t ring_buffer_page_len(void *page
)
335 return local_read(&((struct buffer_data_page
*)page
)->commit
)
340 * Also stolen from mm/slob.c. Thanks to Mathieu Desnoyers for pointing
343 static void free_buffer_page(struct buffer_page
*bpage
)
345 free_page((unsigned long)bpage
->page
);
350 * We need to fit the time_stamp delta into 27 bits.
352 static inline int test_time_stamp(u64 delta
)
354 if (delta
& TS_DELTA_TEST
)
359 #define BUF_PAGE_SIZE (PAGE_SIZE - BUF_PAGE_HDR_SIZE)
361 /* Max payload is BUF_PAGE_SIZE - header (8bytes) */
362 #define BUF_MAX_DATA_SIZE (BUF_PAGE_SIZE - (sizeof(u32) * 2))
364 int ring_buffer_print_page_header(struct trace_seq
*s
)
366 struct buffer_data_page field
;
368 trace_seq_printf(s
, "\tfield: u64 timestamp;\t"
369 "offset:0;\tsize:%u;\tsigned:%u;\n",
370 (unsigned int)sizeof(field
.time_stamp
),
371 (unsigned int)is_signed_type(u64
));
373 trace_seq_printf(s
, "\tfield: local_t commit;\t"
374 "offset:%u;\tsize:%u;\tsigned:%u;\n",
375 (unsigned int)offsetof(typeof(field
), commit
),
376 (unsigned int)sizeof(field
.commit
),
377 (unsigned int)is_signed_type(long));
379 trace_seq_printf(s
, "\tfield: int overwrite;\t"
380 "offset:%u;\tsize:%u;\tsigned:%u;\n",
381 (unsigned int)offsetof(typeof(field
), commit
),
383 (unsigned int)is_signed_type(long));
385 trace_seq_printf(s
, "\tfield: char data;\t"
386 "offset:%u;\tsize:%u;\tsigned:%u;\n",
387 (unsigned int)offsetof(typeof(field
), data
),
388 (unsigned int)BUF_PAGE_SIZE
,
389 (unsigned int)is_signed_type(char));
391 return !trace_seq_has_overflowed(s
);
395 struct irq_work work
;
396 wait_queue_head_t waiters
;
397 wait_queue_head_t full_waiters
;
398 bool waiters_pending
;
399 bool full_waiters_pending
;
404 * Structure to hold event state and handle nested events.
406 struct rb_event_info
{
409 unsigned long length
;
410 struct buffer_page
*tail_page
;
415 * Used for which event context the event is in.
421 * See trace_recursive_lock() comment below for more details.
432 * head_page == tail_page && head == tail then buffer is empty.
434 struct ring_buffer_per_cpu
{
436 atomic_t record_disabled
;
437 struct ring_buffer
*buffer
;
438 raw_spinlock_t reader_lock
; /* serialize readers */
439 arch_spinlock_t lock
;
440 struct lock_class_key lock_key
;
441 struct buffer_data_page
*free_page
;
442 unsigned long nr_pages
;
443 unsigned int current_context
;
444 struct list_head
*pages
;
445 struct buffer_page
*head_page
; /* read from head */
446 struct buffer_page
*tail_page
; /* write to tail */
447 struct buffer_page
*commit_page
; /* committed pages */
448 struct buffer_page
*reader_page
;
449 unsigned long lost_events
;
450 unsigned long last_overrun
;
451 local_t entries_bytes
;
454 local_t commit_overrun
;
455 local_t dropped_events
;
459 unsigned long read_bytes
;
462 /* ring buffer pages to update, > 0 to add, < 0 to remove */
463 long nr_pages_to_update
;
464 struct list_head new_pages
; /* new pages to add */
465 struct work_struct update_pages_work
;
466 struct completion update_done
;
468 struct rb_irq_work irq_work
;
474 atomic_t record_disabled
;
475 atomic_t resize_disabled
;
476 cpumask_var_t cpumask
;
478 struct lock_class_key
*reader_lock_key
;
482 struct ring_buffer_per_cpu
**buffers
;
484 struct hlist_node node
;
487 struct rb_irq_work irq_work
;
490 struct ring_buffer_iter
{
491 struct ring_buffer_per_cpu
*cpu_buffer
;
493 struct buffer_page
*head_page
;
494 struct buffer_page
*cache_reader_page
;
495 unsigned long cache_read
;
500 * rb_wake_up_waiters - wake up tasks waiting for ring buffer input
502 * Schedules a delayed work to wake up any task that is blocked on the
503 * ring buffer waiters queue.
505 static void rb_wake_up_waiters(struct irq_work
*work
)
507 struct rb_irq_work
*rbwork
= container_of(work
, struct rb_irq_work
, work
);
509 wake_up_all(&rbwork
->waiters
);
510 if (rbwork
->wakeup_full
) {
511 rbwork
->wakeup_full
= false;
512 wake_up_all(&rbwork
->full_waiters
);
517 * ring_buffer_wait - wait for input to the ring buffer
518 * @buffer: buffer to wait on
519 * @cpu: the cpu buffer to wait on
520 * @full: wait until a full page is available, if @cpu != RING_BUFFER_ALL_CPUS
522 * If @cpu == RING_BUFFER_ALL_CPUS then the task will wake up as soon
523 * as data is added to any of the @buffer's cpu buffers. Otherwise
524 * it will wait for data to be added to a specific cpu buffer.
526 int ring_buffer_wait(struct ring_buffer
*buffer
, int cpu
, bool full
)
528 struct ring_buffer_per_cpu
*uninitialized_var(cpu_buffer
);
530 struct rb_irq_work
*work
;
534 * Depending on what the caller is waiting for, either any
535 * data in any cpu buffer, or a specific buffer, put the
536 * caller on the appropriate wait queue.
538 if (cpu
== RING_BUFFER_ALL_CPUS
) {
539 work
= &buffer
->irq_work
;
540 /* Full only makes sense on per cpu reads */
543 if (!cpumask_test_cpu(cpu
, buffer
->cpumask
))
545 cpu_buffer
= buffer
->buffers
[cpu
];
546 work
= &cpu_buffer
->irq_work
;
552 prepare_to_wait(&work
->full_waiters
, &wait
, TASK_INTERRUPTIBLE
);
554 prepare_to_wait(&work
->waiters
, &wait
, TASK_INTERRUPTIBLE
);
557 * The events can happen in critical sections where
558 * checking a work queue can cause deadlocks.
559 * After adding a task to the queue, this flag is set
560 * only to notify events to try to wake up the queue
563 * We don't clear it even if the buffer is no longer
564 * empty. The flag only causes the next event to run
565 * irq_work to do the work queue wake up. The worse
566 * that can happen if we race with !trace_empty() is that
567 * an event will cause an irq_work to try to wake up
570 * There's no reason to protect this flag either, as
571 * the work queue and irq_work logic will do the necessary
572 * synchronization for the wake ups. The only thing
573 * that is necessary is that the wake up happens after
574 * a task has been queued. It's OK for spurious wake ups.
577 work
->full_waiters_pending
= true;
579 work
->waiters_pending
= true;
581 if (signal_pending(current
)) {
586 if (cpu
== RING_BUFFER_ALL_CPUS
&& !ring_buffer_empty(buffer
))
589 if (cpu
!= RING_BUFFER_ALL_CPUS
&&
590 !ring_buffer_empty_cpu(buffer
, cpu
)) {
597 raw_spin_lock_irqsave(&cpu_buffer
->reader_lock
, flags
);
598 pagebusy
= cpu_buffer
->reader_page
== cpu_buffer
->commit_page
;
599 raw_spin_unlock_irqrestore(&cpu_buffer
->reader_lock
, flags
);
609 finish_wait(&work
->full_waiters
, &wait
);
611 finish_wait(&work
->waiters
, &wait
);
617 * ring_buffer_poll_wait - poll on buffer input
618 * @buffer: buffer to wait on
619 * @cpu: the cpu buffer to wait on
620 * @filp: the file descriptor
621 * @poll_table: The poll descriptor
623 * If @cpu == RING_BUFFER_ALL_CPUS then the task will wake up as soon
624 * as data is added to any of the @buffer's cpu buffers. Otherwise
625 * it will wait for data to be added to a specific cpu buffer.
627 * Returns POLLIN | POLLRDNORM if data exists in the buffers,
630 int ring_buffer_poll_wait(struct ring_buffer
*buffer
, int cpu
,
631 struct file
*filp
, poll_table
*poll_table
)
633 struct ring_buffer_per_cpu
*cpu_buffer
;
634 struct rb_irq_work
*work
;
636 if (cpu
== RING_BUFFER_ALL_CPUS
)
637 work
= &buffer
->irq_work
;
639 if (!cpumask_test_cpu(cpu
, buffer
->cpumask
))
642 cpu_buffer
= buffer
->buffers
[cpu
];
643 work
= &cpu_buffer
->irq_work
;
646 poll_wait(filp
, &work
->waiters
, poll_table
);
647 work
->waiters_pending
= true;
649 * There's a tight race between setting the waiters_pending and
650 * checking if the ring buffer is empty. Once the waiters_pending bit
651 * is set, the next event will wake the task up, but we can get stuck
652 * if there's only a single event in.
654 * FIXME: Ideally, we need a memory barrier on the writer side as well,
655 * but adding a memory barrier to all events will cause too much of a
656 * performance hit in the fast path. We only need a memory barrier when
657 * the buffer goes from empty to having content. But as this race is
658 * extremely small, and it's not a problem if another event comes in, we
663 if ((cpu
== RING_BUFFER_ALL_CPUS
&& !ring_buffer_empty(buffer
)) ||
664 (cpu
!= RING_BUFFER_ALL_CPUS
&& !ring_buffer_empty_cpu(buffer
, cpu
)))
665 return POLLIN
| POLLRDNORM
;
669 /* buffer may be either ring_buffer or ring_buffer_per_cpu */
670 #define RB_WARN_ON(b, cond) \
672 int _____ret = unlikely(cond); \
674 if (__same_type(*(b), struct ring_buffer_per_cpu)) { \
675 struct ring_buffer_per_cpu *__b = \
677 atomic_inc(&__b->buffer->record_disabled); \
679 atomic_inc(&b->record_disabled); \
685 /* Up this if you want to test the TIME_EXTENTS and normalization */
686 #define DEBUG_SHIFT 0
688 static inline u64
rb_time_stamp(struct ring_buffer
*buffer
)
690 /* shift to debug/test normalization and TIME_EXTENTS */
691 return buffer
->clock() << DEBUG_SHIFT
;
694 u64
ring_buffer_time_stamp(struct ring_buffer
*buffer
, int cpu
)
698 preempt_disable_notrace();
699 time
= rb_time_stamp(buffer
);
700 preempt_enable_no_resched_notrace();
704 EXPORT_SYMBOL_GPL(ring_buffer_time_stamp
);
706 void ring_buffer_normalize_time_stamp(struct ring_buffer
*buffer
,
709 /* Just stupid testing the normalize function and deltas */
712 EXPORT_SYMBOL_GPL(ring_buffer_normalize_time_stamp
);
715 * Making the ring buffer lockless makes things tricky.
716 * Although writes only happen on the CPU that they are on,
717 * and they only need to worry about interrupts. Reads can
720 * The reader page is always off the ring buffer, but when the
721 * reader finishes with a page, it needs to swap its page with
722 * a new one from the buffer. The reader needs to take from
723 * the head (writes go to the tail). But if a writer is in overwrite
724 * mode and wraps, it must push the head page forward.
726 * Here lies the problem.
728 * The reader must be careful to replace only the head page, and
729 * not another one. As described at the top of the file in the
730 * ASCII art, the reader sets its old page to point to the next
731 * page after head. It then sets the page after head to point to
732 * the old reader page. But if the writer moves the head page
733 * during this operation, the reader could end up with the tail.
735 * We use cmpxchg to help prevent this race. We also do something
736 * special with the page before head. We set the LSB to 1.
738 * When the writer must push the page forward, it will clear the
739 * bit that points to the head page, move the head, and then set
740 * the bit that points to the new head page.
742 * We also don't want an interrupt coming in and moving the head
743 * page on another writer. Thus we use the second LSB to catch
746 * head->list->prev->next bit 1 bit 0
749 * Points to head page 0 1
752 * Note we can not trust the prev pointer of the head page, because:
754 * +----+ +-----+ +-----+
755 * | |------>| T |---X--->| N |
757 * +----+ +-----+ +-----+
760 * +----------| R |----------+ |
764 * Key: ---X--> HEAD flag set in pointer
769 * (see __rb_reserve_next() to see where this happens)
771 * What the above shows is that the reader just swapped out
772 * the reader page with a page in the buffer, but before it
773 * could make the new header point back to the new page added
774 * it was preempted by a writer. The writer moved forward onto
775 * the new page added by the reader and is about to move forward
778 * You can see, it is legitimate for the previous pointer of
779 * the head (or any page) not to point back to itself. But only
783 #define RB_PAGE_NORMAL 0UL
784 #define RB_PAGE_HEAD 1UL
785 #define RB_PAGE_UPDATE 2UL
788 #define RB_FLAG_MASK 3UL
790 /* PAGE_MOVED is not part of the mask */
791 #define RB_PAGE_MOVED 4UL
794 * rb_list_head - remove any bit
796 static struct list_head
*rb_list_head(struct list_head
*list
)
798 unsigned long val
= (unsigned long)list
;
800 return (struct list_head
*)(val
& ~RB_FLAG_MASK
);
804 * rb_is_head_page - test if the given page is the head page
806 * Because the reader may move the head_page pointer, we can
807 * not trust what the head page is (it may be pointing to
808 * the reader page). But if the next page is a header page,
809 * its flags will be non zero.
812 rb_is_head_page(struct ring_buffer_per_cpu
*cpu_buffer
,
813 struct buffer_page
*page
, struct list_head
*list
)
817 val
= (unsigned long)list
->next
;
819 if ((val
& ~RB_FLAG_MASK
) != (unsigned long)&page
->list
)
820 return RB_PAGE_MOVED
;
822 return val
& RB_FLAG_MASK
;
828 * The unique thing about the reader page, is that, if the
829 * writer is ever on it, the previous pointer never points
830 * back to the reader page.
832 static bool rb_is_reader_page(struct buffer_page
*page
)
834 struct list_head
*list
= page
->list
.prev
;
836 return rb_list_head(list
->next
) != &page
->list
;
840 * rb_set_list_to_head - set a list_head to be pointing to head.
842 static void rb_set_list_to_head(struct ring_buffer_per_cpu
*cpu_buffer
,
843 struct list_head
*list
)
847 ptr
= (unsigned long *)&list
->next
;
848 *ptr
|= RB_PAGE_HEAD
;
849 *ptr
&= ~RB_PAGE_UPDATE
;
853 * rb_head_page_activate - sets up head page
855 static void rb_head_page_activate(struct ring_buffer_per_cpu
*cpu_buffer
)
857 struct buffer_page
*head
;
859 head
= cpu_buffer
->head_page
;
864 * Set the previous list pointer to have the HEAD flag.
866 rb_set_list_to_head(cpu_buffer
, head
->list
.prev
);
869 static void rb_list_head_clear(struct list_head
*list
)
871 unsigned long *ptr
= (unsigned long *)&list
->next
;
873 *ptr
&= ~RB_FLAG_MASK
;
877 * rb_head_page_dactivate - clears head page ptr (for free list)
880 rb_head_page_deactivate(struct ring_buffer_per_cpu
*cpu_buffer
)
882 struct list_head
*hd
;
884 /* Go through the whole list and clear any pointers found. */
885 rb_list_head_clear(cpu_buffer
->pages
);
887 list_for_each(hd
, cpu_buffer
->pages
)
888 rb_list_head_clear(hd
);
891 static int rb_head_page_set(struct ring_buffer_per_cpu
*cpu_buffer
,
892 struct buffer_page
*head
,
893 struct buffer_page
*prev
,
894 int old_flag
, int new_flag
)
896 struct list_head
*list
;
897 unsigned long val
= (unsigned long)&head
->list
;
902 val
&= ~RB_FLAG_MASK
;
904 ret
= cmpxchg((unsigned long *)&list
->next
,
905 val
| old_flag
, val
| new_flag
);
907 /* check if the reader took the page */
908 if ((ret
& ~RB_FLAG_MASK
) != val
)
909 return RB_PAGE_MOVED
;
911 return ret
& RB_FLAG_MASK
;
914 static int rb_head_page_set_update(struct ring_buffer_per_cpu
*cpu_buffer
,
915 struct buffer_page
*head
,
916 struct buffer_page
*prev
,
919 return rb_head_page_set(cpu_buffer
, head
, prev
,
920 old_flag
, RB_PAGE_UPDATE
);
923 static int rb_head_page_set_head(struct ring_buffer_per_cpu
*cpu_buffer
,
924 struct buffer_page
*head
,
925 struct buffer_page
*prev
,
928 return rb_head_page_set(cpu_buffer
, head
, prev
,
929 old_flag
, RB_PAGE_HEAD
);
932 static int rb_head_page_set_normal(struct ring_buffer_per_cpu
*cpu_buffer
,
933 struct buffer_page
*head
,
934 struct buffer_page
*prev
,
937 return rb_head_page_set(cpu_buffer
, head
, prev
,
938 old_flag
, RB_PAGE_NORMAL
);
941 static inline void rb_inc_page(struct ring_buffer_per_cpu
*cpu_buffer
,
942 struct buffer_page
**bpage
)
944 struct list_head
*p
= rb_list_head((*bpage
)->list
.next
);
946 *bpage
= list_entry(p
, struct buffer_page
, list
);
949 static struct buffer_page
*
950 rb_set_head_page(struct ring_buffer_per_cpu
*cpu_buffer
)
952 struct buffer_page
*head
;
953 struct buffer_page
*page
;
954 struct list_head
*list
;
957 if (RB_WARN_ON(cpu_buffer
, !cpu_buffer
->head_page
))
961 list
= cpu_buffer
->pages
;
962 if (RB_WARN_ON(cpu_buffer
, rb_list_head(list
->prev
->next
) != list
))
965 page
= head
= cpu_buffer
->head_page
;
967 * It is possible that the writer moves the header behind
968 * where we started, and we miss in one loop.
969 * A second loop should grab the header, but we'll do
970 * three loops just because I'm paranoid.
972 for (i
= 0; i
< 3; i
++) {
974 if (rb_is_head_page(cpu_buffer
, page
, page
->list
.prev
)) {
975 cpu_buffer
->head_page
= page
;
978 rb_inc_page(cpu_buffer
, &page
);
979 } while (page
!= head
);
982 RB_WARN_ON(cpu_buffer
, 1);
987 static int rb_head_page_replace(struct buffer_page
*old
,
988 struct buffer_page
*new)
990 unsigned long *ptr
= (unsigned long *)&old
->list
.prev
->next
;
994 val
= *ptr
& ~RB_FLAG_MASK
;
997 ret
= cmpxchg(ptr
, val
, (unsigned long)&new->list
);
1003 * rb_tail_page_update - move the tail page forward
1005 static void rb_tail_page_update(struct ring_buffer_per_cpu
*cpu_buffer
,
1006 struct buffer_page
*tail_page
,
1007 struct buffer_page
*next_page
)
1009 unsigned long old_entries
;
1010 unsigned long old_write
;
1013 * The tail page now needs to be moved forward.
1015 * We need to reset the tail page, but without messing
1016 * with possible erasing of data brought in by interrupts
1017 * that have moved the tail page and are currently on it.
1019 * We add a counter to the write field to denote this.
1021 old_write
= local_add_return(RB_WRITE_INTCNT
, &next_page
->write
);
1022 old_entries
= local_add_return(RB_WRITE_INTCNT
, &next_page
->entries
);
1025 * Just make sure we have seen our old_write and synchronize
1026 * with any interrupts that come in.
1031 * If the tail page is still the same as what we think
1032 * it is, then it is up to us to update the tail
1035 if (tail_page
== READ_ONCE(cpu_buffer
->tail_page
)) {
1036 /* Zero the write counter */
1037 unsigned long val
= old_write
& ~RB_WRITE_MASK
;
1038 unsigned long eval
= old_entries
& ~RB_WRITE_MASK
;
1041 * This will only succeed if an interrupt did
1042 * not come in and change it. In which case, we
1043 * do not want to modify it.
1045 * We add (void) to let the compiler know that we do not care
1046 * about the return value of these functions. We use the
1047 * cmpxchg to only update if an interrupt did not already
1048 * do it for us. If the cmpxchg fails, we don't care.
1050 (void)local_cmpxchg(&next_page
->write
, old_write
, val
);
1051 (void)local_cmpxchg(&next_page
->entries
, old_entries
, eval
);
1054 * No need to worry about races with clearing out the commit.
1055 * it only can increment when a commit takes place. But that
1056 * only happens in the outer most nested commit.
1058 local_set(&next_page
->page
->commit
, 0);
1060 /* Again, either we update tail_page or an interrupt does */
1061 (void)cmpxchg(&cpu_buffer
->tail_page
, tail_page
, next_page
);
1065 static int rb_check_bpage(struct ring_buffer_per_cpu
*cpu_buffer
,
1066 struct buffer_page
*bpage
)
1068 unsigned long val
= (unsigned long)bpage
;
1070 if (RB_WARN_ON(cpu_buffer
, val
& RB_FLAG_MASK
))
1077 * rb_check_list - make sure a pointer to a list has the last bits zero
1079 static int rb_check_list(struct ring_buffer_per_cpu
*cpu_buffer
,
1080 struct list_head
*list
)
1082 if (RB_WARN_ON(cpu_buffer
, rb_list_head(list
->prev
) != list
->prev
))
1084 if (RB_WARN_ON(cpu_buffer
, rb_list_head(list
->next
) != list
->next
))
1090 * rb_check_pages - integrity check of buffer pages
1091 * @cpu_buffer: CPU buffer with pages to test
1093 * As a safety measure we check to make sure the data pages have not
1096 static int rb_check_pages(struct ring_buffer_per_cpu
*cpu_buffer
)
1098 struct list_head
*head
= cpu_buffer
->pages
;
1099 struct buffer_page
*bpage
, *tmp
;
1101 /* Reset the head page if it exists */
1102 if (cpu_buffer
->head_page
)
1103 rb_set_head_page(cpu_buffer
);
1105 rb_head_page_deactivate(cpu_buffer
);
1107 if (RB_WARN_ON(cpu_buffer
, head
->next
->prev
!= head
))
1109 if (RB_WARN_ON(cpu_buffer
, head
->prev
->next
!= head
))
1112 if (rb_check_list(cpu_buffer
, head
))
1115 list_for_each_entry_safe(bpage
, tmp
, head
, list
) {
1116 if (RB_WARN_ON(cpu_buffer
,
1117 bpage
->list
.next
->prev
!= &bpage
->list
))
1119 if (RB_WARN_ON(cpu_buffer
,
1120 bpage
->list
.prev
->next
!= &bpage
->list
))
1122 if (rb_check_list(cpu_buffer
, &bpage
->list
))
1126 rb_head_page_activate(cpu_buffer
);
1131 static int __rb_allocate_pages(long nr_pages
, struct list_head
*pages
, int cpu
)
1133 struct buffer_page
*bpage
, *tmp
;
1136 for (i
= 0; i
< nr_pages
; i
++) {
1139 * __GFP_NORETRY flag makes sure that the allocation fails
1140 * gracefully without invoking oom-killer and the system is
1143 bpage
= kzalloc_node(ALIGN(sizeof(*bpage
), cache_line_size()),
1144 GFP_KERNEL
| __GFP_NORETRY
,
1149 list_add(&bpage
->list
, pages
);
1151 page
= alloc_pages_node(cpu_to_node(cpu
),
1152 GFP_KERNEL
| __GFP_NORETRY
, 0);
1155 bpage
->page
= page_address(page
);
1156 rb_init_page(bpage
->page
);
1162 list_for_each_entry_safe(bpage
, tmp
, pages
, list
) {
1163 list_del_init(&bpage
->list
);
1164 free_buffer_page(bpage
);
1170 static int rb_allocate_pages(struct ring_buffer_per_cpu
*cpu_buffer
,
1171 unsigned long nr_pages
)
1177 if (__rb_allocate_pages(nr_pages
, &pages
, cpu_buffer
->cpu
))
1181 * The ring buffer page list is a circular list that does not
1182 * start and end with a list head. All page list items point to
1185 cpu_buffer
->pages
= pages
.next
;
1188 cpu_buffer
->nr_pages
= nr_pages
;
1190 rb_check_pages(cpu_buffer
);
1195 static struct ring_buffer_per_cpu
*
1196 rb_allocate_cpu_buffer(struct ring_buffer
*buffer
, long nr_pages
, int cpu
)
1198 struct ring_buffer_per_cpu
*cpu_buffer
;
1199 struct buffer_page
*bpage
;
1203 cpu_buffer
= kzalloc_node(ALIGN(sizeof(*cpu_buffer
), cache_line_size()),
1204 GFP_KERNEL
, cpu_to_node(cpu
));
1208 cpu_buffer
->cpu
= cpu
;
1209 cpu_buffer
->buffer
= buffer
;
1210 raw_spin_lock_init(&cpu_buffer
->reader_lock
);
1211 lockdep_set_class(&cpu_buffer
->reader_lock
, buffer
->reader_lock_key
);
1212 cpu_buffer
->lock
= (arch_spinlock_t
)__ARCH_SPIN_LOCK_UNLOCKED
;
1213 INIT_WORK(&cpu_buffer
->update_pages_work
, update_pages_handler
);
1214 init_completion(&cpu_buffer
->update_done
);
1215 init_irq_work(&cpu_buffer
->irq_work
.work
, rb_wake_up_waiters
);
1216 init_waitqueue_head(&cpu_buffer
->irq_work
.waiters
);
1217 init_waitqueue_head(&cpu_buffer
->irq_work
.full_waiters
);
1219 bpage
= kzalloc_node(ALIGN(sizeof(*bpage
), cache_line_size()),
1220 GFP_KERNEL
, cpu_to_node(cpu
));
1222 goto fail_free_buffer
;
1224 rb_check_bpage(cpu_buffer
, bpage
);
1226 cpu_buffer
->reader_page
= bpage
;
1227 page
= alloc_pages_node(cpu_to_node(cpu
), GFP_KERNEL
, 0);
1229 goto fail_free_reader
;
1230 bpage
->page
= page_address(page
);
1231 rb_init_page(bpage
->page
);
1233 INIT_LIST_HEAD(&cpu_buffer
->reader_page
->list
);
1234 INIT_LIST_HEAD(&cpu_buffer
->new_pages
);
1236 ret
= rb_allocate_pages(cpu_buffer
, nr_pages
);
1238 goto fail_free_reader
;
1240 cpu_buffer
->head_page
1241 = list_entry(cpu_buffer
->pages
, struct buffer_page
, list
);
1242 cpu_buffer
->tail_page
= cpu_buffer
->commit_page
= cpu_buffer
->head_page
;
1244 rb_head_page_activate(cpu_buffer
);
1249 free_buffer_page(cpu_buffer
->reader_page
);
1256 static void rb_free_cpu_buffer(struct ring_buffer_per_cpu
*cpu_buffer
)
1258 struct list_head
*head
= cpu_buffer
->pages
;
1259 struct buffer_page
*bpage
, *tmp
;
1261 free_buffer_page(cpu_buffer
->reader_page
);
1263 rb_head_page_deactivate(cpu_buffer
);
1266 list_for_each_entry_safe(bpage
, tmp
, head
, list
) {
1267 list_del_init(&bpage
->list
);
1268 free_buffer_page(bpage
);
1270 bpage
= list_entry(head
, struct buffer_page
, list
);
1271 free_buffer_page(bpage
);
1278 * __ring_buffer_alloc - allocate a new ring_buffer
1279 * @size: the size in bytes per cpu that is needed.
1280 * @flags: attributes to set for the ring buffer.
1282 * Currently the only flag that is available is the RB_FL_OVERWRITE
1283 * flag. This flag means that the buffer will overwrite old data
1284 * when the buffer wraps. If this flag is not set, the buffer will
1285 * drop data when the tail hits the head.
1287 struct ring_buffer
*__ring_buffer_alloc(unsigned long size
, unsigned flags
,
1288 struct lock_class_key
*key
)
1290 struct ring_buffer
*buffer
;
1296 /* keep it in its own cache line */
1297 buffer
= kzalloc(ALIGN(sizeof(*buffer
), cache_line_size()),
1302 if (!zalloc_cpumask_var(&buffer
->cpumask
, GFP_KERNEL
))
1303 goto fail_free_buffer
;
1305 nr_pages
= DIV_ROUND_UP(size
, BUF_PAGE_SIZE
);
1306 buffer
->flags
= flags
;
1307 buffer
->clock
= trace_clock_local
;
1308 buffer
->reader_lock_key
= key
;
1310 init_irq_work(&buffer
->irq_work
.work
, rb_wake_up_waiters
);
1311 init_waitqueue_head(&buffer
->irq_work
.waiters
);
1313 /* need at least two pages */
1317 buffer
->cpus
= nr_cpu_ids
;
1319 bsize
= sizeof(void *) * nr_cpu_ids
;
1320 buffer
->buffers
= kzalloc(ALIGN(bsize
, cache_line_size()),
1322 if (!buffer
->buffers
)
1323 goto fail_free_cpumask
;
1325 cpu
= raw_smp_processor_id();
1326 cpumask_set_cpu(cpu
, buffer
->cpumask
);
1327 buffer
->buffers
[cpu
] = rb_allocate_cpu_buffer(buffer
, nr_pages
, cpu
);
1328 if (!buffer
->buffers
[cpu
])
1329 goto fail_free_buffers
;
1331 ret
= cpuhp_state_add_instance(CPUHP_TRACE_RB_PREPARE
, &buffer
->node
);
1333 goto fail_free_buffers
;
1335 mutex_init(&buffer
->mutex
);
1340 for_each_buffer_cpu(buffer
, cpu
) {
1341 if (buffer
->buffers
[cpu
])
1342 rb_free_cpu_buffer(buffer
->buffers
[cpu
]);
1344 kfree(buffer
->buffers
);
1347 free_cpumask_var(buffer
->cpumask
);
1353 EXPORT_SYMBOL_GPL(__ring_buffer_alloc
);
1356 * ring_buffer_free - free a ring buffer.
1357 * @buffer: the buffer to free.
1360 ring_buffer_free(struct ring_buffer
*buffer
)
1364 cpuhp_state_remove_instance(CPUHP_TRACE_RB_PREPARE
, &buffer
->node
);
1366 for_each_buffer_cpu(buffer
, cpu
)
1367 rb_free_cpu_buffer(buffer
->buffers
[cpu
]);
1369 kfree(buffer
->buffers
);
1370 free_cpumask_var(buffer
->cpumask
);
1374 EXPORT_SYMBOL_GPL(ring_buffer_free
);
1376 void ring_buffer_set_clock(struct ring_buffer
*buffer
,
1379 buffer
->clock
= clock
;
1382 static void rb_reset_cpu(struct ring_buffer_per_cpu
*cpu_buffer
);
1384 static inline unsigned long rb_page_entries(struct buffer_page
*bpage
)
1386 return local_read(&bpage
->entries
) & RB_WRITE_MASK
;
1389 static inline unsigned long rb_page_write(struct buffer_page
*bpage
)
1391 return local_read(&bpage
->write
) & RB_WRITE_MASK
;
1395 rb_remove_pages(struct ring_buffer_per_cpu
*cpu_buffer
, unsigned long nr_pages
)
1397 struct list_head
*tail_page
, *to_remove
, *next_page
;
1398 struct buffer_page
*to_remove_page
, *tmp_iter_page
;
1399 struct buffer_page
*last_page
, *first_page
;
1400 unsigned long nr_removed
;
1401 unsigned long head_bit
;
1406 raw_spin_lock_irq(&cpu_buffer
->reader_lock
);
1407 atomic_inc(&cpu_buffer
->record_disabled
);
1409 * We don't race with the readers since we have acquired the reader
1410 * lock. We also don't race with writers after disabling recording.
1411 * This makes it easy to figure out the first and the last page to be
1412 * removed from the list. We unlink all the pages in between including
1413 * the first and last pages. This is done in a busy loop so that we
1414 * lose the least number of traces.
1415 * The pages are freed after we restart recording and unlock readers.
1417 tail_page
= &cpu_buffer
->tail_page
->list
;
1420 * tail page might be on reader page, we remove the next page
1421 * from the ring buffer
1423 if (cpu_buffer
->tail_page
== cpu_buffer
->reader_page
)
1424 tail_page
= rb_list_head(tail_page
->next
);
1425 to_remove
= tail_page
;
1427 /* start of pages to remove */
1428 first_page
= list_entry(rb_list_head(to_remove
->next
),
1429 struct buffer_page
, list
);
1431 for (nr_removed
= 0; nr_removed
< nr_pages
; nr_removed
++) {
1432 to_remove
= rb_list_head(to_remove
)->next
;
1433 head_bit
|= (unsigned long)to_remove
& RB_PAGE_HEAD
;
1436 next_page
= rb_list_head(to_remove
)->next
;
1439 * Now we remove all pages between tail_page and next_page.
1440 * Make sure that we have head_bit value preserved for the
1443 tail_page
->next
= (struct list_head
*)((unsigned long)next_page
|
1445 next_page
= rb_list_head(next_page
);
1446 next_page
->prev
= tail_page
;
1448 /* make sure pages points to a valid page in the ring buffer */
1449 cpu_buffer
->pages
= next_page
;
1451 /* update head page */
1453 cpu_buffer
->head_page
= list_entry(next_page
,
1454 struct buffer_page
, list
);
1457 * change read pointer to make sure any read iterators reset
1460 cpu_buffer
->read
= 0;
1462 /* pages are removed, resume tracing and then free the pages */
1463 atomic_dec(&cpu_buffer
->record_disabled
);
1464 raw_spin_unlock_irq(&cpu_buffer
->reader_lock
);
1466 RB_WARN_ON(cpu_buffer
, list_empty(cpu_buffer
->pages
));
1468 /* last buffer page to remove */
1469 last_page
= list_entry(rb_list_head(to_remove
), struct buffer_page
,
1471 tmp_iter_page
= first_page
;
1474 to_remove_page
= tmp_iter_page
;
1475 rb_inc_page(cpu_buffer
, &tmp_iter_page
);
1477 /* update the counters */
1478 page_entries
= rb_page_entries(to_remove_page
);
1481 * If something was added to this page, it was full
1482 * since it is not the tail page. So we deduct the
1483 * bytes consumed in ring buffer from here.
1484 * Increment overrun to account for the lost events.
1486 local_add(page_entries
, &cpu_buffer
->overrun
);
1487 local_sub(BUF_PAGE_SIZE
, &cpu_buffer
->entries_bytes
);
1491 * We have already removed references to this list item, just
1492 * free up the buffer_page and its page
1494 free_buffer_page(to_remove_page
);
1497 } while (to_remove_page
!= last_page
);
1499 RB_WARN_ON(cpu_buffer
, nr_removed
);
1501 return nr_removed
== 0;
1505 rb_insert_pages(struct ring_buffer_per_cpu
*cpu_buffer
)
1507 struct list_head
*pages
= &cpu_buffer
->new_pages
;
1508 int retries
, success
;
1510 raw_spin_lock_irq(&cpu_buffer
->reader_lock
);
1512 * We are holding the reader lock, so the reader page won't be swapped
1513 * in the ring buffer. Now we are racing with the writer trying to
1514 * move head page and the tail page.
1515 * We are going to adapt the reader page update process where:
1516 * 1. We first splice the start and end of list of new pages between
1517 * the head page and its previous page.
1518 * 2. We cmpxchg the prev_page->next to point from head page to the
1519 * start of new pages list.
1520 * 3. Finally, we update the head->prev to the end of new list.
1522 * We will try this process 10 times, to make sure that we don't keep
1528 struct list_head
*head_page
, *prev_page
, *r
;
1529 struct list_head
*last_page
, *first_page
;
1530 struct list_head
*head_page_with_bit
;
1532 head_page
= &rb_set_head_page(cpu_buffer
)->list
;
1535 prev_page
= head_page
->prev
;
1537 first_page
= pages
->next
;
1538 last_page
= pages
->prev
;
1540 head_page_with_bit
= (struct list_head
*)
1541 ((unsigned long)head_page
| RB_PAGE_HEAD
);
1543 last_page
->next
= head_page_with_bit
;
1544 first_page
->prev
= prev_page
;
1546 r
= cmpxchg(&prev_page
->next
, head_page_with_bit
, first_page
);
1548 if (r
== head_page_with_bit
) {
1550 * yay, we replaced the page pointer to our new list,
1551 * now, we just have to update to head page's prev
1552 * pointer to point to end of list
1554 head_page
->prev
= last_page
;
1561 INIT_LIST_HEAD(pages
);
1563 * If we weren't successful in adding in new pages, warn and stop
1566 RB_WARN_ON(cpu_buffer
, !success
);
1567 raw_spin_unlock_irq(&cpu_buffer
->reader_lock
);
1569 /* free pages if they weren't inserted */
1571 struct buffer_page
*bpage
, *tmp
;
1572 list_for_each_entry_safe(bpage
, tmp
, &cpu_buffer
->new_pages
,
1574 list_del_init(&bpage
->list
);
1575 free_buffer_page(bpage
);
1581 static void rb_update_pages(struct ring_buffer_per_cpu
*cpu_buffer
)
1585 if (cpu_buffer
->nr_pages_to_update
> 0)
1586 success
= rb_insert_pages(cpu_buffer
);
1588 success
= rb_remove_pages(cpu_buffer
,
1589 -cpu_buffer
->nr_pages_to_update
);
1592 cpu_buffer
->nr_pages
+= cpu_buffer
->nr_pages_to_update
;
1595 static void update_pages_handler(struct work_struct
*work
)
1597 struct ring_buffer_per_cpu
*cpu_buffer
= container_of(work
,
1598 struct ring_buffer_per_cpu
, update_pages_work
);
1599 rb_update_pages(cpu_buffer
);
1600 complete(&cpu_buffer
->update_done
);
1604 * ring_buffer_resize - resize the ring buffer
1605 * @buffer: the buffer to resize.
1606 * @size: the new size.
1607 * @cpu_id: the cpu buffer to resize
1609 * Minimum size is 2 * BUF_PAGE_SIZE.
1611 * Returns 0 on success and < 0 on failure.
1613 int ring_buffer_resize(struct ring_buffer
*buffer
, unsigned long size
,
1616 struct ring_buffer_per_cpu
*cpu_buffer
;
1617 unsigned long nr_pages
;
1621 * Always succeed at resizing a non-existent buffer:
1626 /* Make sure the requested buffer exists */
1627 if (cpu_id
!= RING_BUFFER_ALL_CPUS
&&
1628 !cpumask_test_cpu(cpu_id
, buffer
->cpumask
))
1631 nr_pages
= DIV_ROUND_UP(size
, BUF_PAGE_SIZE
);
1633 /* we need a minimum of two pages */
1637 size
= nr_pages
* BUF_PAGE_SIZE
;
1640 * Don't succeed if resizing is disabled, as a reader might be
1641 * manipulating the ring buffer and is expecting a sane state while
1644 if (atomic_read(&buffer
->resize_disabled
))
1647 /* prevent another thread from changing buffer sizes */
1648 mutex_lock(&buffer
->mutex
);
1650 if (cpu_id
== RING_BUFFER_ALL_CPUS
) {
1651 /* calculate the pages to update */
1652 for_each_buffer_cpu(buffer
, cpu
) {
1653 cpu_buffer
= buffer
->buffers
[cpu
];
1655 cpu_buffer
->nr_pages_to_update
= nr_pages
-
1656 cpu_buffer
->nr_pages
;
1658 * nothing more to do for removing pages or no update
1660 if (cpu_buffer
->nr_pages_to_update
<= 0)
1663 * to add pages, make sure all new pages can be
1664 * allocated without receiving ENOMEM
1666 INIT_LIST_HEAD(&cpu_buffer
->new_pages
);
1667 if (__rb_allocate_pages(cpu_buffer
->nr_pages_to_update
,
1668 &cpu_buffer
->new_pages
, cpu
)) {
1669 /* not enough memory for new pages */
1677 * Fire off all the required work handlers
1678 * We can't schedule on offline CPUs, but it's not necessary
1679 * since we can change their buffer sizes without any race.
1681 for_each_buffer_cpu(buffer
, cpu
) {
1682 cpu_buffer
= buffer
->buffers
[cpu
];
1683 if (!cpu_buffer
->nr_pages_to_update
)
1686 /* Can't run something on an offline CPU. */
1687 if (!cpu_online(cpu
)) {
1688 rb_update_pages(cpu_buffer
);
1689 cpu_buffer
->nr_pages_to_update
= 0;
1691 schedule_work_on(cpu
,
1692 &cpu_buffer
->update_pages_work
);
1696 /* wait for all the updates to complete */
1697 for_each_buffer_cpu(buffer
, cpu
) {
1698 cpu_buffer
= buffer
->buffers
[cpu
];
1699 if (!cpu_buffer
->nr_pages_to_update
)
1702 if (cpu_online(cpu
))
1703 wait_for_completion(&cpu_buffer
->update_done
);
1704 cpu_buffer
->nr_pages_to_update
= 0;
1709 /* Make sure this CPU has been intitialized */
1710 if (!cpumask_test_cpu(cpu_id
, buffer
->cpumask
))
1713 cpu_buffer
= buffer
->buffers
[cpu_id
];
1715 if (nr_pages
== cpu_buffer
->nr_pages
)
1718 cpu_buffer
->nr_pages_to_update
= nr_pages
-
1719 cpu_buffer
->nr_pages
;
1721 INIT_LIST_HEAD(&cpu_buffer
->new_pages
);
1722 if (cpu_buffer
->nr_pages_to_update
> 0 &&
1723 __rb_allocate_pages(cpu_buffer
->nr_pages_to_update
,
1724 &cpu_buffer
->new_pages
, cpu_id
)) {
1731 /* Can't run something on an offline CPU. */
1732 if (!cpu_online(cpu_id
))
1733 rb_update_pages(cpu_buffer
);
1735 schedule_work_on(cpu_id
,
1736 &cpu_buffer
->update_pages_work
);
1737 wait_for_completion(&cpu_buffer
->update_done
);
1740 cpu_buffer
->nr_pages_to_update
= 0;
1746 * The ring buffer resize can happen with the ring buffer
1747 * enabled, so that the update disturbs the tracing as little
1748 * as possible. But if the buffer is disabled, we do not need
1749 * to worry about that, and we can take the time to verify
1750 * that the buffer is not corrupt.
1752 if (atomic_read(&buffer
->record_disabled
)) {
1753 atomic_inc(&buffer
->record_disabled
);
1755 * Even though the buffer was disabled, we must make sure
1756 * that it is truly disabled before calling rb_check_pages.
1757 * There could have been a race between checking
1758 * record_disable and incrementing it.
1760 synchronize_sched();
1761 for_each_buffer_cpu(buffer
, cpu
) {
1762 cpu_buffer
= buffer
->buffers
[cpu
];
1763 rb_check_pages(cpu_buffer
);
1765 atomic_dec(&buffer
->record_disabled
);
1768 mutex_unlock(&buffer
->mutex
);
1772 for_each_buffer_cpu(buffer
, cpu
) {
1773 struct buffer_page
*bpage
, *tmp
;
1775 cpu_buffer
= buffer
->buffers
[cpu
];
1776 cpu_buffer
->nr_pages_to_update
= 0;
1778 if (list_empty(&cpu_buffer
->new_pages
))
1781 list_for_each_entry_safe(bpage
, tmp
, &cpu_buffer
->new_pages
,
1783 list_del_init(&bpage
->list
);
1784 free_buffer_page(bpage
);
1787 mutex_unlock(&buffer
->mutex
);
1790 EXPORT_SYMBOL_GPL(ring_buffer_resize
);
1792 void ring_buffer_change_overwrite(struct ring_buffer
*buffer
, int val
)
1794 mutex_lock(&buffer
->mutex
);
1796 buffer
->flags
|= RB_FL_OVERWRITE
;
1798 buffer
->flags
&= ~RB_FL_OVERWRITE
;
1799 mutex_unlock(&buffer
->mutex
);
1801 EXPORT_SYMBOL_GPL(ring_buffer_change_overwrite
);
1803 static __always_inline
void *
1804 __rb_data_page_index(struct buffer_data_page
*bpage
, unsigned index
)
1806 return bpage
->data
+ index
;
1809 static __always_inline
void *__rb_page_index(struct buffer_page
*bpage
, unsigned index
)
1811 return bpage
->page
->data
+ index
;
1814 static __always_inline
struct ring_buffer_event
*
1815 rb_reader_event(struct ring_buffer_per_cpu
*cpu_buffer
)
1817 return __rb_page_index(cpu_buffer
->reader_page
,
1818 cpu_buffer
->reader_page
->read
);
1821 static __always_inline
struct ring_buffer_event
*
1822 rb_iter_head_event(struct ring_buffer_iter
*iter
)
1824 return __rb_page_index(iter
->head_page
, iter
->head
);
1827 static __always_inline
unsigned rb_page_commit(struct buffer_page
*bpage
)
1829 return local_read(&bpage
->page
->commit
);
1832 /* Size is determined by what has been committed */
1833 static __always_inline
unsigned rb_page_size(struct buffer_page
*bpage
)
1835 return rb_page_commit(bpage
);
1838 static __always_inline
unsigned
1839 rb_commit_index(struct ring_buffer_per_cpu
*cpu_buffer
)
1841 return rb_page_commit(cpu_buffer
->commit_page
);
1844 static __always_inline
unsigned
1845 rb_event_index(struct ring_buffer_event
*event
)
1847 unsigned long addr
= (unsigned long)event
;
1849 return (addr
& ~PAGE_MASK
) - BUF_PAGE_HDR_SIZE
;
1852 static void rb_inc_iter(struct ring_buffer_iter
*iter
)
1854 struct ring_buffer_per_cpu
*cpu_buffer
= iter
->cpu_buffer
;
1857 * The iterator could be on the reader page (it starts there).
1858 * But the head could have moved, since the reader was
1859 * found. Check for this case and assign the iterator
1860 * to the head page instead of next.
1862 if (iter
->head_page
== cpu_buffer
->reader_page
)
1863 iter
->head_page
= rb_set_head_page(cpu_buffer
);
1865 rb_inc_page(cpu_buffer
, &iter
->head_page
);
1867 iter
->read_stamp
= iter
->head_page
->page
->time_stamp
;
1872 * rb_handle_head_page - writer hit the head page
1874 * Returns: +1 to retry page
1879 rb_handle_head_page(struct ring_buffer_per_cpu
*cpu_buffer
,
1880 struct buffer_page
*tail_page
,
1881 struct buffer_page
*next_page
)
1883 struct buffer_page
*new_head
;
1888 entries
= rb_page_entries(next_page
);
1891 * The hard part is here. We need to move the head
1892 * forward, and protect against both readers on
1893 * other CPUs and writers coming in via interrupts.
1895 type
= rb_head_page_set_update(cpu_buffer
, next_page
, tail_page
,
1899 * type can be one of four:
1900 * NORMAL - an interrupt already moved it for us
1901 * HEAD - we are the first to get here.
1902 * UPDATE - we are the interrupt interrupting
1904 * MOVED - a reader on another CPU moved the next
1905 * pointer to its reader page. Give up
1912 * We changed the head to UPDATE, thus
1913 * it is our responsibility to update
1916 local_add(entries
, &cpu_buffer
->overrun
);
1917 local_sub(BUF_PAGE_SIZE
, &cpu_buffer
->entries_bytes
);
1920 * The entries will be zeroed out when we move the
1924 /* still more to do */
1927 case RB_PAGE_UPDATE
:
1929 * This is an interrupt that interrupt the
1930 * previous update. Still more to do.
1933 case RB_PAGE_NORMAL
:
1935 * An interrupt came in before the update
1936 * and processed this for us.
1937 * Nothing left to do.
1942 * The reader is on another CPU and just did
1943 * a swap with our next_page.
1948 RB_WARN_ON(cpu_buffer
, 1); /* WTF??? */
1953 * Now that we are here, the old head pointer is
1954 * set to UPDATE. This will keep the reader from
1955 * swapping the head page with the reader page.
1956 * The reader (on another CPU) will spin till
1959 * We just need to protect against interrupts
1960 * doing the job. We will set the next pointer
1961 * to HEAD. After that, we set the old pointer
1962 * to NORMAL, but only if it was HEAD before.
1963 * otherwise we are an interrupt, and only
1964 * want the outer most commit to reset it.
1966 new_head
= next_page
;
1967 rb_inc_page(cpu_buffer
, &new_head
);
1969 ret
= rb_head_page_set_head(cpu_buffer
, new_head
, next_page
,
1973 * Valid returns are:
1974 * HEAD - an interrupt came in and already set it.
1975 * NORMAL - One of two things:
1976 * 1) We really set it.
1977 * 2) A bunch of interrupts came in and moved
1978 * the page forward again.
1982 case RB_PAGE_NORMAL
:
1986 RB_WARN_ON(cpu_buffer
, 1);
1991 * It is possible that an interrupt came in,
1992 * set the head up, then more interrupts came in
1993 * and moved it again. When we get back here,
1994 * the page would have been set to NORMAL but we
1995 * just set it back to HEAD.
1997 * How do you detect this? Well, if that happened
1998 * the tail page would have moved.
2000 if (ret
== RB_PAGE_NORMAL
) {
2001 struct buffer_page
*buffer_tail_page
;
2003 buffer_tail_page
= READ_ONCE(cpu_buffer
->tail_page
);
2005 * If the tail had moved passed next, then we need
2006 * to reset the pointer.
2008 if (buffer_tail_page
!= tail_page
&&
2009 buffer_tail_page
!= next_page
)
2010 rb_head_page_set_normal(cpu_buffer
, new_head
,
2016 * If this was the outer most commit (the one that
2017 * changed the original pointer from HEAD to UPDATE),
2018 * then it is up to us to reset it to NORMAL.
2020 if (type
== RB_PAGE_HEAD
) {
2021 ret
= rb_head_page_set_normal(cpu_buffer
, next_page
,
2024 if (RB_WARN_ON(cpu_buffer
,
2025 ret
!= RB_PAGE_UPDATE
))
2033 rb_reset_tail(struct ring_buffer_per_cpu
*cpu_buffer
,
2034 unsigned long tail
, struct rb_event_info
*info
)
2036 struct buffer_page
*tail_page
= info
->tail_page
;
2037 struct ring_buffer_event
*event
;
2038 unsigned long length
= info
->length
;
2041 * Only the event that crossed the page boundary
2042 * must fill the old tail_page with padding.
2044 if (tail
>= BUF_PAGE_SIZE
) {
2046 * If the page was filled, then we still need
2047 * to update the real_end. Reset it to zero
2048 * and the reader will ignore it.
2050 if (tail
== BUF_PAGE_SIZE
)
2051 tail_page
->real_end
= 0;
2053 local_sub(length
, &tail_page
->write
);
2057 event
= __rb_page_index(tail_page
, tail
);
2058 kmemcheck_annotate_bitfield(event
, bitfield
);
2060 /* account for padding bytes */
2061 local_add(BUF_PAGE_SIZE
- tail
, &cpu_buffer
->entries_bytes
);
2064 * Save the original length to the meta data.
2065 * This will be used by the reader to add lost event
2068 tail_page
->real_end
= tail
;
2071 * If this event is bigger than the minimum size, then
2072 * we need to be careful that we don't subtract the
2073 * write counter enough to allow another writer to slip
2075 * We put in a discarded commit instead, to make sure
2076 * that this space is not used again.
2078 * If we are less than the minimum size, we don't need to
2081 if (tail
> (BUF_PAGE_SIZE
- RB_EVNT_MIN_SIZE
)) {
2082 /* No room for any events */
2084 /* Mark the rest of the page with padding */
2085 rb_event_set_padding(event
);
2087 /* Set the write back to the previous setting */
2088 local_sub(length
, &tail_page
->write
);
2092 /* Put in a discarded event */
2093 event
->array
[0] = (BUF_PAGE_SIZE
- tail
) - RB_EVNT_HDR_SIZE
;
2094 event
->type_len
= RINGBUF_TYPE_PADDING
;
2095 /* time delta must be non zero */
2096 event
->time_delta
= 1;
2098 /* Set write to end of buffer */
2099 length
= (tail
+ length
) - BUF_PAGE_SIZE
;
2100 local_sub(length
, &tail_page
->write
);
2103 static inline void rb_end_commit(struct ring_buffer_per_cpu
*cpu_buffer
);
2106 * This is the slow path, force gcc not to inline it.
2108 static noinline
struct ring_buffer_event
*
2109 rb_move_tail(struct ring_buffer_per_cpu
*cpu_buffer
,
2110 unsigned long tail
, struct rb_event_info
*info
)
2112 struct buffer_page
*tail_page
= info
->tail_page
;
2113 struct buffer_page
*commit_page
= cpu_buffer
->commit_page
;
2114 struct ring_buffer
*buffer
= cpu_buffer
->buffer
;
2115 struct buffer_page
*next_page
;
2118 next_page
= tail_page
;
2120 rb_inc_page(cpu_buffer
, &next_page
);
2123 * If for some reason, we had an interrupt storm that made
2124 * it all the way around the buffer, bail, and warn
2127 if (unlikely(next_page
== commit_page
)) {
2128 local_inc(&cpu_buffer
->commit_overrun
);
2133 * This is where the fun begins!
2135 * We are fighting against races between a reader that
2136 * could be on another CPU trying to swap its reader
2137 * page with the buffer head.
2139 * We are also fighting against interrupts coming in and
2140 * moving the head or tail on us as well.
2142 * If the next page is the head page then we have filled
2143 * the buffer, unless the commit page is still on the
2146 if (rb_is_head_page(cpu_buffer
, next_page
, &tail_page
->list
)) {
2149 * If the commit is not on the reader page, then
2150 * move the header page.
2152 if (!rb_is_reader_page(cpu_buffer
->commit_page
)) {
2154 * If we are not in overwrite mode,
2155 * this is easy, just stop here.
2157 if (!(buffer
->flags
& RB_FL_OVERWRITE
)) {
2158 local_inc(&cpu_buffer
->dropped_events
);
2162 ret
= rb_handle_head_page(cpu_buffer
,
2171 * We need to be careful here too. The
2172 * commit page could still be on the reader
2173 * page. We could have a small buffer, and
2174 * have filled up the buffer with events
2175 * from interrupts and such, and wrapped.
2177 * Note, if the tail page is also the on the
2178 * reader_page, we let it move out.
2180 if (unlikely((cpu_buffer
->commit_page
!=
2181 cpu_buffer
->tail_page
) &&
2182 (cpu_buffer
->commit_page
==
2183 cpu_buffer
->reader_page
))) {
2184 local_inc(&cpu_buffer
->commit_overrun
);
2190 rb_tail_page_update(cpu_buffer
, tail_page
, next_page
);
2194 rb_reset_tail(cpu_buffer
, tail
, info
);
2196 /* Commit what we have for now. */
2197 rb_end_commit(cpu_buffer
);
2198 /* rb_end_commit() decs committing */
2199 local_inc(&cpu_buffer
->committing
);
2201 /* fail and let the caller try again */
2202 return ERR_PTR(-EAGAIN
);
2206 rb_reset_tail(cpu_buffer
, tail
, info
);
2211 /* Slow path, do not inline */
2212 static noinline
struct ring_buffer_event
*
2213 rb_add_time_stamp(struct ring_buffer_event
*event
, u64 delta
)
2215 event
->type_len
= RINGBUF_TYPE_TIME_EXTEND
;
2217 /* Not the first event on the page? */
2218 if (rb_event_index(event
)) {
2219 event
->time_delta
= delta
& TS_MASK
;
2220 event
->array
[0] = delta
>> TS_SHIFT
;
2222 /* nope, just zero it */
2223 event
->time_delta
= 0;
2224 event
->array
[0] = 0;
2227 return skip_time_extend(event
);
2230 static inline bool rb_event_is_commit(struct ring_buffer_per_cpu
*cpu_buffer
,
2231 struct ring_buffer_event
*event
);
2234 * rb_update_event - update event type and data
2235 * @event: the event to update
2236 * @type: the type of event
2237 * @length: the size of the event field in the ring buffer
2239 * Update the type and data fields of the event. The length
2240 * is the actual size that is written to the ring buffer,
2241 * and with this, we can determine what to place into the
2245 rb_update_event(struct ring_buffer_per_cpu
*cpu_buffer
,
2246 struct ring_buffer_event
*event
,
2247 struct rb_event_info
*info
)
2249 unsigned length
= info
->length
;
2250 u64 delta
= info
->delta
;
2252 /* Only a commit updates the timestamp */
2253 if (unlikely(!rb_event_is_commit(cpu_buffer
, event
)))
2257 * If we need to add a timestamp, then we
2258 * add it to the start of the resevered space.
2260 if (unlikely(info
->add_timestamp
)) {
2261 event
= rb_add_time_stamp(event
, delta
);
2262 length
-= RB_LEN_TIME_EXTEND
;
2266 event
->time_delta
= delta
;
2267 length
-= RB_EVNT_HDR_SIZE
;
2268 if (length
> RB_MAX_SMALL_DATA
|| RB_FORCE_8BYTE_ALIGNMENT
) {
2269 event
->type_len
= 0;
2270 event
->array
[0] = length
;
2272 event
->type_len
= DIV_ROUND_UP(length
, RB_ALIGNMENT
);
2275 static unsigned rb_calculate_event_length(unsigned length
)
2277 struct ring_buffer_event event
; /* Used only for sizeof array */
2279 /* zero length can cause confusions */
2283 if (length
> RB_MAX_SMALL_DATA
|| RB_FORCE_8BYTE_ALIGNMENT
)
2284 length
+= sizeof(event
.array
[0]);
2286 length
+= RB_EVNT_HDR_SIZE
;
2287 length
= ALIGN(length
, RB_ARCH_ALIGNMENT
);
2290 * In case the time delta is larger than the 27 bits for it
2291 * in the header, we need to add a timestamp. If another
2292 * event comes in when trying to discard this one to increase
2293 * the length, then the timestamp will be added in the allocated
2294 * space of this event. If length is bigger than the size needed
2295 * for the TIME_EXTEND, then padding has to be used. The events
2296 * length must be either RB_LEN_TIME_EXTEND, or greater than or equal
2297 * to RB_LEN_TIME_EXTEND + 8, as 8 is the minimum size for padding.
2298 * As length is a multiple of 4, we only need to worry if it
2299 * is 12 (RB_LEN_TIME_EXTEND + 4).
2301 if (length
== RB_LEN_TIME_EXTEND
+ RB_ALIGNMENT
)
2302 length
+= RB_ALIGNMENT
;
2307 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2308 static inline bool sched_clock_stable(void)
2315 rb_try_to_discard(struct ring_buffer_per_cpu
*cpu_buffer
,
2316 struct ring_buffer_event
*event
)
2318 unsigned long new_index
, old_index
;
2319 struct buffer_page
*bpage
;
2320 unsigned long index
;
2323 new_index
= rb_event_index(event
);
2324 old_index
= new_index
+ rb_event_ts_length(event
);
2325 addr
= (unsigned long)event
;
2328 bpage
= READ_ONCE(cpu_buffer
->tail_page
);
2330 if (bpage
->page
== (void *)addr
&& rb_page_write(bpage
) == old_index
) {
2331 unsigned long write_mask
=
2332 local_read(&bpage
->write
) & ~RB_WRITE_MASK
;
2333 unsigned long event_length
= rb_event_length(event
);
2335 * This is on the tail page. It is possible that
2336 * a write could come in and move the tail page
2337 * and write to the next page. That is fine
2338 * because we just shorten what is on this page.
2340 old_index
+= write_mask
;
2341 new_index
+= write_mask
;
2342 index
= local_cmpxchg(&bpage
->write
, old_index
, new_index
);
2343 if (index
== old_index
) {
2344 /* update counters */
2345 local_sub(event_length
, &cpu_buffer
->entries_bytes
);
2350 /* could not discard */
2354 static void rb_start_commit(struct ring_buffer_per_cpu
*cpu_buffer
)
2356 local_inc(&cpu_buffer
->committing
);
2357 local_inc(&cpu_buffer
->commits
);
2360 static __always_inline
void
2361 rb_set_commit_to_write(struct ring_buffer_per_cpu
*cpu_buffer
)
2363 unsigned long max_count
;
2366 * We only race with interrupts and NMIs on this CPU.
2367 * If we own the commit event, then we can commit
2368 * all others that interrupted us, since the interruptions
2369 * are in stack format (they finish before they come
2370 * back to us). This allows us to do a simple loop to
2371 * assign the commit to the tail.
2374 max_count
= cpu_buffer
->nr_pages
* 100;
2376 while (cpu_buffer
->commit_page
!= READ_ONCE(cpu_buffer
->tail_page
)) {
2377 if (RB_WARN_ON(cpu_buffer
, !(--max_count
)))
2379 if (RB_WARN_ON(cpu_buffer
,
2380 rb_is_reader_page(cpu_buffer
->tail_page
)))
2382 local_set(&cpu_buffer
->commit_page
->page
->commit
,
2383 rb_page_write(cpu_buffer
->commit_page
));
2384 rb_inc_page(cpu_buffer
, &cpu_buffer
->commit_page
);
2385 /* Only update the write stamp if the page has an event */
2386 if (rb_page_write(cpu_buffer
->commit_page
))
2387 cpu_buffer
->write_stamp
=
2388 cpu_buffer
->commit_page
->page
->time_stamp
;
2389 /* add barrier to keep gcc from optimizing too much */
2392 while (rb_commit_index(cpu_buffer
) !=
2393 rb_page_write(cpu_buffer
->commit_page
)) {
2395 local_set(&cpu_buffer
->commit_page
->page
->commit
,
2396 rb_page_write(cpu_buffer
->commit_page
));
2397 RB_WARN_ON(cpu_buffer
,
2398 local_read(&cpu_buffer
->commit_page
->page
->commit
) &
2403 /* again, keep gcc from optimizing */
2407 * If an interrupt came in just after the first while loop
2408 * and pushed the tail page forward, we will be left with
2409 * a dangling commit that will never go forward.
2411 if (unlikely(cpu_buffer
->commit_page
!= READ_ONCE(cpu_buffer
->tail_page
)))
2415 static __always_inline
void rb_end_commit(struct ring_buffer_per_cpu
*cpu_buffer
)
2417 unsigned long commits
;
2419 if (RB_WARN_ON(cpu_buffer
,
2420 !local_read(&cpu_buffer
->committing
)))
2424 commits
= local_read(&cpu_buffer
->commits
);
2425 /* synchronize with interrupts */
2427 if (local_read(&cpu_buffer
->committing
) == 1)
2428 rb_set_commit_to_write(cpu_buffer
);
2430 local_dec(&cpu_buffer
->committing
);
2432 /* synchronize with interrupts */
2436 * Need to account for interrupts coming in between the
2437 * updating of the commit page and the clearing of the
2438 * committing counter.
2440 if (unlikely(local_read(&cpu_buffer
->commits
) != commits
) &&
2441 !local_read(&cpu_buffer
->committing
)) {
2442 local_inc(&cpu_buffer
->committing
);
2447 static inline void rb_event_discard(struct ring_buffer_event
*event
)
2449 if (event
->type_len
== RINGBUF_TYPE_TIME_EXTEND
)
2450 event
= skip_time_extend(event
);
2452 /* array[0] holds the actual length for the discarded event */
2453 event
->array
[0] = rb_event_data_length(event
) - RB_EVNT_HDR_SIZE
;
2454 event
->type_len
= RINGBUF_TYPE_PADDING
;
2455 /* time delta must be non zero */
2456 if (!event
->time_delta
)
2457 event
->time_delta
= 1;
2460 static __always_inline
bool
2461 rb_event_is_commit(struct ring_buffer_per_cpu
*cpu_buffer
,
2462 struct ring_buffer_event
*event
)
2464 unsigned long addr
= (unsigned long)event
;
2465 unsigned long index
;
2467 index
= rb_event_index(event
);
2470 return cpu_buffer
->commit_page
->page
== (void *)addr
&&
2471 rb_commit_index(cpu_buffer
) == index
;
2474 static __always_inline
void
2475 rb_update_write_stamp(struct ring_buffer_per_cpu
*cpu_buffer
,
2476 struct ring_buffer_event
*event
)
2481 * The event first in the commit queue updates the
2484 if (rb_event_is_commit(cpu_buffer
, event
)) {
2486 * A commit event that is first on a page
2487 * updates the write timestamp with the page stamp
2489 if (!rb_event_index(event
))
2490 cpu_buffer
->write_stamp
=
2491 cpu_buffer
->commit_page
->page
->time_stamp
;
2492 else if (event
->type_len
== RINGBUF_TYPE_TIME_EXTEND
) {
2493 delta
= event
->array
[0];
2495 delta
+= event
->time_delta
;
2496 cpu_buffer
->write_stamp
+= delta
;
2498 cpu_buffer
->write_stamp
+= event
->time_delta
;
2502 static void rb_commit(struct ring_buffer_per_cpu
*cpu_buffer
,
2503 struct ring_buffer_event
*event
)
2505 local_inc(&cpu_buffer
->entries
);
2506 rb_update_write_stamp(cpu_buffer
, event
);
2507 rb_end_commit(cpu_buffer
);
2510 static __always_inline
void
2511 rb_wakeups(struct ring_buffer
*buffer
, struct ring_buffer_per_cpu
*cpu_buffer
)
2515 if (buffer
->irq_work
.waiters_pending
) {
2516 buffer
->irq_work
.waiters_pending
= false;
2517 /* irq_work_queue() supplies it's own memory barriers */
2518 irq_work_queue(&buffer
->irq_work
.work
);
2521 if (cpu_buffer
->irq_work
.waiters_pending
) {
2522 cpu_buffer
->irq_work
.waiters_pending
= false;
2523 /* irq_work_queue() supplies it's own memory barriers */
2524 irq_work_queue(&cpu_buffer
->irq_work
.work
);
2527 pagebusy
= cpu_buffer
->reader_page
== cpu_buffer
->commit_page
;
2529 if (!pagebusy
&& cpu_buffer
->irq_work
.full_waiters_pending
) {
2530 cpu_buffer
->irq_work
.wakeup_full
= true;
2531 cpu_buffer
->irq_work
.full_waiters_pending
= false;
2532 /* irq_work_queue() supplies it's own memory barriers */
2533 irq_work_queue(&cpu_buffer
->irq_work
.work
);
2538 * The lock and unlock are done within a preempt disable section.
2539 * The current_context per_cpu variable can only be modified
2540 * by the current task between lock and unlock. But it can
2541 * be modified more than once via an interrupt. To pass this
2542 * information from the lock to the unlock without having to
2543 * access the 'in_interrupt()' functions again (which do show
2544 * a bit of overhead in something as critical as function tracing,
2545 * we use a bitmask trick.
2547 * bit 0 = NMI context
2548 * bit 1 = IRQ context
2549 * bit 2 = SoftIRQ context
2550 * bit 3 = normal context.
2552 * This works because this is the order of contexts that can
2553 * preempt other contexts. A SoftIRQ never preempts an IRQ
2556 * When the context is determined, the corresponding bit is
2557 * checked and set (if it was set, then a recursion of that context
2560 * On unlock, we need to clear this bit. To do so, just subtract
2561 * 1 from the current_context and AND it to itself.
2565 * 101 & 100 = 100 (clearing bit zero)
2568 * 1010 & 1001 = 1000 (clearing bit 1)
2570 * The least significant bit can be cleared this way, and it
2571 * just so happens that it is the same bit corresponding to
2572 * the current context.
2575 static __always_inline
int
2576 trace_recursive_lock(struct ring_buffer_per_cpu
*cpu_buffer
)
2578 unsigned int val
= cpu_buffer
->current_context
;
2581 if (in_interrupt()) {
2587 bit
= RB_CTX_SOFTIRQ
;
2589 bit
= RB_CTX_NORMAL
;
2591 if (unlikely(val
& (1 << bit
)))
2595 cpu_buffer
->current_context
= val
;
2600 static __always_inline
void
2601 trace_recursive_unlock(struct ring_buffer_per_cpu
*cpu_buffer
)
2603 cpu_buffer
->current_context
&= cpu_buffer
->current_context
- 1;
2607 * ring_buffer_unlock_commit - commit a reserved
2608 * @buffer: The buffer to commit to
2609 * @event: The event pointer to commit.
2611 * This commits the data to the ring buffer, and releases any locks held.
2613 * Must be paired with ring_buffer_lock_reserve.
2615 int ring_buffer_unlock_commit(struct ring_buffer
*buffer
,
2616 struct ring_buffer_event
*event
)
2618 struct ring_buffer_per_cpu
*cpu_buffer
;
2619 int cpu
= raw_smp_processor_id();
2621 cpu_buffer
= buffer
->buffers
[cpu
];
2623 rb_commit(cpu_buffer
, event
);
2625 rb_wakeups(buffer
, cpu_buffer
);
2627 trace_recursive_unlock(cpu_buffer
);
2629 preempt_enable_notrace();
2633 EXPORT_SYMBOL_GPL(ring_buffer_unlock_commit
);
2635 static noinline
void
2636 rb_handle_timestamp(struct ring_buffer_per_cpu
*cpu_buffer
,
2637 struct rb_event_info
*info
)
2639 WARN_ONCE(info
->delta
> (1ULL << 59),
2640 KERN_WARNING
"Delta way too big! %llu ts=%llu write stamp = %llu\n%s",
2641 (unsigned long long)info
->delta
,
2642 (unsigned long long)info
->ts
,
2643 (unsigned long long)cpu_buffer
->write_stamp
,
2644 sched_clock_stable() ? "" :
2645 "If you just came from a suspend/resume,\n"
2646 "please switch to the trace global clock:\n"
2647 " echo global > /sys/kernel/debug/tracing/trace_clock\n");
2648 info
->add_timestamp
= 1;
2651 static struct ring_buffer_event
*
2652 __rb_reserve_next(struct ring_buffer_per_cpu
*cpu_buffer
,
2653 struct rb_event_info
*info
)
2655 struct ring_buffer_event
*event
;
2656 struct buffer_page
*tail_page
;
2657 unsigned long tail
, write
;
2660 * If the time delta since the last event is too big to
2661 * hold in the time field of the event, then we append a
2662 * TIME EXTEND event ahead of the data event.
2664 if (unlikely(info
->add_timestamp
))
2665 info
->length
+= RB_LEN_TIME_EXTEND
;
2667 /* Don't let the compiler play games with cpu_buffer->tail_page */
2668 tail_page
= info
->tail_page
= READ_ONCE(cpu_buffer
->tail_page
);
2669 write
= local_add_return(info
->length
, &tail_page
->write
);
2671 /* set write to only the index of the write */
2672 write
&= RB_WRITE_MASK
;
2673 tail
= write
- info
->length
;
2676 * If this is the first commit on the page, then it has the same
2677 * timestamp as the page itself.
2682 /* See if we shot pass the end of this buffer page */
2683 if (unlikely(write
> BUF_PAGE_SIZE
))
2684 return rb_move_tail(cpu_buffer
, tail
, info
);
2686 /* We reserved something on the buffer */
2688 event
= __rb_page_index(tail_page
, tail
);
2689 kmemcheck_annotate_bitfield(event
, bitfield
);
2690 rb_update_event(cpu_buffer
, event
, info
);
2692 local_inc(&tail_page
->entries
);
2695 * If this is the first commit on the page, then update
2699 tail_page
->page
->time_stamp
= info
->ts
;
2701 /* account for these added bytes */
2702 local_add(info
->length
, &cpu_buffer
->entries_bytes
);
2707 static __always_inline
struct ring_buffer_event
*
2708 rb_reserve_next_event(struct ring_buffer
*buffer
,
2709 struct ring_buffer_per_cpu
*cpu_buffer
,
2710 unsigned long length
)
2712 struct ring_buffer_event
*event
;
2713 struct rb_event_info info
;
2717 rb_start_commit(cpu_buffer
);
2719 #ifdef CONFIG_RING_BUFFER_ALLOW_SWAP
2721 * Due to the ability to swap a cpu buffer from a buffer
2722 * it is possible it was swapped before we committed.
2723 * (committing stops a swap). We check for it here and
2724 * if it happened, we have to fail the write.
2727 if (unlikely(ACCESS_ONCE(cpu_buffer
->buffer
) != buffer
)) {
2728 local_dec(&cpu_buffer
->committing
);
2729 local_dec(&cpu_buffer
->commits
);
2734 info
.length
= rb_calculate_event_length(length
);
2736 info
.add_timestamp
= 0;
2740 * We allow for interrupts to reenter here and do a trace.
2741 * If one does, it will cause this original code to loop
2742 * back here. Even with heavy interrupts happening, this
2743 * should only happen a few times in a row. If this happens
2744 * 1000 times in a row, there must be either an interrupt
2745 * storm or we have something buggy.
2748 if (RB_WARN_ON(cpu_buffer
, ++nr_loops
> 1000))
2751 info
.ts
= rb_time_stamp(cpu_buffer
->buffer
);
2752 diff
= info
.ts
- cpu_buffer
->write_stamp
;
2754 /* make sure this diff is calculated here */
2757 /* Did the write stamp get updated already? */
2758 if (likely(info
.ts
>= cpu_buffer
->write_stamp
)) {
2760 if (unlikely(test_time_stamp(info
.delta
)))
2761 rb_handle_timestamp(cpu_buffer
, &info
);
2764 event
= __rb_reserve_next(cpu_buffer
, &info
);
2766 if (unlikely(PTR_ERR(event
) == -EAGAIN
)) {
2767 if (info
.add_timestamp
)
2768 info
.length
-= RB_LEN_TIME_EXTEND
;
2778 rb_end_commit(cpu_buffer
);
2783 * ring_buffer_lock_reserve - reserve a part of the buffer
2784 * @buffer: the ring buffer to reserve from
2785 * @length: the length of the data to reserve (excluding event header)
2787 * Returns a reseverd event on the ring buffer to copy directly to.
2788 * The user of this interface will need to get the body to write into
2789 * and can use the ring_buffer_event_data() interface.
2791 * The length is the length of the data needed, not the event length
2792 * which also includes the event header.
2794 * Must be paired with ring_buffer_unlock_commit, unless NULL is returned.
2795 * If NULL is returned, then nothing has been allocated or locked.
2797 struct ring_buffer_event
*
2798 ring_buffer_lock_reserve(struct ring_buffer
*buffer
, unsigned long length
)
2800 struct ring_buffer_per_cpu
*cpu_buffer
;
2801 struct ring_buffer_event
*event
;
2804 /* If we are tracing schedule, we don't want to recurse */
2805 preempt_disable_notrace();
2807 if (unlikely(atomic_read(&buffer
->record_disabled
)))
2810 cpu
= raw_smp_processor_id();
2812 if (unlikely(!cpumask_test_cpu(cpu
, buffer
->cpumask
)))
2815 cpu_buffer
= buffer
->buffers
[cpu
];
2817 if (unlikely(atomic_read(&cpu_buffer
->record_disabled
)))
2820 if (unlikely(length
> BUF_MAX_DATA_SIZE
))
2823 if (unlikely(trace_recursive_lock(cpu_buffer
)))
2826 event
= rb_reserve_next_event(buffer
, cpu_buffer
, length
);
2833 trace_recursive_unlock(cpu_buffer
);
2835 preempt_enable_notrace();
2838 EXPORT_SYMBOL_GPL(ring_buffer_lock_reserve
);
2841 * Decrement the entries to the page that an event is on.
2842 * The event does not even need to exist, only the pointer
2843 * to the page it is on. This may only be called before the commit
2847 rb_decrement_entry(struct ring_buffer_per_cpu
*cpu_buffer
,
2848 struct ring_buffer_event
*event
)
2850 unsigned long addr
= (unsigned long)event
;
2851 struct buffer_page
*bpage
= cpu_buffer
->commit_page
;
2852 struct buffer_page
*start
;
2856 /* Do the likely case first */
2857 if (likely(bpage
->page
== (void *)addr
)) {
2858 local_dec(&bpage
->entries
);
2863 * Because the commit page may be on the reader page we
2864 * start with the next page and check the end loop there.
2866 rb_inc_page(cpu_buffer
, &bpage
);
2869 if (bpage
->page
== (void *)addr
) {
2870 local_dec(&bpage
->entries
);
2873 rb_inc_page(cpu_buffer
, &bpage
);
2874 } while (bpage
!= start
);
2876 /* commit not part of this buffer?? */
2877 RB_WARN_ON(cpu_buffer
, 1);
2881 * ring_buffer_commit_discard - discard an event that has not been committed
2882 * @buffer: the ring buffer
2883 * @event: non committed event to discard
2885 * Sometimes an event that is in the ring buffer needs to be ignored.
2886 * This function lets the user discard an event in the ring buffer
2887 * and then that event will not be read later.
2889 * This function only works if it is called before the the item has been
2890 * committed. It will try to free the event from the ring buffer
2891 * if another event has not been added behind it.
2893 * If another event has been added behind it, it will set the event
2894 * up as discarded, and perform the commit.
2896 * If this function is called, do not call ring_buffer_unlock_commit on
2899 void ring_buffer_discard_commit(struct ring_buffer
*buffer
,
2900 struct ring_buffer_event
*event
)
2902 struct ring_buffer_per_cpu
*cpu_buffer
;
2905 /* The event is discarded regardless */
2906 rb_event_discard(event
);
2908 cpu
= smp_processor_id();
2909 cpu_buffer
= buffer
->buffers
[cpu
];
2912 * This must only be called if the event has not been
2913 * committed yet. Thus we can assume that preemption
2914 * is still disabled.
2916 RB_WARN_ON(buffer
, !local_read(&cpu_buffer
->committing
));
2918 rb_decrement_entry(cpu_buffer
, event
);
2919 if (rb_try_to_discard(cpu_buffer
, event
))
2923 * The commit is still visible by the reader, so we
2924 * must still update the timestamp.
2926 rb_update_write_stamp(cpu_buffer
, event
);
2928 rb_end_commit(cpu_buffer
);
2930 trace_recursive_unlock(cpu_buffer
);
2932 preempt_enable_notrace();
2935 EXPORT_SYMBOL_GPL(ring_buffer_discard_commit
);
2938 * ring_buffer_write - write data to the buffer without reserving
2939 * @buffer: The ring buffer to write to.
2940 * @length: The length of the data being written (excluding the event header)
2941 * @data: The data to write to the buffer.
2943 * This is like ring_buffer_lock_reserve and ring_buffer_unlock_commit as
2944 * one function. If you already have the data to write to the buffer, it
2945 * may be easier to simply call this function.
2947 * Note, like ring_buffer_lock_reserve, the length is the length of the data
2948 * and not the length of the event which would hold the header.
2950 int ring_buffer_write(struct ring_buffer
*buffer
,
2951 unsigned long length
,
2954 struct ring_buffer_per_cpu
*cpu_buffer
;
2955 struct ring_buffer_event
*event
;
2960 preempt_disable_notrace();
2962 if (atomic_read(&buffer
->record_disabled
))
2965 cpu
= raw_smp_processor_id();
2967 if (!cpumask_test_cpu(cpu
, buffer
->cpumask
))
2970 cpu_buffer
= buffer
->buffers
[cpu
];
2972 if (atomic_read(&cpu_buffer
->record_disabled
))
2975 if (length
> BUF_MAX_DATA_SIZE
)
2978 if (unlikely(trace_recursive_lock(cpu_buffer
)))
2981 event
= rb_reserve_next_event(buffer
, cpu_buffer
, length
);
2985 body
= rb_event_data(event
);
2987 memcpy(body
, data
, length
);
2989 rb_commit(cpu_buffer
, event
);
2991 rb_wakeups(buffer
, cpu_buffer
);
2996 trace_recursive_unlock(cpu_buffer
);
2999 preempt_enable_notrace();
3003 EXPORT_SYMBOL_GPL(ring_buffer_write
);
3005 static bool rb_per_cpu_empty(struct ring_buffer_per_cpu
*cpu_buffer
)
3007 struct buffer_page
*reader
= cpu_buffer
->reader_page
;
3008 struct buffer_page
*head
= rb_set_head_page(cpu_buffer
);
3009 struct buffer_page
*commit
= cpu_buffer
->commit_page
;
3011 /* In case of error, head will be NULL */
3012 if (unlikely(!head
))
3015 return reader
->read
== rb_page_commit(reader
) &&
3016 (commit
== reader
||
3018 head
->read
== rb_page_commit(commit
)));
3022 * ring_buffer_record_disable - stop all writes into the buffer
3023 * @buffer: The ring buffer to stop writes to.
3025 * This prevents all writes to the buffer. Any attempt to write
3026 * to the buffer after this will fail and return NULL.
3028 * The caller should call synchronize_sched() after this.
3030 void ring_buffer_record_disable(struct ring_buffer
*buffer
)
3032 atomic_inc(&buffer
->record_disabled
);
3034 EXPORT_SYMBOL_GPL(ring_buffer_record_disable
);
3037 * ring_buffer_record_enable - enable writes to the buffer
3038 * @buffer: The ring buffer to enable writes
3040 * Note, multiple disables will need the same number of enables
3041 * to truly enable the writing (much like preempt_disable).
3043 void ring_buffer_record_enable(struct ring_buffer
*buffer
)
3045 atomic_dec(&buffer
->record_disabled
);
3047 EXPORT_SYMBOL_GPL(ring_buffer_record_enable
);
3050 * ring_buffer_record_off - stop all writes into the buffer
3051 * @buffer: The ring buffer to stop writes to.
3053 * This prevents all writes to the buffer. Any attempt to write
3054 * to the buffer after this will fail and return NULL.
3056 * This is different than ring_buffer_record_disable() as
3057 * it works like an on/off switch, where as the disable() version
3058 * must be paired with a enable().
3060 void ring_buffer_record_off(struct ring_buffer
*buffer
)
3063 unsigned int new_rd
;
3066 rd
= atomic_read(&buffer
->record_disabled
);
3067 new_rd
= rd
| RB_BUFFER_OFF
;
3068 } while (atomic_cmpxchg(&buffer
->record_disabled
, rd
, new_rd
) != rd
);
3070 EXPORT_SYMBOL_GPL(ring_buffer_record_off
);
3073 * ring_buffer_record_on - restart writes into the buffer
3074 * @buffer: The ring buffer to start writes to.
3076 * This enables all writes to the buffer that was disabled by
3077 * ring_buffer_record_off().
3079 * This is different than ring_buffer_record_enable() as
3080 * it works like an on/off switch, where as the enable() version
3081 * must be paired with a disable().
3083 void ring_buffer_record_on(struct ring_buffer
*buffer
)
3086 unsigned int new_rd
;
3089 rd
= atomic_read(&buffer
->record_disabled
);
3090 new_rd
= rd
& ~RB_BUFFER_OFF
;
3091 } while (atomic_cmpxchg(&buffer
->record_disabled
, rd
, new_rd
) != rd
);
3093 EXPORT_SYMBOL_GPL(ring_buffer_record_on
);
3096 * ring_buffer_record_is_on - return true if the ring buffer can write
3097 * @buffer: The ring buffer to see if write is enabled
3099 * Returns true if the ring buffer is in a state that it accepts writes.
3101 int ring_buffer_record_is_on(struct ring_buffer
*buffer
)
3103 return !atomic_read(&buffer
->record_disabled
);
3107 * ring_buffer_record_disable_cpu - stop all writes into the cpu_buffer
3108 * @buffer: The ring buffer to stop writes to.
3109 * @cpu: The CPU buffer to stop
3111 * This prevents all writes to the buffer. Any attempt to write
3112 * to the buffer after this will fail and return NULL.
3114 * The caller should call synchronize_sched() after this.
3116 void ring_buffer_record_disable_cpu(struct ring_buffer
*buffer
, int cpu
)
3118 struct ring_buffer_per_cpu
*cpu_buffer
;
3120 if (!cpumask_test_cpu(cpu
, buffer
->cpumask
))
3123 cpu_buffer
= buffer
->buffers
[cpu
];
3124 atomic_inc(&cpu_buffer
->record_disabled
);
3126 EXPORT_SYMBOL_GPL(ring_buffer_record_disable_cpu
);
3129 * ring_buffer_record_enable_cpu - enable writes to the buffer
3130 * @buffer: The ring buffer to enable writes
3131 * @cpu: The CPU to enable.
3133 * Note, multiple disables will need the same number of enables
3134 * to truly enable the writing (much like preempt_disable).
3136 void ring_buffer_record_enable_cpu(struct ring_buffer
*buffer
, int cpu
)
3138 struct ring_buffer_per_cpu
*cpu_buffer
;
3140 if (!cpumask_test_cpu(cpu
, buffer
->cpumask
))
3143 cpu_buffer
= buffer
->buffers
[cpu
];
3144 atomic_dec(&cpu_buffer
->record_disabled
);
3146 EXPORT_SYMBOL_GPL(ring_buffer_record_enable_cpu
);
3149 * The total entries in the ring buffer is the running counter
3150 * of entries entered into the ring buffer, minus the sum of
3151 * the entries read from the ring buffer and the number of
3152 * entries that were overwritten.
3154 static inline unsigned long
3155 rb_num_of_entries(struct ring_buffer_per_cpu
*cpu_buffer
)
3157 return local_read(&cpu_buffer
->entries
) -
3158 (local_read(&cpu_buffer
->overrun
) + cpu_buffer
->read
);
3162 * ring_buffer_oldest_event_ts - get the oldest event timestamp from the buffer
3163 * @buffer: The ring buffer
3164 * @cpu: The per CPU buffer to read from.
3166 u64
ring_buffer_oldest_event_ts(struct ring_buffer
*buffer
, int cpu
)
3168 unsigned long flags
;
3169 struct ring_buffer_per_cpu
*cpu_buffer
;
3170 struct buffer_page
*bpage
;
3173 if (!cpumask_test_cpu(cpu
, buffer
->cpumask
))
3176 cpu_buffer
= buffer
->buffers
[cpu
];
3177 raw_spin_lock_irqsave(&cpu_buffer
->reader_lock
, flags
);
3179 * if the tail is on reader_page, oldest time stamp is on the reader
3182 if (cpu_buffer
->tail_page
== cpu_buffer
->reader_page
)
3183 bpage
= cpu_buffer
->reader_page
;
3185 bpage
= rb_set_head_page(cpu_buffer
);
3187 ret
= bpage
->page
->time_stamp
;
3188 raw_spin_unlock_irqrestore(&cpu_buffer
->reader_lock
, flags
);
3192 EXPORT_SYMBOL_GPL(ring_buffer_oldest_event_ts
);
3195 * ring_buffer_bytes_cpu - get the number of bytes consumed in a cpu buffer
3196 * @buffer: The ring buffer
3197 * @cpu: The per CPU buffer to read from.
3199 unsigned long ring_buffer_bytes_cpu(struct ring_buffer
*buffer
, int cpu
)
3201 struct ring_buffer_per_cpu
*cpu_buffer
;
3204 if (!cpumask_test_cpu(cpu
, buffer
->cpumask
))
3207 cpu_buffer
= buffer
->buffers
[cpu
];
3208 ret
= local_read(&cpu_buffer
->entries_bytes
) - cpu_buffer
->read_bytes
;
3212 EXPORT_SYMBOL_GPL(ring_buffer_bytes_cpu
);
3215 * ring_buffer_entries_cpu - get the number of entries in a cpu buffer
3216 * @buffer: The ring buffer
3217 * @cpu: The per CPU buffer to get the entries from.
3219 unsigned long ring_buffer_entries_cpu(struct ring_buffer
*buffer
, int cpu
)
3221 struct ring_buffer_per_cpu
*cpu_buffer
;
3223 if (!cpumask_test_cpu(cpu
, buffer
->cpumask
))
3226 cpu_buffer
= buffer
->buffers
[cpu
];
3228 return rb_num_of_entries(cpu_buffer
);
3230 EXPORT_SYMBOL_GPL(ring_buffer_entries_cpu
);
3233 * ring_buffer_overrun_cpu - get the number of overruns caused by the ring
3234 * buffer wrapping around (only if RB_FL_OVERWRITE is on).
3235 * @buffer: The ring buffer
3236 * @cpu: The per CPU buffer to get the number of overruns from
3238 unsigned long ring_buffer_overrun_cpu(struct ring_buffer
*buffer
, int cpu
)
3240 struct ring_buffer_per_cpu
*cpu_buffer
;
3243 if (!cpumask_test_cpu(cpu
, buffer
->cpumask
))
3246 cpu_buffer
= buffer
->buffers
[cpu
];
3247 ret
= local_read(&cpu_buffer
->overrun
);
3251 EXPORT_SYMBOL_GPL(ring_buffer_overrun_cpu
);
3254 * ring_buffer_commit_overrun_cpu - get the number of overruns caused by
3255 * commits failing due to the buffer wrapping around while there are uncommitted
3256 * events, such as during an interrupt storm.
3257 * @buffer: The ring buffer
3258 * @cpu: The per CPU buffer to get the number of overruns from
3261 ring_buffer_commit_overrun_cpu(struct ring_buffer
*buffer
, int cpu
)
3263 struct ring_buffer_per_cpu
*cpu_buffer
;
3266 if (!cpumask_test_cpu(cpu
, buffer
->cpumask
))
3269 cpu_buffer
= buffer
->buffers
[cpu
];
3270 ret
= local_read(&cpu_buffer
->commit_overrun
);
3274 EXPORT_SYMBOL_GPL(ring_buffer_commit_overrun_cpu
);
3277 * ring_buffer_dropped_events_cpu - get the number of dropped events caused by
3278 * the ring buffer filling up (only if RB_FL_OVERWRITE is off).
3279 * @buffer: The ring buffer
3280 * @cpu: The per CPU buffer to get the number of overruns from
3283 ring_buffer_dropped_events_cpu(struct ring_buffer
*buffer
, int cpu
)
3285 struct ring_buffer_per_cpu
*cpu_buffer
;
3288 if (!cpumask_test_cpu(cpu
, buffer
->cpumask
))
3291 cpu_buffer
= buffer
->buffers
[cpu
];
3292 ret
= local_read(&cpu_buffer
->dropped_events
);
3296 EXPORT_SYMBOL_GPL(ring_buffer_dropped_events_cpu
);
3299 * ring_buffer_read_events_cpu - get the number of events successfully read
3300 * @buffer: The ring buffer
3301 * @cpu: The per CPU buffer to get the number of events read
3304 ring_buffer_read_events_cpu(struct ring_buffer
*buffer
, int cpu
)
3306 struct ring_buffer_per_cpu
*cpu_buffer
;
3308 if (!cpumask_test_cpu(cpu
, buffer
->cpumask
))
3311 cpu_buffer
= buffer
->buffers
[cpu
];
3312 return cpu_buffer
->read
;
3314 EXPORT_SYMBOL_GPL(ring_buffer_read_events_cpu
);
3317 * ring_buffer_entries - get the number of entries in a buffer
3318 * @buffer: The ring buffer
3320 * Returns the total number of entries in the ring buffer
3323 unsigned long ring_buffer_entries(struct ring_buffer
*buffer
)
3325 struct ring_buffer_per_cpu
*cpu_buffer
;
3326 unsigned long entries
= 0;
3329 /* if you care about this being correct, lock the buffer */
3330 for_each_buffer_cpu(buffer
, cpu
) {
3331 cpu_buffer
= buffer
->buffers
[cpu
];
3332 entries
+= rb_num_of_entries(cpu_buffer
);
3337 EXPORT_SYMBOL_GPL(ring_buffer_entries
);
3340 * ring_buffer_overruns - get the number of overruns in buffer
3341 * @buffer: The ring buffer
3343 * Returns the total number of overruns in the ring buffer
3346 unsigned long ring_buffer_overruns(struct ring_buffer
*buffer
)
3348 struct ring_buffer_per_cpu
*cpu_buffer
;
3349 unsigned long overruns
= 0;
3352 /* if you care about this being correct, lock the buffer */
3353 for_each_buffer_cpu(buffer
, cpu
) {
3354 cpu_buffer
= buffer
->buffers
[cpu
];
3355 overruns
+= local_read(&cpu_buffer
->overrun
);
3360 EXPORT_SYMBOL_GPL(ring_buffer_overruns
);
3362 static void rb_iter_reset(struct ring_buffer_iter
*iter
)
3364 struct ring_buffer_per_cpu
*cpu_buffer
= iter
->cpu_buffer
;
3366 /* Iterator usage is expected to have record disabled */
3367 iter
->head_page
= cpu_buffer
->reader_page
;
3368 iter
->head
= cpu_buffer
->reader_page
->read
;
3370 iter
->cache_reader_page
= iter
->head_page
;
3371 iter
->cache_read
= cpu_buffer
->read
;
3374 iter
->read_stamp
= cpu_buffer
->read_stamp
;
3376 iter
->read_stamp
= iter
->head_page
->page
->time_stamp
;
3380 * ring_buffer_iter_reset - reset an iterator
3381 * @iter: The iterator to reset
3383 * Resets the iterator, so that it will start from the beginning
3386 void ring_buffer_iter_reset(struct ring_buffer_iter
*iter
)
3388 struct ring_buffer_per_cpu
*cpu_buffer
;
3389 unsigned long flags
;
3394 cpu_buffer
= iter
->cpu_buffer
;
3396 raw_spin_lock_irqsave(&cpu_buffer
->reader_lock
, flags
);
3397 rb_iter_reset(iter
);
3398 raw_spin_unlock_irqrestore(&cpu_buffer
->reader_lock
, flags
);
3400 EXPORT_SYMBOL_GPL(ring_buffer_iter_reset
);
3403 * ring_buffer_iter_empty - check if an iterator has no more to read
3404 * @iter: The iterator to check
3406 int ring_buffer_iter_empty(struct ring_buffer_iter
*iter
)
3408 struct ring_buffer_per_cpu
*cpu_buffer
;
3409 struct buffer_page
*reader
;
3410 struct buffer_page
*head_page
;
3411 struct buffer_page
*commit_page
;
3414 cpu_buffer
= iter
->cpu_buffer
;
3416 /* Remember, trace recording is off when iterator is in use */
3417 reader
= cpu_buffer
->reader_page
;
3418 head_page
= cpu_buffer
->head_page
;
3419 commit_page
= cpu_buffer
->commit_page
;
3420 commit
= rb_page_commit(commit_page
);
3422 return ((iter
->head_page
== commit_page
&& iter
->head
== commit
) ||
3423 (iter
->head_page
== reader
&& commit_page
== head_page
&&
3424 head_page
->read
== commit
&&
3425 iter
->head
== rb_page_commit(cpu_buffer
->reader_page
)));
3427 EXPORT_SYMBOL_GPL(ring_buffer_iter_empty
);
3430 rb_update_read_stamp(struct ring_buffer_per_cpu
*cpu_buffer
,
3431 struct ring_buffer_event
*event
)
3435 switch (event
->type_len
) {
3436 case RINGBUF_TYPE_PADDING
:
3439 case RINGBUF_TYPE_TIME_EXTEND
:
3440 delta
= event
->array
[0];
3442 delta
+= event
->time_delta
;
3443 cpu_buffer
->read_stamp
+= delta
;
3446 case RINGBUF_TYPE_TIME_STAMP
:
3447 /* FIXME: not implemented */
3450 case RINGBUF_TYPE_DATA
:
3451 cpu_buffer
->read_stamp
+= event
->time_delta
;
3461 rb_update_iter_read_stamp(struct ring_buffer_iter
*iter
,
3462 struct ring_buffer_event
*event
)
3466 switch (event
->type_len
) {
3467 case RINGBUF_TYPE_PADDING
:
3470 case RINGBUF_TYPE_TIME_EXTEND
:
3471 delta
= event
->array
[0];
3473 delta
+= event
->time_delta
;
3474 iter
->read_stamp
+= delta
;
3477 case RINGBUF_TYPE_TIME_STAMP
:
3478 /* FIXME: not implemented */
3481 case RINGBUF_TYPE_DATA
:
3482 iter
->read_stamp
+= event
->time_delta
;
3491 static struct buffer_page
*
3492 rb_get_reader_page(struct ring_buffer_per_cpu
*cpu_buffer
)
3494 struct buffer_page
*reader
= NULL
;
3495 unsigned long overwrite
;
3496 unsigned long flags
;
3500 local_irq_save(flags
);
3501 arch_spin_lock(&cpu_buffer
->lock
);
3505 * This should normally only loop twice. But because the
3506 * start of the reader inserts an empty page, it causes
3507 * a case where we will loop three times. There should be no
3508 * reason to loop four times (that I know of).
3510 if (RB_WARN_ON(cpu_buffer
, ++nr_loops
> 3)) {
3515 reader
= cpu_buffer
->reader_page
;
3517 /* If there's more to read, return this page */
3518 if (cpu_buffer
->reader_page
->read
< rb_page_size(reader
))
3521 /* Never should we have an index greater than the size */
3522 if (RB_WARN_ON(cpu_buffer
,
3523 cpu_buffer
->reader_page
->read
> rb_page_size(reader
)))
3526 /* check if we caught up to the tail */
3528 if (cpu_buffer
->commit_page
== cpu_buffer
->reader_page
)
3531 /* Don't bother swapping if the ring buffer is empty */
3532 if (rb_num_of_entries(cpu_buffer
) == 0)
3536 * Reset the reader page to size zero.
3538 local_set(&cpu_buffer
->reader_page
->write
, 0);
3539 local_set(&cpu_buffer
->reader_page
->entries
, 0);
3540 local_set(&cpu_buffer
->reader_page
->page
->commit
, 0);
3541 cpu_buffer
->reader_page
->real_end
= 0;
3545 * Splice the empty reader page into the list around the head.
3547 reader
= rb_set_head_page(cpu_buffer
);
3550 cpu_buffer
->reader_page
->list
.next
= rb_list_head(reader
->list
.next
);
3551 cpu_buffer
->reader_page
->list
.prev
= reader
->list
.prev
;
3554 * cpu_buffer->pages just needs to point to the buffer, it
3555 * has no specific buffer page to point to. Lets move it out
3556 * of our way so we don't accidentally swap it.
3558 cpu_buffer
->pages
= reader
->list
.prev
;
3560 /* The reader page will be pointing to the new head */
3561 rb_set_list_to_head(cpu_buffer
, &cpu_buffer
->reader_page
->list
);
3564 * We want to make sure we read the overruns after we set up our
3565 * pointers to the next object. The writer side does a
3566 * cmpxchg to cross pages which acts as the mb on the writer
3567 * side. Note, the reader will constantly fail the swap
3568 * while the writer is updating the pointers, so this
3569 * guarantees that the overwrite recorded here is the one we
3570 * want to compare with the last_overrun.
3573 overwrite
= local_read(&(cpu_buffer
->overrun
));
3576 * Here's the tricky part.
3578 * We need to move the pointer past the header page.
3579 * But we can only do that if a writer is not currently
3580 * moving it. The page before the header page has the
3581 * flag bit '1' set if it is pointing to the page we want.
3582 * but if the writer is in the process of moving it
3583 * than it will be '2' or already moved '0'.
3586 ret
= rb_head_page_replace(reader
, cpu_buffer
->reader_page
);
3589 * If we did not convert it, then we must try again.
3595 * Yeah! We succeeded in replacing the page.
3597 * Now make the new head point back to the reader page.
3599 rb_list_head(reader
->list
.next
)->prev
= &cpu_buffer
->reader_page
->list
;
3600 rb_inc_page(cpu_buffer
, &cpu_buffer
->head_page
);
3602 /* Finally update the reader page to the new head */
3603 cpu_buffer
->reader_page
= reader
;
3604 cpu_buffer
->reader_page
->read
= 0;
3606 if (overwrite
!= cpu_buffer
->last_overrun
) {
3607 cpu_buffer
->lost_events
= overwrite
- cpu_buffer
->last_overrun
;
3608 cpu_buffer
->last_overrun
= overwrite
;
3614 /* Update the read_stamp on the first event */
3615 if (reader
&& reader
->read
== 0)
3616 cpu_buffer
->read_stamp
= reader
->page
->time_stamp
;
3618 arch_spin_unlock(&cpu_buffer
->lock
);
3619 local_irq_restore(flags
);
3624 static void rb_advance_reader(struct ring_buffer_per_cpu
*cpu_buffer
)
3626 struct ring_buffer_event
*event
;
3627 struct buffer_page
*reader
;
3630 reader
= rb_get_reader_page(cpu_buffer
);
3632 /* This function should not be called when buffer is empty */
3633 if (RB_WARN_ON(cpu_buffer
, !reader
))
3636 event
= rb_reader_event(cpu_buffer
);
3638 if (event
->type_len
<= RINGBUF_TYPE_DATA_TYPE_LEN_MAX
)
3641 rb_update_read_stamp(cpu_buffer
, event
);
3643 length
= rb_event_length(event
);
3644 cpu_buffer
->reader_page
->read
+= length
;
3647 static void rb_advance_iter(struct ring_buffer_iter
*iter
)
3649 struct ring_buffer_per_cpu
*cpu_buffer
;
3650 struct ring_buffer_event
*event
;
3653 cpu_buffer
= iter
->cpu_buffer
;
3656 * Check if we are at the end of the buffer.
3658 if (iter
->head
>= rb_page_size(iter
->head_page
)) {
3659 /* discarded commits can make the page empty */
3660 if (iter
->head_page
== cpu_buffer
->commit_page
)
3666 event
= rb_iter_head_event(iter
);
3668 length
= rb_event_length(event
);
3671 * This should not be called to advance the header if we are
3672 * at the tail of the buffer.
3674 if (RB_WARN_ON(cpu_buffer
,
3675 (iter
->head_page
== cpu_buffer
->commit_page
) &&
3676 (iter
->head
+ length
> rb_commit_index(cpu_buffer
))))
3679 rb_update_iter_read_stamp(iter
, event
);
3681 iter
->head
+= length
;
3683 /* check for end of page padding */
3684 if ((iter
->head
>= rb_page_size(iter
->head_page
)) &&
3685 (iter
->head_page
!= cpu_buffer
->commit_page
))
3689 static int rb_lost_events(struct ring_buffer_per_cpu
*cpu_buffer
)
3691 return cpu_buffer
->lost_events
;
3694 static struct ring_buffer_event
*
3695 rb_buffer_peek(struct ring_buffer_per_cpu
*cpu_buffer
, u64
*ts
,
3696 unsigned long *lost_events
)
3698 struct ring_buffer_event
*event
;
3699 struct buffer_page
*reader
;
3704 * We repeat when a time extend is encountered.
3705 * Since the time extend is always attached to a data event,
3706 * we should never loop more than once.
3707 * (We never hit the following condition more than twice).
3709 if (RB_WARN_ON(cpu_buffer
, ++nr_loops
> 2))
3712 reader
= rb_get_reader_page(cpu_buffer
);
3716 event
= rb_reader_event(cpu_buffer
);
3718 switch (event
->type_len
) {
3719 case RINGBUF_TYPE_PADDING
:
3720 if (rb_null_event(event
))
3721 RB_WARN_ON(cpu_buffer
, 1);
3723 * Because the writer could be discarding every
3724 * event it creates (which would probably be bad)
3725 * if we were to go back to "again" then we may never
3726 * catch up, and will trigger the warn on, or lock
3727 * the box. Return the padding, and we will release
3728 * the current locks, and try again.
3732 case RINGBUF_TYPE_TIME_EXTEND
:
3733 /* Internal data, OK to advance */
3734 rb_advance_reader(cpu_buffer
);
3737 case RINGBUF_TYPE_TIME_STAMP
:
3738 /* FIXME: not implemented */
3739 rb_advance_reader(cpu_buffer
);
3742 case RINGBUF_TYPE_DATA
:
3744 *ts
= cpu_buffer
->read_stamp
+ event
->time_delta
;
3745 ring_buffer_normalize_time_stamp(cpu_buffer
->buffer
,
3746 cpu_buffer
->cpu
, ts
);
3749 *lost_events
= rb_lost_events(cpu_buffer
);
3758 EXPORT_SYMBOL_GPL(ring_buffer_peek
);
3760 static struct ring_buffer_event
*
3761 rb_iter_peek(struct ring_buffer_iter
*iter
, u64
*ts
)
3763 struct ring_buffer
*buffer
;
3764 struct ring_buffer_per_cpu
*cpu_buffer
;
3765 struct ring_buffer_event
*event
;
3768 cpu_buffer
= iter
->cpu_buffer
;
3769 buffer
= cpu_buffer
->buffer
;
3772 * Check if someone performed a consuming read to
3773 * the buffer. A consuming read invalidates the iterator
3774 * and we need to reset the iterator in this case.
3776 if (unlikely(iter
->cache_read
!= cpu_buffer
->read
||
3777 iter
->cache_reader_page
!= cpu_buffer
->reader_page
))
3778 rb_iter_reset(iter
);
3781 if (ring_buffer_iter_empty(iter
))
3785 * We repeat when a time extend is encountered or we hit
3786 * the end of the page. Since the time extend is always attached
3787 * to a data event, we should never loop more than three times.
3788 * Once for going to next page, once on time extend, and
3789 * finally once to get the event.
3790 * (We never hit the following condition more than thrice).
3792 if (RB_WARN_ON(cpu_buffer
, ++nr_loops
> 3))
3795 if (rb_per_cpu_empty(cpu_buffer
))
3798 if (iter
->head
>= rb_page_size(iter
->head_page
)) {
3803 event
= rb_iter_head_event(iter
);
3805 switch (event
->type_len
) {
3806 case RINGBUF_TYPE_PADDING
:
3807 if (rb_null_event(event
)) {
3811 rb_advance_iter(iter
);
3814 case RINGBUF_TYPE_TIME_EXTEND
:
3815 /* Internal data, OK to advance */
3816 rb_advance_iter(iter
);
3819 case RINGBUF_TYPE_TIME_STAMP
:
3820 /* FIXME: not implemented */
3821 rb_advance_iter(iter
);
3824 case RINGBUF_TYPE_DATA
:
3826 *ts
= iter
->read_stamp
+ event
->time_delta
;
3827 ring_buffer_normalize_time_stamp(buffer
,
3828 cpu_buffer
->cpu
, ts
);
3838 EXPORT_SYMBOL_GPL(ring_buffer_iter_peek
);
3840 static inline bool rb_reader_lock(struct ring_buffer_per_cpu
*cpu_buffer
)
3842 if (likely(!in_nmi())) {
3843 raw_spin_lock(&cpu_buffer
->reader_lock
);
3848 * If an NMI die dumps out the content of the ring buffer
3849 * trylock must be used to prevent a deadlock if the NMI
3850 * preempted a task that holds the ring buffer locks. If
3851 * we get the lock then all is fine, if not, then continue
3852 * to do the read, but this can corrupt the ring buffer,
3853 * so it must be permanently disabled from future writes.
3854 * Reading from NMI is a oneshot deal.
3856 if (raw_spin_trylock(&cpu_buffer
->reader_lock
))
3859 /* Continue without locking, but disable the ring buffer */
3860 atomic_inc(&cpu_buffer
->record_disabled
);
3865 rb_reader_unlock(struct ring_buffer_per_cpu
*cpu_buffer
, bool locked
)
3868 raw_spin_unlock(&cpu_buffer
->reader_lock
);
3873 * ring_buffer_peek - peek at the next event to be read
3874 * @buffer: The ring buffer to read
3875 * @cpu: The cpu to peak at
3876 * @ts: The timestamp counter of this event.
3877 * @lost_events: a variable to store if events were lost (may be NULL)
3879 * This will return the event that will be read next, but does
3880 * not consume the data.
3882 struct ring_buffer_event
*
3883 ring_buffer_peek(struct ring_buffer
*buffer
, int cpu
, u64
*ts
,
3884 unsigned long *lost_events
)
3886 struct ring_buffer_per_cpu
*cpu_buffer
= buffer
->buffers
[cpu
];
3887 struct ring_buffer_event
*event
;
3888 unsigned long flags
;
3891 if (!cpumask_test_cpu(cpu
, buffer
->cpumask
))
3895 local_irq_save(flags
);
3896 dolock
= rb_reader_lock(cpu_buffer
);
3897 event
= rb_buffer_peek(cpu_buffer
, ts
, lost_events
);
3898 if (event
&& event
->type_len
== RINGBUF_TYPE_PADDING
)
3899 rb_advance_reader(cpu_buffer
);
3900 rb_reader_unlock(cpu_buffer
, dolock
);
3901 local_irq_restore(flags
);
3903 if (event
&& event
->type_len
== RINGBUF_TYPE_PADDING
)
3910 * ring_buffer_iter_peek - peek at the next event to be read
3911 * @iter: The ring buffer iterator
3912 * @ts: The timestamp counter of this event.
3914 * This will return the event that will be read next, but does
3915 * not increment the iterator.
3917 struct ring_buffer_event
*
3918 ring_buffer_iter_peek(struct ring_buffer_iter
*iter
, u64
*ts
)
3920 struct ring_buffer_per_cpu
*cpu_buffer
= iter
->cpu_buffer
;
3921 struct ring_buffer_event
*event
;
3922 unsigned long flags
;
3925 raw_spin_lock_irqsave(&cpu_buffer
->reader_lock
, flags
);
3926 event
= rb_iter_peek(iter
, ts
);
3927 raw_spin_unlock_irqrestore(&cpu_buffer
->reader_lock
, flags
);
3929 if (event
&& event
->type_len
== RINGBUF_TYPE_PADDING
)
3936 * ring_buffer_consume - return an event and consume it
3937 * @buffer: The ring buffer to get the next event from
3938 * @cpu: the cpu to read the buffer from
3939 * @ts: a variable to store the timestamp (may be NULL)
3940 * @lost_events: a variable to store if events were lost (may be NULL)
3942 * Returns the next event in the ring buffer, and that event is consumed.
3943 * Meaning, that sequential reads will keep returning a different event,
3944 * and eventually empty the ring buffer if the producer is slower.
3946 struct ring_buffer_event
*
3947 ring_buffer_consume(struct ring_buffer
*buffer
, int cpu
, u64
*ts
,
3948 unsigned long *lost_events
)
3950 struct ring_buffer_per_cpu
*cpu_buffer
;
3951 struct ring_buffer_event
*event
= NULL
;
3952 unsigned long flags
;
3956 /* might be called in atomic */
3959 if (!cpumask_test_cpu(cpu
, buffer
->cpumask
))
3962 cpu_buffer
= buffer
->buffers
[cpu
];
3963 local_irq_save(flags
);
3964 dolock
= rb_reader_lock(cpu_buffer
);
3966 event
= rb_buffer_peek(cpu_buffer
, ts
, lost_events
);
3968 cpu_buffer
->lost_events
= 0;
3969 rb_advance_reader(cpu_buffer
);
3972 rb_reader_unlock(cpu_buffer
, dolock
);
3973 local_irq_restore(flags
);
3978 if (event
&& event
->type_len
== RINGBUF_TYPE_PADDING
)
3983 EXPORT_SYMBOL_GPL(ring_buffer_consume
);
3986 * ring_buffer_read_prepare - Prepare for a non consuming read of the buffer
3987 * @buffer: The ring buffer to read from
3988 * @cpu: The cpu buffer to iterate over
3990 * This performs the initial preparations necessary to iterate
3991 * through the buffer. Memory is allocated, buffer recording
3992 * is disabled, and the iterator pointer is returned to the caller.
3994 * Disabling buffer recordng prevents the reading from being
3995 * corrupted. This is not a consuming read, so a producer is not
3998 * After a sequence of ring_buffer_read_prepare calls, the user is
3999 * expected to make at least one call to ring_buffer_read_prepare_sync.
4000 * Afterwards, ring_buffer_read_start is invoked to get things going
4003 * This overall must be paired with ring_buffer_read_finish.
4005 struct ring_buffer_iter
*
4006 ring_buffer_read_prepare(struct ring_buffer
*buffer
, int cpu
)
4008 struct ring_buffer_per_cpu
*cpu_buffer
;
4009 struct ring_buffer_iter
*iter
;
4011 if (!cpumask_test_cpu(cpu
, buffer
->cpumask
))
4014 iter
= kmalloc(sizeof(*iter
), GFP_KERNEL
);
4018 cpu_buffer
= buffer
->buffers
[cpu
];
4020 iter
->cpu_buffer
= cpu_buffer
;
4022 atomic_inc(&buffer
->resize_disabled
);
4023 atomic_inc(&cpu_buffer
->record_disabled
);
4027 EXPORT_SYMBOL_GPL(ring_buffer_read_prepare
);
4030 * ring_buffer_read_prepare_sync - Synchronize a set of prepare calls
4032 * All previously invoked ring_buffer_read_prepare calls to prepare
4033 * iterators will be synchronized. Afterwards, read_buffer_read_start
4034 * calls on those iterators are allowed.
4037 ring_buffer_read_prepare_sync(void)
4039 synchronize_sched();
4041 EXPORT_SYMBOL_GPL(ring_buffer_read_prepare_sync
);
4044 * ring_buffer_read_start - start a non consuming read of the buffer
4045 * @iter: The iterator returned by ring_buffer_read_prepare
4047 * This finalizes the startup of an iteration through the buffer.
4048 * The iterator comes from a call to ring_buffer_read_prepare and
4049 * an intervening ring_buffer_read_prepare_sync must have been
4052 * Must be paired with ring_buffer_read_finish.
4055 ring_buffer_read_start(struct ring_buffer_iter
*iter
)
4057 struct ring_buffer_per_cpu
*cpu_buffer
;
4058 unsigned long flags
;
4063 cpu_buffer
= iter
->cpu_buffer
;
4065 raw_spin_lock_irqsave(&cpu_buffer
->reader_lock
, flags
);
4066 arch_spin_lock(&cpu_buffer
->lock
);
4067 rb_iter_reset(iter
);
4068 arch_spin_unlock(&cpu_buffer
->lock
);
4069 raw_spin_unlock_irqrestore(&cpu_buffer
->reader_lock
, flags
);
4071 EXPORT_SYMBOL_GPL(ring_buffer_read_start
);
4074 * ring_buffer_read_finish - finish reading the iterator of the buffer
4075 * @iter: The iterator retrieved by ring_buffer_start
4077 * This re-enables the recording to the buffer, and frees the
4081 ring_buffer_read_finish(struct ring_buffer_iter
*iter
)
4083 struct ring_buffer_per_cpu
*cpu_buffer
= iter
->cpu_buffer
;
4084 unsigned long flags
;
4087 * Ring buffer is disabled from recording, here's a good place
4088 * to check the integrity of the ring buffer.
4089 * Must prevent readers from trying to read, as the check
4090 * clears the HEAD page and readers require it.
4092 raw_spin_lock_irqsave(&cpu_buffer
->reader_lock
, flags
);
4093 rb_check_pages(cpu_buffer
);
4094 raw_spin_unlock_irqrestore(&cpu_buffer
->reader_lock
, flags
);
4096 atomic_dec(&cpu_buffer
->record_disabled
);
4097 atomic_dec(&cpu_buffer
->buffer
->resize_disabled
);
4100 EXPORT_SYMBOL_GPL(ring_buffer_read_finish
);
4103 * ring_buffer_read - read the next item in the ring buffer by the iterator
4104 * @iter: The ring buffer iterator
4105 * @ts: The time stamp of the event read.
4107 * This reads the next event in the ring buffer and increments the iterator.
4109 struct ring_buffer_event
*
4110 ring_buffer_read(struct ring_buffer_iter
*iter
, u64
*ts
)
4112 struct ring_buffer_event
*event
;
4113 struct ring_buffer_per_cpu
*cpu_buffer
= iter
->cpu_buffer
;
4114 unsigned long flags
;
4116 raw_spin_lock_irqsave(&cpu_buffer
->reader_lock
, flags
);
4118 event
= rb_iter_peek(iter
, ts
);
4122 if (event
->type_len
== RINGBUF_TYPE_PADDING
)
4125 rb_advance_iter(iter
);
4127 raw_spin_unlock_irqrestore(&cpu_buffer
->reader_lock
, flags
);
4131 EXPORT_SYMBOL_GPL(ring_buffer_read
);
4134 * ring_buffer_size - return the size of the ring buffer (in bytes)
4135 * @buffer: The ring buffer.
4137 unsigned long ring_buffer_size(struct ring_buffer
*buffer
, int cpu
)
4140 * Earlier, this method returned
4141 * BUF_PAGE_SIZE * buffer->nr_pages
4142 * Since the nr_pages field is now removed, we have converted this to
4143 * return the per cpu buffer value.
4145 if (!cpumask_test_cpu(cpu
, buffer
->cpumask
))
4148 return BUF_PAGE_SIZE
* buffer
->buffers
[cpu
]->nr_pages
;
4150 EXPORT_SYMBOL_GPL(ring_buffer_size
);
4153 rb_reset_cpu(struct ring_buffer_per_cpu
*cpu_buffer
)
4155 rb_head_page_deactivate(cpu_buffer
);
4157 cpu_buffer
->head_page
4158 = list_entry(cpu_buffer
->pages
, struct buffer_page
, list
);
4159 local_set(&cpu_buffer
->head_page
->write
, 0);
4160 local_set(&cpu_buffer
->head_page
->entries
, 0);
4161 local_set(&cpu_buffer
->head_page
->page
->commit
, 0);
4163 cpu_buffer
->head_page
->read
= 0;
4165 cpu_buffer
->tail_page
= cpu_buffer
->head_page
;
4166 cpu_buffer
->commit_page
= cpu_buffer
->head_page
;
4168 INIT_LIST_HEAD(&cpu_buffer
->reader_page
->list
);
4169 INIT_LIST_HEAD(&cpu_buffer
->new_pages
);
4170 local_set(&cpu_buffer
->reader_page
->write
, 0);
4171 local_set(&cpu_buffer
->reader_page
->entries
, 0);
4172 local_set(&cpu_buffer
->reader_page
->page
->commit
, 0);
4173 cpu_buffer
->reader_page
->read
= 0;
4175 local_set(&cpu_buffer
->entries_bytes
, 0);
4176 local_set(&cpu_buffer
->overrun
, 0);
4177 local_set(&cpu_buffer
->commit_overrun
, 0);
4178 local_set(&cpu_buffer
->dropped_events
, 0);
4179 local_set(&cpu_buffer
->entries
, 0);
4180 local_set(&cpu_buffer
->committing
, 0);
4181 local_set(&cpu_buffer
->commits
, 0);
4182 cpu_buffer
->read
= 0;
4183 cpu_buffer
->read_bytes
= 0;
4185 cpu_buffer
->write_stamp
= 0;
4186 cpu_buffer
->read_stamp
= 0;
4188 cpu_buffer
->lost_events
= 0;
4189 cpu_buffer
->last_overrun
= 0;
4191 rb_head_page_activate(cpu_buffer
);
4195 * ring_buffer_reset_cpu - reset a ring buffer per CPU buffer
4196 * @buffer: The ring buffer to reset a per cpu buffer of
4197 * @cpu: The CPU buffer to be reset
4199 void ring_buffer_reset_cpu(struct ring_buffer
*buffer
, int cpu
)
4201 struct ring_buffer_per_cpu
*cpu_buffer
= buffer
->buffers
[cpu
];
4202 unsigned long flags
;
4204 if (!cpumask_test_cpu(cpu
, buffer
->cpumask
))
4207 atomic_inc(&buffer
->resize_disabled
);
4208 atomic_inc(&cpu_buffer
->record_disabled
);
4210 /* Make sure all commits have finished */
4211 synchronize_sched();
4213 raw_spin_lock_irqsave(&cpu_buffer
->reader_lock
, flags
);
4215 if (RB_WARN_ON(cpu_buffer
, local_read(&cpu_buffer
->committing
)))
4218 arch_spin_lock(&cpu_buffer
->lock
);
4220 rb_reset_cpu(cpu_buffer
);
4222 arch_spin_unlock(&cpu_buffer
->lock
);
4225 raw_spin_unlock_irqrestore(&cpu_buffer
->reader_lock
, flags
);
4227 atomic_dec(&cpu_buffer
->record_disabled
);
4228 atomic_dec(&buffer
->resize_disabled
);
4230 EXPORT_SYMBOL_GPL(ring_buffer_reset_cpu
);
4233 * ring_buffer_reset - reset a ring buffer
4234 * @buffer: The ring buffer to reset all cpu buffers
4236 void ring_buffer_reset(struct ring_buffer
*buffer
)
4240 for_each_buffer_cpu(buffer
, cpu
)
4241 ring_buffer_reset_cpu(buffer
, cpu
);
4243 EXPORT_SYMBOL_GPL(ring_buffer_reset
);
4246 * rind_buffer_empty - is the ring buffer empty?
4247 * @buffer: The ring buffer to test
4249 bool ring_buffer_empty(struct ring_buffer
*buffer
)
4251 struct ring_buffer_per_cpu
*cpu_buffer
;
4252 unsigned long flags
;
4257 /* yes this is racy, but if you don't like the race, lock the buffer */
4258 for_each_buffer_cpu(buffer
, cpu
) {
4259 cpu_buffer
= buffer
->buffers
[cpu
];
4260 local_irq_save(flags
);
4261 dolock
= rb_reader_lock(cpu_buffer
);
4262 ret
= rb_per_cpu_empty(cpu_buffer
);
4263 rb_reader_unlock(cpu_buffer
, dolock
);
4264 local_irq_restore(flags
);
4272 EXPORT_SYMBOL_GPL(ring_buffer_empty
);
4275 * ring_buffer_empty_cpu - is a cpu buffer of a ring buffer empty?
4276 * @buffer: The ring buffer
4277 * @cpu: The CPU buffer to test
4279 bool ring_buffer_empty_cpu(struct ring_buffer
*buffer
, int cpu
)
4281 struct ring_buffer_per_cpu
*cpu_buffer
;
4282 unsigned long flags
;
4286 if (!cpumask_test_cpu(cpu
, buffer
->cpumask
))
4289 cpu_buffer
= buffer
->buffers
[cpu
];
4290 local_irq_save(flags
);
4291 dolock
= rb_reader_lock(cpu_buffer
);
4292 ret
= rb_per_cpu_empty(cpu_buffer
);
4293 rb_reader_unlock(cpu_buffer
, dolock
);
4294 local_irq_restore(flags
);
4298 EXPORT_SYMBOL_GPL(ring_buffer_empty_cpu
);
4300 #ifdef CONFIG_RING_BUFFER_ALLOW_SWAP
4302 * ring_buffer_swap_cpu - swap a CPU buffer between two ring buffers
4303 * @buffer_a: One buffer to swap with
4304 * @buffer_b: The other buffer to swap with
4306 * This function is useful for tracers that want to take a "snapshot"
4307 * of a CPU buffer and has another back up buffer lying around.
4308 * it is expected that the tracer handles the cpu buffer not being
4309 * used at the moment.
4311 int ring_buffer_swap_cpu(struct ring_buffer
*buffer_a
,
4312 struct ring_buffer
*buffer_b
, int cpu
)
4314 struct ring_buffer_per_cpu
*cpu_buffer_a
;
4315 struct ring_buffer_per_cpu
*cpu_buffer_b
;
4318 if (!cpumask_test_cpu(cpu
, buffer_a
->cpumask
) ||
4319 !cpumask_test_cpu(cpu
, buffer_b
->cpumask
))
4322 cpu_buffer_a
= buffer_a
->buffers
[cpu
];
4323 cpu_buffer_b
= buffer_b
->buffers
[cpu
];
4325 /* At least make sure the two buffers are somewhat the same */
4326 if (cpu_buffer_a
->nr_pages
!= cpu_buffer_b
->nr_pages
)
4331 if (atomic_read(&buffer_a
->record_disabled
))
4334 if (atomic_read(&buffer_b
->record_disabled
))
4337 if (atomic_read(&cpu_buffer_a
->record_disabled
))
4340 if (atomic_read(&cpu_buffer_b
->record_disabled
))
4344 * We can't do a synchronize_sched here because this
4345 * function can be called in atomic context.
4346 * Normally this will be called from the same CPU as cpu.
4347 * If not it's up to the caller to protect this.
4349 atomic_inc(&cpu_buffer_a
->record_disabled
);
4350 atomic_inc(&cpu_buffer_b
->record_disabled
);
4353 if (local_read(&cpu_buffer_a
->committing
))
4355 if (local_read(&cpu_buffer_b
->committing
))
4358 buffer_a
->buffers
[cpu
] = cpu_buffer_b
;
4359 buffer_b
->buffers
[cpu
] = cpu_buffer_a
;
4361 cpu_buffer_b
->buffer
= buffer_a
;
4362 cpu_buffer_a
->buffer
= buffer_b
;
4367 atomic_dec(&cpu_buffer_a
->record_disabled
);
4368 atomic_dec(&cpu_buffer_b
->record_disabled
);
4372 EXPORT_SYMBOL_GPL(ring_buffer_swap_cpu
);
4373 #endif /* CONFIG_RING_BUFFER_ALLOW_SWAP */
4376 * ring_buffer_alloc_read_page - allocate a page to read from buffer
4377 * @buffer: the buffer to allocate for.
4378 * @cpu: the cpu buffer to allocate.
4380 * This function is used in conjunction with ring_buffer_read_page.
4381 * When reading a full page from the ring buffer, these functions
4382 * can be used to speed up the process. The calling function should
4383 * allocate a few pages first with this function. Then when it
4384 * needs to get pages from the ring buffer, it passes the result
4385 * of this function into ring_buffer_read_page, which will swap
4386 * the page that was allocated, with the read page of the buffer.
4389 * The page allocated, or NULL on error.
4391 void *ring_buffer_alloc_read_page(struct ring_buffer
*buffer
, int cpu
)
4393 struct ring_buffer_per_cpu
*cpu_buffer
= buffer
->buffers
[cpu
];
4394 struct buffer_data_page
*bpage
= NULL
;
4395 unsigned long flags
;
4398 local_irq_save(flags
);
4399 arch_spin_lock(&cpu_buffer
->lock
);
4401 if (cpu_buffer
->free_page
) {
4402 bpage
= cpu_buffer
->free_page
;
4403 cpu_buffer
->free_page
= NULL
;
4406 arch_spin_unlock(&cpu_buffer
->lock
);
4407 local_irq_restore(flags
);
4412 page
= alloc_pages_node(cpu_to_node(cpu
),
4413 GFP_KERNEL
| __GFP_NORETRY
, 0);
4417 bpage
= page_address(page
);
4420 rb_init_page(bpage
);
4424 EXPORT_SYMBOL_GPL(ring_buffer_alloc_read_page
);
4427 * ring_buffer_free_read_page - free an allocated read page
4428 * @buffer: the buffer the page was allocate for
4429 * @cpu: the cpu buffer the page came from
4430 * @data: the page to free
4432 * Free a page allocated from ring_buffer_alloc_read_page.
4434 void ring_buffer_free_read_page(struct ring_buffer
*buffer
, int cpu
, void *data
)
4436 struct ring_buffer_per_cpu
*cpu_buffer
= buffer
->buffers
[cpu
];
4437 struct buffer_data_page
*bpage
= data
;
4438 unsigned long flags
;
4440 local_irq_save(flags
);
4441 arch_spin_lock(&cpu_buffer
->lock
);
4443 if (!cpu_buffer
->free_page
) {
4444 cpu_buffer
->free_page
= bpage
;
4448 arch_spin_unlock(&cpu_buffer
->lock
);
4449 local_irq_restore(flags
);
4451 free_page((unsigned long)bpage
);
4453 EXPORT_SYMBOL_GPL(ring_buffer_free_read_page
);
4456 * ring_buffer_read_page - extract a page from the ring buffer
4457 * @buffer: buffer to extract from
4458 * @data_page: the page to use allocated from ring_buffer_alloc_read_page
4459 * @len: amount to extract
4460 * @cpu: the cpu of the buffer to extract
4461 * @full: should the extraction only happen when the page is full.
4463 * This function will pull out a page from the ring buffer and consume it.
4464 * @data_page must be the address of the variable that was returned
4465 * from ring_buffer_alloc_read_page. This is because the page might be used
4466 * to swap with a page in the ring buffer.
4469 * rpage = ring_buffer_alloc_read_page(buffer, cpu);
4472 * ret = ring_buffer_read_page(buffer, &rpage, len, cpu, 0);
4474 * process_page(rpage, ret);
4476 * When @full is set, the function will not return true unless
4477 * the writer is off the reader page.
4479 * Note: it is up to the calling functions to handle sleeps and wakeups.
4480 * The ring buffer can be used anywhere in the kernel and can not
4481 * blindly call wake_up. The layer that uses the ring buffer must be
4482 * responsible for that.
4485 * >=0 if data has been transferred, returns the offset of consumed data.
4486 * <0 if no data has been transferred.
4488 int ring_buffer_read_page(struct ring_buffer
*buffer
,
4489 void **data_page
, size_t len
, int cpu
, int full
)
4491 struct ring_buffer_per_cpu
*cpu_buffer
= buffer
->buffers
[cpu
];
4492 struct ring_buffer_event
*event
;
4493 struct buffer_data_page
*bpage
;
4494 struct buffer_page
*reader
;
4495 unsigned long missed_events
;
4496 unsigned long flags
;
4497 unsigned int commit
;
4502 if (!cpumask_test_cpu(cpu
, buffer
->cpumask
))
4506 * If len is not big enough to hold the page header, then
4507 * we can not copy anything.
4509 if (len
<= BUF_PAGE_HDR_SIZE
)
4512 len
-= BUF_PAGE_HDR_SIZE
;
4521 raw_spin_lock_irqsave(&cpu_buffer
->reader_lock
, flags
);
4523 reader
= rb_get_reader_page(cpu_buffer
);
4527 event
= rb_reader_event(cpu_buffer
);
4529 read
= reader
->read
;
4530 commit
= rb_page_commit(reader
);
4532 /* Check if any events were dropped */
4533 missed_events
= cpu_buffer
->lost_events
;
4536 * If this page has been partially read or
4537 * if len is not big enough to read the rest of the page or
4538 * a writer is still on the page, then
4539 * we must copy the data from the page to the buffer.
4540 * Otherwise, we can simply swap the page with the one passed in.
4542 if (read
|| (len
< (commit
- read
)) ||
4543 cpu_buffer
->reader_page
== cpu_buffer
->commit_page
) {
4544 struct buffer_data_page
*rpage
= cpu_buffer
->reader_page
->page
;
4545 unsigned int rpos
= read
;
4546 unsigned int pos
= 0;
4552 if (len
> (commit
- read
))
4553 len
= (commit
- read
);
4555 /* Always keep the time extend and data together */
4556 size
= rb_event_ts_length(event
);
4561 /* save the current timestamp, since the user will need it */
4562 save_timestamp
= cpu_buffer
->read_stamp
;
4564 /* Need to copy one event at a time */
4566 /* We need the size of one event, because
4567 * rb_advance_reader only advances by one event,
4568 * whereas rb_event_ts_length may include the size of
4569 * one or two events.
4570 * We have already ensured there's enough space if this
4571 * is a time extend. */
4572 size
= rb_event_length(event
);
4573 memcpy(bpage
->data
+ pos
, rpage
->data
+ rpos
, size
);
4577 rb_advance_reader(cpu_buffer
);
4578 rpos
= reader
->read
;
4584 event
= rb_reader_event(cpu_buffer
);
4585 /* Always keep the time extend and data together */
4586 size
= rb_event_ts_length(event
);
4587 } while (len
>= size
);
4590 local_set(&bpage
->commit
, pos
);
4591 bpage
->time_stamp
= save_timestamp
;
4593 /* we copied everything to the beginning */
4596 /* update the entry counter */
4597 cpu_buffer
->read
+= rb_page_entries(reader
);
4598 cpu_buffer
->read_bytes
+= BUF_PAGE_SIZE
;
4600 /* swap the pages */
4601 rb_init_page(bpage
);
4602 bpage
= reader
->page
;
4603 reader
->page
= *data_page
;
4604 local_set(&reader
->write
, 0);
4605 local_set(&reader
->entries
, 0);
4610 * Use the real_end for the data size,
4611 * This gives us a chance to store the lost events
4614 if (reader
->real_end
)
4615 local_set(&bpage
->commit
, reader
->real_end
);
4619 cpu_buffer
->lost_events
= 0;
4621 commit
= local_read(&bpage
->commit
);
4623 * Set a flag in the commit field if we lost events
4625 if (missed_events
) {
4626 /* If there is room at the end of the page to save the
4627 * missed events, then record it there.
4629 if (BUF_PAGE_SIZE
- commit
>= sizeof(missed_events
)) {
4630 memcpy(&bpage
->data
[commit
], &missed_events
,
4631 sizeof(missed_events
));
4632 local_add(RB_MISSED_STORED
, &bpage
->commit
);
4633 commit
+= sizeof(missed_events
);
4635 local_add(RB_MISSED_EVENTS
, &bpage
->commit
);
4639 * This page may be off to user land. Zero it out here.
4641 if (commit
< BUF_PAGE_SIZE
)
4642 memset(&bpage
->data
[commit
], 0, BUF_PAGE_SIZE
- commit
);
4645 raw_spin_unlock_irqrestore(&cpu_buffer
->reader_lock
, flags
);
4650 EXPORT_SYMBOL_GPL(ring_buffer_read_page
);
4653 * We only allocate new buffers, never free them if the CPU goes down.
4654 * If we were to free the buffer, then the user would lose any trace that was in
4657 int trace_rb_cpu_prepare(unsigned int cpu
, struct hlist_node
*node
)
4659 struct ring_buffer
*buffer
;
4662 unsigned long nr_pages
;
4664 buffer
= container_of(node
, struct ring_buffer
, node
);
4665 if (cpumask_test_cpu(cpu
, buffer
->cpumask
))
4670 /* check if all cpu sizes are same */
4671 for_each_buffer_cpu(buffer
, cpu_i
) {
4672 /* fill in the size from first enabled cpu */
4674 nr_pages
= buffer
->buffers
[cpu_i
]->nr_pages
;
4675 if (nr_pages
!= buffer
->buffers
[cpu_i
]->nr_pages
) {
4680 /* allocate minimum pages, user can later expand it */
4683 buffer
->buffers
[cpu
] =
4684 rb_allocate_cpu_buffer(buffer
, nr_pages
, cpu
);
4685 if (!buffer
->buffers
[cpu
]) {
4686 WARN(1, "failed to allocate ring buffer on CPU %u\n",
4691 cpumask_set_cpu(cpu
, buffer
->cpumask
);
4695 #ifdef CONFIG_RING_BUFFER_STARTUP_TEST
4697 * This is a basic integrity check of the ring buffer.
4698 * Late in the boot cycle this test will run when configured in.
4699 * It will kick off a thread per CPU that will go into a loop
4700 * writing to the per cpu ring buffer various sizes of data.
4701 * Some of the data will be large items, some small.
4703 * Another thread is created that goes into a spin, sending out
4704 * IPIs to the other CPUs to also write into the ring buffer.
4705 * this is to test the nesting ability of the buffer.
4707 * Basic stats are recorded and reported. If something in the
4708 * ring buffer should happen that's not expected, a big warning
4709 * is displayed and all ring buffers are disabled.
4711 static struct task_struct
*rb_threads
[NR_CPUS
] __initdata
;
4713 struct rb_test_data
{
4714 struct ring_buffer
*buffer
;
4715 unsigned long events
;
4716 unsigned long bytes_written
;
4717 unsigned long bytes_alloc
;
4718 unsigned long bytes_dropped
;
4719 unsigned long events_nested
;
4720 unsigned long bytes_written_nested
;
4721 unsigned long bytes_alloc_nested
;
4722 unsigned long bytes_dropped_nested
;
4723 int min_size_nested
;
4724 int max_size_nested
;
4731 static struct rb_test_data rb_data
[NR_CPUS
] __initdata
;
4734 #define RB_TEST_BUFFER_SIZE 1048576
4736 static char rb_string
[] __initdata
=
4737 "abcdefghijklmnopqrstuvwxyz1234567890!@#$%^&*()?+\\"
4738 "?+|:';\",.<>/?abcdefghijklmnopqrstuvwxyz1234567890"
4739 "!@#$%^&*()?+\\?+|:';\",.<>/?abcdefghijklmnopqrstuv";
4741 static bool rb_test_started __initdata
;
4748 static __init
int rb_write_something(struct rb_test_data
*data
, bool nested
)
4750 struct ring_buffer_event
*event
;
4751 struct rb_item
*item
;
4758 /* Have nested writes different that what is written */
4759 cnt
= data
->cnt
+ (nested
? 27 : 0);
4761 /* Multiply cnt by ~e, to make some unique increment */
4762 size
= (data
->cnt
* 68 / 25) % (sizeof(rb_string
) - 1);
4764 len
= size
+ sizeof(struct rb_item
);
4766 started
= rb_test_started
;
4767 /* read rb_test_started before checking buffer enabled */
4770 event
= ring_buffer_lock_reserve(data
->buffer
, len
);
4772 /* Ignore dropped events before test starts. */
4775 data
->bytes_dropped
+= len
;
4777 data
->bytes_dropped_nested
+= len
;
4782 event_len
= ring_buffer_event_length(event
);
4784 if (RB_WARN_ON(data
->buffer
, event_len
< len
))
4787 item
= ring_buffer_event_data(event
);
4789 memcpy(item
->str
, rb_string
, size
);
4792 data
->bytes_alloc_nested
+= event_len
;
4793 data
->bytes_written_nested
+= len
;
4794 data
->events_nested
++;
4795 if (!data
->min_size_nested
|| len
< data
->min_size_nested
)
4796 data
->min_size_nested
= len
;
4797 if (len
> data
->max_size_nested
)
4798 data
->max_size_nested
= len
;
4800 data
->bytes_alloc
+= event_len
;
4801 data
->bytes_written
+= len
;
4803 if (!data
->min_size
|| len
< data
->min_size
)
4804 data
->max_size
= len
;
4805 if (len
> data
->max_size
)
4806 data
->max_size
= len
;
4810 ring_buffer_unlock_commit(data
->buffer
, event
);
4815 static __init
int rb_test(void *arg
)
4817 struct rb_test_data
*data
= arg
;
4819 while (!kthread_should_stop()) {
4820 rb_write_something(data
, false);
4823 set_current_state(TASK_INTERRUPTIBLE
);
4824 /* Now sleep between a min of 100-300us and a max of 1ms */
4825 usleep_range(((data
->cnt
% 3) + 1) * 100, 1000);
4831 static __init
void rb_ipi(void *ignore
)
4833 struct rb_test_data
*data
;
4834 int cpu
= smp_processor_id();
4836 data
= &rb_data
[cpu
];
4837 rb_write_something(data
, true);
4840 static __init
int rb_hammer_test(void *arg
)
4842 while (!kthread_should_stop()) {
4844 /* Send an IPI to all cpus to write data! */
4845 smp_call_function(rb_ipi
, NULL
, 1);
4846 /* No sleep, but for non preempt, let others run */
4853 static __init
int test_ringbuffer(void)
4855 struct task_struct
*rb_hammer
;
4856 struct ring_buffer
*buffer
;
4860 pr_info("Running ring buffer tests...\n");
4862 buffer
= ring_buffer_alloc(RB_TEST_BUFFER_SIZE
, RB_FL_OVERWRITE
);
4863 if (WARN_ON(!buffer
))
4866 /* Disable buffer so that threads can't write to it yet */
4867 ring_buffer_record_off(buffer
);
4869 for_each_online_cpu(cpu
) {
4870 rb_data
[cpu
].buffer
= buffer
;
4871 rb_data
[cpu
].cpu
= cpu
;
4872 rb_data
[cpu
].cnt
= cpu
;
4873 rb_threads
[cpu
] = kthread_create(rb_test
, &rb_data
[cpu
],
4874 "rbtester/%d", cpu
);
4875 if (WARN_ON(IS_ERR(rb_threads
[cpu
]))) {
4876 pr_cont("FAILED\n");
4877 ret
= PTR_ERR(rb_threads
[cpu
]);
4881 kthread_bind(rb_threads
[cpu
], cpu
);
4882 wake_up_process(rb_threads
[cpu
]);
4885 /* Now create the rb hammer! */
4886 rb_hammer
= kthread_run(rb_hammer_test
, NULL
, "rbhammer");
4887 if (WARN_ON(IS_ERR(rb_hammer
))) {
4888 pr_cont("FAILED\n");
4889 ret
= PTR_ERR(rb_hammer
);
4893 ring_buffer_record_on(buffer
);
4895 * Show buffer is enabled before setting rb_test_started.
4896 * Yes there's a small race window where events could be
4897 * dropped and the thread wont catch it. But when a ring
4898 * buffer gets enabled, there will always be some kind of
4899 * delay before other CPUs see it. Thus, we don't care about
4900 * those dropped events. We care about events dropped after
4901 * the threads see that the buffer is active.
4904 rb_test_started
= true;
4906 set_current_state(TASK_INTERRUPTIBLE
);
4907 /* Just run for 10 seconds */;
4908 schedule_timeout(10 * HZ
);
4910 kthread_stop(rb_hammer
);
4913 for_each_online_cpu(cpu
) {
4914 if (!rb_threads
[cpu
])
4916 kthread_stop(rb_threads
[cpu
]);
4919 ring_buffer_free(buffer
);
4924 pr_info("finished\n");
4925 for_each_online_cpu(cpu
) {
4926 struct ring_buffer_event
*event
;
4927 struct rb_test_data
*data
= &rb_data
[cpu
];
4928 struct rb_item
*item
;
4929 unsigned long total_events
;
4930 unsigned long total_dropped
;
4931 unsigned long total_written
;
4932 unsigned long total_alloc
;
4933 unsigned long total_read
= 0;
4934 unsigned long total_size
= 0;
4935 unsigned long total_len
= 0;
4936 unsigned long total_lost
= 0;
4939 int small_event_size
;
4943 total_events
= data
->events
+ data
->events_nested
;
4944 total_written
= data
->bytes_written
+ data
->bytes_written_nested
;
4945 total_alloc
= data
->bytes_alloc
+ data
->bytes_alloc_nested
;
4946 total_dropped
= data
->bytes_dropped
+ data
->bytes_dropped_nested
;
4948 big_event_size
= data
->max_size
+ data
->max_size_nested
;
4949 small_event_size
= data
->min_size
+ data
->min_size_nested
;
4951 pr_info("CPU %d:\n", cpu
);
4952 pr_info(" events: %ld\n", total_events
);
4953 pr_info(" dropped bytes: %ld\n", total_dropped
);
4954 pr_info(" alloced bytes: %ld\n", total_alloc
);
4955 pr_info(" written bytes: %ld\n", total_written
);
4956 pr_info(" biggest event: %d\n", big_event_size
);
4957 pr_info(" smallest event: %d\n", small_event_size
);
4959 if (RB_WARN_ON(buffer
, total_dropped
))
4964 while ((event
= ring_buffer_consume(buffer
, cpu
, NULL
, &lost
))) {
4966 item
= ring_buffer_event_data(event
);
4967 total_len
+= ring_buffer_event_length(event
);
4968 total_size
+= item
->size
+ sizeof(struct rb_item
);
4969 if (memcmp(&item
->str
[0], rb_string
, item
->size
) != 0) {
4970 pr_info("FAILED!\n");
4971 pr_info("buffer had: %.*s\n", item
->size
, item
->str
);
4972 pr_info("expected: %.*s\n", item
->size
, rb_string
);
4973 RB_WARN_ON(buffer
, 1);
4984 pr_info(" read events: %ld\n", total_read
);
4985 pr_info(" lost events: %ld\n", total_lost
);
4986 pr_info(" total events: %ld\n", total_lost
+ total_read
);
4987 pr_info(" recorded len bytes: %ld\n", total_len
);
4988 pr_info(" recorded size bytes: %ld\n", total_size
);
4990 pr_info(" With dropped events, record len and size may not match\n"
4991 " alloced and written from above\n");
4993 if (RB_WARN_ON(buffer
, total_len
!= total_alloc
||
4994 total_size
!= total_written
))
4997 if (RB_WARN_ON(buffer
, total_lost
+ total_read
!= total_events
))
5003 pr_info("Ring buffer PASSED!\n");
5005 ring_buffer_free(buffer
);
5009 late_initcall(test_ringbuffer
);
5010 #endif /* CONFIG_RING_BUFFER_STARTUP_TEST */