kernel_5.10 no rt

hc

2023-12-11 6778948f9de86c3cfaf36725a7c87dcff9ba247f

 kernel/tools/perf/util/mmap.c
..
..
@@ -1,130 +1,44 @@
1
+// SPDX-License-Identifier: GPL-2.0-only
1
2
 /*
2
3
  * Copyright (C) 2011-2017, Red Hat Inc, Arnaldo Carvalho de Melo <acme@redhat.com>
3
4
  *
4
5
  * Parts came from evlist.c builtin-{top,stat,record}.c, see those files for further
5
6
  * copyright notes.
6
- *
7
- * Released under the GPL v2. (and only v2, not any later version)
8
7
  */
9
8
 
10
9
 #include <sys/mman.h>
11
10
 #include <inttypes.h>
12
11
 #include <asm/bug.h>
12
+#include <linux/zalloc.h>
13
+#include <stdlib.h>
14
+#include <string.h>
15
+#include <unistd.h> // sysconf()
16
+#include <perf/mmap.h>
17
+#ifdef HAVE_LIBNUMA_SUPPORT
18
+#include <numaif.h>
19
+#endif
20
+#include "cpumap.h"
13
21
 #include "debug.h"
14
22
 #include "event.h"
15
23
 #include "mmap.h"
16
-#include "util.h" /* page_size */
24
+#include "../perf.h"
25
+#include <internal/lib.h> /* page_size */
26
+#include <linux/bitmap.h>
17
27
 
18
-size_t perf_mmap__mmap_len(struct perf_mmap *map)
28
+#define MASK_SIZE 1023
29
+void mmap_cpu_mask__scnprintf(struct mmap_cpu_mask *mask, const char *tag)
19
30
 {
20
-	return map->mask + 1 + page_size;
31
+	char buf[MASK_SIZE + 1];
32
+	size_t len;
33
+
34
+	len = bitmap_scnprintf(mask->bits, mask->nbits, buf, MASK_SIZE);
35
+	buf[len] = '\0';
36
+	pr_debug("%p: %s mask[%zd]: %s\n", mask, tag, mask->nbits, buf);
21
37
 }
22
38
 
23
-/* When check_messup is true, 'end' must points to a good entry */
24
-static union perf_event *perf_mmap__read(struct perf_mmap *map,
25
-					 u64 *startp, u64 end)
39
+size_t mmap__mmap_len(struct mmap *map)
26
40
 {
27
-	unsigned char *data = map->base + page_size;
28
-	union perf_event *event = NULL;
29
-	int diff = end - *startp;
30
-
31
-	if (diff >= (int)sizeof(event->header)) {
32
-		size_t size;
33
-
34
-		event = (union perf_event *)&data[*startp & map->mask];
35
-		size = event->header.size;
36
-
37
-		if (size < sizeof(event->header) || diff < (int)size)
38
-			return NULL;
39
-
40
-		/*
41
-		 * Event straddles the mmap boundary -- header should always
42
-		 * be inside due to u64 alignment of output.
43
-		 */
44
-		if ((*startp & map->mask) + size != ((*startp + size) & map->mask)) {
45
-			unsigned int offset = *startp;
46
-			unsigned int len = min(sizeof(*event), size), cpy;
47
-			void *dst = map->event_copy;
48
-
49
-			do {
50
-				cpy = min(map->mask + 1 - (offset & map->mask), len);
51
-				memcpy(dst, &data[offset & map->mask], cpy);
52
-				offset += cpy;
53
-				dst += cpy;
54
-				len -= cpy;
55
-			} while (len);
56
-
57
-			event = (union perf_event *)map->event_copy;
58
-		}
59
-
60
-		*startp += size;
61
-	}
62
-
63
-	return event;
64
-}
65
-
66
-/*
67
- * Read event from ring buffer one by one.
68
- * Return one event for each call.
69
- *
70
- * Usage:
71
- * perf_mmap__read_init()
72
- * while(event = perf_mmap__read_event()) {
73
- *	//process the event
74
- *	perf_mmap__consume()
75
- * }
76
- * perf_mmap__read_done()
77
- */
78
-union perf_event *perf_mmap__read_event(struct perf_mmap *map)
79
-{
80
-	union perf_event *event;
81
-
82
-	/*
83
-	 * Check if event was unmapped due to a POLLHUP/POLLERR.
84
-	 */
85
-	if (!refcount_read(&map->refcnt))
86
-		return NULL;
87
-
88
-	/* non-overwirte doesn't pause the ringbuffer */
89
-	if (!map->overwrite)
90
-		map->end = perf_mmap__read_head(map);
91
-
92
-	event = perf_mmap__read(map, &map->start, map->end);
93
-
94
-	if (!map->overwrite)
95
-		map->prev = map->start;
96
-
97
-	return event;
98
-}
99
-
100
-static bool perf_mmap__empty(struct perf_mmap *map)
101
-{
102
-	return perf_mmap__read_head(map) == map->prev && !map->auxtrace_mmap.base;
103
-}
104
-
105
-void perf_mmap__get(struct perf_mmap *map)
106
-{
107
-	refcount_inc(&map->refcnt);
108
-}
109
-
110
-void perf_mmap__put(struct perf_mmap *map)
111
-{
112
-	BUG_ON(map->base && refcount_read(&map->refcnt) == 0);
113
-
114
-	if (refcount_dec_and_test(&map->refcnt))
115
-		perf_mmap__munmap(map);
116
-}
117
-
118
-void perf_mmap__consume(struct perf_mmap *map)
119
-{
120
-	if (!map->overwrite) {
121
-		u64 old = map->prev;
122
-
123
-		perf_mmap__write_tail(map, old);
124
-	}
125
-
126
-	if (refcount_read(&map->refcnt) == 1 && perf_mmap__empty(map))
127
-		perf_mmap__put(map);
41
+	return perf_mmap__mmap_len(&map->core);
128
42
 }
129
43
 
130
44
 int __weak auxtrace_mmap__mmap(struct auxtrace_mmap *mm __maybe_unused,
..
..
@@ -147,193 +61,292 @@
147
61
 }
148
62
 
149
63
 void __weak auxtrace_mmap_params__set_idx(struct auxtrace_mmap_params *mp __maybe_unused,
150
-					  struct perf_evlist *evlist __maybe_unused,
64
+					  struct evlist *evlist __maybe_unused,
151
65
 					  int idx __maybe_unused,
152
66
 					  bool per_cpu __maybe_unused)
153
67
 {
154
68
 }
155
69
 
156
-void perf_mmap__munmap(struct perf_mmap *map)
70
+#ifdef HAVE_AIO_SUPPORT
71
+static int perf_mmap__aio_enabled(struct mmap *map)
157
72
 {
158
-	if (map->base != NULL) {
159
-		munmap(map->base, perf_mmap__mmap_len(map));
160
-		map->base = NULL;
161
-		map->fd = -1;
162
-		refcount_set(&map->refcnt, 0);
73
+	return map->aio.nr_cblocks > 0;
74
+}
75
+
76
+#ifdef HAVE_LIBNUMA_SUPPORT
77
+static int perf_mmap__aio_alloc(struct mmap *map, int idx)
78
+{
79
+	map->aio.data[idx] = mmap(NULL, mmap__mmap_len(map), PROT_READ|PROT_WRITE,
80
+				  MAP_PRIVATE|MAP_ANONYMOUS, 0, 0);
81
+	if (map->aio.data[idx] == MAP_FAILED) {
82
+		map->aio.data[idx] = NULL;
83
+		return -1;
84
+	}
85
+
86
+	return 0;
87
+}
88
+
89
+static void perf_mmap__aio_free(struct mmap *map, int idx)
90
+{
91
+	if (map->aio.data[idx]) {
92
+		munmap(map->aio.data[idx], mmap__mmap_len(map));
93
+		map->aio.data[idx] = NULL;
94
+	}
95
+}
96
+
97
+static int perf_mmap__aio_bind(struct mmap *map, int idx, int cpu, int affinity)
98
+{
99
+	void *data;
100
+	size_t mmap_len;
101
+	unsigned long *node_mask;
102
+	unsigned long node_index;
103
+	int err = 0;
104
+
105
+	if (affinity != PERF_AFFINITY_SYS && cpu__max_node() > 1) {
106
+		data = map->aio.data[idx];
107
+		mmap_len = mmap__mmap_len(map);
108
+		node_index = cpu__get_node(cpu);
109
+		node_mask = bitmap_alloc(node_index + 1);
110
+		if (!node_mask) {
111
+			pr_err("Failed to allocate node mask for mbind: error %m\n");
112
+			return -1;
113
+		}
114
+		set_bit(node_index, node_mask);
115
+		if (mbind(data, mmap_len, MPOL_BIND, node_mask, node_index + 1 + 1, 0)) {
116
+			pr_err("Failed to bind [%p-%p] AIO buffer to node %lu: error %m\n",
117
+				data, data + mmap_len, node_index);
118
+			err = -1;
119
+		}
120
+		bitmap_free(node_mask);
121
+	}
122
+
123
+	return err;
124
+}
125
+#else /* !HAVE_LIBNUMA_SUPPORT */
126
+static int perf_mmap__aio_alloc(struct mmap *map, int idx)
127
+{
128
+	map->aio.data[idx] = malloc(mmap__mmap_len(map));
129
+	if (map->aio.data[idx] == NULL)
130
+		return -1;
131
+
132
+	return 0;
133
+}
134
+
135
+static void perf_mmap__aio_free(struct mmap *map, int idx)
136
+{
137
+	zfree(&(map->aio.data[idx]));
138
+}
139
+
140
+static int perf_mmap__aio_bind(struct mmap *map __maybe_unused, int idx __maybe_unused,
141
+		int cpu __maybe_unused, int affinity __maybe_unused)
142
+{
143
+	return 0;
144
+}
145
+#endif
146
+
147
+static int perf_mmap__aio_mmap(struct mmap *map, struct mmap_params *mp)
148
+{
149
+	int delta_max, i, prio, ret;
150
+
151
+	map->aio.nr_cblocks = mp->nr_cblocks;
152
+	if (map->aio.nr_cblocks) {
153
+		map->aio.aiocb = calloc(map->aio.nr_cblocks, sizeof(struct aiocb *));
154
+		if (!map->aio.aiocb) {
155
+			pr_debug2("failed to allocate aiocb for data buffer, error %m\n");
156
+			return -1;
157
+		}
158
+		map->aio.cblocks = calloc(map->aio.nr_cblocks, sizeof(struct aiocb));
159
+		if (!map->aio.cblocks) {
160
+			pr_debug2("failed to allocate cblocks for data buffer, error %m\n");
161
+			return -1;
162
+		}
163
+		map->aio.data = calloc(map->aio.nr_cblocks, sizeof(void *));
164
+		if (!map->aio.data) {
165
+			pr_debug2("failed to allocate data buffer, error %m\n");
166
+			return -1;
167
+		}
168
+		delta_max = sysconf(_SC_AIO_PRIO_DELTA_MAX);
169
+		for (i = 0; i < map->aio.nr_cblocks; ++i) {
170
+			ret = perf_mmap__aio_alloc(map, i);
171
+			if (ret == -1) {
172
+				pr_debug2("failed to allocate data buffer area, error %m");
173
+				return -1;
174
+			}
175
+			ret = perf_mmap__aio_bind(map, i, map->core.cpu, mp->affinity);
176
+			if (ret == -1)
177
+				return -1;
178
+			/*
179
+			 * Use cblock.aio_fildes value different from -1
180
+			 * to denote started aio write operation on the
181
+			 * cblock so it requires explicit record__aio_sync()
182
+			 * call prior the cblock may be reused again.
183
+			 */
184
+			map->aio.cblocks[i].aio_fildes = -1;
185
+			/*
186
+			 * Allocate cblocks with priority delta to have
187
+			 * faster aio write system calls because queued requests
188
+			 * are kept in separate per-prio queues and adding
189
+			 * a new request will iterate thru shorter per-prio
190
+			 * list. Blocks with numbers higher than
191
+			 *  _SC_AIO_PRIO_DELTA_MAX go with priority 0.
192
+			 */
193
+			prio = delta_max - i;
194
+			map->aio.cblocks[i].aio_reqprio = prio >= 0 ? prio : 0;
195
+		}
196
+	}
197
+
198
+	return 0;
199
+}
200
+
201
+static void perf_mmap__aio_munmap(struct mmap *map)
202
+{
203
+	int i;
204
+
205
+	for (i = 0; i < map->aio.nr_cblocks; ++i)
206
+		perf_mmap__aio_free(map, i);
207
+	if (map->aio.data)
208
+		zfree(&map->aio.data);
209
+	zfree(&map->aio.cblocks);
210
+	zfree(&map->aio.aiocb);
211
+}
212
+#else /* !HAVE_AIO_SUPPORT */
213
+static int perf_mmap__aio_enabled(struct mmap *map __maybe_unused)
214
+{
215
+	return 0;
216
+}
217
+
218
+static int perf_mmap__aio_mmap(struct mmap *map __maybe_unused,
219
+			       struct mmap_params *mp __maybe_unused)
220
+{
221
+	return 0;
222
+}
223
+
224
+static void perf_mmap__aio_munmap(struct mmap *map __maybe_unused)
225
+{
226
+}
227
+#endif
228
+
229
+void mmap__munmap(struct mmap *map)
230
+{
231
+	bitmap_free(map->affinity_mask.bits);
232
+
233
+	perf_mmap__aio_munmap(map);
234
+	if (map->data != NULL) {
235
+		munmap(map->data, mmap__mmap_len(map));
236
+		map->data = NULL;
163
237
 	}
164
238
 	auxtrace_mmap__munmap(&map->auxtrace_mmap);
165
239
 }
166
240
 
167
-int perf_mmap__mmap(struct perf_mmap *map, struct mmap_params *mp, int fd, int cpu)
241
+static void build_node_mask(int node, struct mmap_cpu_mask *mask)
168
242
 {
169
-	/*
170
-	 * The last one will be done at perf_mmap__consume(), so that we
171
-	 * make sure we don't prevent tools from consuming every last event in
172
-	 * the ring buffer.
173
-	 *
174
-	 * I.e. we can get the POLLHUP meaning that the fd doesn't exist
175
-	 * anymore, but the last events for it are still in the ring buffer,
176
-	 * waiting to be consumed.
177
-	 *
178
-	 * Tools can chose to ignore this at their own discretion, but the
179
-	 * evlist layer can't just drop it when filtering events in
180
-	 * perf_evlist__filter_pollfd().
181
-	 */
182
-	refcount_set(&map->refcnt, 2);
183
-	map->prev = 0;
184
-	map->mask = mp->mask;
185
-	map->base = mmap(NULL, perf_mmap__mmap_len(map), mp->prot,
186
-			 MAP_SHARED, fd, 0);
187
-	if (map->base == MAP_FAILED) {
243
+	int c, cpu, nr_cpus;
244
+	const struct perf_cpu_map *cpu_map = NULL;
245
+
246
+	cpu_map = cpu_map__online();
247
+	if (!cpu_map)
248
+		return;
249
+
250
+	nr_cpus = perf_cpu_map__nr(cpu_map);
251
+	for (c = 0; c < nr_cpus; c++) {
252
+		cpu = cpu_map->map[c]; /* map c index to online cpu index */
253
+		if (cpu__get_node(cpu) == node)
254
+			set_bit(cpu, mask->bits);
255
+	}
256
+}
257
+
258
+static int perf_mmap__setup_affinity_mask(struct mmap *map, struct mmap_params *mp)
259
+{
260
+	map->affinity_mask.nbits = cpu__max_cpu();
261
+	map->affinity_mask.bits = bitmap_alloc(map->affinity_mask.nbits);
262
+	if (!map->affinity_mask.bits)
263
+		return -1;
264
+
265
+	if (mp->affinity == PERF_AFFINITY_NODE && cpu__max_node() > 1)
266
+		build_node_mask(cpu__get_node(map->core.cpu), &map->affinity_mask);
267
+	else if (mp->affinity == PERF_AFFINITY_CPU)
268
+		set_bit(map->core.cpu, map->affinity_mask.bits);
269
+
270
+	return 0;
271
+}
272
+
273
+int mmap__mmap(struct mmap *map, struct mmap_params *mp, int fd, int cpu)
274
+{
275
+	if (perf_mmap__mmap(&map->core, &mp->core, fd, cpu)) {
188
276
 		pr_debug2("failed to mmap perf event ring buffer, error %d\n",
189
277
 			  errno);
190
-		map->base = NULL;
191
278
 		return -1;
192
279
 	}
193
-	map->fd = fd;
194
-	map->cpu = cpu;
280
+
281
+	if (mp->affinity != PERF_AFFINITY_SYS &&
282
+		perf_mmap__setup_affinity_mask(map, mp)) {
283
+		pr_debug2("failed to alloc mmap affinity mask, error %d\n",
284
+			  errno);
285
+		return -1;
286
+	}
287
+
288
+	if (verbose == 2)
289
+		mmap_cpu_mask__scnprintf(&map->affinity_mask, "mmap");
290
+
291
+	map->core.flush = mp->flush;
292
+
293
+	map->comp_level = mp->comp_level;
294
+
295
+	if (map->comp_level && !perf_mmap__aio_enabled(map)) {
296
+		map->data = mmap(NULL, mmap__mmap_len(map), PROT_READ|PROT_WRITE,
297
+				 MAP_PRIVATE|MAP_ANONYMOUS, 0, 0);
298
+		if (map->data == MAP_FAILED) {
299
+			pr_debug2("failed to mmap data buffer, error %d\n",
300
+					errno);
301
+			map->data = NULL;
302
+			return -1;
303
+		}
304
+	}
195
305
 
196
306
 	if (auxtrace_mmap__mmap(&map->auxtrace_mmap,
197
-				&mp->auxtrace_mp, map->base, fd))
307
+				&mp->auxtrace_mp, map->core.base, fd))
198
308
 		return -1;
199
309
 
200
-	return 0;
310
+	return perf_mmap__aio_mmap(map, mp);
201
311
 }
202
312
 
203
-static int overwrite_rb_find_range(void *buf, int mask, u64 *start, u64 *end)
313
+int perf_mmap__push(struct mmap *md, void *to,
314
+		    int push(struct mmap *map, void *to, void *buf, size_t size))
204
315
 {
205
-	struct perf_event_header *pheader;
206
-	u64 evt_head = *start;
207
-	int size = mask + 1;
208
-
209
-	pr_debug2("%s: buf=%p, start=%"PRIx64"\n", __func__, buf, *start);
210
-	pheader = (struct perf_event_header *)(buf + (*start & mask));
211
-	while (true) {
212
-		if (evt_head - *start >= (unsigned int)size) {
213
-			pr_debug("Finished reading overwrite ring buffer: rewind\n");
214
-			if (evt_head - *start > (unsigned int)size)
215
-				evt_head -= pheader->size;
216
-			*end = evt_head;
217
-			return 0;
218
-		}
219
-
220
-		pheader = (struct perf_event_header *)(buf + (evt_head & mask));
221
-
222
-		if (pheader->size == 0) {
223
-			pr_debug("Finished reading overwrite ring buffer: get start\n");
224
-			*end = evt_head;
225
-			return 0;
226
-		}
227
-
228
-		evt_head += pheader->size;
229
-		pr_debug3("move evt_head: %"PRIx64"\n", evt_head);
230
-	}
231
-	WARN_ONCE(1, "Shouldn't get here\n");
232
-	return -1;
233
-}
234
-
235
-/*
236
- * Report the start and end of the available data in ringbuffer
237
- */
238
-static int __perf_mmap__read_init(struct perf_mmap *md)
239
-{
240
-	u64 head = perf_mmap__read_head(md);
241
-	u64 old = md->prev;
242
-	unsigned char *data = md->base + page_size;
243
-	unsigned long size;
244
-
245
-	md->start = md->overwrite ? head : old;
246
-	md->end = md->overwrite ? old : head;
247
-
248
-	if (md->start == md->end)
249
-		return -EAGAIN;
250
-
251
-	size = md->end - md->start;
252
-	if (size > (unsigned long)(md->mask) + 1) {
253
-		if (!md->overwrite) {
254
-			WARN_ONCE(1, "failed to keep up with mmap data. (warn only once)\n");
255
-
256
-			md->prev = head;
257
-			perf_mmap__consume(md);
258
-			return -EAGAIN;
259
-		}
260
-
261
-		/*
262
-		 * Backward ring buffer is full. We still have a chance to read
263
-		 * most of data from it.
264
-		 */
265
-		if (overwrite_rb_find_range(data, md->mask, &md->start, &md->end))
266
-			return -EINVAL;
267
-	}
268
-
269
-	return 0;
270
-}
271
-
272
-int perf_mmap__read_init(struct perf_mmap *map)
273
-{
274
-	/*
275
-	 * Check if event was unmapped due to a POLLHUP/POLLERR.
276
-	 */
277
-	if (!refcount_read(&map->refcnt))
278
-		return -ENOENT;
279
-
280
-	return __perf_mmap__read_init(map);
281
-}
282
-
283
-int perf_mmap__push(struct perf_mmap *md, void *to,
284
-		    int push(void *to, void *buf, size_t size))
285
-{
286
-	u64 head = perf_mmap__read_head(md);
287
-	unsigned char *data = md->base + page_size;
316
+	u64 head = perf_mmap__read_head(&md->core);
317
+	unsigned char *data = md->core.base + page_size;
288
318
 	unsigned long size;
289
319
 	void *buf;
290
320
 	int rc = 0;
291
321
 
292
-	rc = perf_mmap__read_init(md);
322
+	rc = perf_mmap__read_init(&md->core);
293
323
 	if (rc < 0)
294
-		return (rc == -EAGAIN) ? 0 : -1;
324
+		return (rc == -EAGAIN) ? 1 : -1;
295
325
 
296
-	size = md->end - md->start;
326
+	size = md->core.end - md->core.start;
297
327
 
298
-	if ((md->start & md->mask) + size != (md->end & md->mask)) {
299
-		buf = &data[md->start & md->mask];
300
-		size = md->mask + 1 - (md->start & md->mask);
301
-		md->start += size;
328
+	if ((md->core.start & md->core.mask) + size != (md->core.end & md->core.mask)) {
329
+		buf = &data[md->core.start & md->core.mask];
330
+		size = md->core.mask + 1 - (md->core.start & md->core.mask);
331
+		md->core.start += size;
302
332
 
303
-		if (push(to, buf, size) < 0) {
333
+		if (push(md, to, buf, size) < 0) {
304
334
 			rc = -1;
305
335
 			goto out;
306
336
 		}
307
337
 	}
308
338
 
309
-	buf = &data[md->start & md->mask];
310
-	size = md->end - md->start;
311
-	md->start += size;
339
+	buf = &data[md->core.start & md->core.mask];
340
+	size = md->core.end - md->core.start;
341
+	md->core.start += size;
312
342
 
313
-	if (push(to, buf, size) < 0) {
343
+	if (push(md, to, buf, size) < 0) {
314
344
 		rc = -1;
315
345
 		goto out;
316
346
 	}
317
347
 
318
-	md->prev = head;
319
-	perf_mmap__consume(md);
348
+	md->core.prev = head;
349
+	perf_mmap__consume(&md->core);
320
350
 out:
321
351
 	return rc;
322
-}
323
-
324
-/*
325
- * Mandatory for overwrite mode
326
- * The direction of overwrite mode is backward.
327
- * The last perf_mmap__read() will set tail to map->prev.
328
- * Need to correct the map->prev to head which is the end of next read.
329
- */
330
-void perf_mmap__read_done(struct perf_mmap *map)
331
-{
332
-	/*
333
-	 * Check if event was unmapped due to a POLLHUP/POLLERR.
334
-	 */
335
-	if (!refcount_read(&map->refcnt))
336
-		return;
337
-
338
-	map->prev = perf_mmap__read_head(map);
339
352
 }