summaryrefslogtreecommitdiffstats
path: root/src/readahead/readahead-common.c
blob: 41aaff0a3d1532caf323f936bb9577e3a21df24b (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
/*-*- Mode: C; c-basic-offset: 8; indent-tabs-mode: nil -*-*/

/***
  This file is part of systemd.

  Copyright 2010 Lennart Poettering

  systemd is free software; you can redistribute it and/or modify it
  under the terms of the GNU Lesser General Public License as published by
  the Free Software Foundation; either version 2.1 of the License, or
  (at your option) any later version.

  systemd is distributed in the hope that it will be useful, but
  WITHOUT ANY WARRANTY; without even the implied warranty of
  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  Lesser General Public License for more details.

  You should have received a copy of the GNU Lesser General Public License
  along with systemd; If not, see <http://www.gnu.org/licenses/>.
***/

#include <errno.h>
#include <stdlib.h>
#include <string.h>
#include <sys/sysinfo.h>
#include <sys/inotify.h>
#include <fcntl.h>
#include <sys/mman.h>
#include <unistd.h>
#include <libudev.h>

#include "log.h"
#include "readahead-common.h"
#include "util.h"
#include "missing.h"

int file_verify(int fd, const char *fn, off_t file_size_max, struct stat *st) {
        assert(fd >= 0);
        assert(fn);
        assert(st);

        if (fstat(fd, st) < 0) {
                log_warning("fstat(%s) failed: %m", fn);
                return -errno;
        }

        if (!S_ISREG(st->st_mode)) {
                log_debug("Not preloading special file %s", fn);
                return 0;
        }

        if (st->st_size <= 0 || st->st_size > file_size_max) {
                log_debug("Not preloading file %s with size out of bounds %llu", fn, (unsigned long long) st->st_size);
                return 0;
        }

        return 1;
}

int fs_on_ssd(const char *p) {
        struct stat st;
        struct udev *udev = NULL;
        struct udev_device *udev_device = NULL, *look_at = NULL;
        bool b = false;
        const char *devtype, *rotational, *model, *id;
        int r;

        assert(p);

        if (stat(p, &st) < 0)
                return -errno;

        if (major(st.st_dev) == 0) {
                _cleanup_fclose_ FILE *f = NULL;
                int mount_id;
                struct file_handle *h;

                /* Might be btrfs, which exposes "ssd" as mount flag if it is on ssd.
                 *
                 * We first determine the mount ID here, if we can,
                 * and then lookup the mount ID in mountinfo to find
                 * the mount options. */

                h = alloca(MAX_HANDLE_SZ);
                h->handle_bytes = MAX_HANDLE_SZ;
                r = name_to_handle_at(AT_FDCWD, p, h, &mount_id, AT_SYMLINK_FOLLOW);
                if (r < 0)
                        return false;

                f = fopen("/proc/self/mountinfo", "re");
                if (!f)
                        return false;

                for (;;) {
                        char line[LINE_MAX], *e;
                        _cleanup_free_ char *opts = NULL;
                        int mid;

                        if (!fgets(line, sizeof(line), f))
                                return false;

                        truncate_nl(line);

                        if (sscanf(line, "%i", &mid) != 1)
                                continue;

                        if (mid != mount_id)
                                continue;

                        e = strstr(line, " - ");
                        if (!e)
                                continue;

                        if (sscanf(e+3, "%*s %*s %ms", &opts) != 1)
                                continue;

                        if (streq(opts, "ssd") || startswith(opts, "ssd,") || endswith(opts, ",ssd") || strstr(opts, ",ssd,"))
                                return true;
                }

                return false;
        }

        udev = udev_new();
        if (!udev)
                return -ENOMEM;

        udev_device = udev_device_new_from_devnum(udev, 'b', st.st_dev);
        if (!udev_device)
                goto finish;

        devtype = udev_device_get_property_value(udev_device, "DEVTYPE");
        if (devtype && streq(devtype, "partition"))
                look_at = udev_device_get_parent(udev_device);
        else
                look_at = udev_device;

        if (!look_at)
                goto finish;

        /* First, try high-level property */
        id = udev_device_get_property_value(look_at, "ID_SSD");
        if (id) {
                b = streq(id, "1");
                goto finish;
        }

        /* Second, try kernel attribute */
        rotational = udev_device_get_sysattr_value(look_at, "queue/rotational");
        if (rotational) {
                b = streq(rotational, "0");
                goto finish;
        }

        /* Finally, fallback to heuristics */
        look_at = udev_device_get_parent(look_at);
        if (!look_at)
                goto finish;

        model = udev_device_get_sysattr_value(look_at, "model");
        if (model)
                b = !!strstr(model, "SSD");

finish:
        if (udev_device)
                udev_device_unref(udev_device);

        if (udev)
                udev_unref(udev);

        return b;
}

int fs_on_read_only(const char *p) {
        struct stat st;
        struct udev *udev = NULL;
        struct udev_device *udev_device = NULL;
        bool b = false;
        const char *read_only;

        assert(p);

        if (stat(p, &st) < 0)
                return -errno;

        if (major(st.st_dev) == 0)
                return false;

        if (!(udev = udev_new()))
                return -ENOMEM;

        if (!(udev_device = udev_device_new_from_devnum(udev, 'b', st.st_dev)))
                goto finish;

        if ((read_only = udev_device_get_sysattr_value(udev_device, "ro")))
                if ((b = streq(read_only, "1")))
                        goto finish;

finish:
        if (udev_device)
                udev_device_unref(udev_device);

        if (udev)
                udev_unref(udev);

        return b;
}

bool enough_ram(void) {
        struct sysinfo si;

        assert_se(sysinfo(&si) >= 0);

        /* Enable readahead only with at least 128MB memory */
        return si.totalram > 127 * 1024*1024 / si.mem_unit;
}

int open_inotify(void) {
        int fd;

        if ((fd = inotify_init1(IN_CLOEXEC|IN_NONBLOCK)) < 0) {
                log_error("Failed to create inotify handle: %m");
                return -errno;
        }

        mkdir("/run/systemd", 0755);
        mkdir("/run/systemd/readahead", 0755);

        if (inotify_add_watch(fd, "/run/systemd/readahead", IN_CREATE) < 0) {
                log_error("Failed to watch /run/systemd/readahead: %m");
                close_nointr_nofail(fd);
                return -errno;
        }

        return fd;
}

ReadaheadShared *shared_get(void) {
        int fd;
        ReadaheadShared *m = NULL;

        mkdir("/run/systemd", 0755);
        mkdir("/run/systemd/readahead", 0755);

        if ((fd = open("/run/systemd/readahead/shared", O_CREAT|O_RDWR|O_CLOEXEC, 0644)) < 0) {
                log_error("Failed to create shared memory segment: %m");
                goto finish;
        }

        if (ftruncate(fd, sizeof(ReadaheadShared)) < 0) {
                log_error("Failed to truncate shared memory segment: %m");
                goto finish;
        }

        if ((m = mmap(NULL, sizeof(ReadaheadShared), PROT_WRITE|PROT_READ, MAP_SHARED, fd, 0)) == MAP_FAILED) {
                log_error("Failed to mmap shared memory segment: %m");
                m = NULL;
                goto finish;
        }

finish:
        if (fd >= 0)
                close_nointr_nofail(fd);

        return m;
}

/* We use 20K instead of the more human digestable 16K here. Why?
   Simply so that it is more unlikely that users end up picking this
   value too so that we can recognize better whether the user changed
   the value while we had it temporarily bumped. */
#define BUMP_REQUEST_NR (20*1024)

int block_bump_request_nr(const char *p) {
        struct stat st;
        uint64_t u;
        char *ap = NULL, *line = NULL;
        int r;
        dev_t d;

        assert(p);

        if (stat(p, &st) < 0)
                return -errno;

        if (major(st.st_dev) == 0)
                return 0;

        d = st.st_dev;
        block_get_whole_disk(d, &d);

        if (asprintf(&ap, "/sys/dev/block/%u:%u/queue/nr_requests", major(d), minor(d)) < 0) {
                r= -ENOMEM;
                goto finish;
        }

        r = read_one_line_file(ap, &line);
        if (r < 0) {
                if (r == -ENOENT)
                        r = 0;
                goto finish;
        }

        r = safe_atou64(line, &u);
        if (r >= 0 && u >= BUMP_REQUEST_NR) {
                r = 0;
                goto finish;
        }

        free(line);
        line = NULL;

        if (asprintf(&line, "%lu", (unsigned long) BUMP_REQUEST_NR) < 0) {
                r = -ENOMEM;
                goto finish;
        }

        r = write_one_line_file(ap, line);
        if (r < 0)
                goto finish;

        log_info("Bumped block_nr parameter of %u:%u to %lu. This is a temporary hack and should be removed one day.", major(d), minor(d), (unsigned long) BUMP_REQUEST_NR);
        r = 1;

finish:
        free(ap);
        free(line);

        return r;
}

int block_get_readahead(const char *p, uint64_t *bytes) {
        struct stat st;
        char *ap = NULL, *line = NULL;
        int r;
        dev_t d;
        uint64_t u;

        assert(p);
        assert(bytes);

        if (stat(p, &st) < 0)
                return -errno;

        if (major(st.st_dev) == 0)
                return 0;

        d = st.st_dev;
        block_get_whole_disk(d, &d);

        if (asprintf(&ap, "/sys/dev/block/%u:%u/bdi/read_ahead_kb", major(d), minor(d)) < 0) {
                r = -ENOMEM;
                goto finish;
        }

        r = read_one_line_file(ap, &line);
        if (r < 0)
                goto finish;

        r = safe_atou64(line, &u);
        if (r < 0)
                goto finish;

        *bytes = u * 1024ULL;

finish:
        free(ap);
        free(line);

        return r;
}

int block_set_readahead(const char *p, uint64_t bytes) {
        struct stat st;
        char *ap = NULL, *line = NULL;
        int r;
        dev_t d;

        assert(p);
        assert(bytes);

        if (stat(p, &st) < 0)
                return -errno;

        if (major(st.st_dev) == 0)
                return 0;

        d = st.st_dev;
        block_get_whole_disk(d, &d);

        if (asprintf(&ap, "/sys/dev/block/%u:%u/bdi/read_ahead_kb", major(d), minor(d)) < 0) {
                r = -ENOMEM;
                goto finish;
        }

        if (asprintf(&line, "%llu", (unsigned long long) bytes / 1024ULL) < 0) {
                r = -ENOMEM;
                goto finish;
        }

        r = write_one_line_file(ap, line);
        if (r < 0)
                goto finish;

finish:
        free(ap);
        free(line);

        return r;
}