[thirdparty/systemd.git] / src / journal / journal-vacuum.c

/* SPDX-License-Identifier: LGPL-2.1+ */

#include <fcntl.h>
#include <sys/stat.h>
#include <unistd.h>

#include "sd-id128.h"

#include "alloc-util.h"
#include "dirent-util.h"
#include "fd-util.h"
#include "fs-util.h"
#include "journal-def.h"
#include "journal-file.h"
#include "journal-vacuum.h"
#include "parse-util.h"
#include "string-util.h"
#include "time-util.h"
#include "util.h"
#include "xattr-util.h"

struct vacuum_info {
        uint64_t usage;
        char *filename;

        uint64_t realtime;

        sd_id128_t seqnum_id;
        uint64_t seqnum;
        bool have_seqnum;
};

static int vacuum_compare(const struct vacuum_info *a, const struct vacuum_info *b) {
        int r;

        if (a->have_seqnum && b->have_seqnum &&
            sd_id128_equal(a->seqnum_id, b->seqnum_id))
                return CMP(a->seqnum, b->seqnum);

        r = CMP(a->realtime, b->realtime);
        if (r != 0)
                return r;

        if (a->have_seqnum && b->have_seqnum)
                return memcmp(&a->seqnum_id, &b->seqnum_id, 16);

        return strcmp(a->filename, b->filename);
}

static void patch_realtime(
                int fd,
                const char *fn,
                const struct stat *st,
                unsigned long long *realtime) {

        usec_t x, crtime = 0;

        /* The timestamp was determined by the file name, but let's
         * see if the file might actually be older than the file name
         * suggested... */

        assert(fd >= 0);
        assert(fn);
        assert(st);
        assert(realtime);

        x = timespec_load(&st->st_ctim);
        if (x > 0 && x != USEC_INFINITY && x < *realtime)
                *realtime = x;

        x = timespec_load(&st->st_atim);
        if (x > 0 && x != USEC_INFINITY && x < *realtime)
                *realtime = x;

        x = timespec_load(&st->st_mtim);
        if (x > 0 && x != USEC_INFINITY && x < *realtime)
                *realtime = x;

        /* Let's read the original creation time, if possible. Ideally
         * we'd just query the creation time the FS might provide, but
         * unfortunately there's currently no sane API to query
         * it. Hence let's implement this manually... */

        if (fd_getcrtime_at(fd, fn, &crtime, 0) >= 0) {
                if (crtime < *realtime)
                        *realtime = crtime;
        }
}

static int journal_file_empty(int dir_fd, const char *name) {
        _cleanup_close_ int fd;
        struct stat st;
        le64_t n_entries;
        ssize_t n;

        fd = openat(dir_fd, name, O_RDONLY|O_CLOEXEC|O_NOFOLLOW|O_NONBLOCK|O_NOATIME);
        if (fd < 0) {
                /* Maybe failed due to O_NOATIME and lack of privileges? */
                fd = openat(dir_fd, name, O_RDONLY|O_CLOEXEC|O_NOFOLLOW|O_NONBLOCK);
                if (fd < 0)
                        return -errno;
        }

        if (fstat(fd, &st) < 0)
                return -errno;

        /* If an offline file doesn't even have a header we consider it empty */
        if (st.st_size < (off_t) sizeof(Header))
                return 1;

        /* If the number of entries is empty, we consider it empty, too */
        n = pread(fd, &n_entries, sizeof(n_entries), offsetof(Header, n_entries));
        if (n < 0)
                return -errno;
        if (n != sizeof(n_entries))
                return -EIO;

        return le64toh(n_entries) <= 0;
}

int journal_directory_vacuum(
                const char *directory,
                uint64_t max_use,
                uint64_t n_max_files,
                usec_t max_retention_usec,
                usec_t *oldest_usec,
                bool verbose) {

        _cleanup_closedir_ DIR *d = NULL;
        struct vacuum_info *list = NULL;
        unsigned n_list = 0, i, n_active_files = 0;
        size_t n_allocated = 0;
        uint64_t sum = 0, freed = 0;
        usec_t retention_limit = 0;
        char sbytes[FORMAT_BYTES_MAX];
        struct dirent *de;
        int r;

        assert(directory);

        if (max_use <= 0 && max_retention_usec <= 0 && n_max_files <= 0)
                return 0;

        if (max_retention_usec > 0)
                retention_limit = usec_sub_unsigned(now(CLOCK_REALTIME), max_retention_usec);

        d = opendir(directory);
        if (!d)
                return -errno;

        FOREACH_DIRENT_ALL(de, d, r = -errno; goto finish) {

                unsigned long long seqnum = 0, realtime;
                _cleanup_free_ char *p = NULL;
                sd_id128_t seqnum_id;
                bool have_seqnum;
                uint64_t size;
                struct stat st;
                size_t q;

                if (fstatat(dirfd(d), de->d_name, &st, AT_SYMLINK_NOFOLLOW) < 0) {
                        log_debug_errno(errno, "Failed to stat file %s while vacuuming, ignoring: %m", de->d_name);
                        continue;
                }

                if (!S_ISREG(st.st_mode))
                        continue;

                q = strlen(de->d_name);

                if (endswith(de->d_name, ".journal")) {

                        /* Vacuum archived files. Active files are
                         * left around */

                        if (q < 1 + 32 + 1 + 16 + 1 + 16 + 8) {
                                n_active_files++;
                                continue;
                        }

                        if (de->d_name[q-8-16-1] != '-' ||
                            de->d_name[q-8-16-1-16-1] != '-' ||
                            de->d_name[q-8-16-1-16-1-32-1] != '@') {
                                n_active_files++;
                                continue;
                        }

                        p = strdup(de->d_name);
                        if (!p) {
                                r = -ENOMEM;
                                goto finish;
                        }

                        de->d_name[q-8-16-1-16-1] = 0;
                        if (sd_id128_from_string(de->d_name + q-8-16-1-16-1-32, &seqnum_id) < 0) {
                                n_active_files++;
                                continue;
                        }

                        if (sscanf(de->d_name + q-8-16-1-16, "%16llx-%16llx.journal", &seqnum, &realtime) != 2) {
                                n_active_files++;
                                continue;
                        }

                        have_seqnum = true;

                } else if (endswith(de->d_name, ".journal~")) {
                        unsigned long long tmp;

                        /* Vacuum corrupted files */

                        if (q < 1 + 16 + 1 + 16 + 8 + 1) {
                                n_active_files++;
                                continue;
                        }

                        if (de->d_name[q-1-8-16-1] != '-' ||
                            de->d_name[q-1-8-16-1-16-1] != '@') {
                                n_active_files++;
                                continue;
                        }

                        p = strdup(de->d_name);
                        if (!p) {
                                r = -ENOMEM;
                                goto finish;
                        }

                        if (sscanf(de->d_name + q-1-8-16-1-16, "%16llx-%16llx.journal~", &realtime, &tmp) != 2) {
                                n_active_files++;
                                continue;
                        }

                        have_seqnum = false;
                } else {
                        /* We do not vacuum unknown files! */
                        log_debug("Not vacuuming unknown file %s.", de->d_name);
                        continue;
                }

                size = 512UL * (uint64_t) st.st_blocks;

                r = journal_file_empty(dirfd(d), p);
                if (r < 0) {
                        log_debug_errno(r, "Failed check if %s is empty, ignoring: %m", p);
                        continue;
                }
                if (r > 0) {
                        /* Always vacuum empty non-online files. */

                        r = unlinkat_deallocate(dirfd(d), p, 0);
                        if (r >= 0) {

                                log_full(verbose ? LOG_INFO : LOG_DEBUG,
                                         "Deleted empty archived journal %s/%s (%s).", directory, p, format_bytes(sbytes, sizeof(sbytes), size));

                                freed += size;
                        } else if (r != -ENOENT)
                                log_warning_errno(r, "Failed to delete empty archived journal %s/%s: %m", directory, p);

                        continue;
                }

                patch_realtime(dirfd(d), p, &st, &realtime);

                if (!GREEDY_REALLOC(list, n_allocated, n_list + 1)) {
                        r = -ENOMEM;
                        goto finish;
                }

                list[n_list++] = (struct vacuum_info) {
                        .filename = TAKE_PTR(p),
                        .usage = size,
                        .seqnum = seqnum,
                        .realtime = realtime,
                        .seqnum_id = seqnum_id,
                        .have_seqnum = have_seqnum,
                };

                sum += size;
        }

        typesafe_qsort(list, n_list, vacuum_compare);

        for (i = 0; i < n_list; i++) {
                unsigned left;

                left = n_active_files + n_list - i;

                if ((max_retention_usec <= 0 || list[i].realtime >= retention_limit) &&
                    (max_use <= 0 || sum <= max_use) &&
                    (n_max_files <= 0 || left <= n_max_files))
                        break;

                r = unlinkat_deallocate(dirfd(d), list[i].filename, 0);
                if (r >= 0) {
                        log_full(verbose ? LOG_INFO : LOG_DEBUG, "Deleted archived journal %s/%s (%s).", directory, list[i].filename, format_bytes(sbytes, sizeof(sbytes), list[i].usage));
                        freed += list[i].usage;

                        if (list[i].usage < sum)
                                sum -= list[i].usage;
                        else
                                sum = 0;

                } else if (r != -ENOENT)
                        log_warning_errno(r, "Failed to delete archived journal %s/%s: %m", directory, list[i].filename);
        }

        if (oldest_usec && i < n_list && (*oldest_usec == 0 || list[i].realtime < *oldest_usec))
                *oldest_usec = list[i].realtime;

        r = 0;

finish:
        for (i = 0; i < n_list; i++)
                free(list[i].filename);
        free(list);

        log_full(verbose ? LOG_INFO : LOG_DEBUG, "Vacuuming done, freed %s of archived journals from %s.", format_bytes(sbytes, sizeof(sbytes), freed), directory);

        return r;
}
Commit	Line	Data
53e1b683	1	/* SPDX-License-Identifier: LGPL-2.1+ */
0284adc6	2
0284adc6 LP	3	#include <fcntl.h>
0284adc6 LP	4	#include <sys/stat.h>
0284adc6	5	#include <unistd.h>
fb0951b0	6
07630cea LP	7	#include "sd-id128.h"
07630cea LP	8
b5efdb8a	9	#include "alloc-util.h"
a0956174	10	#include "dirent-util.h"
3ffd4af2	11	#include "fd-util.h"
47c073aa	12	#include "fs-util.h"
0284adc6 LP	13	#include "journal-def.h"
0284adc6 LP	14	#include "journal-file.h"
3ffd4af2	15	#include "journal-vacuum.h"
6bedfcbb	16	#include "parse-util.h"
07630cea	17	#include "string-util.h"
6761e0a2	18	#include "time-util.h"
0284adc6	19	#include "util.h"
89a5a90c	20	#include "xattr-util.h"
0284adc6 LP	21
0284adc6 LP	22	struct vacuum_info {
6c142648	23	uint64_t usage;
0284adc6 LP	24	char *filename;
	25
	26	uint64_t realtime;
2e14c544	27
0284adc6 LP	28	sd_id128_t seqnum_id;
0284adc6 LP	29	uint64_t seqnum;
0284adc6 LP	30	bool have_seqnum;
	31	};
	32
93bab288 YW	33	static int vacuum_compare(const struct vacuum_info a, const struct vacuum_info b) {
93bab288 YW	34	int r;
0284adc6 LP	35
0284adc6 LP	36	if (a->have_seqnum && b->have_seqnum &&
93bab288 YW	37	sd_id128_equal(a->seqnum_id, b->seqnum_id))
93bab288 YW	38	return CMP(a->seqnum, b->seqnum);
0284adc6	39
93bab288 YW	40	r = CMP(a->realtime, b->realtime);
	41	if (r != 0)
	42	return r;
	43
	44	if (a->have_seqnum && b->have_seqnum)
0284adc6	45	return memcmp(&a->seqnum_id, &b->seqnum_id, 16);
93bab288 YW	46
93bab288 YW	47	return strcmp(a->filename, b->filename);
0284adc6 LP	48	}
0284adc6 LP	49
fb0951b0	50	static void patch_realtime(
2e14c544	51	int fd,
fb0951b0 LP	52	const char *fn,
	53	const struct stat *st,
	54	unsigned long long *realtime) {
	55
a7f7d1bd	56	usec_t x, crtime = 0;
fb0951b0 LP	57
	58	/* The timestamp was determined by the file name, but let's
	59	* see if the file might actually be older than the file name
	60	* suggested... */
	61
2e14c544	62	assert(fd >= 0);
fb0951b0 LP	63	assert(fn);
	64	assert(st);
	65	assert(realtime);
	66
	67	x = timespec_load(&st->st_ctim);
3a43da28	68	if (x > 0 && x != USEC_INFINITY && x < *realtime)
fb0951b0 LP	69	*realtime = x;
	70
	71	x = timespec_load(&st->st_atim);
3a43da28	72	if (x > 0 && x != USEC_INFINITY && x < *realtime)
fb0951b0 LP	73	*realtime = x;
	74
	75	x = timespec_load(&st->st_mtim);
3a43da28	76	if (x > 0 && x != USEC_INFINITY && x < *realtime)
fb0951b0 LP	77	*realtime = x;
fb0951b0 LP	78
fb0951b0 LP	79	/* Let's read the original creation time, if possible. Ideally
	80	* we'd just query the creation time the FS might provide, but
	81	* unfortunately there's currently no sane API to query
	82	* it. Hence let's implement this manually... */
	83
2e14c544	84	if (fd_getcrtime_at(fd, fn, &crtime, 0) >= 0) {
4a4d89b6	85	if (crtime < *realtime)
fb0951b0 LP	86	*realtime = crtime;
fb0951b0 LP	87	}
fb0951b0 LP	88	}
fb0951b0 LP	89
9d647740	90	static int journal_file_empty(int dir_fd, const char *name) {
48979861	91	_cleanup_close_ int fd;
332076b4 LP	92	struct stat st;
	93	le64_t n_entries;
	94	ssize_t n;
9d647740	95
7b5195e2 LP	96	fd = openat(dir_fd, name, O_RDONLY\|O_CLOEXEC\|O_NOFOLLOW\|O_NONBLOCK\|O_NOATIME);
	97	if (fd < 0) {
	98	/* Maybe failed due to O_NOATIME and lack of privileges? */
	99	fd = openat(dir_fd, name, O_RDONLY\|O_CLOEXEC\|O_NOFOLLOW\|O_NONBLOCK);
	100	if (fd < 0)
	101	return -errno;
	102	}
9d647740	103
332076b4	104	if (fstat(fd, &st) < 0)
9d647740 ZJS	105	return -errno;
9d647740 ZJS	106
332076b4 LP	107	/* If an offline file doesn't even have a header we consider it empty */
	108	if (st.st_size < (off_t) sizeof(Header))
	109	return 1;
	110
	111	/* If the number of entries is empty, we consider it empty, too */
	112	n = pread(fd, &n_entries, sizeof(n_entries), offsetof(Header, n_entries));
	113	if (n < 0)
	114	return -errno;
	115	if (n != sizeof(n_entries))
	116	return -EIO;
9d647740	117
332076b4	118	return le64toh(n_entries) <= 0;
9d647740 ZJS	119	}
9d647740 ZJS	120
fb0951b0 LP	121	int journal_directory_vacuum(
	122	const char *directory,
	123	uint64_t max_use,
8580d1f7	124	uint64_t n_max_files,
fb0951b0	125	usec_t max_retention_usec,
dbd2a83f LP	126	usec_t *oldest_usec,
dbd2a83f LP	127	bool verbose) {
fb0951b0	128
30cb029b	129	_cleanup_closedir_ DIR *d = NULL;
0284adc6	130	struct vacuum_info *list = NULL;
8580d1f7	131	unsigned n_list = 0, i, n_active_files = 0;
30cb029b	132	size_t n_allocated = 0;
289f910e	133	uint64_t sum = 0, freed = 0;
fb0951b0	134	usec_t retention_limit = 0;
dbd2a83f	135	char sbytes[FORMAT_BYTES_MAX];
8580d1f7 LP	136	struct dirent *de;
8580d1f7 LP	137	int r;
0284adc6 LP	138
	139	assert(directory);
	140
8580d1f7	141	if (max_use <= 0 && max_retention_usec <= 0 && n_max_files <= 0)
0284adc6 LP	142	return 0;
0284adc6 LP	143
6761e0a2 LP	144	if (max_retention_usec > 0)
6761e0a2 LP	145	retention_limit = usec_sub_unsigned(now(CLOCK_REALTIME), max_retention_usec);
fb0951b0	146
0284adc6 LP	147	d = opendir(directory);
	148	if (!d)
	149	return -errno;
	150
8580d1f7 LP	151	FOREACH_DIRENT_ALL(de, d, r = -errno; goto finish) {
8580d1f7 LP	152
0284adc6	153	unsigned long long seqnum = 0, realtime;
8580d1f7	154	_cleanup_free_ char *p = NULL;
0284adc6 LP	155	sd_id128_t seqnum_id;
0284adc6 LP	156	bool have_seqnum;
8580d1f7 LP	157	uint64_t size;
	158	struct stat st;
	159	size_t q;
0284adc6	160
8580d1f7 LP	161	if (fstatat(dirfd(d), de->d_name, &st, AT_SYMLINK_NOFOLLOW) < 0) {
8580d1f7 LP	162	log_debug_errno(errno, "Failed to stat file %s while vacuuming, ignoring: %m", de->d_name);
0284adc6	163	continue;
8580d1f7	164	}
0284adc6 LP	165
	166	if (!S_ISREG(st.st_mode))
	167	continue;
	168
	169	q = strlen(de->d_name);
	170
	171	if (endswith(de->d_name, ".journal")) {
	172
8580d1f7 LP	173	/* Vacuum archived files. Active files are
8580d1f7 LP	174	* left around */
0284adc6	175
8580d1f7 LP	176	if (q < 1 + 32 + 1 + 16 + 1 + 16 + 8) {
8580d1f7 LP	177	n_active_files++;
0284adc6	178	continue;
8580d1f7	179	}
0284adc6 LP	180
	181	if (de->d_name[q-8-16-1] != '-' \|\|
	182	de->d_name[q-8-16-1-16-1] != '-' \|\|
8580d1f7 LP	183	de->d_name[q-8-16-1-16-1-32-1] != '@') {
8580d1f7 LP	184	n_active_files++;
0284adc6	185	continue;
8580d1f7	186	}
0284adc6 LP	187
	188	p = strdup(de->d_name);
	189	if (!p) {
	190	r = -ENOMEM;
	191	goto finish;
	192	}
	193
	194	de->d_name[q-8-16-1-16-1] = 0;
	195	if (sd_id128_from_string(de->d_name + q-8-16-1-16-1-32, &seqnum_id) < 0) {
8580d1f7	196	n_active_files++;
0284adc6 LP	197	continue;
	198	}
	199
	200	if (sscanf(de->d_name + q-8-16-1-16, "%16llx-%16llx.journal", &seqnum, &realtime) != 2) {
8580d1f7	201	n_active_files++;
0284adc6 LP	202	continue;
	203	}
	204
	205	have_seqnum = true;
	206
	207	} else if (endswith(de->d_name, ".journal~")) {
	208	unsigned long long tmp;
	209
	210	/* Vacuum corrupted files */
	211
8580d1f7	212	if (q < 1 + 16 + 1 + 16 + 8 + 1) {
313cefa1	213	n_active_files++;
0284adc6	214	continue;
8580d1f7	215	}
0284adc6 LP	216
0284adc6 LP	217	if (de->d_name[q-1-8-16-1] != '-' \|\|
8580d1f7	218	de->d_name[q-1-8-16-1-16-1] != '@') {
313cefa1	219	n_active_files++;
0284adc6	220	continue;
8580d1f7	221	}
0284adc6 LP	222
	223	p = strdup(de->d_name);
	224	if (!p) {
	225	r = -ENOMEM;
	226	goto finish;
	227	}
	228
	229	if (sscanf(de->d_name + q-1-8-16-1-16, "%16llx-%16llx.journal~", &realtime, &tmp) != 2) {
313cefa1	230	n_active_files++;
0284adc6 LP	231	continue;
	232	}
	233
	234	have_seqnum = false;
8580d1f7 LP	235	} else {
	236	/* We do not vacuum unknown files! */
	237	log_debug("Not vacuuming unknown file %s.", de->d_name);
0284adc6	238	continue;
8580d1f7	239	}
0284adc6	240
8580d1f7	241	size = 512UL * (uint64_t) st.st_blocks;
9d647740	242
8580d1f7 LP	243	r = journal_file_empty(dirfd(d), p);
	244	if (r < 0) {
	245	log_debug_errno(r, "Failed check if %s is empty, ignoring: %m", p);
	246	continue;
	247	}
	248	if (r > 0) {
	249	/* Always vacuum empty non-online files. */
289f910e	250
47c073aa LP	251	r = unlinkat_deallocate(dirfd(d), p, 0);
47c073aa LP	252	if (r >= 0) {
8580d1f7 LP	253
	254	log_full(verbose ? LOG_INFO : LOG_DEBUG,
	255	"Deleted empty archived journal %s/%s (%s).", directory, p, format_bytes(sbytes, sizeof(sbytes), size));
	256
289f910e	257	freed += size;
47c073aa LP	258	} else if (r != -ENOENT)
47c073aa LP	259	log_warning_errno(r, "Failed to delete empty archived journal %s/%s: %m", directory, p);
289f910e	260
9d647740 ZJS	261	continue;
	262	}
	263
8580d1f7	264	patch_realtime(dirfd(d), p, &st, &realtime);
fb0951b0	265
26d8ff04	266	if (!GREEDY_REALLOC(list, n_allocated, n_list + 1)) {
26d8ff04 LP	267	r = -ENOMEM;
	268	goto finish;
	269	}
0284adc6	270
ab41da08 LP	271	list[n_list++] = (struct vacuum_info) {
	272	.filename = TAKE_PTR(p),
	273	.usage = size,
	274	.seqnum = seqnum,
	275	.realtime = realtime,
	276	.seqnum_id = seqnum_id,
	277	.have_seqnum = have_seqnum,
	278	};
8580d1f7	279
8580d1f7	280	sum += size;
0284adc6 LP	281	}
0284adc6 LP	282
93bab288	283	typesafe_qsort(list, n_list, vacuum_compare);
0284adc6	284
fb0951b0	285	for (i = 0; i < n_list; i++) {
8580d1f7 LP	286	unsigned left;
	287
	288	left = n_active_files + n_list - i;
	289
fb0951b0	290	if ((max_retention_usec <= 0 \|\| list[i].realtime >= retention_limit) &&
8580d1f7 LP	291	(max_use <= 0 \|\| sum <= max_use) &&
8580d1f7 LP	292	(n_max_files <= 0 \|\| left <= n_max_files))
0284adc6 LP	293	break;
0284adc6 LP	294
47c073aa LP	295	r = unlinkat_deallocate(dirfd(d), list[i].filename, 0);
47c073aa LP	296	if (r >= 0) {
dbd2a83f	297	log_full(verbose ? LOG_INFO : LOG_DEBUG, "Deleted archived journal %s/%s (%s).", directory, list[i].filename, format_bytes(sbytes, sizeof(sbytes), list[i].usage));
289f910e	298	freed += list[i].usage;
4fa25d62	299
6c142648	300	if (list[i].usage < sum)
4fa25d62 LP	301	sum -= list[i].usage;
	302	else
	303	sum = 0;
	304
47c073aa LP	305	} else if (r != -ENOENT)
47c073aa LP	306	log_warning_errno(r, "Failed to delete archived journal %s/%s: %m", directory, list[i].filename);
0284adc6 LP	307	}
0284adc6 LP	308
fb0951b0 LP	309	if (oldest_usec && i < n_list && (oldest_usec == 0 \|\| list[i].realtime < oldest_usec))
	310	*oldest_usec = list[i].realtime;
	311
8580d1f7 LP	312	r = 0;
8580d1f7 LP	313
0284adc6 LP	314	finish:
	315	for (i = 0; i < n_list; i++)
	316	free(list[i].filename);
0284adc6 LP	317	free(list);
0284adc6 LP	318
3cc44bf9	319	log_full(verbose ? LOG_INFO : LOG_DEBUG, "Vacuuming done, freed %s of archived journals from %s.", format_bytes(sbytes, sizeof(sbytes), freed), directory);
289f910e	320
0284adc6 LP	321	return r;
0284adc6 LP	322	}