[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 "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 = now(CLOCK_REALTIME);
                if (retention_limit > max_retention_usec)
                        retention_limit -= max_retention_usec;
                else
                        max_retention_usec = retention_limit = 0;
        }

        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].filename = TAKE_PTR(p);
                list[n_list].usage = size;
                list[n_list].seqnum = seqnum;
                list[n_list].realtime = realtime;
                list[n_list].seqnum_id = seqnum_id;
                list[n_list].have_seqnum = have_seqnum;
                n_list++;

                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"
0284adc6	18	#include "util.h"
89a5a90c	19	#include "xattr-util.h"
0284adc6 LP	20
0284adc6 LP	21	struct vacuum_info {
6c142648	22	uint64_t usage;
0284adc6 LP	23	char *filename;
	24
	25	uint64_t realtime;
2e14c544	26
0284adc6 LP	27	sd_id128_t seqnum_id;
0284adc6 LP	28	uint64_t seqnum;
0284adc6 LP	29	bool have_seqnum;
	30	};
	31
93bab288 YW	32	static int vacuum_compare(const struct vacuum_info a, const struct vacuum_info b) {
93bab288 YW	33	int r;
0284adc6 LP	34
0284adc6 LP	35	if (a->have_seqnum && b->have_seqnum &&
93bab288 YW	36	sd_id128_equal(a->seqnum_id, b->seqnum_id))
93bab288 YW	37	return CMP(a->seqnum, b->seqnum);
0284adc6	38
93bab288 YW	39	r = CMP(a->realtime, b->realtime);
	40	if (r != 0)
	41	return r;
	42
	43	if (a->have_seqnum && b->have_seqnum)
0284adc6	44	return memcmp(&a->seqnum_id, &b->seqnum_id, 16);
93bab288 YW	45
93bab288 YW	46	return strcmp(a->filename, b->filename);
0284adc6 LP	47	}
0284adc6 LP	48
fb0951b0	49	static void patch_realtime(
2e14c544	50	int fd,
fb0951b0 LP	51	const char *fn,
	52	const struct stat *st,
	53	unsigned long long *realtime) {
	54
a7f7d1bd	55	usec_t x, crtime = 0;
fb0951b0 LP	56
	57	/* The timestamp was determined by the file name, but let's
	58	* see if the file might actually be older than the file name
	59	* suggested... */
	60
2e14c544	61	assert(fd >= 0);
fb0951b0 LP	62	assert(fn);
	63	assert(st);
	64	assert(realtime);
	65
	66	x = timespec_load(&st->st_ctim);
3a43da28	67	if (x > 0 && x != USEC_INFINITY && x < *realtime)
fb0951b0 LP	68	*realtime = x;
	69
	70	x = timespec_load(&st->st_atim);
3a43da28	71	if (x > 0 && x != USEC_INFINITY && x < *realtime)
fb0951b0 LP	72	*realtime = x;
	73
	74	x = timespec_load(&st->st_mtim);
3a43da28	75	if (x > 0 && x != USEC_INFINITY && x < *realtime)
fb0951b0 LP	76	*realtime = x;
fb0951b0 LP	77
fb0951b0 LP	78	/* Let's read the original creation time, if possible. Ideally
	79	* we'd just query the creation time the FS might provide, but
	80	* unfortunately there's currently no sane API to query
	81	* it. Hence let's implement this manually... */
	82
2e14c544	83	if (fd_getcrtime_at(fd, fn, &crtime, 0) >= 0) {
4a4d89b6	84	if (crtime < *realtime)
fb0951b0 LP	85	*realtime = crtime;
fb0951b0 LP	86	}
fb0951b0 LP	87	}
fb0951b0 LP	88
9d647740	89	static int journal_file_empty(int dir_fd, const char *name) {
48979861	90	_cleanup_close_ int fd;
332076b4 LP	91	struct stat st;
	92	le64_t n_entries;
	93	ssize_t n;
9d647740	94
7b5195e2 LP	95	fd = openat(dir_fd, name, O_RDONLY\|O_CLOEXEC\|O_NOFOLLOW\|O_NONBLOCK\|O_NOATIME);
	96	if (fd < 0) {
	97	/* Maybe failed due to O_NOATIME and lack of privileges? */
	98	fd = openat(dir_fd, name, O_RDONLY\|O_CLOEXEC\|O_NOFOLLOW\|O_NONBLOCK);
	99	if (fd < 0)
	100	return -errno;
	101	}
9d647740	102
332076b4	103	if (fstat(fd, &st) < 0)
9d647740 ZJS	104	return -errno;
9d647740 ZJS	105
332076b4 LP	106	/* If an offline file doesn't even have a header we consider it empty */
	107	if (st.st_size < (off_t) sizeof(Header))
	108	return 1;
	109
	110	/* If the number of entries is empty, we consider it empty, too */
	111	n = pread(fd, &n_entries, sizeof(n_entries), offsetof(Header, n_entries));
	112	if (n < 0)
	113	return -errno;
	114	if (n != sizeof(n_entries))
	115	return -EIO;
9d647740	116
332076b4	117	return le64toh(n_entries) <= 0;
9d647740 ZJS	118	}
9d647740 ZJS	119
fb0951b0 LP	120	int journal_directory_vacuum(
	121	const char *directory,
	122	uint64_t max_use,
8580d1f7	123	uint64_t n_max_files,
fb0951b0	124	usec_t max_retention_usec,
dbd2a83f LP	125	usec_t *oldest_usec,
dbd2a83f LP	126	bool verbose) {
fb0951b0	127
30cb029b	128	_cleanup_closedir_ DIR *d = NULL;
0284adc6	129	struct vacuum_info *list = NULL;
8580d1f7	130	unsigned n_list = 0, i, n_active_files = 0;
30cb029b	131	size_t n_allocated = 0;
289f910e	132	uint64_t sum = 0, freed = 0;
fb0951b0	133	usec_t retention_limit = 0;
dbd2a83f	134	char sbytes[FORMAT_BYTES_MAX];
8580d1f7 LP	135	struct dirent *de;
8580d1f7 LP	136	int r;
0284adc6 LP	137
	138	assert(directory);
	139
8580d1f7	140	if (max_use <= 0 && max_retention_usec <= 0 && n_max_files <= 0)
0284adc6 LP	141	return 0;
0284adc6 LP	142
fb0951b0 LP	143	if (max_retention_usec > 0) {
	144	retention_limit = now(CLOCK_REALTIME);
	145	if (retention_limit > max_retention_usec)
	146	retention_limit -= max_retention_usec;
	147	else
	148	max_retention_usec = retention_limit = 0;
	149	}
	150
0284adc6 LP	151	d = opendir(directory);
	152	if (!d)
	153	return -errno;
	154
8580d1f7 LP	155	FOREACH_DIRENT_ALL(de, d, r = -errno; goto finish) {
8580d1f7 LP	156
0284adc6	157	unsigned long long seqnum = 0, realtime;
8580d1f7	158	_cleanup_free_ char *p = NULL;
0284adc6 LP	159	sd_id128_t seqnum_id;
0284adc6 LP	160	bool have_seqnum;
8580d1f7 LP	161	uint64_t size;
	162	struct stat st;
	163	size_t q;
0284adc6	164
8580d1f7 LP	165	if (fstatat(dirfd(d), de->d_name, &st, AT_SYMLINK_NOFOLLOW) < 0) {
8580d1f7 LP	166	log_debug_errno(errno, "Failed to stat file %s while vacuuming, ignoring: %m", de->d_name);
0284adc6	167	continue;
8580d1f7	168	}
0284adc6 LP	169
	170	if (!S_ISREG(st.st_mode))
	171	continue;
	172
	173	q = strlen(de->d_name);
	174
	175	if (endswith(de->d_name, ".journal")) {
	176
8580d1f7 LP	177	/* Vacuum archived files. Active files are
8580d1f7 LP	178	* left around */
0284adc6	179
8580d1f7 LP	180	if (q < 1 + 32 + 1 + 16 + 1 + 16 + 8) {
8580d1f7 LP	181	n_active_files++;
0284adc6	182	continue;
8580d1f7	183	}
0284adc6 LP	184
	185	if (de->d_name[q-8-16-1] != '-' \|\|
	186	de->d_name[q-8-16-1-16-1] != '-' \|\|
8580d1f7 LP	187	de->d_name[q-8-16-1-16-1-32-1] != '@') {
8580d1f7 LP	188	n_active_files++;
0284adc6	189	continue;
8580d1f7	190	}
0284adc6 LP	191
	192	p = strdup(de->d_name);
	193	if (!p) {
	194	r = -ENOMEM;
	195	goto finish;
	196	}
	197
	198	de->d_name[q-8-16-1-16-1] = 0;
	199	if (sd_id128_from_string(de->d_name + q-8-16-1-16-1-32, &seqnum_id) < 0) {
8580d1f7	200	n_active_files++;
0284adc6 LP	201	continue;
	202	}
	203
	204	if (sscanf(de->d_name + q-8-16-1-16, "%16llx-%16llx.journal", &seqnum, &realtime) != 2) {
8580d1f7	205	n_active_files++;
0284adc6 LP	206	continue;
	207	}
	208
	209	have_seqnum = true;
	210
	211	} else if (endswith(de->d_name, ".journal~")) {
	212	unsigned long long tmp;
	213
	214	/* Vacuum corrupted files */
	215
8580d1f7	216	if (q < 1 + 16 + 1 + 16 + 8 + 1) {
313cefa1	217	n_active_files++;
0284adc6	218	continue;
8580d1f7	219	}
0284adc6 LP	220
0284adc6 LP	221	if (de->d_name[q-1-8-16-1] != '-' \|\|
8580d1f7	222	de->d_name[q-1-8-16-1-16-1] != '@') {
313cefa1	223	n_active_files++;
0284adc6	224	continue;
8580d1f7	225	}
0284adc6 LP	226
	227	p = strdup(de->d_name);
	228	if (!p) {
	229	r = -ENOMEM;
	230	goto finish;
	231	}
	232
	233	if (sscanf(de->d_name + q-1-8-16-1-16, "%16llx-%16llx.journal~", &realtime, &tmp) != 2) {
313cefa1	234	n_active_files++;
0284adc6 LP	235	continue;
	236	}
	237
	238	have_seqnum = false;
8580d1f7 LP	239	} else {
	240	/* We do not vacuum unknown files! */
	241	log_debug("Not vacuuming unknown file %s.", de->d_name);
0284adc6	242	continue;
8580d1f7	243	}
0284adc6	244
8580d1f7	245	size = 512UL * (uint64_t) st.st_blocks;
9d647740	246
8580d1f7 LP	247	r = journal_file_empty(dirfd(d), p);
	248	if (r < 0) {
	249	log_debug_errno(r, "Failed check if %s is empty, ignoring: %m", p);
	250	continue;
	251	}
	252	if (r > 0) {
	253	/* Always vacuum empty non-online files. */
289f910e	254
47c073aa LP	255	r = unlinkat_deallocate(dirfd(d), p, 0);
47c073aa LP	256	if (r >= 0) {
8580d1f7 LP	257
	258	log_full(verbose ? LOG_INFO : LOG_DEBUG,
	259	"Deleted empty archived journal %s/%s (%s).", directory, p, format_bytes(sbytes, sizeof(sbytes), size));
	260
289f910e	261	freed += size;
47c073aa LP	262	} else if (r != -ENOENT)
47c073aa LP	263	log_warning_errno(r, "Failed to delete empty archived journal %s/%s: %m", directory, p);
289f910e	264
9d647740 ZJS	265	continue;
	266	}
	267
8580d1f7	268	patch_realtime(dirfd(d), p, &st, &realtime);
fb0951b0	269
26d8ff04	270	if (!GREEDY_REALLOC(list, n_allocated, n_list + 1)) {
26d8ff04 LP	271	r = -ENOMEM;
	272	goto finish;
	273	}
0284adc6	274
1cc6c93a	275	list[n_list].filename = TAKE_PTR(p);
8580d1f7	276	list[n_list].usage = size;
0284adc6 LP	277	list[n_list].seqnum = seqnum;
	278	list[n_list].realtime = realtime;
	279	list[n_list].seqnum_id = seqnum_id;
	280	list[n_list].have_seqnum = have_seqnum;
313cefa1	281	n_list++;
8580d1f7	282
8580d1f7	283	sum += size;
0284adc6 LP	284	}
0284adc6 LP	285
93bab288	286	typesafe_qsort(list, n_list, vacuum_compare);
0284adc6	287
fb0951b0	288	for (i = 0; i < n_list; i++) {
8580d1f7 LP	289	unsigned left;
	290
	291	left = n_active_files + n_list - i;
	292
fb0951b0	293	if ((max_retention_usec <= 0 \|\| list[i].realtime >= retention_limit) &&
8580d1f7 LP	294	(max_use <= 0 \|\| sum <= max_use) &&
8580d1f7 LP	295	(n_max_files <= 0 \|\| left <= n_max_files))
0284adc6 LP	296	break;
0284adc6 LP	297
47c073aa LP	298	r = unlinkat_deallocate(dirfd(d), list[i].filename, 0);
47c073aa LP	299	if (r >= 0) {
dbd2a83f	300	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	301	freed += list[i].usage;
4fa25d62	302
6c142648	303	if (list[i].usage < sum)
4fa25d62 LP	304	sum -= list[i].usage;
	305	else
	306	sum = 0;
	307
47c073aa LP	308	} else if (r != -ENOENT)
47c073aa LP	309	log_warning_errno(r, "Failed to delete archived journal %s/%s: %m", directory, list[i].filename);
0284adc6 LP	310	}
0284adc6 LP	311
fb0951b0 LP	312	if (oldest_usec && i < n_list && (oldest_usec == 0 \|\| list[i].realtime < oldest_usec))
	313	*oldest_usec = list[i].realtime;
	314
8580d1f7 LP	315	r = 0;
8580d1f7 LP	316
0284adc6 LP	317	finish:
	318	for (i = 0; i < n_list; i++)
	319	free(list[i].filename);
0284adc6 LP	320	free(list);
0284adc6 LP	321
3cc44bf9	322	log_full(verbose ? LOG_INFO : LOG_DEBUG, "Vacuuming done, freed %s of archived journals from %s.", format_bytes(sbytes, sizeof(sbytes), freed), directory);
289f910e	323
0284adc6 LP	324	return r;
0284adc6 LP	325	}