From 30878268f35298b7a62da5ae0f81024c37b134f7 Mon Sep 17 00:00:00 2001 From: Jan Schmidt Date: Wed, 30 Mar 2011 18:53:12 +0200 Subject: [PATCH 19/32] scrub userland implementation Signed-off-by: Jan Schmidt --- scrub.c | 1568 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 files changed, 1568 insertions(+), 0 deletions(-) create mode 100644 scrub.c diff --git a/scrub.c b/scrub.c new file mode 100644 index 0000000..22052ed --- /dev/null +++ b/scrub.c @@ -0,0 +1,1568 @@ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include "ctree.h" +#include "ioctl.h" +#include "btrfs_cmds.h" +#include "utils.h" +#include "volumes.h" +#include "disk-io.h" + +#define SCRUB_DATA_FILE "/var/btrfs/scrub.status" +#define SCRUB_PROGRESS_SOCKET_PATH "/var/btrfs/scrub.progress" +#define SCRUB_FILE_VERSION_PREFIX "scrub status:" +#define SCRUB_FILE_VERSION "1" + +struct scrub_stats { + time_t t_start; + time_t t_resumed; + u64 duration; + u64 finished; + u64 canceled; +}; + +struct scrub_progress { + struct btrfs_ioctl_scrub_args scrub_args; + int fd; + int ret; + int skip; + struct scrub_stats stats; + struct scrub_file_record *resumed; + int ioctl_errno; + pthread_mutex_t progress_mutex; +}; + +struct scrub_file_record { + u8 fsid[BTRFS_FSID_SIZE]; + u64 devid; + struct scrub_stats stats; + struct btrfs_scrub_progress p; +}; + +struct scrub_progress_cycle { + int fdmnt; + int prg_fd; + int do_record; + struct btrfs_ioctl_fs_info_args *fi; + struct scrub_progress *progress; + struct scrub_progress *shared_progress; + pthread_mutex_t *write_mutex; +}; + +struct scrub_fs_stat { + struct btrfs_scrub_progress p; + struct scrub_stats s; + int i; +}; + +static void print_scrub_full(struct btrfs_scrub_progress *sp) +{ + printf("\tdata_extents_scrubbed: %lld\n", sp->data_extents_scrubbed); + printf("\ttree_extents_scrubbed: %lld\n", sp->tree_extents_scrubbed); + printf("\tdata_bytes_scrubbed: %lld\n", sp->data_bytes_scrubbed); + printf("\ttree_bytes_scrubbed: %lld\n", sp->tree_bytes_scrubbed); + printf("\tread_errors: %lld\n", sp->read_errors); + printf("\tcsum_errors: %lld\n", sp->csum_errors); + printf("\tverify_errors: %lld\n", sp->verify_errors); + printf("\tno_csum: %lld\n", sp->no_csum); + printf("\tcsum_discards: %lld\n", sp->csum_discards); + printf("\tsuper_errors: %lld\n", sp->super_errors); + printf("\tmalloc_errors: %lld\n", sp->malloc_errors); + printf("\tuncorrectable_errors: %lld\n", sp->uncorrectable_errors); + printf("\tcorrected_errors: %lld\n", sp->corrected_errors); + printf("\tlast_physical: %lld\n", sp->last_physical); +} + +#define err(test, ...) do { \ + if (test) \ + fprintf(stderr, __VA_ARGS__); \ +} while (0) + +#define PRINT_SCRUB_ERROR(test, desc) do { \ + if (test) \ + printf(" %s=%llu", desc, test); \ +} while (0) +static void print_scrub_summary(struct btrfs_scrub_progress *p) +{ + u64 err_cnt; + u64 err_cnt2; + + err_cnt = p->read_errors + + p->csum_errors + + p->verify_errors + + p->csum_discards + + p->super_errors + + p->malloc_errors; + + err_cnt2 = p->corrected_errors + p->uncorrectable_errors; + + printf("\ttotal bytes scrubbed: %s with %llu errors\n", + pretty_sizes(p->data_bytes_scrubbed + p->tree_bytes_scrubbed), + max(err_cnt, err_cnt2)); + if (err_cnt || err_cnt2) { + printf("\terror details:"); + PRINT_SCRUB_ERROR(p->read_errors, "read"); + PRINT_SCRUB_ERROR(p->super_errors, "super"); + PRINT_SCRUB_ERROR(p->malloc_errors, "malloc"); + PRINT_SCRUB_ERROR(p->verify_errors, "verify"); + PRINT_SCRUB_ERROR(p->csum_errors, "csum"); + PRINT_SCRUB_ERROR(p->csum_discards, "csum-discards"); + printf("\n"); + printf("\tcorrected errors: %llu, uncorrectable errors: %llu\n", + p->corrected_errors, p->uncorrectable_errors); + } +} + +#define _SCRUB_FS_STAT(p, name, fs_stat) fs_stat->p.name += p->name +#define _SCRUB_FS_STAT_MIN(ss, name, fs_stat) \ +do { \ + if (fs_stat->s.name > ss->name) { \ + fs_stat->s.name = ss->name; \ + } \ +} while (0) +#define _SCRUB_FS_STAT_ZMIN(ss, name, fs_stat) \ +do { \ + if (!fs_stat->s.name || fs_stat->s.name > ss->name) { \ + fs_stat->s.name = ss->name; \ + } \ +} while (0) +#define _SCRUB_FS_STAT_MAX(ss, name, fs_stat) \ +do { \ + if (!fs_stat->s.name || fs_stat->s.name < ss->name) { \ + fs_stat->s.name = ss->name; \ + } \ +} while (0) +static void add_to_fs_stat(struct btrfs_scrub_progress *p, + struct scrub_stats *ss, + struct scrub_fs_stat *fs_stat) +{ + _SCRUB_FS_STAT(p, data_extents_scrubbed, fs_stat); + _SCRUB_FS_STAT(p, tree_extents_scrubbed, fs_stat); + _SCRUB_FS_STAT(p, data_bytes_scrubbed, fs_stat); + _SCRUB_FS_STAT(p, tree_bytes_scrubbed, fs_stat); + _SCRUB_FS_STAT(p, read_errors, fs_stat); + _SCRUB_FS_STAT(p, csum_errors, fs_stat); + _SCRUB_FS_STAT(p, verify_errors, fs_stat); + _SCRUB_FS_STAT(p, no_csum, fs_stat); + _SCRUB_FS_STAT(p, csum_discards, fs_stat); + _SCRUB_FS_STAT(p, super_errors, fs_stat); + _SCRUB_FS_STAT(p, malloc_errors, fs_stat); + _SCRUB_FS_STAT(p, uncorrectable_errors, fs_stat); + _SCRUB_FS_STAT(p, corrected_errors, fs_stat); + _SCRUB_FS_STAT(p, last_physical, fs_stat); + _SCRUB_FS_STAT_ZMIN(ss, t_start, fs_stat); + _SCRUB_FS_STAT_ZMIN(ss, t_resumed, fs_stat); + _SCRUB_FS_STAT_MAX(ss, duration, fs_stat); + _SCRUB_FS_STAT_MAX(ss, canceled, fs_stat); + _SCRUB_FS_STAT_MIN(ss, finished, fs_stat); +} + +static void init_fs_stat(struct scrub_fs_stat *fs_stat) +{ + memset(fs_stat, 0, sizeof(*fs_stat)); + fs_stat->s.finished = 2; +} + +static void _print_scrub_ss(struct scrub_stats *ss) +{ + char t[BTRFS_PATH_NAME_MAX+1]; + struct tm tm; + + if (!ss || !ss->t_start) { + printf("\tno stats available\n"); + return; + } + if (ss->t_resumed) { + localtime_r(&ss->t_resumed, &tm); + strftime(t, sizeof(t), "%c", &tm); + printf("\tscrub resumed at %s", t); + } else { + localtime_r(&ss->t_start, &tm); + strftime(t, sizeof(t), "%c", &tm); + printf("\tscrub started at %s", t); + } + if (ss->finished && !ss->canceled) { + printf(" and finished after %llu seconds\n", + ss->duration); + } else if (ss->canceled) { + printf(" and was aborted after %llu seconds\n", + ss->duration); + } else { + printf(", running for %llu seconds\n", ss->duration); + } +} + +static void print_scrub_dev(struct btrfs_ioctl_dev_info_args *di, + struct btrfs_scrub_progress *p, int raw, + const char *append, struct scrub_stats *ss) +{ + printf("scrub device %s (id %llu) %s\n", di->path, di->devid, + append ? append : ""); + + _print_scrub_ss(ss); + + if (p) { + if (raw) + print_scrub_full(p); + else + print_scrub_summary(p); + } +} + +static void print_fs_stat(struct scrub_fs_stat *fs_stat, int raw) +{ + _print_scrub_ss(&fs_stat->s); + + if (raw) + print_scrub_full(&fs_stat->p); + else + print_scrub_summary(&fs_stat->p); +} + +static void free_history(struct scrub_file_record **last_scrubs) +{ + struct scrub_file_record **l = last_scrubs; + if (!l) + return; + while (*l) + free(*l++); + free(last_scrubs); +} + +static int cancel_fd = -1; +static void scrub_sigint_record_progress(int signal) +{ + ioctl(cancel_fd, BTRFS_IOC_SCRUB_CANCEL, NULL); +} + +static int scrub_handle_sigint_parent(void) +{ + struct sigaction sa = { + .sa_handler = SIG_IGN, + .sa_flags = SA_RESTART, + }; + + return sigaction(SIGINT, &sa, NULL); +} + +static int scrub_handle_sigint_child(int fd) +{ + struct sigaction sa = { + .sa_handler = fd == -1 ? SIG_DFL : scrub_sigint_record_progress, + }; + + cancel_fd = fd; + return sigaction(SIGINT, &sa, NULL); +} + +static int _scrub_datafile(const char *fn_base, const char *fn_local, + const char *fn_tmp, char *datafile, int max) +{ + int ret; + + strncpy(datafile, fn_base, max); + ret = strlen(datafile); + + if (ret + 1 >= max) + return -EOVERFLOW; + + datafile[ret] = '.'; + strncpy(datafile+ret+1, fn_local, max-ret-1); + ret = strlen(datafile); + + if (ret + 1 >= max) + return -EOVERFLOW; + + if (fn_tmp) { + datafile[ret] = '_'; + strncpy(datafile+ret+1, fn_tmp, max-ret-1); + ret = strlen(datafile); + + if (ret >= max) + return -EOVERFLOW; + } + + return 0; +} + +static int _scrub_open_file(const char *datafile, int m) +{ + int fd; + int ret; + + fd = open(datafile, m, 0600); + if (fd < 0) + return -errno; + + ret = flock(fd, LOCK_EX|LOCK_NB); + if (ret) { + ret = errno; + close(fd); + return -ret; + } + + return fd; +} + +static int scrub_open_file_r(const char *fn_base, const char *fn_local) +{ + int ret; + char datafile[BTRFS_PATH_NAME_MAX+1]; + ret = _scrub_datafile(fn_base, fn_local, NULL, + datafile, sizeof(datafile)); + if (ret < 0) + return ret; + return _scrub_open_file(datafile, O_RDONLY); +} + +static int scrub_open_file_w(const char *fn_base, const char *fn_local, + const char *tmp) +{ + int ret; + char datafile[BTRFS_PATH_NAME_MAX+1]; + ret = _scrub_datafile(fn_base, fn_local, tmp, + datafile, sizeof(datafile)); + if (ret < 0) + return ret; + return _scrub_open_file(datafile, O_WRONLY|O_CREAT); +} + +static int scrub_rename_file(const char *fn_base, const char *fn_local, + const char *tmp) +{ + int ret; + char datafile_old[BTRFS_PATH_NAME_MAX+1]; + char datafile_new[BTRFS_PATH_NAME_MAX+1]; + ret = _scrub_datafile(fn_base, fn_local, tmp, + datafile_old, sizeof(datafile_old)); + if (ret < 0) + return ret; + ret = _scrub_datafile(fn_base, fn_local, NULL, + datafile_new, sizeof(datafile_new)); + if (ret < 0) + return ret; + ret = rename(datafile_old, datafile_new); + return ret ? -errno : 0; +} + +#define _SCRUB_KVREAD(i, name, avail, l, dest) \ + _scrub_kvread(i, sizeof(#name), avail, l, #name, dest.name) +#define _SCRUB_KVREAD_STATS(i, name, avail, l, dest) \ + _scrub_kvread(i, sizeof(#name), avail, l, #name, dest->stats.name) +/* + * returns 0 if the key did not match (nothing was read) + * 1 if the key did match (success) + * -1 if the key did match and an error occured + */ +static int _scrub_kvread(int *i, int len, int avail, const char *buf, + const char *key, u64 *dest) +{ + int j; + + if (*i+len+1 < avail && strncmp(&buf[*i], key, len-1) == 0) { + *i += len-1; + if (buf[*i] != ':') { + return -1; + } + *i += 1; + for (j=0; isdigit(buf[*i+j]) && *i+j < avail; ++j) + ; + if (*i+j >= avail) + return -1; + *dest = atoll(&buf[*i]); + *i += j; + return 1; + } + + return 0; +} + +#define _SCRUB_ILLEGAL do { \ + if (report_errors) { \ + fprintf(stderr, "WARNING: illegal data in line %d pos " \ + "%d state %d (near \"%.*s\") at %s:%d\n", \ + lineno, i, state, 20 > avail ? avail : 20, l+i, \ + __FILE__, __LINE__); \ + } \ + goto skip; \ +} while (0) +static struct scrub_file_record **scrub_read_file(int fd, int report_errors) +{ + int avail = 0; + int old_avail = 0; + char l[512]; + int state = 0; + int curr = -1; + int i = 0; + int j; + int ret; + int eof = 0; + int lineno = 0; + u64 version; + char empty_uuid[BTRFS_FSID_SIZE] = {0}; + struct scrub_file_record **p = NULL; + + if (fd < 0) + return ERR_PTR(-EINVAL); + +again: + old_avail = avail-i; + BUG_ON(old_avail < 0); + if (old_avail) + memmove(l, l+i, old_avail); + avail = read(fd, l+old_avail, sizeof(l)-old_avail); + if (avail == 0) { + eof = 1; + } + if (avail + old_avail == 0) { + if (curr >= 0 && + memcmp(p[curr]->fsid, empty_uuid, BTRFS_FSID_SIZE) == 0) { + p[curr] = NULL; + } else if (curr == -1) { + p = ERR_PTR(-ENODATA); + } + return p; + } + if (avail == -1) + return ERR_PTR(-errno); + avail += old_avail; + + i = 0; + while (i < avail) { + switch (state) { + case 0: /* start if file */ + ret = _scrub_kvread(&i, + sizeof(SCRUB_FILE_VERSION_PREFIX)-1, avail, l, + SCRUB_FILE_VERSION_PREFIX, &version); + if (ret != 1) + _SCRUB_ILLEGAL; + if (version != atoll(SCRUB_FILE_VERSION)) + return ERR_PTR(-ENOTSUP); + state = 6; + continue; + case 1: /* start of line, alloc */ + if (!eof && !memchr(l+i, '\n', avail-i)) + goto again; + ++lineno; + if (curr > -1 && memcmp(p[curr]->fsid, empty_uuid, + BTRFS_FSID_SIZE) == 0) { + state = 2; + continue; + } + ++curr; + p = realloc(p, (curr+2)*sizeof(*p)); + if (p) + p[curr] = malloc(sizeof(**p)); + if (!p || !p[curr]) + return ERR_PTR(-errno); + memset(p[curr], 0, sizeof(**p)); + p[curr+1] = NULL; + ++state; + case 2: /* start of line, skip space */ + while (isspace(l[i]) && i= avail || (!eof && !memchr(l+i, '\n', avail-i))) + goto again; + ++state; + case 3: /* read fsid */ + if (i == avail) + continue; + for (j=0; l[i+j] != ':' && i+j < avail; ++j) + ; + if (i+j+1 >= avail) + _SCRUB_ILLEGAL; + if (j != 36) + _SCRUB_ILLEGAL; + l[i+j] = '\0'; + ret = uuid_parse(l+i, p[curr]->fsid); + if (ret) + _SCRUB_ILLEGAL; + i += j + 1; + ++state; + case 4: /* read dev id */ + for (j=0; isdigit(l[i+j]) && i+j < avail; ++j) + ; + if (!j || i+j+1 >= avail) + _SCRUB_ILLEGAL; + p[curr]->devid = atoll(&l[i]); + i += j + 1; + ++state; + case 5: /* read key/value pair */ + ret = _SCRUB_KVREAD(&i, data_extents_scrubbed, avail, + l, &p[curr]->p) || + _SCRUB_KVREAD(&i, data_extents_scrubbed, avail, + l, &p[curr]->p) || + _SCRUB_KVREAD(&i, tree_extents_scrubbed, avail, + l, &p[curr]->p) || + _SCRUB_KVREAD(&i, data_bytes_scrubbed, avail, + l, &p[curr]->p) || + _SCRUB_KVREAD(&i, tree_bytes_scrubbed, avail, + l, &p[curr]->p) || + _SCRUB_KVREAD(&i, read_errors, avail, + l, &p[curr]->p) || + _SCRUB_KVREAD(&i, csum_errors, avail, + l, &p[curr]->p) || + _SCRUB_KVREAD(&i, verify_errors, avail, + l, &p[curr]->p) || + _SCRUB_KVREAD(&i, no_csum, avail, + l, &p[curr]->p) || + _SCRUB_KVREAD(&i, csum_discards, avail, + l, &p[curr]->p) || + _SCRUB_KVREAD(&i, super_errors, avail, + l, &p[curr]->p) || + _SCRUB_KVREAD(&i, malloc_errors, avail, + l, &p[curr]->p) || + _SCRUB_KVREAD(&i, uncorrectable_errors, avail, + l, &p[curr]->p) || + _SCRUB_KVREAD(&i, corrected_errors, avail, + l, &p[curr]->p) || + _SCRUB_KVREAD(&i, last_physical, avail, + l, &p[curr]->p) || + _SCRUB_KVREAD(&i, finished, avail, + l, &p[curr]->stats) || + _SCRUB_KVREAD(&i, t_start, avail, + l, (u64*)&p[curr]->stats) || + _SCRUB_KVREAD(&i, t_resumed, avail, + l, (u64*)&p[curr]->stats) || + _SCRUB_KVREAD(&i, duration, avail, + l, (u64*)&p[curr]->stats) || + _SCRUB_KVREAD(&i, canceled, avail, + l, &p[curr]->stats); + if (ret != 1) + _SCRUB_ILLEGAL; + ++state; + case 6: /* after number */ + if (l[i] == '|') { + state = 5; + } else if (l[i] == '\n') { + state = 1; + } else { + _SCRUB_ILLEGAL; + } + ++i; + continue; + case 99: /* skip rest of line */ +skip: + state = 99; + do { + ++i; + if (l[i-1] == '\n') { + state = 1; + break; + } + } while (i < avail); + continue; + } + BUG(); + } + goto again; +} +#undef _SCRUB_ILLEGAL + +static int _scrub_write_buf(int fd, const void *data, int len) +{ + int ret; + ret = write(fd, data, len); + return ret - len; +} + +static int _scrub_writev(int fd, char *buf, int max, const char *fmt, ...) + __attribute__ ((format (printf, 4, 5))); +static int _scrub_writev(int fd, char *buf, int max, const char *fmt, ...) +{ + int ret; + va_list args; + + va_start(args, fmt); + ret = vsnprintf(buf, max, fmt, args); + va_end(args); + if (ret >= max) + return ret - max; + return _scrub_write_buf(fd, buf, ret); +} + +#define _SCRUB_SUM(dest, data, name) dest->scrub_args.progress.name = \ + data->resumed->p.name + data->scrub_args.progress.name +static struct scrub_progress *_scrub_resumed_stats(struct scrub_progress *data, + struct scrub_progress *dest) +{ + if (!data->resumed || data->skip) + return data; + + _SCRUB_SUM(dest, data, data_extents_scrubbed); + _SCRUB_SUM(dest, data, tree_extents_scrubbed); + _SCRUB_SUM(dest, data, data_bytes_scrubbed); + _SCRUB_SUM(dest, data, tree_bytes_scrubbed); + _SCRUB_SUM(dest, data, read_errors); + _SCRUB_SUM(dest, data, csum_errors); + _SCRUB_SUM(dest, data, verify_errors); + _SCRUB_SUM(dest, data, no_csum); + _SCRUB_SUM(dest, data, csum_discards); + _SCRUB_SUM(dest, data, super_errors); + _SCRUB_SUM(dest, data, malloc_errors); + _SCRUB_SUM(dest, data, uncorrectable_errors); + _SCRUB_SUM(dest, data, corrected_errors); + _SCRUB_SUM(dest, data, last_physical); + dest->stats.canceled = data->stats.canceled; + dest->stats.finished = data->stats.finished; + dest->stats.t_resumed = data->stats.t_start; + dest->stats.t_start = data->resumed->stats.t_start; + dest->stats.duration = data->resumed->stats.duration + + data->stats.duration; + dest->scrub_args.devid = data->scrub_args.devid; + return dest; +} + +#define _SCRUB_KVWRITE(fd, buf, name, use) \ + _scrub_kvwrite(fd, buf, sizeof(buf), #name, \ + use->scrub_args.progress.name) +#define _SCRUB_KVWRITE_STATS(fd, buf, name, use) \ + _scrub_kvwrite(fd, buf, sizeof(buf), #name, \ + use->stats.name) +static int _scrub_kvwrite(int fd, char *buf, int max, + const char *key, u64 val) +{ + return _scrub_writev(fd, buf, max, "|%s:%lld", key, val); +} + +static int scrub_write_file(int fd, const char *fsid, + struct scrub_progress* data, int n) +{ + int ret = 0; + int i; + char buf[1024]; + struct scrub_progress local; + struct scrub_progress *use; + + if (n < 1) { + return -EINVAL; + } + + ret = _scrub_write_buf(fd, SCRUB_FILE_VERSION_PREFIX SCRUB_FILE_VERSION + "\n", sizeof(SCRUB_FILE_VERSION_PREFIX)-1 + + sizeof(SCRUB_FILE_VERSION)-1 + 1); + if (ret) + return -EOVERFLOW; + + for (i=0; iscrub_args.devid) || + _scrub_write_buf(fd, buf, ret) || + _SCRUB_KVWRITE(fd, buf, data_extents_scrubbed, use) || + _SCRUB_KVWRITE(fd, buf, tree_extents_scrubbed, use) || + _SCRUB_KVWRITE(fd, buf, data_bytes_scrubbed, use) || + _SCRUB_KVWRITE(fd, buf, tree_bytes_scrubbed, use) || + _SCRUB_KVWRITE(fd, buf, read_errors, use) || + _SCRUB_KVWRITE(fd, buf, csum_errors, use) || + _SCRUB_KVWRITE(fd, buf, verify_errors, use) || + _SCRUB_KVWRITE(fd, buf, no_csum, use) || + _SCRUB_KVWRITE(fd, buf, csum_discards, use) || + _SCRUB_KVWRITE(fd, buf, super_errors, use) || + _SCRUB_KVWRITE(fd, buf, malloc_errors, use) || + _SCRUB_KVWRITE(fd, buf, uncorrectable_errors, use) || + _SCRUB_KVWRITE(fd, buf, corrected_errors, use) || + _SCRUB_KVWRITE(fd, buf, last_physical, use) || + _SCRUB_KVWRITE_STATS(fd, buf, t_start, use) || + _SCRUB_KVWRITE_STATS(fd, buf, t_resumed, use) || + _SCRUB_KVWRITE_STATS(fd, buf, duration, use) || + _SCRUB_KVWRITE_STATS(fd, buf, canceled, use) || + _SCRUB_KVWRITE_STATS(fd, buf, finished, use) || + _scrub_write_buf(fd, "\n", 1)) { + return -EOVERFLOW; + } + } + + return 0; +} +#undef _SCRUB_KVWRITE + +static int scrub_write_progress(pthread_mutex_t *m, const char *fsid, + struct scrub_progress* data, int n) +{ + int ret; + int fd; + int old; + + ret = pthread_mutex_lock(m); + if (ret) { + ret = -errno; + goto out; + } + + pthread_setcancelstate(PTHREAD_CANCEL_DISABLE, &old); + + fd = scrub_open_file_w(SCRUB_DATA_FILE, fsid, "tmp"); + if (fd < 0) { + ret = fd; + goto out; + } + ret = scrub_write_file(fd, fsid, data, n); + if (ret) + goto out; + ret = scrub_rename_file(SCRUB_DATA_FILE, fsid, "tmp"); + if (ret) + goto out; + ret = close(fd); + if (ret) { + ret = -errno; + goto out; + } + +out: + if (ret) { + pthread_mutex_unlock(m); + } else { + ret = pthread_mutex_unlock(m); + if (ret) + ret = -errno; + } + + pthread_setcancelstate(PTHREAD_CANCEL_ENABLE, &old); + + return ret; +} + +static void *scrub_one_dev(void *ctx) +{ + struct scrub_progress *sp = ctx; + int ret; + struct timeval tv; + + sp->stats.canceled = 0; + sp->stats.duration = 0; + sp->stats.finished = 0; + + ret = ioctl(sp->fd, BTRFS_IOC_SCRUB, &sp->scrub_args); + gettimeofday(&tv, NULL); + sp->ret = ret; + sp->stats.duration = tv.tv_sec - sp->stats.t_start; + sp->stats.canceled = !!ret; + sp->ioctl_errno = errno; + ret = pthread_mutex_lock(&sp->progress_mutex); + if (ret) + return ERR_PTR(-errno); + sp->stats.finished = 1; + ret = pthread_mutex_unlock(&sp->progress_mutex); + if (ret) + return ERR_PTR(-errno); + + + return NULL; +} + +static void *progress_one_dev(void *ctx) +{ + struct scrub_progress *sp = ctx; + + sp->ret = ioctl(sp->fd, BTRFS_IOC_SCRUB_PROGRESS, &sp->scrub_args); + sp->ioctl_errno = errno; + + return NULL; +} + +static void *scrub_progress_cycle(void *ctx) +{ + int ret; + int i; + char fsid[37]; + struct scrub_progress *sp; + struct scrub_progress *sp_last; + struct scrub_progress *sp_shared; + struct timeval tv; + struct scrub_progress_cycle *spc = ctx; + int ndev = spc->fi->num_devices; + int this = 1; + int last = 0; + int peer_fd = -1; + struct pollfd accept_poll_fd = { + .fd = spc->prg_fd, + .events = POLLIN, + .revents = 0, + }; + struct pollfd write_poll_fd = { + .events = POLLOUT, + .revents = 0, + }; + struct sockaddr_un peer; + socklen_t peer_size = sizeof(peer); + + pthread_setcanceltype(PTHREAD_CANCEL_ASYNCHRONOUS, &ret); + uuid_unparse(spc->fi->fsid, fsid); + + for (i=0; iprogress[i]; + sp_last = &spc->progress[i+ndev]; + sp_shared = &spc->shared_progress[i]; + sp->scrub_args.devid = sp_last->scrub_args.devid = + sp_shared->scrub_args.devid; + sp->fd = sp_last->fd = spc->fdmnt; + sp->stats.t_start = sp_last->stats.t_start = + sp_shared->stats.t_start; + sp->resumed = sp_last->resumed = sp_shared->resumed; + sp->skip = sp_last->skip = sp_shared->skip; + sp->stats.finished = sp_last->stats.finished = + sp_shared->stats.finished; + } + + while (1) { + ret = poll(&accept_poll_fd, 1, 5*1000); + if (ret == -1) + return ERR_PTR(-errno); + if (ret) + peer_fd = accept(spc->prg_fd, (struct sockaddr *)&peer, + &peer_size); + gettimeofday(&tv, NULL); + this = (this+1)%2; + last = (last+1)%2; + for (i=0; iprogress[this*ndev+i]; + sp_last = &spc->progress[last*ndev+i]; + sp_shared = &spc->shared_progress[i]; + if (sp->stats.finished) { + continue; + } + progress_one_dev(sp); + sp->stats.duration = tv.tv_sec - sp->stats.t_start; + if (!sp->ret) + continue; + if (sp->ioctl_errno != ENOTCONN && + sp->ioctl_errno != ENODEV) + return ERR_PTR(-sp->ioctl_errno); + /* + * scrub finished or device removed, check the + * finished flag. if unset, just use the last + * result we got for the current write and go + * on. flag should be set on next cycle, then. + */ + ret = pthread_mutex_lock(&sp_shared->progress_mutex); + if (ret) + return ERR_PTR(-errno); + if (!sp_shared->stats.finished) { + ret = pthread_mutex_unlock( + &sp_shared->progress_mutex); + if (ret) + return ERR_PTR(-errno); + memcpy(sp, sp_last, sizeof(*sp)); + continue; + } + ret = pthread_mutex_unlock(&sp_shared->progress_mutex); + if (ret) + return ERR_PTR(-errno); + memcpy(sp, sp_shared, sizeof(*sp)); + memcpy(sp_last, sp_shared, sizeof(*sp)); + } + if (peer_fd != -1) { + write_poll_fd.fd = peer_fd; + ret = poll(&write_poll_fd, 1, 0); + if (ret == -1) + return ERR_PTR(-errno); + if (ret) { + ret = scrub_write_file( + peer_fd, fsid, + &spc->progress[this*ndev], ndev); + if (ret) + return ERR_PTR(ret); + } + close(peer_fd); + peer_fd = -1; + } + if (!spc->do_record) + continue; + ret = scrub_write_progress(spc->write_mutex, fsid, + &spc->progress[this*ndev], ndev); + if (ret) + return ERR_PTR(ret); + } +} + +static struct scrub_file_record *last_dev_scrub( + struct scrub_file_record *const *const past_scrubs, u64 devid) +{ + int i; + + if (!past_scrubs || IS_ERR(past_scrubs)) + return NULL; + + for (i=0; past_scrubs[i]; ++i) + if (past_scrubs[i]->devid == devid) + return past_scrubs[i]; + + return NULL; +} + +static int scrub_device_info(int fd, u64 devid, + struct btrfs_ioctl_dev_info_args *di_args) +{ + int ret; + + di_args->devid = devid; + memset(&di_args->uuid, '\0', sizeof(di_args->uuid)); + + ret = ioctl(fd, BTRFS_IOC_DEV_INFO, di_args); + return ret ? -errno : 0; +} + +static int scrub_fs_info(int fd, char *path, + struct btrfs_ioctl_fs_info_args *fi_args, + struct btrfs_ioctl_dev_info_args **di_ret) +{ + int ret = 0; + int ndevs = 0; + int i = 1; + struct btrfs_fs_devices* fs_devices_mnt = NULL; + struct btrfs_ioctl_dev_info_args *di_args; + char mp[BTRFS_PATH_NAME_MAX+1]; + + memset(fi_args, 0, sizeof(*fi_args)); + + ret = ioctl(fd, BTRFS_IOC_FS_INFO, fi_args); + if (ret && errno == EINVAL) { + /* path is no mounted btrfs. try if it's a device */ + ret = check_mounted_where(fd, path, mp, sizeof(mp), + &fs_devices_mnt); + if (!ret) + return -EINVAL; + fi_args->num_devices = 1; + fi_args->max_id = fs_devices_mnt->latest_devid; + i = fs_devices_mnt->latest_devid; + memcpy(fi_args->fsid, fs_devices_mnt->fsid, BTRFS_FSID_SIZE); + close(fd); + fd = open_file_or_dir(mp); + if (fd < 0) + return -errno; + } else if (ret) { + return -errno; + } + + if (!fi_args->num_devices) + return 0; + + di_args = *di_ret = malloc(fi_args->num_devices*sizeof(*di_args)); + if (!di_args) + return -errno; + + for (; i<=fi_args->max_id; ++i) { + BUG_ON(ndevs >= fi_args->num_devices); + ret = scrub_device_info(fd, i, &di_args[ndevs]); + if (ret == -ENODEV) + continue; + if (ret) + return ret; + ++ndevs; + } + + BUG_ON(ndevs == 0); + + return 0; +} + +int mkdir_p(char *path) +{ + int i; + int ret; + + for (i=1; i= 0) { + past_scrubs = scrub_read_file(fdres, !do_quiet); + if (IS_ERR(past_scrubs)) + err(!do_quiet, "WARNING: failed to read status file: " + "%s\n", strerror(-PTR_ERR(past_scrubs))); + close(fdres); + } + + t_devs = malloc(fi_args.num_devices*sizeof(*t_devs)); + sp = calloc(1, fi_args.num_devices*sizeof(*sp)); + spc.progress = calloc(1, fi_args.num_devices*2*sizeof(*spc.progress)); + + if (!t_devs || !sp || !spc.progress) { + err(!do_quiet, "ERROR: scrub failed: %s", strerror(errno)); + err = 1; + goto out; + } + + ret = pthread_attr_init(&t_attr); + if (ret) { + err(!do_quiet, "ERROR: pthread_attr_init failed: %s\n", + strerror(ret)); + err = 1; + goto out; + } + + for (i = 0; i < fi_args.num_devices; ++i) { + devid = di_args[i].devid; + ret = pthread_mutex_init(&sp[i].progress_mutex, NULL); + if (ret) { + err(!do_quiet, "ERROR: pthread_mutex_init failed: " + "%s\n", strerror(ret)); + err = 1; + goto out; + } + last_scrub = last_dev_scrub(past_scrubs, devid); + sp[i].scrub_args.devid = devid; + sp[i].fd = fdmnt; + if (resume && last_scrub && (last_scrub->stats.canceled || + !last_scrub->stats.finished)) { + ++n_resume; + sp[i].scrub_args.start = last_scrub->p.last_physical; + sp[i].resumed = last_scrub; + } else if (resume) { + ++n_skip; + sp[i].skip = 1; + sp[i].resumed = last_scrub; + continue; + } else { + ++n_start; + sp[i].scrub_args.start = 0ll; + sp[i].resumed = NULL; + } + sp[i].skip = 0; + sp[i].scrub_args.end = (u64)-1ll; + sp[i].scrub_args.flags = readonly ? BTRFS_SCRUB_READONLY : 0; + } + + if (!n_start && !n_resume) { + if (!do_quiet) + printf("scrub: nothing to resume for %s, fsid %s\n", + path, fsid); + err = 0; + goto out; + } + + ret = prg_fd = socket(AF_UNIX, SOCK_STREAM, 0); + while (ret != -1) { + _scrub_datafile(SCRUB_PROGRESS_SOCKET_PATH, fsid, + NULL, sock_path, sizeof(sock_path)); + /* ignore EOVERFLOW, as strncpy follows anyway */ + strncpy(addr.sun_path, sock_path, + sizeof(addr.sun_path)-1); + ret = bind(prg_fd, (struct sockaddr *)&addr, sizeof(addr)); + if (ret != -1 || errno != EADDRINUSE) + break; + ret = connect(prg_fd, (struct sockaddr *)&addr, sizeof(addr)); + if (!ret || errno != ECONNREFUSED) { + fprintf(stderr, "ERROR: scrub already running\n"); + close(prg_fd); + goto out; + } + ret = unlink(sock_path); + } + if (ret != -1) { + ret = listen(prg_fd, 100); + } + if (ret == -1) { + err(!do_quiet, "WARNING: failed to open the progress status " + "socket at %s: %s. Progress cannot be queried\n", + sock_path[0] ? sock_path : SCRUB_PROGRESS_SOCKET_PATH, + strerror(errno)); + if (prg_fd != -1) { + close(prg_fd); + prg_fd = -1; + if (sock_path[0]) + unlink(sock_path); + } + } + + if (do_record) { + /* write all-zero progress file for a start */ + ret = scrub_write_progress(&spc_write_mutex, fsid, sp, + fi_args.num_devices); + if (ret) { + err(!do_quiet, "WARNING: failed to write the progress " + "status file: %s. Status recording disabled\n", + strerror(-ret)); + do_record = 0; + } + } + + if (do_background) { + pid = fork(); + if (pid == -1) { + err(!do_quiet, "ERROR: cannot scrub, fork failed: " + "%s\n", strerror(errno)); + err = 1; + goto out; + } + + if (pid) { + int stat; + scrub_handle_sigint_parent(); + if (!do_quiet) + printf("scrub %s on %s, fsid %s (pid=%d)\n", + n_start ? "started" : "resumed", + path, fsid, pid); + if (!do_wait) { + err = 0; + goto out; + } + ret = wait(&stat); + if (ret != pid) { + err(!do_quiet, "ERROR: wait failed: (ret=%d) " + "%s\n", ret, strerror(errno)); + err = 1; + goto out; + } + if (!WIFEXITED(stat) || WEXITSTATUS(stat)) { + err(!do_quiet, "ERROR: scrub process failed\n"); + err = WIFEXITED(stat) ? WEXITSTATUS(stat) : -1; + goto out; + } + err = 0; + goto out; + } + } + + scrub_handle_sigint_child(fdmnt); + + for (i = 0; i < fi_args.num_devices; ++i) { + if (sp[i].skip) { + sp[i].scrub_args.progress = sp[i].resumed->p; + sp[i].stats = sp[i].resumed->stats; + sp[i].ret = 0; + sp[i].stats.finished = 1; + continue; + } + devid = di_args[i].devid; + gettimeofday(&tv, NULL); + sp[i].stats.t_start = tv.tv_sec; + ret = pthread_create(&t_devs[i], &t_attr, scrub_one_dev,&sp[i]); + if (ret) { + if (do_print) + fprintf(stderr, "ERROR: creating " + "scrub_one_dev[%llu] thread failed: " + "%s\n", devid, strerror(ret)); + err = 1; + goto out; + } + } + + spc.fdmnt = fdmnt; + spc.prg_fd = prg_fd; + spc.do_record = do_record; + spc.write_mutex = &spc_write_mutex; + spc.shared_progress = sp; + spc.fi = &fi_args; + pthread_create(&t_prog, &t_attr, scrub_progress_cycle, &spc); + + err = 0; + for (i = 0; i < fi_args.num_devices; ++i) { + if (sp[i].skip) + continue; + devid = di_args[i].devid; + ret = pthread_join(t_devs[i], NULL); + if (ret) { + if (do_print) + fprintf(stderr, "ERROR: pthread_join failed " + "for scrub_one_dev[%llu]: %s\n", devid, + strerror(ret)); + err++; + continue; + } + if (sp[i].ret && sp[i].ioctl_errno == ENODEV) { + if (do_print) + fprintf(stderr, "WARNING: device %lld not " + "present\n", devid); + continue; + } + if (sp[i].ret && sp[i].ioctl_errno == ECANCELED) { + err++; + } else if (sp[i].ret) { + if (do_print) + fprintf(stderr, "ERROR: scrubbing %s failed " + "for device id %lld (%s)\n", path, + devid, strerror(sp[i].ioctl_errno)); + err++; + continue; + } + } + + if (do_print) { + const char *append = "done"; + if (!do_stats_per_dev) + init_fs_stat(&fs_stat); + for (i = 0; i < fi_args.num_devices; ++i) { + if (do_stats_per_dev) { + print_scrub_dev(&di_args[i], + &sp[i].scrub_args.progress, + print_raw, + sp[i].ret ? "canceled" : "done", + &sp[i].stats); + } else { + if (sp[i].ret) + append = "canceled"; + add_to_fs_stat(&sp[i].scrub_args.progress, + &sp[i].stats, &fs_stat); + } + } + if (!do_stats_per_dev) { + printf("scrub %s for %s\n", append, fsid); + print_fs_stat(&fs_stat, print_raw); + } + } + + pthread_cancel(t_prog); + ret = pthread_join(t_prog, &terr); + if (do_print && terr && terr != PTHREAD_CANCELED) { + fprintf(stderr, "ERROR: recording progress " + "failed: %s\n", strerror(-PTR_ERR(terr))); + } + + if (do_record) { + ret = scrub_write_progress(&spc_write_mutex, fsid, sp, + fi_args.num_devices); + if (ret && do_print) { + fprintf(stderr, "ERROR: failed to record the result: " + "%s\n", strerror(-ret)); + } + } + + scrub_handle_sigint_child(-1); + +out: + free_history(past_scrubs); + free(di_args); + free(t_devs); + free(sp); + free(spc.progress); + if (prg_fd > -1) { + close(prg_fd); + if (sock_path[0]) + unlink(sock_path); + } + close(fdmnt); + + return !!err; +} + +int do_scrub_start(int argc, char **argv) +{ + return scrub_start(argc, argv, 0); +} + +int do_scrub_resume(int argc, char **argv) +{ + return scrub_start(argc, argv, 1); +} + +int do_scrub_cancel(int argc, char **argv) +{ + char *path = argv[1]; + int ret; + int fdmnt; + int err; + char mp[BTRFS_PATH_NAME_MAX+1]; + struct btrfs_fs_devices* fs_devices_mnt = NULL; + + fdmnt = open_file_or_dir(path); + if (fdmnt < 0) { + fprintf(stderr, "ERROR: scrub cancel failed\n"); + return 12; + } + +again: + ret = ioctl(fdmnt, BTRFS_IOC_SCRUB_CANCEL, NULL); + err = errno; + close(fdmnt); + + if (ret && err == EINVAL) { + /* path is no mounted btrfs. try if it's a device */ + ret = check_mounted_where(fdmnt, path, mp, sizeof(mp), + &fs_devices_mnt); + close(fdmnt); + if (ret) { + fdmnt = open_file_or_dir(mp); + if (fdmnt >= 0) { + path = mp; + goto again; + } + } + } + + if (ret) { + fprintf(stderr, "ERROR: scrub cancel failed on %s: %s\n", path, + err == ENOTCONN ? "not running" : strerror(errno)); + return 1; + } + + printf("scrub cancelled\n"); + + return 0; +} + +int do_scrub_status(int argc, char **argv) +{ + + char *path; + struct btrfs_ioctl_fs_info_args fi_args; + struct btrfs_ioctl_dev_info_args *di_args = NULL; + struct scrub_file_record **past_scrubs = NULL; + struct scrub_file_record *last_scrub; + struct scrub_fs_stat fs_stat; + struct sockaddr_un addr = { + .sun_family = AF_UNIX, + }; + int ret; + int fdmnt; + int i; + optind = 1; + int print_raw = 0; + int do_stats_per_dev = 0; + char c; + char fsid[37]; + int fdres = -1; + int err = 0; + + while ((c = getopt(argc, argv, "dR")) != -1) { + switch(c) { + case 'd': + do_stats_per_dev = 1; + break; + case 'R': + print_raw = 1; + break; + case '?': + default: + fprintf(stderr, "ERROR: scrub status args invalid.\n" + " -d stats per device\n"); + return 1; + } + } + + path = argv[optind]; + + fdmnt = open_file_or_dir(path); + if (fdmnt < 0) { + fprintf(stderr, "ERROR: can't access to '%s'\n", path); + return 12; + } + + ret = scrub_fs_info(fdmnt, path, &fi_args, &di_args); + if (ret) { + fprintf(stderr, "ERROR: getting dev info for scrub failed: " + "%s\n", strerror(-ret)); + err = 1; + goto out; + } + if (!fi_args.num_devices) { + fprintf(stderr, "ERROR: no devices found\n"); + err = 1; + goto out; + } + + uuid_unparse(fi_args.fsid, fsid); + + fdres = socket(AF_UNIX, SOCK_STREAM, 0); + if (fdres == -1) { + fprintf(stderr, "ERROR: failed to create socket to " + "receive progress information: %s\n", + strerror(errno)); + err = 1; + goto out; + } + _scrub_datafile(SCRUB_PROGRESS_SOCKET_PATH, fsid, + NULL, addr.sun_path, sizeof(addr.sun_path)-1); + /* ignore EOVERFLOW, just use shorter name and hope for the best */ + ret = connect(fdres, (struct sockaddr *)&addr, sizeof(addr)); + if (ret == -1) { + fdres = scrub_open_file_r(SCRUB_DATA_FILE, fsid); + if (fdres < 0 && fdres != -ENOENT) { + fprintf(stderr, "WARNING: failed to open status file: " + "%s\n", strerror(-fdres)); + err = 1; + goto out; + } + } + + if (fdres >= 0) { + past_scrubs = scrub_read_file(fdres, 1); + if (IS_ERR(past_scrubs)) + fprintf(stderr, "WARNING: failed to read status: %s\n", + strerror(-PTR_ERR(past_scrubs))); + } + + printf("scrub status for %s\n", fsid); + + /* + * TODO: rather communicate with scrub process instead of + * dumping the file stats for instant results + */ + if (do_stats_per_dev) { + for (i = 0; i < fi_args.num_devices; ++i) { + last_scrub = last_dev_scrub(past_scrubs, + di_args[i].devid); + if (!last_scrub) { + print_scrub_dev(&di_args[i], NULL, print_raw, + NULL, NULL); + continue; + } + print_scrub_dev(&di_args[i], &last_scrub->p, print_raw, + last_scrub->stats.finished ? + "history" : "status", + &last_scrub->stats); + } + } else { + init_fs_stat(&fs_stat); + for (i = 0; i < fi_args.num_devices; ++i) { + last_scrub = last_dev_scrub(past_scrubs, + di_args[i].devid); + if (!last_scrub) + continue; + add_to_fs_stat(&last_scrub->p, &last_scrub->stats, + &fs_stat); + } + print_fs_stat(&fs_stat, print_raw); + } + +out: + free_history(past_scrubs); + free(di_args); + close(fdmnt); + if (fdres > -1) + close(fdres); + + return err; +} -- 1.7.5.2.353.g5df3e