static int cmd_scrub_cancel(int argc, char **argv)
{
char *path;
int ret;
int fdmnt = -1;
DIR *dirstream = NULL;
if (check_argc_exact(argc, 2))
usage(cmd_scrub_cancel_usage);
path = argv[1];
fdmnt = open_path_or_dev_mnt(path, &dirstream);
if (fdmnt < 0) {
if (errno == EINVAL)
fprintf(stderr,
"ERROR: '%s' is not a mounted btrfs device\n",
path);
else
fprintf(stderr, "ERROR: can't access '%s': %s\n",
path, strerror(errno));
ret = 1;
goto out;
}
ret = ioctl(fdmnt, BTRFS_IOC_SCRUB_CANCEL, NULL);
if (ret < 0) {
fprintf(stderr, "ERROR: scrub cancel failed on %s: %s\n", path,
errno == ENOTCONN ? "not running" : strerror(errno));
if (errno == ENOTCONN)
ret = 2;
else
ret = 1;
goto out;
}
ret = 0;
printf("scrub cancelled\n");
out:
close_file_or_dir(fdmnt, dirstream);
return ret;
}
static int cmd_scrub_cancel(int argc, char **argv)
{
char *path;
int ret;
int fdmnt = -1;
DIR *dirstream = NULL;
clean_args_no_options(argc, argv, cmd_scrub_cancel_usage);
if (check_argc_exact(argc - optind, 1))
usage(cmd_scrub_cancel_usage);
path = argv[optind];
fdmnt = open_path_or_dev_mnt(path, &dirstream, 1);
if (fdmnt < 0) {
ret = 1;
goto out;
}
ret = ioctl(fdmnt, BTRFS_IOC_SCRUB_CANCEL, NULL);
if (ret < 0) {
error("scrub cancel failed on %s: %s", path,
errno == ENOTCONN ? "not running" : strerror(errno));
if (errno == ENOTCONN)
ret = 2;
else
ret = 1;
goto out;
}
ret = 0;
printf("scrub cancelled\n");
out:
close_file_or_dir(fdmnt, dirstream);
return ret;
}
static int cmd_scrub_cancel(int argc, char **argv)
{
char *path;
int ret;
int fdmnt = -1;
if (check_argc_exact(argc, 2))
usage(cmd_scrub_cancel_usage);
path = argv[1];
fdmnt = open_path_or_dev_mnt(path);
if (fdmnt < 0) {
fprintf(stderr, "ERROR: could not open %s: %s\n",
path, strerror(errno));
ret = 1;
goto out;
}
ret = ioctl(fdmnt, BTRFS_IOC_SCRUB_CANCEL, NULL);
if (ret < 0) {
fprintf(stderr, "ERROR: scrub cancel failed on %s: %s\n", path,
errno == ENOTCONN ? "not running" : strerror(errno));
ret = 1;
goto out;
}
ret = 0;
printf("scrub cancelled\n");
out:
if (fdmnt != -1)
close(fdmnt);
return ret;
}
static int cmd_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 i;
int fdmnt;
int print_raw = 0;
int do_stats_per_dev = 0;
int c;
char fsid[37];
int fdres = -1;
int err = 0;
optind = 1;
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:
usage(cmd_scrub_status_usage);
}
}
if (check_argc_exact(argc - optind, 1))
usage(cmd_scrub_status_usage);
path = argv[optind];
fdmnt = open_path_or_dev_mnt(path);
if (fdmnt < 0) {
fprintf(stderr, "ERROR: can't access to '%s'\n", path);
return 12;
}
ret = get_fs_info(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));
/* ignore EOVERFLOW, just use shorter name and hope for the best */
addr.sun_path[sizeof(addr.sun_path) - 1] = '\0';
ret = connect(fdres, (struct sockaddr *)&addr, sizeof(addr));
if (ret == -1) {
close(fdres);
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);
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);
if (fdres > -1)
close(fdres);
return err;
}
static int scrub_start(int argc, char **argv, int resume)
{
int fdmnt;
int prg_fd = -1;
int fdres = -1;
int ret;
pid_t pid;
int c;
int i;
int err = 0;
int e_uncorrectable = 0;
int e_correctable = 0;
int print_raw = 0;
char *path;
int do_background = 1;
int do_wait = 0;
int do_print = 0;
int do_quiet = 0;
int do_record = 1;
int readonly = 0;
int do_stats_per_dev = 0;
int ioprio_class = IOPRIO_CLASS_IDLE;
int ioprio_classdata = 0;
int n_start = 0;
int n_skip = 0;
int n_resume = 0;
struct btrfs_ioctl_fs_info_args fi_args;
struct btrfs_ioctl_dev_info_args *di_args = NULL;
struct scrub_progress *sp = NULL;
struct scrub_fs_stat fs_stat;
struct timeval tv;
struct sockaddr_un addr = {
.sun_family = AF_UNIX,
};
pthread_t *t_devs = NULL;
pthread_t t_prog;
pthread_attr_t t_attr;
struct scrub_file_record **past_scrubs = NULL;
struct scrub_file_record *last_scrub = NULL;
char *datafile = strdup(SCRUB_DATA_FILE);
char fsid[37];
char sock_path[BTRFS_PATH_NAME_MAX + 1] = "";
struct scrub_progress_cycle spc;
pthread_mutex_t spc_write_mutex = PTHREAD_MUTEX_INITIALIZER;
void *terr;
u64 devid;
optind = 1;
while ((c = getopt(argc, argv, "BdqrRc:n:")) != -1) {
switch (c) {
case 'B':
do_background = 0;
do_wait = 1;
do_print = 1;
break;
case 'd':
do_stats_per_dev = 1;
break;
case 'q':
do_quiet = 1;
break;
case 'r':
readonly = 1;
break;
case 'R':
print_raw = 1;
break;
case 'c':
ioprio_class = (int)strtol(optarg, NULL, 10);
break;
case 'n':
ioprio_classdata = (int)strtol(optarg, NULL, 10);
break;
case '?':
default:
usage(resume ? cmd_scrub_resume_usage :
cmd_scrub_start_usage);
}
}
/* try to catch most error cases before forking */
if (check_argc_exact(argc - optind, 1)) {
usage(resume ? cmd_scrub_resume_usage :
cmd_scrub_start_usage);
}
spc.progress = NULL;
if (do_quiet && do_print)
do_print = 0;
if (mkdir_p(datafile)) {
ERR(!do_quiet, "WARNING: cannot create scrub data "
"file, mkdir %s failed: %s. Status recording "
"disabled\n", datafile, strerror(errno));
do_record = 0;
}
free(datafile);
path = argv[optind];
fdmnt = open_path_or_dev_mnt(path);
if (fdmnt < 0) {
ERR(!do_quiet, "ERROR: can't access '%s'\n", path);
return 12;
}
ret = get_fs_info(path, &fi_args, &di_args);
if (ret) {
ERR(!do_quiet, "ERROR: getting dev info for scrub failed: "
"%s\n", strerror(-ret));
err = 1;
goto out;
}
if (!fi_args.num_devices) {
ERR(!do_quiet, "ERROR: no devices found\n");
err = 1;
goto out;
}
uuid_unparse(fi_args.fsid, fsid);
fdres = scrub_open_file_r(SCRUB_DATA_FILE, fsid);
if (fdres < 0 && fdres != -ENOENT) {
ERR(!do_quiet, "WARNING: failed to open status file: "
"%s\n", strerror(-fdres));
} else if (fdres >= 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(fi_args.num_devices, sizeof(*sp));
spc.progress = calloc(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;
sp[i].ioprio_class = ioprio_class;
sp[i].ioprio_classdata = ioprio_classdata;
}
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) {
ret = scrub_datafile(SCRUB_PROGRESS_SOCKET_PATH, fsid, NULL,
sock_path, sizeof(sock_path));
/* ignore EOVERFLOW, try using a shorter path for the socket */
addr.sun_path[sizeof(addr.sun_path) - 1] = '\0';
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;
/*
* bind failed with EADDRINUSE. so let's see if anyone answers
* when we make a call to the socket ...
*/
ret = connect(prg_fd, (struct sockaddr *)&addr, sizeof(addr));
if (!ret || errno != ECONNREFUSED) {
/* ... yes, so scrub must be running. error out */
fprintf(stderr, "ERROR: scrub already running\n");
close(prg_fd);
prg_fd = -1;
goto out;
}
/*
* ... no, this means someone left us alone with an unused
* socket in the file system. remove it and try again.
*/
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;
ret = pthread_create(&t_prog, &t_attr, scrub_progress_cycle, &spc);
if (ret) {
if (do_print)
fprintf(stderr, "ERROR: creating progress thread "
"failed: %s\n", strerror(ret));
err = 1;
goto out;
}
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 (sp[i].scrub_args.progress.uncorrectable_errors > 0)
e_uncorrectable++;
if (sp[i].scrub_args.progress.corrected_errors > 0
|| sp[i].scrub_args.progress.unverified_errors > 0)
e_correctable++;
}
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);
}
}
ret = pthread_cancel(t_prog);
if (!ret)
ret = pthread_join(t_prog, &terr);
/* check for errors from the handling of the progress thread */
if (do_print && ret) {
fprintf(stderr, "ERROR: progress thread handling failed: %s\n",
strerror(ret));
}
/* check for errors returned from the progress thread itself */
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);
if (err)
return 1;
if (e_correctable)
return 7;
if (e_uncorrectable)
return 8;
return 0;
}
static int cmd_device_stats(int argc, char **argv)
{
char *dev_path;
struct btrfs_ioctl_fs_info_args fi_args;
struct btrfs_ioctl_dev_info_args *di_args = NULL;
int ret;
int fdmnt;
int i;
int err = 0;
int check = 0;
__u64 flags = 0;
DIR *dirstream = NULL;
optind = 0;
while (1) {
int c;
static const struct option long_options[] = {
{"check", no_argument, NULL, 'c'},
{"reset", no_argument, NULL, 'z'},
{NULL, 0, NULL, 0}
};
c = getopt_long(argc, argv, "cz", long_options, NULL);
if (c < 0)
break;
switch (c) {
case 'c':
check = 1;
break;
case 'z':
flags = BTRFS_DEV_STATS_RESET;
break;
case '?':
default:
usage(cmd_device_stats_usage);
}
}
if (check_argc_exact(argc - optind, 1))
usage(cmd_device_stats_usage);
dev_path = argv[optind];
fdmnt = open_path_or_dev_mnt(dev_path, &dirstream, 1);
if (fdmnt < 0)
return 1;
ret = get_fs_info(dev_path, &fi_args, &di_args);
if (ret) {
errno = -ret;
error("getting device info for %s failed: %m", dev_path);
err = 1;
goto out;
}
if (!fi_args.num_devices) {
error("no devices found");
err = 1;
goto out;
}
for (i = 0; i < fi_args.num_devices; i++) {
struct btrfs_ioctl_get_dev_stats args = {0};
char path[BTRFS_DEVICE_PATH_NAME_MAX + 1];
strncpy(path, (char *)di_args[i].path,
BTRFS_DEVICE_PATH_NAME_MAX);
path[BTRFS_DEVICE_PATH_NAME_MAX] = 0;
args.devid = di_args[i].devid;
args.nr_items = BTRFS_DEV_STAT_VALUES_MAX;
args.flags = flags;
if (ioctl(fdmnt, BTRFS_IOC_GET_DEV_STATS, &args) < 0) {
error("device stats ioctl failed on %s: %m",
path);
err |= 1;
} else {
char *canonical_path;
int j;
static const struct {
const char name[32];
u64 num;
} dev_stats[] = {
{ "write_io_errs", BTRFS_DEV_STAT_WRITE_ERRS },
{ "read_io_errs", BTRFS_DEV_STAT_READ_ERRS },
{ "flush_io_errs", BTRFS_DEV_STAT_FLUSH_ERRS },
{ "corruption_errs",
BTRFS_DEV_STAT_CORRUPTION_ERRS },
{ "generation_errs",
BTRFS_DEV_STAT_GENERATION_ERRS },
};
canonical_path = canonicalize_path(path);
/* No path when device is missing. */
if (!canonical_path) {
canonical_path = malloc(32);
if (!canonical_path) {
error("not enough memory for path buffer");
goto out;
}
snprintf(canonical_path, 32,
"devid:%llu", args.devid);
}
for (j = 0; j < ARRAY_SIZE(dev_stats); j++) {
/* We got fewer items than we know */
if (args.nr_items < dev_stats[j].num + 1)
continue;
printf("[%s].%-16s %llu\n", canonical_path,
dev_stats[j].name,
(unsigned long long)
args.values[dev_stats[j].num]);
if ((check == 1)
&& (args.values[dev_stats[j].num] > 0))
err |= 64;
}
free(canonical_path);
}
}
out:
free(di_args);
close_file_or_dir(fdmnt, dirstream);
return err;
}
static int dev_replace_handle_sigint(int fd)
{
struct sigaction sa = {
.sa_handler = fd == -1 ? SIG_DFL : dev_replace_sigint_handler
};
dev_replace_cancel_fd = fd;
return sigaction(SIGINT, &sa, NULL);
}
static const char *const cmd_start_replace_usage[] = {
"btrfs replace start srcdev|devid targetdev [-Bfr] mount_point",
"Replace device of a btrfs filesystem.",
"On a live filesystem, duplicate the data to the target device which",
"is currently stored on the source device. If the source device is not",
"available anymore, or if the -r option is set, the data is built",
"only using the RAID redundancy mechanisms. After completion of the",
"operation, the source device is removed from the filesystem.",
"If the srcdev is a numerical value, it is assumed to be the device id",
"of the filesystem which is mounted at mount_point, otherwise it is",
"the path to the source device. If the source device is disconnected,",
"from the system, you have to use the devid parameter format.",
"The targetdev needs to be same size or larger than the srcdev.",
"",
"-r only read from srcdev if no other zero-defect mirror exists",
" (enable this if your drive has lots of read errors, the access",
" would be very slow)",
"-f force using and overwriting targetdev even if it looks like",
" containing a valid btrfs filesystem. A valid filesystem is",
" assumed if a btrfs superblock is found which contains a",
" correct checksum. Devices which are currently mounted are",
" never allowed to be used as the targetdev",
"-B do not background",
NULL
};
static int cmd_start_replace(int argc, char **argv)
{
struct btrfs_ioctl_dev_replace_args start_args = {0};
struct btrfs_ioctl_dev_replace_args status_args = {0};
int ret;
int i;
int c;
int fdmnt = -1;
int fdsrcdev = -1;
int fddstdev = -1;
char *path;
char *srcdev;
char *dstdev;
int avoid_reading_from_srcdev = 0;
int force_using_targetdev = 0;
u64 total_devs = 1;
struct btrfs_fs_devices *fs_devices_mnt = NULL;
struct stat st;
u64 dstdev_block_count;
int do_not_background = 0;
int mixed = 0;
while ((c = getopt(argc, argv, "Brf")) != -1) {
switch (c) {
case 'B':
do_not_background = 1;
break;
case 'r':
avoid_reading_from_srcdev = 1;
break;
case 'f':
force_using_targetdev = 1;
break;
case '?':
default:
usage(cmd_start_replace_usage);
}
}
start_args.start.cont_reading_from_srcdev_mode =
avoid_reading_from_srcdev ?
BTRFS_IOCTL_DEV_REPLACE_CONT_READING_FROM_SRCDEV_MODE_AVOID :
BTRFS_IOCTL_DEV_REPLACE_CONT_READING_FROM_SRCDEV_MODE_ALWAYS;
if (check_argc_exact(argc - optind, 3))
usage(cmd_start_replace_usage);
path = argv[optind + 2];
fdmnt = open_path_or_dev_mnt(path);
if (fdmnt < 0) {
fprintf(stderr, "ERROR: can't access \"%s\": %s\n",
path, strerror(errno));
goto leave_with_error;
}
/* check for possible errors before backgrounding */
status_args.cmd = BTRFS_IOCTL_DEV_REPLACE_CMD_STATUS;
ret = ioctl(fdmnt, BTRFS_IOC_DEV_REPLACE, &status_args);
if (ret) {
fprintf(stderr,
"ERROR: ioctl(DEV_REPLACE_STATUS) failed on \"%s\": %s, %s\n",
path, strerror(errno),
replace_dev_result2string(status_args.result));
goto leave_with_error;
}
if (status_args.result != BTRFS_IOCTL_DEV_REPLACE_RESULT_NO_ERROR) {
fprintf(stderr,
"ERROR: ioctl(DEV_REPLACE_STATUS) on \"%s\" returns error: %s\n",
path, replace_dev_result2string(status_args.result));
goto leave_with_error;
}
if (status_args.status.replace_state ==
BTRFS_IOCTL_DEV_REPLACE_STATE_STARTED) {
fprintf(stderr,
"ERROR: btrfs replace on \"%s\" already started!\n",
path);
goto leave_with_error;
}
srcdev = argv[optind];
dstdev = argv[optind + 1];
if (is_numerical(srcdev)) {
struct btrfs_ioctl_fs_info_args fi_args;
struct btrfs_ioctl_dev_info_args *di_args = NULL;
if (atoi(srcdev) == 0) {
fprintf(stderr, "Error: Failed to parse the numerical devid value '%s'\n",
srcdev);
goto leave_with_error;
}
start_args.start.srcdevid = (__u64)atoi(srcdev);
ret = get_fs_info(path, &fi_args, &di_args);
if (ret) {
fprintf(stderr, "ERROR: getting dev info for devstats failed: "
"%s\n", strerror(-ret));
free(di_args);
goto leave_with_error;
}
if (!fi_args.num_devices) {
fprintf(stderr, "ERROR: no devices found\n");
free(di_args);
goto leave_with_error;
}
for (i = 0; i < fi_args.num_devices; i++)
if (start_args.start.srcdevid == di_args[i].devid)
break;
free(di_args);
if (i == fi_args.num_devices) {
fprintf(stderr, "Error: '%s' is not a valid devid for filesystem '%s'\n",
srcdev, path);
goto leave_with_error;
}
} else {
fdsrcdev = open(srcdev, O_RDWR);
if (fdsrcdev < 0) {
fprintf(stderr, "Error: Unable to open device '%s'\n",
srcdev);
goto leave_with_error;
}
ret = fstat(fdsrcdev, &st);
if (ret) {
fprintf(stderr, "Error: Unable to stat '%s'\n", srcdev);
goto leave_with_error;
}
if (!S_ISBLK(st.st_mode)) {
fprintf(stderr, "Error: '%s' is not a block device\n",
srcdev);
goto leave_with_error;
}
strncpy((char *)start_args.start.srcdev_name, srcdev,
BTRFS_DEVICE_PATH_NAME_MAX);
close(fdsrcdev);
fdsrcdev = -1;
start_args.start.srcdevid = 0;
}
ret = check_mounted(dstdev);
if (ret < 0) {
fprintf(stderr, "Error checking %s mount status\n", dstdev);
goto leave_with_error;
}
if (ret == 1) {
fprintf(stderr,
"Error, target device %s is in use and currently mounted!\n",
dstdev);
goto leave_with_error;
}
fddstdev = open(dstdev, O_RDWR);
if (fddstdev < 0) {
fprintf(stderr, "Unable to open %s\n", dstdev);
goto leave_with_error;
}
ret = btrfs_scan_one_device(fddstdev, dstdev, &fs_devices_mnt,
&total_devs, BTRFS_SUPER_INFO_OFFSET);
if (ret >= 0 && !force_using_targetdev) {
fprintf(stderr,
"Error, target device %s contains filesystem, use '-f' to force overwriting.\n",
dstdev);
goto leave_with_error;
}
ret = fstat(fddstdev, &st);
if (ret) {
fprintf(stderr, "Error: Unable to stat '%s'\n", dstdev);
goto leave_with_error;
}
if (!S_ISBLK(st.st_mode)) {
fprintf(stderr, "Error: '%s' is not a block device\n", dstdev);
goto leave_with_error;
}
strncpy((char *)start_args.start.tgtdev_name, dstdev,
BTRFS_DEVICE_PATH_NAME_MAX);
if (btrfs_prepare_device(fddstdev, dstdev, 1, &dstdev_block_count, 0,
&mixed, 0)) {
fprintf(stderr, "Error: Failed to prepare device '%s'\n",
dstdev);
goto leave_with_error;
}
close(fddstdev);
fddstdev = -1;
dev_replace_handle_sigint(fdmnt);
if (!do_not_background) {
if (daemon(0, 0) < 0) {
fprintf(stderr, "ERROR, backgrounding failed: %s\n",
strerror(errno));
goto leave_with_error;
}
}
start_args.cmd = BTRFS_IOCTL_DEV_REPLACE_CMD_START;
ret = ioctl(fdmnt, BTRFS_IOC_DEV_REPLACE, &start_args);
if (do_not_background) {
if (ret) {
fprintf(stderr,
"ERROR: ioctl(DEV_REPLACE_START) failed on \"%s\": %s, %s\n",
path, strerror(errno),
replace_dev_result2string(start_args.result));
goto leave_with_error;
}
if (start_args.result !=
BTRFS_IOCTL_DEV_REPLACE_RESULT_NO_ERROR) {
fprintf(stderr,
"ERROR: ioctl(DEV_REPLACE_START) on \"%s\" returns error: %s\n",
path,
replace_dev_result2string(start_args.result));
goto leave_with_error;
}
}
close(fdmnt);
return 0;
leave_with_error:
if (fdmnt != -1)
close(fdmnt);
if (fdsrcdev != -1)
close(fdsrcdev);
if (fddstdev != -1)
close(fddstdev);
return -1;
}
static const char *const cmd_status_replace_usage[] = {
"btrfs replace status mount_point [-1]",
"Print status and progress information of a running device replace",
"operation",
"",
"-1 print once instead of print continously until the replace",
" operation finishes (or is canceled)",
NULL
};
static int cmd_status_replace(int argc, char **argv)
{
int fd;
int e;
int c;
char *path;
int once = 0;
int ret;
while ((c = getopt(argc, argv, "1")) != -1) {
switch (c) {
case '1':
once = 1;
break;
case '?':
default:
usage(cmd_status_replace_usage);
}
}
if (check_argc_exact(argc - optind, 1))
usage(cmd_status_replace_usage);
path = argv[optind];
fd = open_file_or_dir(path);
e = errno;
if (fd < 0) {
fprintf(stderr, "ERROR: can't access \"%s\": %s\n",
path, strerror(e));
return -1;
}
ret = print_replace_status(fd, path, once);
close(fd);
return ret;
}
static int dev_replace_handle_sigint(int fd)
{
struct sigaction sa = {
.sa_handler = fd == -1 ? SIG_DFL : dev_replace_sigint_handler
};
dev_replace_cancel_fd = fd;
return sigaction(SIGINT, &sa, NULL);
}
static const char *const cmd_replace_start_usage[] = {
"btrfs replace start [-Bfr] <srcdev>|<devid> <targetdev> <mount_point>",
"Replace device of a btrfs filesystem.",
"On a live filesystem, duplicate the data to the target device which",
"is currently stored on the source device. If the source device is not",
"available anymore, or if the -r option is set, the data is built",
"only using the RAID redundancy mechanisms. After completion of the",
"operation, the source device is removed from the filesystem.",
"If the <srcdev> is a numerical value, it is assumed to be the device id",
"of the filesystem which is mounted at <mount_point>, otherwise it is",
"the path to the source device. If the source device is disconnected,",
"from the system, you have to use the <devid> parameter format.",
"The <targetdev> needs to be same size or larger than the <srcdev>.",
"",
"-r only read from <srcdev> if no other zero-defect mirror exists",
" (enable this if your drive has lots of read errors, the access",
" would be very slow)",
"-f force using and overwriting <targetdev> even if it looks like",
" containing a valid btrfs filesystem. A valid filesystem is",
" assumed if a btrfs superblock is found which contains a",
" correct checksum. Devices which are currently mounted are",
" never allowed to be used as the <targetdev>",
"-B do not background",
NULL
};
static int cmd_replace_start(int argc, char **argv)
{
struct btrfs_ioctl_dev_replace_args start_args = {0};
struct btrfs_ioctl_dev_replace_args status_args = {0};
int ret;
int i;
int c;
int fdmnt = -1;
int fddstdev = -1;
char *path;
char *srcdev;
char *dstdev = NULL;
int avoid_reading_from_srcdev = 0;
int force_using_targetdev = 0;
u64 dstdev_block_count;
int do_not_background = 0;
DIR *dirstream = NULL;
u64 srcdev_size;
u64 dstdev_size;
while ((c = getopt(argc, argv, "Brf")) != -1) {
switch (c) {
case 'B':
do_not_background = 1;
break;
case 'r':
avoid_reading_from_srcdev = 1;
break;
case 'f':
force_using_targetdev = 1;
break;
case '?':
default:
usage(cmd_replace_start_usage);
}
}
start_args.start.cont_reading_from_srcdev_mode =
avoid_reading_from_srcdev ?
BTRFS_IOCTL_DEV_REPLACE_CONT_READING_FROM_SRCDEV_MODE_AVOID :
BTRFS_IOCTL_DEV_REPLACE_CONT_READING_FROM_SRCDEV_MODE_ALWAYS;
if (check_argc_exact(argc - optind, 3))
usage(cmd_replace_start_usage);
path = argv[optind + 2];
fdmnt = open_path_or_dev_mnt(path, &dirstream, 1);
if (fdmnt < 0)
goto leave_with_error;
/* check for possible errors before backgrounding */
status_args.cmd = BTRFS_IOCTL_DEV_REPLACE_CMD_STATUS;
status_args.result = BTRFS_IOCTL_DEV_REPLACE_RESULT_NO_RESULT;
ret = ioctl(fdmnt, BTRFS_IOC_DEV_REPLACE, &status_args);
if (ret < 0) {
fprintf(stderr,
"ERROR: ioctl(DEV_REPLACE_STATUS) failed on \"%s\": %s",
path, strerror(errno));
if (status_args.result != BTRFS_IOCTL_DEV_REPLACE_RESULT_NO_RESULT)
fprintf(stderr, ", %s\n",
replace_dev_result2string(status_args.result));
else
fprintf(stderr, "\n");
goto leave_with_error;
}
if (status_args.result != BTRFS_IOCTL_DEV_REPLACE_RESULT_NO_ERROR) {
error("ioctl(DEV_REPLACE_STATUS) on '%s' returns error: %s",
path, replace_dev_result2string(status_args.result));
goto leave_with_error;
}
if (status_args.status.replace_state ==
BTRFS_IOCTL_DEV_REPLACE_STATE_STARTED) {
error("device replace on '%s' already started", path);
goto leave_with_error;
}
srcdev = argv[optind];
dstdev = canonicalize_path(argv[optind + 1]);
if (!dstdev) {
error("cannot canonicalize path '%s': %s",
argv[optind + 1], strerror(errno));
goto leave_with_error;
}
if (string_is_numerical(srcdev)) {
struct btrfs_ioctl_fs_info_args fi_args;
struct btrfs_ioctl_dev_info_args *di_args = NULL;
start_args.start.srcdevid = arg_strtou64(srcdev);
ret = get_fs_info(path, &fi_args, &di_args);
if (ret) {
error("failed to get device info: %s", strerror(-ret));
free(di_args);
goto leave_with_error;
}
if (!fi_args.num_devices) {
error("no devices found");
free(di_args);
goto leave_with_error;
}
for (i = 0; i < fi_args.num_devices; i++)
if (start_args.start.srcdevid == di_args[i].devid)
break;
srcdev_size = di_args[i].total_bytes;
free(di_args);
if (i == fi_args.num_devices) {
error("'%s' is not a valid devid for filesystem '%s'",
srcdev, path);
goto leave_with_error;
}
} else if (is_block_device(srcdev) > 0) {
strncpy((char *)start_args.start.srcdev_name, srcdev,
BTRFS_DEVICE_PATH_NAME_MAX);
start_args.start.srcdevid = 0;
srcdev_size = get_partition_size(srcdev);
} else {
error("source device must be a block device or a devid");
goto leave_with_error;
}
ret = test_dev_for_mkfs(dstdev, force_using_targetdev);
if (ret)
goto leave_with_error;
dstdev_size = get_partition_size(dstdev);
if (srcdev_size > dstdev_size) {
error("target device smaller than source device (required %llu bytes)",
srcdev_size);
goto leave_with_error;
}
fddstdev = open(dstdev, O_RDWR);
if (fddstdev < 0) {
error("unable to open %s: %s", dstdev, strerror(errno));
goto leave_with_error;
}
strncpy((char *)start_args.start.tgtdev_name, dstdev,
BTRFS_DEVICE_PATH_NAME_MAX);
ret = btrfs_prepare_device(fddstdev, dstdev, &dstdev_block_count, 0,
PREP_DEVICE_ZERO_END | PREP_DEVICE_VERBOSE);
if (ret)
goto leave_with_error;
close(fddstdev);
fddstdev = -1;
free(dstdev);
dstdev = NULL;
dev_replace_handle_sigint(fdmnt);
if (!do_not_background) {
if (daemon(0, 0) < 0) {
error("backgrounding failed: %s", strerror(errno));
goto leave_with_error;
}
}
start_args.cmd = BTRFS_IOCTL_DEV_REPLACE_CMD_START;
start_args.result = BTRFS_IOCTL_DEV_REPLACE_RESULT_NO_RESULT;
ret = ioctl(fdmnt, BTRFS_IOC_DEV_REPLACE, &start_args);
if (do_not_background) {
if (ret < 0) {
fprintf(stderr,
"ERROR: ioctl(DEV_REPLACE_START) failed on \"%s\": %s",
path, strerror(errno));
if (start_args.result != BTRFS_IOCTL_DEV_REPLACE_RESULT_NO_RESULT)
fprintf(stderr, ", %s\n",
replace_dev_result2string(start_args.result));
else
fprintf(stderr, "\n");
if (errno == EOPNOTSUPP)
warning("device replace of RAID5/6 not supported with this kernel");
goto leave_with_error;
}
if (start_args.result !=
BTRFS_IOCTL_DEV_REPLACE_RESULT_NO_ERROR) {
error("ioctl(DEV_REPLACE_START) on '%s' returns error: %s",
path,
replace_dev_result2string(start_args.result));
goto leave_with_error;
}
}
close_file_or_dir(fdmnt, dirstream);
return 0;
leave_with_error:
if (dstdev)
free(dstdev);
if (fdmnt != -1)
close(fdmnt);
if (fddstdev != -1)
close(fddstdev);
return 1;
}
static const char *const cmd_replace_status_usage[] = {
"btrfs replace status [-1] <mount_point>",
"Print status and progress information of a running device replace",
"operation",
"",
"-1 print once instead of print continuously until the replace",
" operation finishes (or is canceled)",
NULL
};
static int cmd_replace_status(int argc, char **argv)
{
int fd;
int c;
char *path;
int once = 0;
int ret;
DIR *dirstream = NULL;
while ((c = getopt(argc, argv, "1")) != -1) {
switch (c) {
case '1':
once = 1;
break;
case '?':
default:
usage(cmd_replace_status_usage);
}
}
if (check_argc_exact(argc - optind, 1))
usage(cmd_replace_status_usage);
path = argv[optind];
fd = btrfs_open_dir(path, &dirstream, 1);
if (fd < 0)
return 1;
ret = print_replace_status(fd, path, once);
close_file_or_dir(fd, dirstream);
return !!ret;
}
static int scrub_start(int argc, char **argv, int resume)
{
int fdmnt;
int prg_fd = -1;
int fdres = -1;
int ret;
pid_t pid;
int c;
int i;
int err = 0;
int e_uncorrectable = 0;
int e_correctable = 0;
int print_raw = 0;
char *path;
int do_background = 1;
int do_wait = 0;
int do_print = 0;
int do_quiet = 0;
int do_record = 1;
int readonly = 0;
int do_stats_per_dev = 0;
int ioprio_class = IOPRIO_CLASS_IDLE;
int ioprio_classdata = 0;
int n_start = 0;
int n_skip = 0;
int n_resume = 0;
struct btrfs_ioctl_fs_info_args fi_args;
struct btrfs_ioctl_dev_info_args *di_args = NULL;
struct scrub_progress *sp = NULL;
struct scrub_fs_stat fs_stat;
struct timeval tv;
struct sockaddr_un addr = {
.sun_family = AF_UNIX,
};
pthread_t *t_devs = NULL;
pthread_t t_prog;
struct scrub_file_record **past_scrubs = NULL;
struct scrub_file_record *last_scrub = NULL;
char *datafile = strdup(SCRUB_DATA_FILE);
char fsid[BTRFS_UUID_UNPARSED_SIZE];
char sock_path[PATH_MAX] = "";
struct scrub_progress_cycle spc;
pthread_mutex_t spc_write_mutex = PTHREAD_MUTEX_INITIALIZER;
void *terr;
u64 devid;
DIR *dirstream = NULL;
int force = 0;
int nothing_to_resume = 0;
while ((c = getopt(argc, argv, "BdqrRc:n:f")) != -1) {
switch (c) {
case 'B':
do_background = 0;
do_wait = 1;
do_print = 1;
break;
case 'd':
do_stats_per_dev = 1;
break;
case 'q':
do_quiet = 1;
break;
case 'r':
readonly = 1;
break;
case 'R':
print_raw = 1;
break;
case 'c':
ioprio_class = (int)strtol(optarg, NULL, 10);
break;
case 'n':
ioprio_classdata = (int)strtol(optarg, NULL, 10);
break;
case 'f':
force = 1;
break;
case '?':
default:
usage(resume ? cmd_scrub_resume_usage :
cmd_scrub_start_usage);
}
}
/* try to catch most error cases before forking */
if (check_argc_exact(argc - optind, 1)) {
usage(resume ? cmd_scrub_resume_usage :
cmd_scrub_start_usage);
}
spc.progress = NULL;
if (do_quiet && do_print)
do_print = 0;
if (mkdir_p(datafile)) {
warning_on(!do_quiet,
"cannot create scrub data file, mkdir %s failed: %s. Status recording disabled",
datafile, strerror(errno));
do_record = 0;
}
free(datafile);
path = argv[optind];
fdmnt = open_path_or_dev_mnt(path, &dirstream, !do_quiet);
if (fdmnt < 0)
return 1;
ret = get_fs_info(path, &fi_args, &di_args);
if (ret) {
error_on(!do_quiet,
"getting dev info for scrub failed: %s",
strerror(-ret));
err = 1;
goto out;
}
if (!fi_args.num_devices) {
error_on(!do_quiet, "no devices found");
err = 1;
goto out;
}
uuid_unparse(fi_args.fsid, fsid);
fdres = scrub_open_file_r(SCRUB_DATA_FILE, fsid);
if (fdres < 0 && fdres != -ENOENT) {
warning_on(!do_quiet, "failed to open status file: %s",
strerror(-fdres));
} else if (fdres >= 0) {
past_scrubs = scrub_read_file(fdres, !do_quiet);
if (IS_ERR(past_scrubs))
warning_on(!do_quiet, "failed to read status file: %s",
strerror(-PTR_ERR(past_scrubs)));
close(fdres);
}
/*
* Check for stale information in the status file, ie. if it's
* canceled=0, finished=0 but no scrub is running.
*/
if (!is_scrub_running_in_kernel(fdmnt, di_args, fi_args.num_devices))
force = 1;
/*
* check whether any involved device is already busy running a
* scrub. This would cause damaged status messages and the state
* "aborted" without the explanation that a scrub was already
* running. Therefore check it first, prevent it and give some
* feedback to the user if scrub is already running.
* Note that if scrub is started with a block device as the
* parameter, only that particular block device is checked. It
* is a normal mode of operation to start scrub on multiple
* single devices, there is no reason to prevent this.
*/
if (!force && is_scrub_running_on_fs(&fi_args, di_args, past_scrubs)) {
error_on(!do_quiet,
"Scrub is already running.\n"
"To cancel use 'btrfs scrub cancel %s'.\n"
"To see the status use 'btrfs scrub status [-d] %s'",
path, path);
err = 1;
goto out;
}
t_devs = malloc(fi_args.num_devices * sizeof(*t_devs));
sp = calloc(fi_args.num_devices, sizeof(*sp));
spc.progress = calloc(fi_args.num_devices * 2, sizeof(*spc.progress));
if (!t_devs || !sp || !spc.progress) {
error_on(!do_quiet, "scrub failed: %s", strerror(errno));
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) {
error_on(!do_quiet, "pthread_mutex_init failed: %s",
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;
sp[i].ioprio_class = ioprio_class;
sp[i].ioprio_classdata = ioprio_classdata;
}
if (!n_start && !n_resume) {
if (!do_quiet)
printf("scrub: nothing to resume for %s, fsid %s\n",
path, fsid);
nothing_to_resume = 1;
goto out;
}
ret = prg_fd = socket(AF_UNIX, SOCK_STREAM, 0);
while (ret != -1) {
ret = scrub_datafile(SCRUB_PROGRESS_SOCKET_PATH, fsid, NULL,
sock_path, sizeof(sock_path));
/* ignore EOVERFLOW, try using a shorter path for the socket */
addr.sun_path[sizeof(addr.sun_path) - 1] = '\0';
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;
/*
* bind failed with EADDRINUSE. so let's see if anyone answers
* when we make a call to the socket ...
*/
ret = connect(prg_fd, (struct sockaddr *)&addr, sizeof(addr));
if (!ret || errno != ECONNREFUSED) {
/* ... yes, so scrub must be running. error out */
error("scrub already running");
close(prg_fd);
prg_fd = -1;
goto out;
}
/*
* ... no, this means someone left us alone with an unused
* socket in the file system. remove it and try again.
*/
ret = unlink(sock_path);
}
if (ret != -1)
ret = listen(prg_fd, 100);
if (ret == -1) {
warning_on(!do_quiet,
"failed to open the progress status socket at %s: %s. Progress cannot be queried",
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) {
warning_on(!do_quiet,
"failed to write the progress status file: %s. Status recording disabled",
strerror(-ret));
do_record = 0;
}
}
if (do_background) {
pid = fork();
if (pid == -1) {
error_on(!do_quiet, "cannot scrub, fork failed: %s",
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) {
error_on(!do_quiet, "wait failed (ret=%d): %s",
ret, strerror(errno));
err = 1;
goto out;
}
if (!WIFEXITED(stat) || WEXITSTATUS(stat)) {
error_on(!do_quiet, "scrub process failed");
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], NULL,
scrub_one_dev, &sp[i]);
if (ret) {
if (do_print)
error("creating scrub_one_dev[%llu] thread failed: %s",
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;
ret = pthread_create(&t_prog, NULL, scrub_progress_cycle, &spc);
if (ret) {
if (do_print)
error("creating progress thread failed: %s",
strerror(ret));
err = 1;
goto out;
}
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)
error("pthread_join failed for scrub_one_dev[%llu]: %s",
devid, strerror(ret));
++err;
continue;
}
if (sp[i].ret) {
switch (sp[i].ioctl_errno) {
case ENODEV:
if (do_print)
warning("device %lld not present",
devid);
continue;
case ECANCELED:
++err;
break;
default:
if (do_print)
error("scrubbing %s failed for device id %lld: ret=%d, errno=%d (%s)",
path, devid,
sp[i].ret, sp[i].ioctl_errno,
strerror(sp[i].ioctl_errno));
++err;
continue;
}
}
if (sp[i].scrub_args.progress.uncorrectable_errors > 0)
e_uncorrectable++;
if (sp[i].scrub_args.progress.corrected_errors > 0
|| sp[i].scrub_args.progress.unverified_errors > 0)
e_correctable++;
}
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);
}
}
ret = pthread_cancel(t_prog);
if (!ret)
ret = pthread_join(t_prog, &terr);
/* check for errors from the handling of the progress thread */
if (do_print && ret) {
error("progress thread handling failed: %s",
strerror(ret));
}
/* check for errors returned from the progress thread itself */
if (do_print && terr && terr != PTHREAD_CANCELED)
error("recording progress failed: %s",
strerror(-PTR_ERR(terr)));
if (do_record) {
ret = scrub_write_progress(&spc_write_mutex, fsid, sp,
fi_args.num_devices);
if (ret && do_print)
error("failed to record the result: %s",
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_file_or_dir(fdmnt, dirstream);
if (err)
return 1;
if (nothing_to_resume)
return 2;
if (e_uncorrectable) {
error_on(!do_quiet, "there are uncorrectable errors");
return 3;
}
if (e_correctable)
warning_on(!do_quiet,
"errors detected during scrubbing, corrected");
return 0;
}
static int cmd_dev_stats(int argc, char **argv)
{
char *dev_path;
struct btrfs_ioctl_fs_info_args fi_args;
struct btrfs_ioctl_dev_info_args *di_args = NULL;
int ret;
int fdmnt;
int i;
int c;
int err = 0;
__u64 flags = 0;
DIR *dirstream = NULL;
optind = 1;
while ((c = getopt(argc, argv, "z")) != -1) {
switch (c) {
case 'z':
flags = BTRFS_DEV_STATS_RESET;
break;
case '?':
default:
usage(cmd_dev_stats_usage);
}
}
argc = argc - optind;
if (check_argc_exact(argc, 1))
usage(cmd_dev_stats_usage);
dev_path = argv[optind];
fdmnt = open_path_or_dev_mnt(dev_path, &dirstream);
if (fdmnt < 0) {
if (errno == EINVAL)
fprintf(stderr,
"ERROR: '%s' is not a mounted btrfs device\n",
dev_path);
else
fprintf(stderr, "ERROR: can't access '%s': %s\n",
dev_path, strerror(errno));
return 1;
}
ret = get_fs_info(dev_path, &fi_args, &di_args);
if (ret) {
fprintf(stderr, "ERROR: getting dev info for devstats 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;
}
for (i = 0; i < fi_args.num_devices; i++) {
struct btrfs_ioctl_get_dev_stats args = {0};
__u8 path[BTRFS_DEVICE_PATH_NAME_MAX + 1];
strncpy((char *)path, (char *)di_args[i].path,
BTRFS_DEVICE_PATH_NAME_MAX);
path[BTRFS_DEVICE_PATH_NAME_MAX] = '\0';
args.devid = di_args[i].devid;
args.nr_items = BTRFS_DEV_STAT_VALUES_MAX;
args.flags = flags;
if (ioctl(fdmnt, BTRFS_IOC_GET_DEV_STATS, &args) < 0) {
fprintf(stderr,
"ERROR: ioctl(BTRFS_IOC_GET_DEV_STATS) on %s failed: %s\n",
path, strerror(errno));
err = 1;
} else {
char *canonical_path;
canonical_path = canonicalize_path((char *)path);
if (args.nr_items >= BTRFS_DEV_STAT_WRITE_ERRS + 1)
printf("[%s].write_io_errs %llu\n",
canonical_path,
(unsigned long long) args.values[
BTRFS_DEV_STAT_WRITE_ERRS]);
if (args.nr_items >= BTRFS_DEV_STAT_READ_ERRS + 1)
printf("[%s].read_io_errs %llu\n",
canonical_path,
(unsigned long long) args.values[
BTRFS_DEV_STAT_READ_ERRS]);
if (args.nr_items >= BTRFS_DEV_STAT_FLUSH_ERRS + 1)
printf("[%s].flush_io_errs %llu\n",
canonical_path,
(unsigned long long) args.values[
BTRFS_DEV_STAT_FLUSH_ERRS]);
if (args.nr_items >= BTRFS_DEV_STAT_CORRUPTION_ERRS + 1)
printf("[%s].corruption_errs %llu\n",
canonical_path,
(unsigned long long) args.values[
BTRFS_DEV_STAT_CORRUPTION_ERRS]);
if (args.nr_items >= BTRFS_DEV_STAT_GENERATION_ERRS + 1)
printf("[%s].generation_errs %llu\n",
canonical_path,
(unsigned long long) args.values[
BTRFS_DEV_STAT_GENERATION_ERRS]);
free(canonical_path);
}
}
out:
free(di_args);
close_file_or_dir(fdmnt, dirstream);
return err;
}
