mirror of
https://github.com/koverstreet/bcachefs-tools.git
synced 2025-01-22 00:04:31 +03:00
43a1380575
Implement 'bcachefs data scrub', frontend for BCH_IOCTL_DATA.BCH_DATA_OP_scrub. Takes a path to a device, mountpoint, or filesystem uuid. Can be run on a specific device by passing a device, or if run on a filesystem scrubs all devices in parallel. Metadata only scrubbing is supported via -m. Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
313 lines
8.2 KiB
C
313 lines
8.2 KiB
C
#include <fcntl.h>
|
|
#include <getopt.h>
|
|
#include <string.h>
|
|
#include <sys/stat.h>
|
|
#include <sys/types.h>
|
|
|
|
#include "cmds.h"
|
|
#include "libbcachefs.h"
|
|
#include "tools-util.h"
|
|
|
|
#include "libbcachefs/bcachefs.h"
|
|
#include "libbcachefs/btree_iter.h"
|
|
#include "libbcachefs/errcode.h"
|
|
#include "libbcachefs/error.h"
|
|
#include "libbcachefs/journal_io.h"
|
|
#include "libbcachefs/journal_seq_blacklist.h"
|
|
#include "libbcachefs/super.h"
|
|
|
|
static const char *NORMAL = "\x1B[0m";
|
|
static const char *RED = "\x1B[31m";
|
|
|
|
static void list_journal_usage(void)
|
|
{
|
|
puts("bcachefs list_journal - print contents of journal\n"
|
|
"Usage: bcachefs list_journal [OPTION]... <devices>\n"
|
|
"\n"
|
|
"Options:\n"
|
|
" -a Read entire journal, not just dirty entries\n"
|
|
" -n, --nr-entries=nr Number of journal entries to print, starting from the most recent\n"
|
|
" -t, --transaction-filter=bbpos Filter transactions not updating <bbpos>\n"
|
|
" Or entries not matching the range <bbpos-bbpos>\n"
|
|
" -k, --key-filter=btree Filter keys not updating btree\n"
|
|
" -v, --verbose Verbose mode\n"
|
|
" -h, --help Display this help and exit\n"
|
|
"Report bugs to <linux-bcachefs@vger.kernel.org>");
|
|
}
|
|
|
|
static void star_start_of_lines(char *buf)
|
|
{
|
|
char *p = buf;
|
|
|
|
if (*p == ' ')
|
|
*p = '*';
|
|
|
|
while ((p = strstr(p, "\n ")))
|
|
p[1] = '*';
|
|
}
|
|
|
|
static inline bool entry_is_transaction_start(struct jset_entry *entry)
|
|
{
|
|
return entry->type == BCH_JSET_ENTRY_log && !entry->level;
|
|
}
|
|
|
|
static inline bool entry_is_log_msg(struct jset_entry *entry)
|
|
{
|
|
return entry->type == BCH_JSET_ENTRY_log && entry->level;
|
|
}
|
|
|
|
typedef DARRAY(struct bbpos_range) d_bbpos_range;
|
|
typedef DARRAY(enum btree_id) d_btree_id;
|
|
|
|
static bool bkey_matches_filter(d_bbpos_range filter, struct jset_entry *entry, struct bkey_i *k)
|
|
{
|
|
darray_for_each(filter, i) {
|
|
struct bbpos k_start = BBPOS(entry->btree_id, bkey_start_pos(&k->k));
|
|
struct bbpos k_end = BBPOS(entry->btree_id, k->k.p);
|
|
|
|
if (bbpos_cmp(k_start, i->start) >= 0 &&
|
|
bbpos_cmp(k_end, i->end) <= 0)
|
|
return true;
|
|
}
|
|
return false;
|
|
}
|
|
|
|
static bool entry_matches_transaction_filter(struct jset_entry *entry,
|
|
d_bbpos_range filter)
|
|
{
|
|
if (entry->type == BCH_JSET_ENTRY_btree_root ||
|
|
entry->type == BCH_JSET_ENTRY_btree_keys ||
|
|
entry->type == BCH_JSET_ENTRY_overwrite)
|
|
jset_entry_for_each_key(entry, k)
|
|
if (bkey_matches_filter(filter, entry, k))
|
|
return true;
|
|
return false;
|
|
}
|
|
|
|
static bool should_print_transaction(struct jset_entry *entry, struct jset_entry *end,
|
|
darray_str msg_filter,
|
|
d_bbpos_range key_filter)
|
|
{
|
|
struct jset_entry_log *l = container_of(entry, struct jset_entry_log, entry);
|
|
unsigned b = jset_entry_log_msg_bytes(l);
|
|
|
|
darray_for_each(msg_filter, i)
|
|
if (!strncmp(*i, l->d, b))
|
|
return false;
|
|
|
|
if (!key_filter.nr)
|
|
return true;
|
|
|
|
for (entry = vstruct_next(entry);
|
|
entry != end && !entry_is_transaction_start(entry);
|
|
entry = vstruct_next(entry))
|
|
if (entry_is_log_msg(entry) ||
|
|
entry_matches_transaction_filter(entry, key_filter))
|
|
return true;
|
|
|
|
return false;
|
|
}
|
|
|
|
static bool should_print_entry(struct jset_entry *entry, d_btree_id filter)
|
|
{
|
|
if (!filter.nr)
|
|
return true;
|
|
|
|
if (entry->type != BCH_JSET_ENTRY_btree_root &&
|
|
entry->type != BCH_JSET_ENTRY_btree_keys &&
|
|
entry->type != BCH_JSET_ENTRY_overwrite)
|
|
return true;
|
|
|
|
jset_entry_for_each_key(entry, k)
|
|
darray_for_each(filter, id)
|
|
if (entry->btree_id == *id)
|
|
return true;
|
|
|
|
return false;
|
|
}
|
|
|
|
static void journal_entry_header_to_text(struct printbuf *out,
|
|
struct bch_fs *c,
|
|
struct journal_replay *p, bool blacklisted)
|
|
{
|
|
if (blacklisted)
|
|
prt_str(out, "blacklisted ");
|
|
|
|
prt_printf(out,
|
|
"journal entry %llu\n"
|
|
" version %u\n"
|
|
" last seq %llu\n"
|
|
" flush %u\n"
|
|
" written at ",
|
|
le64_to_cpu(p->j.seq),
|
|
le32_to_cpu(p->j.version),
|
|
le64_to_cpu(p->j.last_seq),
|
|
!JSET_NO_FLUSH(&p->j));
|
|
bch2_journal_ptrs_to_text(out, c, p);
|
|
|
|
if (blacklisted)
|
|
star_start_of_lines(out->buf);
|
|
}
|
|
|
|
static void journal_entry_header_print(struct bch_fs *c, struct journal_replay *p, bool blacklisted)
|
|
{
|
|
struct printbuf buf = PRINTBUF;
|
|
journal_entry_header_to_text(&buf, c, p, blacklisted);
|
|
printf("%s\n", buf.buf);
|
|
printbuf_exit(&buf);
|
|
}
|
|
|
|
static void journal_entries_print(struct bch_fs *c, unsigned nr_entries,
|
|
darray_str transaction_msg_filter,
|
|
d_bbpos_range transaction_key_filter,
|
|
d_btree_id key_filter)
|
|
{
|
|
struct journal_replay *p, **_p;
|
|
struct genradix_iter iter;
|
|
struct printbuf buf = PRINTBUF;
|
|
|
|
genradix_for_each(&c->journal_entries, iter, _p) {
|
|
bool printed_header = false;
|
|
|
|
p = *_p;
|
|
if (!p)
|
|
continue;
|
|
|
|
if (le64_to_cpu(p->j.seq) + nr_entries < atomic64_read(&c->journal.seq))
|
|
continue;
|
|
|
|
bool blacklisted = p->ignore_blacklisted ||
|
|
bch2_journal_seq_is_blacklisted(c,
|
|
le64_to_cpu(p->j.seq), false);
|
|
|
|
if (!transaction_msg_filter.nr &&
|
|
!transaction_key_filter.nr) {
|
|
journal_entry_header_print(c, p, blacklisted);
|
|
printed_header = true;
|
|
}
|
|
|
|
struct jset_entry *entry = p->j.start;
|
|
struct jset_entry *end = vstruct_last(&p->j);
|
|
while (entry != end) {
|
|
|
|
/*
|
|
* log entries denote the start of a new transaction
|
|
* commit:
|
|
*/
|
|
if (entry_is_transaction_start(entry)) {
|
|
if (!should_print_transaction(entry, end,
|
|
transaction_msg_filter,
|
|
transaction_key_filter)) {
|
|
do {
|
|
entry = vstruct_next(entry);
|
|
} while (entry != end && !entry_is_transaction_start(entry));
|
|
|
|
continue;
|
|
}
|
|
|
|
prt_newline(&buf);
|
|
}
|
|
|
|
if (!should_print_entry(entry, key_filter))
|
|
goto next;
|
|
|
|
if (!printed_header)
|
|
journal_entry_header_print(c, p, blacklisted);
|
|
printed_header = true;
|
|
|
|
bool highlight = entry_matches_transaction_filter(entry, transaction_key_filter);
|
|
if (highlight)
|
|
fputs(RED, stdout);
|
|
|
|
printbuf_indent_add(&buf, 4);
|
|
bch2_journal_entry_to_text(&buf, c, entry);
|
|
|
|
if (blacklisted)
|
|
star_start_of_lines(buf.buf);
|
|
printf("%s\n", buf.buf);
|
|
printbuf_reset(&buf);
|
|
|
|
if (highlight)
|
|
fputs(NORMAL, stdout);
|
|
next:
|
|
entry = vstruct_next(entry);
|
|
}
|
|
}
|
|
|
|
printbuf_exit(&buf);
|
|
}
|
|
|
|
int cmd_list_journal(int argc, char *argv[])
|
|
{
|
|
static const struct option longopts[] = {
|
|
{ "nr-entries", required_argument, NULL, 'n' },
|
|
{ "transaction-filter", required_argument, NULL, 't' },
|
|
{ "key-filter", required_argument, NULL, 'k' },
|
|
{ "verbose", no_argument, NULL, 'v' },
|
|
{ "help", no_argument, NULL, 'h' },
|
|
{ NULL }
|
|
};
|
|
struct bch_opts opts = bch2_opts_empty();
|
|
u32 nr_entries = U32_MAX;
|
|
darray_str transaction_msg_filter = {};
|
|
d_bbpos_range transaction_key_filter = {};
|
|
d_btree_id key_filter = {};
|
|
int opt;
|
|
|
|
opt_set(opts, noexcl, true);
|
|
opt_set(opts, nochanges, true);
|
|
opt_set(opts, norecovery, true);
|
|
opt_set(opts, read_only, true);
|
|
opt_set(opts, degraded, true);
|
|
opt_set(opts, very_degraded, true);
|
|
opt_set(opts, errors, BCH_ON_ERROR_continue);
|
|
opt_set(opts, fix_errors, FSCK_FIX_yes);
|
|
opt_set(opts, retain_recovery_info ,true);
|
|
opt_set(opts, read_journal_only,true);
|
|
|
|
while ((opt = getopt_long(argc, argv, "an:m:t:k:vh",
|
|
longopts, NULL)) != -1)
|
|
switch (opt) {
|
|
case 'a':
|
|
opt_set(opts, read_entire_journal, true);
|
|
break;
|
|
case 'n':
|
|
if (kstrtouint(optarg, 10, &nr_entries))
|
|
die("error parsing nr_entries");
|
|
opt_set(opts, read_entire_journal, true);
|
|
break;
|
|
case 'm':
|
|
darray_push(&transaction_msg_filter, strdup(optarg));
|
|
break;
|
|
case 't':
|
|
darray_push(&transaction_key_filter, bbpos_range_parse(optarg));
|
|
break;
|
|
case 'k':
|
|
darray_push(&key_filter, read_string_list_or_die(optarg, __bch2_btree_ids, "btree id"));
|
|
break;
|
|
case 'v':
|
|
opt_set(opts, verbose, true);
|
|
break;
|
|
case 'h':
|
|
list_journal_usage();
|
|
exit(EXIT_SUCCESS);
|
|
}
|
|
args_shift(optind);
|
|
|
|
if (!argc)
|
|
die("Please supply device(s) to open");
|
|
|
|
darray_str devs = get_or_split_cmdline_devs(argc, argv);
|
|
|
|
struct bch_fs *c = bch2_fs_open(devs.data, devs.nr, opts);
|
|
if (IS_ERR(c))
|
|
die("error opening %s: %s", argv[0], bch2_err_str(PTR_ERR(c)));
|
|
|
|
journal_entries_print(c, nr_entries,
|
|
transaction_msg_filter,
|
|
transaction_key_filter,
|
|
key_filter);
|
|
bch2_fs_stop(c);
|
|
return 0;
|
|
}
|