diff options
| -rw-r--r-- | fs/bcachefs/alloc_background.c | 2 | ||||
| -rw-r--r-- | fs/bcachefs/btree_io.c | 50 | ||||
| -rw-r--r-- | fs/bcachefs/btree_node_scan.c | 29 | ||||
| -rw-r--r-- | fs/bcachefs/extents.c | 4 | ||||
| -rw-r--r-- | fs/bcachefs/io_write.c | 7 | ||||
| -rw-r--r-- | fs/bcachefs/journal.c | 6 | ||||
| -rw-r--r-- | fs/bcachefs/recovery.c | 7 | ||||
| -rw-r--r-- | fs/bcachefs/super.c | 2 | ||||
| -rw-r--r-- | include/linux/workqueue.h | 12 | ||||
| -rw-r--r-- | kernel/workqueue.c | 14 |
10 files changed, 76 insertions, 57 deletions
diff --git a/fs/bcachefs/alloc_background.c b/fs/bcachefs/alloc_background.c index f1d35b7f3fc5..d5cb07c7f4d6 100644 --- a/fs/bcachefs/alloc_background.c +++ b/fs/bcachefs/alloc_background.c @@ -389,7 +389,7 @@ void __bch2_alloc_to_v4(struct bkey_s_c k, struct bch_alloc_v4 *out) if (k.k->type == KEY_TYPE_alloc_v4) { void *src, *dst; - *out = *bkey_s_c_to_alloc_v4(k).v; + bkey_val_copy(out, bkey_s_c_to_alloc_v4(k)); src = alloc_v4_backpointers(out); SET_BCH_ALLOC_V4_BACKPOINTERS_START(out, BCH_ALLOC_V4_U64s); diff --git a/fs/bcachefs/btree_io.c b/fs/bcachefs/btree_io.c index bd86dd7151a1..2bd8422eb72c 100644 --- a/fs/bcachefs/btree_io.c +++ b/fs/bcachefs/btree_io.c @@ -563,15 +563,11 @@ static int __btree_err(int ret, struct printbuf *err_msg, const char *fmt, ...) { - if (c->recovery.curr_pass == BCH_RECOVERY_PASS_scan_for_btree_nodes) - return ret == -BCH_ERR_btree_node_read_err_fixable - ? bch_err_throw(c, fsck_fix) - : ret; - + bool in_scan = c->recovery.curr_pass == BCH_RECOVERY_PASS_scan_for_btree_nodes; bool have_retry = false; int ret2; - if (ca) { + if (ca && !in_scan) { bch2_mark_btree_validate_failure(failed, ca->dev_idx); struct extent_ptr_decoded pick; @@ -585,12 +581,14 @@ static int __btree_err(int ret, if (!have_retry && ret == -BCH_ERR_btree_node_read_err_must_retry) ret = bch_err_throw(c, btree_node_read_err_bad_node); - bch2_sb_error_count(c, err_type); + if (!in_scan) + bch2_sb_error_count(c, err_type); bool print_deferred = err_msg && rw == READ && - !(test_bit(BCH_FS_in_fsck, &c->flags) && - c->opts.fix_errors == FSCK_FIX_ask); + (!(test_bit(BCH_FS_in_fsck, &c->flags) && + c->opts.fix_errors == FSCK_FIX_ask) || + in_scan); CLASS(printbuf, out)(); bch2_log_msg_start(c, &out); @@ -603,11 +601,17 @@ static int __btree_err(int ret, va_list args; va_start(args, fmt); prt_vprintf(err_msg, fmt, args); - va_end(args); + va_end(args);; - if (print_deferred) { + if (print_deferred) prt_newline(err_msg); + if (in_scan) + return ret == -BCH_ERR_btree_node_read_err_fixable + ? bch_err_throw(c, fsck_fix) + : ret; + + if (print_deferred) { switch (ret) { case -BCH_ERR_btree_node_read_err_fixable: ret2 = bch2_fsck_err_opt(c, FSCK_CAN_FIX, err_type); @@ -1405,10 +1409,8 @@ static void btree_node_read_work(struct work_struct *work) ret = bch2_bkey_pick_read_device(c, bkey_i_to_s_c(&b->key), &failed, &rb->pick, -1); - if (ret <= 0) { - set_btree_node_read_error(b); + if (ret <= 0) break; - } ca = bch2_dev_get_ioref(c, rb->pick.ptr.dev, READ, BCH_DEV_READ_REF_btree_node_read); rb->have_ioref = ca != NULL; @@ -1442,27 +1444,21 @@ start: bch2_maybe_corrupt_bio(bio, bch2_btree_read_corrupt_ratio); ret = bch2_btree_node_read_done(c, ca, b, &failed, &buf); - if (ret == -BCH_ERR_btree_node_read_err_want_retry || - ret == -BCH_ERR_btree_node_read_err_must_retry) - continue; - - if (ret) - set_btree_node_read_error(b); - - break; + if (ret != -BCH_ERR_btree_node_read_err_want_retry && + ret != -BCH_ERR_btree_node_read_err_must_retry) + break; } bch2_io_failures_to_text(&buf, c, &failed); - if (btree_node_read_error(b)) - bch2_btree_lost_data(c, &buf, b->c.btree_id); - /* * only print retry success if we read from a replica with no errors */ - if (btree_node_read_error(b)) + if (ret) { + set_btree_node_read_error(b); + bch2_btree_lost_data(c, &buf, b->c.btree_id); prt_printf(&buf, "ret %s", bch2_err_str(ret)); - else if (failed.nr) { + } else if (failed.nr) { if (!bch2_dev_io_failures(&failed, rb->pick.ptr.dev)) prt_printf(&buf, "retry success"); else diff --git a/fs/bcachefs/btree_node_scan.c b/fs/bcachefs/btree_node_scan.c index f4f958f4615d..c518e18ce5dc 100644 --- a/fs/bcachefs/btree_node_scan.c +++ b/fs/bcachefs/btree_node_scan.c @@ -155,17 +155,6 @@ static void try_read_btree_node(struct find_btree_nodes *f, struct bch_dev *ca, if (BTREE_NODE_LEVEL(bn) >= BTREE_MAX_DEPTH) return; - if (BTREE_NODE_ID(bn) >= BTREE_ID_NR_MAX) - return; - - bio_reset(bio, ca->disk_sb.bdev, REQ_OP_READ); - bio->bi_iter.bi_sector = offset; - bch2_bio_map(bio, b->data, c->opts.btree_node_size); - - submit_time = local_clock(); - submit_bio_wait(bio); - bch2_account_io_completion(ca, BCH_MEMBER_ERROR_read, submit_time, !bio->bi_status); - rcu_read_lock(); struct found_btree_node n = { .btree_id = BTREE_NODE_ID(bn), @@ -182,10 +171,26 @@ static void try_read_btree_node(struct find_btree_nodes *f, struct bch_dev *ca, }; rcu_read_unlock(); + bio_reset(bio, ca->disk_sb.bdev, REQ_OP_READ); + bio->bi_iter.bi_sector = offset; + bch2_bio_map(bio, b->data, c->opts.btree_node_size); + + submit_time = local_clock(); + submit_bio_wait(bio); + bch2_account_io_completion(ca, BCH_MEMBER_ERROR_read, submit_time, !bio->bi_status); + found_btree_node_to_key(&b->key, &n); CLASS(printbuf, buf)(); - if (!bch2_btree_node_read_done(c, ca, b, NULL, &buf)) { + + found_btree_node_to_text(&buf, c, &n); + prt_newline(&buf); + + int ret = bch2_btree_node_read_done(c, ca, b, NULL, &buf); + + bch_verbose(ca, "attempted to read, ret %s\n%s", bch2_err_str(ret), buf.buf); + + if (!ret) { /* read_done will swap out b->data for another buffer */ bn = b->data; /* diff --git a/fs/bcachefs/extents.c b/fs/bcachefs/extents.c index 8152ef1cbbcd..b879a586b7f6 100644 --- a/fs/bcachefs/extents.c +++ b/fs/bcachefs/extents.c @@ -282,9 +282,9 @@ int bch2_bkey_pick_read_device(struct bch_fs *c, struct bkey_s_c k, if (have_pick) return 1; - if (!have_dirty_ptrs) + if (!have_dirty_ptrs && !bkey_is_btree_ptr(k.k)) return 0; - if (have_missing_devs) + if (have_missing_devs || !have_dirty_ptrs) return bch_err_throw(c, no_device_to_read_from); if (have_csum_errors) return bch_err_throw(c, data_read_csum_err); diff --git a/fs/bcachefs/io_write.c b/fs/bcachefs/io_write.c index d7620138e038..44b02d4b6502 100644 --- a/fs/bcachefs/io_write.c +++ b/fs/bcachefs/io_write.c @@ -89,7 +89,12 @@ void bch2_latency_acct(struct bch_dev *ca, u64 submit_time, int rw) new = ewma_add(old, io_latency, 5); } while (!atomic64_try_cmpxchg(latency, &old, new)); - bch2_congested_acct(ca, io_latency, now, rw); + /* + * Only track read latency for congestion accounting: writes are subject + * to heavy queuing delays from page cache writeback: + */ + if (rw == READ) + bch2_congested_acct(ca, io_latency, now, rw); __bch2_time_stats_update(&ca->io_latency[rw].stats, submit_time, now); } diff --git a/fs/bcachefs/journal.c b/fs/bcachefs/journal.c index 97760e89e5a3..69becd504985 100644 --- a/fs/bcachefs/journal.c +++ b/fs/bcachefs/journal.c @@ -1472,6 +1472,10 @@ int bch2_fs_journal_start(struct journal *j, u64 last_seq, u64 cur_seq) last_seq = cur_seq; u64 nr = cur_seq - last_seq; + if (nr * sizeof(journal_entry_pin_list) > 1U << 30) { + bch_err(c, "too many ntjournal fifo (%llu open entries)", nr); + return bch_err_throw(c, ENOMEM_journal_pin_fifo); + } /* * Extra fudge factor, in case we crashed when the journal pin fifo was @@ -1484,7 +1488,7 @@ int bch2_fs_journal_start(struct journal *j, u64 last_seq, u64 cur_seq) nr = max(nr, JOURNAL_PIN); init_fifo(&j->pin, roundup_pow_of_two(nr), GFP_KERNEL); if (!j->pin.data) { - bch_err(c, "error reallocating journal fifo (%llu open entries)", nr); + bch_err(c, "error allocating journal fifo (%llu open entries)", nr); return bch_err_throw(c, ENOMEM_journal_pin_fifo); } diff --git a/fs/bcachefs/recovery.c b/fs/bcachefs/recovery.c index 58c159e5f10d..304473dac268 100644 --- a/fs/bcachefs/recovery.c +++ b/fs/bcachefs/recovery.c @@ -67,13 +67,16 @@ int bch2_btree_lost_data(struct bch_fs *c, ret = __bch2_run_explicit_recovery_pass(c, msg, BCH_RECOVERY_PASS_check_backpointers_to_extents, 0, &write_sb) ?: ret; #endif + write_sb |= !__test_and_set_bit_le64(BCH_FSCK_ERR_lru_entry_bad, ext->errors_silent); + write_sb |= !__test_and_set_bit_le64(BCH_FSCK_ERR_backpointer_to_missing_ptr, ext->errors_silent); + write_sb |= !__test_and_set_bit_le64(BCH_FSCK_ERR_alloc_key_data_type_wrong, ext->errors_silent); + write_sb |= !__test_and_set_bit_le64(BCH_FSCK_ERR_alloc_key_dirty_sectors_wrong, ext->errors_silent); + switch (btree) { case BTREE_ID_alloc: ret = __bch2_run_explicit_recovery_pass(c, msg, BCH_RECOVERY_PASS_check_alloc_info, 0, &write_sb) ?: ret; - write_sb |= !__test_and_set_bit_le64(BCH_FSCK_ERR_alloc_key_data_type_wrong, ext->errors_silent); write_sb |= !__test_and_set_bit_le64(BCH_FSCK_ERR_alloc_key_gen_wrong, ext->errors_silent); - write_sb |= !__test_and_set_bit_le64(BCH_FSCK_ERR_alloc_key_dirty_sectors_wrong, ext->errors_silent); write_sb |= !__test_and_set_bit_le64(BCH_FSCK_ERR_alloc_key_cached_sectors_wrong, ext->errors_silent); write_sb |= !__test_and_set_bit_le64(BCH_FSCK_ERR_alloc_key_stripe_wrong, ext->errors_silent); write_sb |= !__test_and_set_bit_le64(BCH_FSCK_ERR_alloc_key_stripe_redundancy_wrong, ext->errors_silent); diff --git a/fs/bcachefs/super.c b/fs/bcachefs/super.c index 0fc0b2221036..b3b2d8353a36 100644 --- a/fs/bcachefs/super.c +++ b/fs/bcachefs/super.c @@ -729,6 +729,8 @@ void __bch2_fs_stop(struct bch_fs *c) cancel_work_sync(&ca->io_error_work); cancel_work_sync(&c->read_only_work); + + flush_work(&c->btree_interior_update_work); } void bch2_fs_free(struct bch_fs *c) diff --git a/include/linux/workqueue.h b/include/linux/workqueue.h index 6e30f275da77..e907c9bb840c 100644 --- a/include/linux/workqueue.h +++ b/include/linux/workqueue.h @@ -6,6 +6,7 @@ #ifndef _LINUX_WORKQUEUE_H #define _LINUX_WORKQUEUE_H +#include <linux/alloc_tag.h> #include <linux/timer.h> #include <linux/linkage.h> #include <linux/bitops.h> @@ -505,7 +506,8 @@ void workqueue_softirq_dead(unsigned int cpu); * Pointer to the allocated workqueue on success, %NULL on failure. */ __printf(1, 4) struct workqueue_struct * -alloc_workqueue(const char *fmt, unsigned int flags, int max_active, ...); +alloc_workqueue_noprof(const char *fmt, unsigned int flags, int max_active, ...); +#define alloc_workqueue(...) alloc_hooks(alloc_workqueue_noprof(__VA_ARGS__)) #ifdef CONFIG_LOCKDEP /** @@ -544,8 +546,8 @@ alloc_workqueue_lockdep_map(const char *fmt, unsigned int flags, int max_active, * Pointer to the allocated workqueue on success, %NULL on failure. */ #define alloc_ordered_workqueue_lockdep_map(fmt, flags, lockdep_map, args...) \ - alloc_workqueue_lockdep_map(fmt, WQ_UNBOUND | __WQ_ORDERED | (flags), \ - 1, lockdep_map, ##args) + alloc_hooks(alloc_workqueue_lockdep_map(fmt, WQ_UNBOUND | __WQ_ORDERED | (flags),\ + 1, lockdep_map, ##args)) #endif /** @@ -577,7 +579,9 @@ alloc_workqueue_lockdep_map(const char *fmt, unsigned int flags, int max_active, extern void destroy_workqueue(struct workqueue_struct *wq); -struct workqueue_attrs *alloc_workqueue_attrs(void); +struct workqueue_attrs *alloc_workqueue_attrs_noprof(void); +#define alloc_workqueue_attrs(...) alloc_hooks(alloc_workqueue_attrs_noprof(__VA_ARGS__)) + void free_workqueue_attrs(struct workqueue_attrs *attrs); int apply_workqueue_attrs(struct workqueue_struct *wq, const struct workqueue_attrs *attrs); diff --git a/kernel/workqueue.c b/kernel/workqueue.c index 9f9148075828..992cb0467c21 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -4629,7 +4629,7 @@ void free_workqueue_attrs(struct workqueue_attrs *attrs) * * Return: The allocated new workqueue_attr on success. %NULL on failure. */ -struct workqueue_attrs *alloc_workqueue_attrs(void) +struct workqueue_attrs *alloc_workqueue_attrs_noprof(void) { struct workqueue_attrs *attrs; @@ -5682,12 +5682,12 @@ static struct workqueue_struct *__alloc_workqueue(const char *fmt, else wq_size = sizeof(*wq); - wq = kzalloc(wq_size, GFP_KERNEL); + wq = kzalloc_noprof(wq_size, GFP_KERNEL); if (!wq) return NULL; if (flags & WQ_UNBOUND) { - wq->unbound_attrs = alloc_workqueue_attrs(); + wq->unbound_attrs = alloc_workqueue_attrs_noprof(); if (!wq->unbound_attrs) goto err_free_wq; } @@ -5777,9 +5777,9 @@ err_destroy: } __printf(1, 4) -struct workqueue_struct *alloc_workqueue(const char *fmt, - unsigned int flags, - int max_active, ...) +struct workqueue_struct *alloc_workqueue_noprof(const char *fmt, + unsigned int flags, + int max_active, ...) { struct workqueue_struct *wq; va_list args; @@ -5794,7 +5794,7 @@ struct workqueue_struct *alloc_workqueue(const char *fmt, return wq; } -EXPORT_SYMBOL_GPL(alloc_workqueue); +EXPORT_SYMBOL_GPL(alloc_workqueue_noprof); #ifdef CONFIG_LOCKDEP __printf(1, 5) |