summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
Diffstat
-rw-r--r--fs/bcachefs/bcachefs.h6
-rw-r--r--fs/bcachefs/errcode.h1
-rw-r--r--fs/bcachefs/fs-io-buffered.c56
-rw-r--r--fs/bcachefs/fs.c8
-rw-r--r--fs/bcachefs/inode.c5
-rw-r--r--fs/bcachefs/io_read.c55
-rw-r--r--fs/bcachefs/sb-counters_format.h2
-rw-r--r--fs/bcachefs/sb-errors_format.h2
-rw-r--r--fs/bcachefs/super.c2
9 files changed, 91 insertions, 46 deletions
diff --git a/fs/bcachefs/bcachefs.h b/fs/bcachefs/bcachefs.h
index 553031a3b06a..83d6ab9c1a91 100644
--- a/fs/bcachefs/bcachefs.h
+++ b/fs/bcachefs/bcachefs.h
@@ -458,7 +458,6 @@ BCH_DEBUG_PARAMS_ALL()
x(btree_node_compact) \
x(btree_node_merge) \
x(btree_node_sort) \
- x(btree_node_get) \
x(btree_node_read) \
x(btree_node_read_done) \
x(btree_node_write) \
@@ -466,10 +465,6 @@ BCH_DEBUG_PARAMS_ALL()
x(btree_interior_update_total) \
x(btree_gc) \
x(data_write) \
- x(data_write_to_submit) \
- x(data_write_to_queue) \
- x(data_write_to_btree_update) \
- x(data_write_btree_update) \
x(data_read) \
x(data_promote) \
x(journal_flush_write) \
@@ -483,6 +478,7 @@ BCH_DEBUG_PARAMS_ALL()
x(blocked_allocate) \
x(blocked_allocate_open_bucket) \
x(blocked_write_buffer_full) \
+ x(blocked_writeback_throttle) \
x(nocow_lock_contended)
enum bch_time_stats {
diff --git a/fs/bcachefs/errcode.h b/fs/bcachefs/errcode.h
index adc1f9315eab..420f6922dacb 100644
--- a/fs/bcachefs/errcode.h
+++ b/fs/bcachefs/errcode.h
@@ -345,6 +345,7 @@
x(BCH_ERR_data_read, data_read_no_encryption_key) \
x(BCH_ERR_data_read, data_read_buffer_too_small) \
x(BCH_ERR_data_read, data_read_key_overwritten) \
+ x(0, rbio_narrow_crcs_fail) \
x(BCH_ERR_btree_node_read_err, btree_node_read_err_fixable) \
x(BCH_ERR_btree_node_read_err, btree_node_read_err_want_retry) \
x(BCH_ERR_btree_node_read_err, btree_node_read_err_must_retry) \
diff --git a/fs/bcachefs/fs-io-buffered.c b/fs/bcachefs/fs-io-buffered.c
index aab30571b056..fe684adca370 100644
--- a/fs/bcachefs/fs-io-buffered.c
+++ b/fs/bcachefs/fs-io-buffered.c
@@ -532,6 +532,39 @@ static void bch2_writepage_io_alloc(struct bch_fs *c,
op->wbio.bio.bi_opf = wbc_to_write_flags(wbc);
}
+static bool can_write_now(struct bch_fs *c, unsigned replicas_want, struct closure *cl)
+{
+ unsigned reserved = OPEN_BUCKETS_COUNT -
+ (OPEN_BUCKETS_COUNT - bch2_open_buckets_reserved(BCH_WATERMARK_normal)) / 2;
+
+ if (unlikely(c->open_buckets_nr_free <= reserved)) {
+ closure_wait(&c->open_buckets_wait, cl);
+ return false;
+ }
+
+ if (BCH_WATERMARK_normal < c->journal.watermark && !bch2_journal_error(&c->journal)) {
+ closure_wait(&c->journal.async_wait, cl);
+ return false;
+ }
+
+ return true;
+}
+
+static void throttle_writes(struct bch_fs *c, unsigned replicas_want, struct closure *cl)
+{
+ u64 start = 0;
+ while (!can_write_now(c, replicas_want, cl)) {
+ if (!start)
+ start = local_clock();
+ closure_sync(cl);
+ }
+
+ BUG_ON(closure_nr_remaining(cl) > 1);
+
+ if (start)
+ bch2_time_stats_update(&c->times[BCH_TIME_blocked_writeback_throttle], start);
+}
+
static int __bch2_writepage(struct folio *folio,
struct writeback_control *wbc,
void *data)
@@ -667,17 +700,6 @@ do_io:
return 0;
}
-static int bch2_write_cache_pages(struct address_space *mapping,
- struct writeback_control *wbc, void *data)
-{
- struct folio *folio = NULL;
- int error;
-
- while ((folio = writeback_iter(mapping, wbc, folio, &error)))
- error = __bch2_writepage(folio, wbc, data);
- return error;
-}
-
int bch2_writepages(struct address_space *mapping, struct writeback_control *wbc)
{
struct bch_fs *c = mapping->host->i_sb->s_fs_info;
@@ -686,7 +708,17 @@ int bch2_writepages(struct address_space *mapping, struct writeback_control *wbc
bch2_inode_opts_get_inode(c, &to_bch_ei(mapping->host)->ei_inode, &w->opts);
blk_start_plug(&w->plug);
- int ret = bch2_write_cache_pages(mapping, wbc, w);
+
+ struct closure cl;
+ closure_init_stack(&cl);
+
+ struct folio *folio = NULL;
+ int ret = 0;
+
+ while (throttle_writes(c, w->opts.data_replicas, &cl),
+ (folio = writeback_iter(mapping, wbc, folio, &ret)))
+ ret = __bch2_writepage(folio, wbc, w);
+
if (w->io)
bch2_writepage_do_io(w);
blk_finish_plug(&w->plug);
diff --git a/fs/bcachefs/fs.c b/fs/bcachefs/fs.c
index c3839af067f2..d6a2031e17e8 100644
--- a/fs/bcachefs/fs.c
+++ b/fs/bcachefs/fs.c
@@ -2147,9 +2147,11 @@ static void bch2_evict_inode(struct inode *vinode)
KEY_TYPE_QUOTA_WARN);
int ret = bch2_inode_rm(c, inode_inum(inode));
if (ret && !bch2_err_matches(ret, EROFS)) {
- bch_err_msg(c, ret, "VFS incorrectly tried to delete inode %llu:%llu",
- inode->ei_inum.subvol,
- inode->ei_inum.inum);
+ CLASS(printbuf, buf)();
+ bch2_trans_do(c, bch2_inum_to_path(trans, inode->ei_inum, &buf));
+
+ bch_err_msg(c, ret, "VFS incorrectly tried to delete inode %llu:%llu\n%s",
+ inode->ei_inum.subvol, inode->ei_inum.inum, buf.buf);
bch2_sb_error_count(c, BCH_FSCK_ERR_vfs_bad_inode_rm);
}
diff --git a/fs/bcachefs/inode.c b/fs/bcachefs/inode.c
index 655ed90b2a39..543627fb58be 100644
--- a/fs/bcachefs/inode.c
+++ b/fs/bcachefs/inode.c
@@ -1359,7 +1359,7 @@ err:
if (bch2_err_matches(ret, BCH_ERR_transaction_restart))
goto retry;
- return ret ?: bch_err_throw(c, transaction_restart_nested);
+ return ret;
}
/*
@@ -1398,7 +1398,8 @@ next_parent:
int bch2_inode_rm_snapshot(struct btree_trans *trans, u64 inum, u32 snapshot)
{
return __bch2_inode_rm_snapshot(trans, inum, snapshot) ?:
- delete_ancestor_snapshot_inodes(trans, SPOS(0, inum, snapshot));
+ delete_ancestor_snapshot_inodes(trans, SPOS(0, inum, snapshot)) ?:
+ bch_err_throw(trans->c, transaction_restart_nested);
}
static int may_delete_deleted_inode(struct btree_trans *trans, struct bpos pos,
diff --git a/fs/bcachefs/io_read.c b/fs/bcachefs/io_read.c
index 7066be2701c0..e7ba0d0bf5ef 100644
--- a/fs/bcachefs/io_read.c
+++ b/fs/bcachefs/io_read.c
@@ -740,15 +740,13 @@ static void bch2_rbio_error(struct bch_read_bio *rbio,
}
static int __bch2_rbio_narrow_crcs(struct btree_trans *trans,
- struct bch_read_bio *rbio)
+ struct bch_read_bio *rbio,
+ struct bch_extent_crc_unpacked *new_crc)
{
struct bch_fs *c = rbio->c;
u64 data_offset = rbio->data_pos.offset - rbio->pick.crc.offset;
int ret = 0;
- if (crc_is_compressed(rbio->pick.crc))
- return 0;
-
CLASS(btree_iter, iter)(trans, rbio->data_btree, rbio->data_pos, BTREE_ITER_intent);
struct bkey_s_c k = bch2_btree_iter_peek_slot(&iter);
if ((ret = bkey_err(k)))
@@ -756,21 +754,12 @@ static int __bch2_rbio_narrow_crcs(struct btree_trans *trans,
if (bversion_cmp(k.k->bversion, rbio->version) ||
!bch2_bkey_matches_ptr(c, k, rbio->pick.ptr, data_offset))
- return 0;
+ return bch_err_throw(c, rbio_narrow_crcs_fail);
- /* Extent was merged? */
- if (bkey_start_offset(k.k) < data_offset ||
- k.k->p.offset > data_offset + rbio->pick.crc.uncompressed_size)
- return 0;
-
- struct bch_extent_crc_unpacked new_crc;
- if (bch2_rechecksum_bio(c, &rbio->bio, rbio->version,
- rbio->pick.crc, NULL, &new_crc,
- bkey_start_offset(k.k) - data_offset, k.k->size,
- rbio->pick.crc.csum_type)) {
- bch_err(c, "error verifying existing checksum while narrowing checksum (memory corruption?)");
- return 0;
- }
+ /* Extent was trimmed/merged? */
+ if (!bpos_eq(bkey_start_pos(k.k), rbio->data_pos) ||
+ k.k->p.offset != rbio->data_pos.offset + rbio->pick.crc.live_size)
+ return bch_err_throw(c, rbio_narrow_crcs_fail);
/*
* going to be temporarily appending another checksum entry:
@@ -782,17 +771,37 @@ static int __bch2_rbio_narrow_crcs(struct btree_trans *trans,
bkey_reassemble(new, k);
- if (!bch2_bkey_narrow_crcs(new, new_crc))
- return 0;
+ if (!bch2_bkey_narrow_crcs(new, *new_crc))
+ return bch_err_throw(c, rbio_narrow_crcs_fail);
return bch2_trans_update(trans, &iter, new, BTREE_UPDATE_internal_snapshot_node);
}
static noinline void bch2_rbio_narrow_crcs(struct bch_read_bio *rbio)
{
- CLASS(btree_trans, trans)(rbio->c);
- commit_do(trans, NULL, NULL, BCH_TRANS_COMMIT_no_enospc,
- __bch2_rbio_narrow_crcs(trans, rbio));
+ struct bch_fs *c = rbio->c;
+
+ if (crc_is_compressed(rbio->pick.crc))
+ return;
+
+ u64 data_offset = rbio->data_pos.offset - rbio->pick.crc.offset;
+
+ struct bch_extent_crc_unpacked new_crc;
+ if (bch2_rechecksum_bio(c, &rbio->bio, rbio->version,
+ rbio->pick.crc, NULL, &new_crc,
+ rbio->data_pos.offset - data_offset, rbio->pick.crc.live_size,
+ rbio->pick.crc.csum_type)) {
+ bch_err(c, "error verifying existing checksum while narrowing checksum (memory corruption?)");
+ return;
+ }
+
+ CLASS(btree_trans, trans)(c);
+ int ret = commit_do(trans, NULL, NULL, BCH_TRANS_COMMIT_no_enospc,
+ __bch2_rbio_narrow_crcs(trans, rbio, &new_crc));
+ if (!ret)
+ count_event(c, io_read_narrow_crcs);
+ else if (ret == -BCH_ERR_rbio_narrow_crcs_fail)
+ count_event(c, io_read_narrow_crcs_fail);
}
static void bch2_read_decompress_err(struct work_struct *work)
diff --git a/fs/bcachefs/sb-counters_format.h b/fs/bcachefs/sb-counters_format.h
index 17cd617664d9..3907ba7edff2 100644
--- a/fs/bcachefs/sb-counters_format.h
+++ b/fs/bcachefs/sb-counters_format.h
@@ -23,6 +23,8 @@ enum counters_flags {
x(io_read_reuse_race, 34, TYPE_COUNTER) \
x(io_read_retry, 32, TYPE_COUNTER) \
x(io_read_fail_and_poison, 95, TYPE_COUNTER) \
+ x(io_read_narrow_crcs, 97, TYPE_COUNTER) \
+ x(io_read_narrow_crcs_fail, 98, TYPE_COUNTER) \
x(io_write, 1, TYPE_SECTORS) \
x(io_move, 2, TYPE_SECTORS) \
x(io_move_read, 35, TYPE_SECTORS) \
diff --git a/fs/bcachefs/sb-errors_format.h b/fs/bcachefs/sb-errors_format.h
index 7c6f18a1ee2a..728d878057af 100644
--- a/fs/bcachefs/sb-errors_format.h
+++ b/fs/bcachefs/sb-errors_format.h
@@ -160,7 +160,7 @@ enum bch_fsck_flags {
x(extent_ptrs_unwritten, 140, 0) \
x(extent_ptrs_written_and_unwritten, 141, 0) \
x(ptr_to_invalid_device, 142, 0) \
- x(ptr_to_removed_device, 322, 0) \
+ x(ptr_to_removed_device, 322, FSCK_AUTOFIX) \
x(ptr_to_duplicate_device, 143, 0) \
x(ptr_after_last_bucket, 144, 0) \
x(ptr_before_first_bucket, 145, 0) \
diff --git a/fs/bcachefs/super.c b/fs/bcachefs/super.c
index 97eb420cd2d4..473ad4b51180 100644
--- a/fs/bcachefs/super.c
+++ b/fs/bcachefs/super.c
@@ -1211,12 +1211,14 @@ static struct bch_fs *bch2_fs_alloc(struct bch_sb *sb, struct bch_opts *opts,
bch2_opts_apply(&c->opts, *opts);
+#ifdef __KERNEL__
if (!IS_ENABLED(CONFIG_TRANSPARENT_HUGEPAGE) &&
c->opts.block_size > PAGE_SIZE) {
bch_err(c, "cannot mount bs > ps filesystem without CONFIG_TRANSPARENT_HUGEPAGE");
ret = -EINVAL;
goto err;
}
+#endif
c->btree_key_cache_btrees |= 1U << BTREE_ID_alloc;
if (c->opts.inodes_use_key_cache)
generated by cgit v1.2.3 (git 2.46.0) at 2025-09-08 16:41:47 +0000