Update bcachefs sources to 969fbff4ef bcachefs: Print message on btree node read retry success

.bcachefs_revision

@@ -1 +1 @@
-40eaef7e8049b75ff7e5da42227295c754d9c906
+969fbff4ef3a75ae25ef7cca17dd4e028443bfc2

libbcachefs/bcachefs_format.h

@@ -1553,6 +1553,7 @@ LE64_BITMASK(BCH_SB_INODES_USE_KEY_CACHE,struct bch_sb, flags[3], 29, 30);
 LE64_BITMASK(BCH_SB_JOURNAL_FLUSH_DELAY,struct bch_sb, flags[3], 30, 62);
 LE64_BITMASK(BCH_SB_JOURNAL_FLUSH_DISABLED,struct bch_sb, flags[3], 62, 63);
 LE64_BITMASK(BCH_SB_JOURNAL_RECLAIM_DELAY,struct bch_sb, flags[4], 0, 32);
+/* Obsolete, always enabled: */
 LE64_BITMASK(BCH_SB_JOURNAL_TRANSACTION_NAMES,struct bch_sb, flags[4], 32, 33);
 
 /*

libbcachefs/btree_io.c

@@ -1146,10 +1146,12 @@ static void btree_node_read_work(struct work_struct *work)
 	struct bch_io_failures failed = { .nr = 0 };
 	struct printbuf buf = PRINTBUF;
 	bool saw_error = false;
+	bool retry = false;
 	bool can_retry;
 
 	goto start;
 	while (1) {
+		retry = true;
 		bch_info(c, "retrying read");
 		ca = bch_dev_bkey_exists(c, rb->pick.ptr.dev);
 		rb->have_ioref		= bch2_dev_get_ioref(ca, READ);
@@ -1180,8 +1182,11 @@ start:
 				&failed, &rb->pick) > 0;
 
 		if (!bio->bi_status &&
-		    !bch2_btree_node_read_done(c, ca, b, can_retry))
+		    !bch2_btree_node_read_done(c, ca, b, can_retry)) {
+			if (retry)
+				bch_info(c, "retry success");
 			break;
+		}
 
 		saw_error = true;
 

libbcachefs/btree_iter.c

@@ -2248,14 +2248,38 @@ static inline struct bkey_i *btree_trans_peek_updates(struct btree_trans *trans,
 	return NULL;
 }
 
+struct bkey_i *bch2_btree_journal_peek(struct btree_trans *trans,
+				       struct btree_iter *iter,
+				       struct bpos start_pos,
+				       struct bpos end_pos)
+{
+	struct bkey_i *k;
+
+	if (bpos_cmp(start_pos, iter->journal_pos) < 0)
+		iter->journal_idx = 0;
+
+	k = bch2_journal_keys_peek_upto(trans->c, iter->btree_id, 0,
+					start_pos, end_pos,
+					&iter->journal_idx);
+
+	iter->journal_pos = k ? k->k.p : end_pos;
+	return k;
+}
+
+struct bkey_i *bch2_btree_journal_peek_slot(struct btree_trans *trans,
+					    struct btree_iter *iter,
+					    struct bpos pos)
+{
+	return bch2_btree_journal_peek(trans, iter, pos, pos);
+}
+
 static noinline
 struct bkey_s_c btree_trans_peek_journal(struct btree_trans *trans,
 					 struct btree_iter *iter,
 					 struct bkey_s_c k)
 {
 	struct bkey_i *next_journal =
-		bch2_journal_keys_peek_upto(trans->c, iter->btree_id, 0,
+		bch2_btree_journal_peek(trans, iter, iter->path->pos,
-				iter->path->pos,
 				k.k ? k.k->p : iter->path->l[0].b->key.k.p);
 
 	if (next_journal) {
@@ -2804,10 +2828,8 @@ struct bkey_s_c bch2_btree_iter_peek_slot(struct btree_iter *iter)
 		}
 
 		if (unlikely(iter->flags & BTREE_ITER_WITH_JOURNAL) &&
-		    (next_update = bch2_journal_keys_peek_slot(trans->c,
+		    (next_update = bch2_btree_journal_peek_slot(trans,
-					iter->btree_id,
+					iter, iter->pos))) {
-					iter->path->level,
-					iter->pos))) {
 			iter->k = next_update->k;
 			k = bkey_i_to_s_c(next_update);
 			goto out;
@@ -3074,6 +3096,8 @@ static void __bch2_trans_iter_init(struct btree_trans *trans,
 	iter->k.type	= KEY_TYPE_deleted;
 	iter->k.p	= pos;
 	iter->k.size	= 0;
+	iter->journal_idx = 0;
+	iter->journal_pos = POS_MIN;
 #ifdef CONFIG_BCACHEFS_DEBUG
 	iter->ip_allocated = ip;
 #endif

libbcachefs/btree_iter.h

@@ -138,6 +138,9 @@ struct btree_path *bch2_path_get(struct btree_trans *, enum btree_id, struct bpo
 				 unsigned, unsigned, unsigned, unsigned long);
 inline struct bkey_s_c bch2_btree_path_peek_slot(struct btree_path *, struct bkey *);
 
+struct bkey_i *bch2_btree_journal_peek_slot(struct btree_trans *,
+					struct btree_iter *, struct bpos);
+
 #ifdef CONFIG_BCACHEFS_DEBUG
 void bch2_trans_verify_paths(struct btree_trans *);
 void bch2_trans_verify_locks(struct btree_trans *);

libbcachefs/btree_types.h

@@ -292,6 +292,10 @@ struct btree_iter {
 	 * bch2_btree_iter_next_slot() can correctly advance pos.
 	 */
 	struct bkey		k;
+
+	/* BTREE_ITER_WITH_JOURNAL: */
+	size_t			journal_idx;
+	struct bpos		journal_pos;
 #ifdef CONFIG_BCACHEFS_DEBUG
 	unsigned long		ip_allocated;
 #endif
@@ -393,7 +397,6 @@ struct btree_trans {
 	bool			in_traverse_all:1;
 	bool			restarted:1;
 	bool			memory_allocation_failure:1;
-	bool			journal_transaction_names:1;
 	bool			is_initial_gc:1;
 	/*
 	 * For when bch2_trans_update notices we'll be splitting a compressed

libbcachefs/btree_update_leaf.c

@@ -305,7 +305,7 @@ static inline int bch2_trans_journal_res_get(struct btree_trans *trans,
 
 #define JSET_ENTRY_LOG_U64s		4
 
-static noinline void journal_transaction_name(struct btree_trans *trans)
+static void journal_transaction_name(struct btree_trans *trans)
 {
 	struct bch_fs *c = trans->c;
 	struct jset_entry *entry = journal_res_entry(&c->journal, &trans->journal_res);
@@ -684,8 +684,7 @@ bch2_trans_commit_write_locked(struct btree_trans *trans,
 		if (ret)
 			return ret;
 
-		if (unlikely(trans->journal_transaction_names))
+		journal_transaction_name(trans);
-			journal_transaction_name(trans);
 	} else {
 		trans->journal_res.seq = c->journal.replay_journal_seq;
 	}
@@ -1118,10 +1117,8 @@ int __bch2_trans_commit(struct btree_trans *trans)
 	trans->journal_u64s		= trans->extra_journal_entries.nr;
 	trans->journal_preres_u64s	= 0;
 
-	trans->journal_transaction_names = READ_ONCE(c->opts.journal_transaction_names);
+	/* For journalling transaction name: */
-
+	trans->journal_u64s += JSET_ENTRY_LOG_U64s;
-	if (trans->journal_transaction_names)
-		trans->journal_u64s += JSET_ENTRY_LOG_U64s;
 
 	trans_for_each_update(trans, i) {
 		BUG_ON(!i->path->should_be_locked);

libbcachefs/checksum.c

@@ -116,7 +116,12 @@ static inline int do_encrypt(struct crypto_sync_skcipher *tfm,
 {
 	struct scatterlist sg;
 
-	sg_init_one(&sg, buf, len);
+	sg_init_table(&sg, 1);
+	sg_set_page(&sg,
+		    is_vmalloc_addr(buf)
+		    ? vmalloc_to_page(buf)
+		    : virt_to_page(buf),
+		    len, offset_in_page(buf));
 	return do_encrypt_sg(tfm, nonce, &sg, len);
 }
 

libbcachefs/opts.h

@@ -165,22 +165,22 @@ enum opt_type {
 	  OPT_FS|OPT_INODE|OPT_FORMAT|OPT_MOUNT|OPT_RUNTIME,		\
 	  OPT_FN(bch2_opt_target),					\
 	  BCH_SB_METADATA_TARGET,	0,				\
-	  "(target)",	"Device or disk group for metadata writes")	\
+	  "(target)",	"Device or label for metadata writes")		\
 	x(foreground_target,		u16,				\
 	  OPT_FS|OPT_INODE|OPT_FORMAT|OPT_MOUNT|OPT_RUNTIME,		\
 	  OPT_FN(bch2_opt_target),					\
 	  BCH_SB_FOREGROUND_TARGET,	0,				\
-	  "(target)",	"Device or disk group for foreground writes")	\
+	  "(target)",	"Device or label for foreground writes")	\
 	x(background_target,		u16,				\
 	  OPT_FS|OPT_INODE|OPT_FORMAT|OPT_MOUNT|OPT_RUNTIME,		\
 	  OPT_FN(bch2_opt_target),					\
 	  BCH_SB_BACKGROUND_TARGET,	0,				\
-	  "(target)",	"Device or disk group to move data to in the background")\
+	  "(target)",	"Device or label to move data to in the background")\
 	x(promote_target,		u16,				\
 	  OPT_FS|OPT_INODE|OPT_FORMAT|OPT_MOUNT|OPT_RUNTIME,		\
 	  OPT_FN(bch2_opt_target),					\
 	  BCH_SB_PROMOTE_TARGET,	0,				\
-	  "(target)",	"Device or disk group to promote data to on read")\
+	  "(target)",	"Device or label to promote data to on read")	\
 	x(erasure_code,			u16,				\
 	  OPT_FS|OPT_INODE|OPT_FORMAT|OPT_MOUNT|OPT_RUNTIME,		\
 	  OPT_BOOL(),							\
@@ -331,11 +331,6 @@ enum opt_type {
 	  OPT_BOOL(),							\
 	  BCH2_NO_SB_OPT,			false,				\
 	  NULL,		"Only read the journal, skip the rest of recovery")\
-	x(journal_transaction_names,	u8,				\
-	  OPT_FS|OPT_FORMAT|OPT_MOUNT|OPT_RUNTIME,			\
-	  OPT_BOOL(),							\
-	  BCH_SB_JOURNAL_TRANSACTION_NAMES, true,			\
-	  NULL,		"Log transaction function names in journal")	\
 	x(noexcl,			u8,				\
 	  OPT_FS|OPT_MOUNT,						\
 	  OPT_BOOL(),							\

libbcachefs/recovery.c

@@ -86,9 +86,9 @@ static inline struct journal_key *idx_to_key(struct journal_keys *keys, size_t i
 	return keys->d + idx_to_pos(keys, idx);
 }
 
-size_t bch2_journal_key_search(struct journal_keys *keys,
+static size_t bch2_journal_key_search(struct journal_keys *keys,
-			       enum btree_id id, unsigned level,
+				      enum btree_id id, unsigned level,
-			       struct bpos pos)
+				      struct bpos pos)
 {
 	size_t l = 0, r = keys->nr, m;
 
@@ -111,21 +111,31 @@ size_t bch2_journal_key_search(struct journal_keys *keys,
 
 struct bkey_i *bch2_journal_keys_peek_upto(struct bch_fs *c, enum btree_id btree_id,
 					   unsigned level, struct bpos pos,
-					   struct bpos end_pos)
+					   struct bpos end_pos, size_t *idx)
 {
 	struct journal_keys *keys = &c->journal_keys;
-	size_t idx = bch2_journal_key_search(keys, btree_id, level, pos);
+	unsigned iters = 0;
+search:
+	if (!*idx)
+		*idx = bch2_journal_key_search(keys, btree_id, level, pos);
 
-	while (idx < keys->size &&
+	while (*idx < keys->size &&
-	       keys->d[idx].btree_id == btree_id &&
+	       keys->d[*idx].btree_id == btree_id &&
-	       keys->d[idx].level == level &&
+	       keys->d[*idx].level == level &&
-	       bpos_cmp(keys->d[idx].k->k.p, end_pos) <= 0) {
+	       bpos_cmp(keys->d[*idx].k->k.p, end_pos) <= 0) {
-		if (!keys->d[idx].overwritten)
+		if (bpos_cmp(keys->d[*idx].k->k.p, pos) >= 0 &&
-			return keys->d[idx].k;
+		    !keys->d[*idx].overwritten)
+			return keys->d[*idx].k;
 
-		idx++;
+		(*idx)++;
-		if (idx == keys->gap)
+		if (*idx == keys->gap)
-			idx += keys->size - keys->nr;
+			*idx += keys->size - keys->nr;
+
+		iters++;
+		if (iters == 10) {
+			*idx = 0;
+			goto search;
+		}
 	}
 
 	return NULL;
@@ -134,7 +144,9 @@ struct bkey_i *bch2_journal_keys_peek_upto(struct bch_fs *c, enum btree_id btree
 struct bkey_i *bch2_journal_keys_peek_slot(struct bch_fs *c, enum btree_id btree_id,
 					   unsigned level, struct bpos pos)
 {
-	return bch2_journal_keys_peek_upto(c, btree_id, level, pos, pos);
+	size_t idx = 0;
+
+	return bch2_journal_keys_peek_upto(c, btree_id, level, pos, pos, &idx);
 }
 
 static void journal_iters_fix(struct bch_fs *c)
@@ -1246,7 +1258,7 @@ use_clean:
 
 		set_bit(BCH_FS_MAY_GO_RW, &c->flags);
 
-		bch_verbose(c, "starting journal replay, %zu keys", c->journal_keys.nr);
+		bch_info(c, "starting journal replay, %zu keys", c->journal_keys.nr);
 		err = "journal replay failed";
 		ret = bch2_journal_replay(c);
 		if (ret)

libbcachefs/recovery.h

@@ -28,10 +28,8 @@ struct btree_and_journal_iter {
 	}			last;
 };
 
-size_t bch2_journal_key_search(struct journal_keys *, enum btree_id,
-			       unsigned, struct bpos);
 struct bkey_i *bch2_journal_keys_peek_upto(struct bch_fs *, enum btree_id,
-					   unsigned, struct bpos, struct bpos);
+				unsigned, struct bpos, struct bpos, size_t *);
 struct bkey_i *bch2_journal_keys_peek_slot(struct bch_fs *, enum btree_id,
 					   unsigned, struct bpos);
 

libbcachefs/super-io.c

@@ -1058,7 +1058,7 @@ static void bch2_sb_members_to_text(struct printbuf *out, struct bch_sb *sb,
 		       : "unknown");
 		pr_newline(out);
 
-		pr_buf(out, "Group:");
+		pr_buf(out, "Label:");
 		pr_tab(out);
 		if (BCH_MEMBER_GROUP(m)) {
 			unsigned idx = BCH_MEMBER_GROUP(m) - 1;