Update bcachefs sources to b47c5ba55d59 bcachefs: silence userspace build warning

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

.bcachefs_revision

@@ -1 +1 @@
-e02a6a521ba07a404c589f2873fcd9cb8189c76e
+b47c5ba55d592855bdfc77883f41207bdad56248

include/linux/bio.h

@@ -284,7 +284,7 @@ do {						\
 	(dst)->bi_bdev = (src)->bi_bdev;	\
 } while (0)
 
-static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
+static inline void *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
 {
 	return page_address(bvec->bv_page) + bvec->bv_offset;
 }
@@ -294,14 +294,14 @@ static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
 	*flags = 0;
 }
 
-static inline char *bvec_kmap_local(struct bio_vec *bvec)
+static inline void *bvec_kmap_local(struct bio_vec *bvec)
 {
 	return page_address(bvec->bv_page) + bvec->bv_offset;
 }
 
 static inline void bvec_kunmap_local(char *buffer) {}
 
-static inline char *__bio_kmap_irq(struct bio *bio, struct bvec_iter iter,
+static inline void *__bio_kmap_irq(struct bio *bio, struct bvec_iter iter,
 				   unsigned long *flags)
 {
 	return bvec_kmap_irq(&bio_iter_iovec(bio, iter), flags);

include/linux/types.h

@@ -53,6 +53,12 @@ typedef __s16 s16;
 typedef __u8  u8;
 typedef __s8  s8;
 
+typedef unsigned char		unchar;
+typedef unsigned short		ushort;
+typedef unsigned int		uint;
+typedef unsigned long		ulong;
+typedef unsigned long long	ullong;
+
 #ifdef __CHECKER__
 #define __bitwise__ __attribute__((bitwise))
 #else

libbcachefs/backpointers.c

@@ -144,7 +144,8 @@ static noinline int backpointer_mod_err(struct btree_trans *trans,
 	if (!will_check && __bch2_inconsistent_error(c, &buf))
 		ret = bch_err_throw(c, erofs_unfixed_errors);
 
-	bch_err(c, "%s", buf.buf);
+	if (buf.buf)
+		bch_err(c, "%s", buf.buf);
 	printbuf_exit(&buf);
 	return ret;
 }

libbcachefs/btree_cache.c

@@ -86,7 +86,7 @@ void bch2_btree_node_to_freelist(struct bch_fs *c, struct btree *b)
 	six_unlock_intent(&b->c.lock);
 }
 
-static void __btree_node_data_free(struct btree_cache *bc, struct btree *b)
+void __btree_node_data_free(struct btree *b)
 {
 	BUG_ON(!list_empty(&b->list));
 	BUG_ON(btree_node_hashed(b));
@@ -113,16 +113,17 @@ static void __btree_node_data_free(struct btree_cache *bc, struct btree *b)
 	munmap(b->aux_data, btree_aux_data_bytes(b));
 #endif
 	b->aux_data = NULL;
-
-	btree_node_to_freedlist(bc, b);
 }
 
 static void btree_node_data_free(struct btree_cache *bc, struct btree *b)
 {
 	BUG_ON(list_empty(&b->list));
 	list_del_init(&b->list);
+
+	__btree_node_data_free(b);
+
 	--bc->nr_freeable;
-	__btree_node_data_free(bc, b);
+	btree_node_to_freedlist(bc, b);
 }
 
 static int bch2_btree_cache_cmp_fn(struct rhashtable_compare_arg *arg,
@@ -186,10 +187,7 @@ static struct btree *__btree_node_mem_alloc(struct bch_fs *c, gfp_t gfp)
 
 struct btree *__bch2_btree_node_mem_alloc(struct bch_fs *c)
 {
-	struct btree_cache *bc = &c->btree_cache;
-	struct btree *b;
-
-	b = __btree_node_mem_alloc(c, GFP_KERNEL);
+	struct btree *b = __btree_node_mem_alloc(c, GFP_KERNEL);
 	if (!b)
 		return NULL;
 
@@ -199,8 +197,6 @@ struct btree *__bch2_btree_node_mem_alloc(struct bch_fs *c)
 	}
 
 	bch2_btree_lock_init(&b->c, 0, GFP_KERNEL);
-
-	__bch2_btree_node_to_freelist(bc, b);
 	return b;
 }
 
@@ -526,7 +522,8 @@ restart:
 			--touched;;
 		} else if (!btree_node_reclaim(c, b)) {
 			__bch2_btree_node_hash_remove(bc, b);
-			__btree_node_data_free(bc, b);
+			__btree_node_data_free(b);
+			btree_node_to_freedlist(bc, b);
 
 			freed++;
 			bc->nr_freed++;
@@ -667,9 +664,12 @@ int bch2_fs_btree_cache_init(struct bch_fs *c)
 
 	bch2_recalc_btree_reserve(c);
 
-	for (i = 0; i < bc->nr_reserve; i++)
-		if (!__bch2_btree_node_mem_alloc(c))
+	for (i = 0; i < bc->nr_reserve; i++) {
+		struct btree *b = __bch2_btree_node_mem_alloc(c);
+		if (!b)
 			goto err;
+		__bch2_btree_node_to_freelist(bc, b);
+	}
 
 	list_splice_init(&bc->live[0].list, &bc->freeable);
 

libbcachefs/btree_cache.h

@@ -30,6 +30,7 @@ void bch2_btree_node_update_key_early(struct btree_trans *, enum btree_id, unsig
 void bch2_btree_cache_cannibalize_unlock(struct btree_trans *);
 int bch2_btree_cache_cannibalize_lock(struct btree_trans *, struct closure *);
 
+void __btree_node_data_free(struct btree *);
 struct btree *__bch2_btree_node_mem_alloc(struct bch_fs *);
 struct btree *bch2_btree_node_mem_alloc(struct btree_trans *, bool);
 

libbcachefs/btree_io.c

@@ -24,8 +24,15 @@
 #include "super-io.h"
 #include "trace.h"
 
+#include <linux/moduleparam.h>
 #include <linux/sched/mm.h>
 
+#ifdef CONFIG_BCACHEFS_DEBUG
+static unsigned bch2_btree_read_corrupt_ratio;
+module_param_named(btree_read_corrupt_ratio, bch2_btree_read_corrupt_ratio, uint, 0644);
+MODULE_PARM_DESC(btree_read_corrupt_ratio, "");
+#endif
+
 static void bch2_btree_node_header_to_text(struct printbuf *out, struct btree_node *bn)
 {
 	bch2_btree_id_level_to_text(out, BTREE_NODE_ID(bn), BTREE_NODE_LEVEL(bn));
@@ -568,9 +575,9 @@ static int __btree_err(int ret,
 		bch2_mark_btree_validate_failure(failed, ca->dev_idx);
 
 		struct extent_ptr_decoded pick;
-		have_retry = !bch2_bkey_pick_read_device(c,
+		have_retry = bch2_bkey_pick_read_device(c,
 					bkey_i_to_s_c(&b->key),
-					failed, &pick, -1);
+					failed, &pick, -1) == 1;
 	}
 
 	if (!have_retry && ret == -BCH_ERR_btree_node_read_err_want_retry)
@@ -615,7 +622,6 @@ static int __btree_err(int ret,
 			goto out;
 		case -BCH_ERR_btree_node_read_err_bad_node:
 			prt_str(&out, ", ");
-			ret = __bch2_topology_error(c, &out);
 			break;
 		}
 
@@ -644,7 +650,6 @@ static int __btree_err(int ret,
 		goto out;
 	case -BCH_ERR_btree_node_read_err_bad_node:
 		prt_str(&out, ", ");
-		ret = __bch2_topology_error(c, &out);
 		break;
 	}
 print:
@@ -1408,7 +1413,7 @@ 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) {
+		if (ret <= 0) {
 			set_btree_node_read_error(b);
 			break;
 		}
@@ -1439,6 +1444,11 @@ start:
 			continue;
 		}
 
+		memset(&bio->bi_iter, 0, sizeof(bio->bi_iter));
+		bio->bi_iter.bi_size	= btree_buf_bytes(b);
+
+		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)

libbcachefs/btree_journal_iter.c

@@ -717,18 +717,6 @@ static void __journal_keys_sort(struct journal_keys *keys)
 	keys->nr = dst - keys->data;
 }
 
-static bool should_rewind_entry(struct bch_fs *c, struct jset_entry *entry)
-{
-	if (entry->level)
-		return false;
-	if (btree_id_is_alloc(entry->btree_id))
-		return false;
-	if (c->opts.journal_rewind_no_extents &&
-	    entry->btree_id == BTREE_ID_extents)
-		return false;
-	return true;
-}
-
 int bch2_journal_keys_sort(struct bch_fs *c)
 {
 	struct genradix_iter iter;
@@ -747,8 +735,9 @@ int bch2_journal_keys_sort(struct bch_fs *c)
 		cond_resched();
 
 		vstruct_for_each(&i->j, entry) {
-			bool rewind = le64_to_cpu(i->j.seq) >= rewind_seq &&
-				should_rewind_entry(c, entry);
+			bool rewind = !entry->level &&
+				!btree_id_is_alloc(entry->btree_id) &&
+				le64_to_cpu(i->j.seq) >= rewind_seq;
 
 			if (entry->type != (rewind
 					    ? BCH_JSET_ENTRY_overwrite

libbcachefs/btree_node_scan.c

@@ -75,39 +75,6 @@ static inline u64 bkey_journal_seq(struct bkey_s_c k)
 	}
 }
 
-static bool found_btree_node_is_readable(struct btree_trans *trans,
-					 struct found_btree_node *f)
-{
-	struct { __BKEY_PADDED(k, BKEY_BTREE_PTR_VAL_U64s_MAX); } tmp;
-
-	found_btree_node_to_key(&tmp.k, f);
-
-	struct btree *b = bch2_btree_node_get_noiter(trans, &tmp.k, f->btree_id, f->level, false);
-	bool ret = !IS_ERR_OR_NULL(b);
-	if (!ret)
-		return ret;
-
-	f->sectors_written = b->written;
-	f->journal_seq = le64_to_cpu(b->data->keys.journal_seq);
-
-	struct bkey_s_c k;
-	struct bkey unpacked;
-	struct btree_node_iter iter;
-	for_each_btree_node_key_unpack(b, k, &iter, &unpacked)
-		f->journal_seq = max(f->journal_seq, bkey_journal_seq(k));
-
-	six_unlock_read(&b->c.lock);
-
-	/*
-	 * We might update this node's range; if that happens, we need the node
-	 * to be re-read so the read path can trim keys that are no longer in
-	 * this node
-	 */
-	if (b != btree_node_root(trans->c, b))
-		bch2_btree_node_evict(trans, &tmp.k);
-	return ret;
-}
-
 static int found_btree_node_cmp_cookie(const void *_l, const void *_r)
 {
 	const struct found_btree_node *l = _l;
@@ -159,17 +126,17 @@ static const struct min_heap_callbacks found_btree_node_heap_cbs = {
 };
 
 static void try_read_btree_node(struct find_btree_nodes *f, struct bch_dev *ca,
-				struct bio *bio, struct btree_node *bn, u64 offset)
+				struct btree *b, struct bio *bio, u64 offset)
 {
 	struct bch_fs *c = container_of(f, struct bch_fs, found_btree_nodes);
+	struct btree_node *bn = b->data;
 
 	bio_reset(bio, ca->disk_sb.bdev, REQ_OP_READ);
 	bio->bi_iter.bi_sector	= offset;
-	bch2_bio_map(bio, bn, PAGE_SIZE);
+	bch2_bio_map(bio, b->data, c->opts.block_size);
 
 	u64 submit_time = local_clock();
 	submit_bio_wait(bio);
-
 	bch2_account_io_completion(ca, BCH_MEMBER_ERROR_read, submit_time, !bio->bi_status);
 
 	if (bio->bi_status) {
@@ -201,6 +168,14 @@ static void try_read_btree_node(struct find_btree_nodes *f, struct bch_dev *ca,
 	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),
@@ -217,7 +192,20 @@ static void try_read_btree_node(struct find_btree_nodes *f, struct bch_dev *ca,
 	};
 	rcu_read_unlock();
 
-	if (bch2_trans_run(c, found_btree_node_is_readable(trans, &n))) {
+	found_btree_node_to_key(&b->key, &n);
+
+	CLASS(printbuf, buf)();
+	if (!bch2_btree_node_read_done(c, ca, b, NULL, &buf)) {
+		/* read_done will swap out b->data for another buffer */
+		bn = b->data;
+		/*
+		 * Grab journal_seq here because we want the max journal_seq of
+		 * any bset; read_done sorts down to a single set and picks the
+		 * max journal_seq
+		 */
+		n.journal_seq		= le64_to_cpu(bn->keys.journal_seq),
+		n.sectors_written	= b->written;
+
 		mutex_lock(&f->lock);
 		if (BSET_BIG_ENDIAN(&bn->keys) != CPU_BIG_ENDIAN) {
 			bch_err(c, "try_read_btree_node() can't handle endian conversion");
@@ -237,12 +225,20 @@ static int read_btree_nodes_worker(void *p)
 	struct find_btree_nodes_worker *w = p;
 	struct bch_fs *c = container_of(w->f, struct bch_fs, found_btree_nodes);
 	struct bch_dev *ca = w->ca;
-	void *buf = (void *) __get_free_page(GFP_KERNEL);
-	struct bio *bio = bio_alloc(NULL, 1, 0, GFP_KERNEL);
 	unsigned long last_print = jiffies;
+	struct btree *b = NULL;
+	struct bio *bio = NULL;
 
-	if (!buf || !bio) {
-		bch_err(c, "read_btree_nodes_worker: error allocating bio/buf");
+	b = __bch2_btree_node_mem_alloc(c);
+	if (!b) {
+		bch_err(c, "read_btree_nodes_worker: error allocating buf");
+		w->f->ret = -ENOMEM;
+		goto err;
+	}
+
+	bio = bio_alloc(NULL, buf_pages(b->data, c->opts.btree_node_size), 0, GFP_KERNEL);
+	if (!bio) {
+		bch_err(c, "read_btree_nodes_worker: error allocating bio");
 		w->f->ret = -ENOMEM;
 		goto err;
 	}
@@ -266,11 +262,13 @@ static int read_btree_nodes_worker(void *p)
 			    !bch2_dev_btree_bitmap_marked_sectors(ca, sector, btree_sectors(c)))
 				continue;
 
-			try_read_btree_node(w->f, ca, bio, buf, sector);
+			try_read_btree_node(w->f, ca, b, bio, sector);
 		}
 err:
+	if (b)
+		__btree_node_data_free(b);
+	kfree(b);
 	bio_put(bio);
-	free_page((unsigned long) buf);
 	enumerated_ref_put(&ca->io_ref[READ], BCH_DEV_READ_REF_btree_node_scan);
 	closure_put(w->cl);
 	kfree(w);

libbcachefs/btree_trans_commit.c

@@ -1008,7 +1008,7 @@ do_bch2_trans_commit_to_journal_replay(struct btree_trans *trans)
 	return 0;
 }
 
-int __bch2_trans_commit(struct btree_trans *trans, unsigned flags)
+int __bch2_trans_commit(struct btree_trans *trans, enum bch_trans_commit_flags flags)
 {
 	struct btree_insert_entry *errored_at = NULL;
 	struct bch_fs *c = trans->c;

libbcachefs/btree_update.c

@@ -661,21 +661,22 @@ int bch2_btree_insert_trans(struct btree_trans *trans, enum btree_id id,
  * @k:			key to insert
  * @disk_res:		must be non-NULL whenever inserting or potentially
  *			splitting data extents
- * @flags:		transaction commit flags
+ * @commit_flags:	transaction commit flags
  * @iter_flags:		btree iter update trigger flags
  *
  * Returns:		0 on success, error code on failure
  */
 int bch2_btree_insert(struct bch_fs *c, enum btree_id id, struct bkey_i *k,
-		      struct disk_reservation *disk_res, int flags,
+		      struct disk_reservation *disk_res,
+		      enum bch_trans_commit_flags commit_flags,
 		      enum btree_iter_update_trigger_flags iter_flags)
 {
-	return bch2_trans_commit_do(c, disk_res, NULL, flags,
+	return bch2_trans_commit_do(c, disk_res, NULL, commit_flags,
 			     bch2_btree_insert_trans(trans, id, k, iter_flags));
 }
 
-int bch2_btree_delete_at(struct btree_trans *trans,
-			 struct btree_iter *iter, unsigned update_flags)
+int bch2_btree_delete_at(struct btree_trans *trans, struct btree_iter *iter,
+			 enum btree_iter_update_trigger_flags flags)
 {
 	struct bkey_i *k = bch2_trans_kmalloc(trans, sizeof(*k));
 	int ret = PTR_ERR_OR_ZERO(k);
@@ -684,12 +685,12 @@ int bch2_btree_delete_at(struct btree_trans *trans,
 
 	bkey_init(&k->k);
 	k->k.p = iter->pos;
-	return bch2_trans_update(trans, iter, k, update_flags);
+	return bch2_trans_update(trans, iter, k, flags);
 }
 
 int bch2_btree_delete(struct btree_trans *trans,
 		      enum btree_id btree, struct bpos pos,
-		      unsigned update_flags)
+		      enum btree_iter_update_trigger_flags flags)
 {
 	struct btree_iter iter;
 	int ret;
@@ -698,7 +699,7 @@ int bch2_btree_delete(struct btree_trans *trans,
 			     BTREE_ITER_cached|
 			     BTREE_ITER_intent);
 	ret   = bch2_btree_iter_traverse(trans, &iter) ?:
-		bch2_btree_delete_at(trans, &iter, update_flags);
+		bch2_btree_delete_at(trans, &iter, flags);
 	bch2_trans_iter_exit(trans, &iter);
 
 	return ret;
@@ -706,7 +707,7 @@ int bch2_btree_delete(struct btree_trans *trans,
 
 int bch2_btree_delete_range_trans(struct btree_trans *trans, enum btree_id id,
 				  struct bpos start, struct bpos end,
-				  unsigned update_flags,
+				  enum btree_iter_update_trigger_flags flags,
 				  u64 *journal_seq)
 {
 	u32 restart_count = trans->restart_count;
@@ -714,7 +715,7 @@ int bch2_btree_delete_range_trans(struct btree_trans *trans, enum btree_id id,
 	struct bkey_s_c k;
 	int ret = 0;
 
-	bch2_trans_iter_init(trans, &iter, id, start, BTREE_ITER_intent);
+	bch2_trans_iter_init(trans, &iter, id, start, BTREE_ITER_intent|flags);
 	while ((k = bch2_btree_iter_peek_max(trans, &iter, end)).k) {
 		struct disk_reservation disk_res =
 			bch2_disk_reservation_init(trans->c, 0);
@@ -747,7 +748,7 @@ int bch2_btree_delete_range_trans(struct btree_trans *trans, enum btree_id id,
 					bpos_min(end, k.k->p).offset -
 					iter.pos.offset);
 
-		ret   = bch2_trans_update(trans, &iter, &delete, update_flags) ?:
+		ret   = bch2_trans_update(trans, &iter, &delete, flags) ?:
 			bch2_trans_commit(trans, &disk_res, journal_seq,
 					  BCH_TRANS_COMMIT_no_enospc);
 		bch2_disk_reservation_put(trans->c, &disk_res);
@@ -777,12 +778,12 @@ err:
  */
 int bch2_btree_delete_range(struct bch_fs *c, enum btree_id id,
 			    struct bpos start, struct bpos end,
-			    unsigned update_flags,
+			    enum btree_iter_update_trigger_flags flags,
 			    u64 *journal_seq)
 {
 	int ret = bch2_trans_run(c,
 			bch2_btree_delete_range_trans(trans, id, start, end,
-						      update_flags, journal_seq));
+						      flags, journal_seq));
 	if (ret == -BCH_ERR_transaction_restart_nested)
 		ret = 0;
 	return ret;

libbcachefs/btree_update.h

@@ -47,22 +47,27 @@ enum bch_trans_commit_flags {
 
 void bch2_trans_commit_flags_to_text(struct printbuf *, enum bch_trans_commit_flags);
 
-int bch2_btree_delete_at(struct btree_trans *, struct btree_iter *, unsigned);
-int bch2_btree_delete(struct btree_trans *, enum btree_id, struct bpos, unsigned);
+int bch2_btree_delete_at(struct btree_trans *, struct btree_iter *,
+			 enum btree_iter_update_trigger_flags);
+int bch2_btree_delete(struct btree_trans *, enum btree_id, struct bpos,
+		      enum btree_iter_update_trigger_flags);
 
 int bch2_btree_insert_nonextent(struct btree_trans *, enum btree_id,
 				struct bkey_i *, enum btree_iter_update_trigger_flags);
 
 int bch2_btree_insert_trans(struct btree_trans *, enum btree_id, struct bkey_i *,
 			enum btree_iter_update_trigger_flags);
-int bch2_btree_insert(struct bch_fs *, enum btree_id, struct bkey_i *, struct
-		disk_reservation *, int flags, enum
-		btree_iter_update_trigger_flags iter_flags);
+int bch2_btree_insert(struct bch_fs *, enum btree_id, struct bkey_i *,
+		      struct disk_reservation *,
+		      enum bch_trans_commit_flags,
+		      enum btree_iter_update_trigger_flags);
 
 int bch2_btree_delete_range_trans(struct btree_trans *, enum btree_id,
-				  struct bpos, struct bpos, unsigned, u64 *);
+				  struct bpos, struct bpos,
+				  enum btree_iter_update_trigger_flags, u64 *);
 int bch2_btree_delete_range(struct bch_fs *, enum btree_id,
-			    struct bpos, struct bpos, unsigned, u64 *);
+			    struct bpos, struct bpos,
+			    enum btree_iter_update_trigger_flags, u64 *);
 
 int bch2_btree_bit_mod_iter(struct btree_trans *, struct btree_iter *, bool);
 int bch2_btree_bit_mod(struct btree_trans *, enum btree_id, struct bpos, bool);
@@ -226,7 +231,7 @@ static inline int __must_check bch2_trans_update_buffered(struct btree_trans *tr
 
 void bch2_trans_commit_hook(struct btree_trans *,
 			    struct btree_trans_commit_hook *);
-int __bch2_trans_commit(struct btree_trans *, unsigned);
+int __bch2_trans_commit(struct btree_trans *, enum bch_trans_commit_flags);
 
 int bch2_trans_log_str(struct btree_trans *, const char *);
 int bch2_trans_log_msg(struct btree_trans *, struct printbuf *);

libbcachefs/data_update.c

@@ -783,9 +783,6 @@ static int can_write_extent(struct bch_fs *c, struct data_update *m)
 	darray_for_each(m->op.devs_have, i)
 		__clear_bit(*i, devs.d);
 
-	CLASS(printbuf, buf)();
-	buf.atomic++;
-
 	guard(rcu)();
 
 	unsigned nr_replicas = 0, i;
@@ -797,11 +794,7 @@ static int can_write_extent(struct bch_fs *c, struct data_update *m)
 		struct bch_dev_usage usage;
 		bch2_dev_usage_read_fast(ca, &usage);
 
-		u64 nr_free = dev_buckets_free(ca, usage, m->op.watermark);
-
-		prt_printf(&buf, "%s=%llu ", ca->name, nr_free);
-
-		if (!nr_free)
+		if (!dev_buckets_free(ca, usage, m->op.watermark))
 			continue;
 
 		nr_replicas += ca->mi.durability;
@@ -809,10 +802,8 @@ static int can_write_extent(struct bch_fs *c, struct data_update *m)
 			break;
 	}
 
-	if (!nr_replicas) {
-		trace_data_update_done_no_rw_devs(c, buf.buf);
+	if (!nr_replicas)
 		return bch_err_throw(c, data_update_done_no_rw_devs);
-	}
 	if (nr_replicas < m->op.nr_replicas)
 		return bch_err_throw(c, insufficient_devices);
 	return 0;

libbcachefs/debug.c

@@ -153,8 +153,6 @@ void __bch2_btree_verify(struct bch_fs *c, struct btree *b)
 		c->verify_data = __bch2_btree_node_mem_alloc(c);
 		if (!c->verify_data)
 			goto out;
-
-		list_del_init(&c->verify_data->list);
 	}
 
 	BUG_ON(b->nsets != 1);
@@ -586,6 +584,8 @@ static ssize_t bch2_btree_transactions_read(struct file *file, char __user *buf,
 	i->ubuf = buf;
 	i->size	= size;
 	i->ret	= 0;
+
+	int srcu_idx = srcu_read_lock(&c->btree_trans_barrier);
 restart:
 	seqmutex_lock(&c->btree_trans_lock);
 	list_sort(&c->btree_trans_list, list_ptr_order_cmp);
@@ -599,6 +599,11 @@ restart:
 		if (!closure_get_not_zero(&trans->ref))
 			continue;
 
+		if (!trans->srcu_held) {
+			closure_put(&trans->ref);
+			continue;
+		}
+
 		u32 seq = seqmutex_unlock(&c->btree_trans_lock);
 
 		bch2_btree_trans_to_text(&i->buf, trans);
@@ -620,6 +625,8 @@ restart:
 	}
 	seqmutex_unlock(&c->btree_trans_lock);
 unlocked:
+	srcu_read_unlock(&c->btree_trans_barrier, srcu_idx);
+
 	if (i->buf.allocation_failure)
 		ret = -ENOMEM;
 

libbcachefs/errcode.h

@@ -289,7 +289,6 @@
 	x(EIO,				sb_not_downgraded)			\
 	x(EIO,				btree_node_write_all_failed)		\
 	x(EIO,				btree_node_read_error)			\
-	x(EIO,				btree_node_read_validate_error)		\
 	x(EIO,				btree_need_topology_repair)		\
 	x(EIO,				bucket_ref_update)			\
 	x(EIO,				trigger_alloc)				\

libbcachefs/error.c

@@ -103,7 +103,7 @@ int __bch2_topology_error(struct bch_fs *c, struct printbuf *out)
 		return bch_err_throw(c, btree_need_topology_repair);
 	} else {
 		return bch2_run_explicit_recovery_pass(c, out, BCH_RECOVERY_PASS_check_topology, 0) ?:
-			bch_err_throw(c, btree_node_read_validate_error);
+			bch_err_throw(c, btree_need_topology_repair);
 	}
 }
 

libbcachefs/extents.c

@@ -50,19 +50,17 @@ void bch2_io_failures_to_text(struct printbuf *out,
 			      struct bch_io_failures *failed)
 {
 	static const char * const error_types[] = {
-		"io", "checksum", "ec reconstruct", NULL
+		"btree validate", "io", "checksum", "ec reconstruct", NULL
 	};
 
 	for (struct bch_dev_io_failures *f = failed->devs;
 	     f < failed->devs + failed->nr;
 	     f++) {
 		unsigned errflags =
-			((!!f->failed_io)	<< 0) |
-			((!!f->failed_csum_nr)	<< 1) |
-			((!!f->failed_ec)	<< 2);
-
-		if (!errflags)
-			continue;
+			((!!f->failed_btree_validate)	<< 0) |
+			((!!f->failed_io)		<< 1) |
+			((!!f->failed_csum_nr)		<< 2) |
+			((!!f->failed_ec)		<< 3);
 
 		bch2_printbuf_make_room(out, 1024);
 		out->atomic++;
@@ -77,7 +75,9 @@ void bch2_io_failures_to_text(struct printbuf *out,
 
 		prt_char(out, ' ');
 
-		if (is_power_of_2(errflags)) {
+		if (!errflags) {
+			prt_str(out, "no error - confused");
+		} else if (is_power_of_2(errflags)) {
 			prt_bitflags(out, error_types, errflags);
 			prt_str(out, " error");
 		} else {

libbcachefs/fs-io-buffered.c

@@ -257,7 +257,7 @@ err:
 		struct printbuf buf = PRINTBUF;
 		lockrestart_do(trans,
 			bch2_inum_offset_err_msg_trans(trans, &buf, inum, iter.pos.offset << 9));
-		prt_printf(&buf, "read error %i from btree lookup", ret);
+		prt_printf(&buf, "read error %s from btree lookup", bch2_err_str(ret));
 		bch_err_ratelimited(c, "%s", buf.buf);
 		printbuf_exit(&buf);
 

libbcachefs/fsck.c

@@ -1638,7 +1638,8 @@ static int check_i_sectors_notnested(struct btree_trans *trans, struct inode_wal
 			i->count = count2;
 		}
 
-		if (fsck_err_on(!(i->inode.bi_flags & BCH_INODE_i_sectors_dirty),
+		if (fsck_err_on(!(i->inode.bi_flags & BCH_INODE_i_sectors_dirty) &&
+				i->inode.bi_sectors != i->count,
 				trans, inode_i_sectors_wrong,
 				"inode %llu:%u has incorrect i_sectors: got %llu, should be %llu",
 				w->last_pos.inode, i->inode.bi_snapshot,

libbcachefs/io_read.h

@@ -147,7 +147,7 @@ static inline void bch2_read_extent(struct btree_trans *trans,
 	int ret = __bch2_read_extent(trans, rbio, rbio->bio.bi_iter, read_pos,
 				     data_btree, k, offset_into_extent, NULL, flags, -1);
 	/* __bch2_read_extent only returns errors if BCH_READ_in_retry is set */
-	WARN(ret, "unhandled error from __bch2_read_extent()");
+	WARN(ret, "unhandled error from __bch2_read_extent(): %s", bch2_err_str(ret));
 }
 
 int __bch2_read(struct btree_trans *, struct bch_read_bio *, struct bvec_iter,

libbcachefs/io_write.c

@@ -32,6 +32,7 @@
 #include "trace.h"
 
 #include <linux/blkdev.h>
+#include <linux/moduleparam.h>
 #include <linux/prefetch.h>
 #include <linux/random.h>
 #include <linux/sched/mm.h>

libbcachefs/journal.c

@@ -1376,6 +1376,7 @@ int bch2_dev_journal_alloc(struct bch_dev *ca, bool new_fs)
 		return bch_err_throw(c, erofs_filesystem_full);
 	}
 
+	unsigned nr;
 	int ret;
 
 	if (dynamic_fault("bcachefs:add:journal_alloc")) {
@@ -1384,19 +1385,16 @@ int bch2_dev_journal_alloc(struct bch_dev *ca, bool new_fs)
 	}
 
 	/* 1/128th of the device by default: */
-	unsigned nr = ca->mi.nbuckets >> 7;
+	nr = ca->mi.nbuckets >> 7;
 
 	/*
-	 * clamp journal size to 8GB, or 32GB with large_journal option:
+	 * clamp journal size to 8192 buckets or 8GB (in sectors), whichever
+	 * is smaller:
 	 */
-	unsigned max_sectors = 1 << 24;
-
-	if (c->opts.large_journal)
-		max_sectors *= 4;
-
 	nr = clamp_t(unsigned, nr,
 		     BCH_JOURNAL_BUCKETS_MIN,
-		     max_sectors / ca->mi.bucket_size);
+		     min(1 << 13,
+			 (1 << 24) / ca->mi.bucket_size));
 
 	ret = bch2_set_nr_journal_buckets_loop(c, ca, nr, new_fs);
 err:

libbcachefs/journal_io.c

@@ -1245,6 +1245,8 @@ noinline_for_stack
 static void bch2_journal_print_checksum_error(struct bch_fs *c, struct journal_replay *j)
 {
 	struct printbuf buf = PRINTBUF;
+	bch2_log_msg_start(c, &buf);
+
 	enum bch_csum_type csum_type = JSET_CSUM_TYPE(&j->j);
 	bool have_good = false;
 

libbcachefs/opts.h

@@ -343,12 +343,6 @@ enum fsck_err_opts {
 	  OPT_UINT(0, U32_MAX),						\
 	  BCH_SB_JOURNAL_RECLAIM_DELAY,	100,				\
 	  NULL,		"Delay in milliseconds before automatic journal reclaim")\
-	x(large_journal,		bool,				\
-	  OPT_FS|OPT_MOUNT|OPT_FORMAT,					\
-	  OPT_BOOL(),							\
-	  BCH2_NO_SB_OPT,		false,				\
-	  NULL,		"Allocate a bigger than normal journal: recovery from unclean "\
-			"shutdown will be slower, but more info will be available for debugging")\
 	x(move_bytes_in_flight,		u32,				\
 	  OPT_HUMAN_READABLE|OPT_FS|OPT_MOUNT|OPT_RUNTIME,		\
 	  OPT_UINT(1024, U32_MAX),					\
@@ -395,11 +389,6 @@ enum fsck_err_opts {
 	  OPT_UINT(0, U64_MAX),						\
 	  BCH2_NO_SB_OPT,		0,				\
 	  NULL,		"Rewind journal")				\
-	x(journal_rewind_no_extents,	bool,				\
-	  OPT_FS|OPT_MOUNT,						\
-	  OPT_BOOL(),							\
-	  BCH2_NO_SB_OPT,		0,				\
-	  NULL,		"Don't rewind extents when rewinding journal")	\
 	x(recovery_passes,		u64,				\
 	  OPT_FS|OPT_MOUNT,						\
 	  OPT_BITFIELD(bch2_recovery_passes),				\

libbcachefs/recovery.c

@@ -273,24 +273,35 @@ static int bch2_journal_replay_key(struct btree_trans *trans,
 		goto out;
 
 	struct btree_path *path = btree_iter_path(trans, &iter);
-	if (unlikely(!btree_path_node(path, k->level) &&
-		     !k->allocated)) {
+	if (unlikely(!btree_path_node(path, k->level))) {
 		struct bch_fs *c = trans->c;
 
+		CLASS(printbuf, buf)();
+		prt_str(&buf, "btree=");
+		bch2_btree_id_to_text(&buf, k->btree_id);
+		prt_printf(&buf, " level=%u ", k->level);
+		bch2_bkey_val_to_text(&buf, c, bkey_i_to_s_c(k->k));
+
 		if (!(c->recovery.passes_complete & (BIT_ULL(BCH_RECOVERY_PASS_scan_for_btree_nodes)|
 						     BIT_ULL(BCH_RECOVERY_PASS_check_topology)))) {
-			bch_err(c, "have key in journal replay for btree depth that does not exist, confused");
+			bch_err(c, "have key in journal replay for btree depth that does not exist, confused\n%s",
+				buf.buf);
 			ret = -EINVAL;
 		}
-#if 0
+
+		if (!k->allocated) {
+			bch_notice(c, "dropping key in journal replay for depth that does not exist because we're recovering from scan\n%s",
+				   buf.buf);
+			k->overwritten = true;
+			goto out;
+		}
+
 		bch2_trans_iter_exit(trans, &iter);
 		bch2_trans_node_iter_init(trans, &iter, k->btree_id, k->k->k.p,
 					  BTREE_MAX_DEPTH, 0, iter_flags);
 		ret =   bch2_btree_iter_traverse(trans, &iter) ?:
 			bch2_btree_increase_depth(trans, iter.path, 0) ?:
 			-BCH_ERR_transaction_restart_nested;
-#endif
-		k->overwritten = true;
 		goto out;
 	}
 

libbcachefs/sb-members.c

@@ -20,6 +20,7 @@ int bch2_dev_missing_bkey(struct bch_fs *c, struct bkey_s_c k, unsigned dev)
 	prt_printf(&buf, "pointer to %s device %u in key\n",
 		   removed ? "removed" : "nonexistent", dev);
 	bch2_bkey_val_to_text(&buf, c, k);
+	prt_newline(&buf);
 
 	bool print = removed
 		? bch2_count_fsck_err(c, ptr_to_removed_device, &buf)

libbcachefs/super.c

@@ -1974,11 +1974,15 @@ int bch2_dev_add(struct bch_fs *c, const char *path)
 	ca->disk_sb.sb->dev_idx	= dev_idx;
 	bch2_dev_attach(c, ca, dev_idx);
 
+	set_bit(ca->dev_idx, c->online_devs.d);
+
 	if (BCH_MEMBER_GROUP(&dev_mi)) {
 		ret = __bch2_dev_group_set(c, ca, label.buf);
 		bch_err_msg(c, ret, "creating new label");
-		if (ret)
-			goto err_unlock;
+		if (ret) {
+			mutex_unlock(&c->sb_lock);
+			goto err_late;
+		}
 	}
 
 	bch2_write_super(c);
@@ -2526,6 +2530,8 @@ static int bch2_param_get_static_key_t(char *buffer, const struct kernel_param *
 	return sprintf(buffer, "%c\n", static_key_enabled(key) ? 'N' : 'Y');
 }
 
+/* this is unused in userspace - silence the warning */
+__maybe_unused
 static const struct kernel_param_ops bch2_param_ops_static_key_t = {
 	.flags = KERNEL_PARAM_OPS_FL_NOARG,
 	.set = bch2_param_set_static_key_t,

libbcachefs/trace.h

@@ -1330,11 +1330,6 @@ DEFINE_EVENT(fs_str, data_update,
 	TP_ARGS(c, str)
 );
 
-DEFINE_EVENT(fs_str, data_update_done_no_rw_devs,
-	TP_PROTO(struct bch_fs *c, const char *str),
-	TP_ARGS(c, str)
-);
-
 DEFINE_EVENT(fs_str, io_move_pred,
 	TP_PROTO(struct bch_fs *c, const char *str),
 	TP_ARGS(c, str)