Merge tag 'bcachefs-2024-03-13' of https://evilpiepirate.org/git/bcachefs

Pull bcachefs updates from Kent Overstreet:

 - Subvolume children btree; this is needed for providing a userspace
   interface for walking subvolumes, which will come later

 - Lots of improvements to directory structure checking

 - Improved journal pipelining, significantly improving performance on
   high iodepth write workloads

 - Discard path improvements: the discard path is more efficient, and no
   longer flushes the journal unnecessarily

 - Buffered write path can now avoid taking the inode lock

 - new mm helper: memalloc_flags_{save|restore}

 - mempool now does kvmalloc mempools

* tag 'bcachefs-2024-03-13' of https://evilpiepirate.org/git/bcachefs: (128 commits)
  bcachefs: time_stats: shrink time_stat_buffer for better alignment
  bcachefs: time_stats: split stats-with-quantiles into a separate structure
  bcachefs: mean_and_variance: put struct mean_and_variance_weighted on a diet
  bcachefs: time_stats: add larger units
  bcachefs: pull out time_stats.[ch]
  bcachefs: reconstruct_alloc cleanup
  bcachefs: fix bch_folio_sector padding
  bcachefs: Fix btree key cache coherency during replay
  bcachefs: Always flush write buffer in delete_dead_inodes()
  bcachefs: Fix order of gc_done passes
  bcachefs: fix deletion of indirect extents in btree_gc
  bcachefs: Prefer struct_size over open coded arithmetic
  bcachefs: Kill unused flags argument to btree_split()
  bcachefs: Check for writing superblocks with nonsense member seq fields
  bcachefs: fix bch2_journal_buf_to_text()
  lib/generic-radix-tree.c: Make nodes more reasonably sized
  bcachefs: copy_(to|from)_user_errcode()
  bcachefs: Split out bkey_types.h
  bcachefs: fix lost journal buf wakeup due to improved pipelining
  bcachefs: intercept mountoption value for bool type
  ...

Author: torvalds

Documentation/filesystems/bcachefs/errorcodes.rst

@@ -0,0 +1,30 @@
+.. SPDX-License-Identifier: GPL-2.0
+
+bcachefs private error codes
+----------------------------
+
+In bcachefs, as a hard rule we do not throw or directly use standard error
+codes (-EINVAL, -EBUSY, etc.). Instead, we define private error codes as needed
+in fs/bcachefs/errcode.h.
+
+This gives us much better error messages and makes debugging much easier. Any
+direct uses of standard error codes you see in the source code are simply old
+code that has yet to be converted - feel free to clean it up!
+
+Private error codes may subtype another error code, this allows for grouping of
+related errors that should be handled similarly (e.g. transaction restart
+errors), as well as specifying which standard error code should be returned at
+the bcachefs module boundary.
+
+At the module boundary, we use bch2_err_class() to convert to a standard error
+code; this also emits a trace event so that the original error code be
+recovered even if it wasn't logged.
+
+Do not reuse error codes! Generally speaking, a private error code should only
+be thrown in one place. That means that when we see it in a log message we can
+see, unambiguously, exactly which file and line number it was returned from.
+
+Try to give error codes names that are as reasonably descriptive of the error
+as possible. Frequently, the error will be logged at a place far removed from
+where the error was generated; good names for error codes mean much more
+descriptive and useful error messages.

MAINTAINERS

@@ -3555,6 +3555,7 @@ R:	Brian Foster <[email protected]>
 L:	[email protected]
 S:	Supported
 C:	irc://irc.oftc.net/bcache
+T:	git https://evilpiepirate.org/git/bcachefs.git
 F:	fs/bcachefs/
 
 BDISP ST MEDIA DRIVER

fs/bcachefs/Makefile

@@ -82,6 +82,7 @@ bcachefs-y		:=	\
 	super-io.o		\
 	sysfs.o			\
 	tests.o			\
+	time_stats.o		\
 	thread_with_file.o	\
 	trace.o			\
 	two_state_shared_lock.o	\
@@ -90,3 +91,6 @@ bcachefs-y		:=	\
 	xattr.o
 
 obj-$(CONFIG_MEAN_AND_VARIANCE_UNIT_TEST)   += mean_and_variance_test.o
+
+# Silence "note: xyz changed in GCC X.X" messages
+subdir-ccflags-y += $(call cc-disable-warning, psabi)

fs/bcachefs/alloc_background.c

@@ -29,6 +29,8 @@
 #include <linux/sched/task.h>
 #include <linux/sort.h>
 
+static void bch2_discard_one_bucket_fast(struct bch_fs *c, struct bpos bucket);
+
 /* Persistent alloc info: */
 
 static const unsigned BCH_ALLOC_V1_FIELD_BYTES[] = {
@@ -860,23 +862,28 @@ int bch2_trigger_alloc(struct btree_trans *trans,
 			*bucket_gen(ca, new.k->p.offset) = new_a->gen;
 
 		bch2_dev_usage_update(c, ca, old_a, new_a, journal_seq, false);
+		percpu_up_read(&c->mark_lock);
+
+#define eval_state(_a, expr)		({ const struct bch_alloc_v4 *a = _a; expr; })
+#define statechange(expr)		!eval_state(old_a, expr) && eval_state(new_a, expr)
+#define bucket_flushed(a)		(!a->journal_seq || a->journal_seq <= c->journal.flushed_seq_ondisk)
 
-		if (new_a->data_type == BCH_DATA_free &&
-		    (!new_a->journal_seq || new_a->journal_seq < c->journal.flushed_seq_ondisk))
+		if (statechange(a->data_type == BCH_DATA_free) &&
+		    bucket_flushed(new_a))
 			closure_wake_up(&c->freelist_wait);
 
-		if (new_a->data_type == BCH_DATA_need_discard &&
-		    (!bucket_journal_seq || bucket_journal_seq < c->journal.flushed_seq_ondisk))
-			bch2_do_discards(c);
+		if (statechange(a->data_type == BCH_DATA_need_discard) &&
+		    !bch2_bucket_is_open(c, new.k->p.inode, new.k->p.offset) &&
+		    bucket_flushed(new_a))
+			bch2_discard_one_bucket_fast(c, new.k->p);
 
-		if (old_a->data_type != BCH_DATA_cached &&
-		    new_a->data_type == BCH_DATA_cached &&
+		if (statechange(a->data_type == BCH_DATA_cached) &&
+		    !bch2_bucket_is_open(c, new.k->p.inode, new.k->p.offset) &&
 		    should_invalidate_buckets(ca, bch2_dev_usage_read(ca)))
 			bch2_do_invalidates(c);
 
-		if (new_a->data_type == BCH_DATA_need_gc_gens)
+		if (statechange(a->data_type == BCH_DATA_need_gc_gens))
 			bch2_do_gc_gens(c);
-		percpu_up_read(&c->mark_lock);
 	}
 
 	if ((flags & BTREE_TRIGGER_GC) &&
@@ -1045,14 +1052,13 @@ int bch2_check_alloc_key(struct btree_trans *trans,
 	if (ret)
 		goto err;
 
-	if (k.k->type != discard_key_type &&
-	    (c->opts.reconstruct_alloc ||
-	     fsck_err(c, need_discard_key_wrong,
-		      "incorrect key in need_discard btree (got %s should be %s)\n"
-		      "  %s",
-		      bch2_bkey_types[k.k->type],
-		      bch2_bkey_types[discard_key_type],
-		      (bch2_bkey_val_to_text(&buf, c, alloc_k), buf.buf)))) {
+	if (fsck_err_on(k.k->type != discard_key_type,
+			c, need_discard_key_wrong,
+			"incorrect key in need_discard btree (got %s should be %s)\n"
+			"  %s",
+			bch2_bkey_types[k.k->type],
+			bch2_bkey_types[discard_key_type],
+			(bch2_bkey_val_to_text(&buf, c, alloc_k), buf.buf))) {
 		struct bkey_i *update =
 			bch2_trans_kmalloc(trans, sizeof(*update));
 
@@ -1076,15 +1082,14 @@ int bch2_check_alloc_key(struct btree_trans *trans,
 	if (ret)
 		goto err;
 
-	if (k.k->type != freespace_key_type &&
-	    (c->opts.reconstruct_alloc ||
-	     fsck_err(c, freespace_key_wrong,
-		      "incorrect key in freespace btree (got %s should be %s)\n"
-		      "  %s",
-		      bch2_bkey_types[k.k->type],
-		      bch2_bkey_types[freespace_key_type],
-		      (printbuf_reset(&buf),
-		       bch2_bkey_val_to_text(&buf, c, alloc_k), buf.buf)))) {
+	if (fsck_err_on(k.k->type != freespace_key_type,
+			c, freespace_key_wrong,
+			"incorrect key in freespace btree (got %s should be %s)\n"
+			"  %s",
+			bch2_bkey_types[k.k->type],
+			bch2_bkey_types[freespace_key_type],
+			(printbuf_reset(&buf),
+			 bch2_bkey_val_to_text(&buf, c, alloc_k), buf.buf))) {
 		struct bkey_i *update =
 			bch2_trans_kmalloc(trans, sizeof(*update));
 
@@ -1108,14 +1113,13 @@ int bch2_check_alloc_key(struct btree_trans *trans,
 	if (ret)
 		goto err;
 
-	if (a->gen != alloc_gen(k, gens_offset) &&
-	    (c->opts.reconstruct_alloc ||
-	     fsck_err(c, bucket_gens_key_wrong,
-		      "incorrect gen in bucket_gens btree (got %u should be %u)\n"
-		      "  %s",
-		      alloc_gen(k, gens_offset), a->gen,
-		      (printbuf_reset(&buf),
-		       bch2_bkey_val_to_text(&buf, c, alloc_k), buf.buf)))) {
+	if (fsck_err_on(a->gen != alloc_gen(k, gens_offset),
+			c, bucket_gens_key_wrong,
+			"incorrect gen in bucket_gens btree (got %u should be %u)\n"
+			"  %s",
+			alloc_gen(k, gens_offset), a->gen,
+			(printbuf_reset(&buf),
+			 bch2_bkey_val_to_text(&buf, c, alloc_k), buf.buf))) {
 		struct bkey_i_bucket_gens *g =
 			bch2_trans_kmalloc(trans, sizeof(*g));
 
@@ -1167,14 +1171,13 @@ int bch2_check_alloc_hole_freespace(struct btree_trans *trans,
 
 	*end = bkey_min(k.k->p, *end);
 
-	if (k.k->type != KEY_TYPE_set &&
-	    (c->opts.reconstruct_alloc ||
-	     fsck_err(c, freespace_hole_missing,
-		      "hole in alloc btree missing in freespace btree\n"
-		      "  device %llu buckets %llu-%llu",
-		      freespace_iter->pos.inode,
-		      freespace_iter->pos.offset,
-		      end->offset))) {
+	if (fsck_err_on(k.k->type != KEY_TYPE_set,
+			c, freespace_hole_missing,
+			"hole in alloc btree missing in freespace btree\n"
+			"  device %llu buckets %llu-%llu",
+			freespace_iter->pos.inode,
+			freespace_iter->pos.offset,
+			end->offset)) {
 		struct bkey_i *update =
 			bch2_trans_kmalloc(trans, sizeof(*update));
 
@@ -1604,6 +1607,36 @@ int bch2_check_alloc_to_lru_refs(struct bch_fs *c)
 	return ret;
 }
 
+static int discard_in_flight_add(struct bch_fs *c, struct bpos bucket)
+{
+	int ret;
+
+	mutex_lock(&c->discard_buckets_in_flight_lock);
+	darray_for_each(c->discard_buckets_in_flight, i)
+		if (bkey_eq(*i, bucket)) {
+			ret = -EEXIST;
+			goto out;
+		}
+
+	ret = darray_push(&c->discard_buckets_in_flight, bucket);
+out:
+	mutex_unlock(&c->discard_buckets_in_flight_lock);
+	return ret;
+}
+
+static void discard_in_flight_remove(struct bch_fs *c, struct bpos bucket)
+{
+	mutex_lock(&c->discard_buckets_in_flight_lock);
+	darray_for_each(c->discard_buckets_in_flight, i)
+		if (bkey_eq(*i, bucket)) {
+			darray_remove_item(&c->discard_buckets_in_flight, i);
+			goto found;
+		}
+	BUG();
+found:
+	mutex_unlock(&c->discard_buckets_in_flight_lock);
+}
+
 struct discard_buckets_state {
 	u64		seen;
 	u64		open;
@@ -1642,6 +1675,7 @@ static int bch2_discard_one_bucket(struct btree_trans *trans,
 	struct bch_dev *ca;
 	struct bkey_i_alloc_v4 *a;
 	struct printbuf buf = PRINTBUF;
+	bool discard_locked = false;
 	int ret = 0;
 
 	ca = bch_dev_bkey_exists(c, pos.inode);
@@ -1709,6 +1743,11 @@ static int bch2_discard_one_bucket(struct btree_trans *trans,
 		goto out;
 	}
 
+	if (discard_in_flight_add(c, SPOS(iter.pos.inode, iter.pos.offset, true)))
+		goto out;
+
+	discard_locked = true;
+
 	if (!bkey_eq(*discard_pos_done, iter.pos) &&
 	    ca->mi.discard && !c->opts.nochanges) {
 		/*
@@ -1740,6 +1779,8 @@ static int bch2_discard_one_bucket(struct btree_trans *trans,
 	count_event(c, bucket_discard);
 	s->discarded++;
 out:
+	if (discard_locked)
+		discard_in_flight_remove(c, iter.pos);
 	s->seen++;
 	bch2_trans_iter_exit(trans, &iter);
 	percpu_ref_put(&ca->io_ref);
@@ -1779,6 +1820,93 @@ void bch2_do_discards(struct bch_fs *c)
 		bch2_write_ref_put(c, BCH_WRITE_REF_discard);
 }
 
+static int bch2_clear_bucket_needs_discard(struct btree_trans *trans, struct bpos bucket)
+{
+	struct btree_iter iter;
+	bch2_trans_iter_init(trans, &iter, BTREE_ID_alloc, bucket, BTREE_ITER_INTENT);
+	struct bkey_s_c k = bch2_btree_iter_peek_slot(&iter);
+	int ret = bkey_err(k);
+	if (ret)
+		goto err;
+
+	struct bkey_i_alloc_v4 *a = bch2_alloc_to_v4_mut(trans, k);
+	ret = PTR_ERR_OR_ZERO(a);
+	if (ret)
+		goto err;
+
+	SET_BCH_ALLOC_V4_NEED_DISCARD(&a->v, false);
+	a->v.data_type = alloc_data_type(a->v, a->v.data_type);
+
+	ret = bch2_trans_update(trans, &iter, &a->k_i, 0);
+err:
+	bch2_trans_iter_exit(trans, &iter);
+	return ret;
+}
+
+static void bch2_do_discards_fast_work(struct work_struct *work)
+{
+	struct bch_fs *c = container_of(work, struct bch_fs, discard_fast_work);
+
+	while (1) {
+		bool got_bucket = false;
+		struct bpos bucket;
+		struct bch_dev *ca;
+
+		mutex_lock(&c->discard_buckets_in_flight_lock);
+		darray_for_each(c->discard_buckets_in_flight, i) {
+			if (i->snapshot)
+				continue;
+
+			ca = bch_dev_bkey_exists(c, i->inode);
+
+			if (!percpu_ref_tryget(&ca->io_ref)) {
+				darray_remove_item(&c->discard_buckets_in_flight, i);
+				continue;
+			}
+
+			got_bucket = true;
+			bucket = *i;
+			i->snapshot = true;
+			break;
+		}
+		mutex_unlock(&c->discard_buckets_in_flight_lock);
+
+		if (!got_bucket)
+			break;
+
+		if (ca->mi.discard && !c->opts.nochanges)
+			blkdev_issue_discard(ca->disk_sb.bdev,
+					     bucket.offset * ca->mi.bucket_size,
+					     ca->mi.bucket_size,
+					     GFP_KERNEL);
+
+		int ret = bch2_trans_do(c, NULL, NULL,
+					BCH_WATERMARK_btree|
+					BCH_TRANS_COMMIT_no_enospc,
+					bch2_clear_bucket_needs_discard(trans, bucket));
+		bch_err_fn(c, ret);
+
+		percpu_ref_put(&ca->io_ref);
+		discard_in_flight_remove(c, bucket);
+
+		if (ret)
+			break;
+	}
+
+	bch2_write_ref_put(c, BCH_WRITE_REF_discard_fast);
+}
+
+static void bch2_discard_one_bucket_fast(struct bch_fs *c, struct bpos bucket)
+{
+	struct bch_dev *ca = bch_dev_bkey_exists(c, bucket.inode);
+
+	if (!percpu_ref_is_dying(&ca->io_ref) &&
+	    !discard_in_flight_add(c, bucket) &&
+	    bch2_write_ref_tryget(c, BCH_WRITE_REF_discard_fast) &&
+	    !queue_work(c->write_ref_wq, &c->discard_fast_work))
+		bch2_write_ref_put(c, BCH_WRITE_REF_discard_fast);
+}
+
 static int invalidate_one_bucket(struct btree_trans *trans,
 				 struct btree_iter *lru_iter,
 				 struct bkey_s_c lru_k,
@@ -2210,9 +2338,16 @@ void bch2_dev_allocator_add(struct bch_fs *c, struct bch_dev *ca)
 			set_bit(ca->dev_idx, c->rw_devs[i].d);
 }
 
+void bch2_fs_allocator_background_exit(struct bch_fs *c)
+{
+	darray_exit(&c->discard_buckets_in_flight);
+}
+
 void bch2_fs_allocator_background_init(struct bch_fs *c)
 {
 	spin_lock_init(&c->freelist_lock);
+	mutex_init(&c->discard_buckets_in_flight_lock);
 	INIT_WORK(&c->discard_work, bch2_do_discards_work);
+	INIT_WORK(&c->discard_fast_work, bch2_do_discards_fast_work);
 	INIT_WORK(&c->invalidate_work, bch2_do_invalidates_work);
 }

fs/bcachefs/alloc_background.h

@@ -269,6 +269,7 @@ u64 bch2_min_rw_member_capacity(struct bch_fs *);
 void bch2_dev_allocator_remove(struct bch_fs *, struct bch_dev *);
 void bch2_dev_allocator_add(struct bch_fs *, struct bch_dev *);
 
+void bch2_fs_allocator_background_exit(struct bch_fs *);
 void bch2_fs_allocator_background_init(struct bch_fs *);
 
 #endif /* _BCACHEFS_ALLOC_BACKGROUND_H */