bcachefs: Refactor filesystem usage accounting

Various filesystem usage counters are kept in percpu counters, with one
set per in flight journal buffer. Right now all the code that deals with
it assumes that there's only two buffers/sets of counters, but the
number of journal bufs is getting increased to 4 in the next patch - so
refactor that code to not assume a constant.

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

fs/bcachefs/bcachefs.h

@@ -676,7 +676,7 @@ struct bch_fs {
 
 	seqcount_t			usage_lock;
 	struct bch_fs_usage		*usage_base;
-	struct bch_fs_usage __percpu	*usage[2];
+	struct bch_fs_usage __percpu	*usage[JOURNAL_BUF_NR];
 	struct bch_fs_usage __percpu	*usage_gc;
 	u64 __percpu		*online_reserved;
 

fs/bcachefs/btree_gc.c

@@ -603,7 +603,6 @@ static int bch2_gc_done(struct bch_fs *c,
 		struct genradix_iter dst_iter = genradix_iter_init(&c->stripes[0], 0);
 		struct genradix_iter src_iter = genradix_iter_init(&c->stripes[1], 0);
 		struct stripe *dst, *src;
-		unsigned i;
 
 		c->ec_stripes_heap.used = 0;
 

fs/bcachefs/buckets.c

@@ -207,13 +207,13 @@ static inline struct bch_fs_usage *fs_usage_ptr(struct bch_fs *c,
 {
 	return this_cpu_ptr(gc
 			    ? c->usage_gc
-			    : c->usage[journal_seq & 1]);
+			    : c->usage[journal_seq & JOURNAL_BUF_MASK]);
 }
 
 u64 bch2_fs_usage_read_one(struct bch_fs *c, u64 *v)
 {
 	ssize_t offset = v - (u64 *) c->usage_base;
-	unsigned seq;
+	unsigned i, seq;
 	u64 ret;
 
 	BUG_ON(offset < 0 || offset >= fs_usage_u64s(c));
@@ -221,9 +221,10 @@ u64 bch2_fs_usage_read_one(struct bch_fs *c, u64 *v)
 
 	do {
 		seq = read_seqcount_begin(&c->usage_lock);
-		ret = *v +
-			percpu_u64_get((u64 __percpu *) c->usage[0] + offset) +
-			percpu_u64_get((u64 __percpu *) c->usage[1] + offset);
+		ret = *v;
+
+		for (i = 0; i < ARRAY_SIZE(c->usage); i++)
+			ret += percpu_u64_get((u64 __percpu *) c->usage[i] + offset);
 	} while (read_seqcount_retry(&c->usage_lock, seq));
 
 	return ret;
@@ -232,15 +233,20 @@ u64 bch2_fs_usage_read_one(struct bch_fs *c, u64 *v)
 struct bch_fs_usage_online *bch2_fs_usage_read(struct bch_fs *c)
 {
 	struct bch_fs_usage_online *ret;
-	unsigned seq, i, u64s;
+	unsigned seq, i, v, u64s = fs_usage_u64s(c);
+retry:
+	ret = kmalloc(u64s * sizeof(u64), GFP_NOFS);
+	if (unlikely(!ret))
+		return NULL;
 
 	percpu_down_read(&c->mark_lock);
 
-	ret = kmalloc(sizeof(struct bch_fs_usage_online) +
-		      sizeof(u64) + c->replicas.nr, GFP_NOFS);
-	if (unlikely(!ret)) {
+	v = fs_usage_u64s(c);
+	if (unlikely(u64s != v)) {
+		u64s = v;
 		percpu_up_read(&c->mark_lock);
-		return NULL;
+		kfree(ret);
+		goto retry;
 	}
 
 	ret->online_reserved = percpu_u64_get(c->online_reserved);
@@ -248,7 +254,7 @@ struct bch_fs_usage_online *bch2_fs_usage_read(struct bch_fs *c)
 	u64s = fs_usage_u64s(c);
 	do {
 		seq = read_seqcount_begin(&c->usage_lock);
-		memcpy(&ret->u, c->usage_base, u64s * sizeof(u64));
+		memcpy(ret, c->usage_base, u64s * sizeof(u64));
 		for (i = 0; i < ARRAY_SIZE(c->usage); i++)
 			acc_u64s_percpu((u64 *) &ret->u, (u64 __percpu *) c->usage[i], u64s);
 	} while (read_seqcount_retry(&c->usage_lock, seq));

fs/bcachefs/journal_types.h

@@ -11,6 +11,10 @@
 
 struct journal_res;
 
+#define JOURNAL_BUF_BITS	1
+#define JOURNAL_BUF_NR		(1U << JOURNAL_BUF_BITS)
+#define JOURNAL_BUF_MASK	(JOURNAL_BUF_NR - 1)
+
 /*
  * We put two of these in struct journal; we used them for writes to the
  * journal that are being staged or in flight.

fs/bcachefs/replicas.c

@@ -275,7 +275,7 @@ static void __replicas_table_update_pcpu(struct bch_fs_usage __percpu *dst_p,
 static int replicas_table_update(struct bch_fs *c,
 				 struct bch_replicas_cpu *new_r)
 {
-	struct bch_fs_usage __percpu *new_usage[2];
+	struct bch_fs_usage __percpu *new_usage[JOURNAL_BUF_NR];
 	struct bch_fs_usage_online *new_scratch = NULL;
 	struct bch_fs_usage __percpu *new_gc = NULL;
 	struct bch_fs_usage *new_base = NULL;
@@ -283,7 +283,14 @@ static int replicas_table_update(struct bch_fs *c,
 		sizeof(u64) * new_r->nr;
 	unsigned scratch_bytes = sizeof(struct bch_fs_usage_online) +
 		sizeof(u64) * new_r->nr;
-	int ret = -ENOMEM;
+	int ret = 0;
+
+	memset(new_usage, 0, sizeof(new_usage));
+
+	for (i = 0; i < ARRAY_SIZE(new_usage); i++)
+		if (!(new_usage[i] = __alloc_percpu_gfp(bytes,
+					sizeof(u64), GFP_NOIO)))
+			goto err;
 
 	memset(new_usage, 0, sizeof(new_usage));
 
@@ -295,10 +302,8 @@ static int replicas_table_update(struct bch_fs *c,
 	if (!(new_base = kzalloc(bytes, GFP_NOIO)) ||
 	    !(new_scratch  = kmalloc(scratch_bytes, GFP_NOIO)) ||
 	    (c->usage_gc &&
-	     !(new_gc = __alloc_percpu_gfp(bytes, sizeof(u64), GFP_NOIO)))) {
-		bch_err(c, "error updating replicas table: memory allocation failure");
+	     !(new_gc = __alloc_percpu_gfp(bytes, sizeof(u64), GFP_NOIO))))
 		goto err;
-	}
 
 	for (i = 0; i < ARRAY_SIZE(new_usage); i++)
 		if (c->usage[i])
@@ -317,14 +322,17 @@ static int replicas_table_update(struct bch_fs *c,
 	swap(c->usage_scratch,	new_scratch);
 	swap(c->usage_gc,	new_gc);
 	swap(c->replicas,	*new_r);
-	ret = 0;
-err:
+out:
 	free_percpu(new_gc);
 	kfree(new_scratch);
 	free_percpu(new_usage[1]);
 	free_percpu(new_usage[0]);
 	kfree(new_base);
 	return ret;
+err:
+	bch_err(c, "error updating replicas table: memory allocation failure");
+	ret = -ENOMEM;
+	goto out;
 }
 
 static unsigned reserve_journal_replicas(struct bch_fs *c,
@@ -499,9 +507,7 @@ int bch2_replicas_gc_end(struct bch_fs *c, int ret)
 		struct bch_replicas_cpu n;
 
 		if (!__replicas_has_entry(&c->replicas_gc, e) &&
-		    (c->usage_base->replicas[i] ||
-		     percpu_u64_get(&c->usage[0]->replicas[i]) ||
-		     percpu_u64_get(&c->usage[1]->replicas[i]))) {
+		    bch2_fs_usage_read_one(c, &c->usage_base->replicas[i])) {
 			n = cpu_replicas_add_entry(&c->replicas_gc, e);
 			if (!n.entries) {
 				ret = -ENOSPC;
@@ -606,9 +612,7 @@ retry:
 			cpu_replicas_entry(&c->replicas, i);
 
 		if (e->data_type == BCH_DATA_journal ||
-		    c->usage_base->replicas[i] ||
-		    percpu_u64_get(&c->usage[0]->replicas[i]) ||
-		    percpu_u64_get(&c->usage[1]->replicas[i]))
+		    bch2_fs_usage_read_one(c, &c->usage_base->replicas[i]))
 			memcpy(cpu_replicas_entry(&new, new.nr++),
 			       e, new.entry_size);
 	}

fs/bcachefs/super-io.c

@@ -998,7 +998,7 @@ bch2_journal_super_entries_add_common(struct bch_fs *c,
 		for (i = 0; i < ARRAY_SIZE(c->usage); i++)
 			bch2_fs_usage_acc_to_base(c, i);
 	} else {
-		bch2_fs_usage_acc_to_base(c, journal_seq & 1);
+		bch2_fs_usage_acc_to_base(c, journal_seq & JOURNAL_BUF_MASK);
 	}
 
 	{

fs/bcachefs/super.c

@@ -483,8 +483,8 @@ static void __bch2_fs_free(struct bch_fs *c)
 	percpu_free_rwsem(&c->mark_lock);
 	free_percpu(c->online_reserved);
 	kfree(c->usage_scratch);
-	free_percpu(c->usage[1]);
-	free_percpu(c->usage[0]);
+	for (i = 0; i < ARRAY_SIZE(c->usage); i++)
+		free_percpu(c->usage[i]);
 	kfree(c->usage_base);
 
 	if (c->btree_iters_bufs)