bcachefs: Redo filesystem usage ioctls

When disk space accounting was changed to be tracked by replicas entry,
the ioctl interface was never update: this patch finally does that.

Aditionally, the BCH_IOCTL_USAGE ioctl is now broken out into separate
ioctls for filesystem and device usage.

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

4 files changed, 136 insertions, 92 deletions

diff --git a/fs/bcachefs/bcachefs_format.h b/fs/bcachefs/bcachefs_format.h
index e3004593874c..9b8fc265a5c0 100644
--- a/fs/bcachefs/bcachefs_format.h
+++ b/fs/bcachefs/bcachefs_format.h
@@ -1084,6 +1084,9 @@ struct bch_replicas_entry {
 	__u8			devs[];
 } __attribute__((packed));
 
+#define replicas_entry_bytes(_i)					\
+	(offsetof(typeof(*(_i)), devs) + (_i)->nr_devs)
+
 struct bch_sb_field_replicas {
 	struct bch_sb_field	field;
 	struct bch_replicas_entry entries[];
diff --git a/fs/bcachefs/bcachefs_ioctl.h b/fs/bcachefs/bcachefs_ioctl.h
index 8c0599618404..d7f25e52dc71 100644
--- a/fs/bcachefs/bcachefs_ioctl.h
+++ b/fs/bcachefs/bcachefs_ioctl.h
@@ -68,7 +68,8 @@ struct bch_ioctl_incremental {
 #define BCH_IOCTL_DISK_OFFLINE	_IOW(0xbc,	7,  struct bch_ioctl_disk)
 #define BCH_IOCTL_DISK_SET_STATE _IOW(0xbc,	8,  struct bch_ioctl_disk_set_state)
 #define BCH_IOCTL_DATA		_IOW(0xbc,	10, struct bch_ioctl_data)
-#define BCH_IOCTL_USAGE		_IOWR(0xbc,	11, struct bch_ioctl_usage)
+#define BCH_IOCTL_FS_USAGE	_IOWR(0xbc,	11, struct bch_ioctl_fs_usage)
+#define BCH_IOCTL_DEV_USAGE	_IOWR(0xbc,	11, struct bch_ioctl_dev_usage)
 #define BCH_IOCTL_READ_SUPER	_IOW(0xbc,	12, struct bch_ioctl_read_super)
 #define BCH_IOCTL_DISK_GET_IDX	_IOW(0xbc,	13,  struct bch_ioctl_disk_get_idx)
 #define BCH_IOCTL_DISK_RESIZE	_IOW(0xbc,	14,  struct bch_ioctl_disk_resize)
@@ -224,46 +225,59 @@ struct bch_ioctl_data_event {
 	};
 } __attribute__((packed, aligned(8)));
 
-struct bch_ioctl_dev_usage {
-	__u8			state;
-	__u8			alive;
-	__u8			pad[6];
-	__u32			dev;
+struct bch_replicas_usage {
+	__u64			sectors;
+	struct bch_replicas_entry r;
+} __attribute__((packed));
 
-	__u32			bucket_size;
-	__u64			nr_buckets;
-
-	__u64			buckets[BCH_DATA_NR];
-	__u64			sectors[BCH_DATA_NR];
-};
+static inline struct bch_replicas_usage *
+replicas_usage_next(struct bch_replicas_usage *u)
+{
+	return (void *) u + replicas_entry_bytes(&u->r) + 8;
+}
 
+/*
+ * BCH_IOCTL_FS_USAGE: query filesystem disk space usage
+ *
+ * Returns disk space usage broken out by data type, number of replicas, and
+ * by component device
+ *
+ * @replica_entries_bytes - size, in bytes, allocated for replica usage entries
+ *
+ * On success, @replica_entries_bytes will be changed to indicate the number of
+ * bytes actually used.
+ *
+ * Returns -ERANGE if @replica_entries_bytes was too small
+ */
 struct bch_ioctl_fs_usage {
 	__u64			capacity;
 	__u64			used;
 	__u64			online_reserved;
 	__u64			persistent_reserved[BCH_REPLICAS_MAX];
-	__u64			sectors[BCH_DATA_NR][BCH_REPLICAS_MAX];
+
+	__u32			replica_entries_bytes;
+	__u32			pad;
+
+	struct bch_replicas_usage replicas[0];
 };
 
 /*
- * BCH_IOCTL_USAGE: query filesystem disk space usage
- *
- * Returns disk space usage broken out by data type, number of replicas, and
- * by component device
+ * BCH_IOCTL_DEV_USAGE: query device disk space usage
  *
- * @nr_devices	- number of devices userspace allocated space for in @devs
- *
- * On success, @fs and @devs will be filled out appropriately and devs[i].alive
- * will indicate if a device was present in that slot
- *
- * Returns -ERANGE if @nr_devices was too small
+ * Returns disk space usage broken out by data type - both by buckets and
+ * sectors.
  */
-struct bch_ioctl_usage {
-	__u16			nr_devices;
-	__u16			pad[3];
+struct bch_ioctl_dev_usage {
+	__u64			dev;
+	__u32			flags;
+	__u8			state;
+	__u8			pad[7];
+
+	__u32			bucket_size;
+	__u64			nr_buckets;
 
-	struct bch_ioctl_fs_usage fs;
-	struct bch_ioctl_dev_usage devs[0];
+	__u64			buckets[BCH_DATA_NR];
+	__u64			sectors[BCH_DATA_NR];
 };
 
 /*
diff --git a/fs/bcachefs/chardev.c b/fs/bcachefs/chardev.c
index 4d8331022648..084bef5e7997 100644
--- a/fs/bcachefs/chardev.c
+++ b/fs/bcachefs/chardev.c
@@ -6,6 +6,7 @@
 #include "buckets.h"
 #include "chardev.h"
 #include "move.h"
+#include "replicas.h"
 #include "super.h"
 #include "super-io.h"
 
@@ -371,89 +372,116 @@ err:
 	return ret;
 }
 
-static long bch2_ioctl_usage(struct bch_fs *c,
-			     struct bch_ioctl_usage __user *user_arg)
+static long bch2_ioctl_fs_usage(struct bch_fs *c,
+				struct bch_ioctl_fs_usage __user *user_arg)
 {
-	struct bch_ioctl_usage arg;
-	struct bch_dev *ca;
-	unsigned i, j;
-	int ret;
+	struct bch_ioctl_fs_usage *arg = NULL;
+	struct bch_replicas_usage *dst_e, *dst_end;
+	struct bch_fs_usage_online *src;
+	u32 replica_entries_bytes;
+	unsigned i;
+	int ret = 0;
 
 	if (!test_bit(BCH_FS_STARTED, &c->flags))
 		return -EINVAL;
 
-	if (copy_from_user(&arg, user_arg, sizeof(arg)))
+	if (get_user(replica_entries_bytes, &user_arg->replica_entries_bytes))
 		return -EFAULT;
 
-	for (i = 0; i < arg.nr_devices; i++) {
-		struct bch_ioctl_dev_usage dst = { .alive = 0 };
+	arg = kzalloc(sizeof(*arg) + replica_entries_bytes, GFP_KERNEL);
+	if (!arg)
+		return -ENOMEM;
 
-		ret = copy_to_user(&user_arg->devs[i], &dst, sizeof(dst));
-		if (ret)
-			return ret;
+	src = bch2_fs_usage_read(c);
+	if (!src) {
+		ret = -ENOMEM;
+		goto err;
 	}
 
-	{
-		struct bch_fs_usage_online *src;
-		struct bch_ioctl_fs_usage dst = {
-			.capacity		= c->capacity,
-		};
+	arg->capacity		= c->capacity;
+	arg->used		= bch2_fs_sectors_used(c, src);
+	arg->online_reserved	= src->online_reserved;
 
-		src = bch2_fs_usage_read(c);
-		if (!src)
-			return -ENOMEM;
+	for (i = 0; i < BCH_REPLICAS_MAX; i++)
+		arg->persistent_reserved[i] = src->u.persistent_reserved[i];
 
-		dst.used		= bch2_fs_sectors_used(c, src);
-		dst.online_reserved	= src->online_reserved;
+	dst_e	= arg->replicas;
+	dst_end = (void *) arg->replicas + replica_entries_bytes;
 
-		percpu_up_read(&c->mark_lock);
+	for (i = 0; i < c->replicas.nr; i++) {
+		struct bch_replicas_entry *src_e =
+			cpu_replicas_entry(&c->replicas, i);
 
-		for (i = 0; i < BCH_REPLICAS_MAX; i++) {
-			dst.persistent_reserved[i] =
-				src->u.persistent_reserved[i];
-#if 0
-			for (j = 0; j < BCH_DATA_NR; j++)
-				dst.sectors[j][i] = src.replicas[i].data[j];
-#endif
+		if (replicas_usage_next(dst_e) > dst_end) {
+			ret = -ERANGE;
+			break;
 		}
 
-		kfree(src);
+		dst_e->sectors		= src->u.replicas[i];
+		dst_e->r		= *src_e;
+
+		/* recheck after setting nr_devs: */
+		if (replicas_usage_next(dst_e) > dst_end) {
+			ret = -ERANGE;
+			break;
+		}
 
-		ret = copy_to_user(&user_arg->fs, &dst, sizeof(dst));
-		if (ret)
-			return ret;
+		memcpy(dst_e->r.devs, src_e->devs, src_e->nr_devs);
+
+		dst_e = replicas_usage_next(dst_e);
 	}
 
-	for_each_member_device(ca, c, i) {
-		struct bch_dev_usage src = bch2_dev_usage_read(c, ca);
-		struct bch_ioctl_dev_usage dst = {
-			.alive		= 1,
-			.state		= ca->mi.state,
-			.bucket_size	= ca->mi.bucket_size,
-			.nr_buckets	= ca->mi.nbuckets - ca->mi.first_bucket,
-		};
-
-		if (ca->dev_idx >= arg.nr_devices) {
-			percpu_ref_put(&ca->ref);
-			return -ERANGE;
-		}
+	arg->replica_entries_bytes = (void *) dst_e - (void *) arg->replicas;
 
-		if (percpu_ref_tryget(&ca->io_ref)) {
-			dst.dev = huge_encode_dev(ca->disk_sb.bdev->bd_dev);
-			percpu_ref_put(&ca->io_ref);
-		}
+	percpu_up_read(&c->mark_lock);
+	kfree(src);
 
-		for (j = 0; j < BCH_DATA_NR; j++) {
-			dst.buckets[j] = src.buckets[j];
-			dst.sectors[j] = src.sectors[j];
-		}
+	if (!ret)
+		ret = copy_to_user(user_arg, arg,
+			sizeof(*arg) + arg->replica_entries_bytes);
+err:
+	kfree(arg);
+	return ret;
+}
+
+static long bch2_ioctl_dev_usage(struct bch_fs *c,
+				 struct bch_ioctl_dev_usage __user *user_arg)
+{
+	struct bch_ioctl_dev_usage arg;
+	struct bch_dev_usage src;
+	struct bch_dev *ca;
+	unsigned i;
+
+	if (!test_bit(BCH_FS_STARTED, &c->flags))
+		return -EINVAL;
 
-		ret = copy_to_user(&user_arg->devs[i], &dst, sizeof(dst));
-		if (ret)
-			return ret;
+	if (copy_from_user(&arg, user_arg, sizeof(arg)))
+		return -EFAULT;
+
+	if ((arg.flags & ~BCH_BY_INDEX) ||
+	    arg.pad[0] ||
+	    arg.pad[1] ||
+	    arg.pad[2])
+		return -EINVAL;
+
+	ca = bch2_device_lookup(c, arg.dev, arg.flags);
+	if (IS_ERR(ca))
+		return PTR_ERR(ca);
+
+	src = bch2_dev_usage_read(c, ca);
+
+	arg.state	= ca->mi.state;
+	arg.bucket_size	= ca->mi.bucket_size;
+	arg.nr_buckets	= ca->mi.nbuckets - ca->mi.first_bucket;
+
+	for (i = 0; i < BCH_DATA_NR; i++) {
+		arg.buckets[i] = src.buckets[i];
+		arg.sectors[i] = src.sectors[i];
 	}
 
-	return 0;
+	percpu_ref_put(&ca->ref);
+
+	return copy_to_user(user_arg, &arg, sizeof(arg));
 }
 
 static long bch2_ioctl_read_super(struct bch_fs *c,
@@ -547,8 +575,10 @@ long bch2_fs_ioctl(struct bch_fs *c, unsigned cmd, void __user *arg)
 	switch (cmd) {
 	case BCH_IOCTL_QUERY_UUID:
 		return bch2_ioctl_query_uuid(c, arg);
-	case BCH_IOCTL_USAGE:
-		return bch2_ioctl_usage(c, arg);
+	case BCH_IOCTL_FS_USAGE:
+		return bch2_ioctl_fs_usage(c, arg);
+	case BCH_IOCTL_DEV_USAGE:
+		return bch2_ioctl_dev_usage(c, arg);
 	}
 
 	if (!capable(CAP_SYS_ADMIN))
diff --git a/fs/bcachefs/replicas.h b/fs/bcachefs/replicas.h
index 0d6e19126021..8527d82841bb 100644
--- a/fs/bcachefs/replicas.h
+++ b/fs/bcachefs/replicas.h
@@ -72,9 +72,6 @@ int bch2_replicas_set_usage(struct bch_fs *,
 
 /* iterate over superblock replicas - used by userspace tools: */
 
-#define replicas_entry_bytes(_i)					\
-	(offsetof(typeof(*(_i)), devs) + (_i)->nr_devs)
-
 #define replicas_entry_next(_i)						\
 	((typeof(_i)) ((void *) (_i) + replicas_entry_bytes(_i)))