Thread (14 messages) 14 messages, 3 authors, 2021-11-08

Re: [PATCH v6 3/3] btrfs: consolidate device_list_mutex in prepare_sprout to its parent

From: Nikolay Borisov <hidden>
Date: 2021-09-21 11:01:41


On 21.09.21 г. 7:33, Anand Jain wrote:
quoted hunk ↗ jump to hunk
btrfs_prepare_sprout() splices seed devices into its own struct fs_devices,
so that its parent function btrfs_init_new_device() can add the new sprout
device to fs_info->fs_devices.

Both btrfs_prepare_sprout() and btrfs_init_new_device() needs
device_list_mutex. But they are holding it sequentially, thus creates a
small window to an opportunity to race. Close this opportunity and hold
device_list_mutex common to both btrfs_init_new_device() and
btrfs_prepare_sprout().

This patch splits btrfs_prepare_sprout() into btrfs_alloc_sprout() and
btrfs_splice_sprout(). This split is essential because device_list_mutex
shouldn't be held for btrfs_alloc_sprout() but must be held for
btrfs_splice_sprout(). So now a common device_list_mutex can be used
between btrfs_init_new_device() and btrfs_splice_sprout().

This patch also moves the lockdep_assert_held(&uuid_mutex) from the
starting of the function to just above the line where we need this lock.

Signed-off-by: Anand Jain <redacted>
---
v6:
 Remove RFC.
 Split btrfs_prepare_sprout so that the allocation part can be outside
 of the device_list_mutex in the parent function btrfs_init_new_device().

 fs/btrfs/volumes.c | 46 +++++++++++++++++++++++++++++++---------------
 1 file changed, 31 insertions(+), 15 deletions(-)
diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index e4079e25db70..b21eac32ec98 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -2376,19 +2376,13 @@ struct btrfs_device *btrfs_find_device_by_devspec(
 	return btrfs_find_device_by_path(fs_info, device_path);
 }
 
-/*
- * does all the dirty work required for changing file system's UUID.
- */
-static int btrfs_prepare_sprout(struct btrfs_fs_info *fs_info)
+static int btrfs_alloc_sprout(struct btrfs_fs_info *fs_info,
+			      struct btrfs_fs_devices **seed_devices_ret)
Nope, make the function return a struct btrfs_fs_devices *.
quoted hunk ↗ jump to hunk
 {
 	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
 	struct btrfs_fs_devices *old_devices;
 	struct btrfs_fs_devices *seed_devices;
-	struct btrfs_super_block *disk_super = fs_info->super_copy;
-	struct btrfs_device *device;
-	u64 super_flags;
 
-	lockdep_assert_held(&uuid_mutex);
 	if (!fs_devices->seeding)
 		return -EINVAL;
 
@@ -2412,6 +2406,7 @@ static int btrfs_prepare_sprout(struct btrfs_fs_info *fs_info)
 		return PTR_ERR(old_devices);
 	}
 
+	lockdep_assert_held(&uuid_mutex);
 	list_add(&old_devices->fs_list, &fs_uuids);
 
 	memcpy(seed_devices, fs_devices, sizeof(*seed_devices));
@@ -2419,7 +2414,23 @@ static int btrfs_prepare_sprout(struct btrfs_fs_info *fs_info)
 	INIT_LIST_HEAD(&seed_devices->devices);
 	INIT_LIST_HEAD(&seed_devices->alloc_list);
 
-	mutex_lock(&fs_devices->device_list_mutex);
+	*seed_devices_ret = seed_devices;
+
+	return 0;
+}
+
+/*
+ * Splice seed devices into the sprout fs_devices.
+ * Generate a new fsid for the sprouted readwrite btrfs.
+ */
+static void btrfs_splice_sprout(struct btrfs_fs_info *fs_info,
+				struct btrfs_fs_devices *seed_devices)
+{
This function is missing a lockdep_assert_held annotation and it depends
on the device_list_mutex being held.

However looking at the resulting code it doesn't look good, because
btrfs_splice_sporut suggests you simply add the seed device to a bunch
of places, yet looking at the function's body it's evident it actually
finishes some parts of the initialization, changes the uuid of the
fs_devices. I'm not convinced it really makes the code better or at the
very least the 'splice_sprout' needs to be changed, because splicing is
a minot part of what this function really does.


<snip>
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help