| Message ID | 20210113111707.756662-2-nborisov@suse.com (mailing list archive) |
|---|---|
| State | Deferred, archived |
| Headers | show |
| Series | Remove mrlock | expand |
diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c index b7352bc4c815..499e63da935c 100644 --- a/fs/xfs/xfs_inode.c +++ b/fs/xfs/xfs_inode.c @@ -345,6 +345,13 @@ xfs_ilock_demote( } #if defined(DEBUG) || defined(XFS_WARN) +static bool is_rwsem_write_locked(struct rw_semaphore *rwsem) +{ + /* Copy of RWSEM_WRITE_LOCKED from rwsem.c */ +#define LOCK_BIT (1UL << 0) + return atomic_long_read(&rwsem->count) & LOCK_BIT; +} + int xfs_isilocked( xfs_inode_t *ip,
This is going to be used to check if an rwsem is held for write. Mimicking what current mrlock->mr_writer variable provides but using the generic rwsem infrastructure. One might argue that redefining the locked bit is a layering violation since RWSEM_WRITE_LOCKED is not exposed, however I'd argue back that this is only used in a debugging scenario and the physical layout of the rwsem is unlikely to change. Signed-off-by: Nikolay Borisov <nborisov@suse.com> --- fs/xfs/xfs_inode.c | 7 +++++++ 1 file changed, 7 insertions(+)