Clear PF_FSTRANS over spl_filp_fallocate()

The problem described in 2a5d574 also applies to XFS's file or inode
fallocate method.  Both paths may trigger writeback and expose this
issue, see the full stack below.

When layered on XFS a warning will be emitted under CentOS7 when entering
either the file or inode fallocate method with PF_FSTRANS already set.
To avoid triggering this error PF_FSTRANS is cleared and then reset
in vn_space().

WARNING: at fs/xfs/xfs_aops.c:982 xfs_vm_writepage+0x58b/0x5d0

Call Trace:
 [<ffffffff810a1ed5>] warn_slowpath_common+0x95/0xe0
 [<ffffffff810a1f3a>] warn_slowpath_null+0x1a/0x20
 [<ffffffffa0231fdb>] xfs_vm_writepage+0x58b/0x5d0 [xfs]
 [<ffffffff81173ed7>] __writepage+0x17/0x40
 [<ffffffff81176f81>] write_cache_pages+0x251/0x530
 [<ffffffff811772b1>] generic_writepages+0x51/0x80
 [<ffffffffa0230cb0>] xfs_vm_writepages+0x60/0x80 [xfs]
 [<ffffffff81177300>] do_writepages+0x20/0x30
 [<ffffffff8116a5f5>] __filemap_fdatawrite_range+0xb5/0x100
 [<ffffffff8116a6cb>] filemap_write_and_wait_range+0x8b/0xd0
 [<ffffffffa0235bb4>] xfs_free_file_space+0xf4/0x520 [xfs]
 [<ffffffffa023cbce>] xfs_file_fallocate+0x19e/0x2c0 [xfs]
 [<ffffffffa036c6fc>] vn_space+0x3c/0x40 [spl]
 [<ffffffffa0434817>] vdev_file_io_start+0x207/0x260 [zfs]
 [<ffffffffa047170d>] zio_vdev_io_start+0xad/0x2d0 [zfs]
 [<ffffffffa0474942>] zio_execute+0x82/0xe0 [zfs]
 [<ffffffffa036ba7d>] taskq_thread+0x28d/0x5a0 [spl]
 [<ffffffff810c1777>] kthread+0xd7/0xf0
 [<ffffffff8167de2f>] ret_from_fork+0x3f/0x70

Signed-off-by: Brian Behlendorf <[email protected]>
Signed-off-by: Tim Chase <[email protected]>
Signed-off-by: Nikolay Borisov <[email protected]>
Closes openzfs/zfs#4529

module/spl/spl-vnode.c

@@ -578,6 +578,9 @@ int vn_space(vnode_t *vp, int cmd, struct flock *bfp, int flag,
     offset_t offset, void *x6, void *x7)
 {
 	int error = EOPNOTSUPP;
+#ifdef FALLOC_FL_PUNCH_HOLE
+	int fstrans;
+#endif
 
 	if (cmd != F_FREESP || bfp->l_whence != 0)
 		return (EOPNOTSUPP);
@@ -587,13 +590,25 @@ int vn_space(vnode_t *vp, int cmd, struct flock *bfp, int flag,
 	ASSERT(bfp->l_start >= 0 && bfp->l_len > 0);
 
 #ifdef FALLOC_FL_PUNCH_HOLE
+	/*
+	 * May enter XFS which generates a warning when PF_FSTRANS is set.
+	 * To avoid this the flag is cleared over vfs_sync() and then reset.
+	 */
+	fstrans = spl_fstrans_check();
+	if (fstrans)
+		current->flags &= ~(PF_FSTRANS);
+
 	/*
 	 * When supported by the underlying file system preferentially
 	 * use the fallocate() callback to preallocate the space.
 	 */
 	error = -spl_filp_fallocate(vp->v_file,
 	    FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE,
 	    bfp->l_start, bfp->l_len);
+
+	if (fstrans)
+		current->flags |= PF_FSTRANS;
+
 	if (error == 0)
 		return (0);
 #endif