NFSv4/pNFS: Fix a race to wake on NFS_LAYOUT_DRAIN

bcodding-rh · Anna Schumaker · commit c01776287414 · 2025-06-23T11:01:16.000-04:00
We found a few different systems hung up in writeback waiting on the same page lock, and one task waiting on the NFS_LAYOUT_DRAIN bit in pnfs_update_layout(), however the pnfs_layout_hdr's plh_outstanding count was zero. It seems most likely that this is another race between the waiter and waker similar to commit ed0172a ("SUNRPC: Fix a race to wake a sync task"). Fix it up by applying the advised barrier. Fixes: 880265c ("pNFS: Avoid a live lock condition in pnfs_update_layout()") Signed-off-by: Benjamin Coddington <bcodding@redhat.com> Signed-off-by: Anna Schumaker <anna.schumaker@oracle.com>
diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c
@@ -2059,8 +2059,10 @@ static void nfs_layoutget_begin(struct pnfs_layout_hdr *lo)
 static void nfs_layoutget_end(struct pnfs_layout_hdr *lo)
 {
 	if (atomic_dec_and_test(&lo->plh_outstanding) &&
-	    test_and_clear_bit(NFS_LAYOUT_DRAIN, &lo->plh_flags))
+	    test_and_clear_bit(NFS_LAYOUT_DRAIN, &lo->plh_flags)) {
+		smp_mb__after_atomic();
 		wake_up_bit(&lo->plh_flags, NFS_LAYOUT_DRAIN);
+	}
 }
 
 static bool pnfs_is_first_layoutget(struct pnfs_layout_hdr *lo)

Original file line number	Diff line number	Diff line change
`@@ -2059,8 +2059,10 @@ static void nfs_layoutget_begin(struct pnfs_layout_hdr *lo)`
`2059`	`2059`	`static void nfs_layoutget_end(struct pnfs_layout_hdr *lo)`
`2060`	`2060`	`{`
`2061`	`2061`	`if (atomic_dec_and_test(&lo->plh_outstanding) &&`
`2062`		`- test_and_clear_bit(NFS_LAYOUT_DRAIN, &lo->plh_flags))`
	`2062`	`+ test_and_clear_bit(NFS_LAYOUT_DRAIN, &lo->plh_flags)) {`
	`2063`	`+ smp_mb__after_atomic();`
`2063`	`2064`	`wake_up_bit(&lo->plh_flags, NFS_LAYOUT_DRAIN);`
	`2065`	`+ }`
`2064`	`2066`	`}`
`2065`	`2067`
`2066`	`2068`	`static bool pnfs_is_first_layoutget(struct pnfs_layout_hdr *lo)`