[2/2] NFS: Count the bytes of skipped subrequests in nfs_lock_and_join_requests()
diff mbox

Message ID 20170910215242.4212-2-trond.myklebust@primarydata.com
State New
Headers show

Commit Message

Trond Myklebust Sept. 10, 2017, 9:52 p.m. UTC
If we skip a subrequest due to a zero refcount, we should still count
the byte range that it covered so that we accurately reconstruct the
original request size.

Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
---
 fs/nfs/write.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

Patch
diff mbox

diff --git a/fs/nfs/write.c b/fs/nfs/write.c
index 36d34a4c86bd..f68083db63c8 100644
--- a/fs/nfs/write.c
+++ b/fs/nfs/write.c
@@ -504,8 +504,12 @@  nfs_lock_and_join_requests(struct page *page)
 	for (subreq = head->wb_this_page; subreq != head;
 			subreq = subreq->wb_this_page) {
 
-		if (!kref_get_unless_zero(&subreq->wb_kref))
+		if (!kref_get_unless_zero(&subreq->wb_kref)) {
+			if (subreq->wb_offset == head->wb_offset + total_bytes)
+				total_bytes += subreq->wb_bytes;
 			continue;
+		}
+
 		while (!nfs_lock_request(subreq)) {
 			/*
 			 * Unlock page to allow nfs_page_group_sync_on_bit()