protocol/client: Do not fallback to anon-fd if fd is not open

If an open comes on a file when a brick is down and after the brick comes up,
a fop comes on the fd, client xlator would still wind the fop on anon-fd
leading to wrong behavior of the fops in some cases.

Example:
If lk fop is issued on the fd just after the brick is up in the scenario above,
lk fop will be sent on anon-fd instead of failing it on that client xlator.
This lock will never be freed upon close of the fd as flush on anon-fd is
invalid and is not wound below server xlator.

As a fix, failing the fop unless the fd has FALLBACK_TO_ANON_FD flag.

Change-Id: I77692d056660b2858e323bdabdfe0a381807cccc
fixes bz#1390914
Signed-off-by: Pranith Kumar K <pkarampu@redhat.com>

2 files changed, 43 insertions, 1 deletions

diff --git a/tests/bugs/protocol/bug-1390914.t b/tests/bugs/protocol/bug-1390914.t
new file mode 100644
index 00000000000..e3dab92de5a
--- /dev/null
+++ b/tests/bugs/protocol/bug-1390914.t
@@ -0,0 +1,36 @@
+#!/bin/bash
+
+. $(dirname $0)/../../include.rc
+. $(dirname $0)/../../volume.rc
+. $(dirname $0)/../../fileio.rc
+cleanup;
+
+#test that fops are not wound on anon-fd when fd is not open on that brick
+TEST glusterd;
+TEST pidof glusterd
+
+TEST $CLI volume create $V0 replica 3 $H0:$B0/${V0}{1,2,3};
+TEST $CLI volume set $V0 performance.open-behind off
+TEST $CLI volume set $V0 diagnostics.client-log-level DEBUG
+TEST $CLI volume heal $V0 disable
+TEST $CLI volume start $V0
+TEST $CLI volume profile $V0 start
+TEST $GFS -s $H0 --volfile-id=$V0 --direct-io-mode=enable $M0;
+
+TEST touch $M0/1
+TEST kill_brick $V0 $H0 $B0/${V0}1
+TEST fd_open 200 'w' "$M0/1"
+TEST $CLI volume start $V0 force
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 0
+
+#lk should only happen on 2 bricks, if there is a bug, it will plant a lock
+#with anon-fd on first-brick which will never be released because flush won't
+#be wound below server xlator for anon-fd
+TEST flock -x -n 200
+TEST fd_close 200
+
+TEST fd_open 200 'w' "$M0/1"
+#this lock will fail if there is a stale lock
+TEST flock -x -n 200
+TEST fd_close 200
+cleanup;
diff --git a/xlators/protocol/client/src/client-helpers.c b/xlators/protocol/client/src/client-helpers.c
index 24f0cfa6137..53b4484ecc8 100644
--- a/xlators/protocol/client/src/client-helpers.c
+++ b/xlators/protocol/client/src/client-helpers.c
@@ -419,7 +419,13 @@ client_get_remote_fd(xlator_t *this, fd_t *fd, int flags, int64_t *remote_fd)
     {
         fdctx = this_fd_get_ctx(fd, this);
         if (!fdctx) {
-            *remote_fd = GF_ANON_FD_NO;
+            if (fd->anonymous) {
+                *remote_fd = GF_ANON_FD_NO;
+            } else {
+                *remote_fd = -1;
+                gf_msg_debug(this->name, EBADF, "not a valid fd for gfid: %s",
+                             uuid_utoa(fd->inode->gfid));
+            }
         } else {
             if (__is_fd_reopen_in_progress(fdctx))
                 *remote_fd = -1;