From: "J. Bruce Fields" <bfields@redhat.com>
To: linux-nfs@vger.kernel.org
Cc: "J. Bruce Fields" <bfields@redhat.com>
Subject: [PATCH 05/14] rpc: keep backchannel xprt as long as server connection
Date: Tue, 11 Jan 2011 11:21:25 -0500 [thread overview]
Message-ID: <1294762894-21371-6-git-send-email-bfields@redhat.com> (raw)
In-Reply-To: <1294762894-21371-1-git-send-email-bfields@redhat.com>
Multiple backchannels can share the same tcp connection; from rfc 5661 section
2.10.3.1:
A connection's association with a session is not exclusive. A
connection associated with the channel(s) of one session may be
simultaneously associated with the channel(s) of other sessions
including sessions associated with other client IDs.
However, multiple backchannels share a connection, they must all share
the same xid stream (hence the same rpc_xprt); the only way we have to
match replies with calls at the rpc layer is using the xid.
So, keep the rpc_xprt around as long as the connection lasts, in case
we're asked to use the connection as a backchannel again.
Requests to create new backchannel clients over a given server
connection should results in creating new clients that reuse the
existing rpc_xprt.
But to start, just reject attempts to associate multiple rpc_xprt's with
the same underlying bc_xprt.
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
---
net/sunrpc/svc_xprt.c | 4 ++++
net/sunrpc/xprt.c | 2 +-
net/sunrpc/xprtsock.c | 34 ++++++++++++++++++++++++----------
3 files changed, 29 insertions(+), 11 deletions(-)
diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
index 173f3b9..ab86b79 100644
--- a/net/sunrpc/svc_xprt.c
+++ b/net/sunrpc/svc_xprt.c
@@ -13,6 +13,7 @@
#include <linux/sunrpc/stats.h>
#include <linux/sunrpc/svc_xprt.h>
#include <linux/sunrpc/svcsock.h>
+#include <linux/sunrpc/xprt.h>
#define RPCDBG_FACILITY RPCDBG_SVCXPRT
@@ -128,6 +129,9 @@ static void svc_xprt_free(struct kref *kref)
if (test_bit(XPT_CACHE_AUTH, &xprt->xpt_flags))
svcauth_unix_info_release(xprt);
put_net(xprt->xpt_net);
+ /* See comment on corresponding get in xs_setup_bc_tcp(): */
+ if (xprt->xpt_bc_xprt)
+ xprt_put(xprt->xpt_bc_xprt);
xprt->xpt_ops->xpo_free(xprt);
module_put(owner);
}
diff --git a/net/sunrpc/xprt.c b/net/sunrpc/xprt.c
index 4c8f18a..749ad15 100644
--- a/net/sunrpc/xprt.c
+++ b/net/sunrpc/xprt.c
@@ -965,6 +965,7 @@ struct rpc_xprt *xprt_alloc(struct net *net, int size, int max_req)
xprt = kzalloc(size, GFP_KERNEL);
if (xprt == NULL)
goto out;
+ kref_init(&xprt->kref);
xprt->max_reqs = max_req;
xprt->slot = kcalloc(max_req, sizeof(struct rpc_rqst), GFP_KERNEL);
@@ -1102,7 +1103,6 @@ found:
return xprt;
}
- kref_init(&xprt->kref);
spin_lock_init(&xprt->transport_lock);
spin_lock_init(&xprt->reserve_lock);
diff --git a/net/sunrpc/xprtsock.c b/net/sunrpc/xprtsock.c
index 18dc42e..0ef4dd4 100644
--- a/net/sunrpc/xprtsock.c
+++ b/net/sunrpc/xprtsock.c
@@ -2375,16 +2375,6 @@ static struct rpc_xprt *xs_setup_bc_tcp(struct xprt_create *args)
xprt->reestablish_timeout = 0;
xprt->idle_timeout = 0;
- /*
- * The backchannel uses the same socket connection as the
- * forechannel
- */
- args->bc_xprt->xpt_bc_xprt = xprt;
- xprt->bc_xprt = args->bc_xprt;
- bc_sock = container_of(args->bc_xprt, struct svc_sock, sk_xprt);
- transport->sock = bc_sock->sk_sock;
- transport->inet = bc_sock->sk_sk;
-
xprt->ops = &bc_tcp_ops;
switch (addr->sa_family) {
@@ -2407,6 +2397,29 @@ static struct rpc_xprt *xs_setup_bc_tcp(struct xprt_create *args)
xprt->address_strings[RPC_DISPLAY_PROTO]);
/*
+ * The backchannel uses the same socket connection as the
+ * forechannel
+ */
+ if (args->bc_xprt->xpt_bc_xprt) {
+ /* XXX: actually, want to catch this case... */
+ ret = ERR_PTR(-EINVAL);
+ goto out_err;
+ }
+ /*
+ * Once we've associated a backchannel xprt with a connection,
+ * we want to keep it around as long as long as the connection
+ * lasts, in case we need to start using it for a backchannel
+ * again; this reference won't be dropped until bc_xprt is
+ * destroyed.
+ */
+ xprt_get(xprt);
+ args->bc_xprt->xpt_bc_xprt = xprt;
+ xprt->bc_xprt = args->bc_xprt;
+ bc_sock = container_of(args->bc_xprt, struct svc_sock, sk_xprt);
+ transport->sock = bc_sock->sk_sock;
+ transport->inet = bc_sock->sk_sk;
+
+ /*
* Since we don't want connections for the backchannel, we set
* the xprt status to connected
*/
@@ -2415,6 +2428,7 @@ static struct rpc_xprt *xs_setup_bc_tcp(struct xprt_create *args)
if (try_module_get(THIS_MODULE))
return xprt;
+ xprt_put(xprt);
ret = ERR_PTR(-EINVAL);
out_err:
xprt_free(xprt);
--
1.7.1
next prev parent reply other threads:[~2011-01-11 16:21 UTC|newest]
Thread overview: 15+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-01-11 16:21 server sessions patches J. Bruce Fields
2011-01-11 16:21 ` [PATCH 01/14] nfsd4: modify session list under cl_lock J. Bruce Fields
2011-01-11 16:21 ` [PATCH 02/14] nfsd4: support BIND_CONN_TO_SESSION J. Bruce Fields
2011-01-11 16:21 ` [PATCH 03/14] nfsd4: allow backchannel recovery J. Bruce Fields
2011-01-11 16:21 ` [PATCH 04/14] rpc: move sk_bc_xprt to svc_xprt J. Bruce Fields
2011-01-11 16:21 ` J. Bruce Fields [this message]
2011-01-11 16:21 ` [PATCH 06/14] rpc: allow xprt_class->setup to return a preexisting xprt J. Bruce Fields
2011-01-11 16:21 ` [PATCH 07/14] nfsd4: keep finer-grained callback status J. Bruce Fields
2011-01-11 16:21 ` [PATCH 08/14] nfsd4: set sequence flag when backchannel is down J. Bruce Fields
2011-01-11 16:21 ` [PATCH 09/14] nfsd4: re-probe callback on connection loss J. Bruce Fields
2011-01-11 16:21 ` [PATCH 10/14] nfsd4: make sure sequence flags are set after destroy_session J. Bruce Fields
2011-01-11 16:21 ` [PATCH 11/14] nfsd4: add helper function to run callbacks J. Bruce Fields
2011-01-11 16:21 ` [PATCH 12/14] nfsd4: give out delegations more quickly in 4.1 case J. Bruce Fields
2011-01-11 16:21 ` [PATCH 13/14] nfsd4: simplify nfsd4_cb_prepare J. Bruce Fields
2011-01-11 16:21 ` [PATCH 14/14] nfsd4: allow restarting callbacks J. Bruce Fields
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1294762894-21371-6-git-send-email-bfields@redhat.com \
--to=bfields@redhat.com \
--cc=linux-nfs@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).