* [PATCH] Support non reserved ports for NFS client
@ 2003-06-18 14:51 Andi Kleen
2003-06-18 18:32 ` Trond Myklebust
0 siblings, 1 reply; 4+ messages in thread
From: Andi Kleen @ 2003-06-18 14:51 UTC (permalink / raw
To: trond.myklebust, linux-kernel
Currently you cannot have more than 1024 mounts for a single local
IP address because the NFS client always tries to get a "secure"
port <1024.
This patch adds a new noreserved mount option to disable this.
It is off by default because most NFS servers (including linux knfsd)
check it by default, unless special options are given.
Patch for 2.4.21. Also requires a small patch to util-linux/mount,
which I'm sending separately.
-Andi
diff -u linux-2.4.21-work/fs/lockd/host.c-NONRES linux-2.4.21-work/fs/lockd/host.c
--- linux-2.4.21-work/fs/lockd/host.c-NONRES 2002-09-25 01:49:40.000000000 +0200
+++ linux-2.4.21-work/fs/lockd/host.c 2003-06-18 16:09:58.000000000 +0200
@@ -193,7 +193,7 @@
/* Create RPC socket as root user so we get a priv port */
current->fsuid = 0;
cap_raise (current->cap_effective, CAP_NET_BIND_SERVICE);
- xprt = xprt_create_proto(host->h_proto, &host->h_addr, NULL);
+ xprt = xprt_create_proto(host->h_proto, &host->h_addr, NULL, 1);
current->fsuid = saved_fsuid;
current->cap_effective = saved_cap;
if (xprt == NULL)
diff -u linux-2.4.21-work/fs/lockd/mon.c-NONRES linux-2.4.21-work/fs/lockd/mon.c
--- linux-2.4.21-work/fs/lockd/mon.c-NONRES 2002-09-25 01:49:40.000000000 +0200
+++ linux-2.4.21-work/fs/lockd/mon.c 2003-06-18 16:09:58.000000000 +0200
@@ -110,7 +110,7 @@
sin.sin_addr.s_addr = htonl(INADDR_LOOPBACK);
sin.sin_port = 0;
- xprt = xprt_create_proto(IPPROTO_UDP, &sin, NULL);
+ xprt = xprt_create_proto(IPPROTO_UDP, &sin, NULL, 1);
if (!xprt)
goto out;
diff -u linux-2.4.21-work/fs/nfs/inode.c-NONRES linux-2.4.21-work/fs/nfs/inode.c
--- linux-2.4.21-work/fs/nfs/inode.c-NONRES 2002-11-30 00:37:12.000000000 +0100
+++ linux-2.4.21-work/fs/nfs/inode.c 2003-06-18 16:09:58.000000000 +0200
@@ -361,7 +361,8 @@
/* Now create transport and client */
xprt = xprt_create_proto(tcp? IPPROTO_TCP : IPPROTO_UDP,
- &srvaddr, &timeparms);
+ &srvaddr, &timeparms,
+ !(data->flags & NFS_MOUNT_NONRESERVED));
if (xprt == NULL)
goto out_no_xprt;
@@ -560,6 +561,7 @@
char *str;
char *nostr;
} nfs_info[] = {
+ { NFS_MOUNT_NONRESERVED, ",nonreserved", ",reserved" },
{ NFS_MOUNT_SOFT, ",soft", ",hard" },
{ NFS_MOUNT_INTR, ",intr", "" },
{ NFS_MOUNT_POSIX, ",posix", "" },
diff -u linux-2.4.21-work/fs/nfs/mount_clnt.c-NONRES linux-2.4.21-work/fs/nfs/mount_clnt.c
--- linux-2.4.21-work/fs/nfs/mount_clnt.c-NONRES 2002-11-30 00:37:12.000000000 +0100
+++ linux-2.4.21-work/fs/nfs/mount_clnt.c 2003-06-18 16:09:58.000000000 +0200
@@ -70,7 +70,7 @@
struct rpc_xprt *xprt;
struct rpc_clnt *clnt;
- if (!(xprt = xprt_create_proto(protocol, srvaddr, NULL)))
+ if (!(xprt = xprt_create_proto(protocol, srvaddr, NULL, 1)))
return NULL;
clnt = rpc_create_client(xprt, hostname,
diff -u linux-2.4.21-work/include/linux/sunrpc/xprt.h-NONRES linux-2.4.21-work/include/linux/sunrpc/xprt.h
--- linux-2.4.21-work/include/linux/sunrpc/xprt.h-NONRES 2003-05-27 13:44:14.000000000 +0200
+++ linux-2.4.21-work/include/linux/sunrpc/xprt.h 2003-06-18 16:27:23.000000000 +0200
@@ -134,7 +134,8 @@
unsigned long sockstate; /* Socket state */
unsigned char shutdown : 1, /* being shut down */
nocong : 1, /* no congestion control */
- stream : 1; /* TCP */
+ stream : 1, /* TCP */
+ reserved : 1; /* reserved source port */
/*
* State of TCP reply receive stuff
@@ -166,7 +167,7 @@
#ifdef __KERNEL__
struct rpc_xprt * xprt_create_proto(int proto, struct sockaddr_in *addr,
- struct rpc_timeout *toparms);
+ struct rpc_timeout *toparms, int res);
int xprt_destroy(struct rpc_xprt *);
void xprt_shutdown(struct rpc_xprt *);
void xprt_default_timeout(struct rpc_timeout *, int);
diff -u linux-2.4.21-work/include/linux/nfs_mount.h-NONRES linux-2.4.21-work/include/linux/nfs_mount.h
--- linux-2.4.21-work/include/linux/nfs_mount.h-NONRES 2003-05-27 12:04:18.000000000 +0200
+++ linux-2.4.21-work/include/linux/nfs_mount.h 2003-06-18 16:31:16.000000000 +0200
@@ -53,6 +53,8 @@
#define NFS_MOUNT_KERBEROS 0x0100 /* 3 */
#define NFS_MOUNT_NONLM 0x0200 /* 3 */
#define NFS_MOUNT_BROKEN_SUID 0x0400 /* 4 */
+/* 0x800 for NOACL */
+#define NFS_MOUNT_NONRESERVED 0x1000
#define NFS_MOUNT_FLAGMASK 0xFFFF
#endif
diff -u linux-2.4.21-work/net/sunrpc/xprt.c-NONRES linux-2.4.21-work/net/sunrpc/xprt.c
--- linux-2.4.21-work/net/sunrpc/xprt.c-NONRES 2003-06-16 11:37:56.000000000 +0200
+++ linux-2.4.21-work/net/sunrpc/xprt.c 2003-06-18 16:10:01.000000000 +0200
@@ -86,7 +86,7 @@
static void xprt_reserve_status(struct rpc_task *task);
static void xprt_disconnect(struct rpc_xprt *);
static void xprt_reconn_status(struct rpc_task *task);
-static struct socket *xprt_create_socket(int, struct rpc_timeout *);
+static struct socket *xprt_create_socket(int, struct rpc_timeout *, int res);
static int xprt_bind_socket(struct rpc_xprt *, struct socket *);
static int __xprt_get_cong(struct rpc_xprt *, struct rpc_task *);
@@ -450,7 +450,8 @@
status = -ENOTCONN;
if (!(inet = xprt->inet)) {
/* Create an unconnected socket */
- if (!(sock = xprt_create_socket(xprt->prot, &xprt->timeout)))
+ if (!(sock = xprt_create_socket(xprt->prot, &xprt->timeout,
+ xprt->reserved)))
goto defer;
xprt_bind_socket(xprt, sock);
inet = sock->sk;
@@ -1349,7 +1350,7 @@
*/
static struct rpc_xprt *
xprt_setup(struct socket *sock, int proto,
- struct sockaddr_in *ap, struct rpc_timeout *to)
+ struct sockaddr_in *ap, struct rpc_timeout *to, int res)
{
struct rpc_xprt *xprt;
struct rpc_rqst *req;
@@ -1394,7 +1395,8 @@
req->rq_next = req + 1;
req->rq_next = NULL;
xprt->free = xprt->slot;
-
+ xprt->reserved = res;
+
dprintk("RPC: created transport %p\n", xprt);
xprt_bind_socket(xprt, sock);
@@ -1487,7 +1489,7 @@
* Create a client socket given the protocol and peer address.
*/
static struct socket *
-xprt_create_socket(int proto, struct rpc_timeout *to)
+xprt_create_socket(int proto, struct rpc_timeout *to, int res)
{
struct socket *sock;
int type, err;
@@ -1503,7 +1505,7 @@
}
/* If the caller has the capability, bind to a reserved port */
- if (capable(CAP_NET_BIND_SERVICE) && xprt_bindresvport(sock) < 0)
+ if (res && capable(CAP_NET_BIND_SERVICE) && xprt_bindresvport(sock) < 0)
goto failed;
return sock;
@@ -1517,17 +1519,17 @@
* Create an RPC client transport given the protocol and peer address.
*/
struct rpc_xprt *
-xprt_create_proto(int proto, struct sockaddr_in *sap, struct rpc_timeout *to)
+xprt_create_proto(int proto, struct sockaddr_in *sap, struct rpc_timeout *to, int res)
{
struct socket *sock;
struct rpc_xprt *xprt;
dprintk("RPC: xprt_create_proto called\n");
- if (!(sock = xprt_create_socket(proto, to)))
+ if (!(sock = xprt_create_socket(proto, to, res)))
return NULL;
- if (!(xprt = xprt_setup(sock, proto, sap, to)))
+ if (!(xprt = xprt_setup(sock, proto, sap, to, res)))
sock_release(sock);
return xprt;
diff -u linux-2.4.21-work/net/sunrpc/pmap_clnt.c-NONRES linux-2.4.21-work/net/sunrpc/pmap_clnt.c
--- linux-2.4.21-work/net/sunrpc/pmap_clnt.c-NONRES 2002-09-25 01:49:41.000000000 +0200
+++ linux-2.4.21-work/net/sunrpc/pmap_clnt.c 2003-06-18 16:10:01.000000000 +0200
@@ -187,7 +187,8 @@
struct rpc_clnt *clnt;
/* printk("pmap: create xprt\n"); */
- if (!(xprt = xprt_create_proto(proto, srvaddr, NULL)))
+ /* use non reserved ports for now. I hope that's ok. -AK */
+ if (!(xprt = xprt_create_proto(proto, srvaddr, NULL, 0)))
return NULL;
xprt->addr.sin_port = htons(RPC_PMAP_PORT);
^ permalink raw reply [flat|nested] 4+ messages in thread
* [PATCH] Support non reserved ports for NFS client
2003-06-18 14:51 [PATCH] Support non reserved ports for NFS client Andi Kleen
@ 2003-06-18 18:32 ` Trond Myklebust
0 siblings, 0 replies; 4+ messages in thread
From: Trond Myklebust @ 2003-06-18 18:32 UTC (permalink / raw
To: Andi Kleen; +Cc: linux-kernel
>>>>> " " == Andi Kleen <ak@suse.de> writes:
> Currently you cannot have more than 1024 mounts for a single
> local IP address because the NFS client always tries to get a
> "secure" port <1024.
> This patch adds a new noreserved mount option to disable this.
Hi Andi,
I've already got a patch to accomplish most of this in
http://www.fys.uio.no/~trondmy/src/2.4.21/linux-2.4.21-11-fix_tcprace3.dif
It's a backport of some patches that have already gone into 2.5.x to
fix some TCP client reconnection issues. I was hoping to push these
patches to Marcelo for 2.4.22.
Could you therefore please just send me a patch for the
NFS_MOUNT_NONRESERVED mount option alone. Then I'll append it to this
patch series?
This should also make it easy to port the whole thing forward to 2.5.x
(although please note that for 2.5.x I'm also planning some other
improvements that will allow us to share the struct xprt between
different RPC clients).
Cheers,
Trond
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] Support non reserved ports for NFS client
@ 2003-06-18 20:07 Bruce Allan
2003-06-18 20:57 ` Andi Kleen
0 siblings, 1 reply; 4+ messages in thread
From: Bruce Allan @ 2003-06-18 20:07 UTC (permalink / raw
To: trond.myklebust; +Cc: Andi Kleen, linux-kernel
May I make the suggestion that instead of creating a new flag (i.e.
NFS_MOUNT_NONRESERVED) and using a new bit in the flags field of struct
nfs_mount_data, Andi use the already existing and unused NFS_MOUNT_SECURE
assuming that is what was originally intended for this bit. This would
also match in name with the secure/insecure export option on the server.
Regards,
---
Bruce Allan <bruce.allan@us.ibm.com>
Software Engineer, Linux Technology Center
IBM Corporation, Beaverton OR
503-578-4187 IBM Tie-line 775-4187
Trond Myklebust
<trond.myklebust@fys.ui To: Andi Kleen <ak@suse.de>
o.no> cc: linux-kernel@vger.kernel.org
Sent by: Subject: [PATCH] Support non reserved ports for NFS client
linux-kernel-owner@vger
.kernel.org
06/18/2003 11:32 AM
Please respond to
trond.myklebust
>>>>> " " == Andi Kleen <ak@suse.de> writes:
> Currently you cannot have more than 1024 mounts for a single
> local IP address because the NFS client always tries to get a
> "secure" port <1024.
> This patch adds a new noreserved mount option to disable this.
Hi Andi,
I've already got a patch to accomplish most of this in
http://www.fys.uio.no/~trondmy/src/2.4.21
/linux-2.4.21-11-fix_tcprace3.dif
It's a backport of some patches that have already gone into 2.5.x to
fix some TCP client reconnection issues. I was hoping to push these
patches to Marcelo for 2.4.22.
Could you therefore please just send me a patch for the
NFS_MOUNT_NONRESERVED mount option alone. Then I'll append it to this
patch series?
This should also make it easy to port the whole thing forward to 2.5.x
(although please note that for 2.5.x I'm also planning some other
improvements that will allow us to share the struct xprt between
different RPC clients).
Cheers,
Trond
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] Support non reserved ports for NFS client
2003-06-18 20:07 Bruce Allan
@ 2003-06-18 20:57 ` Andi Kleen
0 siblings, 0 replies; 4+ messages in thread
From: Andi Kleen @ 2003-06-18 20:57 UTC (permalink / raw
To: Bruce Allan; +Cc: trond.myklebust, Andi Kleen, linux-kernel
On Wed, Jun 18, 2003 at 01:07:05PM -0700, Bruce Allan wrote:
> May I make the suggestion that instead of creating a new flag (i.e.
> NFS_MOUNT_NONRESERVED) and using a new bit in the flags field of struct
> nfs_mount_data, Andi use the already existing and unused NFS_MOUNT_SECURE
> assuming that is what was originally intended for this bit. This would
> also match in name with the secure/insecure export option on the server.
Makes sense yes. Thanks for the suggestion.
-Andi
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2003-06-18 20:43 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2003-06-18 14:51 [PATCH] Support non reserved ports for NFS client Andi Kleen
2003-06-18 18:32 ` Trond Myklebust
-- strict thread matches above, loose matches on Subject: below --
2003-06-18 20:07 Bruce Allan
2003-06-18 20:57 ` Andi Kleen
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.