[PATCH 00/38] svc: SVC Transport Switch

[PATCH 00/38] svc: SVC Transport Switch

[Tom Tucker]

The Server Side Transport switch implements a pluggable transport
provider layer for RPC servers. This switch allows network transport
drivers to be dynamically loaded and registered, thereby extending the
set of transports that can carry RPC. This patchset includes a
transport provider driver for TCP and UDP sockets.

This version of the patchset includes the following updates to the
original patch series:

- Changed transport hdr len fields and computations to use size_t

- Matched local variables to sign of socket specific write-space
  functions in xpo_wspace transport functions.

- Modified tcp_has_wspace function to include min_wspace consideration

- Changed address length functions to size_t

- Changed port helper function return types to unsigned short

- Changed error return from svc_addr_len to EAFNOSUPPORT from ENOTSUPP

- When modifying function prototypes or adding new functions adhered
  to convention of keeping signature on a single line.

- Removed debug printk from svc_send

- Created separate patch for change that moves the call to
  svc_xprt_received to common code for the accept path

- Combined the portlist update patches into a single patch and updated
  the error codes to give more helpful perror strings when writing
  garbage to the portlist file.

This patchset is based on 2.6.24-rc5 and has been tested on V2, V3 and
V4 over the UDP, TCP and RDMA transports.

-- 
Signed-off-by: Tom Tucker <tom@opengridcomputing.com>

[PATCH 01/38] svc: Add an svc transport class

[Tom Tucker]

The transport class (svc_xprt_class) represents a type of transport, e.g. 
udp, tcp, rdma.  A transport class has a unique name and a set of transport 
operations kept in the svc_xprt_ops structure.

A transport class can be dynamically registered and unregisterd. The 
svc_xprt_class represents the module that implements the transport
type and keeps reference counts on the module to avoid unloading while
there are active users.

The endpoint (svc_xprt) is a generic, transport independent endpoint that can 
be used to send and receive data for an RPC service. It inherits it's 
operations from the transport class. 

A transport driver module registers and unregisters itself with svc sunrpc
by calling svc_reg_xprt_class, and svc_unreg_xprt_class respectively. 

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 include/linux/sunrpc/debug.h    |    1 
 include/linux/sunrpc/svc_xprt.h |   31 +++++++++++++
 net/sunrpc/Makefile             |    3 +
 net/sunrpc/svc_xprt.c           |   95 +++++++++++++++++++++++++++++++++++++++
 4 files changed, 129 insertions(+), 1 deletions(-)

diff --git a/include/linux/sunrpc/debug.h b/include/linux/sunrpc/debug.h
index 3912cf1..092fcfa 100644
--- a/include/linux/sunrpc/debug.h
+++ b/include/linux/sunrpc/debug.h
@@ -21,6 +21,7 @@
 #define RPCDBG_SCHED		0x0040
 #define RPCDBG_TRANS		0x0080
 #define RPCDBG_SVCSOCK		0x0100
+#define RPCDBG_SVCXPRT		0x0100
 #define RPCDBG_SVCDSP		0x0200
 #define RPCDBG_MISC		0x0400
 #define RPCDBG_CACHE		0x0800
diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
new file mode 100644
index 0000000..a8b1da8
--- /dev/null
+++ b/include/linux/sunrpc/svc_xprt.h
@@ -0,0 +1,31 @@
+/*
+ * linux/include/linux/sunrpc/svc_xprt.h
+ *
+ * RPC server transport I/O
+ */
+
+#ifndef SUNRPC_SVC_XPRT_H
+#define SUNRPC_SVC_XPRT_H
+
+#include <linux/sunrpc/svc.h>
+
+struct svc_xprt_ops {
+};
+
+struct svc_xprt_class {
+	const char		*xcl_name;
+	struct module		*xcl_owner;
+	struct svc_xprt_ops	*xcl_ops;
+	struct list_head	xcl_list;
+};
+
+struct svc_xprt {
+	struct svc_xprt_class	*xpt_class;
+	struct svc_xprt_ops	*xpt_ops;
+};
+
+int	svc_reg_xprt_class(struct svc_xprt_class *);
+int	svc_unreg_xprt_class(struct svc_xprt_class *);
+void	svc_xprt_init(struct svc_xprt_class *, struct svc_xprt *);
+
+#endif /* SUNRPC_SVC_XPRT_H */
diff --git a/net/sunrpc/Makefile b/net/sunrpc/Makefile
index 5c69a72..92e1dbe 100644
--- a/net/sunrpc/Makefile
+++ b/net/sunrpc/Makefile
@@ -11,6 +11,7 @@ sunrpc-y := clnt.o xprt.o socklib.o xprtsock.o sched.o \
 	    auth.o auth_null.o auth_unix.o \
 	    svc.o svcsock.o svcauth.o svcauth_unix.o \
 	    rpcb_clnt.o timer.o xdr.o \
-	    sunrpc_syms.o cache.o rpc_pipe.o
+	    sunrpc_syms.o cache.o rpc_pipe.o \
+	    svc_xprt.o
 sunrpc-$(CONFIG_PROC_FS) += stats.o
 sunrpc-$(CONFIG_SYSCTL) += sysctl.o
diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
new file mode 100644
index 0000000..92ea85b
--- /dev/null
+++ b/net/sunrpc/svc_xprt.c
@@ -0,0 +1,95 @@
+/*
+ * linux/net/sunrpc/svc_xprt.c
+ *
+ * Author: Tom Tucker <tom@opengridcomputing.com>
+ */
+
+#include <linux/sched.h>
+#include <linux/errno.h>
+#include <linux/fcntl.h>
+#include <linux/net.h>
+#include <linux/in.h>
+#include <linux/inet.h>
+#include <linux/udp.h>
+#include <linux/tcp.h>
+#include <linux/unistd.h>
+#include <linux/slab.h>
+#include <linux/netdevice.h>
+#include <linux/skbuff.h>
+#include <linux/file.h>
+#include <linux/freezer.h>
+#include <net/sock.h>
+#include <net/checksum.h>
+#include <net/ip.h>
+#include <net/ipv6.h>
+#include <net/tcp_states.h>
+#include <linux/uaccess.h>
+#include <asm/ioctls.h>
+
+#include <linux/sunrpc/types.h>
+#include <linux/sunrpc/clnt.h>
+#include <linux/sunrpc/xdr.h>
+#include <linux/sunrpc/svcsock.h>
+#include <linux/sunrpc/stats.h>
+#include <linux/sunrpc/svc_xprt.h>
+
+#define RPCDBG_FACILITY	RPCDBG_SVCXPRT
+
+/* List of registered transport classes */
+static DEFINE_SPINLOCK(svc_xprt_class_lock);
+static LIST_HEAD(svc_xprt_class_list);
+
+int svc_reg_xprt_class(struct svc_xprt_class *xcl)
+{
+	struct svc_xprt_class *cl;
+	int res = -EEXIST;
+
+	dprintk("svc: Adding svc transport class '%s'\n",
+		xcl->xcl_name);
+
+	INIT_LIST_HEAD(&xcl->xcl_list);
+	spin_lock(&svc_xprt_class_lock);
+	list_for_each_entry(cl, &svc_xprt_class_list, xcl_list) {
+		if (xcl == cl)
+			goto out;
+	}
+	list_add_tail(&xcl->xcl_list, &svc_xprt_class_list);
+	res = 0;
+out:
+	spin_unlock(&svc_xprt_class_lock);
+	return res;
+}
+EXPORT_SYMBOL_GPL(svc_reg_xprt_class);
+
+int svc_unreg_xprt_class(struct svc_xprt_class *xcl)
+{
+	struct svc_xprt_class *cl;
+	int res = 0;
+
+	dprintk("svc: Removing svc transport class '%s'\n", xcl->xcl_name);
+
+	spin_lock(&svc_xprt_class_lock);
+	list_for_each_entry(cl, &svc_xprt_class_list, xcl_list) {
+		if (xcl == cl) {
+			list_del_init(&cl->xcl_list);
+			goto out;
+		}
+	}
+	res = -ENOENT;
+ out:
+	spin_unlock(&svc_xprt_class_lock);
+	return res;
+}
+EXPORT_SYMBOL_GPL(svc_unreg_xprt_class);
+
+/*
+ * Called by transport drivers to initialize the transport independent
+ * portion of the transport instance.
+ */
+void svc_xprt_init(struct svc_xprt_class *xcl, struct svc_xprt *xprt)
+{
+	memset(xprt, 0, sizeof(*xprt));
+	xprt->xpt_class = xcl;
+	xprt->xpt_ops = xcl->xcl_ops;
+}
+EXPORT_SYMBOL_GPL(svc_xprt_init);

Re: [PATCH 01/38] svc: Add an svc transport class

[J. Bruce Fields]

Sorry for joining in a little late....

On Tue, Dec 11, 2007 at 05:31:54PM -0600, Tom Tucker wrote:
> +int svc_reg_xprt_class(struct svc_xprt_class *xcl)

None of the callers appear to check the return value, so this should
probably be a void return.

> +{
> +	struct svc_xprt_class *cl;
> +	int res = -EEXIST;
> +
> +	dprintk("svc: Adding svc transport class '%s'\n",
> +		xcl->xcl_name);
> +
> +	INIT_LIST_HEAD(&xcl->xcl_list);

Unless we care how xcl_list is set in the error case, this is
unnecessary.

> +	spin_lock(&svc_xprt_class_lock);
> +	list_for_each_entry(cl, &svc_xprt_class_list, xcl_list) {
> +		if (xcl == cl)
> +			goto out;

Is this even worth checking?  Accidentally registering the identical
struct svc_xprt_class * seems an unlikely mistake for a transport-class
coder to make, given that they only need call this function once per
transport, in the module initialization.  Maybe make this a BUG()?  Or
check for duplicate xcl_name's if that seems a more likely error?

> +	}
> +	list_add_tail(&xcl->xcl_list, &svc_xprt_class_list);
> +	res = 0;
> +out:
> +	spin_unlock(&svc_xprt_class_lock);
> +	return res;
> +}
> +EXPORT_SYMBOL_GPL(svc_reg_xprt_class);
> +
> +int svc_unreg_xprt_class(struct svc_xprt_class *xcl)
> +{
> +	struct svc_xprt_class *cl;
> +	int res = 0;
> +
> +	dprintk("svc: Removing svc transport class '%s'\n", xcl->xcl_name);
> +
> +	spin_lock(&svc_xprt_class_lock);
> +	list_for_each_entry(cl, &svc_xprt_class_list, xcl_list) {
> +		if (xcl == cl) {
> +			list_del_init(&cl->xcl_list);
> +			goto out;
> +		}
> +	}
> +	res = -ENOENT;

Again, nobody checks this error return, and it seems like an unlikely
programmer error anyway. If we really need the check I'd be inclined
just to ditche the for loop and BUG_ON(list_empty(&cl->xcl_list)), which
will catch double-unregistrations (thanks to the list_del_init()).

> + out:
> +	spin_unlock(&svc_xprt_class_lock);
> +	return res;
> +}
> +EXPORT_SYMBOL_GPL(svc_unreg_xprt_class);
> +
> +/*
> + * Called by transport drivers to initialize the transport independent
> + * portion of the transport instance.
> + */
> +void svc_xprt_init(struct svc_xprt_class *xcl, struct svc_xprt *xprt)
> +{
> +	memset(xprt, 0, sizeof(*xprt));
> +	xprt->xpt_class = xcl;
> +	xprt->xpt_ops = xcl->xcl_ops;
> +}
> +EXPORT_SYMBOL_GPL(svc_xprt_init);

Seems fine otherwise.

--b.

Re: [PATCH 01/38] svc: Add an svc transport class

[Tom Tucker]

On 12/13/07 12:45 PM, "J. Bruce Fields" <bfields@fieldses.org> wrote:

> Sorry for joining in a little late....
> 
> On Tue, Dec 11, 2007 at 05:31:54PM -0600, Tom Tucker wrote:
>> +int svc_reg_xprt_class(struct svc_xprt_class *xcl)
> 
> None of the callers appear to check the return value, so this should
> probably be a void return.
>

I don't feel strongly about this, but what I was trying to catch was two
different transports accidentally colliding on the same name. I doubt there
will be a run on new transports, but it would at least fail the module load
with a reasonable error.

>> +{
>> + struct svc_xprt_class *cl;
>> + int res = -EEXIST;
>> +
>> + dprintk("svc: Adding svc transport class '%s'\n",
>> +  xcl->xcl_name);
>> +
>> + INIT_LIST_HEAD(&xcl->xcl_list);
> 
> Unless we care how xcl_list is set in the error case, this is
> unnecessary.
>

I'll leave this and clean up the un-register path as you suggest below.

 
>> + spin_lock(&svc_xprt_class_lock);
>> + list_for_each_entry(cl, &svc_xprt_class_list, xcl_list) {
>> +  if (xcl == cl)
>> +   goto out;
> 
> Is this even worth checking?  Accidentally registering the identical
> struct svc_xprt_class * seems an unlikely mistake for a transport-class
> coder to make, given that they only need call this function once per
> transport, in the module initialization.  Maybe make this a BUG()?  Or
> check for duplicate xcl_name's if that seems a more likely error?

What I really wanted to check was name == name, not necessarily ptr == ptr.
I should fix this to fail if strcmp(xcl->name, cl->name)==0.

> 
>> + }
>> + list_add_tail(&xcl->xcl_list, &svc_xprt_class_list);
>> + res = 0;
>> +out:
>> + spin_unlock(&svc_xprt_class_lock);
>> + return res;
>> +}
>> +EXPORT_SYMBOL_GPL(svc_reg_xprt_class);
>> +
>> +int svc_unreg_xprt_class(struct svc_xprt_class *xcl)
>> +{
>> + struct svc_xprt_class *cl;
>> + int res = 0;
>> +
>> + dprintk("svc: Removing svc transport class '%s'\n", xcl->xcl_name);
>> +
>> + spin_lock(&svc_xprt_class_lock);
>> + list_for_each_entry(cl, &svc_xprt_class_list, xcl_list) {
>> +  if (xcl == cl) {
>> +   list_del_init(&cl->xcl_list);
>> +   goto out;
>> +  }
>> + }
>> + res = -ENOENT;
> 
> Again, nobody checks this error return, and it seems like an unlikely
> programmer error anyway. If we really need the check I'd be inclined
> just to ditche the for loop and BUG_ON(list_empty(&cl->xcl_list)), which
> will catch double-unregistrations (thanks to the list_del_init()).
>

Yeah, the loop is dumb. I agree with your error assessment, how about if I
change the return type to void and list_del_init(...) sans loop.


>> + out:
>> + spin_unlock(&svc_xprt_class_lock);
>> + return res;
>> +}
>> +EXPORT_SYMBOL_GPL(svc_unreg_xprt_class);
>> +
>> +/*
>> + * Called by transport drivers to initialize the transport independent
>> + * portion of the transport instance.
>> + */
>> +void svc_xprt_init(struct svc_xprt_class *xcl, struct svc_xprt *xprt)
>> +{
>> + memset(xprt, 0, sizeof(*xprt));
>> + xprt->xpt_class = xcl;
>> + xprt->xpt_ops = xcl->xcl_ops;
>> +}
>> +EXPORT_SYMBOL_GPL(svc_xprt_init);
> 
> Seems fine otherwise.
> 
> --b.

Re: [PATCH 01/38] svc: Add an svc transport class

[J. Bruce Fields]

On Mon, Dec 17, 2007 at 03:40:41AM -0600, Tom Tucker wrote:
> 
> 
> 
> On 12/13/07 12:45 PM, "J. Bruce Fields" <bfields@fieldses.org> wrote:
> 
> > Sorry for joining in a little late....
> > 
> > On Tue, Dec 11, 2007 at 05:31:54PM -0600, Tom Tucker wrote:
> >> +int svc_reg_xprt_class(struct svc_xprt_class *xcl)
> > 
> > None of the callers appear to check the return value, so this should
> > probably be a void return.
> >
> 
> I don't feel strongly about this, but what I was trying to catch was two
> different transports accidentally colliding on the same name. I doubt there
> will be a run on new transports, but it would at least fail the module load
> with a reasonable error.

Sounds fine.

Thanks!--b.

Re: [PATCH 01/38] svc: Add an svc transport class

[Tom Tucker]

On Mon, 2007-12-17 at 12:45 -0500, J. Bruce Fields wrote:
> On Mon, Dec 17, 2007 at 03:40:41AM -0600, Tom Tucker wrote:
> > 
> > 
> > 
> > On 12/13/07 12:45 PM, "J. Bruce Fields" <bfields@fieldses.org> wrote:
> > 
> > > Sorry for joining in a little late....
> > > 
> > > On Tue, Dec 11, 2007 at 05:31:54PM -0600, Tom Tucker wrote:
> > >> +int svc_reg_xprt_class(struct svc_xprt_class *xcl)
> > > 
> > > None of the callers appear to check the return value, so this should
> > > probably be a void return.
> > >
> > 
> > I don't feel strongly about this, but what I was trying to catch was two
> > different transports accidentally colliding on the same name. I doubt there
> > will be a run on new transports, but it would at least fail the module load
> > with a reasonable error.
> 
> Sounds fine.

Great, done. 

I've coded all of the changes that you suggested and built a new
patchset. Besides the tests I outlined in my "test matrix" are there any
other tests you would like to see performed?

I could not reproduce the BUG_ON from Bull. I'm wondering if it's a PPC
issue :-\ 

After the testing, I'll publish the revised patchset.

> 
> Thanks!--b.

Re: [PATCH 01/38] svc: Add an svc transport class

[J. Bruce Fields]

On Mon, Dec 17, 2007 at 06:26:24PM -0600, Tom Tucker wrote:
> I've coded all of the changes that you suggested and built a new
> patchset. Besides the tests I outlined in my "test matrix" are there any
> other tests you would like to see performed?

krb5p testing might still be interesting.

> 
> I could not reproduce the BUG_ON from Bull. I'm wondering if it's a PPC
> issue :-\ 

Hm, too bad, I'd like to understand that one.

> 
> After the testing, I'll publish the revised patchset.
> 

OK, thanks.--b.

Re: [PATCH 01/38] svc: Add an svc transport class

[J. Bruce Fields]

By the way, I'm leaving town tommorow, on vacation till the end of the
month.  I might still be online occasionally, but it's hard to predict.

Thanks to everyone for a good 2007, and best wishes for the new year!

--b.

Re: [PATCH 01/38] svc: Add an svc transport class

[Tom Tucker]

On Tue, 2007-12-18 at 09:54 -0500, J. Bruce Fields wrote:
> On Mon, Dec 17, 2007 at 06:26:24PM -0600, Tom Tucker wrote:
> > I've coded all of the changes that you suggested and built a new
> > patchset. Besides the tests I outlined in my "test matrix" are there any
> > other tests you would like to see performed?
> 
> krb5p testing might still be interesting.
> 

I'll add it to my matrix. It will take some time to set up though since
I don't have a KDC or anything set up. Are we under any time
constraints?

> > 
> > I could not reproduce the BUG_ON from Bull. I'm wondering if it's a PPC
> > issue :-\ 
> 
> Hm, too bad, I'd like to understand that one.
> 

Yeah, and I'm having a hard time believing it's PPC related. I think I'm
missing something, I'll keep looking.

> > 
> > After the testing, I'll publish the revised patchset.
> > 
> 
> OK, thanks.--b.

Re: [PATCH 01/38] svc: Add an svc transport class

[J. Bruce Fields]

On Tue, Dec 18, 2007 at 09:04:16AM -0600, Tom Tucker wrote:
> 
> On Tue, 2007-12-18 at 09:54 -0500, J. Bruce Fields wrote:
> > On Mon, Dec 17, 2007 at 06:26:24PM -0600, Tom Tucker wrote:
> > > I've coded all of the changes that you suggested and built a new
> > > patchset. Besides the tests I outlined in my "test matrix" are there any
> > > other tests you would like to see performed?
> > 
> > krb5p testing might still be interesting.
> > 
> 
> I'll add it to my matrix. It will take some time to set up though since
> I don't have a KDC or anything set up. Are we under any time
> constraints?

For krb5p, I don't think it's terifically urgent.

> > > I could not reproduce the BUG_ON from Bull. I'm wondering if it's a PPC
> > > issue :-\ 
> > 
> > Hm, too bad, I'd like to understand that one.
> > 
> 
> Yeah, and I'm having a hard time believing it's PPC related. I think I'm
> missing something, I'll keep looking.

Thanks!

--b.

Re: [PATCH 01/38] svc: Add an svc transport class

[Tom Tucker]

On Tue, 2007-12-18 at 09:04 -0600, Tom Tucker wrote:
> On Tue, 2007-12-18 at 09:54 -0500, J. Bruce Fields wrote:
> > On Mon, Dec 17, 2007 at 06:26:24PM -0600, Tom Tucker wrote:
> > > I've coded all of the changes that you suggested and built a new
> > > patchset. Besides the tests I outlined in my "test matrix" are there any
> > > other tests you would like to see performed?
> > 
> > krb5p testing might still be interesting.
> > 
> 
> I'll add it to my matrix. It will take some time to set up though since
> I don't have a KDC or anything set up. Are we under any time
> constraints?
> 
> > > 
> > > I could not reproduce the BUG_ON from Bull. I'm wondering if it's a PPC
> > > issue :-\ 
> > 
> > Hm, too bad, I'd like to understand that one.
> > 

FYI, I have now reproduced this bug on an AMD64 box with the
CITI_NFS4_ALL-1 tree. 

I am regression testing it on the new build have been running it about
2hrs without failure.

> 
> Yeah, and I'm having a hard time believing it's PPC related. I think I'm
> missing something, I'll keep looking.
> 
> > > 
> > > After the testing, I'll publish the revised patchset.
> > > 
> > 
> > OK, thanks.--b.
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 01/38] svc: Add an svc transport class

[J. Bruce Fields]

On Tue, Dec 18, 2007 at 01:30:38PM -0600, Tom Tucker wrote:
> 
> On Tue, 2007-12-18 at 09:04 -0600, Tom Tucker wrote:
> > On Tue, 2007-12-18 at 09:54 -0500, J. Bruce Fields wrote:
> > > On Mon, Dec 17, 2007 at 06:26:24PM -0600, Tom Tucker wrote:
> > > > I've coded all of the changes that you suggested and built a new
> > > > patchset. Besides the tests I outlined in my "test matrix" are there any
> > > > other tests you would like to see performed?
> > > 
> > > krb5p testing might still be interesting.
> > > 
> > 
> > I'll add it to my matrix. It will take some time to set up though since
> > I don't have a KDC or anything set up. Are we under any time
> > constraints?
> > 
> > > > 
> > > > I could not reproduce the BUG_ON from Bull. I'm wondering if it's a PPC
> > > > issue :-\ 
> > > 
> > > Hm, too bad, I'd like to understand that one.
> > > 
> 
> FYI, I have now reproduced this bug on an AMD64 box with the
> CITI_NFS4_ALL-1 tree. 
> 
> I am regression testing it on the new build have been running it about
> 2hrs without failure.

OK, so probably either something got fixed in the server-xprt-switch
patches since then, or something got fixed in mainline, or the problem
was in one of the other CITI_NFS4_ALL patches.  Hm.

--b.

[PATCH 02/38] svc: Make svc_sock the tcp/udp transport

[Tom Tucker]

Make TCP and UDP svc_sock transports, and register them
with the svc transport core. 

A transport type (svc_sock) has an svc_xprt as its first member, 
and calls svc_xprt_init to initialize this field.

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 include/linux/sunrpc/debug.h   |    1 -
 include/linux/sunrpc/svcsock.h |    4 ++++
 net/sunrpc/sunrpc_syms.c       |    4 +++-
 net/sunrpc/svcsock.c           |   33 ++++++++++++++++++++++++++++++++-
 4 files changed, 39 insertions(+), 3 deletions(-)

diff --git a/include/linux/sunrpc/debug.h b/include/linux/sunrpc/debug.h
index 092fcfa..10709cb 100644
--- a/include/linux/sunrpc/debug.h
+++ b/include/linux/sunrpc/debug.h
@@ -20,7 +20,6 @@
 #define RPCDBG_BIND		0x0020
 #define RPCDBG_SCHED		0x0040
 #define RPCDBG_TRANS		0x0080
-#define RPCDBG_SVCSOCK		0x0100
 #define RPCDBG_SVCXPRT		0x0100
 #define RPCDBG_SVCDSP		0x0200
 #define RPCDBG_MISC		0x0400
diff --git a/include/linux/sunrpc/svcsock.h b/include/linux/sunrpc/svcsock.h
index a53e0fa..1878cbe 100644
--- a/include/linux/sunrpc/svcsock.h
+++ b/include/linux/sunrpc/svcsock.h
@@ -10,11 +10,13 @@
 #define SUNRPC_SVCSOCK_H
 
 #include <linux/sunrpc/svc.h>
+#include <linux/sunrpc/svc_xprt.h>
 
 /*
  * RPC server socket.
  */
 struct svc_sock {
+	struct svc_xprt		sk_xprt;
 	struct list_head	sk_ready;	/* list of ready sockets */
 	struct list_head	sk_list;	/* list of all sockets */
 	struct socket *		sk_sock;	/* berkeley socket layer */
@@ -78,6 +80,8 @@ int		svc_addsock(struct svc_serv *serv,
 			    int fd,
 			    char *name_return,
 			    int *proto);
+void		svc_init_xprt_sock(void);
+void		svc_cleanup_xprt_sock(void);
 
 /*
  * svc_makesock socket characteristics
diff --git a/net/sunrpc/sunrpc_syms.c b/net/sunrpc/sunrpc_syms.c
index 33d89e8..79ea05f 100644
--- a/net/sunrpc/sunrpc_syms.c
+++ b/net/sunrpc/sunrpc_syms.c
@@ -151,7 +151,8 @@ init_sunrpc(void)
 #endif
 	cache_register(&ip_map_cache);
 	cache_register(&unix_gid_cache);
-	init_socket_xprt();
+	svc_init_xprt_sock();	/* svc sock transport */
+	init_socket_xprt();	/* clnt sock transport */
 	rpcauth_init_module();
 out:
 	return err;
@@ -162,6 +163,7 @@ cleanup_sunrpc(void)
 {
 	rpcauth_remove_module();
 	cleanup_socket_xprt();
+	svc_cleanup_xprt_sock();
 	unregister_rpc_pipefs();
 	rpc_destroy_mempool();
 	if (cache_unregister(&ip_map_cache))
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index c75bffe..4755467 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -75,7 +75,7 @@
  *
  */
 
-#define RPCDBG_FACILITY	RPCDBG_SVCSOCK
+#define RPCDBG_FACILITY	RPCDBG_SVCXPRT
 
 
 static struct svc_sock *svc_setup_socket(struct svc_serv *, struct socket *,
@@ -900,12 +900,21 @@ svc_udp_sendto(struct svc_rqst *rqstp)
 	return error;
 }
 
+static struct svc_xprt_ops svc_udp_ops = {
+};
+
+static struct svc_xprt_class svc_udp_class = {
+	.xcl_name = "udp",
+	.xcl_ops = &svc_udp_ops,
+};
+
 static void
 svc_udp_init(struct svc_sock *svsk)
 {
 	int one = 1;
 	mm_segment_t oldfs;
 
+	svc_xprt_init(&svc_udp_class, &svsk->sk_xprt);
 	svsk->sk_sk->sk_data_ready = svc_udp_data_ready;
 	svsk->sk_sk->sk_write_space = svc_write_space;
 	svsk->sk_recvfrom = svc_udp_recvfrom;
@@ -1344,12 +1353,33 @@ svc_tcp_sendto(struct svc_rqst *rqstp)
 	return sent;
 }
 
+static struct svc_xprt_ops svc_tcp_ops = {
+};
+
+static struct svc_xprt_class svc_tcp_class = {
+	.xcl_name = "tcp",
+	.xcl_ops = &svc_tcp_ops,
+};
+
+void svc_init_xprt_sock(void)
+{
+	svc_reg_xprt_class(&svc_tcp_class);
+	svc_reg_xprt_class(&svc_udp_class);
+}
+
+void svc_cleanup_xprt_sock(void)
+{
+	svc_unreg_xprt_class(&svc_tcp_class);
+	svc_unreg_xprt_class(&svc_udp_class);
+}
+
 static void
 svc_tcp_init(struct svc_sock *svsk)
 {
 	struct sock	*sk = svsk->sk_sk;
 	struct tcp_sock *tp = tcp_sk(sk);
 
+	svc_xprt_init(&svc_tcp_class, &svsk->sk_xprt);
 	svsk->sk_recvfrom = svc_tcp_recvfrom;
 	svsk->sk_sendto = svc_tcp_sendto;
 
@@ -1965,3 +1995,4 @@ static struct svc_deferred_req *svc_deferred_dequeue(struct svc_sock *svsk)
 	spin_unlock(&svsk->sk_lock);
 	return dr;
 }
+

Re: [PATCH 02/38] svc: Make svc_sock the tcp/udp transport

[J. Bruce Fields]

On Tue, Dec 11, 2007 at 05:31:56PM -0600, Tom Tucker wrote:
> @@ -1965,3 +1995,4 @@ static struct svc_deferred_req *svc_deferred_dequeue(struct svc_sock *svsk)
>  	spin_unlock(&svsk->sk_lock);
>  	return dr;
>  }
> +

??

--b.

Re: [PATCH 02/38] svc: Make svc_sock the tcp/udp transport

[Tom Tucker]

On 12/13/07 1:01 PM, "J. Bruce Fields" <bfields@fieldses.org> wrote:

> On Tue, Dec 11, 2007 at 05:31:56PM -0600, Tom Tucker wrote:
>> @@ -1965,3 +1995,4 @@ static struct svc_deferred_req
>> *svc_deferred_dequeue(struct svc_sock *svsk)
>> spin_unlock(&svsk->sk_lock);
>> return dr;
>>  }
>> +
> 
> ??

Oops. Fixed.

> 
> --b.

[PATCH 03/38] svc: Change the svc_sock in the rqstp structure to a transport

[Tom Tucker]

The rqstp structure contains a pointer to the transport for the 
RPC request. This functionaly trivial patch adds an unamed union 
with pointers to both svc_sock and svc_xprt. Ultimately the 
union will be removed and only the rq_xprt field will remain. This 
allows incrementally extracting transport independent interfaces without 
one gigundo patch.

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 include/linux/sunrpc/svc.h |    5 ++++-
 1 files changed, 4 insertions(+), 1 deletions(-)

diff --git a/include/linux/sunrpc/svc.h b/include/linux/sunrpc/svc.h
index 8531a70..37f7448 100644
--- a/include/linux/sunrpc/svc.h
+++ b/include/linux/sunrpc/svc.h
@@ -204,7 +204,10 @@ union svc_addr_u {
 struct svc_rqst {
 	struct list_head	rq_list;	/* idle list */
 	struct list_head	rq_all;		/* all threads list */
-	struct svc_sock *	rq_sock;	/* socket */
+	union {
+		struct svc_xprt *	rq_xprt;	/* transport ptr */
+		struct svc_sock *	rq_sock; 	/* socket ptr */
+	};
 	struct sockaddr_storage	rq_addr;	/* peer address */
 	size_t			rq_addrlen;
 

[PATCH 04/38] svc: Add a max payload value to the transport

[Tom Tucker]

The svc_max_payload function currently looks at the socket type
to determine the max payload. Add a max payload value to svc_xprt_class
so it can be returned directly. 

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 include/linux/sunrpc/svc_xprt.h |    1 +
 net/sunrpc/svc.c                |    4 +---
 net/sunrpc/svcsock.c            |    2 ++
 3 files changed, 4 insertions(+), 3 deletions(-)

diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
index a8b1da8..b4ce054 100644
--- a/include/linux/sunrpc/svc_xprt.h
+++ b/include/linux/sunrpc/svc_xprt.h
@@ -17,6 +17,7 @@ struct svc_xprt_class {
 	struct module		*xcl_owner;
 	struct svc_xprt_ops	*xcl_ops;
 	struct list_head	xcl_list;
+	u32			xcl_max_payload;
 };
 
 struct svc_xprt {
diff --git a/net/sunrpc/svc.c b/net/sunrpc/svc.c
index a4a6bf7..ea3fa86 100644
--- a/net/sunrpc/svc.c
+++ b/net/sunrpc/svc.c
@@ -1054,10 +1054,8 @@ err_bad:
  */
 u32 svc_max_payload(const struct svc_rqst *rqstp)
 {
-	int max = RPCSVC_MAXPAYLOAD_TCP;
+	u32 max = rqstp->rq_xprt->xpt_class->xcl_max_payload;
 
-	if (rqstp->rq_sock->sk_sock->type == SOCK_DGRAM)
-		max = RPCSVC_MAXPAYLOAD_UDP;
 	if (rqstp->rq_server->sv_max_payload < max)
 		max = rqstp->rq_server->sv_max_payload;
 	return max;
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index 4755467..ca9b8d8 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -906,6 +906,7 @@ static struct svc_xprt_ops svc_udp_ops = {
 static struct svc_xprt_class svc_udp_class = {
 	.xcl_name = "udp",
 	.xcl_ops = &svc_udp_ops,
+	.xcl_max_payload = RPCSVC_MAXPAYLOAD_UDP,
 };
 
 static void
@@ -1359,6 +1360,7 @@ static struct svc_xprt_ops svc_tcp_ops = {
 static struct svc_xprt_class svc_tcp_class = {
 	.xcl_name = "tcp",
 	.xcl_ops = &svc_tcp_ops,
+	.xcl_max_payload = RPCSVC_MAXPAYLOAD_TCP,
 };
 
 void svc_init_xprt_sock(void)

[PATCH 05/38] svc: Move sk_sendto and sk_recvfrom to svc_xprt_class

[Tom Tucker]

The sk_sendto and sk_recvfrom are function pointers that allow svc_sock
to be used for both UDP and TCP. Move these function pointers to the 
svc_xprt_ops structure.

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 include/linux/sunrpc/svc_xprt.h |    2 ++
 include/linux/sunrpc/svcsock.h  |    3 ---
 net/sunrpc/svcsock.c            |   12 ++++++------
 3 files changed, 8 insertions(+), 9 deletions(-)

diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
index b4ce054..81daa39 100644
--- a/include/linux/sunrpc/svc_xprt.h
+++ b/include/linux/sunrpc/svc_xprt.h
@@ -10,6 +10,8 @@
 #include <linux/sunrpc/svc.h>
 
 struct svc_xprt_ops {
+	int		(*xpo_recvfrom)(struct svc_rqst *);
+	int		(*xpo_sendto)(struct svc_rqst *);
 };
 
 struct svc_xprt_class {
diff --git a/include/linux/sunrpc/svcsock.h b/include/linux/sunrpc/svcsock.h
index 1878cbe..08e78d0 100644
--- a/include/linux/sunrpc/svcsock.h
+++ b/include/linux/sunrpc/svcsock.h
@@ -45,9 +45,6 @@ struct svc_sock {
 						 * be revisted */
 	struct mutex		sk_mutex;	/* to serialize sending data */
 
-	int			(*sk_recvfrom)(struct svc_rqst *rqstp);
-	int			(*sk_sendto)(struct svc_rqst *rqstp);
-
 	/* We keep the old state_change and data_ready CB's here */
 	void			(*sk_ostate)(struct sock *);
 	void			(*sk_odata)(struct sock *, int bytes);
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index ca9b8d8..9c06b15 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -901,6 +901,8 @@ svc_udp_sendto(struct svc_rqst *rqstp)
 }
 
 static struct svc_xprt_ops svc_udp_ops = {
+	.xpo_recvfrom = svc_udp_recvfrom,
+	.xpo_sendto = svc_udp_sendto,
 };
 
 static struct svc_xprt_class svc_udp_class = {
@@ -918,8 +920,6 @@ svc_udp_init(struct svc_sock *svsk)
 	svc_xprt_init(&svc_udp_class, &svsk->sk_xprt);
 	svsk->sk_sk->sk_data_ready = svc_udp_data_ready;
 	svsk->sk_sk->sk_write_space = svc_write_space;
-	svsk->sk_recvfrom = svc_udp_recvfrom;
-	svsk->sk_sendto = svc_udp_sendto;
 
 	/* initialise setting must have enough space to
 	 * receive and respond to one request.
@@ -1355,6 +1355,8 @@ svc_tcp_sendto(struct svc_rqst *rqstp)
 }
 
 static struct svc_xprt_ops svc_tcp_ops = {
+	.xpo_recvfrom = svc_tcp_recvfrom,
+	.xpo_sendto = svc_tcp_sendto,
 };
 
 static struct svc_xprt_class svc_tcp_class = {
@@ -1382,8 +1384,6 @@ svc_tcp_init(struct svc_sock *svsk)
 	struct tcp_sock *tp = tcp_sk(sk);
 
 	svc_xprt_init(&svc_tcp_class, &svsk->sk_xprt);
-	svsk->sk_recvfrom = svc_tcp_recvfrom;
-	svsk->sk_sendto = svc_tcp_sendto;
 
 	if (sk->sk_state == TCP_LISTEN) {
 		dprintk("setting up TCP socket for listening\n");
@@ -1531,7 +1531,7 @@ svc_recv(struct svc_rqst *rqstp, long timeout)
 
 	dprintk("svc: server %p, pool %u, socket %p, inuse=%d\n",
 		 rqstp, pool->sp_id, svsk, atomic_read(&svsk->sk_inuse));
-	len = svsk->sk_recvfrom(rqstp);
+	len = svsk->sk_xprt.xpt_ops->xpo_recvfrom(rqstp);
 	dprintk("svc: got len=%d\n", len);
 
 	/* No data, incomplete (TCP) read, or accept() */
@@ -1591,7 +1591,7 @@ svc_send(struct svc_rqst *rqstp)
 	if (test_bit(SK_DEAD, &svsk->sk_flags))
 		len = -ENOTCONN;
 	else
-		len = svsk->sk_sendto(rqstp);
+		len = svsk->sk_xprt.xpt_ops->xpo_sendto(rqstp);
 	mutex_unlock(&svsk->sk_mutex);
 	svc_sock_release(rqstp);
 

[PATCH 06/38] svc: Add transport specific xpo_release function

[Tom Tucker]

The svc_sock_release function releases pages allocated to a thread. For
UDP, this also returns the receive skb to the stack. For RDMA it will 
post a receive WR and bump the client credit count. 

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 include/linux/sunrpc/svc.h      |    2 +-
 include/linux/sunrpc/svc_xprt.h |    1 +
 net/sunrpc/svcsock.c            |   17 +++++++++--------
 3 files changed, 11 insertions(+), 9 deletions(-)

diff --git a/include/linux/sunrpc/svc.h b/include/linux/sunrpc/svc.h
index 37f7448..cfb2652 100644
--- a/include/linux/sunrpc/svc.h
+++ b/include/linux/sunrpc/svc.h
@@ -217,7 +217,7 @@ struct svc_rqst {
 	struct auth_ops *	rq_authop;	/* authentication flavour */
 	u32			rq_flavor;	/* pseudoflavor */
 	struct svc_cred		rq_cred;	/* auth info */
-	struct sk_buff *	rq_skbuff;	/* fast recv inet buffer */
+	void *			rq_xprt_ctxt;	/* transport specific context ptr */
 	struct svc_deferred_req*rq_deferred;	/* deferred request we are replaying */
 
 	struct xdr_buf		rq_arg;
diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
index 81daa39..e3bd7b1 100644
--- a/include/linux/sunrpc/svc_xprt.h
+++ b/include/linux/sunrpc/svc_xprt.h
@@ -12,6 +12,7 @@
 struct svc_xprt_ops {
 	int		(*xpo_recvfrom)(struct svc_rqst *);
 	int		(*xpo_sendto)(struct svc_rqst *);
+	void		(*xpo_release_rqst)(struct svc_rqst *);
 };
 
 struct svc_xprt_class {
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index 9c06b15..b24c084 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -185,14 +185,13 @@ svc_thread_dequeue(struct svc_pool *pool, struct svc_rqst *rqstp)
 /*
  * Release an skbuff after use
  */
-static inline void
-svc_release_skb(struct svc_rqst *rqstp)
+static void svc_release_skb(struct svc_rqst *rqstp)
 {
-	struct sk_buff *skb = rqstp->rq_skbuff;
+	struct sk_buff *skb = rqstp->rq_xprt_ctxt;
 	struct svc_deferred_req *dr = rqstp->rq_deferred;
 
 	if (skb) {
-		rqstp->rq_skbuff = NULL;
+		rqstp->rq_xprt_ctxt = NULL;
 
 		dprintk("svc: service %p, releasing skb %p\n", rqstp, skb);
 		skb_free_datagram(rqstp->rq_sock->sk_sk, skb);
@@ -395,7 +394,7 @@ svc_sock_release(struct svc_rqst *rqstp)
 {
 	struct svc_sock	*svsk = rqstp->rq_sock;
 
-	svc_release_skb(rqstp);
+	rqstp->rq_xprt->xpt_ops->xpo_release_rqst(rqstp);
 
 	svc_free_res_pages(rqstp);
 	rqstp->rq_res.page_len = 0;
@@ -867,7 +866,7 @@ svc_udp_recvfrom(struct svc_rqst *rqstp)
 			skb_free_datagram(svsk->sk_sk, skb);
 			return 0;
 		}
-		rqstp->rq_skbuff = skb;
+		rqstp->rq_xprt_ctxt = skb;
 	}
 
 	rqstp->rq_arg.page_base = 0;
@@ -903,6 +902,7 @@ svc_udp_sendto(struct svc_rqst *rqstp)
 static struct svc_xprt_ops svc_udp_ops = {
 	.xpo_recvfrom = svc_udp_recvfrom,
 	.xpo_sendto = svc_udp_sendto,
+	.xpo_release_rqst = svc_release_skb,
 };
 
 static struct svc_xprt_class svc_udp_class = {
@@ -1291,7 +1291,7 @@ svc_tcp_recvfrom(struct svc_rqst *rqstp)
 		rqstp->rq_arg.page_len = len - rqstp->rq_arg.head[0].iov_len;
 	}
 
-	rqstp->rq_skbuff      = NULL;
+	rqstp->rq_xprt_ctxt   = NULL;
 	rqstp->rq_prot	      = IPPROTO_TCP;
 
 	/* Reset TCP read info */
@@ -1357,6 +1357,7 @@ svc_tcp_sendto(struct svc_rqst *rqstp)
 static struct svc_xprt_ops svc_tcp_ops = {
 	.xpo_recvfrom = svc_tcp_recvfrom,
 	.xpo_sendto = svc_tcp_sendto,
+	.xpo_release_rqst = svc_release_skb,
 };
 
 static struct svc_xprt_class svc_tcp_class = {
@@ -1578,7 +1579,7 @@ svc_send(struct svc_rqst *rqstp)
 	}
 
 	/* release the receive skb before sending the reply */
-	svc_release_skb(rqstp);
+	rqstp->rq_xprt->xpt_ops->xpo_release_rqst(rqstp);
 
 	/* calculate over-all length */
 	xb = & rqstp->rq_res;

Re: [PATCH 06/38] svc: Add transport specific xpo_release function

[J. Bruce Fields]

On Tue, Dec 11, 2007 at 05:32:06PM -0600, Tom Tucker wrote:
> 
> The svc_sock_release function releases pages allocated to a thread. For
> UDP, this also returns the receive skb to the stack.

The stack?  How about just "frees the receive skb"?

> For RDMA it will 
> post a receive WR and bump the client credit count. 
> 
> Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
> ---
> 
>  include/linux/sunrpc/svc.h      |    2 +-
>  include/linux/sunrpc/svc_xprt.h |    1 +
>  net/sunrpc/svcsock.c            |   17 +++++++++--------
>  3 files changed, 11 insertions(+), 9 deletions(-)
> 
> diff --git a/include/linux/sunrpc/svc.h b/include/linux/sunrpc/svc.h
> index 37f7448..cfb2652 100644
> --- a/include/linux/sunrpc/svc.h
> +++ b/include/linux/sunrpc/svc.h
> @@ -217,7 +217,7 @@ struct svc_rqst {
>  	struct auth_ops *	rq_authop;	/* authentication flavour */
>  	u32			rq_flavor;	/* pseudoflavor */
>  	struct svc_cred		rq_cred;	/* auth info */
> -	struct sk_buff *	rq_skbuff;	/* fast recv inet buffer */
> +	void *			rq_xprt_ctxt;	/* transport specific context ptr */
>  	struct svc_deferred_req*rq_deferred;	/* deferred request we are replaying */
>  
>  	struct xdr_buf		rq_arg;
> diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
> index 81daa39..e3bd7b1 100644
> --- a/include/linux/sunrpc/svc_xprt.h
> +++ b/include/linux/sunrpc/svc_xprt.h
> @@ -12,6 +12,7 @@
>  struct svc_xprt_ops {
>  	int		(*xpo_recvfrom)(struct svc_rqst *);
>  	int		(*xpo_sendto)(struct svc_rqst *);
> +	void		(*xpo_release_rqst)(struct svc_rqst *);
>  };
>  
>  struct svc_xprt_class {
> diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
> index 9c06b15..b24c084 100644
> --- a/net/sunrpc/svcsock.c
> +++ b/net/sunrpc/svcsock.c
> @@ -185,14 +185,13 @@ svc_thread_dequeue(struct svc_pool *pool, struct svc_rqst *rqstp)
>  /*
>   * Release an skbuff after use
>   */
> -static inline void
> -svc_release_skb(struct svc_rqst *rqstp)
> +static void svc_release_skb(struct svc_rqst *rqstp)
>  {
> -	struct sk_buff *skb = rqstp->rq_skbuff;
> +	struct sk_buff *skb = rqstp->rq_xprt_ctxt;
>  	struct svc_deferred_req *dr = rqstp->rq_deferred;
>  
>  	if (skb) {
> -		rqstp->rq_skbuff = NULL;
> +		rqstp->rq_xprt_ctxt = NULL;
>  
>  		dprintk("svc: service %p, releasing skb %p\n", rqstp, skb);
>  		skb_free_datagram(rqstp->rq_sock->sk_sk, skb);
> @@ -395,7 +394,7 @@ svc_sock_release(struct svc_rqst *rqstp)
>  {
>  	struct svc_sock	*svsk = rqstp->rq_sock;
>  
> -	svc_release_skb(rqstp);
> +	rqstp->rq_xprt->xpt_ops->xpo_release_rqst(rqstp);
>  
>  	svc_free_res_pages(rqstp);
>  	rqstp->rq_res.page_len = 0;
> @@ -867,7 +866,7 @@ svc_udp_recvfrom(struct svc_rqst *rqstp)
>  			skb_free_datagram(svsk->sk_sk, skb);
>  			return 0;
>  		}
> -		rqstp->rq_skbuff = skb;
> +		rqstp->rq_xprt_ctxt = skb;
>  	}
>  
>  	rqstp->rq_arg.page_base = 0;
> @@ -903,6 +902,7 @@ svc_udp_sendto(struct svc_rqst *rqstp)
>  static struct svc_xprt_ops svc_udp_ops = {
>  	.xpo_recvfrom = svc_udp_recvfrom,
>  	.xpo_sendto = svc_udp_sendto,
> +	.xpo_release_rqst = svc_release_skb,
>  };
>  
>  static struct svc_xprt_class svc_udp_class = {
> @@ -1291,7 +1291,7 @@ svc_tcp_recvfrom(struct svc_rqst *rqstp)
>  		rqstp->rq_arg.page_len = len - rqstp->rq_arg.head[0].iov_len;
>  	}
>  
> -	rqstp->rq_skbuff      = NULL;
> +	rqstp->rq_xprt_ctxt   = NULL;
>  	rqstp->rq_prot	      = IPPROTO_TCP;
>  
>  	/* Reset TCP read info */
> @@ -1357,6 +1357,7 @@ svc_tcp_sendto(struct svc_rqst *rqstp)
>  static struct svc_xprt_ops svc_tcp_ops = {
>  	.xpo_recvfrom = svc_tcp_recvfrom,
>  	.xpo_sendto = svc_tcp_sendto,
> +	.xpo_release_rqst = svc_release_skb,
>  };
>  
>  static struct svc_xprt_class svc_tcp_class = {
> @@ -1578,7 +1579,7 @@ svc_send(struct svc_rqst *rqstp)
>  	}
>  
>  	/* release the receive skb before sending the reply */
> -	svc_release_skb(rqstp);
> +	rqstp->rq_xprt->xpt_ops->xpo_release_rqst(rqstp);
>  
>  	/* calculate over-all length */
>  	xb = & rqstp->rq_res;

Re: [PATCH 06/38] svc: Add transport specific xpo_release function

[Tom Tucker]

On 12/13/07 1:31 PM, "J. Bruce Fields" <bfields@fieldses.org> wrote:

> On Tue, Dec 11, 2007 at 05:32:06PM -0600, Tom Tucker wrote:
>> 
>> The svc_sock_release function releases pages allocated to a thread. For
>> UDP, this also returns the receive skb to the stack.
> 
> The stack?  How about just "frees the receive skb"?
>

Ok.
 
>> For RDMA it will
>> post a receive WR and bump the client credit count.
>> 
>> Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
>> ---
>> 
>>  include/linux/sunrpc/svc.h      |    2 +-
>>  include/linux/sunrpc/svc_xprt.h |    1 +
>>  net/sunrpc/svcsock.c            |   17 +++++++++--------
>>  3 files changed, 11 insertions(+), 9 deletions(-)
>> 
>> diff --git a/include/linux/sunrpc/svc.h b/include/linux/sunrpc/svc.h
>> index 37f7448..cfb2652 100644
>> --- a/include/linux/sunrpc/svc.h
>> +++ b/include/linux/sunrpc/svc.h
>> @@ -217,7 +217,7 @@ struct svc_rqst {
>> struct auth_ops * rq_authop; /* authentication flavour */
>> u32   rq_flavor; /* pseudoflavor */
>> struct svc_cred  rq_cred; /* auth info */
>> - struct sk_buff * rq_skbuff; /* fast recv inet buffer */
>> + void *   rq_xprt_ctxt; /* transport specific context ptr */
>> struct svc_deferred_req*rq_deferred; /* deferred request we are replaying */
>>  
>> struct xdr_buf  rq_arg;
>> diff --git a/include/linux/sunrpc/svc_xprt.h
>> b/include/linux/sunrpc/svc_xprt.h
>> index 81daa39..e3bd7b1 100644
>> --- a/include/linux/sunrpc/svc_xprt.h
>> +++ b/include/linux/sunrpc/svc_xprt.h
>> @@ -12,6 +12,7 @@
>>  struct svc_xprt_ops {
>> int  (*xpo_recvfrom)(struct svc_rqst *);
>> int  (*xpo_sendto)(struct svc_rqst *);
>> + void  (*xpo_release_rqst)(struct svc_rqst *);
>>  };
>>  
>>  struct svc_xprt_class {
>> diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
>> index 9c06b15..b24c084 100644
>> --- a/net/sunrpc/svcsock.c
>> +++ b/net/sunrpc/svcsock.c
>> @@ -185,14 +185,13 @@ svc_thread_dequeue(struct svc_pool *pool, struct
>> svc_rqst *rqstp)
>>  /*
>>   * Release an skbuff after use
>>   */
>> -static inline void
>> -svc_release_skb(struct svc_rqst *rqstp)
>> +static void svc_release_skb(struct svc_rqst *rqstp)
>>  {
>> - struct sk_buff *skb = rqstp->rq_skbuff;
>> + struct sk_buff *skb = rqstp->rq_xprt_ctxt;
>> struct svc_deferred_req *dr = rqstp->rq_deferred;
>>  
>> if (skb) {
>> -  rqstp->rq_skbuff = NULL;
>> +  rqstp->rq_xprt_ctxt = NULL;
>>  
>> dprintk("svc: service %p, releasing skb %p\n", rqstp, skb);
>> skb_free_datagram(rqstp->rq_sock->sk_sk, skb);
>> @@ -395,7 +394,7 @@ svc_sock_release(struct svc_rqst *rqstp)
>>  {
>> struct svc_sock *svsk = rqstp->rq_sock;
>>  
>> - svc_release_skb(rqstp);
>> + rqstp->rq_xprt->xpt_ops->xpo_release_rqst(rqstp);
>>  
>> svc_free_res_pages(rqstp);
>> rqstp->rq_res.page_len = 0;
>> @@ -867,7 +866,7 @@ svc_udp_recvfrom(struct svc_rqst *rqstp)
>> skb_free_datagram(svsk->sk_sk, skb);
>> return 0;
>> }
>> -  rqstp->rq_skbuff = skb;
>> +  rqstp->rq_xprt_ctxt = skb;
>> }
>>  
>> rqstp->rq_arg.page_base = 0;
>> @@ -903,6 +902,7 @@ svc_udp_sendto(struct svc_rqst *rqstp)
>>  static struct svc_xprt_ops svc_udp_ops = {
>> .xpo_recvfrom = svc_udp_recvfrom,
>> .xpo_sendto = svc_udp_sendto,
>> + .xpo_release_rqst = svc_release_skb,
>>  };
>>  
>>  static struct svc_xprt_class svc_udp_class = {
>> @@ -1291,7 +1291,7 @@ svc_tcp_recvfrom(struct svc_rqst *rqstp)
>> rqstp->rq_arg.page_len = len - rqstp->rq_arg.head[0].iov_len;
>> }
>>  
>> - rqstp->rq_skbuff      = NULL;
>> + rqstp->rq_xprt_ctxt   = NULL;
>> rqstp->rq_prot       = IPPROTO_TCP;
>>  
>> /* Reset TCP read info */
>> @@ -1357,6 +1357,7 @@ svc_tcp_sendto(struct svc_rqst *rqstp)
>>  static struct svc_xprt_ops svc_tcp_ops = {
>> .xpo_recvfrom = svc_tcp_recvfrom,
>> .xpo_sendto = svc_tcp_sendto,
>> + .xpo_release_rqst = svc_release_skb,
>>  };
>>  
>>  static struct svc_xprt_class svc_tcp_class = {
>> @@ -1578,7 +1579,7 @@ svc_send(struct svc_rqst *rqstp)
>> }
>>  
>> /* release the receive skb before sending the reply */
>> - svc_release_skb(rqstp);
>> + rqstp->rq_xprt->xpt_ops->xpo_release_rqst(rqstp);
>>  
>> /* calculate over-all length */
>> xb = & rqstp->rq_res;

[PATCH 07/38] svc: Add per-transport delete functions

[Tom Tucker]

Add transport specific xpo_detach and xpo_free functions. The xpo_detach
function causes the transport to stop delivering data-ready events 
and enqueing the transport for I/O.

The xpo_free function frees all resources associated with the particular 
transport instance.

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 include/linux/sunrpc/svc_xprt.h |    2 +
 net/sunrpc/svcsock.c            |   56 ++++++++++++++++++++++++++++++---------
 2 files changed, 45 insertions(+), 13 deletions(-)

diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
index e3bd7b1..5d7b2a6 100644
--- a/include/linux/sunrpc/svc_xprt.h
+++ b/include/linux/sunrpc/svc_xprt.h
@@ -13,6 +13,8 @@ struct svc_xprt_ops {
 	int		(*xpo_recvfrom)(struct svc_rqst *);
 	int		(*xpo_sendto)(struct svc_rqst *);
 	void		(*xpo_release_rqst)(struct svc_rqst *);
+	void		(*xpo_detach)(struct svc_xprt *);
+	void		(*xpo_free)(struct svc_xprt *);
 };
 
 struct svc_xprt_class {
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index b24c084..478fa33 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -85,6 +85,8 @@ static void		svc_udp_data_ready(struct sock *, int);
 static int		svc_udp_recvfrom(struct svc_rqst *);
 static int		svc_udp_sendto(struct svc_rqst *);
 static void		svc_close_socket(struct svc_sock *svsk);
+static void		svc_sock_detach(struct svc_xprt *);
+static void		svc_sock_free(struct svc_xprt *);
 
 static struct svc_deferred_req *svc_deferred_dequeue(struct svc_sock *svsk);
 static int svc_deferred_recv(struct svc_rqst *rqstp);
@@ -376,16 +378,8 @@ static inline void
 svc_sock_put(struct svc_sock *svsk)
 {
 	if (atomic_dec_and_test(&svsk->sk_inuse)) {
-		BUG_ON(! test_bit(SK_DEAD, &svsk->sk_flags));
-
-		dprintk("svc: releasing dead socket\n");
-		if (svsk->sk_sock->file)
-			sockfd_put(svsk->sk_sock);
-		else
-			sock_release(svsk->sk_sock);
-		if (svsk->sk_info_authunix != NULL)
-			svcauth_unix_info_release(svsk->sk_info_authunix);
-		kfree(svsk);
+		BUG_ON(!test_bit(SK_DEAD, &svsk->sk_flags));
+		svsk->sk_xprt.xpt_ops->xpo_free(&svsk->sk_xprt);
 	}
 }
 
@@ -903,6 +897,8 @@ static struct svc_xprt_ops svc_udp_ops = {
 	.xpo_recvfrom = svc_udp_recvfrom,
 	.xpo_sendto = svc_udp_sendto,
 	.xpo_release_rqst = svc_release_skb,
+	.xpo_detach = svc_sock_detach,
+	.xpo_free = svc_sock_free,
 };
 
 static struct svc_xprt_class svc_udp_class = {
@@ -1358,6 +1354,8 @@ static struct svc_xprt_ops svc_tcp_ops = {
 	.xpo_recvfrom = svc_tcp_recvfrom,
 	.xpo_sendto = svc_tcp_sendto,
 	.xpo_release_rqst = svc_release_skb,
+	.xpo_detach = svc_sock_detach,
+	.xpo_free = svc_sock_free,
 };
 
 static struct svc_xprt_class svc_tcp_class = {
@@ -1815,6 +1813,40 @@ bummer:
 }
 
 /*
+ * Detach the svc_sock from the socket so that no
+ * more callbacks occur.
+ */
+static void svc_sock_detach(struct svc_xprt *xprt)
+{
+	struct svc_sock *svsk = container_of(xprt, struct svc_sock, sk_xprt);
+	struct sock *sk = svsk->sk_sk;
+
+	dprintk("svc: svc_sock_detach(%p)\n", svsk);
+
+	/* put back the old socket callbacks */
+	sk->sk_state_change = svsk->sk_ostate;
+	sk->sk_data_ready = svsk->sk_odata;
+	sk->sk_write_space = svsk->sk_owspace;
+}
+
+/*
+ * Free the svc_sock's socket resources and the svc_sock itself.
+ */
+static void svc_sock_free(struct svc_xprt *xprt)
+{
+	struct svc_sock *svsk = container_of(xprt, struct svc_sock, sk_xprt);
+	dprintk("svc: svc_sock_free(%p)\n", svsk);
+
+	if (svsk->sk_info_authunix != NULL)
+		svcauth_unix_info_release(svsk->sk_info_authunix);
+	if (svsk->sk_sock->file)
+		sockfd_put(svsk->sk_sock);
+	else
+		sock_release(svsk->sk_sock);
+	kfree(svsk);
+}
+
+/*
  * Remove a dead socket
  */
 static void
@@ -1828,9 +1860,7 @@ svc_delete_socket(struct svc_sock *svsk)
 	serv = svsk->sk_server;
 	sk = svsk->sk_sk;
 
-	sk->sk_state_change = svsk->sk_ostate;
-	sk->sk_data_ready = svsk->sk_odata;
-	sk->sk_write_space = svsk->sk_owspace;
+	svsk->sk_xprt.xpt_ops->xpo_detach(&svsk->sk_xprt);
 
 	spin_lock_bh(&serv->sv_lock);
 

[PATCH 08/38] svc: Add xpo_prep_reply_hdr

[Tom Tucker]

Some transports add fields to the RPC header for replies, e.g. the TCP
record length. This function is called when preparing the reply header
to allow each transport to add whatever fields it requires.

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 include/linux/sunrpc/svc_xprt.h |    1 +
 net/sunrpc/svc.c                |    6 +++---
 net/sunrpc/svcsock.c            |   17 +++++++++++++++++
 3 files changed, 21 insertions(+), 3 deletions(-)

diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
index 5d7b2a6..8501115 100644
--- a/include/linux/sunrpc/svc_xprt.h
+++ b/include/linux/sunrpc/svc_xprt.h
@@ -11,6 +11,7 @@
 
 struct svc_xprt_ops {
 	int		(*xpo_recvfrom)(struct svc_rqst *);
+	void		(*xpo_prep_reply_hdr)(struct svc_rqst *);
 	int		(*xpo_sendto)(struct svc_rqst *);
 	void		(*xpo_release_rqst)(struct svc_rqst *);
 	void		(*xpo_detach)(struct svc_xprt *);
diff --git a/net/sunrpc/svc.c b/net/sunrpc/svc.c
index ea3fa86..3cc945d 100644
--- a/net/sunrpc/svc.c
+++ b/net/sunrpc/svc.c
@@ -839,9 +839,9 @@ svc_process(struct svc_rqst *rqstp)
 	rqstp->rq_res.tail[0].iov_len = 0;
 	/* Will be turned off only in gss privacy case: */
 	rqstp->rq_splice_ok = 1;
-	/* tcp needs a space for the record length... */
-	if (rqstp->rq_prot == IPPROTO_TCP)
-		svc_putnl(resv, 0);
+
+	/* Setup reply header */
+	rqstp->rq_xprt->xpt_ops->xpo_prep_reply_hdr(rqstp);
 
 	rqstp->rq_xid = svc_getu32(argv);
 	svc_putu32(resv, rqstp->rq_xid);
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index 478fa33..510ad45 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -893,12 +893,17 @@ svc_udp_sendto(struct svc_rqst *rqstp)
 	return error;
 }
 
+static void svc_udp_prep_reply_hdr(struct svc_rqst *rqstp)
+{
+}
+
 static struct svc_xprt_ops svc_udp_ops = {
 	.xpo_recvfrom = svc_udp_recvfrom,
 	.xpo_sendto = svc_udp_sendto,
 	.xpo_release_rqst = svc_release_skb,
 	.xpo_detach = svc_sock_detach,
 	.xpo_free = svc_sock_free,
+	.xpo_prep_reply_hdr = svc_udp_prep_reply_hdr,
 };
 
 static struct svc_xprt_class svc_udp_class = {
@@ -1350,12 +1355,24 @@ svc_tcp_sendto(struct svc_rqst *rqstp)
 	return sent;
 }
 
+/*
+ * Setup response header. TCP has a 4B record length field.
+ */
+static void svc_tcp_prep_reply_hdr(struct svc_rqst *rqstp)
+{
+	struct kvec *resv = &rqstp->rq_res.head[0];
+
+	/* tcp needs a space for the record length... */
+	svc_putnl(resv, 0);
+}
+
 static struct svc_xprt_ops svc_tcp_ops = {
 	.xpo_recvfrom = svc_tcp_recvfrom,
 	.xpo_sendto = svc_tcp_sendto,
 	.xpo_release_rqst = svc_release_skb,
 	.xpo_detach = svc_sock_detach,
 	.xpo_free = svc_sock_free,
+	.xpo_prep_reply_hdr = svc_tcp_prep_reply_hdr,
 };
 
 static struct svc_xprt_class svc_tcp_class = {

[PATCH 09/38] svc: Add a transport function that checks for write space

[Tom Tucker]

In order to avoid blocking a service thread, the receive side checks
to see if there is sufficient write space to reply to the request.
Each transport has a different mechanism for determining if there is
enough write space to reply.

The code that checked for write space was coupled with code that 
checked for CLOSE and CONN. These checks have been broken out into 
separate statements to make the code easier to read.

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 include/linux/sunrpc/svc_xprt.h |    1 
 net/sunrpc/svcsock.c            |   82 +++++++++++++++++++++++++++------------
 2 files changed, 57 insertions(+), 26 deletions(-)

diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
index 8501115..3adc8f3 100644
--- a/include/linux/sunrpc/svc_xprt.h
+++ b/include/linux/sunrpc/svc_xprt.h
@@ -10,6 +10,7 @@
 #include <linux/sunrpc/svc.h>
 
 struct svc_xprt_ops {
+	int		(*xpo_has_wspace)(struct svc_xprt *);
 	int		(*xpo_recvfrom)(struct svc_rqst *);
 	void		(*xpo_prep_reply_hdr)(struct svc_rqst *);
 	int		(*xpo_sendto)(struct svc_rqst *);
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index 510ad45..e24bf22 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -205,22 +205,6 @@ static void svc_release_skb(struct svc_rqst *rqstp)
 }
 
 /*
- * Any space to write?
- */
-static inline unsigned long
-svc_sock_wspace(struct svc_sock *svsk)
-{
-	int wspace;
-
-	if (svsk->sk_sock->type == SOCK_STREAM)
-		wspace = sk_stream_wspace(svsk->sk_sk);
-	else
-		wspace = sock_wspace(svsk->sk_sk);
-
-	return wspace;
-}
-
-/*
  * Queue up a socket with data pending. If there are idle nfsd
  * processes, wake 'em up.
  *
@@ -269,22 +253,24 @@ svc_sock_enqueue(struct svc_sock *svsk)
 	BUG_ON(svsk->sk_pool != NULL);
 	svsk->sk_pool = pool;
 
-	set_bit(SOCK_NOSPACE, &svsk->sk_sock->flags);
-	if (((atomic_read(&svsk->sk_reserved) + serv->sv_max_mesg)*2
-	     > svc_sock_wspace(svsk))
-	    && !test_bit(SK_CLOSE, &svsk->sk_flags)
-	    && !test_bit(SK_CONN, &svsk->sk_flags)) {
+	/* Handle pending connection */
+	if (test_bit(SK_CONN, &svsk->sk_flags))
+		goto process;
+
+	/* Handle close in-progress */
+	if (test_bit(SK_CLOSE, &svsk->sk_flags))
+		goto process;
+
+	/* Check if we have space to reply to a request */
+	if (!svsk->sk_xprt.xpt_ops->xpo_has_wspace(&svsk->sk_xprt)) {
 		/* Don't enqueue while not enough space for reply */
-		dprintk("svc: socket %p  no space, %d*2 > %ld, not enqueued\n",
-			svsk->sk_sk, atomic_read(&svsk->sk_reserved)+serv->sv_max_mesg,
-			svc_sock_wspace(svsk));
+		dprintk("svc: no write space, socket %p  not enqueued\n", svsk);
 		svsk->sk_pool = NULL;
 		clear_bit(SK_BUSY, &svsk->sk_flags);
 		goto out_unlock;
 	}
-	clear_bit(SOCK_NOSPACE, &svsk->sk_sock->flags);
-
 
+ process:
 	if (!list_empty(&pool->sp_threads)) {
 		rqstp = list_entry(pool->sp_threads.next,
 				   struct svc_rqst,
@@ -897,6 +883,24 @@ static void svc_udp_prep_reply_hdr(struct svc_rqst *rqstp)
 {
 }
 
+static int svc_udp_has_wspace(struct svc_xprt *xprt)
+{
+	struct svc_sock *svsk = container_of(xprt, struct svc_sock, sk_xprt);
+	struct svc_serv	*serv = svsk->sk_server;
+	unsigned long required;
+
+	/*
+	 * Set the SOCK_NOSPACE flag before checking the available
+	 * sock space.
+	 */
+	set_bit(SOCK_NOSPACE, &svsk->sk_sock->flags);
+	required = atomic_read(&svsk->sk_reserved) + serv->sv_max_mesg;
+	if (required*2 > sock_wspace(svsk->sk_sk))
+		return 0;
+	clear_bit(SOCK_NOSPACE, &svsk->sk_sock->flags);
+	return 1;
+}
+
 static struct svc_xprt_ops svc_udp_ops = {
 	.xpo_recvfrom = svc_udp_recvfrom,
 	.xpo_sendto = svc_udp_sendto,
@@ -904,6 +908,7 @@ static struct svc_xprt_ops svc_udp_ops = {
 	.xpo_detach = svc_sock_detach,
 	.xpo_free = svc_sock_free,
 	.xpo_prep_reply_hdr = svc_udp_prep_reply_hdr,
+	.xpo_has_wspace = svc_udp_has_wspace,
 };
 
 static struct svc_xprt_class svc_udp_class = {
@@ -1366,6 +1371,30 @@ static void svc_tcp_prep_reply_hdr(struct svc_rqst *rqstp)
 	svc_putnl(resv, 0);
 }
 
+static int svc_tcp_has_wspace(struct svc_xprt *xprt)
+{
+	struct svc_sock *svsk = container_of(xprt, struct svc_sock, sk_xprt);
+	struct svc_serv	*serv = svsk->sk_server;
+	int required;
+	int wspace;
+
+	/*
+	 * Set the SOCK_NOSPACE flag before checking the available
+	 * sock space.
+	 */
+	set_bit(SOCK_NOSPACE, &svsk->sk_sock->flags);
+	required = atomic_read(&svsk->sk_reserved) + serv->sv_max_mesg;
+	wspace = sk_stream_wspace(svsk->sk_sk);
+
+	if (wspace < sk_stream_min_wspace(svsk->sk_sk))
+		return 0;
+	if (required * 2 > wspace)
+		return 0;
+
+	clear_bit(SOCK_NOSPACE, &svsk->sk_sock->flags);
+	return 1;
+}
+
 static struct svc_xprt_ops svc_tcp_ops = {
 	.xpo_recvfrom = svc_tcp_recvfrom,
 	.xpo_sendto = svc_tcp_sendto,
@@ -1373,6 +1402,7 @@ static struct svc_xprt_ops svc_tcp_ops = {
 	.xpo_detach = svc_sock_detach,
 	.xpo_free = svc_sock_free,
 	.xpo_prep_reply_hdr = svc_tcp_prep_reply_hdr,
+	.xpo_has_wspace = svc_tcp_has_wspace,
 };
 
 static struct svc_xprt_class svc_tcp_class = {

Re: [PATCH 09/38] svc: Add a transport function that checks for write space

[Chuck Lever]

Hi Tom-

On Dec 11, 2007, at 6:32 PM, Tom Tucker wrote:
> In order to avoid blocking a service thread, the receive side checks
> to see if there is sufficient write space to reply to the request.
> Each transport has a different mechanism for determining if there is
> enough write space to reply.
>
> The code that checked for write space was coupled with code that
> checked for CLOSE and CONN. These checks have been broken out into
> separate statements to make the code easier to read.
>
> Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
> ---
>
>  include/linux/sunrpc/svc_xprt.h |    1
>  net/sunrpc/svcsock.c            |   82 ++++++++++++++++++++++++++ 
> +------------
>  2 files changed, 57 insertions(+), 26 deletions(-)
>
> diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/ 
> svc_xprt.h
> index 8501115..3adc8f3 100644
> --- a/include/linux/sunrpc/svc_xprt.h
> +++ b/include/linux/sunrpc/svc_xprt.h
> @@ -10,6 +10,7 @@
>  #include <linux/sunrpc/svc.h>
>
>  struct svc_xprt_ops {
> +	int		(*xpo_has_wspace)(struct svc_xprt *);
>  	int		(*xpo_recvfrom)(struct svc_rqst *);
>  	void		(*xpo_prep_reply_hdr)(struct svc_rqst *);
>  	int		(*xpo_sendto)(struct svc_rqst *);
> diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
> index 510ad45..e24bf22 100644
> --- a/net/sunrpc/svcsock.c
> +++ b/net/sunrpc/svcsock.c
> @@ -205,22 +205,6 @@ static void svc_release_skb(struct svc_rqst  
> *rqstp)
>  }
>
>  /*
> - * Any space to write?
> - */
> -static inline unsigned long
> -svc_sock_wspace(struct svc_sock *svsk)
> -{
> -	int wspace;
> -
> -	if (svsk->sk_sock->type == SOCK_STREAM)
> -		wspace = sk_stream_wspace(svsk->sk_sk);
> -	else
> -		wspace = sock_wspace(svsk->sk_sk);
> -
> -	return wspace;
> -}
> -
> -/*
>   * Queue up a socket with data pending. If there are idle nfsd
>   * processes, wake 'em up.
>   *
> @@ -269,22 +253,24 @@ svc_sock_enqueue(struct svc_sock *svsk)
>  	BUG_ON(svsk->sk_pool != NULL);
>  	svsk->sk_pool = pool;
>
> -	set_bit(SOCK_NOSPACE, &svsk->sk_sock->flags);
> -	if (((atomic_read(&svsk->sk_reserved) + serv->sv_max_mesg)*2
> -	     > svc_sock_wspace(svsk))
> -	    && !test_bit(SK_CLOSE, &svsk->sk_flags)
> -	    && !test_bit(SK_CONN, &svsk->sk_flags)) {
> +	/* Handle pending connection */
> +	if (test_bit(SK_CONN, &svsk->sk_flags))
> +		goto process;
> +
> +	/* Handle close in-progress */
> +	if (test_bit(SK_CLOSE, &svsk->sk_flags))
> +		goto process;
> +
> +	/* Check if we have space to reply to a request */
> +	if (!svsk->sk_xprt.xpt_ops->xpo_has_wspace(&svsk->sk_xprt)) {
>  		/* Don't enqueue while not enough space for reply */
> -		dprintk("svc: socket %p  no space, %d*2 > %ld, not enqueued\n",
> -			svsk->sk_sk, atomic_read(&svsk->sk_reserved)+serv->sv_max_mesg,
> -			svc_sock_wspace(svsk));
> +		dprintk("svc: no write space, socket %p  not enqueued\n", svsk);
>  		svsk->sk_pool = NULL;
>  		clear_bit(SK_BUSY, &svsk->sk_flags);
>  		goto out_unlock;
>  	}
> -	clear_bit(SOCK_NOSPACE, &svsk->sk_sock->flags);
> -
>
> + process:
>  	if (!list_empty(&pool->sp_threads)) {
>  		rqstp = list_entry(pool->sp_threads.next,
>  				   struct svc_rqst,
> @@ -897,6 +883,24 @@ static void svc_udp_prep_reply_hdr(struct  
> svc_rqst *rqstp)
>  {
>  }
>
> +static int svc_udp_has_wspace(struct svc_xprt *xprt)
> +{
> +	struct svc_sock *svsk = container_of(xprt, struct svc_sock,  
> sk_xprt);
> +	struct svc_serv	*serv = svsk->sk_server;
> +	unsigned long required;
> +
> +	/*
> +	 * Set the SOCK_NOSPACE flag before checking the available
> +	 * sock space.
> +	 */
> +	set_bit(SOCK_NOSPACE, &svsk->sk_sock->flags);
> +	required = atomic_read(&svsk->sk_reserved) + serv->sv_max_mesg;
> +	if (required*2 > sock_wspace(svsk->sk_sk))
> +		return 0;
> +	clear_bit(SOCK_NOSPACE, &svsk->sk_sock->flags);
> +	return 1;
> +}
> +
>  static struct svc_xprt_ops svc_udp_ops = {
>  	.xpo_recvfrom = svc_udp_recvfrom,
>  	.xpo_sendto = svc_udp_sendto,
> @@ -904,6 +908,7 @@ static struct svc_xprt_ops svc_udp_ops = {
>  	.xpo_detach = svc_sock_detach,
>  	.xpo_free = svc_sock_free,
>  	.xpo_prep_reply_hdr = svc_udp_prep_reply_hdr,
> +	.xpo_has_wspace = svc_udp_has_wspace,
>  };
>
>  static struct svc_xprt_class svc_udp_class = {
> @@ -1366,6 +1371,30 @@ static void svc_tcp_prep_reply_hdr(struct  
> svc_rqst *rqstp)
>  	svc_putnl(resv, 0);
>  }
>
> +static int svc_tcp_has_wspace(struct svc_xprt *xprt)
> +{
> +	struct svc_sock *svsk = container_of(xprt, struct svc_sock,  
> sk_xprt);
> +	struct svc_serv	*serv = svsk->sk_server;
> +	int required;
> +	int wspace;
> +
> +	/*
> +	 * Set the SOCK_NOSPACE flag before checking the available
> +	 * sock space.
> +	 */
> +	set_bit(SOCK_NOSPACE, &svsk->sk_sock->flags);
> +	required = atomic_read(&svsk->sk_reserved) + serv->sv_max_mesg;
> +	wspace = sk_stream_wspace(svsk->sk_sk);
> +
> +	if (wspace < sk_stream_min_wspace(svsk->sk_sk))
> +		return 0;
> +	if (required * 2 > wspace)
> +		return 0;

Since "required" is an int, this test can behave differently than the  
one it replaces.

What's the test plan to ensure the new test works reasonably?   
(Actually that's a question for the Maintainers too).

> +
> +	clear_bit(SOCK_NOSPACE, &svsk->sk_sock->flags);
> +	return 1;
> +}
> +
>  static struct svc_xprt_ops svc_tcp_ops = {
>  	.xpo_recvfrom = svc_tcp_recvfrom,
>  	.xpo_sendto = svc_tcp_sendto,
> @@ -1373,6 +1402,7 @@ static struct svc_xprt_ops svc_tcp_ops = {
>  	.xpo_detach = svc_sock_detach,
>  	.xpo_free = svc_sock_free,
>  	.xpo_prep_reply_hdr = svc_tcp_prep_reply_hdr,
> +	.xpo_has_wspace = svc_tcp_has_wspace,
>  };
>
>  static struct svc_xprt_class svc_tcp_class = {

--
Chuck Lever
chuck[dot]lever[at]oracle[dot]com

Re: [PATCH 09/38] svc: Add a transport function that checks for write space

[Tom Tucker]

On Wed, 2007-12-12 at 13:10 -0500, Chuck Lever wrote:
> Hi Tom-
> 
> On Dec 11, 2007, at 6:32 PM, Tom Tucker wrote:
> > In order to avoid blocking a service thread, the receive side checks

[...snip...]

> > +static int svc_tcp_has_wspace(struct svc_xprt *xprt)
> > +{
> > +	struct svc_sock *svsk = container_of(xprt, struct svc_sock,  
> > sk_xprt);
> > +	struct svc_serv	*serv = svsk->sk_server;
> > +	int required;
> > +	int wspace;
> > +
> > +	/*
> > +	 * Set the SOCK_NOSPACE flag before checking the available
> > +	 * sock space.
> > +	 */
> > +	set_bit(SOCK_NOSPACE, &svsk->sk_sock->flags);
> > +	required = atomic_read(&svsk->sk_reserved) + serv->sv_max_mesg;
> > +	wspace = sk_stream_wspace(svsk->sk_sk);
> > +
> > +	if (wspace < sk_stream_min_wspace(svsk->sk_sk))
> > +		return 0;
> > +	if (required * 2 > wspace)
> > +		return 0;
> 
> Since "required" is an int, this test can behave differently than the  
> one it replaces.
> 
> What's the test plan to ensure the new test works reasonably?   
> (Actually that's a question for the Maintainers too).
> 

Here's my matrix:

	Connectathon	iozone		Kernel build

V2	udp,tcp
V3	tcp,rdma	tcp,rdma	tcp,rdma
V4	tcp,rdma	tcp,rdma	tcp,rdma

I also kill tests to test the signal race paths, and I run multiple
tests concurrently (e.g. a couple iozones plus a kernel build). 

I guess none of these tests explicitly test the path in question
though.  


> > +
> > +	clear_bit(SOCK_NOSPACE, &svsk->sk_sock->flags);
> > +	return 1;
> > +}
> > +
> >  static struct svc_xprt_ops svc_tcp_ops = {
> >  	.xpo_recvfrom = svc_tcp_recvfrom,
> >  	.xpo_sendto = svc_tcp_sendto,
> > @@ -1373,6 +1402,7 @@ static struct svc_xprt_ops svc_tcp_ops = {
> >  	.xpo_detach = svc_sock_detach,
> >  	.xpo_free = svc_sock_free,
> >  	.xpo_prep_reply_hdr = svc_tcp_prep_reply_hdr,
> > +	.xpo_has_wspace = svc_tcp_has_wspace,
> >  };
> >
> >  static struct svc_xprt_class svc_tcp_class = {
> 
> --
> Chuck Lever
> chuck[dot]lever[at]oracle[dot]com
> -
> To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 09/38] svc: Add a transport function that checks for write space

[J. Bruce Fields]

On Wed, Dec 12, 2007 at 01:10:17PM -0500, Chuck Lever wrote:
> On Dec 11, 2007, at 6:32 PM, Tom Tucker wrote:
>> +	required = atomic_read(&svsk->sk_reserved) + serv->sv_max_mesg;
>> +	wspace = sk_stream_wspace(svsk->sk_sk);
>> +
>> +	if (wspace < sk_stream_min_wspace(svsk->sk_sk))
>> +		return 0;
>> +	if (required * 2 > wspace)
>> +		return 0;
>
> Since "required" is an int, this test can behave differently than the one 
> it replaces.

If sk_reserved can approach half of 2^31, for example, then surely we
have bigger problems?

--b.

Re: [PATCH 09/38] svc: Add a transport function that checks for write space

[Chuck Lever]

On Dec 13, 2007, at 4:33 PM, J. Bruce Fields wrote:
> On Wed, Dec 12, 2007 at 01:10:17PM -0500, Chuck Lever wrote:
>> On Dec 11, 2007, at 6:32 PM, Tom Tucker wrote:
>>> +	required = atomic_read(&svsk->sk_reserved) + serv->sv_max_mesg;
>>> +	wspace = sk_stream_wspace(svsk->sk_sk);
>>> +
>>> +	if (wspace < sk_stream_min_wspace(svsk->sk_sk))
>>> +		return 0;
>>> +	if (required * 2 > wspace)
>>> +		return 0;
>>
>> Since "required" is an int, this test can behave differently than  
>> the one
>> it replaces.
>
> If sk_reserved can approach half of 2^31, for example, then surely we
> have bigger problems?


What stops sk_reserved from going negative?

--
Chuck Lever
chuck[dot]lever[at]oracle[dot]com

Re: [PATCH 09/38] svc: Add a transport function that checks for write space

[Tom Tucker]

On Thu, 2007-12-13 at 16:45 -0500, Chuck Lever wrote:
> On Dec 13, 2007, at 4:33 PM, J. Bruce Fields wrote:
> > On Wed, Dec 12, 2007 at 01:10:17PM -0500, Chuck Lever wrote:
> >> On Dec 11, 2007, at 6:32 PM, Tom Tucker wrote:
> >>> +	required = atomic_read(&svsk->sk_reserved) + serv->sv_max_mesg;
> >>> +	wspace = sk_stream_wspace(svsk->sk_sk);
> >>> +
> >>> +	if (wspace < sk_stream_min_wspace(svsk->sk_sk))
> >>> +		return 0;
> >>> +	if (required * 2 > wspace)
> >>> +		return 0;
> >>
> >> Since "required" is an int, this test can behave differently than  
> >> the one
> >> it replaces.
> >
> > If sk_reserved can approach half of 2^31, for example, then surely we
> > have bigger problems?
> 
> 
> What stops sk_reserved from going negative?

Nothing actively _stops_ it from going negative. Indirectly it is
prevented by the "fact" that sv_max_mesg is always greater than the
amount returned by any read on the socket. If this is not true, then
sk_reserved can go negative. That would occur in svc_recv when the
difference between the amount currently reserved (worst case
sv_max_mesg) and the amount read is subtracted from the amount reserved.
If the amount read were greater than sv_max_mesg then the result would
be negative (both meanings intended).

I could add a BUG_ON and run some tests.

> 
> --
> Chuck Lever
> chuck[dot]lever[at]oracle[dot]com

Re: [PATCH 09/38] svc: Add a transport function that checks for write space

[J. Bruce Fields]

On Thu, Dec 13, 2007 at 04:20:27PM -0600, Tom Tucker wrote:
> 
> On Thu, 2007-12-13 at 16:45 -0500, Chuck Lever wrote:
> > On Dec 13, 2007, at 4:33 PM, J. Bruce Fields wrote:
> > > On Wed, Dec 12, 2007 at 01:10:17PM -0500, Chuck Lever wrote:
> > >> On Dec 11, 2007, at 6:32 PM, Tom Tucker wrote:
> > >>> +	required = atomic_read(&svsk->sk_reserved) + serv->sv_max_mesg;
> > >>> +	wspace = sk_stream_wspace(svsk->sk_sk);
> > >>> +
> > >>> +	if (wspace < sk_stream_min_wspace(svsk->sk_sk))
> > >>> +		return 0;
> > >>> +	if (required * 2 > wspace)
> > >>> +		return 0;
> > >>
> > >> Since "required" is an int, this test can behave differently than  
> > >> the one
> > >> it replaces.
> > >
> > > If sk_reserved can approach half of 2^31, for example, then surely we
> > > have bigger problems?
> > 
> > 
> > What stops sk_reserved from going negative?
> 
> Nothing actively _stops_ it from going negative. Indirectly it is
> prevented by the "fact" that sv_max_mesg is always greater than the
> amount returned by any read on the socket. If this is not true, then
> sk_reserved can go negative. That would occur in svc_recv when the
> difference between the amount currently reserved (worst case
> sv_max_mesg) and the amount read is subtracted from the amount reserved.
> If the amount read were greater than sv_max_mesg then the result would
> be negative (both meanings intended).
> 
> I could add a BUG_ON and run some tests.

I have to admit, I'm completely and utterly lost here.  Chuck, could you
explain exactly, in detail, how you think this change could cause a bug?

--b.

Re: [PATCH 09/38] svc: Add a transport function that checks for write space

[Chuck Lever]

On Dec 13, 2007, at 5:28 PM, J. Bruce Fields wrote:

> On Thu, Dec 13, 2007 at 04:20:27PM -0600, Tom Tucker wrote:
>>
>> On Thu, 2007-12-13 at 16:45 -0500, Chuck Lever wrote:
>>> On Dec 13, 2007, at 4:33 PM, J. Bruce Fields wrote:
>>>> On Wed, Dec 12, 2007 at 01:10:17PM -0500, Chuck Lever wrote:
>>>>> On Dec 11, 2007, at 6:32 PM, Tom Tucker wrote:
>>>>>> +	required = atomic_read(&svsk->sk_reserved) + serv->sv_max_mesg;
>>>>>> +	wspace = sk_stream_wspace(svsk->sk_sk);
>>>>>> +
>>>>>> +	if (wspace < sk_stream_min_wspace(svsk->sk_sk))
>>>>>> +		return 0;
>>>>>> +	if (required * 2 > wspace)
>>>>>> +		return 0;
>>>>>
>>>>> Since "required" is an int, this test can behave differently than
>>>>> the one
>>>>> it replaces.
>>>>
>>>> If sk_reserved can approach half of 2^31, for example, then  
>>>> surely we
>>>> have bigger problems?
>>>
>>>
>>> What stops sk_reserved from going negative?
>>
>> Nothing actively _stops_ it from going negative. Indirectly it is
>> prevented by the "fact" that sv_max_mesg is always greater than the
>> amount returned by any read on the socket. If this is not true, then
>> sk_reserved can go negative. That would occur in svc_recv when the
>> difference between the amount currently reserved (worst case
>> sv_max_mesg) and the amount read is subtracted from the amount  
>> reserved.
>> If the amount read were greater than sv_max_mesg then the result  
>> would
>> be negative (both meanings intended).
>>
>> I could add a BUG_ON and run some tests.
>
> I have to admit, I'm completely and utterly lost here.  Chuck,  
> could you
> explain exactly, in detail, how you think this change could cause a  
> bug?


The original code has this:

static inline unsigned long
svc_sock_wspace(struct svc_sock *svsk)
{
	int wspace;

	if (svsk->sk_sock->type == SOCK_STREAM)
		wspace = sk_stream_wspace(svsk->sk_sk);
	else
		wspace = sock_wspace(svsk->sk_sk);

	return wspace;
}

   ...

	if (((atomic_read(&svsk->sk_reserved) + serv->sv_max_mesg)*2
	     > svc_sock_wspace(svsk))

Since svc_sock_wspace() returns an unsigned long, that means the sum  
above is always converted to unsigned long.  If sk_reserved goes  
negative, then the sum becomes a large positive.

Note that in the TCP case, sk_stream_wspace() returns an int, which  
is also converted to an unsigned long by svc_sock_wspace().   
sk_stream_wspace() can return a negative if sk_wmem_queued happens to  
become larger than sk_sndbuf.

In all the other wspace callbacks I checked, the result of  
sk_stream_wspace() is compared to sk_stream_min_space() -- comparing  
an int to an int -- which means if sk_stream_wspace()'s result *does*  
go negative, those comparisons will catch it and do the right thing.

In the original RPC server implementation, though, sk_stream_wspace 
()'s result is converted to unsigned long; negative return values  
thus become large positives.

In both of these scenarios, the large positives make the comparison  
behave in exactly the opposite way we might expect it to.  Tom's code  
fixes this by making "required" an int for the TCP case so we still  
get an int-to-int comparison.

Because the original is coded the way it is, however, it's difficult  
to tell whether the signedness of these variables was intended for  
some clever effect, or whether it was an oversight.  If this was  
intentional, the new wspace check for TCP could have unintended  
performance consequences.

--
Chuck Lever
chuck[dot]lever[at]oracle[dot]com

Re: [PATCH 09/38] svc: Add a transport function that checks for write space

[Chuck Lever]

On Dec 13, 2007, at 5:20 PM, Tom Tucker wrote:
> On Thu, 2007-12-13 at 16:45 -0500, Chuck Lever wrote:
>> On Dec 13, 2007, at 4:33 PM, J. Bruce Fields wrote:
>>> On Wed, Dec 12, 2007 at 01:10:17PM -0500, Chuck Lever wrote:
>>>> On Dec 11, 2007, at 6:32 PM, Tom Tucker wrote:
>>>>> +	required = atomic_read(&svsk->sk_reserved) + serv->sv_max_mesg;
>>>>> +	wspace = sk_stream_wspace(svsk->sk_sk);
>>>>> +
>>>>> +	if (wspace < sk_stream_min_wspace(svsk->sk_sk))
>>>>> +		return 0;
>>>>> +	if (required * 2 > wspace)
>>>>> +		return 0;
>>>>
>>>> Since "required" is an int, this test can behave differently than
>>>> the one
>>>> it replaces.
>>>
>>> If sk_reserved can approach half of 2^31, for example, then  
>>> surely we
>>> have bigger problems?
>>
>>
>> What stops sk_reserved from going negative?
>
> Nothing actively _stops_ it from going negative. Indirectly it is
> prevented by the "fact" that sv_max_mesg is always greater than the
> amount returned by any read on the socket. If this is not true, then
> sk_reserved can go negative. That would occur in svc_recv when the
> difference between the amount currently reserved (worst case
> sv_max_mesg) and the amount read is subtracted from the amount  
> reserved.
> If the amount read were greater than sv_max_mesg then the result would
> be negative (both meanings intended).
>
> I could add a BUG_ON and run some tests.

My (limited) understanding is that the amount of socket output buffer  
space the server thinks it needs for sending a reply is subtracted  
from sk_reserved.  So it seems like svc_reserve() would be the place  
to assert that sk_reserved does not go negative.

As this is used mostly by svc_reserve_auth(), that suggests that a  
reasonable test case would be running a read/write intensive workload  
with some security flavor that uses large authenticators.

--
Chuck Lever
chuck[dot]lever[at]oracle[dot]com

[PATCH 10/38] svc: Move close processing to a single place

[Tom Tucker]

Close handling was duplicated in the UDP and TCP recvfrom 
methods. This code has been moved to the transport independent
svc_recv function.

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 net/sunrpc/svcsock.c |   24 ++++++++++--------------
 1 files changed, 10 insertions(+), 14 deletions(-)

diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index e24bf22..7373244 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -776,11 +776,6 @@ svc_udp_recvfrom(struct svc_rqst *rqstp)
 		return svc_deferred_recv(rqstp);
 	}
 
-	if (test_bit(SK_CLOSE, &svsk->sk_flags)) {
-		svc_delete_socket(svsk);
-		return 0;
-	}
-
 	clear_bit(SK_DATA, &svsk->sk_flags);
 	skb = NULL;
 	err = kernel_recvmsg(svsk->sk_sock, &msg, NULL,
@@ -1181,11 +1176,6 @@ svc_tcp_recvfrom(struct svc_rqst *rqstp)
 		return svc_deferred_recv(rqstp);
 	}
 
-	if (test_bit(SK_CLOSE, &svsk->sk_flags)) {
-		svc_delete_socket(svsk);
-		return 0;
-	}
-
 	if (svsk->sk_sk->sk_state == TCP_LISTEN) {
 		svc_tcp_accept(svsk);
 		svc_sock_received(svsk);
@@ -1575,10 +1565,16 @@ svc_recv(struct svc_rqst *rqstp, long timeout)
 	}
 	spin_unlock_bh(&pool->sp_lock);
 
-	dprintk("svc: server %p, pool %u, socket %p, inuse=%d\n",
-		 rqstp, pool->sp_id, svsk, atomic_read(&svsk->sk_inuse));
-	len = svsk->sk_xprt.xpt_ops->xpo_recvfrom(rqstp);
-	dprintk("svc: got len=%d\n", len);
+	len = 0;
+	if (test_bit(SK_CLOSE, &svsk->sk_flags)) {
+		dprintk("svc_recv: found SK_CLOSE\n");
+		svc_delete_socket(svsk);
+	} else {
+		dprintk("svc: server %p, pool %u, socket %p, inuse=%d\n",
+			rqstp, pool->sp_id, svsk, atomic_read(&svsk->sk_inuse));
+		len = svsk->sk_xprt.xpt_ops->xpo_recvfrom(rqstp);
+		dprintk("svc: got len=%d\n", len);
+	}
 
 	/* No data, incomplete (TCP) read, or accept() */
 	if (len == 0 || len == -EAGAIN) {

[PATCH 11/38] svc: Add xpo_accept transport function

[Tom Tucker]

Previously, the accept logic looked into the socket state to determine
whether to call accept or recv when data-ready was indicated on an endpoint. 
Since some transports don't use sockets, this logic was changed to use a flag 
bit (SK_LISTENER) to identify listening endpoints. A transport function 
(xpo_accept) was added to allow each transport to define its own accept 
processing. A transport's initialization logic is reponsible for setting the 
SK_LISTENER bit. I didn't see any way to do this in transport independent 
logic since the passive side of a UDP connection doesn't listen and
always recv's.

In the svc_recv function, if the SK_LISTENER bit is set, the transport
xpo_accept function is called to handle accept processing.

Note that all functions are defined even if they don't make sense
for a given transport. For example, accept doesn't mean anything for
UDP. The function is defined anyway and bug checks if called. The
UDP transport should never set the SK_LISTENER bit.

The code that poaches connections when the connection
limit is hit was moved to a subroutine to make the accept logic path
easier to follow. Since this is in the new connection path, it should 
not be a performance issue.

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 include/linux/sunrpc/svc_xprt.h |    1 
 include/linux/sunrpc/svcsock.h  |    1 
 net/sunrpc/svcsock.c            |  128 +++++++++++++++++++++------------------
 3 files changed, 72 insertions(+), 58 deletions(-)

diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
index 3adc8f3..1527ff1 100644
--- a/include/linux/sunrpc/svc_xprt.h
+++ b/include/linux/sunrpc/svc_xprt.h
@@ -10,6 +10,7 @@
 #include <linux/sunrpc/svc.h>
 
 struct svc_xprt_ops {
+	struct svc_xprt	*(*xpo_accept)(struct svc_xprt *);
 	int		(*xpo_has_wspace)(struct svc_xprt *);
 	int		(*xpo_recvfrom)(struct svc_rqst *);
 	void		(*xpo_prep_reply_hdr)(struct svc_rqst *);
diff --git a/include/linux/sunrpc/svcsock.h b/include/linux/sunrpc/svcsock.h
index 08e78d0..9882ce0 100644
--- a/include/linux/sunrpc/svcsock.h
+++ b/include/linux/sunrpc/svcsock.h
@@ -36,6 +36,7 @@ struct svc_sock {
 #define	SK_DEFERRED	8			/* request on sk_deferred */
 #define	SK_OLD		9			/* used for temp socket aging mark+sweep */
 #define	SK_DETACHED	10			/* detached from tempsocks list */
+#define SK_LISTENER	11			/* listening endpoint */
 
 	atomic_t    	    	sk_reserved;	/* space on outq that is reserved */
 
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index 7373244..4585ca5 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -896,6 +896,12 @@ static int svc_udp_has_wspace(struct svc_xprt *xprt)
 	return 1;
 }
 
+static struct svc_xprt *svc_udp_accept(struct svc_xprt *xprt)
+{
+	BUG();
+	return NULL;
+}
+
 static struct svc_xprt_ops svc_udp_ops = {
 	.xpo_recvfrom = svc_udp_recvfrom,
 	.xpo_sendto = svc_udp_sendto,
@@ -904,6 +910,7 @@ static struct svc_xprt_ops svc_udp_ops = {
 	.xpo_free = svc_sock_free,
 	.xpo_prep_reply_hdr = svc_udp_prep_reply_hdr,
 	.xpo_has_wspace = svc_udp_has_wspace,
+	.xpo_accept = svc_udp_accept,
 };
 
 static struct svc_xprt_class svc_udp_class = {
@@ -1028,9 +1035,9 @@ static inline int svc_port_is_privileged(struct sockaddr *sin)
 /*
  * Accept a TCP connection
  */
-static void
-svc_tcp_accept(struct svc_sock *svsk)
+static struct svc_xprt *svc_tcp_accept(struct svc_xprt *xprt)
 {
+	struct svc_sock *svsk = container_of(xprt, struct svc_sock, sk_xprt);
 	struct sockaddr_storage addr;
 	struct sockaddr	*sin = (struct sockaddr *) &addr;
 	struct svc_serv	*serv = svsk->sk_server;
@@ -1042,7 +1049,7 @@ svc_tcp_accept(struct svc_sock *svsk)
 
 	dprintk("svc: tcp_accept %p sock %p\n", svsk, sock);
 	if (!sock)
-		return;
+		return NULL;
 
 	clear_bit(SK_CONN, &svsk->sk_flags);
 	err = kernel_accept(sock, &newsock, O_NONBLOCK);
@@ -1053,11 +1060,10 @@ svc_tcp_accept(struct svc_sock *svsk)
 		else if (err != -EAGAIN && net_ratelimit())
 			printk(KERN_WARNING "%s: accept failed (err %d)!\n",
 				   serv->sv_name, -err);
-		return;
+		return NULL;
 	}
 
 	set_bit(SK_CONN, &svsk->sk_flags);
-	svc_sock_enqueue(svsk);
 
 	err = kernel_getpeername(newsock, sin, &slen);
 	if (err < 0) {
@@ -1099,59 +1105,14 @@ svc_tcp_accept(struct svc_sock *svsk)
 
 	svc_sock_received(newsvsk);
 
-	/* make sure that we don't have too many active connections.
-	 * If we have, something must be dropped.
-	 *
-	 * There's no point in trying to do random drop here for
-	 * DoS prevention. The NFS clients does 1 reconnect in 15
-	 * seconds. An attacker can easily beat that.
-	 *
-	 * The only somewhat efficient mechanism would be if drop
-	 * old connections from the same IP first. But right now
-	 * we don't even record the client IP in svc_sock.
-	 */
-	if (serv->sv_tmpcnt > (serv->sv_nrthreads+3)*20) {
-		struct svc_sock *svsk = NULL;
-		spin_lock_bh(&serv->sv_lock);
-		if (!list_empty(&serv->sv_tempsocks)) {
-			if (net_ratelimit()) {
-				/* Try to help the admin */
-				printk(KERN_NOTICE "%s: too many open TCP "
-					"sockets, consider increasing the "
-					"number of nfsd threads\n",
-						   serv->sv_name);
-				printk(KERN_NOTICE
-				       "%s: last TCP connect from %s\n",
-				       serv->sv_name, __svc_print_addr(sin,
-							buf, sizeof(buf)));
-			}
-			/*
-			 * Always select the oldest socket. It's not fair,
-			 * but so is life
-			 */
-			svsk = list_entry(serv->sv_tempsocks.prev,
-					  struct svc_sock,
-					  sk_list);
-			set_bit(SK_CLOSE, &svsk->sk_flags);
-			atomic_inc(&svsk->sk_inuse);
-		}
-		spin_unlock_bh(&serv->sv_lock);
-
-		if (svsk) {
-			svc_sock_enqueue(svsk);
-			svc_sock_put(svsk);
-		}
-
-	}
-
 	if (serv->sv_stats)
 		serv->sv_stats->nettcpconn++;
 
-	return;
+	return &newsvsk->sk_xprt;
 
 failed:
 	sock_release(newsock);
-	return;
+	return NULL;
 }
 
 /*
@@ -1176,12 +1137,6 @@ svc_tcp_recvfrom(struct svc_rqst *rqstp)
 		return svc_deferred_recv(rqstp);
 	}
 
-	if (svsk->sk_sk->sk_state == TCP_LISTEN) {
-		svc_tcp_accept(svsk);
-		svc_sock_received(svsk);
-		return 0;
-	}
-
 	if (test_and_clear_bit(SK_CHNGBUF, &svsk->sk_flags))
 		/* sndbuf needs to have room for one request
 		 * per thread, otherwise we can stall even when the
@@ -1393,6 +1348,7 @@ static struct svc_xprt_ops svc_tcp_ops = {
 	.xpo_free = svc_sock_free,
 	.xpo_prep_reply_hdr = svc_tcp_prep_reply_hdr,
 	.xpo_has_wspace = svc_tcp_has_wspace,
+	.xpo_accept = svc_tcp_accept,
 };
 
 static struct svc_xprt_class svc_tcp_class = {
@@ -1423,6 +1379,7 @@ svc_tcp_init(struct svc_sock *svsk)
 
 	if (sk->sk_state == TCP_LISTEN) {
 		dprintk("setting up TCP socket for listening\n");
+		set_bit(SK_LISTENER, &svsk->sk_flags);
 		sk->sk_data_ready = svc_tcp_listen_data_ready;
 		set_bit(SK_CONN, &svsk->sk_flags);
 	} else {
@@ -1475,6 +1432,55 @@ svc_sock_update_bufs(struct svc_serv *serv)
 }
 
 /*
+ * Make sure that we don't have too many active connections.  If we
+ * have, something must be dropped.
+ *
+ * There's no point in trying to do random drop here for DoS
+ * prevention. The NFS clients does 1 reconnect in 15 seconds. An
+ * attacker can easily beat that.
+ *
+ * The only somewhat efficient mechanism would be if drop old
+ * connections from the same IP first. But right now we don't even
+ * record the client IP in svc_sock.
+ */
+static void svc_check_conn_limits(struct svc_serv *serv)
+{
+	char	buf[RPC_MAX_ADDRBUFLEN];
+
+	if (serv->sv_tmpcnt > (serv->sv_nrthreads+3)*20) {
+		struct svc_sock *svsk = NULL;
+		spin_lock_bh(&serv->sv_lock);
+		if (!list_empty(&serv->sv_tempsocks)) {
+			if (net_ratelimit()) {
+				/* Try to help the admin */
+				printk(KERN_NOTICE "%s: too many open TCP "
+					"sockets, consider increasing the "
+					"number of nfsd threads\n",
+						   serv->sv_name);
+				printk(KERN_NOTICE
+				       "%s: last TCP connect from %s\n",
+				       serv->sv_name, buf);
+			}
+			/*
+			 * Always select the oldest socket. It's not fair,
+			 * but so is life
+			 */
+			svsk = list_entry(serv->sv_tempsocks.prev,
+					  struct svc_sock,
+					  sk_list);
+			set_bit(SK_CLOSE, &svsk->sk_flags);
+			atomic_inc(&svsk->sk_inuse);
+		}
+		spin_unlock_bh(&serv->sv_lock);
+
+		if (svsk) {
+			svc_sock_enqueue(svsk);
+			svc_sock_put(svsk);
+		}
+	}
+}
+
+/*
  * Receive the next request on any socket.  This code is carefully
  * organised not to touch any cachelines in the shared svc_serv
  * structure, only cachelines in the local svc_pool.
@@ -1569,6 +1575,12 @@ svc_recv(struct svc_rqst *rqstp, long timeout)
 	if (test_bit(SK_CLOSE, &svsk->sk_flags)) {
 		dprintk("svc_recv: found SK_CLOSE\n");
 		svc_delete_socket(svsk);
+	} else if (test_bit(SK_LISTENER, &svsk->sk_flags)) {
+		struct svc_xprt *newxpt;
+		newxpt = svsk->sk_xprt.xpt_ops->xpo_accept(&svsk->sk_xprt);
+		if (newxpt)
+			svc_check_conn_limits(svsk->sk_server);
+		svc_sock_received(svsk);
 	} else {
 		dprintk("svc: server %p, pool %u, socket %p, inuse=%d\n",
 			rqstp, pool->sp_id, svsk, atomic_read(&svsk->sk_inuse));

Re: [PATCH 11/38] svc: Add xpo_accept transport function

[J. Bruce Fields]

On Tue, Dec 11, 2007 at 05:32:17PM -0600, Tom Tucker wrote:
> 
> Previously, the accept logic looked into the socket state to determine
> whether to call accept or recv when data-ready was indicated on an endpoint. 
> Since some transports don't use sockets, this logic was changed to use a flag 
> bit (SK_LISTENER) to identify listening endpoints. A transport function 
> (xpo_accept) was added to allow each transport to define its own accept 
> processing.

Nit: I'd put description of what this patch does in the present tense.

...
> The code that poaches connections when the connection
> limit is hit was moved to a subroutine to make the accept logic path
> easier to follow. Since this is in the new connection path, it should 
> not be a performance issue.

Makes sense.  I'd have done that in a separate patch.

> +static void svc_check_conn_limits(struct svc_serv *serv)
> +{
> +	char	buf[RPC_MAX_ADDRBUFLEN];
> +
> +	if (serv->sv_tmpcnt > (serv->sv_nrthreads+3)*20) {
> +		struct svc_sock *svsk = NULL;
> +		spin_lock_bh(&serv->sv_lock);
> +		if (!list_empty(&serv->sv_tempsocks)) {
> +			if (net_ratelimit()) {
> +				/* Try to help the admin */
> +				printk(KERN_NOTICE "%s: too many open TCP "
> +					"sockets, consider increasing the "
> +					"number of nfsd threads\n",
> +						   serv->sv_name);
> +				printk(KERN_NOTICE
> +				       "%s: last TCP connect from %s\n",
> +				       serv->sv_name, buf);

It looks like the content of "buf" is unitialized here?

--b.

> +			}
> +			/*
> +			 * Always select the oldest socket. It's not fair,
> +			 * but so is life
> +			 */
> +			svsk = list_entry(serv->sv_tempsocks.prev,
> +					  struct svc_sock,
> +					  sk_list);
> +			set_bit(SK_CLOSE, &svsk->sk_flags);
> +			atomic_inc(&svsk->sk_inuse);
> +		}
> +		spin_unlock_bh(&serv->sv_lock);
> +
> +		if (svsk) {
> +			svc_sock_enqueue(svsk);
> +			svc_sock_put(svsk);
> +		}
> +	}
> +}
> +
> +/*

Re: [PATCH 11/38] svc: Add xpo_accept transport function

[Tom Tucker]

On Fri, 2007-12-14 at 14:01 -0500, J. Bruce Fields wrote:
> On Tue, Dec 11, 2007 at 05:32:17PM -0600, Tom Tucker wrote:
> > 
> > Previously, the accept logic looked into the socket state to determine
> > whether to call accept or recv when data-ready was indicated on an endpoint. 
> > Since some transports don't use sockets, this logic was changed to use a flag 
> > bit (SK_LISTENER) to identify listening endpoints. A transport function 
> > (xpo_accept) was added to allow each transport to define its own accept 
> > processing.
> 
> Nit: I'd put description of what this patch does in the present tense.
> 
ok.
> ...
> > The code that poaches connections when the connection
> > limit is hit was moved to a subroutine to make the accept logic path
> > easier to follow. Since this is in the new connection path, it should 
> > not be a performance issue.
> 
> Makes sense.  I'd have done that in a separate patch.
> 
> > +static void svc_check_conn_limits(struct svc_serv *serv)
> > +{
> > +	char	buf[RPC_MAX_ADDRBUFLEN];
> > +
> > +	if (serv->sv_tmpcnt > (serv->sv_nrthreads+3)*20) {
> > +		struct svc_sock *svsk = NULL;
> > +		spin_lock_bh(&serv->sv_lock);
> > +		if (!list_empty(&serv->sv_tempsocks)) {
> > +			if (net_ratelimit()) {
> > +				/* Try to help the admin */
> > +				printk(KERN_NOTICE "%s: too many open TCP "
> > +					"sockets, consider increasing the "
> > +					"number of nfsd threads\n",
> > +						   serv->sv_name);
> > +				printk(KERN_NOTICE
> > +				       "%s: last TCP connect from %s\n",
> > +				       serv->sv_name, buf);
> 
> It looks like the content of "buf" is unitialized here?

This is a bug I introduced. How about if I break this into a separate
patch (like you suggest above) with the appropriate __svc_print_addr
that fills in buf.

> 
> --b.
> 
> > +			}
> > +			/*
> > +			 * Always select the oldest socket. It's not fair,
> > +			 * but so is life
> > +			 */
> > +			svsk = list_entry(serv->sv_tempsocks.prev,
> > +					  struct svc_sock,
> > +					  sk_list);
> > +			set_bit(SK_CLOSE, &svsk->sk_flags);
> > +			atomic_inc(&svsk->sk_inuse);
> > +		}
> > +		spin_unlock_bh(&serv->sv_lock);
> > +
> > +		if (svsk) {
> > +			svc_sock_enqueue(svsk);
> > +			svc_sock_put(svsk);
> > +		}
> > +	}
> > +}
> > +
> > +/*

Re: [PATCH 11/38] svc: Add xpo_accept transport function

[Tom Tucker]

On 12/14/07 1:01 PM, "J. Bruce Fields" <bfields@fieldses.org> wrote:

> On Tue, Dec 11, 2007 at 05:32:17PM -0600, Tom Tucker wrote:
>> 
>> Previously, the accept logic looked into the socket state to determine
>> whether to call accept or recv when data-ready was indicated on an endpoint.
>> Since some transports don't use sockets, this logic was changed to use a flag
>> bit (SK_LISTENER) to identify listening endpoints. A transport function
>> (xpo_accept) was added to allow each transport to define its own accept
>> processing.
> 
> Nit: I'd put description of what this patch does in the present tense.
>

Ok,
 
> ...
>> The code that poaches connections when the connection
>> limit is hit was moved to a subroutine to make the accept logic path
>> easier to follow. Since this is in the new connection path, it should
>> not be a performance issue.
> 
> Makes sense.  I'd have done that in a separate patch.
>

Ok,
 
>> +static void svc_check_conn_limits(struct svc_serv *serv)
>> +{
>> + char buf[RPC_MAX_ADDRBUFLEN];
>> +
>> + if (serv->sv_tmpcnt > (serv->sv_nrthreads+3)*20) {
>> +  struct svc_sock *svsk = NULL;
>> +  spin_lock_bh(&serv->sv_lock);
>> +  if (!list_empty(&serv->sv_tempsocks)) {
>> +   if (net_ratelimit()) {
>> +    /* Try to help the admin */
>> +    printk(KERN_NOTICE "%s: too many open TCP "
>> +     "sockets, consider increasing the "
>> +     "number of nfsd threads\n",
>> +         serv->sv_name);
>> +    printk(KERN_NOTICE
>> +           "%s: last TCP connect from %s\n",
>> +           serv->sv_name, buf);
> 
> It looks like the content of "buf" is unitialized here?
>

Bug, fixed.
 
> --b.
> 
>> +   }
>> +   /*
>> +    * Always select the oldest socket. It's not fair,
>> +    * but so is life
>> +    */
>> +   svsk = list_entry(serv->sv_tempsocks.prev,
>> +       struct svc_sock,
>> +       sk_list);
>> +   set_bit(SK_CLOSE, &svsk->sk_flags);
>> +   atomic_inc(&svsk->sk_inuse);
>> +  }
>> +  spin_unlock_bh(&serv->sv_lock);
>> +
>> +  if (svsk) {
>> +   svc_sock_enqueue(svsk);
>> +   svc_sock_put(svsk);
>> +  }
>> + }
>> +}
>> +
>> +/*

Re: [PATCH 11/38] svc: Add xpo_accept transport function

[J. Bruce Fields]

On Tue, Dec 11, 2007 at 05:32:17PM -0600, Tom Tucker wrote:
> @@ -1053,11 +1060,10 @@ svc_tcp_accept(struct svc_sock *svsk)
>  		else if (err != -EAGAIN && net_ratelimit())
>  			printk(KERN_WARNING "%s: accept failed (err %d)!\n",
>  				   serv->sv_name, -err);
> -		return;
> +		return NULL;
>  	}
>  
>  	set_bit(SK_CONN, &svsk->sk_flags);
> -	svc_sock_enqueue(svsk);
>  
>  	err = kernel_getpeername(newsock, sin, &slen);
>  	if (err < 0) {

Why did we need that svc_sock_enqueue here, and why don't we any more?
(And if we don't, but we still need the set_bit, then we need to fix the
comment at the top of the file claiming the svc_sock_enqueue() is always
required after setting SK_CONN.)

--b.

Re: [PATCH 11/38] svc: Add xpo_accept transport function

[Tom Tucker]

On 12/14/07 1:55 PM, "J. Bruce Fields" <bfields@fieldses.org> wrote:

> On Tue, Dec 11, 2007 at 05:32:17PM -0600, Tom Tucker wrote:
>> @@ -1053,11 +1060,10 @@ svc_tcp_accept(struct svc_sock *svsk)
>> else if (err != -EAGAIN && net_ratelimit())
>> printk(KERN_WARNING "%s: accept failed (err %d)!\n",
>>   serv->sv_name, -err);
>> -  return;
>> +  return NULL;
>> }
>>  
>> set_bit(SK_CONN, &svsk->sk_flags);
>> - svc_sock_enqueue(svsk);
>>  
>> err = kernel_getpeername(newsock, sin, &slen);
>> if (err < 0) {
> 
> Why did we need that svc_sock_enqueue here, and why don't we any more?

Calling svc_sock_enqueue while holding the BUSY bit is a no-op. We can
certainly break this out and update the comment.

> (And if we don't, but we still need the set_bit, then we need to fix the
> comment at the top of the file claiming the svc_sock_enqueue() is always
> required after setting SK_CONN.)

We need to call svc_sock_enqueue, but it is done indirectly through
svc_sock_received -- that calls svc_sock_enqueue after clearing the BUSY
bit. 

> 
> --b.

Re: [PATCH 11/38] svc: Add xpo_accept transport function

[Tom Tucker]

On 12/14/07 1:55 PM, "J. Bruce Fields" <bfields@fieldses.org> wrote:

> On Tue, Dec 11, 2007 at 05:32:17PM -0600, Tom Tucker wrote:
>> @@ -1053,11 +1060,10 @@ svc_tcp_accept(struct svc_sock *svsk)
>> else if (err != -EAGAIN && net_ratelimit())
>> printk(KERN_WARNING "%s: accept failed (err %d)!\n",
>>   serv->sv_name, -err);
>> -  return;
>> +  return NULL;
>> }
>>  
>> set_bit(SK_CONN, &svsk->sk_flags);
>> - svc_sock_enqueue(svsk);
>>  
>> err = kernel_getpeername(newsock, sin, &slen);
>> if (err < 0) {
> 
> Why did we need that svc_sock_enqueue here, and why don't we any more?
> (And if we don't, but we still need the set_bit, then we need to fix the
> comment at the top of the file claiming the svc_sock_enqueue() is always
> required after setting SK_CONN.)

I moved this to it's own patch with a description of why it was removed.

> 
> --b.
> -
> To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH 12/38] svc: Add a generic transport svc_create_xprt function

[Tom Tucker]

The svc_create_xprt function is a transport independent version
of the svc_makesock function. 

Since transport instance creation contains transport dependent and 
independent components, add an xpo_create transport function. The 
transport implementation of this function allocates the memory for the 
endpoint, implements the transport dependent initialization logic, and
calls svc_xprt_init to initialize the transport independent field (svc_xprt) 
in it's data structure.

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 include/linux/sunrpc/svc_xprt.h |    4 +++
 net/sunrpc/svc_xprt.c           |   37 ++++++++++++++++++++++++
 net/sunrpc/svcsock.c            |   59 +++++++++++++++++++++++++++++----------
 3 files changed, 85 insertions(+), 15 deletions(-)

diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
index 1527ff1..3f4a1df 100644
--- a/include/linux/sunrpc/svc_xprt.h
+++ b/include/linux/sunrpc/svc_xprt.h
@@ -10,6 +10,9 @@
 #include <linux/sunrpc/svc.h>
 
 struct svc_xprt_ops {
+	struct svc_xprt	*(*xpo_create)(struct svc_serv *,
+				       struct sockaddr *, int,
+				       int);
 	struct svc_xprt	*(*xpo_accept)(struct svc_xprt *);
 	int		(*xpo_has_wspace)(struct svc_xprt *);
 	int		(*xpo_recvfrom)(struct svc_rqst *);
@@ -36,5 +39,6 @@ struct svc_xprt {
 int	svc_reg_xprt_class(struct svc_xprt_class *);
 int	svc_unreg_xprt_class(struct svc_xprt_class *);
 void	svc_xprt_init(struct svc_xprt_class *, struct svc_xprt *);
+int	svc_create_xprt(struct svc_serv *, char *, unsigned short, int);
 
 #endif /* SUNRPC_SVC_XPRT_H */
diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
index 92ea85b..9136da4 100644
--- a/net/sunrpc/svc_xprt.c
+++ b/net/sunrpc/svc_xprt.c
@@ -93,3 +93,40 @@ void svc_xprt_init(struct svc_xprt_class *xcl, struct svc_xprt *xprt)
 	xprt->xpt_ops = xcl->xcl_ops;
 }
 EXPORT_SYMBOL_GPL(svc_xprt_init);
+
+int svc_create_xprt(struct svc_serv *serv, char *xprt_name, unsigned short port,
+		    int flags)
+{
+	struct svc_xprt_class *xcl;
+	int ret = -ENOENT;
+	struct sockaddr_in sin = {
+		.sin_family		= AF_INET,
+		.sin_addr.s_addr	= INADDR_ANY,
+		.sin_port		= htons(port),
+	};
+	dprintk("svc: creating transport %s[%d]\n", xprt_name, port);
+	spin_lock(&svc_xprt_class_lock);
+	list_for_each_entry(xcl, &svc_xprt_class_list, xcl_list) {
+		if (strcmp(xprt_name, xcl->xcl_name) == 0) {
+			spin_unlock(&svc_xprt_class_lock);
+			if (try_module_get(xcl->xcl_owner)) {
+				struct svc_xprt *newxprt;
+				ret = 0;
+				newxprt = xcl->xcl_ops->xpo_create
+					(serv,
+					 (struct sockaddr *)&sin, sizeof(sin),
+					 flags);
+				if (IS_ERR(newxprt)) {
+					module_put(xcl->xcl_owner);
+					ret = PTR_ERR(newxprt);
+				}
+			}
+			goto out;
+		}
+	}
+	spin_unlock(&svc_xprt_class_lock);
+	dprintk("svc: transport %s not found\n", xprt_name);
+ out:
+	return ret;
+}
+EXPORT_SYMBOL_GPL(svc_create_xprt);
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index 4585ca5..9d99490 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -91,6 +91,8 @@ static void		svc_sock_free(struct svc_xprt *);
 static struct svc_deferred_req *svc_deferred_dequeue(struct svc_sock *svsk);
 static int svc_deferred_recv(struct svc_rqst *rqstp);
 static struct cache_deferred_req *svc_defer(struct cache_req *req);
+static struct svc_xprt *svc_create_socket(struct svc_serv *, int,
+					  struct sockaddr *, int, int);
 
 /* apparently the "standard" is that clients close
  * idle connections after 5 minutes, servers after
@@ -365,6 +367,7 @@ svc_sock_put(struct svc_sock *svsk)
 {
 	if (atomic_dec_and_test(&svsk->sk_inuse)) {
 		BUG_ON(!test_bit(SK_DEAD, &svsk->sk_flags));
+		module_put(svsk->sk_xprt.xpt_class->xcl_owner);
 		svsk->sk_xprt.xpt_ops->xpo_free(&svsk->sk_xprt);
 	}
 }
@@ -902,7 +905,15 @@ static struct svc_xprt *svc_udp_accept(struct svc_xprt *xprt)
 	return NULL;
 }
 
+static struct svc_xprt *svc_udp_create(struct svc_serv *serv,
+				       struct sockaddr *sa, int salen,
+				       int flags)
+{
+	return svc_create_socket(serv, IPPROTO_UDP, sa, salen, flags);
+}
+
 static struct svc_xprt_ops svc_udp_ops = {
+	.xpo_create = svc_udp_create,
 	.xpo_recvfrom = svc_udp_recvfrom,
 	.xpo_sendto = svc_udp_sendto,
 	.xpo_release_rqst = svc_release_skb,
@@ -915,6 +926,7 @@ static struct svc_xprt_ops svc_udp_ops = {
 
 static struct svc_xprt_class svc_udp_class = {
 	.xcl_name = "udp",
+	.xcl_owner = THIS_MODULE,
 	.xcl_ops = &svc_udp_ops,
 	.xcl_max_payload = RPCSVC_MAXPAYLOAD_UDP,
 };
@@ -1340,7 +1352,15 @@ static int svc_tcp_has_wspace(struct svc_xprt *xprt)
 	return 1;
 }
 
+static struct svc_xprt *svc_tcp_create(struct svc_serv *serv,
+				       struct sockaddr *sa, int salen,
+				       int flags)
+{
+	return svc_create_socket(serv, IPPROTO_TCP, sa, salen, flags);
+}
+
 static struct svc_xprt_ops svc_tcp_ops = {
+	.xpo_create = svc_tcp_create,
 	.xpo_recvfrom = svc_tcp_recvfrom,
 	.xpo_sendto = svc_tcp_sendto,
 	.xpo_release_rqst = svc_release_skb,
@@ -1353,6 +1373,7 @@ static struct svc_xprt_ops svc_tcp_ops = {
 
 static struct svc_xprt_class svc_tcp_class = {
 	.xcl_name = "tcp",
+	.xcl_owner = THIS_MODULE,
 	.xcl_ops = &svc_tcp_ops,
 	.xcl_max_payload = RPCSVC_MAXPAYLOAD_TCP,
 };
@@ -1578,8 +1599,14 @@ svc_recv(struct svc_rqst *rqstp, long timeout)
 	} else if (test_bit(SK_LISTENER, &svsk->sk_flags)) {
 		struct svc_xprt *newxpt;
 		newxpt = svsk->sk_xprt.xpt_ops->xpo_accept(&svsk->sk_xprt);
-		if (newxpt)
+		if (newxpt) {
+			/*
+			 * We know this module_get will succeed because the
+			 * listener holds a reference too
+			 */
+			__module_get(newxpt->xpt_class->xcl_owner);
 			svc_check_conn_limits(svsk->sk_server);
+		}
 		svc_sock_received(svsk);
 	} else {
 		dprintk("svc: server %p, pool %u, socket %p, inuse=%d\n",
@@ -1819,8 +1846,10 @@ EXPORT_SYMBOL_GPL(svc_addsock);
 /*
  * Create socket for RPC service.
  */
-static int svc_create_socket(struct svc_serv *serv, int protocol,
-				struct sockaddr *sin, int len, int flags)
+static struct svc_xprt *svc_create_socket(struct svc_serv *serv,
+					  int protocol,
+					  struct sockaddr *sin, int len,
+					  int flags)
 {
 	struct svc_sock	*svsk;
 	struct socket	*sock;
@@ -1835,13 +1864,13 @@ static int svc_create_socket(struct svc_serv *serv, int protocol,
 	if (protocol != IPPROTO_UDP && protocol != IPPROTO_TCP) {
 		printk(KERN_WARNING "svc: only UDP and TCP "
 				"sockets supported\n");
-		return -EINVAL;
+		return ERR_PTR(-EINVAL);
 	}
 	type = (protocol == IPPROTO_UDP)? SOCK_DGRAM : SOCK_STREAM;
 
 	error = sock_create_kern(sin->sa_family, type, protocol, &sock);
 	if (error < 0)
-		return error;
+		return ERR_PTR(error);
 
 	svc_reclassify_socket(sock);
 
@@ -1858,13 +1887,13 @@ static int svc_create_socket(struct svc_serv *serv, int protocol,
 
 	if ((svsk = svc_setup_socket(serv, sock, &error, flags)) != NULL) {
 		svc_sock_received(svsk);
-		return ntohs(inet_sk(svsk->sk_sk)->sport);
+		return (struct svc_xprt *)svsk;
 	}
 
 bummer:
 	dprintk("svc: svc_create_socket error = %d\n", -error);
 	sock_release(sock);
-	return error;
+	return ERR_PTR(error);
 }
 
 /*
@@ -1975,15 +2004,15 @@ void svc_force_close_socket(struct svc_sock *svsk)
 int svc_makesock(struct svc_serv *serv, int protocol, unsigned short port,
 			int flags)
 {
-	struct sockaddr_in sin = {
-		.sin_family		= AF_INET,
-		.sin_addr.s_addr	= INADDR_ANY,
-		.sin_port		= htons(port),
-	};
-
 	dprintk("svc: creating socket proto = %d\n", protocol);
-	return svc_create_socket(serv, protocol, (struct sockaddr *) &sin,
-							sizeof(sin), flags);
+	switch (protocol) {
+	case IPPROTO_TCP:
+		return svc_create_xprt(serv, "tcp", port, flags);
+	case IPPROTO_UDP:
+		return svc_create_xprt(serv, "udp", port, flags);
+	default:
+		return -EINVAL;
+	}
 }
 
 /*

[PATCH 13/38] svc: Change services to use new svc_create_xprt service

[Tom Tucker]

Modify the various kernel RPC svcs to use the svc_create_xprt service.

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 fs/lockd/svc.c                 |   17 ++++++++---------
 fs/nfs/callback.c              |    4 ++--
 fs/nfsd/nfssvc.c               |    4 ++--
 include/linux/sunrpc/svcsock.h |    1 -
 net/sunrpc/sunrpc_syms.c       |    1 -
 net/sunrpc/svcsock.c           |   22 ----------------------
 6 files changed, 12 insertions(+), 37 deletions(-)

diff --git a/fs/lockd/svc.c b/fs/lockd/svc.c
index 82e2192..8686915 100644
--- a/fs/lockd/svc.c
+++ b/fs/lockd/svc.c
@@ -219,13 +219,12 @@ lockd(struct svc_rqst *rqstp)
 	module_put_and_exit(0);
 }
 
-
-static int find_socket(struct svc_serv *serv, int proto)
+static int find_xprt(struct svc_serv *serv, char *proto)
 {
 	struct svc_sock *svsk;
 	int found = 0;
 	list_for_each_entry(svsk, &serv->sv_permsocks, sk_list)
-		if (svsk->sk_sk->sk_protocol == proto) {
+		if (strcmp(svsk->sk_xprt.xpt_class->xcl_name, proto) == 0) {
 			found = 1;
 			break;
 		}
@@ -243,13 +242,13 @@ static int make_socks(struct svc_serv *serv, int proto)
 	int err = 0;
 
 	if (proto == IPPROTO_UDP || nlm_udpport)
-		if (!find_socket(serv, IPPROTO_UDP))
-			err = svc_makesock(serv, IPPROTO_UDP, nlm_udpport,
-						SVC_SOCK_DEFAULTS);
+		if (!find_xprt(serv, "udp"))
+			err = svc_create_xprt(serv, "udp", nlm_udpport,
+					      SVC_SOCK_DEFAULTS);
 	if (err >= 0 && (proto == IPPROTO_TCP || nlm_tcpport))
-		if (!find_socket(serv, IPPROTO_TCP))
-			err = svc_makesock(serv, IPPROTO_TCP, nlm_tcpport,
-						SVC_SOCK_DEFAULTS);
+		if (!find_xprt(serv, "tcp"))
+			err = svc_create_xprt(serv, "tcp", nlm_tcpport,
+					      SVC_SOCK_DEFAULTS);
 
 	if (err >= 0) {
 		warned = 0;
diff --git a/fs/nfs/callback.c b/fs/nfs/callback.c
index a796be5..e27ca14 100644
--- a/fs/nfs/callback.c
+++ b/fs/nfs/callback.c
@@ -123,8 +123,8 @@ int nfs_callback_up(void)
 	if (!serv)
 		goto out_err;
 
-	ret = svc_makesock(serv, IPPROTO_TCP, nfs_callback_set_tcpport,
-							SVC_SOCK_ANONYMOUS);
+	ret = svc_create_xprt(serv, "tcp", nfs_callback_set_tcpport,
+			      SVC_SOCK_ANONYMOUS);
 	if (ret <= 0)
 		goto out_destroy;
 	nfs_callback_tcpport = ret;
diff --git a/fs/nfsd/nfssvc.c b/fs/nfsd/nfssvc.c
index 1190aea..a828b0b 100644
--- a/fs/nfsd/nfssvc.c
+++ b/fs/nfsd/nfssvc.c
@@ -236,7 +236,7 @@ static int nfsd_init_socks(int port)
 
 	error = lockd_up(IPPROTO_UDP);
 	if (error >= 0) {
-		error = svc_makesock(nfsd_serv, IPPROTO_UDP, port,
+		error = svc_create_xprt(nfsd_serv, "udp", port,
 					SVC_SOCK_DEFAULTS);
 		if (error < 0)
 			lockd_down();
@@ -247,7 +247,7 @@ static int nfsd_init_socks(int port)
 #ifdef CONFIG_NFSD_TCP
 	error = lockd_up(IPPROTO_TCP);
 	if (error >= 0) {
-		error = svc_makesock(nfsd_serv, IPPROTO_TCP, port,
+		error = svc_create_xprt(nfsd_serv, "tcp", port,
 					SVC_SOCK_DEFAULTS);
 		if (error < 0)
 			lockd_down();
diff --git a/include/linux/sunrpc/svcsock.h b/include/linux/sunrpc/svcsock.h
index 9882ce0..3181d9d 100644
--- a/include/linux/sunrpc/svcsock.h
+++ b/include/linux/sunrpc/svcsock.h
@@ -67,7 +67,6 @@ struct svc_sock {
 /*
  * Function prototypes.
  */
-int		svc_makesock(struct svc_serv *, int, unsigned short, int flags);
 void		svc_force_close_socket(struct svc_sock *);
 int		svc_recv(struct svc_rqst *, long);
 int		svc_send(struct svc_rqst *);
diff --git a/net/sunrpc/sunrpc_syms.c b/net/sunrpc/sunrpc_syms.c
index 79ea05f..52eea54 100644
--- a/net/sunrpc/sunrpc_syms.c
+++ b/net/sunrpc/sunrpc_syms.c
@@ -72,7 +72,6 @@ EXPORT_SYMBOL(svc_drop);
 EXPORT_SYMBOL(svc_process);
 EXPORT_SYMBOL(svc_recv);
 EXPORT_SYMBOL(svc_wake_up);
-EXPORT_SYMBOL(svc_makesock);
 EXPORT_SYMBOL(svc_reserve);
 EXPORT_SYMBOL(svc_auth_register);
 EXPORT_SYMBOL(auth_domain_lookup);
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index 9d99490..f9fc03f 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -1993,28 +1993,6 @@ void svc_force_close_socket(struct svc_sock *svsk)
 	svc_close_socket(svsk);
 }
 
-/**
- * svc_makesock - Make a socket for nfsd and lockd
- * @serv: RPC server structure
- * @protocol: transport protocol to use
- * @port: port to use
- * @flags: requested socket characteristics
- *
- */
-int svc_makesock(struct svc_serv *serv, int protocol, unsigned short port,
-			int flags)
-{
-	dprintk("svc: creating socket proto = %d\n", protocol);
-	switch (protocol) {
-	case IPPROTO_TCP:
-		return svc_create_xprt(serv, "tcp", port, flags);
-	case IPPROTO_UDP:
-		return svc_create_xprt(serv, "udp", port, flags);
-	default:
-		return -EINVAL;
-	}
-}
-
 /*
  * Handle defer and revisit of requests
  */

[PATCH 14/38] svc: Change sk_inuse to a kref

[Tom Tucker]

Change the atomic_t reference count to a kref and move it to the 
transport indepenent svc_xprt structure. Change the reference count
wrapper names to be generic.

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 include/linux/sunrpc/svc_xprt.h |    8 ++++++
 include/linux/sunrpc/svcsock.h  |    1 -
 net/sunrpc/svc_xprt.c           |   17 ++++++++++++
 net/sunrpc/svcsock.c            |   54 +++++++++++++++------------------------
 4 files changed, 46 insertions(+), 34 deletions(-)

diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
index 3f4a1df..eb801ad 100644
--- a/include/linux/sunrpc/svc_xprt.h
+++ b/include/linux/sunrpc/svc_xprt.h
@@ -8,6 +8,7 @@
 #define SUNRPC_SVC_XPRT_H
 
 #include <linux/sunrpc/svc.h>
+#include <linux/module.h>
 
 struct svc_xprt_ops {
 	struct svc_xprt	*(*xpo_create)(struct svc_serv *,
@@ -34,11 +35,18 @@ struct svc_xprt_class {
 struct svc_xprt {
 	struct svc_xprt_class	*xpt_class;
 	struct svc_xprt_ops	*xpt_ops;
+	struct kref		xpt_ref;
 };
 
 int	svc_reg_xprt_class(struct svc_xprt_class *);
 int	svc_unreg_xprt_class(struct svc_xprt_class *);
 void	svc_xprt_init(struct svc_xprt_class *, struct svc_xprt *);
 int	svc_create_xprt(struct svc_serv *, char *, unsigned short, int);
+void	svc_xprt_put(struct svc_xprt *xprt);
+
+static inline void svc_xprt_get(struct svc_xprt *xprt)
+{
+	kref_get(&xprt->xpt_ref);
+}
 
 #endif /* SUNRPC_SVC_XPRT_H */
diff --git a/include/linux/sunrpc/svcsock.h b/include/linux/sunrpc/svcsock.h
index 3181d9d..ba07d50 100644
--- a/include/linux/sunrpc/svcsock.h
+++ b/include/linux/sunrpc/svcsock.h
@@ -24,7 +24,6 @@ struct svc_sock {
 
 	struct svc_pool *	sk_pool;	/* current pool iff queued */
 	struct svc_serv *	sk_server;	/* service for this socket */
-	atomic_t		sk_inuse;	/* use count */
 	unsigned long		sk_flags;
 #define	SK_BUSY		0			/* enqueued/receiving */
 #define	SK_CONN		1			/* conn pending */
diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
index 9136da4..43418cf 100644
--- a/net/sunrpc/svc_xprt.c
+++ b/net/sunrpc/svc_xprt.c
@@ -82,6 +82,22 @@ int svc_unreg_xprt_class(struct svc_xprt_class *xcl)
 }
 EXPORT_SYMBOL_GPL(svc_unreg_xprt_class);
 
+static void svc_xprt_free(struct kref *kref)
+{
+	struct svc_xprt *xprt =
+		container_of(kref, struct svc_xprt, xpt_ref);
+	struct module *owner = xprt->xpt_class->xcl_owner;
+	BUG_ON(atomic_read(&kref->refcount));
+	xprt->xpt_ops->xpo_free(xprt);
+	module_put(owner);
+}
+
+void svc_xprt_put(struct svc_xprt *xprt)
+{
+	kref_put(&xprt->xpt_ref, svc_xprt_free);
+}
+EXPORT_SYMBOL_GPL(svc_xprt_put);
+
 /*
  * Called by transport drivers to initialize the transport independent
  * portion of the transport instance.
@@ -91,6 +107,7 @@ void svc_xprt_init(struct svc_xprt_class *xcl, struct svc_xprt *xprt)
 	memset(xprt, 0, sizeof(*xprt));
 	xprt->xpt_class = xcl;
 	xprt->xpt_ops = xcl->xcl_ops;
+	kref_init(&xprt->xpt_ref);
 }
 EXPORT_SYMBOL_GPL(svc_xprt_init);
 
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index f9fc03f..8e355cc 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -66,8 +66,8 @@
  *		after a clear, the socket must be read/accepted
  *		 if this succeeds, it must be set again.
  *	SK_CLOSE can set at any time. It is never cleared.
- *      sk_inuse contains a bias of '1' until SK_DEAD is set.
- *             so when sk_inuse hits zero, we know the socket is dead
+ *      xpt_ref contains a bias of '1' until SK_DEAD is set.
+ *             so when xprt_ref hits zero, we know the transport is dead
  *             and no-one is using it.
  *      SK_DEAD can only be set while SK_BUSY is held which ensures
  *             no other thread will be using the socket or will try to
@@ -285,7 +285,7 @@ svc_sock_enqueue(struct svc_sock *svsk)
 				"svc_sock_enqueue: server %p, rq_sock=%p!\n",
 				rqstp, rqstp->rq_sock);
 		rqstp->rq_sock = svsk;
-		atomic_inc(&svsk->sk_inuse);
+		svc_xprt_get(&svsk->sk_xprt);
 		rqstp->rq_reserved = serv->sv_max_mesg;
 		atomic_add(rqstp->rq_reserved, &svsk->sk_reserved);
 		BUG_ON(svsk->sk_pool != pool);
@@ -316,7 +316,7 @@ svc_sock_dequeue(struct svc_pool *pool)
 	list_del_init(&svsk->sk_ready);
 
 	dprintk("svc: socket %p dequeued, inuse=%d\n",
-		svsk->sk_sk, atomic_read(&svsk->sk_inuse));
+		svsk->sk_sk, atomic_read(&svsk->sk_xprt.xpt_ref.refcount));
 
 	return svsk;
 }
@@ -359,19 +359,6 @@ void svc_reserve(struct svc_rqst *rqstp, int space)
 	}
 }
 
-/*
- * Release a socket after use.
- */
-static inline void
-svc_sock_put(struct svc_sock *svsk)
-{
-	if (atomic_dec_and_test(&svsk->sk_inuse)) {
-		BUG_ON(!test_bit(SK_DEAD, &svsk->sk_flags));
-		module_put(svsk->sk_xprt.xpt_class->xcl_owner);
-		svsk->sk_xprt.xpt_ops->xpo_free(&svsk->sk_xprt);
-	}
-}
-
 static void
 svc_sock_release(struct svc_rqst *rqstp)
 {
@@ -398,7 +385,7 @@ svc_sock_release(struct svc_rqst *rqstp)
 	svc_reserve(rqstp, 0);
 	rqstp->rq_sock = NULL;
 
-	svc_sock_put(svsk);
+	svc_xprt_put(&svsk->sk_xprt);
 }
 
 /*
@@ -1490,13 +1477,13 @@ static void svc_check_conn_limits(struct svc_serv *serv)
 					  struct svc_sock,
 					  sk_list);
 			set_bit(SK_CLOSE, &svsk->sk_flags);
-			atomic_inc(&svsk->sk_inuse);
+			svc_xprt_get(&svsk->sk_xprt);
 		}
 		spin_unlock_bh(&serv->sv_lock);
 
 		if (svsk) {
 			svc_sock_enqueue(svsk);
-			svc_sock_put(svsk);
+			svc_xprt_put(&svsk->sk_xprt);
 		}
 	}
 }
@@ -1561,7 +1548,7 @@ svc_recv(struct svc_rqst *rqstp, long timeout)
 	spin_lock_bh(&pool->sp_lock);
 	if ((svsk = svc_sock_dequeue(pool)) != NULL) {
 		rqstp->rq_sock = svsk;
-		atomic_inc(&svsk->sk_inuse);
+		svc_xprt_get(&svsk->sk_xprt);
 		rqstp->rq_reserved = serv->sv_max_mesg;
 		atomic_add(rqstp->rq_reserved, &svsk->sk_reserved);
 	} else {
@@ -1610,7 +1597,8 @@ svc_recv(struct svc_rqst *rqstp, long timeout)
 		svc_sock_received(svsk);
 	} else {
 		dprintk("svc: server %p, pool %u, socket %p, inuse=%d\n",
-			rqstp, pool->sp_id, svsk, atomic_read(&svsk->sk_inuse));
+			rqstp, pool->sp_id, svsk,
+			atomic_read(&svsk->sk_xprt.xpt_ref.refcount));
 		len = svsk->sk_xprt.xpt_ops->xpo_recvfrom(rqstp);
 		dprintk("svc: got len=%d\n", len);
 	}
@@ -1707,9 +1695,10 @@ svc_age_temp_sockets(unsigned long closure)
 
 		if (!test_and_set_bit(SK_OLD, &svsk->sk_flags))
 			continue;
-		if (atomic_read(&svsk->sk_inuse) > 1 || test_bit(SK_BUSY, &svsk->sk_flags))
+		if (atomic_read(&svsk->sk_xprt.xpt_ref.refcount) > 1
+		    || test_bit(SK_BUSY, &svsk->sk_flags))
 			continue;
-		atomic_inc(&svsk->sk_inuse);
+		svc_xprt_get(&svsk->sk_xprt);
 		list_move(le, &to_be_aged);
 		set_bit(SK_CLOSE, &svsk->sk_flags);
 		set_bit(SK_DETACHED, &svsk->sk_flags);
@@ -1727,7 +1716,7 @@ svc_age_temp_sockets(unsigned long closure)
 
 		/* a thread will dequeue and close it soon */
 		svc_sock_enqueue(svsk);
-		svc_sock_put(svsk);
+		svc_xprt_put(&svsk->sk_xprt);
 	}
 
 	mod_timer(&serv->sv_temptimer, jiffies + svc_conn_age_period * HZ);
@@ -1772,7 +1761,6 @@ static struct svc_sock *svc_setup_socket(struct svc_serv *serv,
 	svsk->sk_odata = inet->sk_data_ready;
 	svsk->sk_owspace = inet->sk_write_space;
 	svsk->sk_server = serv;
-	atomic_set(&svsk->sk_inuse, 1);
 	svsk->sk_lastrecv = get_seconds();
 	spin_lock_init(&svsk->sk_lock);
 	INIT_LIST_HEAD(&svsk->sk_deferred);
@@ -1958,8 +1946,8 @@ svc_delete_socket(struct svc_sock *svsk)
 	 * is about to be destroyed (in svc_destroy).
 	 */
 	if (!test_and_set_bit(SK_DEAD, &svsk->sk_flags)) {
-		BUG_ON(atomic_read(&svsk->sk_inuse)<2);
-		atomic_dec(&svsk->sk_inuse);
+		BUG_ON(atomic_read(&svsk->sk_xprt.xpt_ref.refcount) < 2);
+		svc_xprt_put(&svsk->sk_xprt);
 		if (test_bit(SK_TEMP, &svsk->sk_flags))
 			serv->sv_tmpcnt--;
 	}
@@ -1974,10 +1962,10 @@ static void svc_close_socket(struct svc_sock *svsk)
 		/* someone else will have to effect the close */
 		return;
 
-	atomic_inc(&svsk->sk_inuse);
+	svc_xprt_get(&svsk->sk_xprt);
 	svc_delete_socket(svsk);
 	clear_bit(SK_BUSY, &svsk->sk_flags);
-	svc_sock_put(svsk);
+	svc_xprt_put(&svsk->sk_xprt);
 }
 
 void svc_force_close_socket(struct svc_sock *svsk)
@@ -2003,7 +1991,7 @@ static void svc_revisit(struct cache_deferred_req *dreq, int too_many)
 	struct svc_sock *svsk;
 
 	if (too_many) {
-		svc_sock_put(dr->svsk);
+		svc_xprt_put(&dr->svsk->sk_xprt);
 		kfree(dr);
 		return;
 	}
@@ -2015,7 +2003,7 @@ static void svc_revisit(struct cache_deferred_req *dreq, int too_many)
 	spin_unlock(&svsk->sk_lock);
 	set_bit(SK_DEFERRED, &svsk->sk_flags);
 	svc_sock_enqueue(svsk);
-	svc_sock_put(svsk);
+	svc_xprt_put(&svsk->sk_xprt);
 }
 
 static struct cache_deferred_req *
@@ -2045,7 +2033,7 @@ svc_defer(struct cache_req *req)
 		dr->argslen = rqstp->rq_arg.len >> 2;
 		memcpy(dr->args, rqstp->rq_arg.head[0].iov_base-skip, dr->argslen<<2);
 	}
-	atomic_inc(&rqstp->rq_sock->sk_inuse);
+	svc_xprt_get(rqstp->rq_xprt);
 	dr->svsk = rqstp->rq_sock;
 
 	dr->handle.revisit = svc_revisit;

Re: [PATCH 14/38] svc: Change sk_inuse to a kref

[J. Bruce Fields]

On Tue, Dec 11, 2007 at 05:32:25PM -0600, Tom Tucker wrote:
> 
> Change the atomic_t reference count to a kref and move it to the 
> transport indepenent svc_xprt structure. Change the reference count
> wrapper names to be generic.
> 
> Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
> ---
> 
>  include/linux/sunrpc/svc_xprt.h |    8 ++++++
>  include/linux/sunrpc/svcsock.h  |    1 -
>  net/sunrpc/svc_xprt.c           |   17 ++++++++++++
>  net/sunrpc/svcsock.c            |   54 +++++++++++++++------------------------
>  4 files changed, 46 insertions(+), 34 deletions(-)
> 
> diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
> index 3f4a1df..eb801ad 100644
> --- a/include/linux/sunrpc/svc_xprt.h
> +++ b/include/linux/sunrpc/svc_xprt.h
> @@ -8,6 +8,7 @@
>  #define SUNRPC_SVC_XPRT_H
>  
>  #include <linux/sunrpc/svc.h>
> +#include <linux/module.h>
>  
>  struct svc_xprt_ops {
>  	struct svc_xprt	*(*xpo_create)(struct svc_serv *,
> @@ -34,11 +35,18 @@ struct svc_xprt_class {
>  struct svc_xprt {
>  	struct svc_xprt_class	*xpt_class;
>  	struct svc_xprt_ops	*xpt_ops;
> +	struct kref		xpt_ref;
>  };
>  
>  int	svc_reg_xprt_class(struct svc_xprt_class *);
>  int	svc_unreg_xprt_class(struct svc_xprt_class *);
>  void	svc_xprt_init(struct svc_xprt_class *, struct svc_xprt *);
>  int	svc_create_xprt(struct svc_serv *, char *, unsigned short, int);
> +void	svc_xprt_put(struct svc_xprt *xprt);
> +
> +static inline void svc_xprt_get(struct svc_xprt *xprt)
> +{
> +	kref_get(&xprt->xpt_ref);
> +}
>  
>  #endif /* SUNRPC_SVC_XPRT_H */
> diff --git a/include/linux/sunrpc/svcsock.h b/include/linux/sunrpc/svcsock.h
> index 3181d9d..ba07d50 100644
> --- a/include/linux/sunrpc/svcsock.h
> +++ b/include/linux/sunrpc/svcsock.h
> @@ -24,7 +24,6 @@ struct svc_sock {
>  
>  	struct svc_pool *	sk_pool;	/* current pool iff queued */
>  	struct svc_serv *	sk_server;	/* service for this socket */
> -	atomic_t		sk_inuse;	/* use count */
>  	unsigned long		sk_flags;
>  #define	SK_BUSY		0			/* enqueued/receiving */
>  #define	SK_CONN		1			/* conn pending */
> diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
> index 9136da4..43418cf 100644
> --- a/net/sunrpc/svc_xprt.c
> +++ b/net/sunrpc/svc_xprt.c
> @@ -82,6 +82,22 @@ int svc_unreg_xprt_class(struct svc_xprt_class *xcl)
>  }
>  EXPORT_SYMBOL_GPL(svc_unreg_xprt_class);
>  
> +static void svc_xprt_free(struct kref *kref)
> +{
> +	struct svc_xprt *xprt =
> +		container_of(kref, struct svc_xprt, xpt_ref);
> +	struct module *owner = xprt->xpt_class->xcl_owner;
> +	BUG_ON(atomic_read(&kref->refcount));

All this BUG_ON() is doing is checking for what looks like a bizarrely
unlikely bug in kref_put().  The BUG_ON from the original code is:

	BUG_ON(!test_bit(SK_DEAD, &svsk->sk_flags));

But I don't have any strong feelings as to whether that's necessary.

--b.

Re: [PATCH 14/38] svc: Change sk_inuse to a kref

[Tom Tucker]

On Fri, 2007-12-14 at 15:52 -0500, J. Bruce Fields wrote:
> On Tue, Dec 11, 2007 at 05:32:25PM -0600, Tom Tucker wrote:
> > 
> > Change the atomic_t reference count to a kref and move it to the 
> > transport indepenent svc_xprt structure. Change the reference count
> > wrapper names to be generic.
> > 
> > Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
> > ---
> > 
> >  include/linux/sunrpc/svc_xprt.h |    8 ++++++
> >  include/linux/sunrpc/svcsock.h  |    1 -
> >  net/sunrpc/svc_xprt.c           |   17 ++++++++++++
> >  net/sunrpc/svcsock.c            |   54 +++++++++++++++------------------------
> >  4 files changed, 46 insertions(+), 34 deletions(-)
> > 
> > diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
> > index 3f4a1df..eb801ad 100644
> > --- a/include/linux/sunrpc/svc_xprt.h
> > +++ b/include/linux/sunrpc/svc_xprt.h
> > @@ -8,6 +8,7 @@
> >  #define SUNRPC_SVC_XPRT_H
> >  
> >  #include <linux/sunrpc/svc.h>
> > +#include <linux/module.h>
> >  
> >  struct svc_xprt_ops {
> >  	struct svc_xprt	*(*xpo_create)(struct svc_serv *,
> > @@ -34,11 +35,18 @@ struct svc_xprt_class {
> >  struct svc_xprt {
> >  	struct svc_xprt_class	*xpt_class;
> >  	struct svc_xprt_ops	*xpt_ops;
> > +	struct kref		xpt_ref;
> >  };
> >  
> >  int	svc_reg_xprt_class(struct svc_xprt_class *);
> >  int	svc_unreg_xprt_class(struct svc_xprt_class *);
> >  void	svc_xprt_init(struct svc_xprt_class *, struct svc_xprt *);
> >  int	svc_create_xprt(struct svc_serv *, char *, unsigned short, int);
> > +void	svc_xprt_put(struct svc_xprt *xprt);
> > +
> > +static inline void svc_xprt_get(struct svc_xprt *xprt)
> > +{
> > +	kref_get(&xprt->xpt_ref);
> > +}
> >  
> >  #endif /* SUNRPC_SVC_XPRT_H */
> > diff --git a/include/linux/sunrpc/svcsock.h b/include/linux/sunrpc/svcsock.h
> > index 3181d9d..ba07d50 100644
> > --- a/include/linux/sunrpc/svcsock.h
> > +++ b/include/linux/sunrpc/svcsock.h
> > @@ -24,7 +24,6 @@ struct svc_sock {
> >  
> >  	struct svc_pool *	sk_pool;	/* current pool iff queued */
> >  	struct svc_serv *	sk_server;	/* service for this socket */
> > -	atomic_t		sk_inuse;	/* use count */
> >  	unsigned long		sk_flags;
> >  #define	SK_BUSY		0			/* enqueued/receiving */
> >  #define	SK_CONN		1			/* conn pending */
> > diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
> > index 9136da4..43418cf 100644
> > --- a/net/sunrpc/svc_xprt.c
> > +++ b/net/sunrpc/svc_xprt.c
> > @@ -82,6 +82,22 @@ int svc_unreg_xprt_class(struct svc_xprt_class *xcl)
> >  }
> >  EXPORT_SYMBOL_GPL(svc_unreg_xprt_class);
> >  
> > +static void svc_xprt_free(struct kref *kref)
> > +{
> > +	struct svc_xprt *xprt =
> > +		container_of(kref, struct svc_xprt, xpt_ref);
> > +	struct module *owner = xprt->xpt_class->xcl_owner;
> > +	BUG_ON(atomic_read(&kref->refcount));
> 
> All this BUG_ON() is doing is checking for what looks like a bizarrely
> unlikely bug in kref_put().  The BUG_ON from the original code is:
> 
> 	BUG_ON(!test_bit(SK_DEAD, &svsk->sk_flags));
> 
> But I don't have any strong feelings as to whether that's necessary.
> 

It's not. 

> --b.

[PATCH 15/38] svc: Move sk_flags to the svc_xprt structure

[Tom Tucker]

This functionally trivial change moves the transport independent sk_flags 
field to the transport independent svc_xprt structure.

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 include/linux/sunrpc/svc_xprt.h |   12 +++
 include/linux/sunrpc/svcsock.h  |   13 ---
 net/sunrpc/svcsock.c            |  149 ++++++++++++++++++++-------------------
 3 files changed, 87 insertions(+), 87 deletions(-)

diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
index eb801ad..f391d21 100644
--- a/include/linux/sunrpc/svc_xprt.h
+++ b/include/linux/sunrpc/svc_xprt.h
@@ -36,6 +36,18 @@ struct svc_xprt {
 	struct svc_xprt_class	*xpt_class;
 	struct svc_xprt_ops	*xpt_ops;
 	struct kref		xpt_ref;
+	unsigned long		xpt_flags;
+#define	XPT_BUSY	0		/* enqueued/receiving */
+#define	XPT_CONN	1		/* conn pending */
+#define	XPT_CLOSE	2		/* dead or dying */
+#define	XPT_DATA	3		/* data pending */
+#define	XPT_TEMP	4		/* connected transport */
+#define	XPT_DEAD	6		/* transport closed */
+#define	XPT_CHNGBUF	7		/* need to change snd/rcv buf sizes */
+#define	XPT_DEFERRED	8		/* deferred request pending */
+#define	XPT_OLD		9		/* used for xprt aging mark+sweep */
+#define	XPT_DETACHED	10		/* detached from tempsocks list */
+#define XPT_LISTENER	11		/* listening endpoint */
 };
 
 int	svc_reg_xprt_class(struct svc_xprt_class *);
diff --git a/include/linux/sunrpc/svcsock.h b/include/linux/sunrpc/svcsock.h
index ba07d50..b8a8496 100644
--- a/include/linux/sunrpc/svcsock.h
+++ b/include/linux/sunrpc/svcsock.h
@@ -24,19 +24,6 @@ struct svc_sock {
 
 	struct svc_pool *	sk_pool;	/* current pool iff queued */
 	struct svc_serv *	sk_server;	/* service for this socket */
-	unsigned long		sk_flags;
-#define	SK_BUSY		0			/* enqueued/receiving */
-#define	SK_CONN		1			/* conn pending */
-#define	SK_CLOSE	2			/* dead or dying */
-#define	SK_DATA		3			/* data pending */
-#define	SK_TEMP		4			/* temp (TCP) socket */
-#define	SK_DEAD		6			/* socket closed */
-#define	SK_CHNGBUF	7			/* need to change snd/rcv buffer sizes */
-#define	SK_DEFERRED	8			/* request on sk_deferred */
-#define	SK_OLD		9			/* used for temp socket aging mark+sweep */
-#define	SK_DETACHED	10			/* detached from tempsocks list */
-#define SK_LISTENER	11			/* listening endpoint */
-
 	atomic_t    	    	sk_reserved;	/* space on outq that is reserved */
 
 	spinlock_t		sk_lock;	/* protects sk_deferred and
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index 8e355cc..3a625c4 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -56,22 +56,23 @@
  *	BKL protects svc_serv->sv_nrthread.
  *	svc_sock->sk_lock protects the svc_sock->sk_deferred list
  *             and the ->sk_info_authunix cache.
- *	svc_sock->sk_flags.SK_BUSY prevents a svc_sock being enqueued multiply.
+ *	svc_sock->sk_xprt.xpt_flags.XPT_BUSY prevents a svc_sock being
+ *	enqueued multiply.
  *
  *	Some flags can be set to certain values at any time
  *	providing that certain rules are followed:
  *
- *	SK_CONN, SK_DATA, can be set or cleared at any time.
+ *	XPT_CONN, XPT_DATA, can be set or cleared at any time.
  *		after a set, svc_sock_enqueue must be called.
  *		after a clear, the socket must be read/accepted
  *		 if this succeeds, it must be set again.
- *	SK_CLOSE can set at any time. It is never cleared.
- *      xpt_ref contains a bias of '1' until SK_DEAD is set.
+ *	XPT_CLOSE can set at any time. It is never cleared.
+ *      xpt_ref contains a bias of '1' until XPT_DEAD is set.
  *             so when xprt_ref hits zero, we know the transport is dead
  *             and no-one is using it.
- *      SK_DEAD can only be set while SK_BUSY is held which ensures
+ *      XPT_DEAD can only be set while XPT_BUSY is held which ensures
  *             no other thread will be using the socket or will try to
- *	       set SK_DEAD.
+ *	       set XPT_DEAD.
  *
  */
 
@@ -219,10 +220,10 @@ svc_sock_enqueue(struct svc_sock *svsk)
 	struct svc_rqst	*rqstp;
 	int cpu;
 
-	if (!(svsk->sk_flags &
-	      ( (1<<SK_CONN)|(1<<SK_DATA)|(1<<SK_CLOSE)|(1<<SK_DEFERRED)) ))
+	if (!(svsk->sk_xprt.xpt_flags &
+	      ((1<<XPT_CONN)|(1<<XPT_DATA)|(1<<XPT_CLOSE)|(1<<XPT_DEFERRED))))
 		return;
-	if (test_bit(SK_DEAD, &svsk->sk_flags))
+	if (test_bit(XPT_DEAD, &svsk->sk_xprt.xpt_flags))
 		return;
 
 	cpu = get_cpu();
@@ -236,7 +237,7 @@ svc_sock_enqueue(struct svc_sock *svsk)
 		printk(KERN_ERR
 			"svc_sock_enqueue: threads and sockets both waiting??\n");
 
-	if (test_bit(SK_DEAD, &svsk->sk_flags)) {
+	if (test_bit(XPT_DEAD, &svsk->sk_xprt.xpt_flags)) {
 		/* Don't enqueue dead sockets */
 		dprintk("svc: socket %p is dead, not enqueued\n", svsk->sk_sk);
 		goto out_unlock;
@@ -244,10 +245,10 @@ svc_sock_enqueue(struct svc_sock *svsk)
 
 	/* Mark socket as busy. It will remain in this state until the
 	 * server has processed all pending data and put the socket back
-	 * on the idle list.  We update SK_BUSY atomically because
+	 * on the idle list.  We update XPT_BUSY atomically because
 	 * it also guards against trying to enqueue the svc_sock twice.
 	 */
-	if (test_and_set_bit(SK_BUSY, &svsk->sk_flags)) {
+	if (test_and_set_bit(XPT_BUSY, &svsk->sk_xprt.xpt_flags)) {
 		/* Don't enqueue socket while already enqueued */
 		dprintk("svc: socket %p busy, not enqueued\n", svsk->sk_sk);
 		goto out_unlock;
@@ -256,11 +257,11 @@ svc_sock_enqueue(struct svc_sock *svsk)
 	svsk->sk_pool = pool;
 
 	/* Handle pending connection */
-	if (test_bit(SK_CONN, &svsk->sk_flags))
+	if (test_bit(XPT_CONN, &svsk->sk_xprt.xpt_flags))
 		goto process;
 
 	/* Handle close in-progress */
-	if (test_bit(SK_CLOSE, &svsk->sk_flags))
+	if (test_bit(XPT_CLOSE, &svsk->sk_xprt.xpt_flags))
 		goto process;
 
 	/* Check if we have space to reply to a request */
@@ -268,7 +269,7 @@ svc_sock_enqueue(struct svc_sock *svsk)
 		/* Don't enqueue while not enough space for reply */
 		dprintk("svc: no write space, socket %p  not enqueued\n", svsk);
 		svsk->sk_pool = NULL;
-		clear_bit(SK_BUSY, &svsk->sk_flags);
+		clear_bit(XPT_BUSY, &svsk->sk_xprt.xpt_flags);
 		goto out_unlock;
 	}
 
@@ -324,14 +325,14 @@ svc_sock_dequeue(struct svc_pool *pool)
 /*
  * Having read something from a socket, check whether it
  * needs to be re-enqueued.
- * Note: SK_DATA only gets cleared when a read-attempt finds
+ * Note: XPT_DATA only gets cleared when a read-attempt finds
  * no (or insufficient) data.
  */
 static inline void
 svc_sock_received(struct svc_sock *svsk)
 {
 	svsk->sk_pool = NULL;
-	clear_bit(SK_BUSY, &svsk->sk_flags);
+	clear_bit(XPT_BUSY, &svsk->sk_xprt.xpt_flags);
 	svc_sock_enqueue(svsk);
 }
 
@@ -680,8 +681,9 @@ svc_udp_data_ready(struct sock *sk, int count)
 
 	if (svsk) {
 		dprintk("svc: socket %p(inet %p), count=%d, busy=%d\n",
-			svsk, sk, count, test_bit(SK_BUSY, &svsk->sk_flags));
-		set_bit(SK_DATA, &svsk->sk_flags);
+			svsk, sk, count,
+			test_bit(XPT_BUSY, &svsk->sk_xprt.xpt_flags));
+		set_bit(XPT_DATA, &svsk->sk_xprt.xpt_flags);
 		svc_sock_enqueue(svsk);
 	}
 	if (sk->sk_sleep && waitqueue_active(sk->sk_sleep))
@@ -698,7 +700,7 @@ svc_write_space(struct sock *sk)
 
 	if (svsk) {
 		dprintk("svc: socket %p(inet %p), write_space busy=%d\n",
-			svsk, sk, test_bit(SK_BUSY, &svsk->sk_flags));
+			svsk, sk, test_bit(XPT_BUSY, &svsk->sk_xprt.xpt_flags));
 		svc_sock_enqueue(svsk);
 	}
 
@@ -748,7 +750,7 @@ svc_udp_recvfrom(struct svc_rqst *rqstp)
 		.msg_flags = MSG_DONTWAIT,
 	};
 
-	if (test_and_clear_bit(SK_CHNGBUF, &svsk->sk_flags))
+	if (test_and_clear_bit(XPT_CHNGBUF, &svsk->sk_xprt.xpt_flags))
 	    /* udp sockets need large rcvbuf as all pending
 	     * requests are still in that buffer.  sndbuf must
 	     * also be large enough that there is enough space
@@ -766,7 +768,7 @@ svc_udp_recvfrom(struct svc_rqst *rqstp)
 		return svc_deferred_recv(rqstp);
 	}
 
-	clear_bit(SK_DATA, &svsk->sk_flags);
+	clear_bit(XPT_DATA, &svsk->sk_xprt.xpt_flags);
 	skb = NULL;
 	err = kernel_recvmsg(svsk->sk_sock, &msg, NULL,
 			     0, 0, MSG_PEEK | MSG_DONTWAIT);
@@ -777,7 +779,7 @@ svc_udp_recvfrom(struct svc_rqst *rqstp)
 		if (err != -EAGAIN) {
 			/* possibly an icmp error */
 			dprintk("svc: recvfrom returned error %d\n", -err);
-			set_bit(SK_DATA, &svsk->sk_flags);
+			set_bit(XPT_DATA, &svsk->sk_xprt.xpt_flags);
 		}
 		svc_sock_received(svsk);
 		return -EAGAIN;
@@ -789,7 +791,7 @@ svc_udp_recvfrom(struct svc_rqst *rqstp)
 		   need that much accuracy */
 	}
 	svsk->sk_sk->sk_stamp = skb->tstamp;
-	set_bit(SK_DATA, &svsk->sk_flags); /* there may be more data... */
+	set_bit(XPT_DATA, &svsk->sk_xprt.xpt_flags); /* there may be more data... */
 
 	/*
 	 * Maybe more packets - kick another thread ASAP.
@@ -936,8 +938,8 @@ svc_udp_init(struct svc_sock *svsk)
 			    3 * svsk->sk_server->sv_max_mesg,
 			    3 * svsk->sk_server->sv_max_mesg);
 
-	set_bit(SK_DATA, &svsk->sk_flags); /* might have come in before data_ready set up */
-	set_bit(SK_CHNGBUF, &svsk->sk_flags);
+	set_bit(XPT_DATA, &svsk->sk_xprt.xpt_flags); /* might have come in before data_ready set up */
+	set_bit(XPT_CHNGBUF, &svsk->sk_xprt.xpt_flags);
 
 	oldfs = get_fs();
 	set_fs(KERNEL_DS);
@@ -971,7 +973,7 @@ svc_tcp_listen_data_ready(struct sock *sk, int count_unused)
 	 */
 	if (sk->sk_state == TCP_LISTEN) {
 		if (svsk) {
-			set_bit(SK_CONN, &svsk->sk_flags);
+			set_bit(XPT_CONN, &svsk->sk_xprt.xpt_flags);
 			svc_sock_enqueue(svsk);
 		} else
 			printk("svc: socket %p: no user data\n", sk);
@@ -995,7 +997,7 @@ svc_tcp_state_change(struct sock *sk)
 	if (!svsk)
 		printk("svc: socket %p: no user data\n", sk);
 	else {
-		set_bit(SK_CLOSE, &svsk->sk_flags);
+		set_bit(XPT_CLOSE, &svsk->sk_xprt.xpt_flags);
 		svc_sock_enqueue(svsk);
 	}
 	if (sk->sk_sleep && waitqueue_active(sk->sk_sleep))
@@ -1010,7 +1012,7 @@ svc_tcp_data_ready(struct sock *sk, int count)
 	dprintk("svc: socket %p TCP data ready (svsk %p)\n",
 		sk, sk->sk_user_data);
 	if (svsk) {
-		set_bit(SK_DATA, &svsk->sk_flags);
+		set_bit(XPT_DATA, &svsk->sk_xprt.xpt_flags);
 		svc_sock_enqueue(svsk);
 	}
 	if (sk->sk_sleep && waitqueue_active(sk->sk_sleep))
@@ -1050,7 +1052,7 @@ static struct svc_xprt *svc_tcp_accept(struct svc_xprt *xprt)
 	if (!sock)
 		return NULL;
 
-	clear_bit(SK_CONN, &svsk->sk_flags);
+	clear_bit(XPT_CONN, &svsk->sk_xprt.xpt_flags);
 	err = kernel_accept(sock, &newsock, O_NONBLOCK);
 	if (err < 0) {
 		if (err == -ENOMEM)
@@ -1061,8 +1063,7 @@ static struct svc_xprt *svc_tcp_accept(struct svc_xprt *xprt)
 				   serv->sv_name, -err);
 		return NULL;
 	}
-
-	set_bit(SK_CONN, &svsk->sk_flags);
+	set_bit(XPT_CONN, &svsk->sk_xprt.xpt_flags);
 
 	err = kernel_getpeername(newsock, sin, &slen);
 	if (err < 0) {
@@ -1127,16 +1128,16 @@ svc_tcp_recvfrom(struct svc_rqst *rqstp)
 	int pnum, vlen;
 
 	dprintk("svc: tcp_recv %p data %d conn %d close %d\n",
-		svsk, test_bit(SK_DATA, &svsk->sk_flags),
-		test_bit(SK_CONN, &svsk->sk_flags),
-		test_bit(SK_CLOSE, &svsk->sk_flags));
+		svsk, test_bit(XPT_DATA, &svsk->sk_xprt.xpt_flags),
+		test_bit(XPT_CONN, &svsk->sk_xprt.xpt_flags),
+		test_bit(XPT_CLOSE, &svsk->sk_xprt.xpt_flags));
 
 	if ((rqstp->rq_deferred = svc_deferred_dequeue(svsk))) {
 		svc_sock_received(svsk);
 		return svc_deferred_recv(rqstp);
 	}
 
-	if (test_and_clear_bit(SK_CHNGBUF, &svsk->sk_flags))
+	if (test_and_clear_bit(XPT_CHNGBUF, &svsk->sk_xprt.xpt_flags))
 		/* sndbuf needs to have room for one request
 		 * per thread, otherwise we can stall even when the
 		 * network isn't a bottleneck.
@@ -1153,7 +1154,7 @@ svc_tcp_recvfrom(struct svc_rqst *rqstp)
 				    (serv->sv_nrthreads+3) * serv->sv_max_mesg,
 				    3 * serv->sv_max_mesg);
 
-	clear_bit(SK_DATA, &svsk->sk_flags);
+	clear_bit(XPT_DATA, &svsk->sk_xprt.xpt_flags);
 
 	/* Receive data. If we haven't got the record length yet, get
 	 * the next four bytes. Otherwise try to gobble up as much as
@@ -1212,7 +1213,7 @@ svc_tcp_recvfrom(struct svc_rqst *rqstp)
 		return -EAGAIN;	/* record not complete */
 	}
 	len = svsk->sk_reclen;
-	set_bit(SK_DATA, &svsk->sk_flags);
+	set_bit(XPT_DATA, &svsk->sk_xprt.xpt_flags);
 
 	vec = rqstp->rq_vec;
 	vec[0] = rqstp->rq_arg.head[0];
@@ -1288,7 +1289,7 @@ svc_tcp_sendto(struct svc_rqst *rqstp)
 	reclen = htonl(0x80000000|((xbufp->len ) - 4));
 	memcpy(xbufp->head[0].iov_base, &reclen, 4);
 
-	if (test_bit(SK_DEAD, &rqstp->rq_sock->sk_flags))
+	if (test_bit(XPT_DEAD, &rqstp->rq_sock->sk_xprt.xpt_flags))
 		return -ENOTCONN;
 
 	sent = svc_sendto(rqstp, &rqstp->rq_res);
@@ -1297,7 +1298,7 @@ svc_tcp_sendto(struct svc_rqst *rqstp)
 		       rqstp->rq_sock->sk_server->sv_name,
 		       (sent<0)?"got error":"sent only",
 		       sent, xbufp->len);
-		set_bit(SK_CLOSE, &rqstp->rq_sock->sk_flags);
+		set_bit(XPT_CLOSE, &rqstp->rq_sock->sk_xprt.xpt_flags);
 		svc_sock_enqueue(rqstp->rq_sock);
 		sent = -EAGAIN;
 	}
@@ -1387,9 +1388,9 @@ svc_tcp_init(struct svc_sock *svsk)
 
 	if (sk->sk_state == TCP_LISTEN) {
 		dprintk("setting up TCP socket for listening\n");
-		set_bit(SK_LISTENER, &svsk->sk_flags);
+		set_bit(XPT_LISTENER, &svsk->sk_xprt.xpt_flags);
 		sk->sk_data_ready = svc_tcp_listen_data_ready;
-		set_bit(SK_CONN, &svsk->sk_flags);
+		set_bit(XPT_CONN, &svsk->sk_xprt.xpt_flags);
 	} else {
 		dprintk("setting up TCP socket for reading\n");
 		sk->sk_state_change = svc_tcp_state_change;
@@ -1409,10 +1410,10 @@ svc_tcp_init(struct svc_sock *svsk)
 				    3 * svsk->sk_server->sv_max_mesg,
 				    3 * svsk->sk_server->sv_max_mesg);
 
-		set_bit(SK_CHNGBUF, &svsk->sk_flags);
-		set_bit(SK_DATA, &svsk->sk_flags);
+		set_bit(XPT_CHNGBUF, &svsk->sk_xprt.xpt_flags);
+		set_bit(XPT_DATA, &svsk->sk_xprt.xpt_flags);
 		if (sk->sk_state != TCP_ESTABLISHED)
-			set_bit(SK_CLOSE, &svsk->sk_flags);
+			set_bit(XPT_CLOSE, &svsk->sk_xprt.xpt_flags);
 	}
 }
 
@@ -1429,12 +1430,12 @@ svc_sock_update_bufs(struct svc_serv *serv)
 	list_for_each(le, &serv->sv_permsocks) {
 		struct svc_sock *svsk =
 			list_entry(le, struct svc_sock, sk_list);
-		set_bit(SK_CHNGBUF, &svsk->sk_flags);
+		set_bit(XPT_CHNGBUF, &svsk->sk_xprt.xpt_flags);
 	}
 	list_for_each(le, &serv->sv_tempsocks) {
 		struct svc_sock *svsk =
 			list_entry(le, struct svc_sock, sk_list);
-		set_bit(SK_CHNGBUF, &svsk->sk_flags);
+		set_bit(XPT_CHNGBUF, &svsk->sk_xprt.xpt_flags);
 	}
 	spin_unlock_bh(&serv->sv_lock);
 }
@@ -1476,7 +1477,7 @@ static void svc_check_conn_limits(struct svc_serv *serv)
 			svsk = list_entry(serv->sv_tempsocks.prev,
 					  struct svc_sock,
 					  sk_list);
-			set_bit(SK_CLOSE, &svsk->sk_flags);
+			set_bit(XPT_CLOSE, &svsk->sk_xprt.xpt_flags);
 			svc_xprt_get(&svsk->sk_xprt);
 		}
 		spin_unlock_bh(&serv->sv_lock);
@@ -1580,10 +1581,10 @@ svc_recv(struct svc_rqst *rqstp, long timeout)
 	spin_unlock_bh(&pool->sp_lock);
 
 	len = 0;
-	if (test_bit(SK_CLOSE, &svsk->sk_flags)) {
-		dprintk("svc_recv: found SK_CLOSE\n");
+	if (test_bit(XPT_CLOSE, &svsk->sk_xprt.xpt_flags)) {
+		dprintk("svc_recv: found XPT_CLOSE\n");
 		svc_delete_socket(svsk);
-	} else if (test_bit(SK_LISTENER, &svsk->sk_flags)) {
+	} else if (test_bit(XPT_LISTENER, &svsk->sk_xprt.xpt_flags)) {
 		struct svc_xprt *newxpt;
 		newxpt = svsk->sk_xprt.xpt_ops->xpo_accept(&svsk->sk_xprt);
 		if (newxpt) {
@@ -1610,7 +1611,7 @@ svc_recv(struct svc_rqst *rqstp, long timeout)
 		return -EAGAIN;
 	}
 	svsk->sk_lastrecv = get_seconds();
-	clear_bit(SK_OLD, &svsk->sk_flags);
+	clear_bit(XPT_OLD, &svsk->sk_xprt.xpt_flags);
 
 	rqstp->rq_secure = svc_port_is_privileged(svc_addr(rqstp));
 	rqstp->rq_chandle.defer = svc_defer;
@@ -1657,7 +1658,7 @@ svc_send(struct svc_rqst *rqstp)
 
 	/* Grab svsk->sk_mutex to serialize outgoing data. */
 	mutex_lock(&svsk->sk_mutex);
-	if (test_bit(SK_DEAD, &svsk->sk_flags))
+	if (test_bit(XPT_DEAD, &svsk->sk_xprt.xpt_flags))
 		len = -ENOTCONN;
 	else
 		len = svsk->sk_xprt.xpt_ops->xpo_sendto(rqstp);
@@ -1693,21 +1694,21 @@ svc_age_temp_sockets(unsigned long closure)
 	list_for_each_safe(le, next, &serv->sv_tempsocks) {
 		svsk = list_entry(le, struct svc_sock, sk_list);
 
-		if (!test_and_set_bit(SK_OLD, &svsk->sk_flags))
+		if (!test_and_set_bit(XPT_OLD, &svsk->sk_xprt.xpt_flags))
 			continue;
 		if (atomic_read(&svsk->sk_xprt.xpt_ref.refcount) > 1
-		    || test_bit(SK_BUSY, &svsk->sk_flags))
+		    || test_bit(XPT_BUSY, &svsk->sk_xprt.xpt_flags))
 			continue;
 		svc_xprt_get(&svsk->sk_xprt);
 		list_move(le, &to_be_aged);
-		set_bit(SK_CLOSE, &svsk->sk_flags);
-		set_bit(SK_DETACHED, &svsk->sk_flags);
+		set_bit(XPT_CLOSE, &svsk->sk_xprt.xpt_flags);
+		set_bit(XPT_DETACHED, &svsk->sk_xprt.xpt_flags);
 	}
 	spin_unlock_bh(&serv->sv_lock);
 
 	while (!list_empty(&to_be_aged)) {
 		le = to_be_aged.next;
-		/* fiddling the sk_list node is safe 'cos we're SK_DETACHED */
+		/* fiddling the sk_list node is safe 'cos we're XPT_DETACHED */
 		list_del_init(le);
 		svsk = list_entry(le, struct svc_sock, sk_list);
 
@@ -1753,7 +1754,7 @@ static struct svc_sock *svc_setup_socket(struct svc_serv *serv,
 		return NULL;
 	}
 
-	set_bit(SK_BUSY, &svsk->sk_flags);
+	set_bit(XPT_BUSY, &svsk->sk_xprt.xpt_flags);
 	inet->sk_user_data = svsk;
 	svsk->sk_sock = sock;
 	svsk->sk_sk = inet;
@@ -1775,7 +1776,7 @@ static struct svc_sock *svc_setup_socket(struct svc_serv *serv,
 
 	spin_lock_bh(&serv->sv_lock);
 	if (is_temporary) {
-		set_bit(SK_TEMP, &svsk->sk_flags);
+		set_bit(XPT_TEMP, &svsk->sk_xprt.xpt_flags);
 		list_add(&svsk->sk_list, &serv->sv_tempsocks);
 		serv->sv_tmpcnt++;
 		if (serv->sv_temptimer.function == NULL) {
@@ -1786,7 +1787,7 @@ static struct svc_sock *svc_setup_socket(struct svc_serv *serv,
 					jiffies + svc_conn_age_period * HZ);
 		}
 	} else {
-		clear_bit(SK_TEMP, &svsk->sk_flags);
+		clear_bit(XPT_TEMP, &svsk->sk_xprt.xpt_flags);
 		list_add(&svsk->sk_list, &serv->sv_permsocks);
 	}
 	spin_unlock_bh(&serv->sv_lock);
@@ -1936,7 +1937,7 @@ svc_delete_socket(struct svc_sock *svsk)
 
 	spin_lock_bh(&serv->sv_lock);
 
-	if (!test_and_set_bit(SK_DETACHED, &svsk->sk_flags))
+	if (!test_and_set_bit(XPT_DETACHED, &svsk->sk_xprt.xpt_flags))
 		list_del_init(&svsk->sk_list);
 	/*
 	 * We used to delete the svc_sock from whichever list
@@ -1945,10 +1946,10 @@ svc_delete_socket(struct svc_sock *svsk)
 	 * while still attached to a queue, the queue itself
 	 * is about to be destroyed (in svc_destroy).
 	 */
-	if (!test_and_set_bit(SK_DEAD, &svsk->sk_flags)) {
+	if (!test_and_set_bit(XPT_DEAD, &svsk->sk_xprt.xpt_flags)) {
 		BUG_ON(atomic_read(&svsk->sk_xprt.xpt_ref.refcount) < 2);
 		svc_xprt_put(&svsk->sk_xprt);
-		if (test_bit(SK_TEMP, &svsk->sk_flags))
+		if (test_bit(XPT_TEMP, &svsk->sk_xprt.xpt_flags))
 			serv->sv_tmpcnt--;
 	}
 
@@ -1957,26 +1958,26 @@ svc_delete_socket(struct svc_sock *svsk)
 
 static void svc_close_socket(struct svc_sock *svsk)
 {
-	set_bit(SK_CLOSE, &svsk->sk_flags);
-	if (test_and_set_bit(SK_BUSY, &svsk->sk_flags))
+	set_bit(XPT_CLOSE, &svsk->sk_xprt.xpt_flags);
+	if (test_and_set_bit(XPT_BUSY, &svsk->sk_xprt.xpt_flags))
 		/* someone else will have to effect the close */
 		return;
 
 	svc_xprt_get(&svsk->sk_xprt);
 	svc_delete_socket(svsk);
-	clear_bit(SK_BUSY, &svsk->sk_flags);
+	clear_bit(XPT_BUSY, &svsk->sk_xprt.xpt_flags);
 	svc_xprt_put(&svsk->sk_xprt);
 }
 
 void svc_force_close_socket(struct svc_sock *svsk)
 {
-	set_bit(SK_CLOSE, &svsk->sk_flags);
-	if (test_bit(SK_BUSY, &svsk->sk_flags)) {
+	set_bit(XPT_CLOSE, &svsk->sk_xprt.xpt_flags);
+	if (test_bit(XPT_BUSY, &svsk->sk_xprt.xpt_flags)) {
 		/* Waiting to be processed, but no threads left,
 		 * So just remove it from the waiting list
 		 */
 		list_del_init(&svsk->sk_ready);
-		clear_bit(SK_BUSY, &svsk->sk_flags);
+		clear_bit(XPT_BUSY, &svsk->sk_xprt.xpt_flags);
 	}
 	svc_close_socket(svsk);
 }
@@ -2001,7 +2002,7 @@ static void svc_revisit(struct cache_deferred_req *dreq, int too_many)
 	spin_lock(&svsk->sk_lock);
 	list_add(&dr->handle.recent, &svsk->sk_deferred);
 	spin_unlock(&svsk->sk_lock);
-	set_bit(SK_DEFERRED, &svsk->sk_flags);
+	set_bit(XPT_DEFERRED, &svsk->sk_xprt.xpt_flags);
 	svc_sock_enqueue(svsk);
 	svc_xprt_put(&svsk->sk_xprt);
 }
@@ -2064,16 +2065,16 @@ static struct svc_deferred_req *svc_deferred_dequeue(struct svc_sock *svsk)
 {
 	struct svc_deferred_req *dr = NULL;
 
-	if (!test_bit(SK_DEFERRED, &svsk->sk_flags))
+	if (!test_bit(XPT_DEFERRED, &svsk->sk_xprt.xpt_flags))
 		return NULL;
 	spin_lock(&svsk->sk_lock);
-	clear_bit(SK_DEFERRED, &svsk->sk_flags);
+	clear_bit(XPT_DEFERRED, &svsk->sk_xprt.xpt_flags);
 	if (!list_empty(&svsk->sk_deferred)) {
 		dr = list_entry(svsk->sk_deferred.next,
 				struct svc_deferred_req,
 				handle.recent);
 		list_del_init(&dr->handle.recent);
-		set_bit(SK_DEFERRED, &svsk->sk_flags);
+		set_bit(XPT_DEFERRED, &svsk->sk_xprt.xpt_flags);
 	}
 	spin_unlock(&svsk->sk_lock);
 	return dr;

[PATCH 16/38] svc: Move sk_server and sk_pool to svc_xprt

[Tom Tucker]

This is another incremental change that moves transport independent 
fields from svc_sock to the svc_xprt structure. The changes 
should be functionally null.

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 include/linux/sunrpc/svc_xprt.h |    6 +++-
 include/linux/sunrpc/svcsock.h  |    2 -
 net/sunrpc/svc_xprt.c           |    4 ++-
 net/sunrpc/svcsock.c            |   57 ++++++++++++++++++---------------------
 4 files changed, 35 insertions(+), 34 deletions(-)

diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
index f391d21..4f7dbbc 100644
--- a/include/linux/sunrpc/svc_xprt.h
+++ b/include/linux/sunrpc/svc_xprt.h
@@ -48,11 +48,15 @@ struct svc_xprt {
 #define	XPT_OLD		9		/* used for xprt aging mark+sweep */
 #define	XPT_DETACHED	10		/* detached from tempsocks list */
 #define XPT_LISTENER	11		/* listening endpoint */
+
+	struct svc_pool		*xpt_pool;	/* current pool iff queued */
+	struct svc_serv		*xpt_server;	/* service for transport */
 };
 
 int	svc_reg_xprt_class(struct svc_xprt_class *);
 int	svc_unreg_xprt_class(struct svc_xprt_class *);
-void	svc_xprt_init(struct svc_xprt_class *, struct svc_xprt *);
+void	svc_xprt_init(struct svc_xprt_class *, struct svc_xprt *,
+		      struct svc_serv *);
 int	svc_create_xprt(struct svc_serv *, char *, unsigned short, int);
 void	svc_xprt_put(struct svc_xprt *xprt);
 
diff --git a/include/linux/sunrpc/svcsock.h b/include/linux/sunrpc/svcsock.h
index b8a8496..92d4cc9 100644
--- a/include/linux/sunrpc/svcsock.h
+++ b/include/linux/sunrpc/svcsock.h
@@ -22,8 +22,6 @@ struct svc_sock {
 	struct socket *		sk_sock;	/* berkeley socket layer */
 	struct sock *		sk_sk;		/* INET layer */
 
-	struct svc_pool *	sk_pool;	/* current pool iff queued */
-	struct svc_serv *	sk_server;	/* service for this socket */
 	atomic_t    	    	sk_reserved;	/* space on outq that is reserved */
 
 	spinlock_t		sk_lock;	/* protects sk_deferred and
diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
index 43418cf..e366add 100644
--- a/net/sunrpc/svc_xprt.c
+++ b/net/sunrpc/svc_xprt.c
@@ -102,12 +102,14 @@ EXPORT_SYMBOL_GPL(svc_xprt_put);
  * Called by transport drivers to initialize the transport independent
  * portion of the transport instance.
  */
-void svc_xprt_init(struct svc_xprt_class *xcl, struct svc_xprt *xprt)
+void svc_xprt_init(struct svc_xprt_class *xcl, struct svc_xprt *xprt,
+		   struct svc_serv *serv)
 {
 	memset(xprt, 0, sizeof(*xprt));
 	xprt->xpt_class = xcl;
 	xprt->xpt_ops = xcl->xcl_ops;
 	kref_init(&xprt->xpt_ref);
+	xprt->xpt_server = serv;
 }
 EXPORT_SYMBOL_GPL(svc_xprt_init);
 
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index 3a625c4..400cf6e 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -215,7 +215,7 @@ static void svc_release_skb(struct svc_rqst *rqstp)
 static void
 svc_sock_enqueue(struct svc_sock *svsk)
 {
-	struct svc_serv	*serv = svsk->sk_server;
+	struct svc_serv	*serv = svsk->sk_xprt.xpt_server;
 	struct svc_pool *pool;
 	struct svc_rqst	*rqstp;
 	int cpu;
@@ -227,7 +227,7 @@ svc_sock_enqueue(struct svc_sock *svsk)
 		return;
 
 	cpu = get_cpu();
-	pool = svc_pool_for_cpu(svsk->sk_server, cpu);
+	pool = svc_pool_for_cpu(svsk->sk_xprt.xpt_server, cpu);
 	put_cpu();
 
 	spin_lock_bh(&pool->sp_lock);
@@ -253,8 +253,8 @@ svc_sock_enqueue(struct svc_sock *svsk)
 		dprintk("svc: socket %p busy, not enqueued\n", svsk->sk_sk);
 		goto out_unlock;
 	}
-	BUG_ON(svsk->sk_pool != NULL);
-	svsk->sk_pool = pool;
+	BUG_ON(svsk->sk_xprt.xpt_pool != NULL);
+	svsk->sk_xprt.xpt_pool = pool;
 
 	/* Handle pending connection */
 	if (test_bit(XPT_CONN, &svsk->sk_xprt.xpt_flags))
@@ -268,7 +268,7 @@ svc_sock_enqueue(struct svc_sock *svsk)
 	if (!svsk->sk_xprt.xpt_ops->xpo_has_wspace(&svsk->sk_xprt)) {
 		/* Don't enqueue while not enough space for reply */
 		dprintk("svc: no write space, socket %p  not enqueued\n", svsk);
-		svsk->sk_pool = NULL;
+		svsk->sk_xprt.xpt_pool = NULL;
 		clear_bit(XPT_BUSY, &svsk->sk_xprt.xpt_flags);
 		goto out_unlock;
 	}
@@ -289,12 +289,12 @@ svc_sock_enqueue(struct svc_sock *svsk)
 		svc_xprt_get(&svsk->sk_xprt);
 		rqstp->rq_reserved = serv->sv_max_mesg;
 		atomic_add(rqstp->rq_reserved, &svsk->sk_reserved);
-		BUG_ON(svsk->sk_pool != pool);
+		BUG_ON(svsk->sk_xprt.xpt_pool != pool);
 		wake_up(&rqstp->rq_wait);
 	} else {
 		dprintk("svc: socket %p put into queue\n", svsk->sk_sk);
 		list_add_tail(&svsk->sk_ready, &pool->sp_sockets);
-		BUG_ON(svsk->sk_pool != pool);
+		BUG_ON(svsk->sk_xprt.xpt_pool != pool);
 	}
 
 out_unlock:
@@ -331,7 +331,7 @@ svc_sock_dequeue(struct svc_pool *pool)
 static inline void
 svc_sock_received(struct svc_sock *svsk)
 {
-	svsk->sk_pool = NULL;
+	svsk->sk_xprt.xpt_pool = NULL;
 	clear_bit(XPT_BUSY, &svsk->sk_xprt.xpt_flags);
 	svc_sock_enqueue(svsk);
 }
@@ -735,7 +735,7 @@ static int
 svc_udp_recvfrom(struct svc_rqst *rqstp)
 {
 	struct svc_sock	*svsk = rqstp->rq_sock;
-	struct svc_serv	*serv = svsk->sk_server;
+	struct svc_serv	*serv = svsk->sk_xprt.xpt_server;
 	struct sk_buff	*skb;
 	union {
 		struct cmsghdr	hdr;
@@ -873,7 +873,7 @@ static void svc_udp_prep_reply_hdr(struct svc_rqst *rqstp)
 static int svc_udp_has_wspace(struct svc_xprt *xprt)
 {
 	struct svc_sock *svsk = container_of(xprt, struct svc_sock, sk_xprt);
-	struct svc_serv	*serv = svsk->sk_server;
+	struct svc_serv	*serv = xprt->xpt_server;
 	unsigned long required;
 
 	/*
@@ -920,13 +920,12 @@ static struct svc_xprt_class svc_udp_class = {
 	.xcl_max_payload = RPCSVC_MAXPAYLOAD_UDP,
 };
 
-static void
-svc_udp_init(struct svc_sock *svsk)
+static void svc_udp_init(struct svc_sock *svsk, struct svc_serv *serv)
 {
 	int one = 1;
 	mm_segment_t oldfs;
 
-	svc_xprt_init(&svc_udp_class, &svsk->sk_xprt);
+	svc_xprt_init(&svc_udp_class, &svsk->sk_xprt, serv);
 	svsk->sk_sk->sk_data_ready = svc_udp_data_ready;
 	svsk->sk_sk->sk_write_space = svc_write_space;
 
@@ -935,8 +934,8 @@ svc_udp_init(struct svc_sock *svsk)
 	 * svc_udp_recvfrom will re-adjust if necessary
 	 */
 	svc_sock_setbufsize(svsk->sk_sock,
-			    3 * svsk->sk_server->sv_max_mesg,
-			    3 * svsk->sk_server->sv_max_mesg);
+			    3 * svsk->sk_xprt.xpt_server->sv_max_mesg,
+			    3 * svsk->sk_xprt.xpt_server->sv_max_mesg);
 
 	set_bit(XPT_DATA, &svsk->sk_xprt.xpt_flags); /* might have come in before data_ready set up */
 	set_bit(XPT_CHNGBUF, &svsk->sk_xprt.xpt_flags);
@@ -1041,7 +1040,7 @@ static struct svc_xprt *svc_tcp_accept(struct svc_xprt *xprt)
 	struct svc_sock *svsk = container_of(xprt, struct svc_sock, sk_xprt);
 	struct sockaddr_storage addr;
 	struct sockaddr	*sin = (struct sockaddr *) &addr;
-	struct svc_serv	*serv = svsk->sk_server;
+	struct svc_serv	*serv = svsk->sk_xprt.xpt_server;
 	struct socket	*sock = svsk->sk_sock;
 	struct socket	*newsock;
 	struct svc_sock	*newsvsk;
@@ -1122,7 +1121,7 @@ static int
 svc_tcp_recvfrom(struct svc_rqst *rqstp)
 {
 	struct svc_sock	*svsk = rqstp->rq_sock;
-	struct svc_serv	*serv = svsk->sk_server;
+	struct svc_serv	*serv = svsk->sk_xprt.xpt_server;
 	int		len;
 	struct kvec *vec;
 	int pnum, vlen;
@@ -1265,7 +1264,7 @@ svc_tcp_recvfrom(struct svc_rqst *rqstp)
 		svc_sock_received(svsk);
 	} else {
 		printk(KERN_NOTICE "%s: recvfrom returned errno %d\n",
-					svsk->sk_server->sv_name, -len);
+		       svsk->sk_xprt.xpt_server->sv_name, -len);
 		goto err_delete;
 	}
 
@@ -1295,7 +1294,7 @@ svc_tcp_sendto(struct svc_rqst *rqstp)
 	sent = svc_sendto(rqstp, &rqstp->rq_res);
 	if (sent != xbufp->len) {
 		printk(KERN_NOTICE "rpc-srv/tcp: %s: %s %d when sending %d bytes - shutting down socket\n",
-		       rqstp->rq_sock->sk_server->sv_name,
+		       rqstp->rq_sock->sk_xprt.xpt_server->sv_name,
 		       (sent<0)?"got error":"sent only",
 		       sent, xbufp->len);
 		set_bit(XPT_CLOSE, &rqstp->rq_sock->sk_xprt.xpt_flags);
@@ -1319,7 +1318,7 @@ static void svc_tcp_prep_reply_hdr(struct svc_rqst *rqstp)
 static int svc_tcp_has_wspace(struct svc_xprt *xprt)
 {
 	struct svc_sock *svsk = container_of(xprt, struct svc_sock, sk_xprt);
-	struct svc_serv	*serv = svsk->sk_server;
+	struct svc_serv	*serv = svsk->sk_xprt.xpt_server;
 	int required;
 	int wspace;
 
@@ -1378,13 +1377,12 @@ void svc_cleanup_xprt_sock(void)
 	svc_unreg_xprt_class(&svc_udp_class);
 }
 
-static void
-svc_tcp_init(struct svc_sock *svsk)
+static void svc_tcp_init(struct svc_sock *svsk, struct svc_serv *serv)
 {
 	struct sock	*sk = svsk->sk_sk;
 	struct tcp_sock *tp = tcp_sk(sk);
 
-	svc_xprt_init(&svc_tcp_class, &svsk->sk_xprt);
+	svc_xprt_init(&svc_tcp_class, &svsk->sk_xprt, serv);
 
 	if (sk->sk_state == TCP_LISTEN) {
 		dprintk("setting up TCP socket for listening\n");
@@ -1407,8 +1405,8 @@ svc_tcp_init(struct svc_sock *svsk)
 		 * svc_tcp_recvfrom will re-adjust if necessary
 		 */
 		svc_sock_setbufsize(svsk->sk_sock,
-				    3 * svsk->sk_server->sv_max_mesg,
-				    3 * svsk->sk_server->sv_max_mesg);
+				    3 * svsk->sk_xprt.xpt_server->sv_max_mesg,
+				    3 * svsk->sk_xprt.xpt_server->sv_max_mesg);
 
 		set_bit(XPT_CHNGBUF, &svsk->sk_xprt.xpt_flags);
 		set_bit(XPT_DATA, &svsk->sk_xprt.xpt_flags);
@@ -1593,7 +1591,7 @@ svc_recv(struct svc_rqst *rqstp, long timeout)
 			 * listener holds a reference too
 			 */
 			__module_get(newxpt->xpt_class->xcl_owner);
-			svc_check_conn_limits(svsk->sk_server);
+			svc_check_conn_limits(svsk->sk_xprt.xpt_server);
 		}
 		svc_sock_received(svsk);
 	} else {
@@ -1761,7 +1759,6 @@ static struct svc_sock *svc_setup_socket(struct svc_serv *serv,
 	svsk->sk_ostate = inet->sk_state_change;
 	svsk->sk_odata = inet->sk_data_ready;
 	svsk->sk_owspace = inet->sk_write_space;
-	svsk->sk_server = serv;
 	svsk->sk_lastrecv = get_seconds();
 	spin_lock_init(&svsk->sk_lock);
 	INIT_LIST_HEAD(&svsk->sk_deferred);
@@ -1770,9 +1767,9 @@ static struct svc_sock *svc_setup_socket(struct svc_serv *serv,
 
 	/* Initialize the socket */
 	if (sock->type == SOCK_DGRAM)
-		svc_udp_init(svsk);
+		svc_udp_init(svsk, serv);
 	else
-		svc_tcp_init(svsk);
+		svc_tcp_init(svsk, serv);
 
 	spin_lock_bh(&serv->sv_lock);
 	if (is_temporary) {
@@ -1930,7 +1927,7 @@ svc_delete_socket(struct svc_sock *svsk)
 
 	dprintk("svc: svc_delete_socket(%p)\n", svsk);
 
-	serv = svsk->sk_server;
+	serv = svsk->sk_xprt.xpt_server;
 	sk = svsk->sk_sk;
 
 	svsk->sk_xprt.xpt_ops->xpo_detach(&svsk->sk_xprt);

[PATCH 17/38] svc: Make close transport independent

[Tom Tucker]

Move sk_list and sk_ready to svc_xprt. This involves close because these
lists are walked by svcs when closing all their transports. So I combined
the moving of these lists to svc_xprt with making close transport independent.

The svc_force_sock_close has been changed to svc_close_all and takes a list
as an argument. This removes some svc internals knowledge from the svcs. 

This code races with module removal and transport addition. 

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 fs/lockd/svc.c                  |    6 +-
 fs/nfsd/nfssvc.c                |    4 +
 include/linux/sunrpc/svc_xprt.h |    2 +
 include/linux/sunrpc/svcsock.h  |    4 -
 net/sunrpc/svc.c                |    9 +--
 net/sunrpc/svc_xprt.c           |    2 +
 net/sunrpc/svcsock.c            |  106 +++++++++++++++++++--------------------
 7 files changed, 63 insertions(+), 70 deletions(-)

diff --git a/fs/lockd/svc.c b/fs/lockd/svc.c
index 8686915..a8e79a9 100644
--- a/fs/lockd/svc.c
+++ b/fs/lockd/svc.c
@@ -221,10 +221,10 @@ lockd(struct svc_rqst *rqstp)
 
 static int find_xprt(struct svc_serv *serv, char *proto)
 {
-	struct svc_sock *svsk;
+	struct svc_xprt *xprt;
 	int found = 0;
-	list_for_each_entry(svsk, &serv->sv_permsocks, sk_list)
-		if (strcmp(svsk->sk_xprt.xpt_class->xcl_name, proto) == 0) {
+	list_for_each_entry(xprt, &serv->sv_permsocks, xpt_list)
+		if (strcmp(xprt->xpt_class->xcl_name, proto) == 0) {
 			found = 1;
 			break;
 		}
diff --git a/fs/nfsd/nfssvc.c b/fs/nfsd/nfssvc.c
index a828b0b..9647b0f 100644
--- a/fs/nfsd/nfssvc.c
+++ b/fs/nfsd/nfssvc.c
@@ -155,8 +155,8 @@ static int killsig;	/* signal that was used to kill last nfsd */
 static void nfsd_last_thread(struct svc_serv *serv)
 {
 	/* When last nfsd thread exits we need to do some clean-up */
-	struct svc_sock *svsk;
-	list_for_each_entry(svsk, &serv->sv_permsocks, sk_list)
+	struct svc_xprt *xprt;
+	list_for_each_entry(xprt, &serv->sv_permsocks, xpt_list)
 		lockd_down();
 	nfsd_serv = NULL;
 	nfsd_racache_shutdown();
diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
index 4f7dbbc..21fa6ad 100644
--- a/include/linux/sunrpc/svc_xprt.h
+++ b/include/linux/sunrpc/svc_xprt.h
@@ -36,6 +36,8 @@ struct svc_xprt {
 	struct svc_xprt_class	*xpt_class;
 	struct svc_xprt_ops	*xpt_ops;
 	struct kref		xpt_ref;
+	struct list_head	xpt_list;
+	struct list_head	xpt_ready;
 	unsigned long		xpt_flags;
 #define	XPT_BUSY	0		/* enqueued/receiving */
 #define	XPT_CONN	1		/* conn pending */
diff --git a/include/linux/sunrpc/svcsock.h b/include/linux/sunrpc/svcsock.h
index 92d4cc9..060508b 100644
--- a/include/linux/sunrpc/svcsock.h
+++ b/include/linux/sunrpc/svcsock.h
@@ -17,8 +17,6 @@
  */
 struct svc_sock {
 	struct svc_xprt		sk_xprt;
-	struct list_head	sk_ready;	/* list of ready sockets */
-	struct list_head	sk_list;	/* list of all sockets */
 	struct socket *		sk_sock;	/* berkeley socket layer */
 	struct sock *		sk_sk;		/* INET layer */
 
@@ -51,7 +49,7 @@ struct svc_sock {
 /*
  * Function prototypes.
  */
-void		svc_force_close_socket(struct svc_sock *);
+void		svc_close_all(struct list_head *);
 int		svc_recv(struct svc_rqst *, long);
 int		svc_send(struct svc_rqst *);
 void		svc_drop(struct svc_rqst *);
diff --git a/net/sunrpc/svc.c b/net/sunrpc/svc.c
index 3cc945d..be18734 100644
--- a/net/sunrpc/svc.c
+++ b/net/sunrpc/svc.c
@@ -458,9 +458,6 @@ svc_create_pooled(struct svc_program *prog, unsigned int bufsize,
 void
 svc_destroy(struct svc_serv *serv)
 {
-	struct svc_sock	*svsk;
-	struct svc_sock *tmp;
-
 	dprintk("svc: svc_destroy(%s, %d)\n",
 				serv->sv_program->pg_name,
 				serv->sv_nrthreads);
@@ -475,14 +472,12 @@ svc_destroy(struct svc_serv *serv)
 
 	del_timer_sync(&serv->sv_temptimer);
 
-	list_for_each_entry_safe(svsk, tmp, &serv->sv_tempsocks, sk_list)
-		svc_force_close_socket(svsk);
+	svc_close_all(&serv->sv_tempsocks);
 
 	if (serv->sv_shutdown)
 		serv->sv_shutdown(serv);
 
-	list_for_each_entry_safe(svsk, tmp, &serv->sv_permsocks, sk_list)
-		svc_force_close_socket(svsk);
+	svc_close_all(&serv->sv_permsocks);
 
 	BUG_ON(!list_empty(&serv->sv_permsocks));
 	BUG_ON(!list_empty(&serv->sv_tempsocks));
diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
index e366add..bbdada7 100644
--- a/net/sunrpc/svc_xprt.c
+++ b/net/sunrpc/svc_xprt.c
@@ -110,6 +110,8 @@ void svc_xprt_init(struct svc_xprt_class *xcl, struct svc_xprt *xprt,
 	xprt->xpt_ops = xcl->xcl_ops;
 	kref_init(&xprt->xpt_ref);
 	xprt->xpt_server = serv;
+	INIT_LIST_HEAD(&xprt->xpt_list);
+	INIT_LIST_HEAD(&xprt->xpt_ready);
 }
 EXPORT_SYMBOL_GPL(svc_xprt_init);
 
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index 400cf6e..0541c55 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -81,11 +81,11 @@
 
 static struct svc_sock *svc_setup_socket(struct svc_serv *, struct socket *,
 					 int *errp, int flags);
-static void		svc_delete_socket(struct svc_sock *svsk);
+static void		svc_delete_xprt(struct svc_xprt *xprt);
 static void		svc_udp_data_ready(struct sock *, int);
 static int		svc_udp_recvfrom(struct svc_rqst *);
 static int		svc_udp_sendto(struct svc_rqst *);
-static void		svc_close_socket(struct svc_sock *svsk);
+static void		svc_close_xprt(struct svc_xprt *xprt);
 static void		svc_sock_detach(struct svc_xprt *);
 static void		svc_sock_free(struct svc_xprt *);
 
@@ -293,7 +293,7 @@ svc_sock_enqueue(struct svc_sock *svsk)
 		wake_up(&rqstp->rq_wait);
 	} else {
 		dprintk("svc: socket %p put into queue\n", svsk->sk_sk);
-		list_add_tail(&svsk->sk_ready, &pool->sp_sockets);
+		list_add_tail(&svsk->sk_xprt.xpt_ready, &pool->sp_sockets);
 		BUG_ON(svsk->sk_xprt.xpt_pool != pool);
 	}
 
@@ -313,8 +313,8 @@ svc_sock_dequeue(struct svc_pool *pool)
 		return NULL;
 
 	svsk = list_entry(pool->sp_sockets.next,
-			  struct svc_sock, sk_ready);
-	list_del_init(&svsk->sk_ready);
+			  struct svc_sock, sk_xprt.xpt_ready);
+	list_del_init(&svsk->sk_xprt.xpt_ready);
 
 	dprintk("svc: socket %p dequeued, inuse=%d\n",
 		svsk->sk_sk, atomic_read(&svsk->sk_xprt.xpt_ref.refcount));
@@ -572,7 +572,7 @@ svc_sock_names(char *buf, struct svc_serv *serv, char *toclose)
 	if (!serv)
 		return 0;
 	spin_lock_bh(&serv->sv_lock);
-	list_for_each_entry(svsk, &serv->sv_permsocks, sk_list) {
+	list_for_each_entry(svsk, &serv->sv_permsocks, sk_xprt.xpt_list) {
 		int onelen = one_sock_name(buf+len, svsk);
 		if (toclose && strcmp(toclose, buf+len) == 0)
 			closesk = svsk;
@@ -584,7 +584,7 @@ svc_sock_names(char *buf, struct svc_serv *serv, char *toclose)
 		/* Should unregister with portmap, but you cannot
 		 * unregister just one protocol...
 		 */
-		svc_close_socket(closesk);
+		svc_close_xprt(&closesk->sk_xprt);
 	else if (toclose)
 		return -ENOENT;
 	return len;
@@ -1255,7 +1255,7 @@ svc_tcp_recvfrom(struct svc_rqst *rqstp)
 	return len;
 
  err_delete:
-	svc_delete_socket(svsk);
+	svc_delete_xprt(&svsk->sk_xprt);
 	return -EAGAIN;
 
  error:
@@ -1427,12 +1427,12 @@ svc_sock_update_bufs(struct svc_serv *serv)
 	spin_lock_bh(&serv->sv_lock);
 	list_for_each(le, &serv->sv_permsocks) {
 		struct svc_sock *svsk =
-			list_entry(le, struct svc_sock, sk_list);
+			list_entry(le, struct svc_sock, sk_xprt.xpt_list);
 		set_bit(XPT_CHNGBUF, &svsk->sk_xprt.xpt_flags);
 	}
 	list_for_each(le, &serv->sv_tempsocks) {
 		struct svc_sock *svsk =
-			list_entry(le, struct svc_sock, sk_list);
+			list_entry(le, struct svc_sock, sk_xprt.xpt_list);
 		set_bit(XPT_CHNGBUF, &svsk->sk_xprt.xpt_flags);
 	}
 	spin_unlock_bh(&serv->sv_lock);
@@ -1474,7 +1474,7 @@ static void svc_check_conn_limits(struct svc_serv *serv)
 			 */
 			svsk = list_entry(serv->sv_tempsocks.prev,
 					  struct svc_sock,
-					  sk_list);
+					  sk_xprt.xpt_list);
 			set_bit(XPT_CLOSE, &svsk->sk_xprt.xpt_flags);
 			svc_xprt_get(&svsk->sk_xprt);
 		}
@@ -1581,7 +1581,7 @@ svc_recv(struct svc_rqst *rqstp, long timeout)
 	len = 0;
 	if (test_bit(XPT_CLOSE, &svsk->sk_xprt.xpt_flags)) {
 		dprintk("svc_recv: found XPT_CLOSE\n");
-		svc_delete_socket(svsk);
+		svc_delete_xprt(&svsk->sk_xprt);
 	} else if (test_bit(XPT_LISTENER, &svsk->sk_xprt.xpt_flags)) {
 		struct svc_xprt *newxpt;
 		newxpt = svsk->sk_xprt.xpt_ops->xpo_accept(&svsk->sk_xprt);
@@ -1690,7 +1690,7 @@ svc_age_temp_sockets(unsigned long closure)
 	}
 
 	list_for_each_safe(le, next, &serv->sv_tempsocks) {
-		svsk = list_entry(le, struct svc_sock, sk_list);
+		svsk = list_entry(le, struct svc_sock, sk_xprt.xpt_list);
 
 		if (!test_and_set_bit(XPT_OLD, &svsk->sk_xprt.xpt_flags))
 			continue;
@@ -1706,9 +1706,9 @@ svc_age_temp_sockets(unsigned long closure)
 
 	while (!list_empty(&to_be_aged)) {
 		le = to_be_aged.next;
-		/* fiddling the sk_list node is safe 'cos we're XPT_DETACHED */
+		/* fiddling the sk_xprt.xpt_list node is safe 'cos we're XPT_DETACHED */
 		list_del_init(le);
-		svsk = list_entry(le, struct svc_sock, sk_list);
+		svsk = list_entry(le, struct svc_sock, sk_xprt.xpt_list);
 
 		dprintk("queuing svsk %p for closing, %lu seconds old\n",
 			svsk, get_seconds() - svsk->sk_lastrecv);
@@ -1762,7 +1762,6 @@ static struct svc_sock *svc_setup_socket(struct svc_serv *serv,
 	svsk->sk_lastrecv = get_seconds();
 	spin_lock_init(&svsk->sk_lock);
 	INIT_LIST_HEAD(&svsk->sk_deferred);
-	INIT_LIST_HEAD(&svsk->sk_ready);
 	mutex_init(&svsk->sk_mutex);
 
 	/* Initialize the socket */
@@ -1774,7 +1773,7 @@ static struct svc_sock *svc_setup_socket(struct svc_serv *serv,
 	spin_lock_bh(&serv->sv_lock);
 	if (is_temporary) {
 		set_bit(XPT_TEMP, &svsk->sk_xprt.xpt_flags);
-		list_add(&svsk->sk_list, &serv->sv_tempsocks);
+		list_add(&svsk->sk_xprt.xpt_list, &serv->sv_tempsocks);
 		serv->sv_tmpcnt++;
 		if (serv->sv_temptimer.function == NULL) {
 			/* setup timer to age temp sockets */
@@ -1785,7 +1784,7 @@ static struct svc_sock *svc_setup_socket(struct svc_serv *serv,
 		}
 	} else {
 		clear_bit(XPT_TEMP, &svsk->sk_xprt.xpt_flags);
-		list_add(&svsk->sk_list, &serv->sv_permsocks);
+		list_add(&svsk->sk_xprt.xpt_list, &serv->sv_permsocks);
 	}
 	spin_unlock_bh(&serv->sv_lock);
 
@@ -1917,66 +1916,63 @@ static void svc_sock_free(struct svc_xprt *xprt)
 }
 
 /*
- * Remove a dead socket
+ * Remove a dead transport
  */
-static void
-svc_delete_socket(struct svc_sock *svsk)
+static void svc_delete_xprt(struct svc_xprt *xprt)
 {
-	struct svc_serv	*serv;
-	struct sock	*sk;
-
-	dprintk("svc: svc_delete_socket(%p)\n", svsk);
-
-	serv = svsk->sk_xprt.xpt_server;
-	sk = svsk->sk_sk;
+	struct svc_serv	*serv = xprt->xpt_server;
 
-	svsk->sk_xprt.xpt_ops->xpo_detach(&svsk->sk_xprt);
+	dprintk("svc: svc_delete_xprt(%p)\n", xprt);
+	xprt->xpt_ops->xpo_detach(xprt);
 
 	spin_lock_bh(&serv->sv_lock);
-
-	if (!test_and_set_bit(XPT_DETACHED, &svsk->sk_xprt.xpt_flags))
-		list_del_init(&svsk->sk_list);
+	if (!test_and_set_bit(XPT_DETACHED, &xprt->xpt_flags))
+		list_del_init(&xprt->xpt_list);
 	/*
-	 * We used to delete the svc_sock from whichever list
-	 * it's sk_ready node was on, but we don't actually
+	 * We used to delete the transport from whichever list
+	 * it's sk_xprt.xpt_ready node was on, but we don't actually
 	 * need to.  This is because the only time we're called
 	 * while still attached to a queue, the queue itself
 	 * is about to be destroyed (in svc_destroy).
 	 */
-	if (!test_and_set_bit(XPT_DEAD, &svsk->sk_xprt.xpt_flags)) {
-		BUG_ON(atomic_read(&svsk->sk_xprt.xpt_ref.refcount) < 2);
-		svc_xprt_put(&svsk->sk_xprt);
-		if (test_bit(XPT_TEMP, &svsk->sk_xprt.xpt_flags))
+	if (!test_and_set_bit(XPT_DEAD, &xprt->xpt_flags)) {
+		BUG_ON(atomic_read(&xprt->xpt_ref.refcount) < 2);
+		svc_xprt_put(xprt);
+		if (test_bit(XPT_TEMP, &xprt->xpt_flags))
 			serv->sv_tmpcnt--;
 	}
-
 	spin_unlock_bh(&serv->sv_lock);
 }
 
-static void svc_close_socket(struct svc_sock *svsk)
+static void svc_close_xprt(struct svc_xprt *xprt)
 {
-	set_bit(XPT_CLOSE, &svsk->sk_xprt.xpt_flags);
-	if (test_and_set_bit(XPT_BUSY, &svsk->sk_xprt.xpt_flags))
+	set_bit(XPT_CLOSE, &xprt->xpt_flags);
+	if (test_and_set_bit(XPT_BUSY, &xprt->xpt_flags))
 		/* someone else will have to effect the close */
 		return;
 
-	svc_xprt_get(&svsk->sk_xprt);
-	svc_delete_socket(svsk);
-	clear_bit(XPT_BUSY, &svsk->sk_xprt.xpt_flags);
-	svc_xprt_put(&svsk->sk_xprt);
+	svc_xprt_get(xprt);
+	svc_delete_xprt(xprt);
+	clear_bit(XPT_BUSY, &xprt->xpt_flags);
+	svc_xprt_put(xprt);
 }
 
-void svc_force_close_socket(struct svc_sock *svsk)
+void svc_close_all(struct list_head *xprt_list)
 {
-	set_bit(XPT_CLOSE, &svsk->sk_xprt.xpt_flags);
-	if (test_bit(XPT_BUSY, &svsk->sk_xprt.xpt_flags)) {
-		/* Waiting to be processed, but no threads left,
-		 * So just remove it from the waiting list
-		 */
-		list_del_init(&svsk->sk_ready);
-		clear_bit(XPT_BUSY, &svsk->sk_xprt.xpt_flags);
+	struct svc_xprt *xprt;
+	struct svc_xprt *tmp;
+
+	list_for_each_entry_safe(xprt, tmp, xprt_list, xpt_list) {
+		set_bit(XPT_CLOSE, &xprt->xpt_flags);
+		if (test_bit(XPT_BUSY, &xprt->xpt_flags)) {
+			/* Waiting to be processed, but no threads left,
+			 * So just remove it from the waiting list
+			 */
+			list_del_init(&xprt->xpt_ready);
+			clear_bit(XPT_BUSY, &xprt->xpt_flags);
+		}
+		svc_close_xprt(xprt);
 	}
-	svc_close_socket(svsk);
 }
 
 /*

[PATCH 18/38] svc: Move sk_reserved to svc_xprt

[Tom Tucker]

This functionally trivial patch moves the sk_reserved field to the 
transport independent svc_xprt structure. 

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 include/linux/sunrpc/svc_xprt.h |    1 +
 include/linux/sunrpc/svcsock.h  |    2 --
 net/sunrpc/svcsock.c            |   10 +++++-----
 3 files changed, 6 insertions(+), 7 deletions(-)

diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
index 21fa6ad..c9892d5 100644
--- a/include/linux/sunrpc/svc_xprt.h
+++ b/include/linux/sunrpc/svc_xprt.h
@@ -53,6 +53,7 @@ struct svc_xprt {
 
 	struct svc_pool		*xpt_pool;	/* current pool iff queued */
 	struct svc_serv		*xpt_server;	/* service for transport */
+	atomic_t    	    	xpt_reserved;	/* space on outq that is rsvd */
 };
 
 int	svc_reg_xprt_class(struct svc_xprt_class *);
diff --git a/include/linux/sunrpc/svcsock.h b/include/linux/sunrpc/svcsock.h
index 060508b..ba41f11 100644
--- a/include/linux/sunrpc/svcsock.h
+++ b/include/linux/sunrpc/svcsock.h
@@ -20,8 +20,6 @@ struct svc_sock {
 	struct socket *		sk_sock;	/* berkeley socket layer */
 	struct sock *		sk_sk;		/* INET layer */
 
-	atomic_t    	    	sk_reserved;	/* space on outq that is reserved */
-
 	spinlock_t		sk_lock;	/* protects sk_deferred and
 						 * sk_info_authunix */
 	struct list_head	sk_deferred;	/* deferred requests that need to
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index 0541c55..954689e 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -288,7 +288,7 @@ svc_sock_enqueue(struct svc_sock *svsk)
 		rqstp->rq_sock = svsk;
 		svc_xprt_get(&svsk->sk_xprt);
 		rqstp->rq_reserved = serv->sv_max_mesg;
-		atomic_add(rqstp->rq_reserved, &svsk->sk_reserved);
+		atomic_add(rqstp->rq_reserved, &svsk->sk_xprt.xpt_reserved);
 		BUG_ON(svsk->sk_xprt.xpt_pool != pool);
 		wake_up(&rqstp->rq_wait);
 	} else {
@@ -353,7 +353,7 @@ void svc_reserve(struct svc_rqst *rqstp, int space)
 
 	if (space < rqstp->rq_reserved) {
 		struct svc_sock *svsk = rqstp->rq_sock;
-		atomic_sub((rqstp->rq_reserved - space), &svsk->sk_reserved);
+		atomic_sub((rqstp->rq_reserved - space), &svsk->sk_xprt.xpt_reserved);
 		rqstp->rq_reserved = space;
 
 		svc_sock_enqueue(svsk);
@@ -881,7 +881,7 @@ static int svc_udp_has_wspace(struct svc_xprt *xprt)
 	 * sock space.
 	 */
 	set_bit(SOCK_NOSPACE, &svsk->sk_sock->flags);
-	required = atomic_read(&svsk->sk_reserved) + serv->sv_max_mesg;
+	required = atomic_read(&svsk->sk_xprt.xpt_reserved) + serv->sv_max_mesg;
 	if (required*2 > sock_wspace(svsk->sk_sk))
 		return 0;
 	clear_bit(SOCK_NOSPACE, &svsk->sk_sock->flags);
@@ -1327,7 +1327,7 @@ static int svc_tcp_has_wspace(struct svc_xprt *xprt)
 	 * sock space.
 	 */
 	set_bit(SOCK_NOSPACE, &svsk->sk_sock->flags);
-	required = atomic_read(&svsk->sk_reserved) + serv->sv_max_mesg;
+	required = atomic_read(&svsk->sk_xprt.xpt_reserved) + serv->sv_max_mesg;
 	wspace = sk_stream_wspace(svsk->sk_sk);
 
 	if (wspace < sk_stream_min_wspace(svsk->sk_sk))
@@ -1549,7 +1549,7 @@ svc_recv(struct svc_rqst *rqstp, long timeout)
 		rqstp->rq_sock = svsk;
 		svc_xprt_get(&svsk->sk_xprt);
 		rqstp->rq_reserved = serv->sv_max_mesg;
-		atomic_add(rqstp->rq_reserved, &svsk->sk_reserved);
+		atomic_add(rqstp->rq_reserved, &svsk->sk_xprt.xpt_reserved);
 	} else {
 		/* No data pending. Go to sleep */
 		svc_thread_enqueue(pool, rqstp);

[PATCH 19/38] svc: Make the enqueue service transport neutral and export it.

[Tom Tucker]

The svc_sock_enqueue function is now transport independent since all of
the fields it touches have been moved to the transport independent svc_xprt
structure. Change the function to use the svc_xprt structure directly
instead of the transport specific svc_sock structure.

Transport specific data-ready handlers need to call this function, so
export it.

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 net/sunrpc/svcsock.c |   94 ++++++++++++++++++++++++++------------------------
 1 files changed, 48 insertions(+), 46 deletions(-)

diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index 954689e..d8df8c7 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -5,7 +5,7 @@
  *
  * The server scheduling algorithm does not always distribute the load
  * evenly when servicing a single client. May need to modify the
- * svc_sock_enqueue procedure...
+ * svc_xprt_enqueue procedure...
  *
  * TCP support is largely untested and may be a little slow. The problem
  * is that we currently do two separate recvfrom's, one for the 4-byte
@@ -63,7 +63,7 @@
  *	providing that certain rules are followed:
  *
  *	XPT_CONN, XPT_DATA, can be set or cleared at any time.
- *		after a set, svc_sock_enqueue must be called.
+ *		after a set, svc_xprt_enqueue must be called.
  *		after a clear, the socket must be read/accepted
  *		 if this succeeds, it must be set again.
  *	XPT_CLOSE can set at any time. It is never cleared.
@@ -212,22 +212,21 @@ static void svc_release_skb(struct svc_rqst *rqstp)
  * processes, wake 'em up.
  *
  */
-static void
-svc_sock_enqueue(struct svc_sock *svsk)
+void svc_xprt_enqueue(struct svc_xprt *xprt)
 {
-	struct svc_serv	*serv = svsk->sk_xprt.xpt_server;
+	struct svc_serv	*serv = xprt->xpt_server;
 	struct svc_pool *pool;
 	struct svc_rqst	*rqstp;
 	int cpu;
 
-	if (!(svsk->sk_xprt.xpt_flags &
+	if (!(xprt->xpt_flags &
 	      ((1<<XPT_CONN)|(1<<XPT_DATA)|(1<<XPT_CLOSE)|(1<<XPT_DEFERRED))))
 		return;
-	if (test_bit(XPT_DEAD, &svsk->sk_xprt.xpt_flags))
+	if (test_bit(XPT_DEAD, &xprt->xpt_flags))
 		return;
 
 	cpu = get_cpu();
-	pool = svc_pool_for_cpu(svsk->sk_xprt.xpt_server, cpu);
+	pool = svc_pool_for_cpu(xprt->xpt_server, cpu);
 	put_cpu();
 
 	spin_lock_bh(&pool->sp_lock);
@@ -235,11 +234,12 @@ svc_sock_enqueue(struct svc_sock *svsk)
 	if (!list_empty(&pool->sp_threads) &&
 	    !list_empty(&pool->sp_sockets))
 		printk(KERN_ERR
-			"svc_sock_enqueue: threads and sockets both waiting??\n");
+		       "svc_xprt_enqueue: "
+		       "threads and transports both waiting??\n");
 
-	if (test_bit(XPT_DEAD, &svsk->sk_xprt.xpt_flags)) {
+	if (test_bit(XPT_DEAD, &xprt->xpt_flags)) {
 		/* Don't enqueue dead sockets */
-		dprintk("svc: socket %p is dead, not enqueued\n", svsk->sk_sk);
+		dprintk("svc: transport %p is dead, not enqueued\n", xprt);
 		goto out_unlock;
 	}
 
@@ -248,28 +248,29 @@ svc_sock_enqueue(struct svc_sock *svsk)
 	 * on the idle list.  We update XPT_BUSY atomically because
 	 * it also guards against trying to enqueue the svc_sock twice.
 	 */
-	if (test_and_set_bit(XPT_BUSY, &svsk->sk_xprt.xpt_flags)) {
+	if (test_and_set_bit(XPT_BUSY, &xprt->xpt_flags)) {
 		/* Don't enqueue socket while already enqueued */
-		dprintk("svc: socket %p busy, not enqueued\n", svsk->sk_sk);
+		dprintk("svc: transport %p busy, not enqueued\n", xprt);
 		goto out_unlock;
 	}
-	BUG_ON(svsk->sk_xprt.xpt_pool != NULL);
-	svsk->sk_xprt.xpt_pool = pool;
+	BUG_ON(xprt->xpt_pool != NULL);
+	xprt->xpt_pool = pool;
 
 	/* Handle pending connection */
-	if (test_bit(XPT_CONN, &svsk->sk_xprt.xpt_flags))
+	if (test_bit(XPT_CONN, &xprt->xpt_flags))
 		goto process;
 
 	/* Handle close in-progress */
-	if (test_bit(XPT_CLOSE, &svsk->sk_xprt.xpt_flags))
+	if (test_bit(XPT_CLOSE, &xprt->xpt_flags))
 		goto process;
 
 	/* Check if we have space to reply to a request */
-	if (!svsk->sk_xprt.xpt_ops->xpo_has_wspace(&svsk->sk_xprt)) {
+	if (!xprt->xpt_ops->xpo_has_wspace(xprt)) {
 		/* Don't enqueue while not enough space for reply */
-		dprintk("svc: no write space, socket %p  not enqueued\n", svsk);
-		svsk->sk_xprt.xpt_pool = NULL;
-		clear_bit(XPT_BUSY, &svsk->sk_xprt.xpt_flags);
+		dprintk("svc: no write space, transport %p  not enqueued\n",
+			xprt);
+		xprt->xpt_pool = NULL;
+		clear_bit(XPT_BUSY, &xprt->xpt_flags);
 		goto out_unlock;
 	}
 
@@ -278,28 +279,29 @@ svc_sock_enqueue(struct svc_sock *svsk)
 		rqstp = list_entry(pool->sp_threads.next,
 				   struct svc_rqst,
 				   rq_list);
-		dprintk("svc: socket %p served by daemon %p\n",
-			svsk->sk_sk, rqstp);
+		dprintk("svc: transport %p served by daemon %p\n",
+			xprt, rqstp);
 		svc_thread_dequeue(pool, rqstp);
-		if (rqstp->rq_sock)
+		if (rqstp->rq_xprt)
 			printk(KERN_ERR
-				"svc_sock_enqueue: server %p, rq_sock=%p!\n",
-				rqstp, rqstp->rq_sock);
-		rqstp->rq_sock = svsk;
-		svc_xprt_get(&svsk->sk_xprt);
+				"svc_xprt_enqueue: server %p, rq_xprt=%p!\n",
+				rqstp, rqstp->rq_xprt);
+		rqstp->rq_xprt = xprt;
+		svc_xprt_get(xprt);
 		rqstp->rq_reserved = serv->sv_max_mesg;
-		atomic_add(rqstp->rq_reserved, &svsk->sk_xprt.xpt_reserved);
-		BUG_ON(svsk->sk_xprt.xpt_pool != pool);
+		atomic_add(rqstp->rq_reserved, &xprt->xpt_reserved);
+		BUG_ON(xprt->xpt_pool != pool);
 		wake_up(&rqstp->rq_wait);
 	} else {
-		dprintk("svc: socket %p put into queue\n", svsk->sk_sk);
-		list_add_tail(&svsk->sk_xprt.xpt_ready, &pool->sp_sockets);
-		BUG_ON(svsk->sk_xprt.xpt_pool != pool);
+		dprintk("svc: transport %p put into queue\n", xprt);
+		list_add_tail(&xprt->xpt_ready, &pool->sp_sockets);
+		BUG_ON(xprt->xpt_pool != pool);
 	}
 
 out_unlock:
 	spin_unlock_bh(&pool->sp_lock);
 }
+EXPORT_SYMBOL_GPL(svc_xprt_enqueue);
 
 /*
  * Dequeue the first socket.  Must be called with the pool->sp_lock held.
@@ -333,7 +335,7 @@ svc_sock_received(struct svc_sock *svsk)
 {
 	svsk->sk_xprt.xpt_pool = NULL;
 	clear_bit(XPT_BUSY, &svsk->sk_xprt.xpt_flags);
-	svc_sock_enqueue(svsk);
+	svc_xprt_enqueue(&svsk->sk_xprt);
 }
 
 
@@ -352,11 +354,11 @@ void svc_reserve(struct svc_rqst *rqstp, int space)
 	space += rqstp->rq_res.head[0].iov_len;
 
 	if (space < rqstp->rq_reserved) {
-		struct svc_sock *svsk = rqstp->rq_sock;
-		atomic_sub((rqstp->rq_reserved - space), &svsk->sk_xprt.xpt_reserved);
+		struct svc_xprt *xprt = rqstp->rq_xprt;
+		atomic_sub((rqstp->rq_reserved - space), &xprt->xpt_reserved);
 		rqstp->rq_reserved = space;
 
-		svc_sock_enqueue(svsk);
+		svc_xprt_enqueue(xprt);
 	}
 }
 
@@ -684,7 +686,7 @@ svc_udp_data_ready(struct sock *sk, int count)
 			svsk, sk, count,
 			test_bit(XPT_BUSY, &svsk->sk_xprt.xpt_flags));
 		set_bit(XPT_DATA, &svsk->sk_xprt.xpt_flags);
-		svc_sock_enqueue(svsk);
+		svc_xprt_enqueue(&svsk->sk_xprt);
 	}
 	if (sk->sk_sleep && waitqueue_active(sk->sk_sleep))
 		wake_up_interruptible(sk->sk_sleep);
@@ -701,7 +703,7 @@ svc_write_space(struct sock *sk)
 	if (svsk) {
 		dprintk("svc: socket %p(inet %p), write_space busy=%d\n",
 			svsk, sk, test_bit(XPT_BUSY, &svsk->sk_xprt.xpt_flags));
-		svc_sock_enqueue(svsk);
+		svc_xprt_enqueue(&svsk->sk_xprt);
 	}
 
 	if (sk->sk_sleep && waitqueue_active(sk->sk_sleep)) {
@@ -973,7 +975,7 @@ svc_tcp_listen_data_ready(struct sock *sk, int count_unused)
 	if (sk->sk_state == TCP_LISTEN) {
 		if (svsk) {
 			set_bit(XPT_CONN, &svsk->sk_xprt.xpt_flags);
-			svc_sock_enqueue(svsk);
+			svc_xprt_enqueue(&svsk->sk_xprt);
 		} else
 			printk("svc: socket %p: no user data\n", sk);
 	}
@@ -997,7 +999,7 @@ svc_tcp_state_change(struct sock *sk)
 		printk("svc: socket %p: no user data\n", sk);
 	else {
 		set_bit(XPT_CLOSE, &svsk->sk_xprt.xpt_flags);
-		svc_sock_enqueue(svsk);
+		svc_xprt_enqueue(&svsk->sk_xprt);
 	}
 	if (sk->sk_sleep && waitqueue_active(sk->sk_sleep))
 		wake_up_interruptible_all(sk->sk_sleep);
@@ -1012,7 +1014,7 @@ svc_tcp_data_ready(struct sock *sk, int count)
 		sk, sk->sk_user_data);
 	if (svsk) {
 		set_bit(XPT_DATA, &svsk->sk_xprt.xpt_flags);
-		svc_sock_enqueue(svsk);
+		svc_xprt_enqueue(&svsk->sk_xprt);
 	}
 	if (sk->sk_sleep && waitqueue_active(sk->sk_sleep))
 		wake_up_interruptible(sk->sk_sleep);
@@ -1298,7 +1300,7 @@ svc_tcp_sendto(struct svc_rqst *rqstp)
 		       (sent<0)?"got error":"sent only",
 		       sent, xbufp->len);
 		set_bit(XPT_CLOSE, &rqstp->rq_sock->sk_xprt.xpt_flags);
-		svc_sock_enqueue(rqstp->rq_sock);
+		svc_xprt_enqueue(rqstp->rq_xprt);
 		sent = -EAGAIN;
 	}
 	return sent;
@@ -1481,7 +1483,7 @@ static void svc_check_conn_limits(struct svc_serv *serv)
 		spin_unlock_bh(&serv->sv_lock);
 
 		if (svsk) {
-			svc_sock_enqueue(svsk);
+			svc_xprt_enqueue(&svsk->sk_xprt);
 			svc_xprt_put(&svsk->sk_xprt);
 		}
 	}
@@ -1714,7 +1716,7 @@ svc_age_temp_sockets(unsigned long closure)
 			svsk, get_seconds() - svsk->sk_lastrecv);
 
 		/* a thread will dequeue and close it soon */
-		svc_sock_enqueue(svsk);
+		svc_xprt_enqueue(&svsk->sk_xprt);
 		svc_xprt_put(&svsk->sk_xprt);
 	}
 
@@ -1996,7 +1998,7 @@ static void svc_revisit(struct cache_deferred_req *dreq, int too_many)
 	list_add(&dr->handle.recent, &svsk->sk_deferred);
 	spin_unlock(&svsk->sk_lock);
 	set_bit(XPT_DEFERRED, &svsk->sk_xprt.xpt_flags);
-	svc_sock_enqueue(svsk);
+	svc_xprt_enqueue(&svsk->sk_xprt);
 	svc_xprt_put(&svsk->sk_xprt);
 }
 

[PATCH 20/38] svc: Make svc_send transport neutral

[Tom Tucker]

Move the sk_mutex field to the transport independent svc_xprt structure.
Now all the fields that svc_send touches are transport neutral. Change the 
svc_send function to use the transport independent svc_xprt directly instead 
of the transport dependent svc_sock structure.

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 include/linux/sunrpc/svc_xprt.h |    1 +
 include/linux/sunrpc/svcsock.h  |    1 -
 net/sunrpc/svc_xprt.c           |    1 +
 net/sunrpc/svcsock.c            |   19 ++++++++-----------
 4 files changed, 10 insertions(+), 12 deletions(-)

diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
index c9892d5..d5ef902 100644
--- a/include/linux/sunrpc/svc_xprt.h
+++ b/include/linux/sunrpc/svc_xprt.h
@@ -54,6 +54,7 @@ struct svc_xprt {
 	struct svc_pool		*xpt_pool;	/* current pool iff queued */
 	struct svc_serv		*xpt_server;	/* service for transport */
 	atomic_t    	    	xpt_reserved;	/* space on outq that is rsvd */
+	struct mutex		xpt_mutex;	/* to serialize sending data */
 };
 
 int	svc_reg_xprt_class(struct svc_xprt_class *);
diff --git a/include/linux/sunrpc/svcsock.h b/include/linux/sunrpc/svcsock.h
index ba41f11..41c2dfa 100644
--- a/include/linux/sunrpc/svcsock.h
+++ b/include/linux/sunrpc/svcsock.h
@@ -24,7 +24,6 @@ struct svc_sock {
 						 * sk_info_authunix */
 	struct list_head	sk_deferred;	/* deferred requests that need to
 						 * be revisted */
-	struct mutex		sk_mutex;	/* to serialize sending data */
 
 	/* We keep the old state_change and data_ready CB's here */
 	void			(*sk_ostate)(struct sock *);
diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
index bbdada7..7544102 100644
--- a/net/sunrpc/svc_xprt.c
+++ b/net/sunrpc/svc_xprt.c
@@ -112,6 +112,7 @@ void svc_xprt_init(struct svc_xprt_class *xcl, struct svc_xprt *xprt,
 	xprt->xpt_server = serv;
 	INIT_LIST_HEAD(&xprt->xpt_list);
 	INIT_LIST_HEAD(&xprt->xpt_ready);
+	mutex_init(&xprt->xpt_mutex);
 }
 EXPORT_SYMBOL_GPL(svc_xprt_init);
 
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index d8df8c7..d546709 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -1637,15 +1637,13 @@ svc_drop(struct svc_rqst *rqstp)
 int
 svc_send(struct svc_rqst *rqstp)
 {
-	struct svc_sock	*svsk;
+	struct svc_xprt	*xprt;
 	int		len;
 	struct xdr_buf	*xb;
 
-	if ((svsk = rqstp->rq_sock) == NULL) {
-		printk(KERN_WARNING "NULL socket pointer in %s:%d\n",
-				__FILE__, __LINE__);
+	xprt = rqstp->rq_xprt;
+	if (!xprt)
 		return -EFAULT;
-	}
 
 	/* release the receive skb before sending the reply */
 	rqstp->rq_xprt->xpt_ops->xpo_release_rqst(rqstp);
@@ -1656,13 +1654,13 @@ svc_send(struct svc_rqst *rqstp)
 		xb->page_len +
 		xb->tail[0].iov_len;
 
-	/* Grab svsk->sk_mutex to serialize outgoing data. */
-	mutex_lock(&svsk->sk_mutex);
-	if (test_bit(XPT_DEAD, &svsk->sk_xprt.xpt_flags))
+	/* Grab mutex to serialize outgoing data. */
+	mutex_lock(&xprt->xpt_mutex);
+	if (test_bit(XPT_DEAD, &xprt->xpt_flags))
 		len = -ENOTCONN;
 	else
-		len = svsk->sk_xprt.xpt_ops->xpo_sendto(rqstp);
-	mutex_unlock(&svsk->sk_mutex);
+		len = xprt->xpt_ops->xpo_sendto(rqstp);
+	mutex_unlock(&xprt->xpt_mutex);
 	svc_sock_release(rqstp);
 
 	if (len == -ECONNREFUSED || len == -ENOTCONN || len == -EAGAIN)
@@ -1764,7 +1762,6 @@ static struct svc_sock *svc_setup_socket(struct svc_serv *serv,
 	svsk->sk_lastrecv = get_seconds();
 	spin_lock_init(&svsk->sk_lock);
 	INIT_LIST_HEAD(&svsk->sk_deferred);
-	mutex_init(&svsk->sk_mutex);
 
 	/* Initialize the socket */
 	if (sock->type == SOCK_DGRAM)

[PATCH 21/38] svc: Change svc_sock_received to svc_xprt_received and export it

[Tom Tucker]

All fields touched by svc_sock_received are now transport independent.
Change it to use svc_xprt directly. This function is called from
transport dependent code, so export it. 

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 include/linux/sunrpc/svc_xprt.h |    2 +-
 net/sunrpc/svcsock.c            |   35 +++++++++++++++++------------------
 2 files changed, 18 insertions(+), 19 deletions(-)

diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
index d5ef902..c416d05 100644
--- a/include/linux/sunrpc/svc_xprt.h
+++ b/include/linux/sunrpc/svc_xprt.h
@@ -62,8 +62,8 @@ int	svc_unreg_xprt_class(struct svc_xprt_class *);
 void	svc_xprt_init(struct svc_xprt_class *, struct svc_xprt *,
 		      struct svc_serv *);
 int	svc_create_xprt(struct svc_serv *, char *, unsigned short, int);
+void	svc_xprt_received(struct svc_xprt *);
 void	svc_xprt_put(struct svc_xprt *xprt);
-
 static inline void svc_xprt_get(struct svc_xprt *xprt)
 {
 	kref_get(&xprt->xpt_ref);
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index d546709..d8d88c1 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -330,14 +330,13 @@ svc_sock_dequeue(struct svc_pool *pool)
  * Note: XPT_DATA only gets cleared when a read-attempt finds
  * no (or insufficient) data.
  */
-static inline void
-svc_sock_received(struct svc_sock *svsk)
+void svc_xprt_received(struct svc_xprt *xprt)
 {
-	svsk->sk_xprt.xpt_pool = NULL;
-	clear_bit(XPT_BUSY, &svsk->sk_xprt.xpt_flags);
-	svc_xprt_enqueue(&svsk->sk_xprt);
+	xprt->xpt_pool = NULL;
+	clear_bit(XPT_BUSY, &xprt->xpt_flags);
+	svc_xprt_enqueue(xprt);
 }
-
+EXPORT_SYMBOL_GPL(svc_xprt_received);
 
 /**
  * svc_reserve - change the space reserved for the reply to a request.
@@ -766,7 +765,7 @@ svc_udp_recvfrom(struct svc_rqst *rqstp)
 				(serv->sv_nrthreads+3) * serv->sv_max_mesg);
 
 	if ((rqstp->rq_deferred = svc_deferred_dequeue(svsk))) {
-		svc_sock_received(svsk);
+		svc_xprt_received(&svsk->sk_xprt);
 		return svc_deferred_recv(rqstp);
 	}
 
@@ -783,7 +782,7 @@ svc_udp_recvfrom(struct svc_rqst *rqstp)
 			dprintk("svc: recvfrom returned error %d\n", -err);
 			set_bit(XPT_DATA, &svsk->sk_xprt.xpt_flags);
 		}
-		svc_sock_received(svsk);
+		svc_xprt_received(&svsk->sk_xprt);
 		return -EAGAIN;
 	}
 	rqstp->rq_addrlen = sizeof(rqstp->rq_addr);
@@ -798,7 +797,7 @@ svc_udp_recvfrom(struct svc_rqst *rqstp)
 	/*
 	 * Maybe more packets - kick another thread ASAP.
 	 */
-	svc_sock_received(svsk);
+	svc_xprt_received(&svsk->sk_xprt);
 
 	len  = skb->len - sizeof(struct udphdr);
 	rqstp->rq_arg.len = len;
@@ -1104,7 +1103,7 @@ static struct svc_xprt *svc_tcp_accept(struct svc_xprt *xprt)
 	}
 	memcpy(&newsvsk->sk_local, sin, slen);
 
-	svc_sock_received(newsvsk);
+	svc_xprt_received(&newsvsk->sk_xprt);
 
 	if (serv->sv_stats)
 		serv->sv_stats->nettcpconn++;
@@ -1134,7 +1133,7 @@ svc_tcp_recvfrom(struct svc_rqst *rqstp)
 		test_bit(XPT_CLOSE, &svsk->sk_xprt.xpt_flags));
 
 	if ((rqstp->rq_deferred = svc_deferred_dequeue(svsk))) {
-		svc_sock_received(svsk);
+		svc_xprt_received(&svsk->sk_xprt);
 		return svc_deferred_recv(rqstp);
 	}
 
@@ -1174,7 +1173,7 @@ svc_tcp_recvfrom(struct svc_rqst *rqstp)
 		if (len < want) {
 			dprintk("svc: short recvfrom while reading record length (%d of %lu)\n",
 				len, want);
-			svc_sock_received(svsk);
+			svc_xprt_received(&svsk->sk_xprt);
 			return -EAGAIN; /* record header not complete */
 		}
 
@@ -1210,7 +1209,7 @@ svc_tcp_recvfrom(struct svc_rqst *rqstp)
 	if (len < svsk->sk_reclen) {
 		dprintk("svc: incomplete TCP record (%d of %d)\n",
 			len, svsk->sk_reclen);
-		svc_sock_received(svsk);
+		svc_xprt_received(&svsk->sk_xprt);
 		return -EAGAIN;	/* record not complete */
 	}
 	len = svsk->sk_reclen;
@@ -1250,7 +1249,7 @@ svc_tcp_recvfrom(struct svc_rqst *rqstp)
 	svsk->sk_reclen = 0;
 	svsk->sk_tcplen = 0;
 
-	svc_sock_received(svsk);
+	svc_xprt_received(&svsk->sk_xprt);
 	if (serv->sv_stats)
 		serv->sv_stats->nettcpcnt++;
 
@@ -1263,7 +1262,7 @@ svc_tcp_recvfrom(struct svc_rqst *rqstp)
  error:
 	if (len == -EAGAIN) {
 		dprintk("RPC: TCP recvfrom got EAGAIN\n");
-		svc_sock_received(svsk);
+		svc_xprt_received(&svsk->sk_xprt);
 	} else {
 		printk(KERN_NOTICE "%s: recvfrom returned errno %d\n",
 		       svsk->sk_xprt.xpt_server->sv_name, -len);
@@ -1595,7 +1594,7 @@ svc_recv(struct svc_rqst *rqstp, long timeout)
 			__module_get(newxpt->xpt_class->xcl_owner);
 			svc_check_conn_limits(svsk->sk_xprt.xpt_server);
 		}
-		svc_sock_received(svsk);
+		svc_xprt_received(&svsk->sk_xprt);
 	} else {
 		dprintk("svc: server %p, pool %u, socket %p, inuse=%d\n",
 			rqstp, pool->sp_id, svsk,
@@ -1814,7 +1813,7 @@ int svc_addsock(struct svc_serv *serv,
 	else {
 		svsk = svc_setup_socket(serv, so, &err, SVC_SOCK_DEFAULTS);
 		if (svsk) {
-			svc_sock_received(svsk);
+			svc_xprt_received(&svsk->sk_xprt);
 			err = 0;
 		}
 	}
@@ -1870,7 +1869,7 @@ static struct svc_xprt *svc_create_socket(struct svc_serv *serv,
 	}
 
 	if ((svsk = svc_setup_socket(serv, sock, &error, flags)) != NULL) {
-		svc_sock_received(svsk);
+		svc_xprt_received(&svsk->sk_xprt);
 		return (struct svc_xprt *)svsk;
 	}
 

[PATCH 22/38] svc: Move accept call to svc_xprt_received to common code

[Tom Tucker]

Now that the svc_xprt_received function handles transports, the call
to svc_xprt_received in the xpo_tcp_accept function can be moved to
common code.

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 net/sunrpc/svcsock.c |    3 +--
 1 files changed, 1 insertions(+), 2 deletions(-)

diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index d8d88c1..c637acf 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -1103,8 +1103,6 @@ static struct svc_xprt *svc_tcp_accept(struct svc_xprt *xprt)
 	}
 	memcpy(&newsvsk->sk_local, sin, slen);
 
-	svc_xprt_received(&newsvsk->sk_xprt);
-
 	if (serv->sv_stats)
 		serv->sv_stats->nettcpconn++;
 
@@ -1593,6 +1591,7 @@ svc_recv(struct svc_rqst *rqstp, long timeout)
 			 */
 			__module_get(newxpt->xpt_class->xcl_owner);
 			svc_check_conn_limits(svsk->sk_xprt.xpt_server);
+			svc_xprt_received(newxpt);
 		}
 		svc_xprt_received(&svsk->sk_xprt);
 	} else {

[PATCH 23/38] svc: Remove sk_lastrecv

[Tom Tucker]

With the implementation of the new mark and sweep algorithm for shutting
down old connections, the sk_lastrecv field is no longer needed.

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 include/linux/sunrpc/svcsock.h |    1 -
 net/sunrpc/svcsock.c           |    5 +----
 2 files changed, 1 insertions(+), 5 deletions(-)

diff --git a/include/linux/sunrpc/svcsock.h b/include/linux/sunrpc/svcsock.h
index 41c2dfa..406d003 100644
--- a/include/linux/sunrpc/svcsock.h
+++ b/include/linux/sunrpc/svcsock.h
@@ -33,7 +33,6 @@ struct svc_sock {
 	/* private TCP part */
 	int			sk_reclen;	/* length of record */
 	int			sk_tcplen;	/* current read length */
-	time_t			sk_lastrecv;	/* time of last received request */
 
 	/* cache of various info for TCP sockets */
 	void			*sk_info_authunix;
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index c637acf..a7af4cf 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -1608,7 +1608,6 @@ svc_recv(struct svc_rqst *rqstp, long timeout)
 		svc_sock_release(rqstp);
 		return -EAGAIN;
 	}
-	svsk->sk_lastrecv = get_seconds();
 	clear_bit(XPT_OLD, &svsk->sk_xprt.xpt_flags);
 
 	rqstp->rq_secure = svc_port_is_privileged(svc_addr(rqstp));
@@ -1708,8 +1707,7 @@ svc_age_temp_sockets(unsigned long closure)
 		list_del_init(le);
 		svsk = list_entry(le, struct svc_sock, sk_xprt.xpt_list);
 
-		dprintk("queuing svsk %p for closing, %lu seconds old\n",
-			svsk, get_seconds() - svsk->sk_lastrecv);
+		dprintk("queuing svsk %p for closing\n", svsk);
 
 		/* a thread will dequeue and close it soon */
 		svc_xprt_enqueue(&svsk->sk_xprt);
@@ -1757,7 +1755,6 @@ static struct svc_sock *svc_setup_socket(struct svc_serv *serv,
 	svsk->sk_ostate = inet->sk_state_change;
 	svsk->sk_odata = inet->sk_data_ready;
 	svsk->sk_owspace = inet->sk_write_space;
-	svsk->sk_lastrecv = get_seconds();
 	spin_lock_init(&svsk->sk_lock);
 	INIT_LIST_HEAD(&svsk->sk_deferred);
 

[PATCH 24/38] svc: Move the authinfo cache to svc_xprt.

[Tom Tucker]

Move the authinfo cache to svc_xprt. This allows both the TCP and RDMA
transports to share this logic. A flag bit is used to determine if
auth information is to be cached or not. Previously, this code looked
at the transport protocol.

I've also changed the spin_lock/unlock logic so that a lock is not taken for 
transports that are not caching auth info.  

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 include/linux/sunrpc/svc_xprt.h |    4 +++
 include/linux/sunrpc/svcsock.h  |    5 ----
 net/sunrpc/svc_xprt.c           |    4 +++
 net/sunrpc/svcauth_unix.c       |   54 +++++++++++++++++++++------------------
 net/sunrpc/svcsock.c            |   22 +++++++++-------
 5 files changed, 49 insertions(+), 40 deletions(-)

diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
index c416d05..dfb1d4d 100644
--- a/include/linux/sunrpc/svc_xprt.h
+++ b/include/linux/sunrpc/svc_xprt.h
@@ -50,11 +50,15 @@ struct svc_xprt {
 #define	XPT_OLD		9		/* used for xprt aging mark+sweep */
 #define	XPT_DETACHED	10		/* detached from tempsocks list */
 #define XPT_LISTENER	11		/* listening endpoint */
+#define XPT_CACHE_AUTH	12		/* cache auth info */
 
 	struct svc_pool		*xpt_pool;	/* current pool iff queued */
 	struct svc_serv		*xpt_server;	/* service for transport */
 	atomic_t    	    	xpt_reserved;	/* space on outq that is rsvd */
 	struct mutex		xpt_mutex;	/* to serialize sending data */
+	spinlock_t		xpt_lock;	/* protects sk_deferred
+						 * and xpt_auth_cache */
+	void			*xpt_auth_cache;/* auth cache */
 };
 
 int	svc_reg_xprt_class(struct svc_xprt_class *);
diff --git a/include/linux/sunrpc/svcsock.h b/include/linux/sunrpc/svcsock.h
index 406d003..f2ed6a2 100644
--- a/include/linux/sunrpc/svcsock.h
+++ b/include/linux/sunrpc/svcsock.h
@@ -20,8 +20,6 @@ struct svc_sock {
 	struct socket *		sk_sock;	/* berkeley socket layer */
 	struct sock *		sk_sk;		/* INET layer */
 
-	spinlock_t		sk_lock;	/* protects sk_deferred and
-						 * sk_info_authunix */
 	struct list_head	sk_deferred;	/* deferred requests that need to
 						 * be revisted */
 
@@ -34,9 +32,6 @@ struct svc_sock {
 	int			sk_reclen;	/* length of record */
 	int			sk_tcplen;	/* current read length */
 
-	/* cache of various info for TCP sockets */
-	void			*sk_info_authunix;
-
 	struct sockaddr_storage	sk_local;	/* local address */
 	struct sockaddr_storage	sk_remote;	/* remote peer's address */
 	int			sk_remotelen;	/* length of address */
diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
index 7544102..79009c2 100644
--- a/net/sunrpc/svc_xprt.c
+++ b/net/sunrpc/svc_xprt.c
@@ -88,6 +88,9 @@ static void svc_xprt_free(struct kref *kref)
 		container_of(kref, struct svc_xprt, xpt_ref);
 	struct module *owner = xprt->xpt_class->xcl_owner;
 	BUG_ON(atomic_read(&kref->refcount));
+	if (test_bit(XPT_CACHE_AUTH, &xprt->xpt_flags)
+	    && xprt->xpt_auth_cache != NULL)
+		svcauth_unix_info_release(xprt->xpt_auth_cache);
 	xprt->xpt_ops->xpo_free(xprt);
 	module_put(owner);
 }
@@ -113,6 +116,7 @@ void svc_xprt_init(struct svc_xprt_class *xcl, struct svc_xprt *xprt,
 	INIT_LIST_HEAD(&xprt->xpt_list);
 	INIT_LIST_HEAD(&xprt->xpt_ready);
 	mutex_init(&xprt->xpt_mutex);
+	spin_lock_init(&xprt->xpt_lock);
 }
 EXPORT_SYMBOL_GPL(svc_xprt_init);
 
diff --git a/net/sunrpc/svcauth_unix.c b/net/sunrpc/svcauth_unix.c
index 4114794..6815157 100644
--- a/net/sunrpc/svcauth_unix.c
+++ b/net/sunrpc/svcauth_unix.c
@@ -384,41 +384,45 @@ void svcauth_unix_purge(void)
 static inline struct ip_map *
 ip_map_cached_get(struct svc_rqst *rqstp)
 {
-	struct ip_map *ipm;
-	struct svc_sock *svsk = rqstp->rq_sock;
-	spin_lock(&svsk->sk_lock);
-	ipm = svsk->sk_info_authunix;
-	if (ipm != NULL) {
-		if (!cache_valid(&ipm->h)) {
-			/*
-			 * The entry has been invalidated since it was
-			 * remembered, e.g. by a second mount from the
-			 * same IP address.
-			 */
-			svsk->sk_info_authunix = NULL;
-			spin_unlock(&svsk->sk_lock);
-			cache_put(&ipm->h, &ip_map_cache);
-			return NULL;
+	struct ip_map *ipm = NULL;
+	struct svc_xprt *xprt = rqstp->rq_xprt;
+
+	if (test_bit(XPT_CACHE_AUTH, &xprt->xpt_flags)) {
+		spin_lock(&xprt->xpt_lock);
+		ipm = xprt->xpt_auth_cache;
+		if (ipm != NULL) {
+			if (!cache_valid(&ipm->h)) {
+				/*
+				 * The entry has been invalidated since it was
+				 * remembered, e.g. by a second mount from the
+				 * same IP address.
+				 */
+				xprt->xpt_auth_cache = NULL;
+				spin_unlock(&xprt->xpt_lock);
+				cache_put(&ipm->h, &ip_map_cache);
+				return NULL;
+			}
+			cache_get(&ipm->h);
 		}
-		cache_get(&ipm->h);
+		spin_unlock(&xprt->xpt_lock);
 	}
-	spin_unlock(&svsk->sk_lock);
 	return ipm;
 }
 
 static inline void
 ip_map_cached_put(struct svc_rqst *rqstp, struct ip_map *ipm)
 {
-	struct svc_sock *svsk = rqstp->rq_sock;
+	struct svc_xprt *xprt = rqstp->rq_xprt;
 
-	spin_lock(&svsk->sk_lock);
-	if (svsk->sk_sock->type == SOCK_STREAM &&
-	    svsk->sk_info_authunix == NULL) {
-		/* newly cached, keep the reference */
-		svsk->sk_info_authunix = ipm;
-		ipm = NULL;
+	if (test_bit(XPT_CACHE_AUTH, &xprt->xpt_flags)) {
+		spin_lock(&xprt->xpt_lock);
+		if (xprt->xpt_auth_cache == NULL) {
+			/* newly cached, keep the reference */
+			xprt->xpt_auth_cache = ipm;
+			ipm = NULL;
+		}
+		spin_unlock(&xprt->xpt_lock);
 	}
-	spin_unlock(&svsk->sk_lock);
 	if (ipm)
 		cache_put(&ipm->h, &ip_map_cache);
 }
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index a7af4cf..328155a 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -113,12 +113,16 @@ static inline void svc_reclassify_socket(struct socket *sock)
 	switch (sk->sk_family) {
 	case AF_INET:
 		sock_lock_init_class_and_name(sk, "slock-AF_INET-NFSD",
-		    &svc_slock_key[0], "sk_lock-AF_INET-NFSD", &svc_key[0]);
+					      &svc_slock_key[0],
+					      "sk_xprt.xpt_lock-AF_INET-NFSD",
+					      &svc_key[0]);
 		break;
 
 	case AF_INET6:
 		sock_lock_init_class_and_name(sk, "slock-AF_INET6-NFSD",
-		    &svc_slock_key[1], "sk_lock-AF_INET6-NFSD", &svc_key[1]);
+					      &svc_slock_key[1],
+					      "sk_xprt.xpt_lock-AF_INET6-NFSD",
+					      &svc_key[1]);
 		break;
 
 	default:
@@ -927,6 +931,7 @@ static void svc_udp_init(struct svc_sock *svsk, struct svc_serv *serv)
 	mm_segment_t oldfs;
 
 	svc_xprt_init(&svc_udp_class, &svsk->sk_xprt, serv);
+	clear_bit(XPT_CACHE_AUTH, &svsk->sk_xprt.xpt_flags);
 	svsk->sk_sk->sk_data_ready = svc_udp_data_ready;
 	svsk->sk_sk->sk_write_space = svc_write_space;
 
@@ -1382,7 +1387,7 @@ static void svc_tcp_init(struct svc_sock *svsk, struct svc_serv *serv)
 	struct tcp_sock *tp = tcp_sk(sk);
 
 	svc_xprt_init(&svc_tcp_class, &svsk->sk_xprt, serv);
-
+	set_bit(XPT_CACHE_AUTH, &svsk->sk_xprt.xpt_flags);
 	if (sk->sk_state == TCP_LISTEN) {
 		dprintk("setting up TCP socket for listening\n");
 		set_bit(XPT_LISTENER, &svsk->sk_xprt.xpt_flags);
@@ -1755,7 +1760,6 @@ static struct svc_sock *svc_setup_socket(struct svc_serv *serv,
 	svsk->sk_ostate = inet->sk_state_change;
 	svsk->sk_odata = inet->sk_data_ready;
 	svsk->sk_owspace = inet->sk_write_space;
-	spin_lock_init(&svsk->sk_lock);
 	INIT_LIST_HEAD(&svsk->sk_deferred);
 
 	/* Initialize the socket */
@@ -1900,8 +1904,6 @@ static void svc_sock_free(struct svc_xprt *xprt)
 	struct svc_sock *svsk = container_of(xprt, struct svc_sock, sk_xprt);
 	dprintk("svc: svc_sock_free(%p)\n", svsk);
 
-	if (svsk->sk_info_authunix != NULL)
-		svcauth_unix_info_release(svsk->sk_info_authunix);
 	if (svsk->sk_sock->file)
 		sockfd_put(svsk->sk_sock);
 	else
@@ -1986,9 +1988,9 @@ static void svc_revisit(struct cache_deferred_req *dreq, int too_many)
 	dprintk("revisit queued\n");
 	svsk = dr->svsk;
 	dr->svsk = NULL;
-	spin_lock(&svsk->sk_lock);
+	spin_lock(&svsk->sk_xprt.xpt_lock);
 	list_add(&dr->handle.recent, &svsk->sk_deferred);
-	spin_unlock(&svsk->sk_lock);
+	spin_unlock(&svsk->sk_xprt.xpt_lock);
 	set_bit(XPT_DEFERRED, &svsk->sk_xprt.xpt_flags);
 	svc_xprt_enqueue(&svsk->sk_xprt);
 	svc_xprt_put(&svsk->sk_xprt);
@@ -2054,7 +2056,7 @@ static struct svc_deferred_req *svc_deferred_dequeue(struct svc_sock *svsk)
 
 	if (!test_bit(XPT_DEFERRED, &svsk->sk_xprt.xpt_flags))
 		return NULL;
-	spin_lock(&svsk->sk_lock);
+	spin_lock(&svsk->sk_xprt.xpt_lock);
 	clear_bit(XPT_DEFERRED, &svsk->sk_xprt.xpt_flags);
 	if (!list_empty(&svsk->sk_deferred)) {
 		dr = list_entry(svsk->sk_deferred.next,
@@ -2063,7 +2065,7 @@ static struct svc_deferred_req *svc_deferred_dequeue(struct svc_sock *svsk)
 		list_del_init(&dr->handle.recent);
 		set_bit(XPT_DEFERRED, &svsk->sk_xprt.xpt_flags);
 	}
-	spin_unlock(&svsk->sk_lock);
+	spin_unlock(&svsk->sk_xprt.xpt_lock);
 	return dr;
 }
 

[PATCH 25/38] svc: Make deferral processing xprt independent

[Tom Tucker]

This patch moves the transport independent sk_deferred list to the svc_xprt
structure and updates the svc_deferred_req structure to keep pointers to
svc_xprt's directly. The deferral processing code is also moved out of the
transport dependent recvfrom functions and into the generic svc_recv path.

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 include/linux/sunrpc/svc.h      |    2 +
 include/linux/sunrpc/svc_xprt.h |    2 +
 include/linux/sunrpc/svcsock.h  |    3 --
 net/sunrpc/svc_xprt.c           |    1 +
 net/sunrpc/svcsock.c            |   58 +++++++++++++++++----------------------
 5 files changed, 29 insertions(+), 37 deletions(-)

diff --git a/include/linux/sunrpc/svc.h b/include/linux/sunrpc/svc.h
index cfb2652..40adc9d 100644
--- a/include/linux/sunrpc/svc.h
+++ b/include/linux/sunrpc/svc.h
@@ -320,7 +320,7 @@ static inline void svc_free_res_pages(struct svc_rqst *rqstp)
 
 struct svc_deferred_req {
 	u32			prot;	/* protocol (UDP or TCP) */
-	struct svc_sock		*svsk;
+	struct svc_xprt		*xprt;
 	struct sockaddr_storage	addr;	/* where reply must go */
 	size_t			addrlen;
 	union svc_addr_u	daddr;	/* where reply must come from */
diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
index dfb1d4d..d93ae27 100644
--- a/include/linux/sunrpc/svc_xprt.h
+++ b/include/linux/sunrpc/svc_xprt.h
@@ -59,6 +59,8 @@ struct svc_xprt {
 	spinlock_t		xpt_lock;	/* protects sk_deferred
 						 * and xpt_auth_cache */
 	void			*xpt_auth_cache;/* auth cache */
+	struct list_head	xpt_deferred;	/* deferred requests that need
+						 * to be revisted */
 };
 
 int	svc_reg_xprt_class(struct svc_xprt_class *);
diff --git a/include/linux/sunrpc/svcsock.h b/include/linux/sunrpc/svcsock.h
index f2ed6a2..96a229e 100644
--- a/include/linux/sunrpc/svcsock.h
+++ b/include/linux/sunrpc/svcsock.h
@@ -20,9 +20,6 @@ struct svc_sock {
 	struct socket *		sk_sock;	/* berkeley socket layer */
 	struct sock *		sk_sk;		/* INET layer */
 
-	struct list_head	sk_deferred;	/* deferred requests that need to
-						 * be revisted */
-
 	/* We keep the old state_change and data_ready CB's here */
 	void			(*sk_ostate)(struct sock *);
 	void			(*sk_odata)(struct sock *, int bytes);
diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
index 79009c2..fdf0d8c 100644
--- a/net/sunrpc/svc_xprt.c
+++ b/net/sunrpc/svc_xprt.c
@@ -115,6 +115,7 @@ void svc_xprt_init(struct svc_xprt_class *xcl, struct svc_xprt *xprt,
 	xprt->xpt_server = serv;
 	INIT_LIST_HEAD(&xprt->xpt_list);
 	INIT_LIST_HEAD(&xprt->xpt_ready);
+	INIT_LIST_HEAD(&xprt->xpt_deferred);
 	mutex_init(&xprt->xpt_mutex);
 	spin_lock_init(&xprt->xpt_lock);
 }
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index 328155a..887729f 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -89,7 +89,7 @@ static void		svc_close_xprt(struct svc_xprt *xprt);
 static void		svc_sock_detach(struct svc_xprt *);
 static void		svc_sock_free(struct svc_xprt *);
 
-static struct svc_deferred_req *svc_deferred_dequeue(struct svc_sock *svsk);
+static struct svc_deferred_req *svc_deferred_dequeue(struct svc_xprt *xprt);
 static int svc_deferred_recv(struct svc_rqst *rqstp);
 static struct cache_deferred_req *svc_defer(struct cache_req *req);
 static struct svc_xprt *svc_create_socket(struct svc_serv *, int,
@@ -768,11 +768,6 @@ svc_udp_recvfrom(struct svc_rqst *rqstp)
 				(serv->sv_nrthreads+3) * serv->sv_max_mesg,
 				(serv->sv_nrthreads+3) * serv->sv_max_mesg);
 
-	if ((rqstp->rq_deferred = svc_deferred_dequeue(svsk))) {
-		svc_xprt_received(&svsk->sk_xprt);
-		return svc_deferred_recv(rqstp);
-	}
-
 	clear_bit(XPT_DATA, &svsk->sk_xprt.xpt_flags);
 	skb = NULL;
 	err = kernel_recvmsg(svsk->sk_sock, &msg, NULL,
@@ -1135,11 +1130,6 @@ svc_tcp_recvfrom(struct svc_rqst *rqstp)
 		test_bit(XPT_CONN, &svsk->sk_xprt.xpt_flags),
 		test_bit(XPT_CLOSE, &svsk->sk_xprt.xpt_flags));
 
-	if ((rqstp->rq_deferred = svc_deferred_dequeue(svsk))) {
-		svc_xprt_received(&svsk->sk_xprt);
-		return svc_deferred_recv(rqstp);
-	}
-
 	if (test_and_clear_bit(XPT_CHNGBUF, &svsk->sk_xprt.xpt_flags))
 		/* sndbuf needs to have room for one request
 		 * per thread, otherwise we can stall even when the
@@ -1603,7 +1593,12 @@ svc_recv(struct svc_rqst *rqstp, long timeout)
 		dprintk("svc: server %p, pool %u, socket %p, inuse=%d\n",
 			rqstp, pool->sp_id, svsk,
 			atomic_read(&svsk->sk_xprt.xpt_ref.refcount));
-		len = svsk->sk_xprt.xpt_ops->xpo_recvfrom(rqstp);
+		rqstp->rq_deferred = svc_deferred_dequeue(&svsk->sk_xprt);
+		if (rqstp->rq_deferred) {
+			svc_xprt_received(&svsk->sk_xprt);
+			len = svc_deferred_recv(rqstp);
+		} else
+			len = svsk->sk_xprt.xpt_ops->xpo_recvfrom(rqstp);
 		dprintk("svc: got len=%d\n", len);
 	}
 
@@ -1760,7 +1755,6 @@ static struct svc_sock *svc_setup_socket(struct svc_serv *serv,
 	svsk->sk_ostate = inet->sk_state_change;
 	svsk->sk_odata = inet->sk_data_ready;
 	svsk->sk_owspace = inet->sk_write_space;
-	INIT_LIST_HEAD(&svsk->sk_deferred);
 
 	/* Initialize the socket */
 	if (sock->type == SOCK_DGRAM)
@@ -1978,22 +1972,21 @@ void svc_close_all(struct list_head *xprt_list)
 static void svc_revisit(struct cache_deferred_req *dreq, int too_many)
 {
 	struct svc_deferred_req *dr = container_of(dreq, struct svc_deferred_req, handle);
-	struct svc_sock *svsk;
+	struct svc_xprt *xprt = dr->xprt;
 
 	if (too_many) {
-		svc_xprt_put(&dr->svsk->sk_xprt);
+		svc_xprt_put(xprt);
 		kfree(dr);
 		return;
 	}
 	dprintk("revisit queued\n");
-	svsk = dr->svsk;
-	dr->svsk = NULL;
-	spin_lock(&svsk->sk_xprt.xpt_lock);
-	list_add(&dr->handle.recent, &svsk->sk_deferred);
-	spin_unlock(&svsk->sk_xprt.xpt_lock);
-	set_bit(XPT_DEFERRED, &svsk->sk_xprt.xpt_flags);
-	svc_xprt_enqueue(&svsk->sk_xprt);
-	svc_xprt_put(&svsk->sk_xprt);
+	dr->xprt = NULL;
+	spin_lock(&xprt->xpt_lock);
+	list_add(&dr->handle.recent, &xprt->xpt_deferred);
+	spin_unlock(&xprt->xpt_lock);
+	set_bit(XPT_DEFERRED, &xprt->xpt_flags);
+	svc_xprt_enqueue(xprt);
+	svc_xprt_put(xprt);
 }
 
 static struct cache_deferred_req *
@@ -2024,7 +2017,7 @@ svc_defer(struct cache_req *req)
 		memcpy(dr->args, rqstp->rq_arg.head[0].iov_base-skip, dr->argslen<<2);
 	}
 	svc_xprt_get(rqstp->rq_xprt);
-	dr->svsk = rqstp->rq_sock;
+	dr->xprt = rqstp->rq_xprt;
 
 	dr->handle.revisit = svc_revisit;
 	return &dr->handle;
@@ -2050,22 +2043,21 @@ static int svc_deferred_recv(struct svc_rqst *rqstp)
 }
 
 
-static struct svc_deferred_req *svc_deferred_dequeue(struct svc_sock *svsk)
+static struct svc_deferred_req *svc_deferred_dequeue(struct svc_xprt *xprt)
 {
 	struct svc_deferred_req *dr = NULL;
 
-	if (!test_bit(XPT_DEFERRED, &svsk->sk_xprt.xpt_flags))
+	if (!test_bit(XPT_DEFERRED, &xprt->xpt_flags))
 		return NULL;
-	spin_lock(&svsk->sk_xprt.xpt_lock);
-	clear_bit(XPT_DEFERRED, &svsk->sk_xprt.xpt_flags);
-	if (!list_empty(&svsk->sk_deferred)) {
-		dr = list_entry(svsk->sk_deferred.next,
+	spin_lock(&xprt->xpt_lock);
+	clear_bit(XPT_DEFERRED, &xprt->xpt_flags);
+	if (!list_empty(&xprt->xpt_deferred)) {
+		dr = list_entry(xprt->xpt_deferred.next,
 				struct svc_deferred_req,
 				handle.recent);
 		list_del_init(&dr->handle.recent);
-		set_bit(XPT_DEFERRED, &svsk->sk_xprt.xpt_flags);
+		set_bit(XPT_DEFERRED, &xprt->xpt_flags);
 	}
-	spin_unlock(&svsk->sk_xprt.xpt_lock);
+	spin_unlock(&xprt->xpt_lock);
 	return dr;
 }
-

[PATCH 26/38] svc: Move the sockaddr information to svc_xprt

[Tom Tucker]

This patch moves the transport sockaddr to the svc_xprt
structure.  Convenience functions are added to set and
get the local and remote addresses of a transport from
the transport provider as well as determine the length
of a sockaddr.

A transport is responsible for setting the xpt_local
and xpt_remote addresses in the svc_xprt structure as
part of transport creation and xpo_accept processing. This
cannot be done in a generic way and in fact varies
between TCP, UDP and RDMA. A set of xpo_ functions
(e.g. getlocalname, getremotename) could have been
added but this would have resulted in additional
caching and copying of the addresses around.  Note that
the xpt_local address should also be set on listening
endpoints; for TCP/RDMA this is done as part of
endpoint creation.

For connected transports like TCP and RDMA, the addresses
never change and can be set once and copied into the
rqstp structure for each request. For UDP, however, the
local and remote addresses may change for each request. In
this case, the address information is obtained from the
UDP recvmsg info and copied into the rqstp structure from
there. 

A svc_xprt_local_port function was also added that returns
the local port given a transport. This is used by
svc_create_xprt when returning the port associated with
a newly created transport, and later when creating a
generic find transport service to check if a service is
already listening on a given port.

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 include/linux/sunrpc/svc_xprt.h |   51 ++++++++++++++++++++++++++++++++++++
 include/linux/sunrpc/svcsock.h  |    4 ---
 net/sunrpc/svc_xprt.c           |   31 ++++++++++++++++++++--
 net/sunrpc/svcsock.c            |   56 +++++++++++++++++++++------------------
 4 files changed, 110 insertions(+), 32 deletions(-)

diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
index d93ae27..a1d324a 100644
--- a/include/linux/sunrpc/svc_xprt.h
+++ b/include/linux/sunrpc/svc_xprt.h
@@ -61,6 +61,10 @@ struct svc_xprt {
 	void			*xpt_auth_cache;/* auth cache */
 	struct list_head	xpt_deferred;	/* deferred requests that need
 						 * to be revisted */
+	struct sockaddr_storage	xpt_local;	/* local address */
+	size_t			xpt_locallen;	/* length of address */
+	struct sockaddr_storage	xpt_remote;	/* remote peer's address */
+	size_t			xpt_remotelen;	/* length of address */
 };
 
 int	svc_reg_xprt_class(struct svc_xprt_class *);
@@ -70,9 +74,56 @@ void	svc_xprt_init(struct svc_xprt_class *, struct svc_xprt *,
 int	svc_create_xprt(struct svc_serv *, char *, unsigned short, int);
 void	svc_xprt_received(struct svc_xprt *);
 void	svc_xprt_put(struct svc_xprt *xprt);
+void	svc_xprt_copy_addrs(struct svc_rqst *rqstp, struct svc_xprt *xprt);
 static inline void svc_xprt_get(struct svc_xprt *xprt)
 {
 	kref_get(&xprt->xpt_ref);
 }
+static inline void svc_xprt_set_local(struct svc_xprt *xprt,
+				      struct sockaddr *sa, int salen)
+{
+	memcpy(&xprt->xpt_local, sa, salen);
+	xprt->xpt_locallen = salen;
+}
+static inline void svc_xprt_set_remote(struct svc_xprt *xprt,
+				       struct sockaddr *sa, int salen)
+{
+	memcpy(&xprt->xpt_remote, sa, salen);
+	xprt->xpt_remotelen = salen;
+}
+static inline unsigned short svc_addr_port(struct sockaddr *sa)
+{
+	unsigned short ret = 0;
+	switch (sa->sa_family) {
+	case AF_INET:
+		ret = ntohs(((struct sockaddr_in *)sa)->sin_port);
+		break;
+	case AF_INET6:
+		ret = ntohs(((struct sockaddr_in6 *)sa)->sin6_port);
+		break;
+	}
+	return ret;
+}
+
+static inline size_t svc_addr_len(struct sockaddr *sa)
+{
+	switch (sa->sa_family) {
+	case AF_INET:
+		return sizeof(struct sockaddr_in);
+	case AF_INET6:
+		return sizeof(struct sockaddr_in6);
+	}
+	return -EAFNOSUPPORT;
+}
+
+static inline unsigned short svc_xprt_local_port(struct svc_xprt *xprt)
+{
+	return svc_addr_port((struct sockaddr *)&xprt->xpt_local);
+}
+
+static inline unsigned short svc_xprt_remote_port(struct svc_xprt *xprt)
+{
+	return svc_addr_port((struct sockaddr *)&xprt->xpt_remote);
+}
 
 #endif /* SUNRPC_SVC_XPRT_H */
diff --git a/include/linux/sunrpc/svcsock.h b/include/linux/sunrpc/svcsock.h
index 96a229e..206f092 100644
--- a/include/linux/sunrpc/svcsock.h
+++ b/include/linux/sunrpc/svcsock.h
@@ -28,10 +28,6 @@ struct svc_sock {
 	/* private TCP part */
 	int			sk_reclen;	/* length of record */
 	int			sk_tcplen;	/* current read length */
-
-	struct sockaddr_storage	sk_local;	/* local address */
-	struct sockaddr_storage	sk_remote;	/* remote peer's address */
-	int			sk_remotelen;	/* length of address */
 };
 
 /*
diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
index fdf0d8c..d0cbfe0 100644
--- a/net/sunrpc/svc_xprt.c
+++ b/net/sunrpc/svc_xprt.c
@@ -138,7 +138,6 @@ int svc_create_xprt(struct svc_serv *serv, char *xprt_name, unsigned short port,
 			spin_unlock(&svc_xprt_class_lock);
 			if (try_module_get(xcl->xcl_owner)) {
 				struct svc_xprt *newxprt;
-				ret = 0;
 				newxprt = xcl->xcl_ops->xpo_create
 					(serv,
 					 (struct sockaddr *)&sin, sizeof(sin),
@@ -146,7 +145,8 @@ int svc_create_xprt(struct svc_serv *serv, char *xprt_name, unsigned short port,
 				if (IS_ERR(newxprt)) {
 					module_put(xcl->xcl_owner);
 					ret = PTR_ERR(newxprt);
-				}
+				} else
+					ret = svc_xprt_local_port(newxprt);
 			}
 			goto out;
 		}
@@ -157,3 +157,30 @@ int svc_create_xprt(struct svc_serv *serv, char *xprt_name, unsigned short port,
 	return ret;
 }
 EXPORT_SYMBOL_GPL(svc_create_xprt);
+
+/*
+ * Copy the local and remote xprt addresses to the rqstp structure
+ */
+void svc_xprt_copy_addrs(struct svc_rqst *rqstp, struct svc_xprt *xprt)
+{
+	struct sockaddr *sin;
+
+	memcpy(&rqstp->rq_addr, &xprt->xpt_remote, xprt->xpt_remotelen);
+	rqstp->rq_addrlen = xprt->xpt_remotelen;
+
+	/*
+	 * Destination address in request is needed for binding the
+	 * source address in RPC replies/callbacks later.
+	 */
+	sin = (struct sockaddr *)&xprt->xpt_local;
+	switch (sin->sa_family) {
+	case AF_INET:
+		rqstp->rq_daddr.addr = ((struct sockaddr_in *)sin)->sin_addr;
+		break;
+	case AF_INET6:
+		rqstp->rq_daddr.addr6 = ((struct sockaddr_in6 *)sin)->sin6_addr;
+		break;
+	}
+}
+EXPORT_SYMBOL_GPL(svc_xprt_copy_addrs);
+
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index 887729f..8828c74 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -620,33 +620,13 @@ svc_recvfrom(struct svc_rqst *rqstp, struct kvec *iov, int nr, int buflen)
 	struct msghdr msg = {
 		.msg_flags	= MSG_DONTWAIT,
 	};
-	struct sockaddr *sin;
 	int len;
 
 	len = kernel_recvmsg(svsk->sk_sock, &msg, iov, nr, buflen,
 				msg.msg_flags);
 
-	/* sock_recvmsg doesn't fill in the name/namelen, so we must..
-	 */
-	memcpy(&rqstp->rq_addr, &svsk->sk_remote, svsk->sk_remotelen);
-	rqstp->rq_addrlen = svsk->sk_remotelen;
-
-	/* Destination address in request is needed for binding the
-	 * source address in RPC callbacks later.
-	 */
-	sin = (struct sockaddr *)&svsk->sk_local;
-	switch (sin->sa_family) {
-	case AF_INET:
-		rqstp->rq_daddr.addr = ((struct sockaddr_in *)sin)->sin_addr;
-		break;
-	case AF_INET6:
-		rqstp->rq_daddr.addr6 = ((struct sockaddr_in6 *)sin)->sin6_addr;
-		break;
-	}
-
 	dprintk("svc: socket %p recvfrom(%p, %Zu) = %d\n",
 		svsk, iov[0].iov_base, iov[0].iov_len, len);
-
 	return len;
 }
 
@@ -716,8 +696,15 @@ svc_write_space(struct sock *sk)
 	}
 }
 
-static inline void svc_udp_get_dest_address(struct svc_rqst *rqstp,
-					    struct cmsghdr *cmh)
+/*
+ * Copy the UDP datagram's destination address to the rqstp structure.
+ * The 'destination' address in this case is the address to which the
+ * peer sent the datagram, i.e. our local address. For multihomed
+ * hosts, this can change from msg to msg. Note that only the IP
+ * address changes, the port number should remain the same.
+ */
+static void svc_udp_get_dest_address(struct svc_rqst *rqstp,
+				     struct cmsghdr *cmh)
 {
 	switch (rqstp->rq_sock->sk_sk->sk_family) {
 	case AF_INET: {
@@ -784,7 +771,10 @@ svc_udp_recvfrom(struct svc_rqst *rqstp)
 		svc_xprt_received(&svsk->sk_xprt);
 		return -EAGAIN;
 	}
-	rqstp->rq_addrlen = sizeof(rqstp->rq_addr);
+	len = svc_addr_len(svc_addr(rqstp));
+	if (len < 0)
+		return len;
+	rqstp->rq_addrlen = len;
 	if (skb->tstamp.tv64 == 0) {
 		skb->tstamp = ktime_get_real();
 		/* Don't enable netstamp, sunrpc doesn't
@@ -1094,14 +1084,13 @@ static struct svc_xprt *svc_tcp_accept(struct svc_xprt *xprt)
 	if (!(newsvsk = svc_setup_socket(serv, newsock, &err,
 				 (SVC_SOCK_ANONYMOUS | SVC_SOCK_TEMPORARY))))
 		goto failed;
-	memcpy(&newsvsk->sk_remote, sin, slen);
-	newsvsk->sk_remotelen = slen;
+	svc_xprt_set_remote(&newsvsk->sk_xprt, sin, slen);
 	err = kernel_getsockname(newsock, sin, &slen);
 	if (unlikely(err < 0)) {
 		dprintk("svc_tcp_accept: kernel_getsockname error %d\n", -err);
 		slen = offsetof(struct sockaddr, sa_data);
 	}
-	memcpy(&newsvsk->sk_local, sin, slen);
+	svc_xprt_set_local(&newsvsk->sk_xprt, sin, slen);
 
 	if (serv->sv_stats)
 		serv->sv_stats->nettcpconn++;
@@ -1242,6 +1231,7 @@ svc_tcp_recvfrom(struct svc_rqst *rqstp)
 	svsk->sk_reclen = 0;
 	svsk->sk_tcplen = 0;
 
+	svc_xprt_copy_addrs(rqstp, &svsk->sk_xprt);
 	svc_xprt_received(&svsk->sk_xprt);
 	if (serv->sv_stats)
 		serv->sv_stats->nettcpcnt++;
@@ -1807,6 +1797,11 @@ int svc_addsock(struct svc_serv *serv,
 	else {
 		svsk = svc_setup_socket(serv, so, &err, SVC_SOCK_DEFAULTS);
 		if (svsk) {
+			struct sockaddr_storage addr;
+			struct sockaddr *sin = (struct sockaddr *)&addr;
+			int salen;
+			if (kernel_getsockname(svsk->sk_sock, sin, &salen) == 0)
+				svc_xprt_set_local(&svsk->sk_xprt, sin, salen);
 			svc_xprt_received(&svsk->sk_xprt);
 			err = 0;
 		}
@@ -1833,6 +1828,9 @@ static struct svc_xprt *svc_create_socket(struct svc_serv *serv,
 	int		error;
 	int		type;
 	char		buf[RPC_MAX_ADDRBUFLEN];
+	struct sockaddr_storage addr;
+	struct sockaddr *newsin = (struct sockaddr *)&addr;
+	int		newlen;
 
 	dprintk("svc: svc_create_socket(%s, %d, %s)\n",
 			serv->sv_program->pg_name, protocol,
@@ -1857,12 +1855,18 @@ static struct svc_xprt *svc_create_socket(struct svc_serv *serv,
 	if (error < 0)
 		goto bummer;
 
+	newlen = len;
+	error = kernel_getsockname(sock, newsin, &newlen);
+	if (error < 0)
+		goto bummer;
+
 	if (protocol == IPPROTO_TCP) {
 		if ((error = kernel_listen(sock, 64)) < 0)
 			goto bummer;
 	}
 
 	if ((svsk = svc_setup_socket(serv, sock, &error, flags)) != NULL) {
+		svc_xprt_set_local(&svsk->sk_xprt, newsin, newlen);
 		svc_xprt_received(&svsk->sk_xprt);
 		return (struct svc_xprt *)svsk;
 	}

[PATCH 27/38] svc: Make svc_sock_release svc_xprt_release

[Tom Tucker]

The svc_sock_release function only touches transport independent fields.
Change the function to manipulate svc_xprt directly instead of the transport
dependent svc_sock structure. 

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 net/sunrpc/svcsock.c |   16 +++++++---------
 1 files changed, 7 insertions(+), 9 deletions(-)

diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index 8828c74..378c970 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -365,10 +365,9 @@ void svc_reserve(struct svc_rqst *rqstp, int space)
 	}
 }
 
-static void
-svc_sock_release(struct svc_rqst *rqstp)
+static void svc_xprt_release(struct svc_rqst *rqstp)
 {
-	struct svc_sock	*svsk = rqstp->rq_sock;
+	struct svc_xprt	*xprt = rqstp->rq_xprt;
 
 	rqstp->rq_xprt->xpt_ops->xpo_release_rqst(rqstp);
 
@@ -376,7 +375,6 @@ svc_sock_release(struct svc_rqst *rqstp)
 	rqstp->rq_res.page_len = 0;
 	rqstp->rq_res.page_base = 0;
 
-
 	/* Reset response buffer and release
 	 * the reservation.
 	 * But first, check that enough space was reserved
@@ -389,9 +387,9 @@ svc_sock_release(struct svc_rqst *rqstp)
 
 	rqstp->rq_res.head[0].iov_len = 0;
 	svc_reserve(rqstp, 0);
-	rqstp->rq_sock = NULL;
+	rqstp->rq_xprt = NULL;
 
-	svc_xprt_put(&svsk->sk_xprt);
+	svc_xprt_put(xprt);
 }
 
 /*
@@ -1595,7 +1593,7 @@ svc_recv(struct svc_rqst *rqstp, long timeout)
 	/* No data, incomplete (TCP) read, or accept() */
 	if (len == 0 || len == -EAGAIN) {
 		rqstp->rq_res.len = 0;
-		svc_sock_release(rqstp);
+		svc_xprt_release(rqstp);
 		return -EAGAIN;
 	}
 	clear_bit(XPT_OLD, &svsk->sk_xprt.xpt_flags);
@@ -1615,7 +1613,7 @@ void
 svc_drop(struct svc_rqst *rqstp)
 {
 	dprintk("svc: socket %p dropped request\n", rqstp->rq_sock);
-	svc_sock_release(rqstp);
+	svc_xprt_release(rqstp);
 }
 
 /*
@@ -1648,7 +1646,7 @@ svc_send(struct svc_rqst *rqstp)
 	else
 		len = xprt->xpt_ops->xpo_sendto(rqstp);
 	mutex_unlock(&xprt->xpt_mutex);
-	svc_sock_release(rqstp);
+	svc_xprt_release(rqstp);
 
 	if (len == -ECONNREFUSED || len == -ENOTCONN || len == -EAGAIN)
 		return 0;

[PATCH 28/38] svc: Make svc_recv transport neutral

[Tom Tucker]

All of the transport field and functions used by svc_recv are now
transport independent. Change the svc_recv function to use the svc_xprt
structure directly instead of the transport specific svc_sock structure.

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 net/sunrpc/svcsock.c |   61 +++++++++++++++++++++++++-------------------------
 1 files changed, 31 insertions(+), 30 deletions(-)

diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index 378c970..a6793b5 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -310,22 +310,21 @@ EXPORT_SYMBOL_GPL(svc_xprt_enqueue);
 /*
  * Dequeue the first socket.  Must be called with the pool->sp_lock held.
  */
-static inline struct svc_sock *
-svc_sock_dequeue(struct svc_pool *pool)
+static struct svc_xprt *svc_xprt_dequeue(struct svc_pool *pool)
 {
-	struct svc_sock	*svsk;
+	struct svc_xprt	*xprt;
 
 	if (list_empty(&pool->sp_sockets))
 		return NULL;
 
-	svsk = list_entry(pool->sp_sockets.next,
-			  struct svc_sock, sk_xprt.xpt_ready);
-	list_del_init(&svsk->sk_xprt.xpt_ready);
+	xprt = list_entry(pool->sp_sockets.next,
+			  struct svc_xprt, xpt_ready);
+	list_del_init(&xprt->xpt_ready);
 
-	dprintk("svc: socket %p dequeued, inuse=%d\n",
-		svsk->sk_sk, atomic_read(&svsk->sk_xprt.xpt_ref.refcount));
+	dprintk("svc: transport %p dequeued, inuse=%d\n",
+		xprt, atomic_read(&xprt->xpt_ref.refcount));
 
-	return svsk;
+	return xprt;
 }
 
 /*
@@ -1477,20 +1476,20 @@ static void svc_check_conn_limits(struct svc_serv *serv)
 int
 svc_recv(struct svc_rqst *rqstp, long timeout)
 {
-	struct svc_sock		*svsk = NULL;
+	struct svc_xprt		*xprt = NULL;
 	struct svc_serv		*serv = rqstp->rq_server;
 	struct svc_pool		*pool = rqstp->rq_pool;
 	int			len, i;
-	int 			pages;
+	int			pages;
 	struct xdr_buf		*arg;
 	DECLARE_WAITQUEUE(wait, current);
 
 	dprintk("svc: server %p waiting for data (to = %ld)\n",
 		rqstp, timeout);
 
-	if (rqstp->rq_sock)
+	if (rqstp->rq_xprt)
 		printk(KERN_ERR
-			"svc_recv: service %p, socket not NULL!\n",
+			"svc_recv: service %p, transport not NULL!\n",
 			 rqstp);
 	if (waitqueue_active(&rqstp->rq_wait))
 		printk(KERN_ERR
@@ -1527,11 +1526,12 @@ svc_recv(struct svc_rqst *rqstp, long timeout)
 		return -EINTR;
 
 	spin_lock_bh(&pool->sp_lock);
-	if ((svsk = svc_sock_dequeue(pool)) != NULL) {
-		rqstp->rq_sock = svsk;
-		svc_xprt_get(&svsk->sk_xprt);
+	xprt = svc_xprt_dequeue(pool);
+	if (xprt) {
+		rqstp->rq_xprt = xprt;
+		svc_xprt_get(xprt);
 		rqstp->rq_reserved = serv->sv_max_mesg;
-		atomic_add(rqstp->rq_reserved, &svsk->sk_xprt.xpt_reserved);
+		atomic_add(rqstp->rq_reserved, &xprt->xpt_reserved);
 	} else {
 		/* No data pending. Go to sleep */
 		svc_thread_enqueue(pool, rqstp);
@@ -1551,7 +1551,8 @@ svc_recv(struct svc_rqst *rqstp, long timeout)
 		spin_lock_bh(&pool->sp_lock);
 		remove_wait_queue(&rqstp->rq_wait, &wait);
 
-		if (!(svsk = rqstp->rq_sock)) {
+		xprt = rqstp->rq_xprt;
+		if (!xprt) {
 			svc_thread_dequeue(pool, rqstp);
 			spin_unlock_bh(&pool->sp_lock);
 			dprintk("svc: server %p, no data yet\n", rqstp);
@@ -1561,12 +1562,12 @@ svc_recv(struct svc_rqst *rqstp, long timeout)
 	spin_unlock_bh(&pool->sp_lock);
 
 	len = 0;
-	if (test_bit(XPT_CLOSE, &svsk->sk_xprt.xpt_flags)) {
+	if (test_bit(XPT_CLOSE, &xprt->xpt_flags)) {
 		dprintk("svc_recv: found XPT_CLOSE\n");
-		svc_delete_xprt(&svsk->sk_xprt);
-	} else if (test_bit(XPT_LISTENER, &svsk->sk_xprt.xpt_flags)) {
+		svc_delete_xprt(xprt);
+	} else if (test_bit(XPT_LISTENER, &xprt->xpt_flags)) {
 		struct svc_xprt *newxpt;
-		newxpt = svsk->sk_xprt.xpt_ops->xpo_accept(&svsk->sk_xprt);
+		newxpt = xprt->xpt_ops->xpo_accept(xprt);
 		if (newxpt) {
 			/*
 			 * We know this module_get will succeed because the
@@ -1576,17 +1577,17 @@ svc_recv(struct svc_rqst *rqstp, long timeout)
 			svc_check_conn_limits(svsk->sk_xprt.xpt_server);
 			svc_xprt_received(newxpt);
 		}
-		svc_xprt_received(&svsk->sk_xprt);
+		svc_xprt_received(xprt);
 	} else {
-		dprintk("svc: server %p, pool %u, socket %p, inuse=%d\n",
-			rqstp, pool->sp_id, svsk,
-			atomic_read(&svsk->sk_xprt.xpt_ref.refcount));
-		rqstp->rq_deferred = svc_deferred_dequeue(&svsk->sk_xprt);
+		dprintk("svc: server %p, pool %u, transport %p, inuse=%d\n",
+			rqstp, pool->sp_id, xprt,
+			atomic_read(&xprt->xpt_ref.refcount));
+		rqstp->rq_deferred = svc_deferred_dequeue(xprt);
 		if (rqstp->rq_deferred) {
-			svc_xprt_received(&svsk->sk_xprt);
+			svc_xprt_received(xprt);
 			len = svc_deferred_recv(rqstp);
 		} else
-			len = svsk->sk_xprt.xpt_ops->xpo_recvfrom(rqstp);
+			len = xprt->xpt_ops->xpo_recvfrom(rqstp);
 		dprintk("svc: got len=%d\n", len);
 	}
 
@@ -1596,7 +1597,7 @@ svc_recv(struct svc_rqst *rqstp, long timeout)
 		svc_xprt_release(rqstp);
 		return -EAGAIN;
 	}
-	clear_bit(XPT_OLD, &svsk->sk_xprt.xpt_flags);
+	clear_bit(XPT_OLD, &xprt->xpt_flags);
 
 	rqstp->rq_secure = svc_port_is_privileged(svc_addr(rqstp));
 	rqstp->rq_chandle.defer = svc_defer;

[PATCH 29/38] svc: Make svc_age_temp_sockets svc_age_temp_transports

[Tom Tucker]

This function is transport independent. Change it to use svc_xprt directly
and change it's name to reflect this.

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 net/sunrpc/svcsock.c |   37 +++++++++++++++++++------------------
 1 files changed, 19 insertions(+), 18 deletions(-)

diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index a6793b5..27cf09f 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -1658,49 +1658,50 @@ svc_send(struct svc_rqst *rqstp)
  * Timer function to close old temporary sockets, using
  * a mark-and-sweep algorithm.
  */
-static void
-svc_age_temp_sockets(unsigned long closure)
+static void svc_age_temp_xprts(unsigned long closure)
 {
 	struct svc_serv *serv = (struct svc_serv *)closure;
-	struct svc_sock *svsk;
+	struct svc_xprt *xprt;
 	struct list_head *le, *next;
 	LIST_HEAD(to_be_aged);
 
-	dprintk("svc_age_temp_sockets\n");
+	dprintk("svc_age_temp_xprts\n");
 
 	if (!spin_trylock_bh(&serv->sv_lock)) {
 		/* busy, try again 1 sec later */
-		dprintk("svc_age_temp_sockets: busy\n");
+		dprintk("svc_age_temp_xprts: busy\n");
 		mod_timer(&serv->sv_temptimer, jiffies + HZ);
 		return;
 	}
 
 	list_for_each_safe(le, next, &serv->sv_tempsocks) {
-		svsk = list_entry(le, struct svc_sock, sk_xprt.xpt_list);
+		xprt = list_entry(le, struct svc_xprt, xpt_list);
 
-		if (!test_and_set_bit(XPT_OLD, &svsk->sk_xprt.xpt_flags))
+		/* First time through, just mark it OLD. Second time
+		 * through, close it. */
+		if (!test_and_set_bit(XPT_OLD, &xprt->xpt_flags))
 			continue;
-		if (atomic_read(&svsk->sk_xprt.xpt_ref.refcount) > 1
-		    || test_bit(XPT_BUSY, &svsk->sk_xprt.xpt_flags))
+		if (atomic_read(&xprt->xpt_ref.refcount) > 1
+		    || test_bit(XPT_BUSY, &xprt->xpt_flags))
 			continue;
-		svc_xprt_get(&svsk->sk_xprt);
+		svc_xprt_get(xprt);
 		list_move(le, &to_be_aged);
-		set_bit(XPT_CLOSE, &svsk->sk_xprt.xpt_flags);
-		set_bit(XPT_DETACHED, &svsk->sk_xprt.xpt_flags);
+		set_bit(XPT_CLOSE, &xprt->xpt_flags);
+		set_bit(XPT_DETACHED, &xprt->xpt_flags);
 	}
 	spin_unlock_bh(&serv->sv_lock);
 
 	while (!list_empty(&to_be_aged)) {
 		le = to_be_aged.next;
-		/* fiddling the sk_xprt.xpt_list node is safe 'cos we're XPT_DETACHED */
+		/* fiddling the xpt_list node is safe 'cos we're XPT_DETACHED */
 		list_del_init(le);
-		svsk = list_entry(le, struct svc_sock, sk_xprt.xpt_list);
+		xprt = list_entry(le, struct svc_xprt, xpt_list);
 
-		dprintk("queuing svsk %p for closing\n", svsk);
+		dprintk("queuing xprt %p for closing\n", xprt);
 
 		/* a thread will dequeue and close it soon */
-		svc_xprt_enqueue(&svsk->sk_xprt);
-		svc_xprt_put(&svsk->sk_xprt);
+		svc_xprt_enqueue(xprt);
+		svc_xprt_put(xprt);
 	}
 
 	mod_timer(&serv->sv_temptimer, jiffies + svc_conn_age_period * HZ);
@@ -1758,7 +1759,7 @@ static struct svc_sock *svc_setup_socket(struct svc_serv *serv,
 		serv->sv_tmpcnt++;
 		if (serv->sv_temptimer.function == NULL) {
 			/* setup timer to age temp sockets */
-			setup_timer(&serv->sv_temptimer, svc_age_temp_sockets,
+			setup_timer(&serv->sv_temptimer, svc_age_temp_xprts,
 					(unsigned long)serv);
 			mod_timer(&serv->sv_temptimer,
 					jiffies + svc_conn_age_period * HZ);

[PATCH 30/38] svc: Move common create logic to common code

[Tom Tucker]

Move the code that adds a transport instance to the sv_tempsocks and
sv_permsocks lists out of the transport specific functions and into core
logic. 

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 net/sunrpc/svc_xprt.c |    9 ++++++++-
 net/sunrpc/svcsock.c  |   39 +++++++++++++++++++--------------------
 2 files changed, 27 insertions(+), 21 deletions(-)

diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
index d0cbfe0..924df63 100644
--- a/net/sunrpc/svc_xprt.c
+++ b/net/sunrpc/svc_xprt.c
@@ -145,8 +145,15 @@ int svc_create_xprt(struct svc_serv *serv, char *xprt_name, unsigned short port,
 				if (IS_ERR(newxprt)) {
 					module_put(xcl->xcl_owner);
 					ret = PTR_ERR(newxprt);
-				} else
+				} else {
+					clear_bit(XPT_TEMP,
+						  &newxprt->xpt_flags);
+					spin_lock_bh(&serv->sv_lock);
+					list_add(&newxprt->xpt_list,
+						 &serv->sv_permsocks);
+					spin_unlock_bh(&serv->sv_lock);
 					ret = svc_xprt_local_port(newxprt);
+				}
 			}
 			goto out;
 		}
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index 27cf09f..f284bdf 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -94,6 +94,7 @@ static int svc_deferred_recv(struct svc_rqst *rqstp);
 static struct cache_deferred_req *svc_defer(struct cache_req *req);
 static struct svc_xprt *svc_create_socket(struct svc_serv *, int,
 					  struct sockaddr *, int, int);
+static void svc_age_temp_xprts(unsigned long closure);
 
 /* apparently the "standard" is that clients close
  * idle connections after 5 minutes, servers after
@@ -1574,7 +1575,20 @@ svc_recv(struct svc_rqst *rqstp, long timeout)
 			 * listener holds a reference too
 			 */
 			__module_get(newxpt->xpt_class->xcl_owner);
-			svc_check_conn_limits(svsk->sk_xprt.xpt_server);
+			svc_check_conn_limits(xprt->xpt_server);
+			spin_lock_bh(&serv->sv_lock);
+			set_bit(XPT_TEMP, &newxpt->xpt_flags);
+			list_add(&newxpt->xpt_list, &serv->sv_tempsocks);
+			serv->sv_tmpcnt++;
+			if (serv->sv_temptimer.function == NULL) {
+				/* setup timer to age temp sockets */
+				setup_timer(&serv->sv_temptimer,
+					    svc_age_temp_xprts,
+					    (unsigned long)serv);
+				mod_timer(&serv->sv_temptimer,
+					  jiffies + svc_conn_age_period * HZ);
+			}
+			spin_unlock_bh(&serv->sv_lock);
 			svc_xprt_received(newxpt);
 		}
 		svc_xprt_received(xprt);
@@ -1718,7 +1732,6 @@ static struct svc_sock *svc_setup_socket(struct svc_serv *serv,
 	struct svc_sock	*svsk;
 	struct sock	*inet;
 	int		pmap_register = !(flags & SVC_SOCK_ANONYMOUS);
-	int		is_temporary = flags & SVC_SOCK_TEMPORARY;
 
 	dprintk("svc: svc_setup_socket %p\n", sock);
 	if (!(svsk = kzalloc(sizeof(*svsk), GFP_KERNEL))) {
@@ -1752,24 +1765,6 @@ static struct svc_sock *svc_setup_socket(struct svc_serv *serv,
 	else
 		svc_tcp_init(svsk, serv);
 
-	spin_lock_bh(&serv->sv_lock);
-	if (is_temporary) {
-		set_bit(XPT_TEMP, &svsk->sk_xprt.xpt_flags);
-		list_add(&svsk->sk_xprt.xpt_list, &serv->sv_tempsocks);
-		serv->sv_tmpcnt++;
-		if (serv->sv_temptimer.function == NULL) {
-			/* setup timer to age temp sockets */
-			setup_timer(&serv->sv_temptimer, svc_age_temp_xprts,
-					(unsigned long)serv);
-			mod_timer(&serv->sv_temptimer,
-					jiffies + svc_conn_age_period * HZ);
-		}
-	} else {
-		clear_bit(XPT_TEMP, &svsk->sk_xprt.xpt_flags);
-		list_add(&svsk->sk_xprt.xpt_list, &serv->sv_permsocks);
-	}
-	spin_unlock_bh(&serv->sv_lock);
-
 	dprintk("svc: svc_setup_socket created %p (inet %p)\n",
 				svsk, svsk->sk_sk);
 
@@ -1805,6 +1800,10 @@ int svc_addsock(struct svc_serv *serv,
 			svc_xprt_received(&svsk->sk_xprt);
 			err = 0;
 		}
+		clear_bit(XPT_TEMP, &svsk->sk_xprt.xpt_flags);
+		spin_lock_bh(&serv->sv_lock);
+		list_add(&svsk->sk_xprt.xpt_list, &serv->sv_permsocks);
+		spin_unlock_bh(&serv->sv_lock);
 	}
 	if (err) {
 		sockfd_put(so);

Re: [PATCH 30/38] svc: Move common create logic to common code

[J. Bruce Fields]

On Tue, Dec 11, 2007 at 05:33:00PM -0600, Tom Tucker wrote:
> 
> Move the code that adds a transport instance to the sv_tempsocks and
> sv_permsocks lists out of the transport specific functions and into core
> logic. 
> 
> Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
> ---
> 
>  net/sunrpc/svc_xprt.c |    9 ++++++++-
>  net/sunrpc/svcsock.c  |   39 +++++++++++++++++++--------------------
>  2 files changed, 27 insertions(+), 21 deletions(-)
> 
> diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
> index d0cbfe0..924df63 100644
> --- a/net/sunrpc/svc_xprt.c
> +++ b/net/sunrpc/svc_xprt.c
> @@ -145,8 +145,15 @@ int svc_create_xprt(struct svc_serv *serv, char *xprt_name, unsigned short port,
>  				if (IS_ERR(newxprt)) {
>  					module_put(xcl->xcl_owner);
>  					ret = PTR_ERR(newxprt);
> -				} else
> +				} else {
> +					clear_bit(XPT_TEMP,
> +						  &newxprt->xpt_flags);
> +					spin_lock_bh(&serv->sv_lock);
> +					list_add(&newxprt->xpt_list,
> +						 &serv->sv_permsocks);
> +					spin_unlock_bh(&serv->sv_lock);
>  					ret = svc_xprt_local_port(newxprt);
> +				}
>  			}
>  			goto out;
>  		}

The nesting here is getting kind of deep.

A bit more idiomatic for kernel code would be (untested) something like:

	list_for_each_entry(xcl, &svc_xprt_class_list, xcl_list) {
		struct svc_xprt *newxprt;

		if (strcmp(xprt_name, xcl->xcl_name) != 0)
			continue;
		spin_unlock(&svc_xprt_class_lock);
		if (!try_module_get(xcl->xcl_owner))
			goto out;
		# (assuming that's the right order)
		newxprt = xcl->xcl_ops->xpo_create (serv,
				 (struct sockaddr *)&sin, sizeof(sin), flags);
		if (IS_ERR(newxprt)) {
			module_put(xcl->xcl_owner);
			ret = PTR_ERR(newxprt);
			goto out;
		}
		clear_bit(XPT_TEMP, &newxprt->xpt_flags);
		spin_lock_bh(&serv->sv_lock);
		list_add(&newxprt->xpt_list, &serv->sv_permsocks);
		spin_unlock_bh(&serv->sv_lock);
		ret = svc_xprt_local_port(newxprt);
		goto out;
	}

At least to my eyes, that also makes it a lot easier to see what the
important (succesful) case is, since that's the code that stays
unindented all the way to the end.

I'd also be inclined to replace those "got out"'s by "returns".  I don't
really see the point of the label when it's leads to something that's
literally just a return.

--b.

Re: [PATCH 30/38] svc: Move common create logic to common code

[Tom Tucker]

On 12/14/07 5:19 PM, "J. Bruce Fields" <bfields@fieldses.org> wrote:

> On Tue, Dec 11, 2007 at 05:33:00PM -0600, Tom Tucker wrote:
>> 
>> Move the code that adds a transport instance to the sv_tempsocks and
>> sv_permsocks lists out of the transport specific functions and into core
>> logic. 
>> 
>> Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
>> ---
>> 
>>  net/sunrpc/svc_xprt.c |    9 ++++++++-
>>  net/sunrpc/svcsock.c  |   39 +++++++++++++++++++--------------------
>>  2 files changed, 27 insertions(+), 21 deletions(-)
>> 
>> diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
>> index d0cbfe0..924df63 100644
>> --- a/net/sunrpc/svc_xprt.c
>> +++ b/net/sunrpc/svc_xprt.c
>> @@ -145,8 +145,15 @@ int svc_create_xprt(struct svc_serv *serv, char
>> *xprt_name, unsigned short port,
>> if (IS_ERR(newxprt)) {
>> module_put(xcl->xcl_owner);
>> ret = PTR_ERR(newxprt);
>> -    } else
>> +    } else {
>> +     clear_bit(XPT_TEMP,
>> +        &newxprt->xpt_flags);
>> +     spin_lock_bh(&serv->sv_lock);
>> +     list_add(&newxprt->xpt_list,
>> +       &serv->sv_permsocks);
>> +     spin_unlock_bh(&serv->sv_lock);
>> ret = svc_xprt_local_port(newxprt);
>> +    }
>> }
>> goto out;
>> }
> 
> The nesting here is getting kind of deep.
> 
> A bit more idiomatic for kernel code would be (untested) something like:
>

I agree. I'll look at refactoring this.
 
 
> list_for_each_entry(xcl, &svc_xprt_class_list, xcl_list) {
> struct svc_xprt *newxprt;
> 
> if (strcmp(xprt_name, xcl->xcl_name) != 0)
> continue;
> spin_unlock(&svc_xprt_class_lock);
> if (!try_module_get(xcl->xcl_owner))
> goto out;
> # (assuming that's the right order)
> newxprt = xcl->xcl_ops->xpo_create (serv,
> (struct sockaddr *)&sin, sizeof(sin), flags);
> if (IS_ERR(newxprt)) {
> module_put(xcl->xcl_owner);
> ret = PTR_ERR(newxprt);
> goto out;
> }
> clear_bit(XPT_TEMP, &newxprt->xpt_flags);
> spin_lock_bh(&serv->sv_lock);
> list_add(&newxprt->xpt_list, &serv->sv_permsocks);
> spin_unlock_bh(&serv->sv_lock);
> ret = svc_xprt_local_port(newxprt);
> goto out;
> }
> 
> At least to my eyes, that also makes it a lot easier to see what the
> important (succesful) case is, since that's the code that stays
> unindented all the way to the end.
> 
> I'd also be inclined to replace those "got out"'s by "returns".  I don't
> really see the point of the label when it's leads to something that's
> literally just a return.
> 
> --b.
> -
> To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 30/38] svc: Move common create logic to common code

[Tom Tucker]

On 12/14/07 5:19 PM, "J. Bruce Fields" <bfields@fieldses.org> wrote:

> On Tue, Dec 11, 2007 at 05:33:00PM -0600, Tom Tucker wrote:
>> 
>> Move the code that adds a transport instance to the sv_tempsocks and
>> sv_permsocks lists out of the transport specific functions and into core
>> logic. 
>> 
>> Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
>> ---
>> 
>>  net/sunrpc/svc_xprt.c |    9 ++++++++-
>>  net/sunrpc/svcsock.c  |   39 +++++++++++++++++++--------------------
>>  2 files changed, 27 insertions(+), 21 deletions(-)
>> 
>> diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
>> index d0cbfe0..924df63 100644
>> --- a/net/sunrpc/svc_xprt.c
>> +++ b/net/sunrpc/svc_xprt.c
>> @@ -145,8 +145,15 @@ int svc_create_xprt(struct svc_serv *serv, char
>> *xprt_name, unsigned short port,
>> if (IS_ERR(newxprt)) {
>> module_put(xcl->xcl_owner);
>> ret = PTR_ERR(newxprt);
>> -    } else
>> +    } else {
>> +     clear_bit(XPT_TEMP,
>> +        &newxprt->xpt_flags);
>> +     spin_lock_bh(&serv->sv_lock);
>> +     list_add(&newxprt->xpt_list,
>> +       &serv->sv_permsocks);
>> +     spin_unlock_bh(&serv->sv_lock);
>> ret = svc_xprt_local_port(newxprt);
>> +    }
>> }
>> goto out;
>> }
> 
> The nesting here is getting kind of deep.
> 
> A bit more idiomatic for kernel code would be (untested) something like:
> 
> list_for_each_entry(xcl, &svc_xprt_class_list, xcl_list) {
> struct svc_xprt *newxprt;
> 
> if (strcmp(xprt_name, xcl->xcl_name) != 0)
> continue;
> spin_unlock(&svc_xprt_class_lock);
> if (!try_module_get(xcl->xcl_owner))
> goto out;
> # (assuming that's the right order)
> newxprt = xcl->xcl_ops->xpo_create (serv,
> (struct sockaddr *)&sin, sizeof(sin), flags);
> if (IS_ERR(newxprt)) {
> module_put(xcl->xcl_owner);
> ret = PTR_ERR(newxprt);
> goto out;
> }
> clear_bit(XPT_TEMP, &newxprt->xpt_flags);
> spin_lock_bh(&serv->sv_lock);
> list_add(&newxprt->xpt_list, &serv->sv_permsocks);
> spin_unlock_bh(&serv->sv_lock);
> ret = svc_xprt_local_port(newxprt);
> goto out;
> }
> 
> At least to my eyes, that also makes it a lot easier to see what the
> important (succesful) case is, since that's the code that stays
> unindented all the way to the end.
> 
> I'd also be inclined to replace those "got out"'s by "returns".  I don't
> really see the point of the label when it's leads to something that's
> literally just a return.
> 

Agree. Fixed.

> --b.
> -
> To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH 31/38] svc: Removing remaining references to rq_sock in rqstp

[Tom Tucker]

This functionally empty patch removes rq_sock and unamed union 
from rqstp structure. 

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 include/linux/sunrpc/svc.h |    5 +----
 net/sunrpc/svcsock.c       |   38 ++++++++++++++++++++++++--------------
 2 files changed, 25 insertions(+), 18 deletions(-)

diff --git a/include/linux/sunrpc/svc.h b/include/linux/sunrpc/svc.h
index 40adc9d..04eb20e 100644
--- a/include/linux/sunrpc/svc.h
+++ b/include/linux/sunrpc/svc.h
@@ -204,10 +204,7 @@ union svc_addr_u {
 struct svc_rqst {
 	struct list_head	rq_list;	/* idle list */
 	struct list_head	rq_all;		/* all threads list */
-	union {
-		struct svc_xprt *	rq_xprt;	/* transport ptr */
-		struct svc_sock *	rq_sock; 	/* socket ptr */
-	};
+	struct svc_xprt *	rq_xprt;	/* transport ptr */
 	struct sockaddr_storage	rq_addr;	/* peer address */
 	size_t			rq_addrlen;
 
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index f284bdf..a785055 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -201,10 +201,12 @@ static void svc_release_skb(struct svc_rqst *rqstp)
 	struct svc_deferred_req *dr = rqstp->rq_deferred;
 
 	if (skb) {
+		struct svc_sock *svsk =
+			container_of(rqstp->rq_xprt, struct svc_sock, sk_xprt);
 		rqstp->rq_xprt_ctxt = NULL;
 
 		dprintk("svc: service %p, releasing skb %p\n", rqstp, skb);
-		skb_free_datagram(rqstp->rq_sock->sk_sk, skb);
+		skb_free_datagram(svsk->sk_sk, skb);
 	}
 	if (dr) {
 		rqstp->rq_deferred = NULL;
@@ -415,7 +417,7 @@ svc_wake_up(struct svc_serv *serv)
 			dprintk("svc: daemon %p woken up.\n", rqstp);
 			/*
 			svc_thread_dequeue(pool, rqstp);
-			rqstp->rq_sock = NULL;
+			rqstp->rq_xprt = NULL;
 			 */
 			wake_up(&rqstp->rq_wait);
 		}
@@ -432,7 +434,9 @@ union svc_pktinfo_u {
 
 static void svc_set_cmsg_data(struct svc_rqst *rqstp, struct cmsghdr *cmh)
 {
-	switch (rqstp->rq_sock->sk_sk->sk_family) {
+	struct svc_sock *svsk =
+		container_of(rqstp->rq_xprt, struct svc_sock, sk_xprt);
+	switch (svsk->sk_sk->sk_family) {
 	case AF_INET: {
 			struct in_pktinfo *pki = CMSG_DATA(cmh);
 
@@ -465,7 +469,8 @@ static void svc_set_cmsg_data(struct svc_rqst *rqstp, struct cmsghdr *cmh)
 static int
 svc_sendto(struct svc_rqst *rqstp, struct xdr_buf *xdr)
 {
-	struct svc_sock	*svsk = rqstp->rq_sock;
+	struct svc_sock	*svsk =
+		container_of(rqstp->rq_xprt, struct svc_sock, sk_xprt);
 	struct socket	*sock = svsk->sk_sock;
 	int		slen;
 	union {
@@ -538,7 +543,7 @@ svc_sendto(struct svc_rqst *rqstp, struct xdr_buf *xdr)
 	}
 out:
 	dprintk("svc: socket %p sendto([%p %Zu... ], %d) = %d (addr %s)\n",
-		rqstp->rq_sock, xdr->head[0].iov_base, xdr->head[0].iov_len,
+		svsk, xdr->head[0].iov_base, xdr->head[0].iov_len,
 		xdr->len, len, svc_print_addr(rqstp, buf, sizeof(buf)));
 
 	return len;
@@ -614,7 +619,8 @@ svc_recv_available(struct svc_sock *svsk)
 static int
 svc_recvfrom(struct svc_rqst *rqstp, struct kvec *iov, int nr, int buflen)
 {
-	struct svc_sock *svsk = rqstp->rq_sock;
+	struct svc_sock *svsk =
+		container_of(rqstp->rq_xprt, struct svc_sock, sk_xprt);
 	struct msghdr msg = {
 		.msg_flags	= MSG_DONTWAIT,
 	};
@@ -704,7 +710,9 @@ svc_write_space(struct sock *sk)
 static void svc_udp_get_dest_address(struct svc_rqst *rqstp,
 				     struct cmsghdr *cmh)
 {
-	switch (rqstp->rq_sock->sk_sk->sk_family) {
+	struct svc_sock *svsk =
+		container_of(rqstp->rq_xprt, struct svc_sock, sk_xprt);
+	switch (svsk->sk_sk->sk_family) {
 	case AF_INET: {
 		struct in_pktinfo *pki = CMSG_DATA(cmh);
 		rqstp->rq_daddr.addr.s_addr = pki->ipi_spec_dst.s_addr;
@@ -724,7 +732,8 @@ static void svc_udp_get_dest_address(struct svc_rqst *rqstp,
 static int
 svc_udp_recvfrom(struct svc_rqst *rqstp)
 {
-	struct svc_sock	*svsk = rqstp->rq_sock;
+	struct svc_sock	*svsk =
+		container_of(rqstp->rq_xprt, struct svc_sock, sk_xprt);
 	struct svc_serv	*serv = svsk->sk_xprt.xpt_server;
 	struct sk_buff	*skb;
 	union {
@@ -1106,7 +1115,8 @@ failed:
 static int
 svc_tcp_recvfrom(struct svc_rqst *rqstp)
 {
-	struct svc_sock	*svsk = rqstp->rq_sock;
+	struct svc_sock	*svsk =
+		container_of(rqstp->rq_xprt, struct svc_sock, sk_xprt);
 	struct svc_serv	*serv = svsk->sk_xprt.xpt_server;
 	int		len;
 	struct kvec *vec;
@@ -1270,16 +1280,16 @@ svc_tcp_sendto(struct svc_rqst *rqstp)
 	reclen = htonl(0x80000000|((xbufp->len ) - 4));
 	memcpy(xbufp->head[0].iov_base, &reclen, 4);
 
-	if (test_bit(XPT_DEAD, &rqstp->rq_sock->sk_xprt.xpt_flags))
+	if (test_bit(XPT_DEAD, &rqstp->rq_xprt->xpt_flags))
 		return -ENOTCONN;
 
 	sent = svc_sendto(rqstp, &rqstp->rq_res);
 	if (sent != xbufp->len) {
 		printk(KERN_NOTICE "rpc-srv/tcp: %s: %s %d when sending %d bytes - shutting down socket\n",
-		       rqstp->rq_sock->sk_xprt.xpt_server->sv_name,
+		       rqstp->rq_xprt->xpt_server->sv_name,
 		       (sent<0)?"got error":"sent only",
 		       sent, xbufp->len);
-		set_bit(XPT_CLOSE, &rqstp->rq_sock->sk_xprt.xpt_flags);
+		set_bit(XPT_CLOSE, &rqstp->rq_xprt->xpt_flags);
 		svc_xprt_enqueue(rqstp->rq_xprt);
 		sent = -EAGAIN;
 	}
@@ -1299,7 +1309,7 @@ static void svc_tcp_prep_reply_hdr(struct svc_rqst *rqstp)
 
 static int svc_tcp_has_wspace(struct svc_xprt *xprt)
 {
-	struct svc_sock *svsk = container_of(xprt, struct svc_sock, sk_xprt);
+	struct svc_sock *svsk =	container_of(xprt, struct svc_sock, sk_xprt);
 	struct svc_serv	*serv = svsk->sk_xprt.xpt_server;
 	int required;
 	int wspace;
@@ -1627,7 +1637,7 @@ svc_recv(struct svc_rqst *rqstp, long timeout)
 void
 svc_drop(struct svc_rqst *rqstp)
 {
-	dprintk("svc: socket %p dropped request\n", rqstp->rq_sock);
+	dprintk("svc: xprt %p dropped request\n", rqstp->rq_xprt);
 	svc_xprt_release(rqstp);
 }
 

[PATCH 32/38] svc: Make svc_check_conn_limits xprt independent

[Tom Tucker]

The svc_check_conn_limits function only manipulates xprt fields. Change references
to svc_sock->sk_xprt to svc_xprt directly.

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 net/sunrpc/svcsock.c |   28 ++++++++++++++--------------
 1 files changed, 14 insertions(+), 14 deletions(-)

diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index a785055..8578936 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -1447,34 +1447,34 @@ static void svc_check_conn_limits(struct svc_serv *serv)
 	char	buf[RPC_MAX_ADDRBUFLEN];
 
 	if (serv->sv_tmpcnt > (serv->sv_nrthreads+3)*20) {
-		struct svc_sock *svsk = NULL;
+		struct svc_xprt *xprt = NULL;
 		spin_lock_bh(&serv->sv_lock);
 		if (!list_empty(&serv->sv_tempsocks)) {
 			if (net_ratelimit()) {
 				/* Try to help the admin */
-				printk(KERN_NOTICE "%s: too many open TCP "
-					"sockets, consider increasing the "
+				printk(KERN_NOTICE "%s: too many open  "
+					"connections, consider increasing the "
 					"number of nfsd threads\n",
-						   serv->sv_name);
+				       serv->sv_name);
 				printk(KERN_NOTICE
-				       "%s: last TCP connect from %s\n",
+				       "%s: last connection from %s\n",
 				       serv->sv_name, buf);
 			}
 			/*
-			 * Always select the oldest socket. It's not fair,
+			 * Always select the oldest connection. It's not fair,
 			 * but so is life
 			 */
-			svsk = list_entry(serv->sv_tempsocks.prev,
-					  struct svc_sock,
-					  sk_xprt.xpt_list);
-			set_bit(XPT_CLOSE, &svsk->sk_xprt.xpt_flags);
-			svc_xprt_get(&svsk->sk_xprt);
+			xprt = list_entry(serv->sv_tempsocks.prev,
+					  struct svc_xprt,
+					  xpt_list);
+			set_bit(XPT_CLOSE, &xprt->xpt_flags);
+			svc_xprt_get(xprt);
 		}
 		spin_unlock_bh(&serv->sv_lock);
 
-		if (svsk) {
-			svc_xprt_enqueue(&svsk->sk_xprt);
-			svc_xprt_put(&svsk->sk_xprt);
+		if (xprt) {
+			svc_xprt_enqueue(xprt);
+			svc_xprt_put(xprt);
 		}
 	}
 }

[PATCH 33/38] svc: Move the xprt independent code to the svc_xprt.c file

[Tom Tucker]

This functionally trivial patch moves all of the transport independent
functions from the svcsock.c file to the transport independent svc_xprt.c
file.

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 include/linux/sunrpc/svc_xprt.h |   27 +
 net/sunrpc/svc_xprt.c           |  726 +++++++++++++++++++++++++++++++++++
 net/sunrpc/svcsock.c            |  809 +--------------------------------------
 3 files changed, 776 insertions(+), 786 deletions(-)

diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
index a1d324a..db8bcf5 100644
--- a/include/linux/sunrpc/svc_xprt.h
+++ b/include/linux/sunrpc/svc_xprt.h
@@ -72,9 +72,13 @@ int	svc_unreg_xprt_class(struct svc_xprt_class *);
 void	svc_xprt_init(struct svc_xprt_class *, struct svc_xprt *,
 		      struct svc_serv *);
 int	svc_create_xprt(struct svc_serv *, char *, unsigned short, int);
+void	svc_xprt_enqueue(struct svc_xprt *xprt);
 void	svc_xprt_received(struct svc_xprt *);
 void	svc_xprt_put(struct svc_xprt *xprt);
 void	svc_xprt_copy_addrs(struct svc_rqst *rqstp, struct svc_xprt *xprt);
+void	svc_close_xprt(struct svc_xprt *xprt);
+void	svc_delete_xprt(struct svc_xprt *xprt);
+int	svc_port_is_privileged(struct sockaddr *sin);
 static inline void svc_xprt_get(struct svc_xprt *xprt)
 {
 	kref_get(&xprt->xpt_ref);
@@ -126,4 +130,27 @@ static inline unsigned short svc_xprt_remote_port(struct svc_xprt *xprt)
 	return svc_addr_port((struct sockaddr *)&xprt->xpt_remote);
 }
 
+static inline char *__svc_print_addr(struct sockaddr *addr,
+				     char *buf, size_t len)
+{
+	switch (addr->sa_family) {
+	case AF_INET:
+		snprintf(buf, len, "%u.%u.%u.%u, port=%u",
+			NIPQUAD(((struct sockaddr_in *) addr)->sin_addr),
+			ntohs(((struct sockaddr_in *) addr)->sin_port));
+		break;
+
+	case AF_INET6:
+		snprintf(buf, len, "%x:%x:%x:%x:%x:%x:%x:%x, port=%u",
+			NIP6(((struct sockaddr_in6 *) addr)->sin6_addr),
+			ntohs(((struct sockaddr_in6 *) addr)->sin6_port));
+		break;
+
+	default:
+		snprintf(buf, len, "unknown address type: %d", addr->sa_family);
+		break;
+	}
+	return buf;
+}
+
 #endif /* SUNRPC_SVC_XPRT_H */
diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
index 924df63..a477883 100644
--- a/net/sunrpc/svc_xprt.c
+++ b/net/sunrpc/svc_xprt.c
@@ -35,6 +35,18 @@
 
 #define RPCDBG_FACILITY	RPCDBG_SVCXPRT
 
+static struct svc_deferred_req *svc_deferred_dequeue(struct svc_xprt *xprt);
+static int svc_deferred_recv(struct svc_rqst *rqstp);
+static struct cache_deferred_req *svc_defer(struct cache_req *req);
+static void svc_age_temp_xprts(unsigned long closure);
+
+/* apparently the "standard" is that clients close
+ * idle connections after 5 minutes, servers after
+ * 6 minutes
+ *   http://www.connectathon.org/talks96/nfstcp.pdf
+ */
+static int svc_conn_age_period = 6*60;
+
 /* List of registered transport classes */
 static DEFINE_SPINLOCK(svc_xprt_class_lock);
 static LIST_HEAD(svc_xprt_class_list);
@@ -191,3 +203,717 @@ void svc_xprt_copy_addrs(struct svc_rqst *rqstp, struct svc_xprt *xprt)
 }
 EXPORT_SYMBOL_GPL(svc_xprt_copy_addrs);
 
+/**
+ * svc_print_addr - Format rq_addr field for printing
+ * @rqstp: svc_rqst struct containing address to print
+ * @buf: target buffer for formatted address
+ * @len: length of target buffer
+ *
+ */
+char *svc_print_addr(struct svc_rqst *rqstp, char *buf, size_t len)
+{
+	return __svc_print_addr(svc_addr(rqstp), buf, len);
+}
+EXPORT_SYMBOL_GPL(svc_print_addr);
+
+/*
+ * Queue up an idle server thread.  Must have pool->sp_lock held.
+ * Note: this is really a stack rather than a queue, so that we only
+ * use as many different threads as we need, and the rest don't pollute
+ * the cache.
+ */
+static void svc_thread_enqueue(struct svc_pool *pool, struct svc_rqst *rqstp)
+{
+	list_add(&rqstp->rq_list, &pool->sp_threads);
+}
+
+/*
+ * Dequeue an nfsd thread.  Must have pool->sp_lock held.
+ */
+static void svc_thread_dequeue(struct svc_pool *pool, struct svc_rqst *rqstp)
+{
+	list_del(&rqstp->rq_list);
+}
+
+/*
+ * Queue up a socket with data pending. If there are idle nfsd
+ * processes, wake 'em up.
+ *
+ */
+void svc_xprt_enqueue(struct svc_xprt *xprt)
+{
+	struct svc_serv	*serv = xprt->xpt_server;
+	struct svc_pool *pool;
+	struct svc_rqst	*rqstp;
+	int cpu;
+
+	if (!(xprt->xpt_flags &
+	      ((1<<XPT_CONN)|(1<<XPT_DATA)|(1<<XPT_CLOSE)|(1<<XPT_DEFERRED))))
+		return;
+	if (test_bit(XPT_DEAD, &xprt->xpt_flags))
+		return;
+
+	cpu = get_cpu();
+	pool = svc_pool_for_cpu(xprt->xpt_server, cpu);
+	put_cpu();
+
+	spin_lock_bh(&pool->sp_lock);
+
+	if (!list_empty(&pool->sp_threads) &&
+	    !list_empty(&pool->sp_sockets))
+		printk(KERN_ERR
+		       "svc_xprt_enqueue: "
+		       "threads and transports both waiting??\n");
+
+	if (test_bit(XPT_DEAD, &xprt->xpt_flags)) {
+		/* Don't enqueue dead sockets */
+		dprintk("svc: transport %p is dead, not enqueued\n", xprt);
+		goto out_unlock;
+	}
+
+	/* Mark socket as busy. It will remain in this state until the
+	 * server has processed all pending data and put the socket back
+	 * on the idle list.  We update XPT_BUSY atomically because
+	 * it also guards against trying to enqueue the svc_sock twice.
+	 */
+	if (test_and_set_bit(XPT_BUSY, &xprt->xpt_flags)) {
+		/* Don't enqueue socket while already enqueued */
+		dprintk("svc: transport %p busy, not enqueued\n", xprt);
+		goto out_unlock;
+	}
+	BUG_ON(xprt->xpt_pool != NULL);
+	xprt->xpt_pool = pool;
+
+	/* Handle pending connection */
+	if (test_bit(XPT_CONN, &xprt->xpt_flags))
+		goto process;
+
+	/* Handle close in-progress */
+	if (test_bit(XPT_CLOSE, &xprt->xpt_flags))
+		goto process;
+
+	/* Check if we have space to reply to a request */
+	if (!xprt->xpt_ops->xpo_has_wspace(xprt)) {
+		/* Don't enqueue while not enough space for reply */
+		dprintk("svc: no write space, transport %p  not enqueued\n",
+			xprt);
+		xprt->xpt_pool = NULL;
+		clear_bit(XPT_BUSY, &xprt->xpt_flags);
+		goto out_unlock;
+	}
+
+ process:
+	if (!list_empty(&pool->sp_threads)) {
+		rqstp = list_entry(pool->sp_threads.next,
+				   struct svc_rqst,
+				   rq_list);
+		dprintk("svc: transport %p served by daemon %p\n",
+			xprt, rqstp);
+		svc_thread_dequeue(pool, rqstp);
+		if (rqstp->rq_xprt)
+			printk(KERN_ERR
+				"svc_xprt_enqueue: server %p, rq_xprt=%p!\n",
+				rqstp, rqstp->rq_xprt);
+		rqstp->rq_xprt = xprt;
+		svc_xprt_get(xprt);
+		rqstp->rq_reserved = serv->sv_max_mesg;
+		atomic_add(rqstp->rq_reserved, &xprt->xpt_reserved);
+		BUG_ON(xprt->xpt_pool != pool);
+		wake_up(&rqstp->rq_wait);
+	} else {
+		dprintk("svc: transport %p put into queue\n", xprt);
+		list_add_tail(&xprt->xpt_ready, &pool->sp_sockets);
+		BUG_ON(xprt->xpt_pool != pool);
+	}
+
+out_unlock:
+	spin_unlock_bh(&pool->sp_lock);
+}
+EXPORT_SYMBOL_GPL(svc_xprt_enqueue);
+
+/*
+ * Dequeue the first socket.  Must be called with the pool->sp_lock held.
+ */
+static struct svc_xprt *svc_xprt_dequeue(struct svc_pool *pool)
+{
+	struct svc_xprt	*xprt;
+
+	if (list_empty(&pool->sp_sockets))
+		return NULL;
+
+	xprt = list_entry(pool->sp_sockets.next,
+			  struct svc_xprt, xpt_ready);
+	list_del_init(&xprt->xpt_ready);
+
+	dprintk("svc: transport %p dequeued, inuse=%d\n",
+		xprt, atomic_read(&xprt->xpt_ref.refcount));
+
+	return xprt;
+}
+
+/*
+ * Having read something from a socket, check whether it
+ * needs to be re-enqueued.
+ * Note: XPT_DATA only gets cleared when a read-attempt finds
+ * no (or insufficient) data.
+ */
+void svc_xprt_received(struct svc_xprt *xprt)
+{
+	xprt->xpt_pool = NULL;
+	clear_bit(XPT_BUSY, &xprt->xpt_flags);
+	svc_xprt_enqueue(xprt);
+}
+EXPORT_SYMBOL_GPL(svc_xprt_received);
+
+/**
+ * svc_reserve - change the space reserved for the reply to a request.
+ * @rqstp:  The request in question
+ * @space: new max space to reserve
+ *
+ * Each request reserves some space on the output queue of the socket
+ * to make sure the reply fits.  This function reduces that reserved
+ * space to be the amount of space used already, plus @space.
+ *
+ */
+void svc_reserve(struct svc_rqst *rqstp, int space)
+{
+	space += rqstp->rq_res.head[0].iov_len;
+
+	if (space < rqstp->rq_reserved) {
+		struct svc_xprt *xprt = rqstp->rq_xprt;
+		atomic_sub((rqstp->rq_reserved - space), &xprt->xpt_reserved);
+		rqstp->rq_reserved = space;
+
+		svc_xprt_enqueue(xprt);
+	}
+}
+
+static void svc_xprt_release(struct svc_rqst *rqstp)
+{
+	struct svc_xprt	*xprt = rqstp->rq_xprt;
+
+	rqstp->rq_xprt->xpt_ops->xpo_release_rqst(rqstp);
+
+	svc_free_res_pages(rqstp);
+	rqstp->rq_res.page_len = 0;
+	rqstp->rq_res.page_base = 0;
+
+	/* Reset response buffer and release
+	 * the reservation.
+	 * But first, check that enough space was reserved
+	 * for the reply, otherwise we have a bug!
+	 */
+	if ((rqstp->rq_res.len) >  rqstp->rq_reserved)
+		printk(KERN_ERR "RPC request reserved %d but used %d\n",
+		       rqstp->rq_reserved,
+		       rqstp->rq_res.len);
+
+	rqstp->rq_res.head[0].iov_len = 0;
+	svc_reserve(rqstp, 0);
+	rqstp->rq_xprt = NULL;
+
+	svc_xprt_put(xprt);
+}
+
+/*
+ * External function to wake up a server waiting for data
+ * This really only makes sense for services like lockd
+ * which have exactly one thread anyway.
+ */
+void svc_wake_up(struct svc_serv *serv)
+{
+	struct svc_rqst	*rqstp;
+	unsigned int i;
+	struct svc_pool *pool;
+
+	for (i = 0; i < serv->sv_nrpools; i++) {
+		pool = &serv->sv_pools[i];
+
+		spin_lock_bh(&pool->sp_lock);
+		if (!list_empty(&pool->sp_threads)) {
+			rqstp = list_entry(pool->sp_threads.next,
+					   struct svc_rqst,
+					   rq_list);
+			dprintk("svc: daemon %p woken up.\n", rqstp);
+			/*
+			svc_thread_dequeue(pool, rqstp);
+			rqstp->rq_xprt = NULL;
+			 */
+			wake_up(&rqstp->rq_wait);
+		}
+		spin_unlock_bh(&pool->sp_lock);
+	}
+}
+
+int svc_port_is_privileged(struct sockaddr *sin)
+{
+	switch (sin->sa_family) {
+	case AF_INET:
+		return ntohs(((struct sockaddr_in *)sin)->sin_port)
+			< PROT_SOCK;
+	case AF_INET6:
+		return ntohs(((struct sockaddr_in6 *)sin)->sin6_port)
+			< PROT_SOCK;
+	default:
+		return 0;
+	}
+}
+
+/*
+ * Make sure that we don't have too many active connections.  If we
+ * have, something must be dropped.
+ *
+ * There's no point in trying to do random drop here for DoS
+ * prevention. The NFS clients does 1 reconnect in 15 seconds. An
+ * attacker can easily beat that.
+ *
+ * The only somewhat efficient mechanism would be if drop old
+ * connections from the same IP first. But right now we don't even
+ * record the client IP in svc_sock.
+ */
+static void svc_check_conn_limits(struct svc_serv *serv)
+{
+	char	buf[RPC_MAX_ADDRBUFLEN];
+
+	if (serv->sv_tmpcnt > (serv->sv_nrthreads+3)*20) {
+		struct svc_xprt *xprt = NULL;
+		spin_lock_bh(&serv->sv_lock);
+		if (!list_empty(&serv->sv_tempsocks)) {
+			if (net_ratelimit()) {
+				/* Try to help the admin */
+				printk(KERN_NOTICE "%s: too many open  "
+					"connections, consider increasing the "
+					"number of nfsd threads\n",
+				       serv->sv_name);
+				printk(KERN_NOTICE
+				       "%s: last connection from %s\n",
+				       serv->sv_name, buf);
+			}
+			/*
+			 * Always select the oldest connection. It's not fair,
+			 * but so is life
+			 */
+			xprt = list_entry(serv->sv_tempsocks.prev,
+					  struct svc_xprt,
+					  xpt_list);
+			set_bit(XPT_CLOSE, &xprt->xpt_flags);
+			svc_xprt_get(xprt);
+		}
+		spin_unlock_bh(&serv->sv_lock);
+
+		if (xprt) {
+			svc_xprt_enqueue(xprt);
+			svc_xprt_put(xprt);
+		}
+	}
+}
+
+/*
+ * Receive the next request on any socket.  This code is carefully
+ * organised not to touch any cachelines in the shared svc_serv
+ * structure, only cachelines in the local svc_pool.
+ */
+int svc_recv(struct svc_rqst *rqstp, long timeout)
+{
+	struct svc_xprt		*xprt = NULL;
+	struct svc_serv		*serv = rqstp->rq_server;
+	struct svc_pool		*pool = rqstp->rq_pool;
+	int			len, i;
+	int			pages;
+	struct xdr_buf		*arg;
+	DECLARE_WAITQUEUE(wait, current);
+
+	dprintk("svc: server %p waiting for data (to = %ld)\n",
+		rqstp, timeout);
+
+	if (rqstp->rq_xprt)
+		printk(KERN_ERR
+			"svc_recv: service %p, transport not NULL!\n",
+			 rqstp);
+	if (waitqueue_active(&rqstp->rq_wait))
+		printk(KERN_ERR
+			"svc_recv: service %p, wait queue active!\n",
+			 rqstp);
+
+
+	/* now allocate needed pages.  If we get a failure, sleep briefly */
+	pages = (serv->sv_max_mesg + PAGE_SIZE) / PAGE_SIZE;
+	for (i = 0; i < pages ; i++)
+		while (rqstp->rq_pages[i] == NULL) {
+			struct page *p = alloc_page(GFP_KERNEL);
+			if (!p) {
+				int jiffies = msecs_to_jiffies(500);
+				schedule_timeout_uninterruptible(jiffies);
+			}
+			rqstp->rq_pages[i] = p;
+		}
+	rqstp->rq_pages[i++] = NULL; /* this might be seen in nfs_read_actor */
+	BUG_ON(pages >= RPCSVC_MAXPAGES);
+
+	/* Make arg->head point to first page and arg->pages point to rest */
+	arg = &rqstp->rq_arg;
+	arg->head[0].iov_base = page_address(rqstp->rq_pages[0]);
+	arg->head[0].iov_len = PAGE_SIZE;
+	arg->pages = rqstp->rq_pages + 1;
+	arg->page_base = 0;
+	/* save at least one page for response */
+	arg->page_len = (pages-2)*PAGE_SIZE;
+	arg->len = (pages-1)*PAGE_SIZE;
+	arg->tail[0].iov_len = 0;
+
+	try_to_freeze();
+	cond_resched();
+	if (signalled())
+		return -EINTR;
+
+	spin_lock_bh(&pool->sp_lock);
+	xprt = svc_xprt_dequeue(pool);
+	if (xprt) {
+		rqstp->rq_xprt = xprt;
+		svc_xprt_get(xprt);
+		rqstp->rq_reserved = serv->sv_max_mesg;
+		atomic_add(rqstp->rq_reserved, &xprt->xpt_reserved);
+	} else {
+		/* No data pending. Go to sleep */
+		svc_thread_enqueue(pool, rqstp);
+
+		/*
+		 * We have to be able to interrupt this wait
+		 * to bring down the daemons ...
+		 */
+		set_current_state(TASK_INTERRUPTIBLE);
+		add_wait_queue(&rqstp->rq_wait, &wait);
+		spin_unlock_bh(&pool->sp_lock);
+
+		schedule_timeout(timeout);
+
+		try_to_freeze();
+
+		spin_lock_bh(&pool->sp_lock);
+		remove_wait_queue(&rqstp->rq_wait, &wait);
+
+		xprt = rqstp->rq_xprt;
+		if (!xprt) {
+			svc_thread_dequeue(pool, rqstp);
+			spin_unlock_bh(&pool->sp_lock);
+			dprintk("svc: server %p, no data yet\n", rqstp);
+			return signalled()? -EINTR : -EAGAIN;
+		}
+	}
+	spin_unlock_bh(&pool->sp_lock);
+
+	len = 0;
+	if (test_bit(XPT_CLOSE, &xprt->xpt_flags)) {
+		dprintk("svc_recv: found XPT_CLOSE\n");
+		svc_delete_xprt(xprt);
+	} else if (test_bit(XPT_LISTENER, &xprt->xpt_flags)) {
+		struct svc_xprt *newxpt;
+		newxpt = xprt->xpt_ops->xpo_accept(xprt);
+		if (newxpt) {
+			/*
+			 * We know this module_get will succeed because the
+			 * listener holds a reference too
+			 */
+			__module_get(newxpt->xpt_class->xcl_owner);
+			svc_check_conn_limits(xprt->xpt_server);
+			spin_lock_bh(&serv->sv_lock);
+			set_bit(XPT_TEMP, &newxpt->xpt_flags);
+			list_add(&newxpt->xpt_list, &serv->sv_tempsocks);
+			serv->sv_tmpcnt++;
+			if (serv->sv_temptimer.function == NULL) {
+				/* setup timer to age temp sockets */
+				setup_timer(&serv->sv_temptimer,
+					    svc_age_temp_xprts,
+					    (unsigned long)serv);
+				mod_timer(&serv->sv_temptimer,
+					  jiffies + svc_conn_age_period * HZ);
+			}
+			spin_unlock_bh(&serv->sv_lock);
+			svc_xprt_received(newxpt);
+		}
+		svc_xprt_received(xprt);
+	} else {
+		dprintk("svc: server %p, pool %u, transport %p, inuse=%d\n",
+			rqstp, pool->sp_id, xprt,
+			atomic_read(&xprt->xpt_ref.refcount));
+		rqstp->rq_deferred = svc_deferred_dequeue(xprt);
+		if (rqstp->rq_deferred) {
+			svc_xprt_received(xprt);
+			len = svc_deferred_recv(rqstp);
+		} else
+			len = xprt->xpt_ops->xpo_recvfrom(rqstp);
+		dprintk("svc: got len=%d\n", len);
+	}
+
+	/* No data, incomplete (TCP) read, or accept() */
+	if (len == 0 || len == -EAGAIN) {
+		rqstp->rq_res.len = 0;
+		svc_xprt_release(rqstp);
+		return -EAGAIN;
+	}
+	clear_bit(XPT_OLD, &xprt->xpt_flags);
+
+	rqstp->rq_secure = svc_port_is_privileged(svc_addr(rqstp));
+	rqstp->rq_chandle.defer = svc_defer;
+
+	if (serv->sv_stats)
+		serv->sv_stats->netcnt++;
+	return len;
+}
+
+/*
+ * Drop request
+ */
+void svc_drop(struct svc_rqst *rqstp)
+{
+	dprintk("svc: xprt %p dropped request\n", rqstp->rq_xprt);
+	svc_xprt_release(rqstp);
+}
+
+/*
+ * Return reply to client.
+ */
+int svc_send(struct svc_rqst *rqstp)
+{
+	struct svc_xprt	*xprt;
+	int		len;
+	struct xdr_buf	*xb;
+
+	xprt = rqstp->rq_xprt;
+	if (!xprt)
+		return -EFAULT;
+
+	/* release the receive skb before sending the reply */
+	rqstp->rq_xprt->xpt_ops->xpo_release_rqst(rqstp);
+
+	/* calculate over-all length */
+	xb = &rqstp->rq_res;
+	xb->len = xb->head[0].iov_len +
+		xb->page_len +
+		xb->tail[0].iov_len;
+
+	/* Grab mutex to serialize outgoing data. */
+	mutex_lock(&xprt->xpt_mutex);
+	if (test_bit(XPT_DEAD, &xprt->xpt_flags))
+		len = -ENOTCONN;
+	else
+		len = xprt->xpt_ops->xpo_sendto(rqstp);
+	mutex_unlock(&xprt->xpt_mutex);
+	svc_xprt_release(rqstp);
+
+	if (len == -ECONNREFUSED || len == -ENOTCONN || len == -EAGAIN)
+		return 0;
+	return len;
+}
+
+/*
+ * Timer function to close old temporary sockets, using
+ * a mark-and-sweep algorithm.
+ */
+static void svc_age_temp_xprts(unsigned long closure)
+{
+	struct svc_serv *serv = (struct svc_serv *)closure;
+	struct svc_xprt *xprt;
+	struct list_head *le, *next;
+	LIST_HEAD(to_be_aged);
+
+	dprintk("svc_age_temp_xprts\n");
+
+	if (!spin_trylock_bh(&serv->sv_lock)) {
+		/* busy, try again 1 sec later */
+		dprintk("svc_age_temp_xprts: busy\n");
+		mod_timer(&serv->sv_temptimer, jiffies + HZ);
+		return;
+	}
+
+	list_for_each_safe(le, next, &serv->sv_tempsocks) {
+		xprt = list_entry(le, struct svc_xprt, xpt_list);
+
+		/* First time through, just mark it OLD. Second time
+		 * through, close it. */
+		if (!test_and_set_bit(XPT_OLD, &xprt->xpt_flags))
+			continue;
+		if (atomic_read(&xprt->xpt_ref.refcount) > 1
+		    || test_bit(XPT_BUSY, &xprt->xpt_flags))
+			continue;
+		svc_xprt_get(xprt);
+		list_move(le, &to_be_aged);
+		set_bit(XPT_CLOSE, &xprt->xpt_flags);
+		set_bit(XPT_DETACHED, &xprt->xpt_flags);
+	}
+	spin_unlock_bh(&serv->sv_lock);
+
+	while (!list_empty(&to_be_aged)) {
+		le = to_be_aged.next;
+		/* fiddling the xpt_list node is safe 'cos we're XPT_DETACHED */
+		list_del_init(le);
+		xprt = list_entry(le, struct svc_xprt, xpt_list);
+
+		dprintk("queuing xprt %p for closing\n", xprt);
+
+		/* a thread will dequeue and close it soon */
+		svc_xprt_enqueue(xprt);
+		svc_xprt_put(xprt);
+	}
+
+	mod_timer(&serv->sv_temptimer, jiffies + svc_conn_age_period * HZ);
+}
+
+/*
+ * Remove a dead transport
+ */
+void svc_delete_xprt(struct svc_xprt *xprt)
+{
+	struct svc_serv	*serv = xprt->xpt_server;
+
+	dprintk("svc: svc_delete_xprt(%p)\n", xprt);
+	xprt->xpt_ops->xpo_detach(xprt);
+
+	spin_lock_bh(&serv->sv_lock);
+	if (!test_and_set_bit(XPT_DETACHED, &xprt->xpt_flags))
+		list_del_init(&xprt->xpt_list);
+	/*
+	 * We used to delete the transport from whichever list
+	 * it's sk_xprt.xpt_ready node was on, but we don't actually
+	 * need to.  This is because the only time we're called
+	 * while still attached to a queue, the queue itself
+	 * is about to be destroyed (in svc_destroy).
+	 */
+	if (!test_and_set_bit(XPT_DEAD, &xprt->xpt_flags)) {
+		BUG_ON(atomic_read(&xprt->xpt_ref.refcount) < 2);
+		svc_xprt_put(xprt);
+		if (test_bit(XPT_TEMP, &xprt->xpt_flags))
+			serv->sv_tmpcnt--;
+	}
+	spin_unlock_bh(&serv->sv_lock);
+}
+
+void svc_close_xprt(struct svc_xprt *xprt)
+{
+	set_bit(XPT_CLOSE, &xprt->xpt_flags);
+	if (test_and_set_bit(XPT_BUSY, &xprt->xpt_flags))
+		/* someone else will have to effect the close */
+		return;
+
+	svc_xprt_get(xprt);
+	svc_delete_xprt(xprt);
+	clear_bit(XPT_BUSY, &xprt->xpt_flags);
+	svc_xprt_put(xprt);
+}
+
+void svc_close_all(struct list_head *xprt_list)
+{
+	struct svc_xprt *xprt;
+	struct svc_xprt *tmp;
+
+	list_for_each_entry_safe(xprt, tmp, xprt_list, xpt_list) {
+		set_bit(XPT_CLOSE, &xprt->xpt_flags);
+		if (test_bit(XPT_BUSY, &xprt->xpt_flags)) {
+			/* Waiting to be processed, but no threads left,
+			 * So just remove it from the waiting list
+			 */
+			list_del_init(&xprt->xpt_ready);
+			clear_bit(XPT_BUSY, &xprt->xpt_flags);
+		}
+		svc_close_xprt(xprt);
+	}
+}
+
+/*
+ * Handle defer and revisit of requests
+ */
+
+static void svc_revisit(struct cache_deferred_req *dreq, int too_many)
+{
+	struct svc_deferred_req *dr =
+		container_of(dreq, struct svc_deferred_req, handle);
+	struct svc_xprt *xprt = dr->xprt;
+
+	if (too_many) {
+		svc_xprt_put(xprt);
+		kfree(dr);
+		return;
+	}
+	dprintk("revisit queued\n");
+	dr->xprt = NULL;
+	spin_lock(&xprt->xpt_lock);
+	list_add(&dr->handle.recent, &xprt->xpt_deferred);
+	spin_unlock(&xprt->xpt_lock);
+	set_bit(XPT_DEFERRED, &xprt->xpt_flags);
+	svc_xprt_enqueue(xprt);
+	svc_xprt_put(xprt);
+}
+
+static struct cache_deferred_req *svc_defer(struct cache_req *req)
+{
+	struct svc_rqst *rqstp = container_of(req, struct svc_rqst, rq_chandle);
+	int size = sizeof(struct svc_deferred_req) + (rqstp->rq_arg.len);
+	struct svc_deferred_req *dr;
+
+	if (rqstp->rq_arg.page_len)
+		return NULL; /* if more than a page, give up FIXME */
+	if (rqstp->rq_deferred) {
+		dr = rqstp->rq_deferred;
+		rqstp->rq_deferred = NULL;
+	} else {
+		int skip  = rqstp->rq_arg.len - rqstp->rq_arg.head[0].iov_len;
+		/* FIXME maybe discard if size too large */
+		dr = kmalloc(size, GFP_KERNEL);
+		if (dr == NULL)
+			return NULL;
+
+		dr->handle.owner = rqstp->rq_server;
+		dr->prot = rqstp->rq_prot;
+		memcpy(&dr->addr, &rqstp->rq_addr, rqstp->rq_addrlen);
+		dr->addrlen = rqstp->rq_addrlen;
+		dr->daddr = rqstp->rq_daddr;
+		dr->argslen = rqstp->rq_arg.len >> 2;
+		memcpy(dr->args, rqstp->rq_arg.head[0].iov_base-skip,
+		       dr->argslen<<2);
+	}
+	svc_xprt_get(rqstp->rq_xprt);
+	dr->xprt = rqstp->rq_xprt;
+
+	dr->handle.revisit = svc_revisit;
+	return &dr->handle;
+}
+
+/*
+ * recv data from a deferred request into an active one
+ */
+static int svc_deferred_recv(struct svc_rqst *rqstp)
+{
+	struct svc_deferred_req *dr = rqstp->rq_deferred;
+
+	rqstp->rq_arg.head[0].iov_base = dr->args;
+	rqstp->rq_arg.head[0].iov_len = dr->argslen<<2;
+	rqstp->rq_arg.page_len = 0;
+	rqstp->rq_arg.len = dr->argslen<<2;
+	rqstp->rq_prot        = dr->prot;
+	memcpy(&rqstp->rq_addr, &dr->addr, dr->addrlen);
+	rqstp->rq_addrlen     = dr->addrlen;
+	rqstp->rq_daddr       = dr->daddr;
+	rqstp->rq_respages    = rqstp->rq_pages;
+	return dr->argslen<<2;
+}
+
+
+static struct svc_deferred_req *svc_deferred_dequeue(struct svc_xprt *xprt)
+{
+	struct svc_deferred_req *dr = NULL;
+
+	if (!test_bit(XPT_DEFERRED, &xprt->xpt_flags))
+		return NULL;
+	spin_lock(&xprt->xpt_lock);
+	clear_bit(XPT_DEFERRED, &xprt->xpt_flags);
+	if (!list_empty(&xprt->xpt_deferred)) {
+		dr = list_entry(xprt->xpt_deferred.next,
+				struct svc_deferred_req,
+				handle.recent);
+		list_del_init(&dr->handle.recent);
+		set_bit(XPT_DEFERRED, &xprt->xpt_flags);
+	}
+	spin_unlock(&xprt->xpt_lock);
+	return dr;
+}
+
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index 8578936..0ab68f1 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -51,7 +51,7 @@
 /* SMP locking strategy:
  *
  *	svc_pool->sp_lock protects most of the fields of that pool.
- * 	svc_serv->sv_lock protects sv_tempsocks, sv_permsocks, sv_tmpcnt.
+ *	svc_serv->sv_lock protects sv_tempsocks, sv_permsocks, sv_tmpcnt.
  *	when both need to be taken (rare), svc_serv->sv_lock is first.
  *	BKL protects svc_serv->sv_nrthread.
  *	svc_sock->sk_lock protects the svc_sock->sk_deferred list
@@ -81,33 +81,20 @@
 
 static struct svc_sock *svc_setup_socket(struct svc_serv *, struct socket *,
 					 int *errp, int flags);
-static void		svc_delete_xprt(struct svc_xprt *xprt);
 static void		svc_udp_data_ready(struct sock *, int);
 static int		svc_udp_recvfrom(struct svc_rqst *);
 static int		svc_udp_sendto(struct svc_rqst *);
-static void		svc_close_xprt(struct svc_xprt *xprt);
 static void		svc_sock_detach(struct svc_xprt *);
 static void		svc_sock_free(struct svc_xprt *);
 
-static struct svc_deferred_req *svc_deferred_dequeue(struct svc_xprt *xprt);
-static int svc_deferred_recv(struct svc_rqst *rqstp);
-static struct cache_deferred_req *svc_defer(struct cache_req *req);
 static struct svc_xprt *svc_create_socket(struct svc_serv *, int,
 					  struct sockaddr *, int, int);
-static void svc_age_temp_xprts(unsigned long closure);
-
-/* apparently the "standard" is that clients close
- * idle connections after 5 minutes, servers after
- * 6 minutes
- *   http://www.connectathon.org/talks96/nfstcp.pdf
- */
-static int svc_conn_age_period = 6*60;
 
 #ifdef CONFIG_DEBUG_LOCK_ALLOC
 static struct lock_class_key svc_key[2];
 static struct lock_class_key svc_slock_key[2];
 
-static inline void svc_reclassify_socket(struct socket *sock)
+static void svc_reclassify_socket(struct socket *sock)
 {
 	struct sock *sk = sock->sk;
 	BUG_ON(sock_owned_by_user(sk));
@@ -131,67 +118,11 @@ static inline void svc_reclassify_socket(struct socket *sock)
 	}
 }
 #else
-static inline void svc_reclassify_socket(struct socket *sock)
+static void svc_reclassify_socket(struct socket *sock)
 {
 }
 #endif
 
-static char *__svc_print_addr(struct sockaddr *addr, char *buf, size_t len)
-{
-	switch (addr->sa_family) {
-	case AF_INET:
-		snprintf(buf, len, "%u.%u.%u.%u, port=%u",
-			NIPQUAD(((struct sockaddr_in *) addr)->sin_addr),
-			ntohs(((struct sockaddr_in *) addr)->sin_port));
-		break;
-
-	case AF_INET6:
-		snprintf(buf, len, "%x:%x:%x:%x:%x:%x:%x:%x, port=%u",
-			NIP6(((struct sockaddr_in6 *) addr)->sin6_addr),
-			ntohs(((struct sockaddr_in6 *) addr)->sin6_port));
-		break;
-
-	default:
-		snprintf(buf, len, "unknown address type: %d", addr->sa_family);
-		break;
-	}
-	return buf;
-}
-
-/**
- * svc_print_addr - Format rq_addr field for printing
- * @rqstp: svc_rqst struct containing address to print
- * @buf: target buffer for formatted address
- * @len: length of target buffer
- *
- */
-char *svc_print_addr(struct svc_rqst *rqstp, char *buf, size_t len)
-{
-	return __svc_print_addr(svc_addr(rqstp), buf, len);
-}
-EXPORT_SYMBOL_GPL(svc_print_addr);
-
-/*
- * Queue up an idle server thread.  Must have pool->sp_lock held.
- * Note: this is really a stack rather than a queue, so that we only
- * use as many different threads as we need, and the rest don't pollute
- * the cache.
- */
-static inline void
-svc_thread_enqueue(struct svc_pool *pool, struct svc_rqst *rqstp)
-{
-	list_add(&rqstp->rq_list, &pool->sp_threads);
-}
-
-/*
- * Dequeue an nfsd thread.  Must have pool->sp_lock held.
- */
-static inline void
-svc_thread_dequeue(struct svc_pool *pool, struct svc_rqst *rqstp)
-{
-	list_del(&rqstp->rq_list);
-}
-
 /*
  * Release an skbuff after use
  */
@@ -214,217 +145,6 @@ static void svc_release_skb(struct svc_rqst *rqstp)
 	}
 }
 
-/*
- * Queue up a socket with data pending. If there are idle nfsd
- * processes, wake 'em up.
- *
- */
-void svc_xprt_enqueue(struct svc_xprt *xprt)
-{
-	struct svc_serv	*serv = xprt->xpt_server;
-	struct svc_pool *pool;
-	struct svc_rqst	*rqstp;
-	int cpu;
-
-	if (!(xprt->xpt_flags &
-	      ((1<<XPT_CONN)|(1<<XPT_DATA)|(1<<XPT_CLOSE)|(1<<XPT_DEFERRED))))
-		return;
-	if (test_bit(XPT_DEAD, &xprt->xpt_flags))
-		return;
-
-	cpu = get_cpu();
-	pool = svc_pool_for_cpu(xprt->xpt_server, cpu);
-	put_cpu();
-
-	spin_lock_bh(&pool->sp_lock);
-
-	if (!list_empty(&pool->sp_threads) &&
-	    !list_empty(&pool->sp_sockets))
-		printk(KERN_ERR
-		       "svc_xprt_enqueue: "
-		       "threads and transports both waiting??\n");
-
-	if (test_bit(XPT_DEAD, &xprt->xpt_flags)) {
-		/* Don't enqueue dead sockets */
-		dprintk("svc: transport %p is dead, not enqueued\n", xprt);
-		goto out_unlock;
-	}
-
-	/* Mark socket as busy. It will remain in this state until the
-	 * server has processed all pending data and put the socket back
-	 * on the idle list.  We update XPT_BUSY atomically because
-	 * it also guards against trying to enqueue the svc_sock twice.
-	 */
-	if (test_and_set_bit(XPT_BUSY, &xprt->xpt_flags)) {
-		/* Don't enqueue socket while already enqueued */
-		dprintk("svc: transport %p busy, not enqueued\n", xprt);
-		goto out_unlock;
-	}
-	BUG_ON(xprt->xpt_pool != NULL);
-	xprt->xpt_pool = pool;
-
-	/* Handle pending connection */
-	if (test_bit(XPT_CONN, &xprt->xpt_flags))
-		goto process;
-
-	/* Handle close in-progress */
-	if (test_bit(XPT_CLOSE, &xprt->xpt_flags))
-		goto process;
-
-	/* Check if we have space to reply to a request */
-	if (!xprt->xpt_ops->xpo_has_wspace(xprt)) {
-		/* Don't enqueue while not enough space for reply */
-		dprintk("svc: no write space, transport %p  not enqueued\n",
-			xprt);
-		xprt->xpt_pool = NULL;
-		clear_bit(XPT_BUSY, &xprt->xpt_flags);
-		goto out_unlock;
-	}
-
- process:
-	if (!list_empty(&pool->sp_threads)) {
-		rqstp = list_entry(pool->sp_threads.next,
-				   struct svc_rqst,
-				   rq_list);
-		dprintk("svc: transport %p served by daemon %p\n",
-			xprt, rqstp);
-		svc_thread_dequeue(pool, rqstp);
-		if (rqstp->rq_xprt)
-			printk(KERN_ERR
-				"svc_xprt_enqueue: server %p, rq_xprt=%p!\n",
-				rqstp, rqstp->rq_xprt);
-		rqstp->rq_xprt = xprt;
-		svc_xprt_get(xprt);
-		rqstp->rq_reserved = serv->sv_max_mesg;
-		atomic_add(rqstp->rq_reserved, &xprt->xpt_reserved);
-		BUG_ON(xprt->xpt_pool != pool);
-		wake_up(&rqstp->rq_wait);
-	} else {
-		dprintk("svc: transport %p put into queue\n", xprt);
-		list_add_tail(&xprt->xpt_ready, &pool->sp_sockets);
-		BUG_ON(xprt->xpt_pool != pool);
-	}
-
-out_unlock:
-	spin_unlock_bh(&pool->sp_lock);
-}
-EXPORT_SYMBOL_GPL(svc_xprt_enqueue);
-
-/*
- * Dequeue the first socket.  Must be called with the pool->sp_lock held.
- */
-static struct svc_xprt *svc_xprt_dequeue(struct svc_pool *pool)
-{
-	struct svc_xprt	*xprt;
-
-	if (list_empty(&pool->sp_sockets))
-		return NULL;
-
-	xprt = list_entry(pool->sp_sockets.next,
-			  struct svc_xprt, xpt_ready);
-	list_del_init(&xprt->xpt_ready);
-
-	dprintk("svc: transport %p dequeued, inuse=%d\n",
-		xprt, atomic_read(&xprt->xpt_ref.refcount));
-
-	return xprt;
-}
-
-/*
- * Having read something from a socket, check whether it
- * needs to be re-enqueued.
- * Note: XPT_DATA only gets cleared when a read-attempt finds
- * no (or insufficient) data.
- */
-void svc_xprt_received(struct svc_xprt *xprt)
-{
-	xprt->xpt_pool = NULL;
-	clear_bit(XPT_BUSY, &xprt->xpt_flags);
-	svc_xprt_enqueue(xprt);
-}
-EXPORT_SYMBOL_GPL(svc_xprt_received);
-
-/**
- * svc_reserve - change the space reserved for the reply to a request.
- * @rqstp:  The request in question
- * @space: new max space to reserve
- *
- * Each request reserves some space on the output queue of the socket
- * to make sure the reply fits.  This function reduces that reserved
- * space to be the amount of space used already, plus @space.
- *
- */
-void svc_reserve(struct svc_rqst *rqstp, int space)
-{
-	space += rqstp->rq_res.head[0].iov_len;
-
-	if (space < rqstp->rq_reserved) {
-		struct svc_xprt *xprt = rqstp->rq_xprt;
-		atomic_sub((rqstp->rq_reserved - space), &xprt->xpt_reserved);
-		rqstp->rq_reserved = space;
-
-		svc_xprt_enqueue(xprt);
-	}
-}
-
-static void svc_xprt_release(struct svc_rqst *rqstp)
-{
-	struct svc_xprt	*xprt = rqstp->rq_xprt;
-
-	rqstp->rq_xprt->xpt_ops->xpo_release_rqst(rqstp);
-
-	svc_free_res_pages(rqstp);
-	rqstp->rq_res.page_len = 0;
-	rqstp->rq_res.page_base = 0;
-
-	/* Reset response buffer and release
-	 * the reservation.
-	 * But first, check that enough space was reserved
-	 * for the reply, otherwise we have a bug!
-	 */
-	if ((rqstp->rq_res.len) >  rqstp->rq_reserved)
-		printk(KERN_ERR "RPC request reserved %d but used %d\n",
-		       rqstp->rq_reserved,
-		       rqstp->rq_res.len);
-
-	rqstp->rq_res.head[0].iov_len = 0;
-	svc_reserve(rqstp, 0);
-	rqstp->rq_xprt = NULL;
-
-	svc_xprt_put(xprt);
-}
-
-/*
- * External function to wake up a server waiting for data
- * This really only makes sense for services like lockd
- * which have exactly one thread anyway.
- */
-void
-svc_wake_up(struct svc_serv *serv)
-{
-	struct svc_rqst	*rqstp;
-	unsigned int i;
-	struct svc_pool *pool;
-
-	for (i = 0; i < serv->sv_nrpools; i++) {
-		pool = &serv->sv_pools[i];
-
-		spin_lock_bh(&pool->sp_lock);
-		if (!list_empty(&pool->sp_threads)) {
-			rqstp = list_entry(pool->sp_threads.next,
-					   struct svc_rqst,
-					   rq_list);
-			dprintk("svc: daemon %p woken up.\n", rqstp);
-			/*
-			svc_thread_dequeue(pool, rqstp);
-			rqstp->rq_xprt = NULL;
-			 */
-			wake_up(&rqstp->rq_wait);
-		}
-		spin_unlock_bh(&pool->sp_lock);
-	}
-}
-
 union svc_pktinfo_u {
 	struct in_pktinfo pkti;
 	struct in6_pktinfo pkti6;
@@ -466,8 +186,7 @@ static void svc_set_cmsg_data(struct svc_rqst *rqstp, struct cmsghdr *cmh)
 /*
  * Generic sendto routine
  */
-static int
-svc_sendto(struct svc_rqst *rqstp, struct xdr_buf *xdr)
+static int svc_sendto(struct svc_rqst *rqstp, struct xdr_buf *xdr)
 {
 	struct svc_sock	*svsk =
 		container_of(rqstp->rq_xprt, struct svc_sock, sk_xprt);
@@ -571,8 +290,7 @@ static int one_sock_name(char *buf, struct svc_sock *svsk)
 	return len;
 }
 
-int
-svc_sock_names(char *buf, struct svc_serv *serv, char *toclose)
+int svc_sock_names(char *buf, struct svc_serv *serv, char *toclose)
 {
 	struct svc_sock *svsk, *closesk = NULL;
 	int len = 0;
@@ -602,8 +320,7 @@ EXPORT_SYMBOL(svc_sock_names);
 /*
  * Check input queue length
  */
-static int
-svc_recv_available(struct svc_sock *svsk)
+static int svc_recv_available(struct svc_sock *svsk)
 {
 	struct socket	*sock = svsk->sk_sock;
 	int		avail, err;
@@ -616,8 +333,8 @@ svc_recv_available(struct svc_sock *svsk)
 /*
  * Generic recvfrom routine.
  */
-static int
-svc_recvfrom(struct svc_rqst *rqstp, struct kvec *iov, int nr, int buflen)
+static int svc_recvfrom(struct svc_rqst *rqstp, struct kvec *iov, int nr,
+			int buflen)
 {
 	struct svc_sock *svsk =
 		container_of(rqstp->rq_xprt, struct svc_sock, sk_xprt);
@@ -637,8 +354,8 @@ svc_recvfrom(struct svc_rqst *rqstp, struct kvec *iov, int nr, int buflen)
 /*
  * Set socket snd and rcv buffer lengths
  */
-static inline void
-svc_sock_setbufsize(struct socket *sock, unsigned int snd, unsigned int rcv)
+static void svc_sock_setbufsize(struct socket *sock,
+				unsigned int snd, unsigned int rcv)
 {
 #if 0
 	mm_segment_t	oldfs;
@@ -663,8 +380,7 @@ svc_sock_setbufsize(struct socket *sock, unsigned int snd, unsigned int rcv)
 /*
  * INET callback when data has been received on the socket.
  */
-static void
-svc_udp_data_ready(struct sock *sk, int count)
+static void svc_udp_data_ready(struct sock *sk, int count)
 {
 	struct svc_sock	*svsk = (struct svc_sock *)sk->sk_user_data;
 
@@ -682,8 +398,7 @@ svc_udp_data_ready(struct sock *sk, int count)
 /*
  * INET callback when space is newly available on the socket.
  */
-static void
-svc_write_space(struct sock *sk)
+static void svc_write_space(struct sock *sk)
 {
 	struct svc_sock	*svsk = (struct svc_sock *)(sk->sk_user_data);
 
@@ -729,8 +444,7 @@ static void svc_udp_get_dest_address(struct svc_rqst *rqstp,
 /*
  * Receive a datagram from a UDP socket.
  */
-static int
-svc_udp_recvfrom(struct svc_rqst *rqstp)
+static int svc_udp_recvfrom(struct svc_rqst *rqstp)
 {
 	struct svc_sock	*svsk =
 		container_of(rqstp->rq_xprt, struct svc_sock, sk_xprt);
@@ -850,8 +564,7 @@ svc_udp_recvfrom(struct svc_rqst *rqstp)
 	return len;
 }
 
-static int
-svc_udp_sendto(struct svc_rqst *rqstp)
+static int svc_udp_sendto(struct svc_rqst *rqstp)
 {
 	int		error;
 
@@ -950,8 +663,7 @@ static void svc_udp_init(struct svc_sock *svsk, struct svc_serv *serv)
  * A data_ready event on a listening socket means there's a connection
  * pending. Do not use state_change as a substitute for it.
  */
-static void
-svc_tcp_listen_data_ready(struct sock *sk, int count_unused)
+static void svc_tcp_listen_data_ready(struct sock *sk, int count_unused)
 {
 	struct svc_sock	*svsk = (struct svc_sock *)sk->sk_user_data;
 
@@ -983,8 +695,7 @@ svc_tcp_listen_data_ready(struct sock *sk, int count_unused)
 /*
  * A state change on a connected socket means it's dying or dead.
  */
-static void
-svc_tcp_state_change(struct sock *sk)
+static void svc_tcp_state_change(struct sock *sk)
 {
 	struct svc_sock	*svsk = (struct svc_sock *)sk->sk_user_data;
 
@@ -1001,8 +712,7 @@ svc_tcp_state_change(struct sock *sk)
 		wake_up_interruptible_all(sk->sk_sleep);
 }
 
-static void
-svc_tcp_data_ready(struct sock *sk, int count)
+static void svc_tcp_data_ready(struct sock *sk, int count)
 {
 	struct svc_sock *svsk = (struct svc_sock *)sk->sk_user_data;
 
@@ -1016,20 +726,6 @@ svc_tcp_data_ready(struct sock *sk, int count)
 		wake_up_interruptible(sk->sk_sleep);
 }
 
-static inline int svc_port_is_privileged(struct sockaddr *sin)
-{
-	switch (sin->sa_family) {
-	case AF_INET:
-		return ntohs(((struct sockaddr_in *)sin)->sin_port)
-			< PROT_SOCK;
-	case AF_INET6:
-		return ntohs(((struct sockaddr_in6 *)sin)->sin6_port)
-			< PROT_SOCK;
-	default:
-		return 0;
-	}
-}
-
 /*
  * Accept a TCP connection
  */
@@ -1112,8 +808,7 @@ failed:
 /*
  * Receive data from a TCP socket.
  */
-static int
-svc_tcp_recvfrom(struct svc_rqst *rqstp)
+static int svc_tcp_recvfrom(struct svc_rqst *rqstp)
 {
 	struct svc_sock	*svsk =
 		container_of(rqstp->rq_xprt, struct svc_sock, sk_xprt);
@@ -1266,8 +961,7 @@ svc_tcp_recvfrom(struct svc_rqst *rqstp)
 /*
  * Send out data on TCP socket.
  */
-static int
-svc_tcp_sendto(struct svc_rqst *rqstp)
+static int svc_tcp_sendto(struct svc_rqst *rqstp)
 {
 	struct xdr_buf	*xbufp = &rqstp->rq_res;
 	int sent;
@@ -1285,7 +979,9 @@ svc_tcp_sendto(struct svc_rqst *rqstp)
 
 	sent = svc_sendto(rqstp, &rqstp->rq_res);
 	if (sent != xbufp->len) {
-		printk(KERN_NOTICE "rpc-srv/tcp: %s: %s %d when sending %d bytes - shutting down socket\n",
+		printk(KERN_NOTICE
+		       "rpc-srv/tcp: %s: %s %d when sending %d bytes - "
+		       "shutting down socket\n",
 		       rqstp->rq_xprt->xpt_server->sv_name,
 		       (sent<0)?"got error":"sent only",
 		       sent, xbufp->len);
@@ -1407,8 +1103,7 @@ static void svc_tcp_init(struct svc_sock *svsk, struct svc_serv *serv)
 	}
 }
 
-void
-svc_sock_update_bufs(struct svc_serv *serv)
+void svc_sock_update_bufs(struct svc_serv *serv)
 {
 	/*
 	 * The number of server threads has changed. Update
@@ -1431,307 +1126,6 @@ svc_sock_update_bufs(struct svc_serv *serv)
 }
 
 /*
- * Make sure that we don't have too many active connections.  If we
- * have, something must be dropped.
- *
- * There's no point in trying to do random drop here for DoS
- * prevention. The NFS clients does 1 reconnect in 15 seconds. An
- * attacker can easily beat that.
- *
- * The only somewhat efficient mechanism would be if drop old
- * connections from the same IP first. But right now we don't even
- * record the client IP in svc_sock.
- */
-static void svc_check_conn_limits(struct svc_serv *serv)
-{
-	char	buf[RPC_MAX_ADDRBUFLEN];
-
-	if (serv->sv_tmpcnt > (serv->sv_nrthreads+3)*20) {
-		struct svc_xprt *xprt = NULL;
-		spin_lock_bh(&serv->sv_lock);
-		if (!list_empty(&serv->sv_tempsocks)) {
-			if (net_ratelimit()) {
-				/* Try to help the admin */
-				printk(KERN_NOTICE "%s: too many open  "
-					"connections, consider increasing the "
-					"number of nfsd threads\n",
-				       serv->sv_name);
-				printk(KERN_NOTICE
-				       "%s: last connection from %s\n",
-				       serv->sv_name, buf);
-			}
-			/*
-			 * Always select the oldest connection. It's not fair,
-			 * but so is life
-			 */
-			xprt = list_entry(serv->sv_tempsocks.prev,
-					  struct svc_xprt,
-					  xpt_list);
-			set_bit(XPT_CLOSE, &xprt->xpt_flags);
-			svc_xprt_get(xprt);
-		}
-		spin_unlock_bh(&serv->sv_lock);
-
-		if (xprt) {
-			svc_xprt_enqueue(xprt);
-			svc_xprt_put(xprt);
-		}
-	}
-}
-
-/*
- * Receive the next request on any socket.  This code is carefully
- * organised not to touch any cachelines in the shared svc_serv
- * structure, only cachelines in the local svc_pool.
- */
-int
-svc_recv(struct svc_rqst *rqstp, long timeout)
-{
-	struct svc_xprt		*xprt = NULL;
-	struct svc_serv		*serv = rqstp->rq_server;
-	struct svc_pool		*pool = rqstp->rq_pool;
-	int			len, i;
-	int			pages;
-	struct xdr_buf		*arg;
-	DECLARE_WAITQUEUE(wait, current);
-
-	dprintk("svc: server %p waiting for data (to = %ld)\n",
-		rqstp, timeout);
-
-	if (rqstp->rq_xprt)
-		printk(KERN_ERR
-			"svc_recv: service %p, transport not NULL!\n",
-			 rqstp);
-	if (waitqueue_active(&rqstp->rq_wait))
-		printk(KERN_ERR
-			"svc_recv: service %p, wait queue active!\n",
-			 rqstp);
-
-
-	/* now allocate needed pages.  If we get a failure, sleep briefly */
-	pages = (serv->sv_max_mesg + PAGE_SIZE) / PAGE_SIZE;
-	for (i=0; i < pages ; i++)
-		while (rqstp->rq_pages[i] == NULL) {
-			struct page *p = alloc_page(GFP_KERNEL);
-			if (!p)
-				schedule_timeout_uninterruptible(msecs_to_jiffies(500));
-			rqstp->rq_pages[i] = p;
-		}
-	rqstp->rq_pages[i++] = NULL; /* this might be seen in nfs_read_actor */
-	BUG_ON(pages >= RPCSVC_MAXPAGES);
-
-	/* Make arg->head point to first page and arg->pages point to rest */
-	arg = &rqstp->rq_arg;
-	arg->head[0].iov_base = page_address(rqstp->rq_pages[0]);
-	arg->head[0].iov_len = PAGE_SIZE;
-	arg->pages = rqstp->rq_pages + 1;
-	arg->page_base = 0;
-	/* save at least one page for response */
-	arg->page_len = (pages-2)*PAGE_SIZE;
-	arg->len = (pages-1)*PAGE_SIZE;
-	arg->tail[0].iov_len = 0;
-
-	try_to_freeze();
-	cond_resched();
-	if (signalled())
-		return -EINTR;
-
-	spin_lock_bh(&pool->sp_lock);
-	xprt = svc_xprt_dequeue(pool);
-	if (xprt) {
-		rqstp->rq_xprt = xprt;
-		svc_xprt_get(xprt);
-		rqstp->rq_reserved = serv->sv_max_mesg;
-		atomic_add(rqstp->rq_reserved, &xprt->xpt_reserved);
-	} else {
-		/* No data pending. Go to sleep */
-		svc_thread_enqueue(pool, rqstp);
-
-		/*
-		 * We have to be able to interrupt this wait
-		 * to bring down the daemons ...
-		 */
-		set_current_state(TASK_INTERRUPTIBLE);
-		add_wait_queue(&rqstp->rq_wait, &wait);
-		spin_unlock_bh(&pool->sp_lock);
-
-		schedule_timeout(timeout);
-
-		try_to_freeze();
-
-		spin_lock_bh(&pool->sp_lock);
-		remove_wait_queue(&rqstp->rq_wait, &wait);
-
-		xprt = rqstp->rq_xprt;
-		if (!xprt) {
-			svc_thread_dequeue(pool, rqstp);
-			spin_unlock_bh(&pool->sp_lock);
-			dprintk("svc: server %p, no data yet\n", rqstp);
-			return signalled()? -EINTR : -EAGAIN;
-		}
-	}
-	spin_unlock_bh(&pool->sp_lock);
-
-	len = 0;
-	if (test_bit(XPT_CLOSE, &xprt->xpt_flags)) {
-		dprintk("svc_recv: found XPT_CLOSE\n");
-		svc_delete_xprt(xprt);
-	} else if (test_bit(XPT_LISTENER, &xprt->xpt_flags)) {
-		struct svc_xprt *newxpt;
-		newxpt = xprt->xpt_ops->xpo_accept(xprt);
-		if (newxpt) {
-			/*
-			 * We know this module_get will succeed because the
-			 * listener holds a reference too
-			 */
-			__module_get(newxpt->xpt_class->xcl_owner);
-			svc_check_conn_limits(xprt->xpt_server);
-			spin_lock_bh(&serv->sv_lock);
-			set_bit(XPT_TEMP, &newxpt->xpt_flags);
-			list_add(&newxpt->xpt_list, &serv->sv_tempsocks);
-			serv->sv_tmpcnt++;
-			if (serv->sv_temptimer.function == NULL) {
-				/* setup timer to age temp sockets */
-				setup_timer(&serv->sv_temptimer,
-					    svc_age_temp_xprts,
-					    (unsigned long)serv);
-				mod_timer(&serv->sv_temptimer,
-					  jiffies + svc_conn_age_period * HZ);
-			}
-			spin_unlock_bh(&serv->sv_lock);
-			svc_xprt_received(newxpt);
-		}
-		svc_xprt_received(xprt);
-	} else {
-		dprintk("svc: server %p, pool %u, transport %p, inuse=%d\n",
-			rqstp, pool->sp_id, xprt,
-			atomic_read(&xprt->xpt_ref.refcount));
-		rqstp->rq_deferred = svc_deferred_dequeue(xprt);
-		if (rqstp->rq_deferred) {
-			svc_xprt_received(xprt);
-			len = svc_deferred_recv(rqstp);
-		} else
-			len = xprt->xpt_ops->xpo_recvfrom(rqstp);
-		dprintk("svc: got len=%d\n", len);
-	}
-
-	/* No data, incomplete (TCP) read, or accept() */
-	if (len == 0 || len == -EAGAIN) {
-		rqstp->rq_res.len = 0;
-		svc_xprt_release(rqstp);
-		return -EAGAIN;
-	}
-	clear_bit(XPT_OLD, &xprt->xpt_flags);
-
-	rqstp->rq_secure = svc_port_is_privileged(svc_addr(rqstp));
-	rqstp->rq_chandle.defer = svc_defer;
-
-	if (serv->sv_stats)
-		serv->sv_stats->netcnt++;
-	return len;
-}
-
-/*
- * Drop request
- */
-void
-svc_drop(struct svc_rqst *rqstp)
-{
-	dprintk("svc: xprt %p dropped request\n", rqstp->rq_xprt);
-	svc_xprt_release(rqstp);
-}
-
-/*
- * Return reply to client.
- */
-int
-svc_send(struct svc_rqst *rqstp)
-{
-	struct svc_xprt	*xprt;
-	int		len;
-	struct xdr_buf	*xb;
-
-	xprt = rqstp->rq_xprt;
-	if (!xprt)
-		return -EFAULT;
-
-	/* release the receive skb before sending the reply */
-	rqstp->rq_xprt->xpt_ops->xpo_release_rqst(rqstp);
-
-	/* calculate over-all length */
-	xb = & rqstp->rq_res;
-	xb->len = xb->head[0].iov_len +
-		xb->page_len +
-		xb->tail[0].iov_len;
-
-	/* Grab mutex to serialize outgoing data. */
-	mutex_lock(&xprt->xpt_mutex);
-	if (test_bit(XPT_DEAD, &xprt->xpt_flags))
-		len = -ENOTCONN;
-	else
-		len = xprt->xpt_ops->xpo_sendto(rqstp);
-	mutex_unlock(&xprt->xpt_mutex);
-	svc_xprt_release(rqstp);
-
-	if (len == -ECONNREFUSED || len == -ENOTCONN || len == -EAGAIN)
-		return 0;
-	return len;
-}
-
-/*
- * Timer function to close old temporary sockets, using
- * a mark-and-sweep algorithm.
- */
-static void svc_age_temp_xprts(unsigned long closure)
-{
-	struct svc_serv *serv = (struct svc_serv *)closure;
-	struct svc_xprt *xprt;
-	struct list_head *le, *next;
-	LIST_HEAD(to_be_aged);
-
-	dprintk("svc_age_temp_xprts\n");
-
-	if (!spin_trylock_bh(&serv->sv_lock)) {
-		/* busy, try again 1 sec later */
-		dprintk("svc_age_temp_xprts: busy\n");
-		mod_timer(&serv->sv_temptimer, jiffies + HZ);
-		return;
-	}
-
-	list_for_each_safe(le, next, &serv->sv_tempsocks) {
-		xprt = list_entry(le, struct svc_xprt, xpt_list);
-
-		/* First time through, just mark it OLD. Second time
-		 * through, close it. */
-		if (!test_and_set_bit(XPT_OLD, &xprt->xpt_flags))
-			continue;
-		if (atomic_read(&xprt->xpt_ref.refcount) > 1
-		    || test_bit(XPT_BUSY, &xprt->xpt_flags))
-			continue;
-		svc_xprt_get(xprt);
-		list_move(le, &to_be_aged);
-		set_bit(XPT_CLOSE, &xprt->xpt_flags);
-		set_bit(XPT_DETACHED, &xprt->xpt_flags);
-	}
-	spin_unlock_bh(&serv->sv_lock);
-
-	while (!list_empty(&to_be_aged)) {
-		le = to_be_aged.next;
-		/* fiddling the xpt_list node is safe 'cos we're XPT_DETACHED */
-		list_del_init(le);
-		xprt = list_entry(le, struct svc_xprt, xpt_list);
-
-		dprintk("queuing xprt %p for closing\n", xprt);
-
-		/* a thread will dequeue and close it soon */
-		svc_xprt_enqueue(xprt);
-		svc_xprt_put(xprt);
-	}
-
-	mod_timer(&serv->sv_temptimer, jiffies + svc_conn_age_period * HZ);
-}
-
-/*
  * Initialize socket for RPC use and create svc_sock struct
  * XXX: May want to setsockopt SO_SNDBUF and SO_RCVBUF.
  */
@@ -1917,160 +1311,3 @@ static void svc_sock_free(struct svc_xprt *xprt)
 		sock_release(svsk->sk_sock);
 	kfree(svsk);
 }
-
-/*
- * Remove a dead transport
- */
-static void svc_delete_xprt(struct svc_xprt *xprt)
-{
-	struct svc_serv	*serv = xprt->xpt_server;
-
-	dprintk("svc: svc_delete_xprt(%p)\n", xprt);
-	xprt->xpt_ops->xpo_detach(xprt);
-
-	spin_lock_bh(&serv->sv_lock);
-	if (!test_and_set_bit(XPT_DETACHED, &xprt->xpt_flags))
-		list_del_init(&xprt->xpt_list);
-	/*
-	 * We used to delete the transport from whichever list
-	 * it's sk_xprt.xpt_ready node was on, but we don't actually
-	 * need to.  This is because the only time we're called
-	 * while still attached to a queue, the queue itself
-	 * is about to be destroyed (in svc_destroy).
-	 */
-	if (!test_and_set_bit(XPT_DEAD, &xprt->xpt_flags)) {
-		BUG_ON(atomic_read(&xprt->xpt_ref.refcount) < 2);
-		svc_xprt_put(xprt);
-		if (test_bit(XPT_TEMP, &xprt->xpt_flags))
-			serv->sv_tmpcnt--;
-	}
-	spin_unlock_bh(&serv->sv_lock);
-}
-
-static void svc_close_xprt(struct svc_xprt *xprt)
-{
-	set_bit(XPT_CLOSE, &xprt->xpt_flags);
-	if (test_and_set_bit(XPT_BUSY, &xprt->xpt_flags))
-		/* someone else will have to effect the close */
-		return;
-
-	svc_xprt_get(xprt);
-	svc_delete_xprt(xprt);
-	clear_bit(XPT_BUSY, &xprt->xpt_flags);
-	svc_xprt_put(xprt);
-}
-
-void svc_close_all(struct list_head *xprt_list)
-{
-	struct svc_xprt *xprt;
-	struct svc_xprt *tmp;
-
-	list_for_each_entry_safe(xprt, tmp, xprt_list, xpt_list) {
-		set_bit(XPT_CLOSE, &xprt->xpt_flags);
-		if (test_bit(XPT_BUSY, &xprt->xpt_flags)) {
-			/* Waiting to be processed, but no threads left,
-			 * So just remove it from the waiting list
-			 */
-			list_del_init(&xprt->xpt_ready);
-			clear_bit(XPT_BUSY, &xprt->xpt_flags);
-		}
-		svc_close_xprt(xprt);
-	}
-}
-
-/*
- * Handle defer and revisit of requests
- */
-
-static void svc_revisit(struct cache_deferred_req *dreq, int too_many)
-{
-	struct svc_deferred_req *dr = container_of(dreq, struct svc_deferred_req, handle);
-	struct svc_xprt *xprt = dr->xprt;
-
-	if (too_many) {
-		svc_xprt_put(xprt);
-		kfree(dr);
-		return;
-	}
-	dprintk("revisit queued\n");
-	dr->xprt = NULL;
-	spin_lock(&xprt->xpt_lock);
-	list_add(&dr->handle.recent, &xprt->xpt_deferred);
-	spin_unlock(&xprt->xpt_lock);
-	set_bit(XPT_DEFERRED, &xprt->xpt_flags);
-	svc_xprt_enqueue(xprt);
-	svc_xprt_put(xprt);
-}
-
-static struct cache_deferred_req *
-svc_defer(struct cache_req *req)
-{
-	struct svc_rqst *rqstp = container_of(req, struct svc_rqst, rq_chandle);
-	int size = sizeof(struct svc_deferred_req) + (rqstp->rq_arg.len);
-	struct svc_deferred_req *dr;
-
-	if (rqstp->rq_arg.page_len)
-		return NULL; /* if more than a page, give up FIXME */
-	if (rqstp->rq_deferred) {
-		dr = rqstp->rq_deferred;
-		rqstp->rq_deferred = NULL;
-	} else {
-		int skip  = rqstp->rq_arg.len - rqstp->rq_arg.head[0].iov_len;
-		/* FIXME maybe discard if size too large */
-		dr = kmalloc(size, GFP_KERNEL);
-		if (dr == NULL)
-			return NULL;
-
-		dr->handle.owner = rqstp->rq_server;
-		dr->prot = rqstp->rq_prot;
-		memcpy(&dr->addr, &rqstp->rq_addr, rqstp->rq_addrlen);
-		dr->addrlen = rqstp->rq_addrlen;
-		dr->daddr = rqstp->rq_daddr;
-		dr->argslen = rqstp->rq_arg.len >> 2;
-		memcpy(dr->args, rqstp->rq_arg.head[0].iov_base-skip, dr->argslen<<2);
-	}
-	svc_xprt_get(rqstp->rq_xprt);
-	dr->xprt = rqstp->rq_xprt;
-
-	dr->handle.revisit = svc_revisit;
-	return &dr->handle;
-}
-
-/*
- * recv data from a deferred request into an active one
- */
-static int svc_deferred_recv(struct svc_rqst *rqstp)
-{
-	struct svc_deferred_req *dr = rqstp->rq_deferred;
-
-	rqstp->rq_arg.head[0].iov_base = dr->args;
-	rqstp->rq_arg.head[0].iov_len = dr->argslen<<2;
-	rqstp->rq_arg.page_len = 0;
-	rqstp->rq_arg.len = dr->argslen<<2;
-	rqstp->rq_prot        = dr->prot;
-	memcpy(&rqstp->rq_addr, &dr->addr, dr->addrlen);
-	rqstp->rq_addrlen     = dr->addrlen;
-	rqstp->rq_daddr       = dr->daddr;
-	rqstp->rq_respages    = rqstp->rq_pages;
-	return dr->argslen<<2;
-}
-
-
-static struct svc_deferred_req *svc_deferred_dequeue(struct svc_xprt *xprt)
-{
-	struct svc_deferred_req *dr = NULL;
-
-	if (!test_bit(XPT_DEFERRED, &xprt->xpt_flags))
-		return NULL;
-	spin_lock(&xprt->xpt_lock);
-	clear_bit(XPT_DEFERRED, &xprt->xpt_flags);
-	if (!list_empty(&xprt->xpt_deferred)) {
-		dr = list_entry(xprt->xpt_deferred.next,
-				struct svc_deferred_req,
-				handle.recent);
-		list_del_init(&dr->handle.recent);
-		set_bit(XPT_DEFERRED, &xprt->xpt_flags);
-	}
-	spin_unlock(&xprt->xpt_lock);
-	return dr;
-}

[PATCH 34/38] svc: Add transport hdr size for defer/revisit

[Tom Tucker]

Some transports have a header in front of the RPC header. The current
defer/revisit processing considers only the iov_len and arg_len to 
determine how much to back up when saving the original request
to revisit. Add a field to the rqstp structure to save the size
of the transport header so svc_defer can correctly compute
the start of a request. 

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 include/linux/sunrpc/svc.h |    2 ++
 net/sunrpc/svc_xprt.c      |   36 +++++++++++++++++++++++++++---------
 net/sunrpc/svcsock.c       |    2 ++
 3 files changed, 31 insertions(+), 9 deletions(-)

diff --git a/include/linux/sunrpc/svc.h b/include/linux/sunrpc/svc.h
index 04eb20e..742ab46 100644
--- a/include/linux/sunrpc/svc.h
+++ b/include/linux/sunrpc/svc.h
@@ -217,6 +217,7 @@ struct svc_rqst {
 	void *			rq_xprt_ctxt;	/* transport specific context ptr */
 	struct svc_deferred_req*rq_deferred;	/* deferred request we are replaying */
 
+	size_t			rq_xprt_hlen;	/* xprt header len */
 	struct xdr_buf		rq_arg;
 	struct xdr_buf		rq_res;
 	struct page *		rq_pages[RPCSVC_MAXPAGES];
@@ -322,6 +323,7 @@ struct svc_deferred_req {
 	size_t			addrlen;
 	union svc_addr_u	daddr;	/* where reply must come from */
 	struct cache_deferred_req handle;
+	size_t			xprt_hlen;
 	int			argslen;
 	__be32			args[0];
 };
diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
index a477883..fdaac77 100644
--- a/net/sunrpc/svc_xprt.c
+++ b/net/sunrpc/svc_xprt.c
@@ -29,7 +29,6 @@
 #include <linux/sunrpc/types.h>
 #include <linux/sunrpc/clnt.h>
 #include <linux/sunrpc/xdr.h>
-#include <linux/sunrpc/svcsock.h>
 #include <linux/sunrpc/stats.h>
 #include <linux/sunrpc/svc_xprt.h>
 
@@ -844,10 +843,18 @@ static void svc_revisit(struct cache_deferred_req *dreq, int too_many)
 	svc_xprt_put(xprt);
 }
 
+/*
+ * Save the request off for later processing. The request buffer looks
+ * like this:
+ *
+ * <xprt-header><rpc-header><rpc-pagelist><rpc-tail>
+ *
+ * This code can only handle requests that consist of an xprt-header
+ * and rpc-header.
+ */
 static struct cache_deferred_req *svc_defer(struct cache_req *req)
 {
 	struct svc_rqst *rqstp = container_of(req, struct svc_rqst, rq_chandle);
-	int size = sizeof(struct svc_deferred_req) + (rqstp->rq_arg.len);
 	struct svc_deferred_req *dr;
 
 	if (rqstp->rq_arg.page_len)
@@ -856,8 +863,10 @@ static struct cache_deferred_req *svc_defer(struct cache_req *req)
 		dr = rqstp->rq_deferred;
 		rqstp->rq_deferred = NULL;
 	} else {
-		int skip  = rqstp->rq_arg.len - rqstp->rq_arg.head[0].iov_len;
+		size_t skip;
+		size_t size;
 		/* FIXME maybe discard if size too large */
+		size = sizeof(struct svc_deferred_req) + rqstp->rq_arg.len;
 		dr = kmalloc(size, GFP_KERNEL);
 		if (dr == NULL)
 			return NULL;
@@ -868,8 +877,12 @@ static struct cache_deferred_req *svc_defer(struct cache_req *req)
 		dr->addrlen = rqstp->rq_addrlen;
 		dr->daddr = rqstp->rq_daddr;
 		dr->argslen = rqstp->rq_arg.len >> 2;
-		memcpy(dr->args, rqstp->rq_arg.head[0].iov_base-skip,
-		       dr->argslen<<2);
+		dr->xprt_hlen = rqstp->rq_xprt_hlen;
+
+		/* back up head to the start of the buffer and copy */
+		skip = rqstp->rq_arg.len - rqstp->rq_arg.head[0].iov_len;
+		memcpy(dr->args, rqstp->rq_arg.head[0].iov_base - skip,
+		       dr->argslen << 2);
 	}
 	svc_xprt_get(rqstp->rq_xprt);
 	dr->xprt = rqstp->rq_xprt;
@@ -885,16 +898,21 @@ static int svc_deferred_recv(struct svc_rqst *rqstp)
 {
 	struct svc_deferred_req *dr = rqstp->rq_deferred;
 
-	rqstp->rq_arg.head[0].iov_base = dr->args;
-	rqstp->rq_arg.head[0].iov_len = dr->argslen<<2;
+	/* setup iov_base past transport header */
+	rqstp->rq_arg.head[0].iov_base = dr->args + (dr->xprt_hlen>>2);
+	/* The iov_len does not include the transport header bytes */
+	rqstp->rq_arg.head[0].iov_len = (dr->argslen<<2) - dr->xprt_hlen;
 	rqstp->rq_arg.page_len = 0;
-	rqstp->rq_arg.len = dr->argslen<<2;
+	/* The rq_arg.len includes the transport header bytes */
+	rqstp->rq_arg.len     = dr->argslen<<2;
 	rqstp->rq_prot        = dr->prot;
 	memcpy(&rqstp->rq_addr, &dr->addr, dr->addrlen);
 	rqstp->rq_addrlen     = dr->addrlen;
+	/* Save off transport header len in case we get deferred again */
+	rqstp->rq_xprt_hlen   = dr->xprt_hlen;
 	rqstp->rq_daddr       = dr->daddr;
 	rqstp->rq_respages    = rqstp->rq_pages;
-	return dr->argslen<<2;
+	return (dr->argslen<<2) - dr->xprt_hlen;
 }
 
 
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index 0ab68f1..ac9eea9 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -343,6 +343,8 @@ static int svc_recvfrom(struct svc_rqst *rqstp, struct kvec *iov, int nr,
 	};
 	int len;
 
+	rqstp->rq_xprt_hlen = 0;
+
 	len = kernel_recvmsg(svsk->sk_sock, &msg, iov, nr, buflen,
 				msg.msg_flags);
 

Re: [PATCH 34/38] svc: Add transport hdr size for defer/revisit

[J. Bruce Fields]

On Tue, Dec 11, 2007 at 05:33:09PM -0600, Tom Tucker wrote:
> 
> Some transports have a header in front of the RPC header. The current
> defer/revisit processing considers only the iov_len and arg_len to 
> determine how much to back up when saving the original request
> to revisit. Add a field to the rqstp structure to save the size
> of the transport header so svc_defer can correctly compute
> the start of a request. 

I know I asked before, and can't remember what happened: has this been
tested with krb5p?  (I know nobody cares much whether krb5p/rdma works,
I just want to make sure krb5p/rdma doesn't oops, and that
krb5p/{tcp,udp} are unchanged.)

I'll try it eventually if you don't....

--b.

Re: [PATCH 34/38] svc: Add transport hdr size for defer/revisit

[Tom Tucker]

On 12/14/07 5:37 PM, "J. Bruce Fields" <bfields@fieldses.org> wrote:

> On Tue, Dec 11, 2007 at 05:33:09PM -0600, Tom Tucker wrote:
>> 
>> Some transports have a header in front of the RPC header. The current
>> defer/revisit processing considers only the iov_len and arg_len to
>> determine how much to back up when saving the original request
>> to revisit. Add a field to the rqstp structure to save the size
>> of the transport header so svc_defer can correctly compute
>> the start of a request.
> 
> I know I asked before, and can't remember what happened: has this been
> tested with krb5p?  (I know nobody cares much whether krb5p/rdma works,
> I just want to make sure krb5p/rdma doesn't oops, and that
> krb5p/{tcp,udp} are unchanged.)

I haven't tested this combination. I definitely don't have a kerberos setup,
but I guess I need to sooner or later...

> 
> I'll try it eventually if you don't....
> 
> --b.

[PATCH 35/38] svc: Add /proc/sys/sunrpc/transport files

[Tom Tucker]

Add a file that when read lists the set of registered svc
transports.

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 include/linux/sunrpc/svc_xprt.h |    1 +
 net/sunrpc/svc_xprt.c           |   28 ++++++++++++++++++++++++++++
 net/sunrpc/sysctl.c             |   31 +++++++++++++++++++++++++++++++
 3 files changed, 60 insertions(+), 0 deletions(-)

diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
index db8bcf5..e60e8f8 100644
--- a/include/linux/sunrpc/svc_xprt.h
+++ b/include/linux/sunrpc/svc_xprt.h
@@ -79,6 +79,7 @@ void	svc_xprt_copy_addrs(struct svc_rqst *rqstp, struct svc_xprt *xprt);
 void	svc_close_xprt(struct svc_xprt *xprt);
 void	svc_delete_xprt(struct svc_xprt *xprt);
 int	svc_port_is_privileged(struct sockaddr *sin);
+int	svc_print_xprts(char *buf, int maxlen);
 static inline void svc_xprt_get(struct svc_xprt *xprt)
 {
 	kref_get(&xprt->xpt_ref);
diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
index fdaac77..dc3a3fc 100644
--- a/net/sunrpc/svc_xprt.c
+++ b/net/sunrpc/svc_xprt.c
@@ -93,6 +93,34 @@ int svc_unreg_xprt_class(struct svc_xprt_class *xcl)
 }
 EXPORT_SYMBOL_GPL(svc_unreg_xprt_class);
 
+/*
+ * Format the transport list for printing
+ */
+int svc_print_xprts(char *buf, int maxlen)
+{
+	struct list_head *le;
+	char tmpstr[80];
+	int len = 0;
+	buf[0] = '\0';
+
+	spin_lock(&svc_xprt_class_lock);
+	list_for_each(le, &svc_xprt_class_list) {
+		int slen;
+		struct svc_xprt_class *xcl =
+			list_entry(le, struct svc_xprt_class, xcl_list);
+
+		sprintf(tmpstr, "%s %d\n", xcl->xcl_name, xcl->xcl_max_payload);
+		slen = strlen(tmpstr);
+		if (len + slen > maxlen)
+			break;
+		len += slen;
+		strcat(buf, tmpstr);
+	}
+	spin_unlock(&svc_xprt_class_lock);
+
+	return len;
+}
+
 static void svc_xprt_free(struct kref *kref)
 {
 	struct svc_xprt *xprt =
diff --git a/net/sunrpc/sysctl.c b/net/sunrpc/sysctl.c
index 2be714e..fd7cf59 100644
--- a/net/sunrpc/sysctl.c
+++ b/net/sunrpc/sysctl.c
@@ -18,6 +18,7 @@
 #include <linux/sunrpc/types.h>
 #include <linux/sunrpc/sched.h>
 #include <linux/sunrpc/stats.h>
+#include <linux/sunrpc/svc_xprt.h>
 
 /*
  * Declare the debug flags here
@@ -48,6 +49,30 @@ rpc_unregister_sysctl(void)
 	}
 }
 
+static int proc_do_xprt(ctl_table *table, int write, struct file *file,
+			void __user *buffer, size_t *lenp, loff_t *ppos)
+{
+	char tmpbuf[256];
+	int len;
+	if ((*ppos && !write) || !*lenp) {
+		*lenp = 0;
+		return 0;
+	}
+	if (write)
+		return -EINVAL;
+	else {
+		len = svc_print_xprts(tmpbuf, sizeof(tmpbuf));
+		if (!access_ok(VERIFY_WRITE, buffer, len))
+			return -EFAULT;
+
+		if (__copy_to_user(buffer, tmpbuf, len))
+			return -EFAULT;
+	}
+	*lenp -= len;
+	*ppos += len;
+	return 0;
+}
+
 static int
 proc_dodebug(ctl_table *table, int write, struct file *file,
 				void __user *buffer, size_t *lenp, loff_t *ppos)
@@ -140,6 +165,12 @@ static ctl_table debug_table[] = {
 		.mode		= 0644,
 		.proc_handler	= &proc_dodebug
 	},
+	{
+		.procname	= "transports",
+		.maxlen		= 256,
+		.mode		= 0444,
+		.proc_handler	= &proc_do_xprt,
+	},
 	{ .ctl_name = 0 }
 };
 

[PATCH 36/38] svc: Add svc API that queries for a transport instance

[Tom Tucker]

Add a new svc function that allows a service to query whether a 
transport instance has already been created. This is used in lockd 
to determine whether or not a transport needs to be created when
a lockd instance is brought up. 

Specifying 0 for the address family or port is effectively a wild-card,
and will result in matching the first transport in the service's list
that has a matching class name.

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 fs/lockd/svc.c                  |   16 ++--------------
 include/linux/sunrpc/svc_xprt.h |    1 +
 net/sunrpc/svc_xprt.c           |   34 ++++++++++++++++++++++++++++++++++
 3 files changed, 37 insertions(+), 14 deletions(-)

diff --git a/fs/lockd/svc.c b/fs/lockd/svc.c
index a8e79a9..470af01 100644
--- a/fs/lockd/svc.c
+++ b/fs/lockd/svc.c
@@ -219,18 +219,6 @@ lockd(struct svc_rqst *rqstp)
 	module_put_and_exit(0);
 }
 
-static int find_xprt(struct svc_serv *serv, char *proto)
-{
-	struct svc_xprt *xprt;
-	int found = 0;
-	list_for_each_entry(xprt, &serv->sv_permsocks, xpt_list)
-		if (strcmp(xprt->xpt_class->xcl_name, proto) == 0) {
-			found = 1;
-			break;
-		}
-	return found;
-}
-
 /*
  * Make any sockets that are needed but not present.
  * If nlm_udpport or nlm_tcpport were set as module
@@ -242,11 +230,11 @@ static int make_socks(struct svc_serv *serv, int proto)
 	int err = 0;
 
 	if (proto == IPPROTO_UDP || nlm_udpport)
-		if (!find_xprt(serv, "udp"))
+		if (!svc_find_xprt(serv, "udp", 0, 0))
 			err = svc_create_xprt(serv, "udp", nlm_udpport,
 					      SVC_SOCK_DEFAULTS);
 	if (err >= 0 && (proto == IPPROTO_TCP || nlm_tcpport))
-		if (!find_xprt(serv, "tcp"))
+		if (!svc_find_xprt(serv, "tcp", 0, 0))
 			err = svc_create_xprt(serv, "tcp", nlm_tcpport,
 					      SVC_SOCK_DEFAULTS);
 
diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
index e60e8f8..840b5c4 100644
--- a/include/linux/sunrpc/svc_xprt.h
+++ b/include/linux/sunrpc/svc_xprt.h
@@ -80,6 +80,7 @@ void	svc_close_xprt(struct svc_xprt *xprt);
 void	svc_delete_xprt(struct svc_xprt *xprt);
 int	svc_port_is_privileged(struct sockaddr *sin);
 int	svc_print_xprts(char *buf, int maxlen);
+struct	svc_xprt *svc_find_xprt(struct svc_serv *, char *, int, int);
 static inline void svc_xprt_get(struct svc_xprt *xprt)
 {
 	kref_get(&xprt->xpt_ref);
diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
index dc3a3fc..9815c6f 100644
--- a/net/sunrpc/svc_xprt.c
+++ b/net/sunrpc/svc_xprt.c
@@ -963,3 +963,37 @@ static struct svc_deferred_req *svc_deferred_dequeue(struct svc_xprt *xprt)
 	return dr;
 }
 
+/*
+ * Return the transport instance pointer for the endpoint accepting
+ * connections/peer traffic from the specified transport class,
+ * address family and port.
+ *
+ * Specifying 0 for the address family or port is effectively a
+ * wild-card, and will result in matching the first transport in the
+ * service's list that has a matching class name.
+ */
+struct svc_xprt *svc_find_xprt(struct svc_serv *serv, char *xcl_name,
+			       int af, int port)
+{
+	struct svc_xprt *xprt;
+	struct svc_xprt *found = NULL;
+
+	/* Sanity check the args */
+	if (!serv || !xcl_name)
+		return found;
+
+	spin_lock_bh(&serv->sv_lock);
+	list_for_each_entry(xprt, &serv->sv_permsocks, xpt_list) {
+		if (strcmp(xprt->xpt_class->xcl_name, xcl_name))
+			continue;
+		if (af != AF_UNSPEC && af != xprt->xpt_local.ss_family)
+			continue;
+		if (port && port != svc_xprt_local_port(xprt))
+			continue;
+		found = xprt;
+		break;
+	}
+	spin_unlock_bh(&serv->sv_lock);
+	return found;
+}
+EXPORT_SYMBOL_GPL(svc_find_xprt);

[PATCH 37/38] knfsd: Support adding transports by writing portlist file

[Tom Tucker]

Update the write handler for the portlist file to allow creating new
listening endpoints on a transport. The general form of the string is:

<transport_name><space><port number>

For example:

echo "tcp 2049" > /proc/fs/nfsd/portlist

This is intended to support the creation of a listening endpoint for
RDMA transports without adding #ifdef code to the nfssvc.c file.

Transports can also be removed as follows:

'-'<transport_name><space><port number>

For example:

echo "-tcp 2049" > /proc/fs/nfsd/portlist

Attempting to add a listener with an invalid transport string results
in EPROTONOSUPPORT and a perror string of "Protocol not supported". 

Attempting to remove an non-existent listener (.e.g. bad proto or port)
results in ENOTCONN and a perror string of 
"Transport endpoint is not connected"

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 fs/nfsd/nfsctl.c      |   52 ++++++++++++++++++++++++++++++++++++++++++++++++-
 net/sunrpc/svc_xprt.c |    1 +
 2 files changed, 52 insertions(+), 1 deletions(-)

diff --git a/fs/nfsd/nfsctl.c b/fs/nfsd/nfsctl.c
index 77dc989..5b9ed0e 100644
--- a/fs/nfsd/nfsctl.c
+++ b/fs/nfsd/nfsctl.c
@@ -540,7 +540,7 @@ static ssize_t write_ports(struct file *file, char *buf, size_t size)
 		}
 		return err < 0 ? err : 0;
 	}
-	if (buf[0] == '-') {
+	if (buf[0] == '-' && isdigit(buf[1])) {
 		char *toclose = kstrdup(buf+1, GFP_KERNEL);
 		int len = 0;
 		if (!toclose)
@@ -554,6 +554,56 @@ static ssize_t write_ports(struct file *file, char *buf, size_t size)
 		kfree(toclose);
 		return len;
 	}
+	/*
+	 * Add a transport listener by writing it's transport name
+	 */
+	if (isalpha(buf[0])) {
+		int err;
+		char transport[16];
+		int port;
+		if (sscanf(buf, "%15s %4d", transport, &port) == 2) {
+			err = nfsd_create_serv();
+			if (!err) {
+				if (svc_find_xprt(nfsd_serv, transport,
+						  AF_UNSPEC, port))
+					return -EADDRINUSE;
+
+				err = svc_create_xprt(nfsd_serv,
+						      transport, port,
+						      SVC_SOCK_ANONYMOUS);
+				if (err == -ENOENT)
+					/* Give a reasonable perror msg for
+					 * bad transport string */
+					err = -EPROTONOSUPPORT;
+			}
+			return err < 0 ? err : 0;
+		}
+	}
+	/*
+	 * Remove a transport by writing it's transport name and port number
+	 */
+	if (buf[0] == '-' && isalpha(buf[1])) {
+		struct svc_xprt *xprt;
+		int err = -EINVAL;
+		char transport[16];
+		int port;
+		if (sscanf(&buf[1], "%15s %4d", transport, &port) == 2) {
+			if (port == 0)
+				return -EINVAL;
+			lock_kernel();
+			if (nfsd_serv) {
+				xprt = svc_find_xprt(nfsd_serv, transport,
+						     AF_UNSPEC, port);
+				if (xprt) {
+					svc_close_xprt(xprt);
+					err = 0;
+				} else
+					err = -ENOTCONN;
+			}
+			unlock_kernel();
+			return err < 0 ? err : 0;
+		}
+	}
 	return -EINVAL;
 }
 
diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
index 9815c6f..6708aa2 100644
--- a/net/sunrpc/svc_xprt.c
+++ b/net/sunrpc/svc_xprt.c
@@ -827,6 +827,7 @@ void svc_close_xprt(struct svc_xprt *xprt)
 	clear_bit(XPT_BUSY, &xprt->xpt_flags);
 	svc_xprt_put(xprt);
 }
+EXPORT_SYMBOL_GPL(svc_close_xprt);
 
 void svc_close_all(struct list_head *xprt_list)
 {

Re: [PATCH 37/38] knfsd: Support adding transports by writing portlist file

[J. Bruce Fields]

On Tue, Dec 11, 2007 at 05:33:16PM -0600, Tom Tucker wrote:
> 
> Update the write handler for the portlist file to allow creating new
> listening endpoints on a transport. The general form of the string is:
> 
> <transport_name><space><port number>
> 
> For example:
> 
> echo "tcp 2049" > /proc/fs/nfsd/portlist
> 
> This is intended to support the creation of a listening endpoint for
> RDMA transports without adding #ifdef code to the nfssvc.c file.
> 
> Transports can also be removed as follows:
> 
> '-'<transport_name><space><port number>
> 
> For example:
> 
> echo "-tcp 2049" > /proc/fs/nfsd/portlist
> 
> Attempting to add a listener with an invalid transport string results
> in EPROTONOSUPPORT and a perror string of "Protocol not supported". 
> 
> Attempting to remove an non-existent listener (.e.g. bad proto or port)
> results in ENOTCONN and a perror string of 
> "Transport endpoint is not connected"
> 
> Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
> ---
> 
>  fs/nfsd/nfsctl.c      |   52 ++++++++++++++++++++++++++++++++++++++++++++++++-
>  net/sunrpc/svc_xprt.c |    1 +
>  2 files changed, 52 insertions(+), 1 deletions(-)
> 
> diff --git a/fs/nfsd/nfsctl.c b/fs/nfsd/nfsctl.c
> index 77dc989..5b9ed0e 100644
> --- a/fs/nfsd/nfsctl.c
> +++ b/fs/nfsd/nfsctl.c
> @@ -540,7 +540,7 @@ static ssize_t write_ports(struct file *file, char *buf, size_t size)
>  		}
>  		return err < 0 ? err : 0;
>  	}
> -	if (buf[0] == '-') {
> +	if (buf[0] == '-' && isdigit(buf[1])) {
>  		char *toclose = kstrdup(buf+1, GFP_KERNEL);
>  		int len = 0;
>  		if (!toclose)
> @@ -554,6 +554,56 @@ static ssize_t write_ports(struct file *file, char *buf, size_t size)
>  		kfree(toclose);
>  		return len;
>  	}
> +	/*
> +	 * Add a transport listener by writing it's transport name
> +	 */
> +	if (isalpha(buf[0])) {
> +		int err;
> +		char transport[16];
> +		int port;
> +		if (sscanf(buf, "%15s %4d", transport, &port) == 2) {
> +			err = nfsd_create_serv();
> +			if (!err) {
> +				if (svc_find_xprt(nfsd_serv, transport,
> +						  AF_UNSPEC, port))
> +					return -EADDRINUSE;

Shouldn't svc_create_xprt do this check for us itself?  (And if it
doesn't, isn't there some minor race between two writers both checking,
finding nothing, and then both svc_create_xprt()ing?)

> +
> +				err = svc_create_xprt(nfsd_serv,
> +						      transport, port,
> +						      SVC_SOCK_ANONYMOUS);
> +				if (err == -ENOENT)
> +					/* Give a reasonable perror msg for
> +					 * bad transport string */
> +					err = -EPROTONOSUPPORT;
> +			}
> +			return err < 0 ? err : 0;
> +		}
> +	}
> +	/*
> +	 * Remove a transport by writing it's transport name and port number
> +	 */
> +	if (buf[0] == '-' && isalpha(buf[1])) {
> +		struct svc_xprt *xprt;
> +		int err = -EINVAL;
> +		char transport[16];
> +		int port;
> +		if (sscanf(&buf[1], "%15s %4d", transport, &port) == 2) {
> +			if (port == 0)
> +				return -EINVAL;
> +			lock_kernel();
> +			if (nfsd_serv) {
> +				xprt = svc_find_xprt(nfsd_serv, transport,
> +						     AF_UNSPEC, port);
> +				if (xprt) {
> +					svc_close_xprt(xprt);
> +					err = 0;

Hm, also: I'd have thought that svc_find_xprt should be incrementing the
reference count on the returned xprt, if we're expecting the caller to
do anything with it other than check it against NULL.

--b.


> +				} else
> +					err = -ENOTCONN;
> +			}
> +			unlock_kernel();
> +			return err < 0 ? err : 0;
> +		}
> +	}
>  	return -EINVAL;
>  }
>  
> diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
> index 9815c6f..6708aa2 100644
> --- a/net/sunrpc/svc_xprt.c
> +++ b/net/sunrpc/svc_xprt.c
> @@ -827,6 +827,7 @@ void svc_close_xprt(struct svc_xprt *xprt)
>  	clear_bit(XPT_BUSY, &xprt->xpt_flags);
>  	svc_xprt_put(xprt);
>  }
> +EXPORT_SYMBOL_GPL(svc_close_xprt);
>  
>  void svc_close_all(struct list_head *xprt_list)
>  {

Re: [PATCH 37/38] knfsd: Support adding transports by writing portlist file

[Tom Tucker]

On Fri, 2007-12-14 at 18:51 -0500, J. Bruce Fields wrote:
> On Tue, Dec 11, 2007 at 05:33:16PM -0600, Tom Tucker wrote:
> > 
> > Update the write handler for the portlist file to allow creating new
> > listening endpoints on a transport. The general form of the string is:
> > 
> > <transport_name><space><port number>
> > 
> > For example:
> > 
> > echo "tcp 2049" > /proc/fs/nfsd/portlist
> > 
> > This is intended to support the creation of a listening endpoint for
> > RDMA transports without adding #ifdef code to the nfssvc.c file.
> > 
> > Transports can also be removed as follows:
> > 
> > '-'<transport_name><space><port number>
> > 
> > For example:
> > 
> > echo "-tcp 2049" > /proc/fs/nfsd/portlist
> > 
> > Attempting to add a listener with an invalid transport string results
> > in EPROTONOSUPPORT and a perror string of "Protocol not supported". 
> > 
> > Attempting to remove an non-existent listener (.e.g. bad proto or port)
> > results in ENOTCONN and a perror string of 
> > "Transport endpoint is not connected"
> > 
> > Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
> > ---
> > 
> >  fs/nfsd/nfsctl.c      |   52 ++++++++++++++++++++++++++++++++++++++++++++++++-
> >  net/sunrpc/svc_xprt.c |    1 +
> >  2 files changed, 52 insertions(+), 1 deletions(-)
> > 
> > diff --git a/fs/nfsd/nfsctl.c b/fs/nfsd/nfsctl.c
> > index 77dc989..5b9ed0e 100644
> > --- a/fs/nfsd/nfsctl.c
> > +++ b/fs/nfsd/nfsctl.c
> > @@ -540,7 +540,7 @@ static ssize_t write_ports(struct file *file, char *buf, size_t size)
> >  		}
> >  		return err < 0 ? err : 0;
> >  	}
> > -	if (buf[0] == '-') {
> > +	if (buf[0] == '-' && isdigit(buf[1])) {
> >  		char *toclose = kstrdup(buf+1, GFP_KERNEL);
> >  		int len = 0;
> >  		if (!toclose)
> > @@ -554,6 +554,56 @@ static ssize_t write_ports(struct file *file, char *buf, size_t size)
> >  		kfree(toclose);
> >  		return len;
> >  	}
> > +	/*
> > +	 * Add a transport listener by writing it's transport name
> > +	 */
> > +	if (isalpha(buf[0])) {
> > +		int err;
> > +		char transport[16];
> > +		int port;
> > +		if (sscanf(buf, "%15s %4d", transport, &port) == 2) {
> > +			err = nfsd_create_serv();
> > +			if (!err) {
> > +				if (svc_find_xprt(nfsd_serv, transport,
> > +						  AF_UNSPEC, port))
> > +					return -EADDRINUSE;
> 
> Shouldn't svc_create_xprt do this check for us itself?  (And if it
> doesn't, isn't there some minor race between two writers both checking,
> finding nothing, and then both svc_create_xprt()ing?)
> 

You're right. fixed.

> > +
> > +				err = svc_create_xprt(nfsd_serv,
> > +						      transport, port,
> > +						      SVC_SOCK_ANONYMOUS);
> > +				if (err == -ENOENT)
> > +					/* Give a reasonable perror msg for
> > +					 * bad transport string */
> > +					err = -EPROTONOSUPPORT;
> > +			}
> > +			return err < 0 ? err : 0;
> > +		}
> > +	}
> > +	/*
> > +	 * Remove a transport by writing it's transport name and port number
> > +	 */
> > +	if (buf[0] == '-' && isalpha(buf[1])) {
> > +		struct svc_xprt *xprt;
> > +		int err = -EINVAL;
> > +		char transport[16];
> > +		int port;
> > +		if (sscanf(&buf[1], "%15s %4d", transport, &port) == 2) {
> > +			if (port == 0)
> > +				return -EINVAL;
> > +			lock_kernel();
> > +			if (nfsd_serv) {
> > +				xprt = svc_find_xprt(nfsd_serv, transport,
> > +						     AF_UNSPEC, port);
> > +				if (xprt) {
> > +					svc_close_xprt(xprt);
> > +					err = 0;
> 
> Hm, also: I'd have thought that svc_find_xprt should be incrementing the
> reference count on the returned xprt, if we're expecting the caller to
> do anything with it other than check it against NULL.
> 
> --b.
> 
> 
> > +				} else
> > +					err = -ENOTCONN;
> > +			}
> > +			unlock_kernel();
> > +			return err < 0 ? err : 0;
> > +		}
> > +	}
> >  	return -EINVAL;
> >  }
> >  
> > diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
> > index 9815c6f..6708aa2 100644
> > --- a/net/sunrpc/svc_xprt.c
> > +++ b/net/sunrpc/svc_xprt.c
> > @@ -827,6 +827,7 @@ void svc_close_xprt(struct svc_xprt *xprt)
> >  	clear_bit(XPT_BUSY, &xprt->xpt_flags);
> >  	svc_xprt_put(xprt);
> >  }
> > +EXPORT_SYMBOL_GPL(svc_close_xprt);
> >  
> >  void svc_close_all(struct list_head *xprt_list)
> >  {
> -
> To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 37/38] knfsd: Support adding transports by writing portlist file

[J. Bruce Fields]

On Fri, Dec 21, 2007 at 11:52:30AM -0600, Tom Tucker wrote:
> 
> On Fri, 2007-12-14 at 18:51 -0500, J. Bruce Fields wrote:
> > On Tue, Dec 11, 2007 at 05:33:16PM -0600, Tom Tucker wrote:
> > > 
> > > Update the write handler for the portlist file to allow creating new
> > > listening endpoints on a transport. The general form of the string is:
> > > 
> > > <transport_name><space><port number>
> > > 
> > > For example:
> > > 
> > > echo "tcp 2049" > /proc/fs/nfsd/portlist
> > > 
> > > This is intended to support the creation of a listening endpoint for
> > > RDMA transports without adding #ifdef code to the nfssvc.c file.
> > > 
> > > Transports can also be removed as follows:
> > > 
> > > '-'<transport_name><space><port number>
> > > 
> > > For example:
> > > 
> > > echo "-tcp 2049" > /proc/fs/nfsd/portlist
> > > 
> > > Attempting to add a listener with an invalid transport string results
> > > in EPROTONOSUPPORT and a perror string of "Protocol not supported". 
> > > 
> > > Attempting to remove an non-existent listener (.e.g. bad proto or port)
> > > results in ENOTCONN and a perror string of 
> > > "Transport endpoint is not connected"
> > > 
> > > Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
> > > ---
> > > 
> > >  fs/nfsd/nfsctl.c      |   52 ++++++++++++++++++++++++++++++++++++++++++++++++-
> > >  net/sunrpc/svc_xprt.c |    1 +
> > >  2 files changed, 52 insertions(+), 1 deletions(-)
> > > 
> > > diff --git a/fs/nfsd/nfsctl.c b/fs/nfsd/nfsctl.c
> > > index 77dc989..5b9ed0e 100644
> > > --- a/fs/nfsd/nfsctl.c
> > > +++ b/fs/nfsd/nfsctl.c
> > > @@ -540,7 +540,7 @@ static ssize_t write_ports(struct file *file, char *buf, size_t size)
> > >  		}
> > >  		return err < 0 ? err : 0;
> > >  	}
> > > -	if (buf[0] == '-') {
> > > +	if (buf[0] == '-' && isdigit(buf[1])) {
> > >  		char *toclose = kstrdup(buf+1, GFP_KERNEL);
> > >  		int len = 0;
> > >  		if (!toclose)
> > > @@ -554,6 +554,56 @@ static ssize_t write_ports(struct file *file, char *buf, size_t size)
> > >  		kfree(toclose);
> > >  		return len;
> > >  	}
> > > +	/*
> > > +	 * Add a transport listener by writing it's transport name
> > > +	 */
> > > +	if (isalpha(buf[0])) {
> > > +		int err;
> > > +		char transport[16];
> > > +		int port;
> > > +		if (sscanf(buf, "%15s %4d", transport, &port) == 2) {
> > > +			err = nfsd_create_serv();
> > > +			if (!err) {
> > > +				if (svc_find_xprt(nfsd_serv, transport,
> > > +						  AF_UNSPEC, port))
> > > +					return -EADDRINUSE;
> > 
> > Shouldn't svc_create_xprt do this check for us itself?  (And if it
> > doesn't, isn't there some minor race between two writers both checking,
> > finding nothing, and then both svc_create_xprt()ing?)
> > 
> 
> You're right. fixed.

Thanks!  But:

> 
> > > +
> > > +				err = svc_create_xprt(nfsd_serv,
> > > +						      transport, port,
> > > +						      SVC_SOCK_ANONYMOUS);
> > > +				if (err == -ENOENT)
> > > +					/* Give a reasonable perror msg for
> > > +					 * bad transport string */
> > > +					err = -EPROTONOSUPPORT;
> > > +			}
> > > +			return err < 0 ? err : 0;
> > > +		}
> > > +	}
> > > +	/*
> > > +	 * Remove a transport by writing it's transport name and port number
> > > +	 */
> > > +	if (buf[0] == '-' && isalpha(buf[1])) {
> > > +		struct svc_xprt *xprt;
> > > +		int err = -EINVAL;
> > > +		char transport[16];
> > > +		int port;
> > > +		if (sscanf(&buf[1], "%15s %4d", transport, &port) == 2) {
> > > +			if (port == 0)
> > > +				return -EINVAL;
> > > +			lock_kernel();
> > > +			if (nfsd_serv) {
> > > +				xprt = svc_find_xprt(nfsd_serv, transport,
> > > +						     AF_UNSPEC, port);
> > > +				if (xprt) {
> > > +					svc_close_xprt(xprt);
> > > +					err = 0;
> > 
> > Hm, also: I'd have thought that svc_find_xprt should be incrementing the
> > reference count on the returned xprt, if we're expecting the caller to
> > do anything with it other than check it against NULL.

... there's still a race here, isn't there?

--b.

Re: [PATCH 37/38] knfsd: Support adding transports by writing portlist file

[Tom Tucker]

On Thu, 2008-01-03 at 16:13 -0500, J. Bruce Fields wrote:
> On Fri, Dec 21, 2007 at 11:52:30AM -0600, Tom Tucker wrote:

[...snip...]

> > > > +						     AF_UNSPEC, port);
> > > > +				if (xprt) {
> > > > +					svc_close_xprt(xprt);
> > > > +					err = 0;
> > > 
> > > Hm, also: I'd have thought that svc_find_xprt should be incrementing the
> > > reference count on the returned xprt, if we're expecting the caller to
> > > do anything with it other than check it against NULL.
> 
> ... there's still a race here, isn't there?
> 

I think it's a good idea to have the svc_find_xprt service take a
reference. I'll code this up....

> --b.

[PATCH 38/38] svc: Add svc_xprt_names service to replace svc_sock_names

[Tom Tucker]

Create a transport independent version of the svc_sock_names function.

The toclose capability of the svc_sock_names service can be implemented
using the svc_xprt_find and svc_xprt_close services.

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
---

 fs/nfsd/nfsctl.c                |    2 +-
 include/linux/sunrpc/svc_xprt.h |    1 +
 net/sunrpc/svc_xprt.c           |   35 +++++++++++++++++++++++++++++++++++
 3 files changed, 37 insertions(+), 1 deletions(-)

diff --git a/fs/nfsd/nfsctl.c b/fs/nfsd/nfsctl.c
index 5b9ed0e..86d084e 100644
--- a/fs/nfsd/nfsctl.c
+++ b/fs/nfsd/nfsctl.c
@@ -503,7 +503,7 @@ static ssize_t write_ports(struct file *file, char *buf, size_t size)
 		int len = 0;
 		lock_kernel();
 		if (nfsd_serv)
-			len = svc_sock_names(buf, nfsd_serv, NULL);
+			len = svc_xprt_names(nfsd_serv, buf, 0);
 		unlock_kernel();
 		return len;
 	}
diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
index 840b5c4..ed9791a 100644
--- a/include/linux/sunrpc/svc_xprt.h
+++ b/include/linux/sunrpc/svc_xprt.h
@@ -81,6 +81,7 @@ void	svc_delete_xprt(struct svc_xprt *xprt);
 int	svc_port_is_privileged(struct sockaddr *sin);
 int	svc_print_xprts(char *buf, int maxlen);
 struct	svc_xprt *svc_find_xprt(struct svc_serv *, char *, int, int);
+int	svc_xprt_names(struct svc_serv *serv, char *buf, int buflen);
 static inline void svc_xprt_get(struct svc_xprt *xprt)
 {
 	kref_get(&xprt->xpt_ref);
diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
index 6708aa2..2e3a672 100644
--- a/net/sunrpc/svc_xprt.c
+++ b/net/sunrpc/svc_xprt.c
@@ -998,3 +998,38 @@ struct svc_xprt *svc_find_xprt(struct svc_serv *serv, char *xcl_name,
 	return found;
 }
 EXPORT_SYMBOL_GPL(svc_find_xprt);
+
+/*
+ * Format a buffer with a list of the active transports. A zero for
+ * the buflen parameter disables target buffer overflow checking.
+ */
+int svc_xprt_names(struct svc_serv *serv, char *buf, int buflen)
+{
+	struct svc_xprt *xprt;
+	char xprt_str[64];
+	int totlen = 0;
+	int len;
+
+	/* Sanity check args */
+	if (!serv)
+		return 0;
+
+	spin_lock_bh(&serv->sv_lock);
+	list_for_each_entry(xprt, &serv->sv_permsocks, xpt_list) {
+		len = snprintf(xprt_str, sizeof(xprt_str),
+			       "%s %d\n", xprt->xpt_class->xcl_name,
+			       svc_xprt_local_port(xprt));
+		/* If the string was truncated, replace with error string */
+		if (len >= sizeof(xprt_str))
+			strcpy(xprt_str, "name-too-long\n");
+		/* Don't overflow buffer */
+		len = strlen(xprt_str);
+		if (buflen && (len + totlen >= buflen))
+			break;
+		strcpy(buf+totlen, xprt_str);
+		totlen += len;
+	}
+	spin_unlock_bh(&serv->sv_lock);
+	return totlen;
+}
+EXPORT_SYMBOL_GPL(svc_xprt_names);

Re: [PATCH 38/38] svc: Add svc_xprt_names service to replace svc_sock_names

[J. Bruce Fields]

On Tue, Dec 11, 2007 at 05:33:18PM -0600, Tom Tucker wrote:
> 
> Create a transport independent version of the svc_sock_names function.
> 
> The toclose capability of the svc_sock_names service can be implemented
> using the svc_xprt_find and svc_xprt_close services.

Should we delete the toclose checks from svc_sock_names(), then, under
the assumption it's always called with toclose non-NULL now?

And why can't we just completely replace svc_sock_names() at this point?

--b.

> 
> Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
> ---
> 
>  fs/nfsd/nfsctl.c                |    2 +-
>  include/linux/sunrpc/svc_xprt.h |    1 +
>  net/sunrpc/svc_xprt.c           |   35 +++++++++++++++++++++++++++++++++++
>  3 files changed, 37 insertions(+), 1 deletions(-)
> 
> diff --git a/fs/nfsd/nfsctl.c b/fs/nfsd/nfsctl.c
> index 5b9ed0e..86d084e 100644
> --- a/fs/nfsd/nfsctl.c
> +++ b/fs/nfsd/nfsctl.c
> @@ -503,7 +503,7 @@ static ssize_t write_ports(struct file *file, char *buf, size_t size)
>  		int len = 0;
>  		lock_kernel();
>  		if (nfsd_serv)
> -			len = svc_sock_names(buf, nfsd_serv, NULL);
> +			len = svc_xprt_names(nfsd_serv, buf, 0);
>  		unlock_kernel();
>  		return len;
>  	}
> diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
> index 840b5c4..ed9791a 100644
> --- a/include/linux/sunrpc/svc_xprt.h
> +++ b/include/linux/sunrpc/svc_xprt.h
> @@ -81,6 +81,7 @@ void	svc_delete_xprt(struct svc_xprt *xprt);
>  int	svc_port_is_privileged(struct sockaddr *sin);
>  int	svc_print_xprts(char *buf, int maxlen);
>  struct	svc_xprt *svc_find_xprt(struct svc_serv *, char *, int, int);
> +int	svc_xprt_names(struct svc_serv *serv, char *buf, int buflen);
>  static inline void svc_xprt_get(struct svc_xprt *xprt)
>  {
>  	kref_get(&xprt->xpt_ref);
> diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
> index 6708aa2..2e3a672 100644
> --- a/net/sunrpc/svc_xprt.c
> +++ b/net/sunrpc/svc_xprt.c
> @@ -998,3 +998,38 @@ struct svc_xprt *svc_find_xprt(struct svc_serv *serv, char *xcl_name,
>  	return found;
>  }
>  EXPORT_SYMBOL_GPL(svc_find_xprt);
> +
> +/*
> + * Format a buffer with a list of the active transports. A zero for
> + * the buflen parameter disables target buffer overflow checking.
> + */
> +int svc_xprt_names(struct svc_serv *serv, char *buf, int buflen)
> +{
> +	struct svc_xprt *xprt;
> +	char xprt_str[64];
> +	int totlen = 0;
> +	int len;
> +
> +	/* Sanity check args */
> +	if (!serv)
> +		return 0;
> +
> +	spin_lock_bh(&serv->sv_lock);
> +	list_for_each_entry(xprt, &serv->sv_permsocks, xpt_list) {
> +		len = snprintf(xprt_str, sizeof(xprt_str),
> +			       "%s %d\n", xprt->xpt_class->xcl_name,
> +			       svc_xprt_local_port(xprt));
> +		/* If the string was truncated, replace with error string */
> +		if (len >= sizeof(xprt_str))
> +			strcpy(xprt_str, "name-too-long\n");
> +		/* Don't overflow buffer */
> +		len = strlen(xprt_str);
> +		if (buflen && (len + totlen >= buflen))
> +			break;
> +		strcpy(buf+totlen, xprt_str);
> +		totlen += len;
> +	}
> +	spin_unlock_bh(&serv->sv_lock);
> +	return totlen;
> +}
> +EXPORT_SYMBOL_GPL(svc_xprt_names);

Re: [PATCH 38/38] svc: Add svc_xprt_names service to replace svc_sock_names

[Tom Tucker]

On Fri, 2007-12-14 at 19:03 -0500, J. Bruce Fields wrote:
> On Tue, Dec 11, 2007 at 05:33:18PM -0600, Tom Tucker wrote:
> > 
> > Create a transport independent version of the svc_sock_names function.
> > 
> > The toclose capability of the svc_sock_names service can be implemented
> > using the svc_xprt_find and svc_xprt_close services.
> 
> Should we delete the toclose checks from svc_sock_names(), then, under
> the assumption it's always called with toclose non-NULL now?
> 
> And why can't we just completely replace svc_sock_names() at this point?
> 

IMO we could, but there is currently a difference in behavior between
svc_sock_names and svc_find_xprt/close. svc_find_xprt doesn't care what
the IP address is, it only compares transport name, address family and
port. 

Presently in NFS, we only ever listen on zero, but you can destroy an
endpoint on a particular interface (IP address). Can anyone shed some
light on why this is? 

> --b.
> 
> > 
> > Signed-off-by: Tom Tucker <tom@opengridcomputing.com>
> > ---
> > 
> >  fs/nfsd/nfsctl.c                |    2 +-
> >  include/linux/sunrpc/svc_xprt.h |    1 +
> >  net/sunrpc/svc_xprt.c           |   35 +++++++++++++++++++++++++++++++++++
> >  3 files changed, 37 insertions(+), 1 deletions(-)
> > 
> > diff --git a/fs/nfsd/nfsctl.c b/fs/nfsd/nfsctl.c
> > index 5b9ed0e..86d084e 100644
> > --- a/fs/nfsd/nfsctl.c
> > +++ b/fs/nfsd/nfsctl.c
> > @@ -503,7 +503,7 @@ static ssize_t write_ports(struct file *file, char *buf, size_t size)
> >  		int len = 0;
> >  		lock_kernel();
> >  		if (nfsd_serv)
> > -			len = svc_sock_names(buf, nfsd_serv, NULL);
> > +			len = svc_xprt_names(nfsd_serv, buf, 0);
> >  		unlock_kernel();
> >  		return len;
> >  	}
> > diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
> > index 840b5c4..ed9791a 100644
> > --- a/include/linux/sunrpc/svc_xprt.h
> > +++ b/include/linux/sunrpc/svc_xprt.h
> > @@ -81,6 +81,7 @@ void	svc_delete_xprt(struct svc_xprt *xprt);
> >  int	svc_port_is_privileged(struct sockaddr *sin);
> >  int	svc_print_xprts(char *buf, int maxlen);
> >  struct	svc_xprt *svc_find_xprt(struct svc_serv *, char *, int, int);
> > +int	svc_xprt_names(struct svc_serv *serv, char *buf, int buflen);
> >  static inline void svc_xprt_get(struct svc_xprt *xprt)
> >  {
> >  	kref_get(&xprt->xpt_ref);
> > diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
> > index 6708aa2..2e3a672 100644
> > --- a/net/sunrpc/svc_xprt.c
> > +++ b/net/sunrpc/svc_xprt.c
> > @@ -998,3 +998,38 @@ struct svc_xprt *svc_find_xprt(struct svc_serv *serv, char *xcl_name,
> >  	return found;
> >  }
> >  EXPORT_SYMBOL_GPL(svc_find_xprt);
> > +
> > +/*
> > + * Format a buffer with a list of the active transports. A zero for
> > + * the buflen parameter disables target buffer overflow checking.
> > + */
> > +int svc_xprt_names(struct svc_serv *serv, char *buf, int buflen)
> > +{
> > +	struct svc_xprt *xprt;
> > +	char xprt_str[64];
> > +	int totlen = 0;
> > +	int len;
> > +
> > +	/* Sanity check args */
> > +	if (!serv)
> > +		return 0;
> > +
> > +	spin_lock_bh(&serv->sv_lock);
> > +	list_for_each_entry(xprt, &serv->sv_permsocks, xpt_list) {
> > +		len = snprintf(xprt_str, sizeof(xprt_str),
> > +			       "%s %d\n", xprt->xpt_class->xcl_name,
> > +			       svc_xprt_local_port(xprt));
> > +		/* If the string was truncated, replace with error string */
> > +		if (len >= sizeof(xprt_str))
> > +			strcpy(xprt_str, "name-too-long\n");
> > +		/* Don't overflow buffer */
> > +		len = strlen(xprt_str);
> > +		if (buflen && (len + totlen >= buflen))
> > +			break;
> > +		strcpy(buf+totlen, xprt_str);
> > +		totlen += len;
> > +	}
> > +	spin_unlock_bh(&serv->sv_lock);
> > +	return totlen;
> > +}
> > +EXPORT_SYMBOL_GPL(svc_xprt_names);
> -
> To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 38/38] svc: Add svc_xprt_names service to replace svc_sock_names

[J. Bruce Fields]

On Fri, Dec 21, 2007 at 11:51:47AM -0600, Tom Tucker wrote:
> 
> On Fri, 2007-12-14 at 19:03 -0500, J. Bruce Fields wrote:
> > On Tue, Dec 11, 2007 at 05:33:18PM -0600, Tom Tucker wrote:
> > > 
> > > Create a transport independent version of the svc_sock_names function.
> > > 
> > > The toclose capability of the svc_sock_names service can be implemented
> > > using the svc_xprt_find and svc_xprt_close services.
> > 
> > Should we delete the toclose checks from svc_sock_names(), then, under
> > the assumption it's always called with toclose non-NULL now?
> > 
> > And why can't we just completely replace svc_sock_names() at this point?
> > 
> 
> IMO we could, but there is currently a difference in behavior between
> svc_sock_names and svc_find_xprt/close. svc_find_xprt doesn't care what
> the IP address is, it only compares transport name, address family and
> port. 
> 
> Presently in NFS, we only ever listen on zero, but you can destroy an
> endpoint on a particular interface (IP address). Can anyone shed some
> light on why this is? 

Dunno.  Neil?

--b.

Re: [PATCH 38/38] svc: Add svc_xprt_names service to replace svc_sock_names

[Neil Brown]

On Thursday January 3, bfields@fieldses.org wrote:
> On Fri, Dec 21, 2007 at 11:51:47AM -0600, Tom Tucker wrote:
> > 
> > On Fri, 2007-12-14 at 19:03 -0500, J. Bruce Fields wrote:
> > > On Tue, Dec 11, 2007 at 05:33:18PM -0600, Tom Tucker wrote:
> > > > 
> > > > Create a transport independent version of the svc_sock_names function.
> > > > 
> > > > The toclose capability of the svc_sock_names service can be implemented
> > > > using the svc_xprt_find and svc_xprt_close services.
> > > 
> > > Should we delete the toclose checks from svc_sock_names(), then, under
> > > the assumption it's always called with toclose non-NULL now?
> > > 
> > > And why can't we just completely replace svc_sock_names() at this point?
> > > 
> > 
> > IMO we could, but there is currently a difference in behavior between
> > svc_sock_names and svc_find_xprt/close. svc_find_xprt doesn't care what
> > the IP address is, it only compares transport name, address family and
> > port. 
> > 
> > Presently in NFS, we only ever listen on zero, but you can destroy an
> > endpoint on a particular interface (IP address). Can anyone shed some
> > light on why this is? 
> 
> Dunno.  Neil?

The "-H" option to "rpc.nfsd" allow you to bind nfsd to a specific
interface.
I don't think any user-space tools current allow you to unbind a
specific interface, but I wouldn't want to lose that functionality.

Why are we changing this.  Clear the toclose functionality of the
svc_sock_names service can *NOT* be impletemented using svc_xprt_find
and svc_xprt_close as _find_ doesn't check the port - so just leave
things the way they are??? I haven't really been following this update
much so maybe there is something I'm missing....

NeilBrown

Re: [PATCH 38/38] svc: Add svc_xprt_names service to replace svc_sock_names

[Tom Tucker]

Sorry for the delay...

On 1/3/08 6:47 PM, "Neil Brown" <neilb@suse.de> wrote:

> On Thursday January 3, bfields@fieldses.org wrote:
>> On Fri, Dec 21, 2007 at 11:51:47AM -0600, Tom Tucker wrote:
>>> 
>>> On Fri, 2007-12-14 at 19:03 -0500, J. Bruce Fields wrote:
>>>> On Tue, Dec 11, 2007 at 05:33:18PM -0600, Tom Tucker wrote:
>>>>> 
>>>>> Create a transport independent version of the svc_sock_names function.
>>>>> 
>>>>> The toclose capability of the svc_sock_names service can be implemented
>>>>> using the svc_xprt_find and svc_xprt_close services.
>>>> 
>>>> Should we delete the toclose checks from svc_sock_names(), then, under
>>>> the assumption it's always called with toclose non-NULL now?
>>>> 
>>>> And why can't we just completely replace svc_sock_names() at this point?
>>>> 
>>> 
>>> IMO we could, but there is currently a difference in behavior between
>>> svc_sock_names and svc_find_xprt/close. svc_find_xprt doesn't care what
>>> the IP address is, it only compares transport name, address family and
>>> port. 
>>> 
>>> Presently in NFS, we only ever listen on zero, but you can destroy an
>>> endpoint on a particular interface (IP address). Can anyone shed some
>>> light on why this is?
>> 
>> Dunno.  Neil?
> 
> The "-H" option to "rpc.nfsd" allow you to bind nfsd to a specific
> interface.
> I don't think any user-space tools current allow you to unbind a
> specific interface, but I wouldn't want to lose that functionality.
> 
> Why are we changing this.  Clear the toclose functionality of the
> svc_sock_names service can *NOT* be impletemented using svc_xprt_find
> and svc_xprt_close as _find_ doesn't check the port - so just leave
> things the way they are???

I think you mean check the IP address ... Correct?

> I haven't really been following this update
> much so maybe there is something I'm missing....

I don't think you're missing anything, however, I think there are two pieces
to this: a) what should the new API support, and b) should we move the
existing code to the new API.

The new API (svc_xprt_find) does not currently search based on IP address
and therefore cannot be used for b). I think it should, however, since
svc_xprt_create can be used to create a listener on a specific IP
address/interface. 

 
> 
> NeilBrown
> -
> To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html