[PATCH] knfsd: Prepare knfsd for support of rsize/wsize of up to 1MB, over TCP

The limit over UDP remains at 32K.  Also, make some of the apparently
arbitrary sizing constants clearer.

The biggest change here involves replacing NFSSVC_MAXBLKSIZE by a function of
the rqstp.  This allows it to be different for different protocols (udp/tcp)
and also allows it to depend on the servers declared sv_bufsiz.

Note that we don't actually increase sv_bufsz for nfs yet.  That comes next.

Signed-off-by: Greg Banks <gnb@melbourne.sgi.com>
Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

11 files changed, 119 insertions, 34 deletions

diff --git a/fs/nfsd/nfs3proc.c b/fs/nfsd/nfs3proc.c
index 0c4fab018791e4..a5ebc7dbb3842d 100644
--- a/fs/nfsd/nfs3proc.c
+++ b/fs/nfsd/nfs3proc.c
@@ -160,6 +160,7 @@ nfsd3_proc_read(struct svc_rqst *rqstp, struct nfsd3_readargs *argp,
 				        struct nfsd3_readres  *resp)
 {
 	int	nfserr;
+	u32	max_blocksize = svc_max_payload(rqstp);
 
 	dprintk("nfsd: READ(3) %s %lu bytes at %lu\n",
 				SVCFH_fmt(&argp->fh),
@@ -172,8 +173,8 @@ nfsd3_proc_read(struct svc_rqst *rqstp, struct nfsd3_readargs *argp,
 	 */
 
 	resp->count = argp->count;
-	if (NFSSVC_MAXBLKSIZE < resp->count)
-		resp->count = NFSSVC_MAXBLKSIZE;
+	if (max_blocksize < resp->count)
+		resp->count = max_blocksize;
 
 	svc_reserve(rqstp, ((1 + NFS3_POST_OP_ATTR_WORDS + 3)<<2) + resp->count +4);
 
@@ -538,15 +539,16 @@ nfsd3_proc_fsinfo(struct svc_rqst * rqstp, struct nfsd_fhandle    *argp,
 					   struct nfsd3_fsinfores *resp)
 {
 	int	nfserr;
+	u32	max_blocksize = svc_max_payload(rqstp);
 
 	dprintk("nfsd: FSINFO(3)   %s\n",
 				SVCFH_fmt(&argp->fh));
 
-	resp->f_rtmax  = NFSSVC_MAXBLKSIZE;
-	resp->f_rtpref = NFSSVC_MAXBLKSIZE;
+	resp->f_rtmax  = max_blocksize;
+	resp->f_rtpref = max_blocksize;
 	resp->f_rtmult = PAGE_SIZE;
-	resp->f_wtmax  = NFSSVC_MAXBLKSIZE;
-	resp->f_wtpref = NFSSVC_MAXBLKSIZE;
+	resp->f_wtmax  = max_blocksize;
+	resp->f_wtpref = max_blocksize;
 	resp->f_wtmult = PAGE_SIZE;
 	resp->f_dtpref = PAGE_SIZE;
 	resp->f_maxfilesize = ~(u32) 0;
diff --git a/fs/nfsd/nfs3xdr.c b/fs/nfsd/nfs3xdr.c
index 4b9aefbcc93c5e..247d518248bf71 100644
--- a/fs/nfsd/nfs3xdr.c
+++ b/fs/nfsd/nfs3xdr.c
@@ -330,6 +330,7 @@ nfs3svc_decode_readargs(struct svc_rqst *rqstp, u32 *p,
 {
 	unsigned int len;
 	int v,pn;
+	u32 max_blocksize = svc_max_payload(rqstp);
 
 	if (!(p = decode_fh(p, &args->fh))
 	 || !(p = xdr_decode_hyper(p, &args->offset)))
@@ -337,8 +338,8 @@ nfs3svc_decode_readargs(struct svc_rqst *rqstp, u32 *p,
 
 	len = args->count = ntohl(*p++);
 
-	if (len > NFSSVC_MAXBLKSIZE)
-		len = NFSSVC_MAXBLKSIZE;
+	if (len > max_blocksize)
+		len = max_blocksize;
 
 	/* set up the kvec */
 	v=0;
@@ -358,6 +359,7 @@ nfs3svc_decode_writeargs(struct svc_rqst *rqstp, u32 *p,
 					struct nfsd3_writeargs *args)
 {
 	unsigned int len, v, hdr;
+	u32 max_blocksize = svc_max_payload(rqstp);
 
 	if (!(p = decode_fh(p, &args->fh))
 	 || !(p = xdr_decode_hyper(p, &args->offset)))
@@ -375,8 +377,8 @@ nfs3svc_decode_writeargs(struct svc_rqst *rqstp, u32 *p,
 	rqstp->rq_vec[0].iov_base = (void*)p;
 	rqstp->rq_vec[0].iov_len = rqstp->rq_arg.head[0].iov_len - hdr;
 
-	if (len > NFSSVC_MAXBLKSIZE)
-		len = NFSSVC_MAXBLKSIZE;
+	if (len > max_blocksize)
+		len = max_blocksize;
 	v=  0;
 	while (len > rqstp->rq_vec[v].iov_len) {
 		len -= rqstp->rq_vec[v].iov_len;
@@ -564,6 +566,7 @@ nfs3svc_decode_readdirplusargs(struct svc_rqst *rqstp, u32 *p,
 					struct nfsd3_readdirargs *args)
 {
 	int len, pn;
+	u32 max_blocksize = svc_max_payload(rqstp);
 
 	if (!(p = decode_fh(p, &args->fh)))
 		return 0;
@@ -572,7 +575,7 @@ nfs3svc_decode_readdirplusargs(struct svc_rqst *rqstp, u32 *p,
 	args->dircount = ntohl(*p++);
 	args->count    = ntohl(*p++);
 
-	len = (args->count > NFSSVC_MAXBLKSIZE) ? NFSSVC_MAXBLKSIZE :
+	len = (args->count > max_blocksize) ? max_blocksize :
 						  args->count;
 	args->count = len;
 
diff --git a/fs/nfsd/nfs4xdr.c b/fs/nfsd/nfs4xdr.c
index df341956254eab..4cfacc557b4087 100644
--- a/fs/nfsd/nfs4xdr.c
+++ b/fs/nfsd/nfs4xdr.c
@@ -1536,12 +1536,12 @@ out_acl:
 	if (bmval0 & FATTR4_WORD0_MAXREAD) {
 		if ((buflen -= 8) < 0)
 			goto out_resource;
-		WRITE64((u64) NFSSVC_MAXBLKSIZE);
+		WRITE64((u64) svc_max_payload(rqstp));
 	}
 	if (bmval0 & FATTR4_WORD0_MAXWRITE) {
 		if ((buflen -= 8) < 0)
 			goto out_resource;
-		WRITE64((u64) NFSSVC_MAXBLKSIZE);
+		WRITE64((u64) svc_max_payload(rqstp));
 	}
 	if (bmval1 & FATTR4_WORD1_MODE) {
 		if ((buflen -= 4) < 0)
@@ -2055,7 +2055,7 @@ nfsd4_encode_read(struct nfsd4_compoundres *resp, int nfserr,
 
 	RESERVE_SPACE(8); /* eof flag and byte count */
 
-	maxcount = NFSSVC_MAXBLKSIZE;
+	maxcount = svc_max_payload(resp->rqstp);
 	if (maxcount > read->rd_length)
 		maxcount = read->rd_length;
 
diff --git a/fs/nfsd/nfsproc.c b/fs/nfsd/nfsproc.c
index ca4973150218fb..9ee1dab5d44adc 100644
--- a/fs/nfsd/nfsproc.c
+++ b/fs/nfsd/nfsproc.c
@@ -146,13 +146,13 @@ nfsd_proc_read(struct svc_rqst *rqstp, struct nfsd_readargs *argp,
 	 * status, 17 words for fattr, and 1 word for the byte count.
 	 */
 
-	if (NFSSVC_MAXBLKSIZE < argp->count) {
+	if (NFSSVC_MAXBLKSIZE_V2 < argp->count) {
 		printk(KERN_NOTICE
 			"oversized read request from %u.%u.%u.%u:%d (%d bytes)\n",
 				NIPQUAD(rqstp->rq_addr.sin_addr.s_addr),
 				ntohs(rqstp->rq_addr.sin_port),
 				argp->count);
-		argp->count = NFSSVC_MAXBLKSIZE;
+		argp->count = NFSSVC_MAXBLKSIZE_V2;
 	}
 	svc_reserve(rqstp, (19<<2) + argp->count + 4);
 
@@ -553,7 +553,7 @@ static struct svc_procedure		nfsd_procedures2[18] = {
   PROC(none,	 void,		void,		none,		RC_NOCACHE, ST),
   PROC(lookup,	 diropargs,	diropres,	fhandle,	RC_NOCACHE, ST+FH+AT),
   PROC(readlink, readlinkargs,	readlinkres,	none,		RC_NOCACHE, ST+1+NFS_MAXPATHLEN/4),
-  PROC(read,	 readargs,	readres,	fhandle,	RC_NOCACHE, ST+AT+1+NFSSVC_MAXBLKSIZE/4),
+  PROC(read,	 readargs,	readres,	fhandle,	RC_NOCACHE, ST+AT+1+NFSSVC_MAXBLKSIZE_V2/4),
   PROC(none,	 void,		void,		none,		RC_NOCACHE, ST),
   PROC(write,	 writeargs,	attrstat,	fhandle,	RC_REPLBUFF, ST+AT),
   PROC(create,	 createargs,	diropres,	fhandle,	RC_REPLBUFF, ST+FH+AT),
diff --git a/fs/nfsd/nfsxdr.c b/fs/nfsd/nfsxdr.c
index ab6745e78d16d4..1135c0d145574a 100644
--- a/fs/nfsd/nfsxdr.c
+++ b/fs/nfsd/nfsxdr.c
@@ -254,8 +254,8 @@ nfssvc_decode_readargs(struct svc_rqst *rqstp, u32 *p,
 	len = args->count     = ntohl(*p++);
 	p++; /* totalcount - unused */
 
-	if (len > NFSSVC_MAXBLKSIZE)
-		len = NFSSVC_MAXBLKSIZE;
+	if (len > NFSSVC_MAXBLKSIZE_V2)
+		len = NFSSVC_MAXBLKSIZE_V2;
 
 	/* set up somewhere to store response.
 	 * We take pages, put them on reslist and include in iovec
@@ -288,8 +288,8 @@ nfssvc_decode_writeargs(struct svc_rqst *rqstp, u32 *p,
 	rqstp->rq_vec[0].iov_base = (void*)p;
 	rqstp->rq_vec[0].iov_len = rqstp->rq_arg.head[0].iov_len -
 				(((void*)p) - rqstp->rq_arg.head[0].iov_base);
-	if (len > NFSSVC_MAXBLKSIZE)
-		len = NFSSVC_MAXBLKSIZE;
+	if (len > NFSSVC_MAXBLKSIZE_V2)
+		len = NFSSVC_MAXBLKSIZE_V2;
 	v = 0;
 	while (len > rqstp->rq_vec[v].iov_len) {
 		len -= rqstp->rq_vec[v].iov_len;
@@ -458,7 +458,7 @@ nfssvc_encode_statfsres(struct svc_rqst *rqstp, u32 *p,
 {
 	struct kstatfs	*stat = &resp->stats;
 
-	*p++ = htonl(NFSSVC_MAXBLKSIZE);	/* max transfer size */
+	*p++ = htonl(NFSSVC_MAXBLKSIZE_V2);	/* max transfer size */
 	*p++ = htonl(stat->f_bsize);
 	*p++ = htonl(stat->f_blocks);
 	*p++ = htonl(stat->f_bfree);
diff --git a/include/linux/nfsd/const.h b/include/linux/nfsd/const.h
index b75bb1b38d09a2..adbddf007898d6 100644
--- a/include/linux/nfsd/const.h
+++ b/include/linux/nfsd/const.h
@@ -13,6 +13,7 @@
 #include <linux/nfs2.h>
 #include <linux/nfs3.h>
 #include <linux/nfs4.h>
+#include <linux/sunrpc/msg_prot.h>
 
 /*
  * Maximum protocol version supported by knfsd
@@ -23,6 +24,8 @@
  * Maximum blocksize supported by daemon currently at 32K
  */
 #define NFSSVC_MAXBLKSIZE	(32*1024)
+/* NFSv2 is limited by the protocol specification, see RFC 1094 */
+#define NFSSVC_MAXBLKSIZE_V2	(8*1024)
 
 #ifdef __KERNEL__
 
@@ -30,7 +33,17 @@
 # define NFS_SUPER_MAGIC	0x6969
 #endif
 
-#define NFSD_BUFSIZE		(1024 + NFSSVC_MAXBLKSIZE)
+/*
+ * Largest number of bytes we need to allocate for an NFS
+ * call or reply.  Used to control buffer sizes.  We use
+ * the length of v3 WRITE, READDIR and READDIR replies
+ * which are an RPC header, up to 26 XDR units of reply
+ * data, and some page data.
+ *
+ * Note that accuracy here doesn't matter too much as the
+ * size is rounded up to a page size when allocating space.
+ */
+#define NFSD_BUFSIZE		((RPC_MAX_HEADER_WITH_AUTH+26)*XDR_UNIT + NFSSVC_MAXBLKSIZE)
 
 #ifdef CONFIG_NFSD_V4
 # define NFSSVC_XDRSIZE		NFS4_SVC_XDRSIZE
diff --git a/include/linux/sunrpc/auth.h b/include/linux/sunrpc/auth.h
index 862c0d8c83817c..534cdc7be58dbb 100644
--- a/include/linux/sunrpc/auth.h
+++ b/include/linux/sunrpc/auth.h
@@ -20,9 +20,6 @@
 /* size of the nodename buffer */
 #define UNX_MAXNODENAME	32
 
-/* Maximum size (in bytes) of an rpc credential or verifier */
-#define RPC_MAX_AUTH_SIZE (400)
-
 /* Work around the lack of a VFS credential */
 struct auth_cred {
 	uid_t	uid;
diff --git a/include/linux/sunrpc/msg_prot.h b/include/linux/sunrpc/msg_prot.h
index 8d10d148834e0e..1e65f2dd80e5d5 100644
--- a/include/linux/sunrpc/msg_prot.h
+++ b/include/linux/sunrpc/msg_prot.h
@@ -11,6 +11,9 @@
 
 #define RPC_VERSION 2
 
+/* size of an XDR encoding unit in bytes, i.e. 32bit */
+#define XDR_UNIT	(4)
+
 /* spec defines authentication flavor as an unsigned 32 bit integer */
 typedef u32	rpc_authflavor_t;
 
@@ -34,6 +37,9 @@ enum rpc_auth_flavors {
 	RPC_AUTH_GSS_SPKMP = 390011,
 };
 
+/* Maximum size (in bytes) of an rpc credential or verifier */
+#define RPC_MAX_AUTH_SIZE (400)
+
 enum rpc_msg_type {
 	RPC_CALL = 0,
 	RPC_REPLY = 1
@@ -101,5 +107,39 @@ typedef __be32	rpc_fraghdr;
 #define	RPC_FRAGMENT_SIZE_MASK		(~RPC_LAST_STREAM_FRAGMENT)
 #define	RPC_MAX_FRAGMENT_SIZE		((1U << 31) - 1)
 
+/*
+ * RPC call and reply header size as number of 32bit words (verifier
+ * size computed separately, see below)
+ */
+#define RPC_CALLHDRSIZE		(6)
+#define RPC_REPHDRSIZE		(4)
+
+
+/*
+ * Maximum RPC header size, including authentication,
+ * as number of 32bit words (see RFCs 1831, 1832).
+ *
+ *	xid			    1 xdr unit = 4 bytes
+ *	mtype			    1
+ *	rpc_version		    1
+ *	program			    1
+ *	prog_version		    1
+ *	procedure		    1
+ *	cred {
+ *	    flavor		    1
+ *	    length		    1
+ *	    body<RPC_MAX_AUTH_SIZE> 100 xdr units = 400 bytes
+ *	}
+ *	verf {
+ *	    flavor		    1
+ *	    length		    1
+ *	    body<RPC_MAX_AUTH_SIZE> 100 xdr units = 400 bytes
+ *	}
+ *	TOTAL			    210 xdr units = 840 bytes
+ */
+#define RPC_MAX_HEADER_WITH_AUTH \
+	(RPC_CALLHDRSIZE + 2*(2+RPC_MAX_AUTH_SIZE/4))
+
+
 #endif /* __KERNEL__ */
 #endif /* _LINUX_SUNRPC_MSGPROT_H_ */
diff --git a/include/linux/sunrpc/svc.h b/include/linux/sunrpc/svc.h
index cb0ed9beb22726..74e52c245da4b1 100644
--- a/include/linux/sunrpc/svc.h
+++ b/include/linux/sunrpc/svc.h
@@ -13,6 +13,7 @@
 #include <linux/in.h>
 #include <linux/sunrpc/types.h>
 #include <linux/sunrpc/xdr.h>
+#include <linux/sunrpc/auth.h>
 #include <linux/sunrpc/svcauth.h>
 #include <linux/wait.h>
 #include <linux/mm.h>
@@ -95,8 +96,28 @@ static inline void svc_get(struct svc_serv *serv)
  * Maximum payload size supported by a kernel RPC server.
  * This is use to determine the max number of pages nfsd is
  * willing to return in a single READ operation.
+ *
+ * These happen to all be powers of 2, which is not strictly
+ * necessary but helps enforce the real limitation, which is
+ * that they should be multiples of PAGE_CACHE_SIZE.
+ *
+ * For UDP transports, a block plus NFS,RPC, and UDP headers
+ * has to fit into the IP datagram limit of 64K.  The largest
+ * feasible number for all known page sizes is probably 48K,
+ * but we choose 32K here.  This is the same as the historical
+ * Linux limit; someone who cares more about NFS/UDP performance
+ * can test a larger number.
+ *
+ * For TCP transports we have more freedom.  A size of 1MB is
+ * chosen to match the client limit.  Other OSes are known to
+ * have larger limits, but those numbers are probably beyond
+ * the point of diminishing returns.
  */
-#define RPCSVC_MAXPAYLOAD	(64*1024u)
+#define RPCSVC_MAXPAYLOAD	(1*1024*1024u)
+#define RPCSVC_MAXPAYLOAD_TCP	RPCSVC_MAXPAYLOAD
+#define RPCSVC_MAXPAYLOAD_UDP	(32*1024u)
+
+extern u32 svc_max_payload(const struct svc_rqst *rqstp);
 
 /*
  * RPC Requsts and replies are stored in one or more pages.
diff --git a/include/linux/sunrpc/xprt.h b/include/linux/sunrpc/xprt.h
index 6cf62658075298..60394fbc4c704d 100644
--- a/include/linux/sunrpc/xprt.h
+++ b/include/linux/sunrpc/xprt.h
@@ -15,6 +15,7 @@
 #include <linux/kref.h>
 #include <linux/sunrpc/sched.h>
 #include <linux/sunrpc/xdr.h>
+#include <linux/sunrpc/msg_prot.h>
 
 extern unsigned int xprt_udp_slot_table_entries;
 extern unsigned int xprt_tcp_slot_table_entries;
@@ -24,13 +25,6 @@ extern unsigned int xprt_tcp_slot_table_entries;
 #define RPC_MAX_SLOT_TABLE	(128U)
 
 /*
- * RPC call and reply header size as number of 32bit words (verifier
- * size computed separately)
- */
-#define RPC_CALLHDRSIZE		6
-#define RPC_REPHDRSIZE		4
-
-/*
  * Parameters for choosing a free port
  */
 extern unsigned int xprt_min_resvport;
diff --git a/net/sunrpc/svc.c b/net/sunrpc/svc.c
index f4a509a925b5ee..b252401c86016d 100644
--- a/net/sunrpc/svc.c
+++ b/net/sunrpc/svc.c
@@ -919,3 +919,18 @@ err_bad:
 	svc_putnl(resv, ntohl(rpc_stat));
 	goto sendit;
 }
+
+/*
+ * Return (transport-specific) limit on the rpc payload.
+ */
+u32 svc_max_payload(const struct svc_rqst *rqstp)
+{
+	int max = RPCSVC_MAXPAYLOAD_TCP;
+
+	if (rqstp->rq_sock->sk_sock->type == SOCK_DGRAM)
+		max = RPCSVC_MAXPAYLOAD_UDP;
+	if (rqstp->rq_server->sv_bufsz < max)
+		max = rqstp->rq_server->sv_bufsz;
+	return max;
+}
+EXPORT_SYMBOL_GPL(svc_max_payload);