NFS: pnfs IPv6 support

Handle ipv6 remote addresses from GETDEVICEINFO

 - supports netid "tcp" for ipv4 and "tcp6" for ipv6 as rfc 5665 specifies
 - added ds_remotestr to avoid having to handle different AFs in every dprintk
 - tested against pynfs 4.1 server, submitting ipv6 support patch to pynfs
 - tested with IPv6 disabled, it compiles cleanly and relies on rpc_pton to
   refuse to accept IPv6 addresses

Signed-off-by: Weston Andros Adamson <dros@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>

1 files changed, 185 insertions, 70 deletions

diff --git a/fs/nfs/nfs4filelayoutdev.c b/fs/nfs/nfs4filelayoutdev.c
index 3b7bf1377264..1a8308b90108 100644
--- a/fs/nfs/nfs4filelayoutdev.c
+++ b/fs/nfs/nfs4filelayoutdev.c
@@ -56,28 +56,56 @@ print_ds(struct nfs4_pnfs_ds *ds)
                 printk("%s NULL device\n", __func__);
                 return;
         }
-        printk("        ip_addr %x port %hu\n"
+        printk("        ds %s\n"
                 "        ref count %d\n"
                 "        client %p\n"
                 "        cl_exchange_flags %x\n",
-                ntohl(ds->ds_ip_addr), ntohs(ds->ds_port),
+                ds->ds_remotestr,
                 atomic_read(&ds->ds_count), ds->ds_clp,
                 ds->ds_clp ? ds->ds_clp->cl_exchange_flags : 0);
 }
 
 /* nfs4_ds_cache_lock is held */
 static struct nfs4_pnfs_ds *
-_data_server_lookup_locked(u32 ip_addr, u32 port)
+_data_server_lookup_locked(struct sockaddr *addr, size_t addrlen)
 {
         struct nfs4_pnfs_ds *ds;
-
-        dprintk("_data_server_lookup: ip_addr=%x port=%hu\n",
-                        ntohl(ip_addr), ntohs(port));
+        struct sockaddr_in *a, *b;
+        struct sockaddr_in6 *a6, *b6;
 
         list_for_each_entry(ds, &nfs4_data_server_cache, ds_node) {
-                if (ds->ds_ip_addr == ip_addr &&
-                    ds->ds_port == port) {
-                        return ds;
+                if (addr->sa_family != ds->ds_addr.ss_family)
+                        continue;
+
+                switch (addr->sa_family) {
+                case AF_INET:
+                        a = (struct sockaddr_in *)addr;
+                        b = (struct sockaddr_in *)&ds->ds_addr;
+
+                        if (a->sin_addr.s_addr == b->sin_addr.s_addr &&
+                            a->sin_port == b->sin_port)
+                                return ds;
+                        break;
+
+                case AF_INET6:
+                        a6 = (struct sockaddr_in6 *)addr;
+                        b6 = (struct sockaddr_in6 *)&ds->ds_addr;
+
+                        /* LINKLOCAL addresses must have matching scope_id */
+                        if (ipv6_addr_scope(&a6->sin6_addr) ==
+                            IPV6_ADDR_SCOPE_LINKLOCAL &&
+                            a6->sin6_scope_id != b6->sin6_scope_id)
+                                continue;
+
+                        if (ipv6_addr_equal(&a6->sin6_addr, &b6->sin6_addr) &&
+                            a6->sin6_port == b6->sin6_port)
+                                return ds;
+                        break;
+
+                default:
+                        dprintk("%s: unhandled address family: %u\n",
+                                __func__, addr->sa_family);
+                        return NULL;
                 }
         }
         return NULL;
@@ -91,19 +119,14 @@ static int
 nfs4_ds_connect(struct nfs_server *mds_srv, struct nfs4_pnfs_ds *ds)
 {
         struct nfs_client *clp;
-        struct sockaddr_in sin;
         int status = 0;
 
-        dprintk("--> %s ip:port %x:%hu au_flavor %d\n", __func__,
-                ntohl(ds->ds_ip_addr), ntohs(ds->ds_port),
+        dprintk("--> %s addr %s au_flavor %d\n", __func__, ds->ds_remotestr,
                 mds_srv->nfs_client->cl_rpcclient->cl_auth->au_flavor);
 
-        sin.sin_family = AF_INET;
-        sin.sin_addr.s_addr = ds->ds_ip_addr;
-        sin.sin_port = ds->ds_port;
-
-        clp = nfs4_set_ds_client(mds_srv->nfs_client, (struct sockaddr *)&sin,
-                                 sizeof(sin), IPPROTO_TCP);
+        clp = nfs4_set_ds_client(mds_srv->nfs_client,
+                                 (struct sockaddr *)&ds->ds_addr,
+                                 ds->ds_addrlen, IPPROTO_TCP);
         if (IS_ERR(clp)) {
                 status = PTR_ERR(clp);
                 goto out;
@@ -115,8 +138,8 @@ nfs4_ds_connect(struct nfs_server *mds_srv, struct nfs4_pnfs_ds *ds)
                         goto out_put;
                 }
                 ds->ds_clp = clp;
-                dprintk("%s [existing] ip=%x, port=%hu\n", __func__,
-                        ntohl(ds->ds_ip_addr), ntohs(ds->ds_port));
+                dprintk("%s [existing] server=%s\n", __func__,
+                        ds->ds_remotestr);
                 goto out;
         }
 
@@ -135,8 +158,7 @@ nfs4_ds_connect(struct nfs_server *mds_srv, struct nfs4_pnfs_ds *ds)
                 goto out_put;
 
         ds->ds_clp = clp;
-        dprintk("%s [new] ip=%x, port=%hu\n", __func__, ntohl(ds->ds_ip_addr),
-                ntohs(ds->ds_port));
+        dprintk("%s [new] addr: %s\n", __func__, ds->ds_remotestr);
 out:
         return status;
 out_put:
@@ -153,6 +175,7 @@ destroy_ds(struct nfs4_pnfs_ds *ds)
 
         if (ds->ds_clp)
                 nfs_put_client(ds->ds_clp);
+        kfree(ds->ds_remotestr);
         kfree(ds);
 }
 
@@ -179,31 +202,85 @@ nfs4_fl_free_deviceid(struct nfs4_file_layout_dsaddr *dsaddr)
         kfree(dsaddr);
 }
 
+/*
+ * Create a string with a human readable address and port to avoid
+ * complicated setup around many dprinks.
+ */
+static char *
+nfs4_pnfs_remotestr(struct sockaddr *ds_addr, gfp_t gfp_flags)
+{
+        char buf[INET6_ADDRSTRLEN + IPV6_SCOPE_ID_LEN];
+        char *remotestr;
+        char *startsep = "";
+        char *endsep = "";
+        size_t len;
+        uint16_t port;
+
+        switch (ds_addr->sa_family) {
+        case AF_INET:
+                port = ((struct sockaddr_in *)ds_addr)->sin_port;
+                break;
+        case AF_INET6:
+                startsep = "[";
+                endsep = "]";
+                port = ((struct sockaddr_in6 *)ds_addr)->sin6_port;
+                break;
+        default:
+                dprintk("%s: Unknown address family %u\n",
+                        __func__, ds_addr->sa_family);
+                return NULL;
+        }
+
+        if (!rpc_ntop((struct sockaddr *)ds_addr, buf, sizeof(buf))) {
+                dprintk("%s: error printing addr\n", __func__);
+                return NULL;
+        }
+
+        len = strlen(buf) + strlen(startsep) + strlen(endsep) + 1 + 5 + 1;
+        remotestr = kzalloc(len, gfp_flags);
+
+        if (unlikely(!remotestr)) {
+                dprintk("%s: couldn't alloc remotestr\n", __func__);
+                return NULL;
+        }
+
+        snprintf(remotestr, len, "%s%s%s:%u",
+                 startsep, buf, endsep, ntohs(port));
+
+        return remotestr;
+}
+
 static struct nfs4_pnfs_ds *
-nfs4_pnfs_ds_add(struct inode *inode, u32 ip_addr, u32 port, gfp_t gfp_flags)
+nfs4_pnfs_ds_add(struct sockaddr *addr, size_t addrlen, gfp_t gfp_flags)
 {
-        struct nfs4_pnfs_ds *tmp_ds, *ds;
+        struct nfs4_pnfs_ds *tmp_ds, *ds = NULL;
+        char *remotestr;
 
         ds = kzalloc(sizeof(*tmp_ds), gfp_flags);
         if (!ds)
                 goto out;
 
+        /* this is only used for debugging, so it's ok if its NULL */
+        remotestr = nfs4_pnfs_remotestr(addr, gfp_flags);
+
         spin_lock(&nfs4_ds_cache_lock);
-        tmp_ds = _data_server_lookup_locked(ip_addr, port);
+        tmp_ds = _data_server_lookup_locked(addr, addrlen);
         if (tmp_ds == NULL) {
-                ds->ds_ip_addr = ip_addr;
-                ds->ds_port = port;
+                memcpy(&ds->ds_addr, addr, addrlen);
+                ds->ds_addrlen = addrlen;
+                ds->ds_remotestr = remotestr;
                 atomic_set(&ds->ds_count, 1);
                 INIT_LIST_HEAD(&ds->ds_node);
                 ds->ds_clp = NULL;
                 list_add(&ds->ds_node, &nfs4_data_server_cache);
-                dprintk("%s add new data server ip 0x%x\n", __func__,
-                        ds->ds_ip_addr);
+                dprintk("%s add new data server %s\n", __func__,
+                        ds->ds_remotestr);
         } else {
+                kfree(remotestr);
                 kfree(ds);
                 atomic_inc(&tmp_ds->ds_count);
-                dprintk("%s data server found ip 0x%x, inc'ed ds_count to %d\n",
-                        __func__, tmp_ds->ds_ip_addr,
+                dprintk("%s data server %s found, inc'ed ds_count to %d\n",
+                        __func__, tmp_ds->ds_remotestr,
                         atomic_read(&tmp_ds->ds_count));
                 ds = tmp_ds;
         }
@@ -213,18 +290,21 @@ out:
 }
 
 /*
- * Currently only support ipv4, and one multi-path address.
+ * Currently only supports ipv4, ipv6 and one multi-path address.
  */
 static struct nfs4_pnfs_ds *
 decode_and_add_ds(struct xdr_stream *streamp, struct inode *inode, gfp_t gfp_flags)
 {
         struct nfs4_pnfs_ds *ds = NULL;
-        char *buf;
-        const char *ipend, *pstr;
-        u32 ip_addr, port;
-        int nlen, rlen, i;
+        char *buf, *portstr;
+        struct sockaddr_storage ss;
+        size_t sslen;
+        u32 port;
+        int nlen, rlen;
         int tmp[2];
         __be32 *p;
+        char *netid, *match_netid;
+        size_t match_netid_len;
 
         /* r_netid */
         p = xdr_inline_decode(streamp, 4);
@@ -236,62 +316,97 @@ decode_and_add_ds(struct xdr_stream *streamp, struct inode *inode, gfp_t gfp_fla
         if (unlikely(!p))
                 goto out_err;
 
-        /* Check that netid is "tcp" */
-        if (nlen != 3 ||  memcmp((char *)p, "tcp", 3)) {
-                dprintk("%s: ERROR: non ipv4 TCP r_netid\n", __func__);
+        netid = kmalloc(nlen+1, gfp_flags);
+        if (unlikely(!netid))
                 goto out_err;
-        }
 
-        /* r_addr */
+        netid[nlen] = '\0';
+        memcpy(netid, p, nlen);
+
+        /* r_addr: ip/ip6addr with port in dec octets - see RFC 5665 */
         p = xdr_inline_decode(streamp, 4);
         if (unlikely(!p))
-                goto out_err;
+                goto out_free_netid;
         rlen = be32_to_cpup(p);
 
         p = xdr_inline_decode(streamp, rlen);
         if (unlikely(!p))
-                goto out_err;
+                goto out_free_netid;
 
-        /* ipv6 length plus port is legal */
-        if (rlen > INET6_ADDRSTRLEN + 8) {
+        /* port is ".ABC.DEF", 8 chars max */
+        if (rlen > INET6_ADDRSTRLEN + IPV6_SCOPE_ID_LEN + 8) {
                 dprintk("%s: Invalid address, length %d\n", __func__,
                         rlen);
-                goto out_err;
+                goto out_free_netid;
         }
         buf = kmalloc(rlen + 1, gfp_flags);
         if (!buf) {
                 dprintk("%s: Not enough memory\n", __func__);
-                goto out_err;
+                goto out_free_netid;
         }
         buf[rlen] = '\0';
         memcpy(buf, p, rlen);
 
-        /* replace the port dots with dashes for the in4_pton() delimiter*/
-        for (i = 0; i < 2; i++) {
-                char *res = strrchr(buf, '.');
-                if (!res) {
-                        dprintk("%s: Failed finding expected dots in port\n",
-                                __func__);
-                        goto out_free;
-                }
-                *res = '-';
+        /* replace port '.' with '-' */
+        portstr = strrchr(buf, '.');
+        if (!portstr) {
+                dprintk("%s: Failed finding expected dot in port\n",
+                        __func__);
+                goto out_free_buf;
+        }
+        *portstr = '-';
+
+        /* find '.' between address and port */
+        portstr = strrchr(buf, '.');
+        if (!portstr) {
+                dprintk("%s: Failed finding expected dot between address and "
+                        "port\n", __func__);
+                goto out_free_buf;
         }
+        *portstr = '\0';
 
-        /* Currently only support ipv4 address */
-        if (in4_pton(buf, rlen, (u8 *)&ip_addr, '-', &ipend) == 0) {
-                dprintk("%s: Only ipv4 addresses supported\n", __func__);
-                goto out_free;
+        if (!rpc_pton(buf, portstr-buf, (struct sockaddr *)&ss, sizeof(ss))) {
+                dprintk("%s: Error parsing address %s\n", __func__, buf);
+                goto out_free_buf;
         }
 
-        /* port */
-        pstr = ipend;
-        sscanf(pstr, "-%d-%d", &tmp[0], &tmp[1]);
+        portstr++;
+        sscanf(portstr, "%d-%d", &tmp[0], &tmp[1]);
         port = htons((tmp[0] << 8) | (tmp[1]));
 
-        ds = nfs4_pnfs_ds_add(inode, ip_addr, port, gfp_flags);
-        dprintk("%s: Decoded address and port %s\n", __func__, buf);
-out_free:
+        switch (ss.ss_family) {
+        case AF_INET:
+                ((struct sockaddr_in *)&ss)->sin_port = port;
+                sslen = sizeof(struct sockaddr_in);
+                match_netid = "tcp";
+                match_netid_len = 3;
+                break;
+
+        case AF_INET6:
+                ((struct sockaddr_in6 *)&ss)->sin6_port = port;
+                sslen = sizeof(struct sockaddr_in6);
+                match_netid = "tcp6";
+                match_netid_len = 4;
+                break;
+
+        default:
+                dprintk("%s: unsupported address family: %u\n",
+                        __func__, ss.ss_family);
+                goto out_free_buf;
+        }
+
+        if (nlen != match_netid_len || strncmp(netid, match_netid, nlen)) {
+                dprintk("%s: ERROR: r_netid \"%s\" != \"%s\"\n",
+                        __func__, netid, match_netid);
+                goto out_free_buf;
+        }
+
+        ds = nfs4_pnfs_ds_add((struct sockaddr *)&ss, sslen, gfp_flags);
+        dprintk("%s: Added DS %s\n", __func__, ds->ds_remotestr);
+out_free_buf:
         kfree(buf);
+out_free_netid:
+        kfree(netid);
 out_err:
         return ds;
 }
@@ -591,13 +706,13 @@ nfs4_fl_select_ds_fh(struct pnfs_layout_segment *lseg, u32 j)
 
 static void
 filelayout_mark_devid_negative(struct nfs4_file_layout_dsaddr *dsaddr,
-                               int err, u32 ds_addr)
+                               int err, const char *ds_remotestr)
 {
         u32 *p = (u32 *)&dsaddr->id_node.deviceid;
 
-        printk(KERN_ERR "NFS: data server %x connection error %d."
+        printk(KERN_ERR "NFS: data server %s connection error %d."
                 " Deviceid [%x%x%x%x] marked out of use.\n",
-                ds_addr, err, p[0], p[1], p[2], p[3]);
+                ds_remotestr, err, p[0], p[1], p[2], p[3]);
 
         spin_lock(&nfs4_ds_cache_lock);
         dsaddr->flags |= NFS4_DEVICE_ID_NEG_ENTRY;
@@ -628,7 +743,7 @@ nfs4_fl_prepare_ds(struct pnfs_layout_segment *lseg, u32 ds_idx)
                 err = nfs4_ds_connect(s, ds);
                 if (err) {
                         filelayout_mark_devid_negative(dsaddr, err,
-                                                       ntohl(ds->ds_ip_addr));
+                                                       ds->ds_remotestr);
                         return NULL;
                 }
         }