From 11fd165c68b73434ca1273e21f21db5eecc90926 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 28 Jul 2011 20:04:09 +0200 Subject: sunrpc: use better NUMA affinities Use NUMA aware allocations to reduce latencies and increase throughput. sunrpc kthreads can use kthread_create_on_node() if pool_mode is "percpu" or "pernode", and svc_prepare_thread()/svc_init_buffer() can also take into account NUMA node affinity for memory allocations. Signed-off-by: Eric Dumazet CC: "J. Bruce Fields" CC: Neil Brown CC: David Miller Reviewed-by: Greg Banks [bfields@redhat.com: fix up caller nfs41_callback_up] Signed-off-by: J. Bruce Fields --- net/sunrpc/svc.c | 33 ++++++++++++++++++++++++--------- 1 file changed, 24 insertions(+), 9 deletions(-) (limited to 'net/sunrpc') diff --git a/net/sunrpc/svc.c b/net/sunrpc/svc.c index 6a69a1131fb7..30d70abb4e2c 100644 --- a/net/sunrpc/svc.c +++ b/net/sunrpc/svc.c @@ -295,6 +295,18 @@ svc_pool_map_put(void) } +static int svc_pool_map_get_node(unsigned int pidx) +{ + const struct svc_pool_map *m = &svc_pool_map; + + if (m->count) { + if (m->mode == SVC_POOL_PERCPU) + return cpu_to_node(m->pool_to[pidx]); + if (m->mode == SVC_POOL_PERNODE) + return m->pool_to[pidx]; + } + return NUMA_NO_NODE; +} /* * Set the given thread's cpus_allowed mask so that it * will only run on cpus in the given pool. @@ -499,7 +511,7 @@ EXPORT_SYMBOL_GPL(svc_destroy); * We allocate pages and place them in rq_argpages. */ static int -svc_init_buffer(struct svc_rqst *rqstp, unsigned int size) +svc_init_buffer(struct svc_rqst *rqstp, unsigned int size, int node) { unsigned int pages, arghi; @@ -513,7 +525,7 @@ svc_init_buffer(struct svc_rqst *rqstp, unsigned int size) arghi = 0; BUG_ON(pages > RPCSVC_MAXPAGES); while (pages) { - struct page *p = alloc_page(GFP_KERNEL); + struct page *p = alloc_pages_node(node, GFP_KERNEL, 0); if (!p) break; rqstp->rq_pages[arghi++] = p; @@ -536,11 +548,11 @@ svc_release_buffer(struct svc_rqst *rqstp) } struct svc_rqst * -svc_prepare_thread(struct svc_serv *serv, struct svc_pool *pool) +svc_prepare_thread(struct svc_serv *serv, struct svc_pool *pool, int node) { struct svc_rqst *rqstp; - rqstp = kzalloc(sizeof(*rqstp), GFP_KERNEL); + rqstp = kzalloc_node(sizeof(*rqstp), GFP_KERNEL, node); if (!rqstp) goto out_enomem; @@ -554,15 +566,15 @@ svc_prepare_thread(struct svc_serv *serv, struct svc_pool *pool) rqstp->rq_server = serv; rqstp->rq_pool = pool; - rqstp->rq_argp = kmalloc(serv->sv_xdrsize, GFP_KERNEL); + rqstp->rq_argp = kmalloc_node(serv->sv_xdrsize, GFP_KERNEL, node); if (!rqstp->rq_argp) goto out_thread; - rqstp->rq_resp = kmalloc(serv->sv_xdrsize, GFP_KERNEL); + rqstp->rq_resp = kmalloc_node(serv->sv_xdrsize, GFP_KERNEL, node); if (!rqstp->rq_resp) goto out_thread; - if (!svc_init_buffer(rqstp, serv->sv_max_mesg)) + if (!svc_init_buffer(rqstp, serv->sv_max_mesg, node)) goto out_thread; return rqstp; @@ -647,6 +659,7 @@ svc_set_num_threads(struct svc_serv *serv, struct svc_pool *pool, int nrservs) struct svc_pool *chosen_pool; int error = 0; unsigned int state = serv->sv_nrthreads-1; + int node; if (pool == NULL) { /* The -1 assumes caller has done a svc_get() */ @@ -662,14 +675,16 @@ svc_set_num_threads(struct svc_serv *serv, struct svc_pool *pool, int nrservs) nrservs--; chosen_pool = choose_pool(serv, pool, &state); - rqstp = svc_prepare_thread(serv, chosen_pool); + node = svc_pool_map_get_node(chosen_pool->sp_id); + rqstp = svc_prepare_thread(serv, chosen_pool, node); if (IS_ERR(rqstp)) { error = PTR_ERR(rqstp); break; } __module_get(serv->sv_module); - task = kthread_create(serv->sv_function, rqstp, serv->sv_name); + task = kthread_create_on_node(serv->sv_function, rqstp, + node, serv->sv_name); if (IS_ERR(task)) { error = PTR_ERR(task); module_put(serv->sv_module); -- cgit v1.2.3-70-g09d2 From 849a1cf13d4394d398d91752166e92e9ecd64f8d Mon Sep 17 00:00:00 2001 From: Mi Jinlong Date: Tue, 30 Aug 2011 17:18:41 +0800 Subject: SUNRPC: Replace svc_addr_u by sockaddr_storage For IPv6 local address, lockd can not callback to client for missing scope id when binding address at inet6_bind: 324 if (addr_type & IPV6_ADDR_LINKLOCAL) { 325 if (addr_len >= sizeof(struct sockaddr_in6) && 326 addr->sin6_scope_id) { 327 /* Override any existing binding, if another one 328 * is supplied by user. 329 */ 330 sk->sk_bound_dev_if = addr->sin6_scope_id; 331 } 332 333 /* Binding to link-local address requires an interface */ 334 if (!sk->sk_bound_dev_if) { 335 err = -EINVAL; 336 goto out_unlock; 337 } Replacing svc_addr_u by sockaddr_storage, let rqstp->rq_daddr contains more info besides address. Reviewed-by: Jeff Layton Reviewed-by: Chuck Lever Signed-off-by: Mi Jinlong Signed-off-by: J. Bruce Fields --- fs/lockd/host.c | 25 ++----------------------- fs/nfsd/nfs4state.c | 16 +--------------- include/linux/sunrpc/svc.h | 30 +++++++++++++++++++++--------- net/sunrpc/svc_xprt.c | 13 ++----------- net/sunrpc/svcsock.c | 23 +++++++++++++++++------ 5 files changed, 43 insertions(+), 64 deletions(-) (limited to 'net/sunrpc') diff --git a/fs/lockd/host.c b/fs/lockd/host.c index b7c99bfb3da6..6f29836ec0cb 100644 --- a/fs/lockd/host.c +++ b/fs/lockd/host.c @@ -316,14 +316,8 @@ struct nlm_host *nlmsvc_lookup_host(const struct svc_rqst *rqstp, struct hlist_node *pos; struct nlm_host *host = NULL; struct nsm_handle *nsm = NULL; - struct sockaddr_in sin = { - .sin_family = AF_INET, - }; - struct sockaddr_in6 sin6 = { - .sin6_family = AF_INET6, - }; - struct sockaddr *src_sap; - size_t src_len = rqstp->rq_addrlen; + struct sockaddr *src_sap = svc_daddr(rqstp); + size_t src_len = rqstp->rq_daddrlen; struct nlm_lookup_host_info ni = { .server = 1, .sap = svc_addr(rqstp), @@ -340,21 +334,6 @@ struct nlm_host *nlmsvc_lookup_host(const struct svc_rqst *rqstp, mutex_lock(&nlm_host_mutex); - switch (ni.sap->sa_family) { - case AF_INET: - sin.sin_addr.s_addr = rqstp->rq_daddr.addr.s_addr; - src_sap = (struct sockaddr *)&sin; - break; - case AF_INET6: - ipv6_addr_copy(&sin6.sin6_addr, &rqstp->rq_daddr.addr6); - src_sap = (struct sockaddr *)&sin6; - break; - default: - dprintk("lockd: %s failed; unrecognized address family\n", - __func__); - goto out; - } - if (time_after_eq(jiffies, next_gc)) nlm_gc_hosts(); diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c index 0cd346477f29..e7f83bd9b4a8 100644 --- a/fs/nfsd/nfs4state.c +++ b/fs/nfsd/nfs4state.c @@ -1257,20 +1257,6 @@ find_unconfirmed_client_by_str(const char *dname, unsigned int hashval) return NULL; } -static void rpc_svcaddr2sockaddr(struct sockaddr *sa, unsigned short family, union svc_addr_u *svcaddr) -{ - switch (family) { - case AF_INET: - ((struct sockaddr_in *)sa)->sin_family = AF_INET; - ((struct sockaddr_in *)sa)->sin_addr = svcaddr->addr; - return; - case AF_INET6: - ((struct sockaddr_in6 *)sa)->sin6_family = AF_INET6; - ((struct sockaddr_in6 *)sa)->sin6_addr = svcaddr->addr6; - return; - } -} - static void gen_callback(struct nfs4_client *clp, struct nfsd4_setclientid *se, struct svc_rqst *rqstp) { @@ -1302,7 +1288,7 @@ gen_callback(struct nfs4_client *clp, struct nfsd4_setclientid *se, struct svc_r conn->cb_prog = se->se_callback_prog; conn->cb_ident = se->se_callback_ident; - rpc_svcaddr2sockaddr((struct sockaddr *)&conn->cb_saddr, expected_family, &rqstp->rq_daddr); + memcpy(&conn->cb_saddr, &rqstp->rq_daddr, rqstp->rq_daddrlen); return; out_err: conn->cb_addr.ss_family = AF_UNSPEC; diff --git a/include/linux/sunrpc/svc.h b/include/linux/sunrpc/svc.h index a78a51e93373..d8d5d93071b3 100644 --- a/include/linux/sunrpc/svc.h +++ b/include/linux/sunrpc/svc.h @@ -212,11 +212,6 @@ static inline void svc_putu32(struct kvec *iov, __be32 val) iov->iov_len += sizeof(__be32); } -union svc_addr_u { - struct in_addr addr; - struct in6_addr addr6; -}; - /* * The context of a single thread, including the request currently being * processed. @@ -225,8 +220,12 @@ struct svc_rqst { struct list_head rq_list; /* idle list */ struct list_head rq_all; /* all threads list */ struct svc_xprt * rq_xprt; /* transport ptr */ + struct sockaddr_storage rq_addr; /* peer address */ size_t rq_addrlen; + struct sockaddr_storage rq_daddr; /* dest addr of request + * - reply from here */ + size_t rq_daddrlen; struct svc_serv * rq_server; /* RPC service definition */ struct svc_pool * rq_pool; /* thread pool */ @@ -255,9 +254,6 @@ struct svc_rqst { unsigned short rq_secure : 1; /* secure port */ - union svc_addr_u rq_daddr; /* dest addr of request - * - reply from here */ - void * rq_argp; /* decoded arguments */ void * rq_resp; /* xdr'd results */ void * rq_auth_data; /* flavor-specific data */ @@ -300,6 +296,21 @@ static inline struct sockaddr *svc_addr(const struct svc_rqst *rqst) return (struct sockaddr *) &rqst->rq_addr; } +static inline struct sockaddr_in *svc_daddr_in(const struct svc_rqst *rqst) +{ + return (struct sockaddr_in *) &rqst->rq_daddr; +} + +static inline struct sockaddr_in6 *svc_daddr_in6(const struct svc_rqst *rqst) +{ + return (struct sockaddr_in6 *) &rqst->rq_daddr; +} + +static inline struct sockaddr *svc_daddr(const struct svc_rqst *rqst) +{ + return (struct sockaddr *) &rqst->rq_daddr; +} + /* * Check buffer bounds after decoding arguments */ @@ -340,7 +351,8 @@ struct svc_deferred_req { 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 */ + struct sockaddr_storage daddr; /* where reply must come from */ + size_t daddrlen; struct cache_deferred_req handle; size_t xprt_hlen; int argslen; diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c index bd31208bbb61..d86bb673e1f6 100644 --- a/net/sunrpc/svc_xprt.c +++ b/net/sunrpc/svc_xprt.c @@ -254,8 +254,6 @@ EXPORT_SYMBOL_GPL(svc_create_xprt); */ 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; @@ -263,15 +261,8 @@ void svc_xprt_copy_addrs(struct svc_rqst *rqstp, struct svc_xprt *xprt) * 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; - } + memcpy(&rqstp->rq_daddr, &xprt->xpt_local, xprt->xpt_locallen); + rqstp->rq_daddrlen = xprt->xpt_locallen; } EXPORT_SYMBOL_GPL(svc_xprt_copy_addrs); diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c index 767d494de7a2..dfd686eb0b7f 100644 --- a/net/sunrpc/svcsock.c +++ b/net/sunrpc/svcsock.c @@ -143,19 +143,20 @@ static void svc_set_cmsg_data(struct svc_rqst *rqstp, struct cmsghdr *cmh) cmh->cmsg_level = SOL_IP; cmh->cmsg_type = IP_PKTINFO; pki->ipi_ifindex = 0; - pki->ipi_spec_dst.s_addr = rqstp->rq_daddr.addr.s_addr; + pki->ipi_spec_dst.s_addr = + svc_daddr_in(rqstp)->sin_addr.s_addr; cmh->cmsg_len = CMSG_LEN(sizeof(*pki)); } break; case AF_INET6: { struct in6_pktinfo *pki = CMSG_DATA(cmh); + struct sockaddr_in6 *daddr = svc_daddr_in6(rqstp); cmh->cmsg_level = SOL_IPV6; cmh->cmsg_type = IPV6_PKTINFO; - pki->ipi6_ifindex = 0; - ipv6_addr_copy(&pki->ipi6_addr, - &rqstp->rq_daddr.addr6); + pki->ipi6_ifindex = daddr->sin6_scope_id; + ipv6_addr_copy(&pki->ipi6_addr, &daddr->sin6_addr); cmh->cmsg_len = CMSG_LEN(sizeof(*pki)); } break; @@ -498,9 +499,13 @@ static int svc_udp_get_dest_address4(struct svc_rqst *rqstp, struct cmsghdr *cmh) { struct in_pktinfo *pki = CMSG_DATA(cmh); + struct sockaddr_in *daddr = svc_daddr_in(rqstp); + if (cmh->cmsg_type != IP_PKTINFO) return 0; - rqstp->rq_daddr.addr.s_addr = pki->ipi_spec_dst.s_addr; + + daddr->sin_family = AF_INET; + daddr->sin_addr.s_addr = pki->ipi_spec_dst.s_addr; return 1; } @@ -511,9 +516,14 @@ static int svc_udp_get_dest_address6(struct svc_rqst *rqstp, struct cmsghdr *cmh) { struct in6_pktinfo *pki = CMSG_DATA(cmh); + struct sockaddr_in6 *daddr = svc_daddr_in6(rqstp); + if (cmh->cmsg_type != IPV6_PKTINFO) return 0; - ipv6_addr_copy(&rqstp->rq_daddr.addr6, &pki->ipi6_addr); + + daddr->sin6_family = AF_INET6; + ipv6_addr_copy(&daddr->sin6_addr, &pki->ipi6_addr); + daddr->sin6_scope_id = pki->ipi6_ifindex; return 1; } @@ -614,6 +624,7 @@ static int svc_udp_recvfrom(struct svc_rqst *rqstp) skb_free_datagram_locked(svsk->sk_sk, skb); return 0; } + rqstp->rq_daddrlen = svc_addr_len(svc_daddr(rqstp)); if (skb_is_nonlinear(skb)) { /* we have to copy */ -- cgit v1.2.3-70-g09d2 From dcbf8c30344dfa01915c34f88318d464e116f004 Mon Sep 17 00:00:00 2001 From: Michal Schmidt Date: Sat, 2 Jul 2011 00:47:12 +0200 Subject: sunrpc: add MODULE_ALIAS to match the filesystem name sunrpc implements the rpc_pipefs filesystem type. Add the alias to have the module requested automatically by the kernel when the filesystem is mounted. Signed-off-by: Michal Schmidt Signed-off-by: J. Bruce Fields --- net/sunrpc/rpc_pipe.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'net/sunrpc') diff --git a/net/sunrpc/rpc_pipe.c b/net/sunrpc/rpc_pipe.c index b181e3441323..3156ba864519 100644 --- a/net/sunrpc/rpc_pipe.c +++ b/net/sunrpc/rpc_pipe.c @@ -1084,3 +1084,6 @@ void unregister_rpc_pipefs(void) kmem_cache_destroy(rpc_inode_cachep); unregister_filesystem(&rpc_pipe_fs_type); } + +/* Make 'mount -t rpc_pipefs ...' autoload this module. */ +MODULE_ALIAS("rpc_pipefs"); -- cgit v1.2.3-70-g09d2