X-Git-Url: https://git.whamcloud.com/?a=blobdiff_plain;f=lnet%2Flnet%2Facceptor.c;h=94d16fad242a83b75a9ead9e9c7044b8b48e57e3;hb=38e6ee6fb779c34c7ef3eb4a010332385b3b84c1;hp=3585d20dd1bcbca6fb68907da94009104ab244f6;hpb=7574e32e64b77851163d0a8d3141b406d03621be;p=fs%2Flustre-release.git diff --git a/lnet/lnet/acceptor.c b/lnet/lnet/acceptor.c index 3585d20..94d16fa 100644 --- a/lnet/lnet/acceptor.c +++ b/lnet/lnet/acceptor.c @@ -15,11 +15,7 @@ * * You should have received a copy of the GNU General Public License * version 2 along with this program; If not, see - * http://www.sun.com/software/products/lustre/docs/GPLv2.pdf - * - * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara, - * CA 95054 USA or visit www.sun.com if you need additional information or - * have any questions. + * http://www.gnu.org/licenses/gpl-2.0.html * * GPL HEADER END */ @@ -27,7 +23,7 @@ * Copyright (c) 2007, 2010, Oracle and/or its affiliates. All rights reserved. * Use is subject to license terms. * - * Copyright (c) 2011, 2014, Intel Corporation. + * Copyright (c) 2011, 2017, Intel Corporation. */ /* * This file is part of Lustre, http://www.lustre.org/ @@ -35,7 +31,9 @@ */ #define DEBUG_SUBSYSTEM S_LNET + #include +#include #include static int accept_port = 988; @@ -46,6 +44,7 @@ static struct { int pta_shutdown; struct socket *pta_sock; struct completion pta_signal; + struct net *pta_ns; } lnet_acceptor_state = { .pta_shutdown = 1 }; @@ -53,160 +52,154 @@ static struct { int lnet_acceptor_port(void) { - return accept_port; + return accept_port; } static inline int lnet_accept_magic(__u32 magic, __u32 constant) { - return (magic == constant || - magic == __swab32(constant)); + return (magic == constant || + magic == __swab32(constant)); } EXPORT_SYMBOL(lnet_acceptor_port); static char *accept = "secure"; -CFS_MODULE_PARM(accept, "s", charp, 0444, - "Accept connections (secure|all|none)"); -CFS_MODULE_PARM(accept_port, "i", int, 0444, - "Acceptor's port (same on all nodes)"); -CFS_MODULE_PARM(accept_backlog, "i", int, 0444, - "Acceptor's listen backlog"); -CFS_MODULE_PARM(accept_timeout, "i", int, 0644, - "Acceptor's timeout (seconds)"); +module_param(accept, charp, 0444); +MODULE_PARM_DESC(accept, "Accept connections (secure|all|none)"); +module_param(accept_port, int, 0444); +MODULE_PARM_DESC(accept_port, "Acceptor's port (same on all nodes)"); +module_param(accept_backlog, int, 0444); +MODULE_PARM_DESC(accept_backlog, "Acceptor's listen backlog"); +module_param(accept_timeout, int, 0644); +MODULE_PARM_DESC(accept_timeout, "Acceptor's timeout (seconds)"); static char *accept_type = NULL; static int lnet_acceptor_get_tunables(void) { - /* Userland acceptor uses 'accept_type' instead of 'accept', due to - * conflict with 'accept(2)', but kernel acceptor still uses 'accept' - * for compatibility. Hence the trick. */ - accept_type = accept; - return 0; + /* Userland acceptor uses 'accept_type' instead of 'accept', due to + * conflict with 'accept(2)', but kernel acceptor still uses 'accept' + * for compatibility. Hence the trick. */ + accept_type = accept; + return 0; } int lnet_acceptor_timeout(void) { - return accept_timeout; + return accept_timeout; } EXPORT_SYMBOL(lnet_acceptor_timeout); void lnet_connect_console_error (int rc, lnet_nid_t peer_nid, - __u32 peer_ip, int peer_port) + __u32 peer_ip, int peer_port) { - switch (rc) { - /* "normal" errors */ - case -ECONNREFUSED: + switch (rc) { + /* "normal" errors */ + case -ECONNREFUSED: CNETERR("Connection to %s at host %pI4h on port %d was " - "refused: check that Lustre is running on that node.\n", + "refused: check that Lustre is running on that node.\n", libcfs_nid2str(peer_nid), &peer_ip, peer_port); - break; - case -EHOSTUNREACH: - case -ENETUNREACH: + break; + case -EHOSTUNREACH: + case -ENETUNREACH: CNETERR("Connection to %s at host %pI4h " - "was unreachable: the network or that node may " - "be down, or Lustre may be misconfigured.\n", + "was unreachable: the network or that node may " + "be down, or Lustre may be misconfigured.\n", libcfs_nid2str(peer_nid), &peer_ip); - break; - case -ETIMEDOUT: + break; + case -ETIMEDOUT: CNETERR("Connection to %s at host %pI4h on " - "port %d took too long: that node may be hung " - "or experiencing high load.\n", + "port %d took too long: that node may be hung " + "or experiencing high load.\n", libcfs_nid2str(peer_nid), &peer_ip, peer_port); - break; - case -ECONNRESET: + break; + case -ECONNRESET: LCONSOLE_ERROR_MSG(0x11b, "Connection to %s at host %pI4h" - " on port %d was reset: " - "is it running a compatible version of " - "Lustre and is %s one of its NIDs?\n", + " on port %d was reset: " + "is it running a compatible version of " + "Lustre and is %s one of its NIDs?\n", libcfs_nid2str(peer_nid), &peer_ip, peer_port, libcfs_nid2str(peer_nid)); - break; - case -EPROTO: - LCONSOLE_ERROR_MSG(0x11c, "Protocol error connecting to %s at " + break; + case -EPROTO: + LCONSOLE_ERROR_MSG(0x11c, "Protocol error connecting to %s at " "host %pI4h on port %d: is it running " - "a compatible version of Lustre?\n", + "a compatible version of Lustre?\n", libcfs_nid2str(peer_nid), &peer_ip, peer_port); - break; - case -EADDRINUSE: - LCONSOLE_ERROR_MSG(0x11d, "No privileged ports available to " + break; + case -EADDRINUSE: + LCONSOLE_ERROR_MSG(0x11d, "No privileged ports available to " "connect to %s at host %pI4h on port " - "%d\n", libcfs_nid2str(peer_nid), + "%d\n", libcfs_nid2str(peer_nid), &peer_ip, peer_port); - break; - default: - LCONSOLE_ERROR_MSG(0x11e, "Unexpected error %d connecting to %s" + break; + default: + LCONSOLE_ERROR_MSG(0x11e, "Unexpected error %d connecting to %s" " at host %pI4h on port %d\n", rc, - libcfs_nid2str(peer_nid), + libcfs_nid2str(peer_nid), &peer_ip, peer_port); - break; - } + break; + } } EXPORT_SYMBOL(lnet_connect_console_error); int lnet_connect(struct socket **sockp, lnet_nid_t peer_nid, - __u32 local_ip, __u32 peer_ip, int peer_port) + __u32 local_ip, __u32 peer_ip, int peer_port, struct net *ns) { - lnet_acceptor_connreq_t cr; - struct socket *sock; - int rc; - int port; - int fatal; + struct lnet_acceptor_connreq cr; + struct socket *sock; + int rc; + int port; + int fatal; - CLASSERT (sizeof(cr) <= 16); /* not too big to be on the stack */ + BUILD_BUG_ON(sizeof(cr) > 16); /* not too big to be on the stack */ - for (port = LNET_ACCEPTOR_MAX_RESERVED_PORT; - port >= LNET_ACCEPTOR_MIN_RESERVED_PORT; - --port) { - /* Iterate through reserved ports. */ + for (port = LNET_ACCEPTOR_MAX_RESERVED_PORT; + port >= LNET_ACCEPTOR_MIN_RESERVED_PORT; + --port) { + /* Iterate through reserved ports. */ rc = lnet_sock_connect(&sock, &fatal, - local_ip, port, - peer_ip, peer_port); - if (rc != 0) { - if (fatal) - goto failed; - continue; - } - - CLASSERT (LNET_PROTO_ACCEPTOR_VERSION == 1); - - cr.acr_magic = LNET_PROTO_ACCEPTOR_MAGIC; - cr.acr_version = LNET_PROTO_ACCEPTOR_VERSION; - cr.acr_nid = peer_nid; - - if (the_lnet.ln_testprotocompat != 0) { - /* single-shot proto check */ - lnet_net_lock(LNET_LOCK_EX); - if ((the_lnet.ln_testprotocompat & 4) != 0) { - cr.acr_version++; - the_lnet.ln_testprotocompat &= ~4; - } - if ((the_lnet.ln_testprotocompat & 8) != 0) { - cr.acr_magic = LNET_PROTO_MAGIC; - the_lnet.ln_testprotocompat &= ~8; - } - lnet_net_unlock(LNET_LOCK_EX); - } + local_ip, port, + peer_ip, peer_port, ns); + if (rc != 0) { + if (fatal) + goto failed; + continue; + } + + BUILD_BUG_ON(LNET_PROTO_ACCEPTOR_VERSION != 1); + + cr.acr_magic = LNET_PROTO_ACCEPTOR_MAGIC; + cr.acr_version = LNET_PROTO_ACCEPTOR_VERSION; + cr.acr_nid = peer_nid; + + if (the_lnet.ln_testprotocompat) { + /* single-shot proto check */ + if (test_and_clear_bit(2, &the_lnet.ln_testprotocompat)) + cr.acr_version++; + if (test_and_clear_bit(3, &the_lnet.ln_testprotocompat)) + cr.acr_magic = LNET_PROTO_MAGIC; + } rc = lnet_sock_write(sock, &cr, sizeof(cr), - accept_timeout); - if (rc != 0) - goto failed_sock; + accept_timeout); + if (rc != 0) + goto failed_sock; - *sockp = sock; - return 0; - } + *sockp = sock; + return 0; + } - rc = -EADDRINUSE; - goto failed; + rc = -EADDRINUSE; + goto failed; failed_sock: sock_release(sock); @@ -219,167 +212,166 @@ EXPORT_SYMBOL(lnet_connect); static int lnet_accept(struct socket *sock, __u32 magic) { - lnet_acceptor_connreq_t cr; - __u32 peer_ip; - int peer_port; - int rc; - int flip; - lnet_ni_t *ni; - char *str; + struct lnet_acceptor_connreq cr; + __u32 peer_ip; + int peer_port; + int rc; + int flip; + struct lnet_ni *ni; + char *str; - LASSERT (sizeof(cr) <= 16); /* not too big for the stack */ + LASSERT(sizeof(cr) <= 16); /* not too big for the stack */ rc = lnet_sock_getaddr(sock, true, &peer_ip, &peer_port); - LASSERT (rc == 0); /* we succeeded before */ + LASSERT(rc == 0); /* we succeeded before */ - if (!lnet_accept_magic(magic, LNET_PROTO_ACCEPTOR_MAGIC)) { + if (!lnet_accept_magic(magic, LNET_PROTO_ACCEPTOR_MAGIC)) { - if (lnet_accept_magic(magic, LNET_PROTO_MAGIC)) { - /* future version compatibility! - * When LNET unifies protocols over all LNDs, the first - * thing sent will be a version query. I send back - * LNET_PROTO_ACCEPTOR_MAGIC to tell her I'm "old" */ + if (lnet_accept_magic(magic, LNET_PROTO_MAGIC)) { + /* future version compatibility! + * When LNET unifies protocols over all LNDs, the first + * thing sent will be a version query. I send back + * LNET_PROTO_ACCEPTOR_MAGIC to tell her I'm "old" */ - memset (&cr, 0, sizeof(cr)); - cr.acr_magic = LNET_PROTO_ACCEPTOR_MAGIC; - cr.acr_version = LNET_PROTO_ACCEPTOR_VERSION; + memset(&cr, 0, sizeof(cr)); + cr.acr_magic = LNET_PROTO_ACCEPTOR_MAGIC; + cr.acr_version = LNET_PROTO_ACCEPTOR_VERSION; rc = lnet_sock_write(sock, &cr, sizeof(cr), - accept_timeout); + accept_timeout); - if (rc != 0) - CERROR("Error sending magic+version in response" - "to LNET magic from %pI4h: %d\n", + if (rc != 0) + CERROR("Error sending magic+version in response to LNET magic from %pI4h: %d\n", &peer_ip, rc); - return -EPROTO; - } + return -EPROTO; + } - if (magic == le32_to_cpu(LNET_PROTO_TCP_MAGIC)) - str = "'old' socknal/tcpnal"; - else - str = "unrecognised"; + if (lnet_accept_magic(magic, LNET_PROTO_TCP_MAGIC)) + str = "'old' socknal/tcpnal"; + else + str = "unrecognised"; LCONSOLE_ERROR_MSG(0x11f, "Refusing connection from %pI4h" - " magic %08x: %s acceptor protocol\n", + " magic %08x: %s acceptor protocol\n", &peer_ip, magic, str); - return -EPROTO; - } + return -EPROTO; + } - flip = (magic != LNET_PROTO_ACCEPTOR_MAGIC); + flip = (magic != LNET_PROTO_ACCEPTOR_MAGIC); rc = lnet_sock_read(sock, &cr.acr_version, - sizeof(cr.acr_version), - accept_timeout); - if (rc != 0) { - CERROR("Error %d reading connection request version from " + sizeof(cr.acr_version), + accept_timeout); + if (rc != 0) { + CERROR("Error %d reading connection request version from " "%pI4h\n", rc, &peer_ip); - return -EIO; - } + return -EIO; + } - if (flip) - __swab32s(&cr.acr_version); + if (flip) + __swab32s(&cr.acr_version); - if (cr.acr_version != LNET_PROTO_ACCEPTOR_VERSION) { - /* future version compatibility! - * An acceptor-specific protocol rev will first send a version - * query. I send back my current version to tell her I'm - * "old". */ - int peer_version = cr.acr_version; + if (cr.acr_version != LNET_PROTO_ACCEPTOR_VERSION) { + /* future version compatibility! + * An acceptor-specific protocol rev will first send a version + * query. I send back my current version to tell her I'm + * "old". */ + int peer_version = cr.acr_version; - memset (&cr, 0, sizeof(cr)); - cr.acr_magic = LNET_PROTO_ACCEPTOR_MAGIC; - cr.acr_version = LNET_PROTO_ACCEPTOR_VERSION; + memset(&cr, 0, sizeof(cr)); + cr.acr_magic = LNET_PROTO_ACCEPTOR_MAGIC; + cr.acr_version = LNET_PROTO_ACCEPTOR_VERSION; rc = lnet_sock_write(sock, &cr, sizeof(cr), - accept_timeout); + accept_timeout); - if (rc != 0) - CERROR("Error sending magic+version in response" - "to version %d from %pI4h: %d\n", + if (rc != 0) + CERROR("Error sending magic+version in response to version %d from %pI4h: %d\n", peer_version, &peer_ip, rc); - return -EPROTO; - } + return -EPROTO; + } rc = lnet_sock_read(sock, &cr.acr_nid, - sizeof(cr) - - offsetof(lnet_acceptor_connreq_t, acr_nid), - accept_timeout); - if (rc != 0) { - CERROR("Error %d reading connection request from " + sizeof(cr) - + offsetof(struct lnet_acceptor_connreq, acr_nid), + accept_timeout); + if (rc != 0) { + CERROR("Error %d reading connection request from " "%pI4h\n", rc, &peer_ip); - return -EIO; - } + return -EIO; + } - if (flip) - __swab64s(&cr.acr_nid); + if (flip) + __swab64s(&cr.acr_nid); - ni = lnet_net2ni(LNET_NIDNET(cr.acr_nid)); - if (ni == NULL || /* no matching net */ - ni->ni_nid != cr.acr_nid) { /* right NET, wrong NID! */ - if (ni != NULL) - lnet_ni_decref(ni); + ni = lnet_nid2ni_addref(cr.acr_nid); + if (ni == NULL || /* no matching net */ + ni->ni_nid != cr.acr_nid) { /* right NET, wrong NID! */ + if (ni != NULL) + lnet_ni_decref(ni); LCONSOLE_ERROR_MSG(0x120, "Refusing connection from %pI4h " "for %s: No matching NI\n", &peer_ip, libcfs_nid2str(cr.acr_nid)); - return -EPERM; - } + return -EPERM; + } - if (ni->ni_lnd->lnd_accept == NULL) { - /* This catches a request for the loopback LND */ - lnet_ni_decref(ni); + if (ni->ni_net->net_lnd->lnd_accept == NULL) { + /* This catches a request for the loopback LND */ + lnet_ni_decref(ni); LCONSOLE_ERROR_MSG(0x121, "Refusing connection from %pI4h " "for %s: NI doesn not accept IP connections\n", &peer_ip, libcfs_nid2str(cr.acr_nid)); - return -EPERM; - } + return -EPERM; + } CDEBUG(D_NET, "Accept %s from %pI4h\n", libcfs_nid2str(cr.acr_nid), &peer_ip); - rc = ni->ni_lnd->lnd_accept(ni, sock); + rc = ni->ni_net->net_lnd->lnd_accept(ni, sock); - lnet_ni_decref(ni); - return rc; + lnet_ni_decref(ni); + return rc; } static int lnet_acceptor(void *arg) { struct socket *newsock; - int rc; - __u32 magic; - __u32 peer_ip; - int peer_port; - int secure = (int)((long_ptr_t)arg); + int rc; + __u32 magic; + __u32 peer_ip; + int peer_port; + int secure = (int)((uintptr_t)arg); - LASSERT (lnet_acceptor_state.pta_sock == NULL); + LASSERT(lnet_acceptor_state.pta_sock == NULL); - cfs_block_allsigs(); + cfs_block_allsigs(); rc = lnet_sock_listen(&lnet_acceptor_state.pta_sock, - 0, accept_port, accept_backlog); - if (rc != 0) { - if (rc == -EADDRINUSE) - LCONSOLE_ERROR_MSG(0x122, "Can't start acceptor on port" - " %d: port already in use\n", - accept_port); - else - LCONSOLE_ERROR_MSG(0x123, "Can't start acceptor on port " - "%d: unexpected error %d\n", - accept_port, rc); - - lnet_acceptor_state.pta_sock = NULL; - } else { - LCONSOLE(0, "Accept %s, port %d\n", accept_type, accept_port); - } + 0, accept_port, accept_backlog, + lnet_acceptor_state.pta_ns); + if (rc != 0) { + if (rc == -EADDRINUSE) + LCONSOLE_ERROR_MSG(0x122, "Can't start acceptor on port" + " %d: port already in use\n", + accept_port); + else + LCONSOLE_ERROR_MSG(0x123, "Can't start acceptor on port " + "%d: unexpected error %d\n", + accept_port, rc); + + lnet_acceptor_state.pta_sock = NULL; + } else { + LCONSOLE(0, "Accept %s, port %d\n", accept_type, accept_port); + } /* set init status and unblock parent */ lnet_acceptor_state.pta_shutdown = rc; complete(&lnet_acceptor_state.pta_signal); - if (rc != 0) - return rc; + if (rc != 0) + return rc; - while (!lnet_acceptor_state.pta_shutdown) { + while (!lnet_acceptor_state.pta_shutdown) { rc = lnet_sock_accept(&newsock, lnet_acceptor_state.pta_sock); if (rc != 0) { @@ -398,30 +390,30 @@ lnet_acceptor(void *arg) } rc = lnet_sock_getaddr(newsock, true, &peer_ip, &peer_port); - if (rc != 0) { - CERROR("Can't determine new connection's address\n"); - goto failed; - } + if (rc != 0) { + CERROR("Can't determine new connection's address\n"); + goto failed; + } - if (secure && peer_port > LNET_ACCEPTOR_MAX_RESERVED_PORT) { + if (secure && peer_port > LNET_ACCEPTOR_MAX_RESERVED_PORT) { CERROR("Refusing connection from %pI4h: " "insecure port %d\n", &peer_ip, peer_port); - goto failed; - } + goto failed; + } rc = lnet_sock_read(newsock, &magic, sizeof(magic), - accept_timeout); - if (rc != 0) { - CERROR("Error %d reading connection request from " + accept_timeout); + if (rc != 0) { + CERROR("Error %d reading connection request from " "%pI4h\n", rc, &peer_ip); - goto failed; - } + goto failed; + } - rc = lnet_accept(newsock, magic); - if (rc != 0) - goto failed; + rc = lnet_accept(newsock, magic); + if (rc != 0) + goto failed; - continue; + continue; failed: sock_release(newsock); @@ -440,53 +432,55 @@ failed: static inline int accept2secure(const char *acc, long *sec) { - if (!strcmp(acc, "secure")) { - *sec = 1; - return 1; - } else if (!strcmp(acc, "all")) { - *sec = 0; - return 1; - } else if (!strcmp(acc, "none")) { - return 0; - } else { - LCONSOLE_ERROR_MSG(0x124, "Can't parse 'accept=\"%s\"'\n", - acc); - return -EINVAL; - } + if (!strcmp(acc, "secure")) { + *sec = 1; + return 1; + } else if (!strcmp(acc, "all")) { + *sec = 0; + return 1; + } else if (!strcmp(acc, "none")) { + return 0; + } else { + LCONSOLE_ERROR_MSG(0x124, "Can't parse 'accept=\"%s\"'\n", + acc); + return -EINVAL; + } } int lnet_acceptor_start(void) { struct task_struct *task; - int rc; - long rc2; - long secure; + int rc; + long rc2; + long secure; /* if acceptor is already running return immediately */ if (!lnet_acceptor_state.pta_shutdown) return 0; - LASSERT (lnet_acceptor_state.pta_sock == NULL); + LASSERT(lnet_acceptor_state.pta_sock == NULL); - rc = lnet_acceptor_get_tunables(); - if (rc != 0) - return rc; + rc = lnet_acceptor_get_tunables(); + if (rc != 0) + return rc; init_completion(&lnet_acceptor_state.pta_signal); rc = accept2secure(accept_type, &secure); if (rc <= 0) return rc; - if (lnet_count_acceptor_nis() == 0) /* not required */ + if (lnet_count_acceptor_nets() == 0) /* not required */ return 0; - - task = kthread_run(lnet_acceptor, (void *)(ulong_ptr_t)secure, + if (current->nsproxy && current->nsproxy->net_ns) + lnet_acceptor_state.pta_ns = current->nsproxy->net_ns; + else + lnet_acceptor_state.pta_ns = &init_net; + task = kthread_run(lnet_acceptor, (void *)(uintptr_t)secure, "acceptor_%03ld", secure); if (IS_ERR(task)) { rc2 = PTR_ERR(task); CERROR("Can't start acceptor thread: %ld\n", rc2); - return -ESRCH; }