X-Git-Url: https://git.whamcloud.com/?p=fs%2Flustre-release.git;a=blobdiff_plain;f=lnet%2Finclude%2Flnet%2Flib-lnet.h;h=97ca7a457249ef234351e6a748a086af460e08af;hp=9cebc3af8fff5dc1a37b113b883f6e9b7ce96e39;hb=63c3e512987308e45005800a7e080ce2a55b1301;hpb=e8a1124e74f027779068eb7931f10035e0bc6f1d diff --git a/lnet/include/lnet/lib-lnet.h b/lnet/include/lnet/lib-lnet.h index 9cebc3a..97ca7a4 100644 --- a/lnet/include/lnet/lib-lnet.h +++ b/lnet/include/lnet/lib-lnet.h @@ -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) 2003, 2010, Oracle and/or its affiliates. All rights reserved. * Use is subject to license terms. * - * Copyright (c) 2012, 2013, Intel Corporation. + * Copyright (c) 2012, 2016, Intel Corporation. */ /* * This file is part of Lustre, http://www.lustre.org/ @@ -41,91 +37,78 @@ #ifndef __LNET_LIB_LNET_H__ #define __LNET_LIB_LNET_H__ -#if defined(__linux__) -#include -#elif defined(__APPLE__) -#include -#elif defined(__WINNT__) -#include -#else -#error Unsupported Operating System +#ifndef __KERNEL__ +# error This include is only for kernel use. #endif #include -#include +#include #include #include -extern lnet_t the_lnet; /* THE network */ - -#if !defined(__KERNEL__) || defined(LNET_USE_LIB_FREELIST) -/* 1 CPT, simplify implementation... */ -# define LNET_CPT_MAX_BITS 0 - -#else /* KERNEL and no freelist */ +extern lnet_t the_lnet; /* THE network */ -# if (BITS_PER_LONG == 32) +#if (BITS_PER_LONG == 32) /* 2 CPTs, allowing more CPTs might make us under memory pressure */ -# define LNET_CPT_MAX_BITS 1 +# define LNET_CPT_MAX_BITS 1 -# else /* 64-bit system */ +#else /* 64-bit system */ /* * 256 CPTs for thousands of CPUs, allowing more CPTs might make us * under risk of consuming all lh_cookie. */ -# define LNET_CPT_MAX_BITS 8 -# endif /* BITS_PER_LONG == 32 */ -#endif +# define LNET_CPT_MAX_BITS 8 +#endif /* BITS_PER_LONG == 32 */ /* max allowed CPT number */ -#define LNET_CPT_MAX (1 << LNET_CPT_MAX_BITS) +#define LNET_CPT_MAX (1 << LNET_CPT_MAX_BITS) -#define LNET_CPT_NUMBER (the_lnet.ln_cpt_number) -#define LNET_CPT_BITS (the_lnet.ln_cpt_bits) -#define LNET_CPT_MASK ((1ULL << LNET_CPT_BITS) - 1) +#define LNET_CPT_NUMBER (the_lnet.ln_cpt_number) +#define LNET_CPT_BITS (the_lnet.ln_cpt_bits) +#define LNET_CPT_MASK ((1ULL << LNET_CPT_BITS) - 1) /** exclusive lock */ -#define LNET_LOCK_EX CFS_PERCPT_LOCK_EX +#define LNET_LOCK_EX CFS_PERCPT_LOCK_EX static inline int lnet_is_route_alive(lnet_route_t *route) { - if (!route->lr_gateway->lp_alive) + if (!route->lr_gateway->lpni_alive) return 0; /* gateway is down */ - if ((route->lr_gateway->lp_ping_feats & + if ((route->lr_gateway->lpni_ping_feats & LNET_PING_FEAT_NI_STATUS) == 0) return 1; /* no NI status, assume it's alive */ /* has NI status, check # down NIs */ return route->lr_downis == 0; } -static inline int lnet_is_wire_handle_none (lnet_handle_wire_t *wh) +static inline int lnet_is_wire_handle_none(struct lnet_handle_wire *wh) { - return (wh->wh_interface_cookie == LNET_WIRE_HANDLE_COOKIE_NONE && - wh->wh_object_cookie == LNET_WIRE_HANDLE_COOKIE_NONE); + return (wh->wh_interface_cookie == LNET_WIRE_HANDLE_COOKIE_NONE && + wh->wh_object_cookie == LNET_WIRE_HANDLE_COOKIE_NONE); } static inline int lnet_md_exhausted (lnet_libmd_t *md) { - return (md->md_threshold == 0 || - ((md->md_options & LNET_MD_MAX_SIZE) != 0 && - md->md_offset + md->md_max_size > md->md_length)); + return (md->md_threshold == 0 || + ((md->md_options & LNET_MD_MAX_SIZE) != 0 && + md->md_offset + md->md_max_size > md->md_length)); } static inline int lnet_md_unlinkable (lnet_libmd_t *md) { - /* Should unlink md when its refcount is 0 and either: - * - md has been flagged for deletion (by auto unlink or LNetM[DE]Unlink, - * in the latter case md may not be exhausted). - * - auto unlink is on and md is exhausted. - */ - if (md->md_refcount != 0) - return 0; + /* Should unlink md when its refcount is 0 and either: + * - md has been flagged for deletion (by auto unlink or LNetM[DE]Unlink, + * in the latter case md may not be exhausted). + * - auto unlink is on and md is exhausted. + */ + if (md->md_refcount != 0) + return 0; - if ((md->md_flags & LNET_MD_FLAG_ZOMBIE) != 0) - return 1; + if ((md->md_flags & LNET_MD_FLAG_ZOMBIE) != 0) + return 1; - return ((md->md_flags & LNET_MD_FLAG_AUTO_UNLINK) != 0 && - lnet_md_exhausted(md)); + return ((md->md_flags & LNET_MD_FLAG_AUTO_UNLINK) != 0 && + lnet_md_exhausted(md)); } #define lnet_cpt_table() (the_lnet.ln_cpt_table) @@ -186,301 +169,79 @@ lnet_net_lock_current(void) #define LNET_LOCK() lnet_net_lock(LNET_LOCK_EX) #define LNET_UNLOCK() lnet_net_unlock(LNET_LOCK_EX) -#ifdef __KERNEL__ - #define lnet_ptl_lock(ptl) spin_lock(&(ptl)->ptl_lock) #define lnet_ptl_unlock(ptl) spin_unlock(&(ptl)->ptl_lock) #define lnet_eq_wait_lock() spin_lock(&the_lnet.ln_eq_wait_lock) #define lnet_eq_wait_unlock() spin_unlock(&the_lnet.ln_eq_wait_lock) #define lnet_ni_lock(ni) spin_lock(&(ni)->ni_lock) #define lnet_ni_unlock(ni) spin_unlock(&(ni)->ni_lock) -#define LNET_MUTEX_LOCK(m) mutex_lock(m) -#define LNET_MUTEX_UNLOCK(m) mutex_unlock(m) - -#else /* !__KERNEL__ */ - -# ifndef HAVE_LIBPTHREAD -#define LNET_SINGLE_THREADED_LOCK(l) \ -do { \ - LASSERT ((l) == 0); \ - (l) = 1; \ -} while (0) - -#define LNET_SINGLE_THREADED_UNLOCK(l) \ -do { \ - LASSERT ((l) == 1); \ - (l) = 0; \ -} while (0) - -#define LNET_MUTEX_LOCK(m) LNET_SINGLE_THREADED_LOCK(*(m)) -#define LNET_MUTEX_UNLOCK(m) LNET_SINGLE_THREADED_UNLOCK(*(m)) - -#define lnet_ptl_lock(ptl) \ - LNET_SINGLE_THREADED_LOCK((ptl)->ptl_lock) -#define lnet_ptl_unlock(ptl) \ - LNET_SINGLE_THREADED_UNLOCK((ptl)->ptl_lock) - -#define lnet_eq_wait_lock() \ - LNET_SINGLE_THREADED_LOCK(the_lnet.ln_eq_wait_lock) -#define lnet_eq_wait_unlock() \ - LNET_SINGLE_THREADED_UNLOCK(the_lnet.ln_eq_wait_lock) - -#define lnet_ni_lock(ni) \ - LNET_SINGLE_THREADED_LOCK((ni)->ni_lock) -#define lnet_ni_unlock(ni) \ - LNET_SINGLE_THREADED_UNLOCK((ni)->ni_lock) - -# else /* HAVE_LIBPTHREAD */ - -#define LNET_MUTEX_LOCK(m) pthread_mutex_lock(m) -#define LNET_MUTEX_UNLOCK(m) pthread_mutex_unlock(m) - -#define lnet_ptl_lock(ptl) pthread_mutex_lock(&(ptl)->ptl_lock) -#define lnet_ptl_unlock(ptl) pthread_mutex_unlock(&(ptl)->ptl_lock) - -#define lnet_eq_wait_lock() pthread_mutex_lock(&the_lnet.ln_eq_wait_lock) -#define lnet_eq_wait_unlock() pthread_mutex_unlock(&the_lnet.ln_eq_wait_lock) - -#define lnet_ni_lock(ni) pthread_mutex_lock(&(ni)->ni_lock) -#define lnet_ni_unlock(ni) pthread_mutex_unlock(&(ni)->ni_lock) - -# endif /* HAVE_LIBPTHREAD */ -#endif /* __KERNEL__ */ - -#define MAX_PORTALS 64 - -/* these are only used by code with LNET_USE_LIB_FREELIST, but we still - * exported them to !LNET_USE_LIB_FREELIST for easy implemetation */ -#define LNET_FL_MAX_MES 2048 -#define LNET_FL_MAX_MDS 2048 -#define LNET_FL_MAX_EQS 512 -#define LNET_FL_MAX_MSGS 2048 /* Outstanding messages */ - -#ifdef LNET_USE_LIB_FREELIST - -int lnet_freelist_init(lnet_freelist_t *fl, int n, int size); -void lnet_freelist_fini(lnet_freelist_t *fl); - -static inline void * -lnet_freelist_alloc (lnet_freelist_t *fl) -{ - /* ALWAYS called with liblock held */ - lnet_freeobj_t *o; - - if (cfs_list_empty (&fl->fl_list)) - return (NULL); - - o = cfs_list_entry (fl->fl_list.next, lnet_freeobj_t, fo_list); - cfs_list_del (&o->fo_list); - return ((void *)&o->fo_contents); -} - -static inline void -lnet_freelist_free (lnet_freelist_t *fl, void *obj) -{ - /* ALWAYS called with liblock held */ - lnet_freeobj_t *o = cfs_list_entry (obj, lnet_freeobj_t, fo_contents); - - cfs_list_add (&o->fo_list, &fl->fl_list); -} - - -static inline lnet_eq_t * -lnet_eq_alloc (void) -{ - /* NEVER called with resource lock held */ - struct lnet_res_container *rec = &the_lnet.ln_eq_container; - lnet_eq_t *eq; - - LASSERT(LNET_CPT_NUMBER == 1); - - lnet_res_lock(0); - eq = (lnet_eq_t *)lnet_freelist_alloc(&rec->rec_freelist); - lnet_res_unlock(0); - - return eq; -} - -static inline void -lnet_eq_free_locked(lnet_eq_t *eq) -{ - /* ALWAYS called with resource lock held */ - struct lnet_res_container *rec = &the_lnet.ln_eq_container; - - LASSERT(LNET_CPT_NUMBER == 1); - lnet_freelist_free(&rec->rec_freelist, eq); -} - -static inline void -lnet_eq_free(lnet_eq_t *eq) -{ - lnet_res_lock(0); - lnet_eq_free_locked(eq); - lnet_res_unlock(0); -} - -static inline lnet_libmd_t * -lnet_md_alloc (lnet_md_t *umd) -{ - /* NEVER called with resource lock held */ - struct lnet_res_container *rec = the_lnet.ln_md_containers[0]; - lnet_libmd_t *md; - - LASSERT(LNET_CPT_NUMBER == 1); - - lnet_res_lock(0); - md = (lnet_libmd_t *)lnet_freelist_alloc(&rec->rec_freelist); - lnet_res_unlock(0); - - if (md != NULL) - CFS_INIT_LIST_HEAD(&md->md_list); - - return md; -} - -static inline void -lnet_md_free_locked(lnet_libmd_t *md) -{ - /* ALWAYS called with resource lock held */ - struct lnet_res_container *rec = the_lnet.ln_md_containers[0]; - - LASSERT(LNET_CPT_NUMBER == 1); - lnet_freelist_free(&rec->rec_freelist, md); -} - -static inline void -lnet_md_free(lnet_libmd_t *md) -{ - lnet_res_lock(0); - lnet_md_free_locked(md); - lnet_res_unlock(0); -} - -static inline lnet_me_t * -lnet_me_alloc(void) -{ - /* NEVER called with resource lock held */ - struct lnet_res_container *rec = the_lnet.ln_me_containers[0]; - lnet_me_t *me; - - LASSERT(LNET_CPT_NUMBER == 1); - - lnet_res_lock(0); - me = (lnet_me_t *)lnet_freelist_alloc(&rec->rec_freelist); - lnet_res_unlock(0); - - return me; -} - -static inline void -lnet_me_free_locked(lnet_me_t *me) -{ - /* ALWAYS called with resource lock held */ - struct lnet_res_container *rec = the_lnet.ln_me_containers[0]; - - LASSERT(LNET_CPT_NUMBER == 1); - lnet_freelist_free(&rec->rec_freelist, me); -} - -static inline void -lnet_me_free(lnet_me_t *me) -{ - lnet_res_lock(0); - lnet_me_free_locked(me); - lnet_res_unlock(0); -} - -static inline lnet_msg_t * -lnet_msg_alloc (void) -{ - /* NEVER called with network lock held */ - struct lnet_msg_container *msc = the_lnet.ln_msg_containers[0]; - lnet_msg_t *msg; - - LASSERT(LNET_CPT_NUMBER == 1); - - lnet_net_lock(0); - msg = (lnet_msg_t *)lnet_freelist_alloc(&msc->msc_freelist); - lnet_net_unlock(0); - - if (msg != NULL) { - /* NULL pointers, clear flags etc */ - memset(msg, 0, sizeof(*msg)); - } - return msg; -} - -static inline void -lnet_msg_free_locked(lnet_msg_t *msg) -{ - /* ALWAYS called with network lock held */ - struct lnet_msg_container *msc = the_lnet.ln_msg_containers[0]; - LASSERT(LNET_CPT_NUMBER == 1); - LASSERT(!msg->msg_onactivelist); - lnet_freelist_free(&msc->msc_freelist, msg); -} +#define MAX_PORTALS 64 -static inline void -lnet_msg_free (lnet_msg_t *msg) -{ - lnet_net_lock(0); - lnet_msg_free_locked(msg); - lnet_net_unlock(0); -} - -#else /* !LNET_USE_LIB_FREELIST */ +#define LNET_SMALL_MD_SIZE offsetof(lnet_libmd_t, md_iov.iov[1]) +extern struct kmem_cache *lnet_mes_cachep; /* MEs kmem_cache */ +extern struct kmem_cache *lnet_small_mds_cachep; /* <= LNET_SMALL_MD_SIZE bytes + * MDs kmem_cache */ static inline lnet_eq_t * lnet_eq_alloc (void) { - /* NEVER called with liblock held */ - lnet_eq_t *eq; + lnet_eq_t *eq; - LIBCFS_ALLOC(eq, sizeof(*eq)); - return (eq); + LIBCFS_ALLOC(eq, sizeof(*eq)); + return (eq); } static inline void lnet_eq_free(lnet_eq_t *eq) { - /* ALWAYS called with resource lock held */ LIBCFS_FREE(eq, sizeof(*eq)); } static inline lnet_libmd_t * lnet_md_alloc (lnet_md_t *umd) { - /* NEVER called with liblock held */ - lnet_libmd_t *md; - unsigned int size; - unsigned int niov; - - if ((umd->options & LNET_MD_KIOV) != 0) { - niov = umd->length; - size = offsetof(lnet_libmd_t, md_iov.kiov[niov]); - } else { - niov = ((umd->options & LNET_MD_IOVEC) != 0) ? - umd->length : 1; - size = offsetof(lnet_libmd_t, md_iov.iov[niov]); - } + lnet_libmd_t *md; + unsigned int size; + unsigned int niov; + + if ((umd->options & LNET_MD_KIOV) != 0) { + niov = umd->length; + size = offsetof(lnet_libmd_t, md_iov.kiov[niov]); + } else { + niov = ((umd->options & LNET_MD_IOVEC) != 0) ? + umd->length : 1; + size = offsetof(lnet_libmd_t, md_iov.iov[niov]); + } - LIBCFS_ALLOC(md, size); + if (size <= LNET_SMALL_MD_SIZE) { + md = kmem_cache_alloc(lnet_small_mds_cachep, + GFP_NOFS | __GFP_ZERO); + if (md) { + CDEBUG(D_MALLOC, "slab-alloced 'md' of size %u at " + "%p.\n", size, md); + } else { + CDEBUG(D_MALLOC, "failed to allocate 'md' of size %u\n", + size); + return NULL; + } + } else { + LIBCFS_ALLOC(md, size); + } - if (md != NULL) { - /* Set here in case of early free */ - md->md_options = umd->options; - md->md_niov = niov; - CFS_INIT_LIST_HEAD(&md->md_list); - } + if (md != NULL) { + /* Set here in case of early free */ + md->md_options = umd->options; + md->md_niov = niov; + INIT_LIST_HEAD(&md->md_list); + } - return (md); + return md; } static inline void lnet_md_free(lnet_libmd_t *md) { - /* ALWAYS called with resource lock held */ unsigned int size; if ((md->md_options & LNET_MD_KIOV) != 0) @@ -488,53 +249,36 @@ lnet_md_free(lnet_libmd_t *md) else size = offsetof(lnet_libmd_t, md_iov.iov[md->md_niov]); - LIBCFS_FREE(md, size); + if (size <= LNET_SMALL_MD_SIZE) { + CDEBUG(D_MALLOC, "slab-freed 'md' at %p.\n", md); + kmem_cache_free(lnet_small_mds_cachep, md); + } else { + LIBCFS_FREE(md, size); + } } static inline lnet_me_t * lnet_me_alloc (void) { - /* NEVER called with liblock held */ - lnet_me_t *me; - - LIBCFS_ALLOC(me, sizeof(*me)); - return (me); -} - -static inline void -lnet_me_free(lnet_me_t *me) -{ - /* ALWAYS called with resource lock held */ - LIBCFS_FREE(me, sizeof(*me)); -} + lnet_me_t *me; -static inline lnet_msg_t * -lnet_msg_alloc(void) -{ - /* NEVER called with liblock held */ - lnet_msg_t *msg; + me = kmem_cache_alloc(lnet_mes_cachep, GFP_NOFS | __GFP_ZERO); - LIBCFS_ALLOC(msg, sizeof(*msg)); + if (me) + CDEBUG(D_MALLOC, "slab-alloced 'me' at %p.\n", me); + else + CDEBUG(D_MALLOC, "failed to allocate 'me'\n"); - /* no need to zero, LIBCFS_ALLOC does for us */ - return (msg); + return me; } static inline void -lnet_msg_free(lnet_msg_t *msg) +lnet_me_free(lnet_me_t *me) { - /* ALWAYS called with network lock held */ - LASSERT(!msg->msg_onactivelist); - LIBCFS_FREE(msg, sizeof(*msg)); + CDEBUG(D_MALLOC, "slab-freed 'me' at %p.\n", me); + kmem_cache_free(lnet_mes_cachep, me); } -#define lnet_eq_free_locked(eq) lnet_eq_free(eq) -#define lnet_md_free_locked(md) lnet_md_free(md) -#define lnet_me_free_locked(me) lnet_me_free(me) -#define lnet_msg_free_locked(msg) lnet_msg_free(msg) - -#endif /* LNET_USE_LIB_FREELIST */ - lnet_libhandle_t *lnet_res_lh_lookup(struct lnet_res_container *rec, __u64 cookie); void lnet_res_lh_initialize(struct lnet_res_container *rec, @@ -544,18 +288,18 @@ lnet_res_lh_invalidate(lnet_libhandle_t *lh) { /* ALWAYS called with resource lock held */ /* NB: cookie is still useful, don't reset it */ - cfs_list_del(&lh->lh_hash_chain); + list_del(&lh->lh_hash_chain); } static inline void lnet_eq2handle (lnet_handle_eq_t *handle, lnet_eq_t *eq) { - if (eq == NULL) { - LNetInvalidateHandle(handle); - return; - } + if (eq == NULL) { + LNetInvalidateHandle(handle); + return; + } - handle->cookie = eq->eq_lh.lh_cookie; + handle->cookie = eq->eq_lh.lh_cookie; } static inline lnet_eq_t * @@ -574,7 +318,7 @@ lnet_handle2eq(lnet_handle_eq_t *handle) static inline void lnet_md2handle (lnet_handle_md_t *handle, lnet_libmd_t *md) { - handle->cookie = md->md_lh.lh_cookie; + handle->cookie = md->md_lh.lh_cookie; } static inline lnet_libmd_t * @@ -594,7 +338,7 @@ lnet_handle2md(lnet_handle_md_t *handle) } static inline lnet_libmd_t * -lnet_wire_handle2md(lnet_handle_wire_t *wh) +lnet_wire_handle2md(struct lnet_handle_wire *wh) { /* ALWAYS called with resource lock held */ lnet_libhandle_t *lh; @@ -615,7 +359,7 @@ lnet_wire_handle2md(lnet_handle_wire_t *wh) static inline void lnet_me2handle (lnet_handle_me_t *handle, lnet_me_t *me) { - handle->cookie = me->me_lh.lh_cookie; + handle->cookie = me->me_lh.lh_cookie; } static inline lnet_me_t * @@ -635,27 +379,27 @@ lnet_handle2me(lnet_handle_me_t *handle) } static inline void -lnet_peer_addref_locked(lnet_peer_t *lp) +lnet_peer_ni_addref_locked(struct lnet_peer_ni *lp) { - LASSERT (lp->lp_refcount > 0); - lp->lp_refcount++; + LASSERT (atomic_read(&lp->lpni_refcount) > 0); + atomic_inc(&lp->lpni_refcount); } -extern void lnet_destroy_peer_locked(lnet_peer_t *lp); +extern void lnet_destroy_peer_ni_locked(struct lnet_peer_ni *lp); static inline void -lnet_peer_decref_locked(lnet_peer_t *lp) +lnet_peer_ni_decref_locked(struct lnet_peer_ni *lp) { - LASSERT (lp->lp_refcount > 0); - lp->lp_refcount--; - if (lp->lp_refcount == 0) - lnet_destroy_peer_locked(lp); + LASSERT (atomic_read(&lp->lpni_refcount) > 0); + atomic_dec(&lp->lpni_refcount); + if (atomic_read(&lp->lpni_refcount) == 0) + lnet_destroy_peer_ni_locked(lp); } static inline int -lnet_isrouter(lnet_peer_t *lp) +lnet_isrouter(struct lnet_peer_ni *lp) { - return lp->lp_rtr_refcount != 0; + return lp->lpni_rtr_refcount != 0; } static inline void @@ -692,15 +436,44 @@ lnet_ni_decref(lnet_ni_t *ni) lnet_net_unlock(0); } -void lnet_ni_free(lnet_ni_t *ni); +static inline lnet_msg_t * +lnet_msg_alloc(void) +{ + lnet_msg_t *msg; + + LIBCFS_ALLOC(msg, sizeof(*msg)); + + /* no need to zero, LIBCFS_ALLOC does for us */ + return (msg); +} + +static inline void +lnet_msg_free(lnet_msg_t *msg) +{ + LASSERT(!msg->msg_onactivelist); + LIBCFS_FREE(msg, sizeof(*msg)); +} + +void lnet_ni_free(struct lnet_ni *ni); +void lnet_net_free(struct lnet_net *net); + +struct lnet_net * +lnet_net_alloc(__u32 net_type, struct list_head *netlist); + +struct lnet_ni * +lnet_ni_alloc(struct lnet_net *net, struct cfs_expr_list *el, + char *iface); +struct lnet_ni * +lnet_ni_alloc_w_cpt_array(struct lnet_net *net, __u32 *cpts, __u32 ncpts, + char *iface); static inline int lnet_nid2peerhash(lnet_nid_t nid) { - return cfs_hash_long(nid, LNET_PEER_HASH_BITS); + return hash_long(nid, LNET_PEER_HASH_BITS); } -static inline cfs_list_t * +static inline struct list_head * lnet_net2rnethash(__u32 net) { return &the_lnet.ln_remote_nets_hash[(LNET_NETNUM(net) + @@ -709,46 +482,69 @@ lnet_net2rnethash(__u32 net) } extern lnd_t the_lolnd; +extern int avoid_asym_router_failure; -#ifndef __KERNEL__ -/* unconditional registration */ -#define LNET_REGISTER_ULND(lnd) \ -do { \ - extern lnd_t lnd; \ - \ - lnet_register_lnd(&(lnd)); \ -} while (0) - -/* conditional registration */ -#define LNET_REGISTER_ULND_IF_PRESENT(lnd) \ -do { \ - extern lnd_t lnd __attribute__ ((weak, alias("the_lolnd"))); \ - \ - if (&(lnd) != &the_lolnd) \ - lnet_register_lnd(&(lnd)); \ -} while (0) -#endif - -extern int lnet_cpt_of_nid_locked(lnet_nid_t nid); -extern int lnet_cpt_of_nid(lnet_nid_t nid); +extern unsigned int lnet_nid_cpt_hash(lnet_nid_t nid, unsigned int number); +extern int lnet_cpt_of_nid_locked(lnet_nid_t nid, struct lnet_ni *ni); +extern int lnet_cpt_of_nid(lnet_nid_t nid, struct lnet_ni *ni); extern lnet_ni_t *lnet_nid2ni_locked(lnet_nid_t nid, int cpt); +extern lnet_ni_t *lnet_nid2ni_addref(lnet_nid_t nid); extern lnet_ni_t *lnet_net2ni_locked(__u32 net, int cpt); -extern lnet_ni_t *lnet_net2ni(__u32 net); +extern lnet_ni_t *lnet_net2ni_addref(__u32 net); +bool lnet_is_ni_healthy_locked(struct lnet_ni *ni); +struct lnet_net *lnet_get_net_locked(__u32 net_id); + +int lnet_lib_init(void); +void lnet_lib_exit(void); + +extern int portal_rotor; int lnet_notify(lnet_ni_t *ni, lnet_nid_t peer, int alive, cfs_time_t when); -void lnet_notify_locked(lnet_peer_t *lp, int notifylnd, int alive, cfs_time_t when); -int lnet_add_route(__u32 net, unsigned int hops, lnet_nid_t gateway_nid, +void lnet_notify_locked(struct lnet_peer_ni *lp, int notifylnd, int alive, + cfs_time_t when); +int lnet_add_route(__u32 net, __u32 hops, lnet_nid_t gateway_nid, unsigned int priority); int lnet_check_routes(void); int lnet_del_route(__u32 net, lnet_nid_t gw_nid); void lnet_destroy_routes(void); int lnet_get_route(int idx, __u32 *net, __u32 *hops, lnet_nid_t *gateway, __u32 *alive, __u32 *priority); +int lnet_get_rtr_pool_cfg(int idx, struct lnet_ioctl_pool_cfg *pool_cfg); +struct lnet_ni *lnet_get_next_ni_locked(struct lnet_net *mynet, + struct lnet_ni *prev); +struct lnet_ni *lnet_get_ni_idx_locked(int idx); + +struct libcfs_ioctl_handler { + struct list_head item; + int (*handle_ioctl)(unsigned int cmd, struct libcfs_ioctl_hdr *hdr); +}; + +#define DECLARE_IOCTL_HANDLER(ident, func) \ + static struct libcfs_ioctl_handler ident = { \ + /* .item = */ LIST_HEAD_INIT(ident.item), \ + /* .handle_ioctl = */ func \ + } + +extern int libcfs_register_ioctl(struct libcfs_ioctl_handler *hand); +extern int libcfs_deregister_ioctl(struct libcfs_ioctl_handler *hand); +extern int libcfs_ioctl_getdata(struct libcfs_ioctl_hdr **hdr_pp, + struct libcfs_ioctl_hdr __user *uparam); + void lnet_proc_init(void); void lnet_proc_fini(void); int lnet_rtrpools_alloc(int im_a_router); -void lnet_rtrpools_free(void); -lnet_remotenet_t *lnet_find_net_locked (__u32 net); +void lnet_destroy_rtrbuf(lnet_rtrbuf_t *rb, int npages); +int lnet_rtrpools_adjust(int tiny, int small, int large); +int lnet_rtrpools_enable(void); +void lnet_rtrpools_disable(void); +void lnet_rtrpools_free(int keep_pools); +lnet_remotenet_t *lnet_find_rnet_locked(__u32 net); +int lnet_dyn_add_net(struct lnet_ioctl_config_data *conf); +int lnet_dyn_del_net(__u32 net); +int lnet_dyn_add_ni(struct lnet_ioctl_config_ni *conf); +int lnet_dyn_del_ni(struct lnet_ioctl_config_ni *conf); +int lnet_clear_lazy_portal(struct lnet_ni *ni, int portal, char *reason); +struct lnet_net *lnet_get_net_locked(__u32 net_id); int lnet_islocalnid(lnet_nid_t nid); int lnet_islocalnet(__u32 net); @@ -763,10 +559,12 @@ void lnet_msg_decommit(lnet_msg_t *msg, int cpt, int status); void lnet_eq_enqueue_event(lnet_eq_t *eq, lnet_event_t *ev); void lnet_prep_send(lnet_msg_t *msg, int type, lnet_process_id_t target, - unsigned int offset, unsigned int len); + unsigned int offset, unsigned int len); int lnet_send(lnet_nid_t nid, lnet_msg_t *msg, lnet_nid_t rtr_nid); void lnet_return_tx_credits_locked(lnet_msg_t *msg); void lnet_return_rx_credits_locked(lnet_msg_t *msg); +void lnet_schedule_blocked_locked(lnet_rtrbufpool_t *rbp); +void lnet_drop_routed_msgs_locked(struct list_head *list, int cpt); /* portals functions */ /* portals attributes */ @@ -801,7 +599,7 @@ lnet_ptl_unsetopt(lnet_portal_t *ptl, int opt) } /* match-table functions */ -cfs_list_t *lnet_mt_match_head(struct lnet_match_table *mtable, +struct list_head *lnet_mt_match_head(struct lnet_match_table *mtable, lnet_process_id_t id, __u64 mbits); struct lnet_match_table *lnet_mt_of_attach(unsigned int index, lnet_process_id_t id, __u64 mbits, @@ -812,7 +610,7 @@ int lnet_mt_match_md(struct lnet_match_table *mtable, /* portals match/attach functions */ void lnet_ptl_attach_md(lnet_me_t *me, lnet_libmd_t *md, - cfs_list_t *matches, cfs_list_t *drops); + struct list_head *matches, struct list_head *drops); void lnet_ptl_detach_md(lnet_me_t *me, lnet_libmd_t *md); int lnet_ptl_match_md(struct lnet_match_info *info, struct lnet_msg *msg); @@ -822,14 +620,25 @@ void lnet_portals_destroy(void); /* message functions */ int lnet_parse (lnet_ni_t *ni, lnet_hdr_t *hdr, - lnet_nid_t fromnid, void *private, int rdma_req); + lnet_nid_t fromnid, void *private, int rdma_req); +int lnet_parse_local(lnet_ni_t *ni, lnet_msg_t *msg); +int lnet_parse_forward_locked(lnet_ni_t *ni, lnet_msg_t *msg); + void lnet_recv(lnet_ni_t *ni, void *private, lnet_msg_t *msg, int delayed, - unsigned int offset, unsigned int mlen, unsigned int rlen); + unsigned int offset, unsigned int mlen, unsigned int rlen); +void lnet_ni_recv(lnet_ni_t *ni, void *private, lnet_msg_t *msg, + int delayed, unsigned int offset, + unsigned int mlen, unsigned int rlen); + lnet_msg_t *lnet_create_reply_msg (lnet_ni_t *ni, lnet_msg_t *get_msg); void lnet_set_reply_msg_len(lnet_ni_t *ni, lnet_msg_t *msg, unsigned int len); + void lnet_finalize(lnet_ni_t *ni, lnet_msg_t *msg, int rc); -void lnet_drop_delayed_msg_list(cfs_list_t *head, char *reason); -void lnet_recv_delayed_msg_list(cfs_list_t *head); + +void lnet_drop_message(lnet_ni_t *ni, int cpt, void *private, + unsigned int nob); +void lnet_drop_delayed_msg_list(struct list_head *head, char *reason); +void lnet_recv_delayed_msg_list(struct list_head *head); int lnet_msg_container_setup(struct lnet_msg_container *container, int cpt); void lnet_msg_container_cleanup(struct lnet_msg_container *container); @@ -840,74 +649,94 @@ char *lnet_msgtyp2str (int type); void lnet_print_hdr (lnet_hdr_t * hdr); int lnet_fail_nid(lnet_nid_t nid, unsigned int threshold); +/** \addtogroup lnet_fault_simulation @{ */ + +int lnet_fault_ctl(int cmd, struct libcfs_ioctl_data *data); +int lnet_fault_init(void); +void lnet_fault_fini(void); + +bool lnet_drop_rule_match(lnet_hdr_t *hdr); + +int lnet_delay_rule_add(struct lnet_fault_attr *attr); +int lnet_delay_rule_del(lnet_nid_t src, lnet_nid_t dst, bool shutdown); +int lnet_delay_rule_list(int pos, struct lnet_fault_attr *attr, + struct lnet_fault_stat *stat); +void lnet_delay_rule_reset(void); +void lnet_delay_rule_check(void); +bool lnet_delay_rule_match_locked(lnet_hdr_t *hdr, struct lnet_msg *msg); + +/** @} lnet_fault_simulation */ + void lnet_counters_get(lnet_counters_t *counters); void lnet_counters_reset(void); -unsigned int lnet_iov_nob (unsigned int niov, struct iovec *iov); -int lnet_extract_iov (int dst_niov, struct iovec *dst, - int src_niov, struct iovec *src, - unsigned int offset, unsigned int len); +unsigned int lnet_iov_nob(unsigned int niov, struct kvec *iov); +int lnet_extract_iov(int dst_niov, struct kvec *dst, + int src_niov, struct kvec *src, + unsigned int offset, unsigned int len); unsigned int lnet_kiov_nob (unsigned int niov, lnet_kiov_t *iov); -int lnet_extract_kiov (int dst_niov, lnet_kiov_t *dst, - int src_niov, lnet_kiov_t *src, - unsigned int offset, unsigned int len); - -void lnet_copy_iov2iov (unsigned int ndiov, struct iovec *diov, - unsigned int doffset, - unsigned int nsiov, struct iovec *siov, - unsigned int soffset, unsigned int nob); -void lnet_copy_kiov2iov (unsigned int niov, struct iovec *iov, - unsigned int iovoffset, - unsigned int nkiov, lnet_kiov_t *kiov, - unsigned int kiovoffset, unsigned int nob); -void lnet_copy_iov2kiov (unsigned int nkiov, lnet_kiov_t *kiov, - unsigned int kiovoffset, - unsigned int niov, struct iovec *iov, - unsigned int iovoffset, unsigned int nob); -void lnet_copy_kiov2kiov (unsigned int ndkiov, lnet_kiov_t *dkiov, - unsigned int doffset, - unsigned int nskiov, lnet_kiov_t *skiov, - unsigned int soffset, unsigned int nob); +int lnet_extract_kiov(int dst_niov, lnet_kiov_t *dst, + int src_niov, lnet_kiov_t *src, + unsigned int offset, unsigned int len); + +void lnet_copy_iov2iov(unsigned int ndiov, struct kvec *diov, + unsigned int doffset, + unsigned int nsiov, struct kvec *siov, + unsigned int soffset, unsigned int nob); +void lnet_copy_kiov2iov(unsigned int niov, struct kvec *iov, + unsigned int iovoffset, + unsigned int nkiov, lnet_kiov_t *kiov, + unsigned int kiovoffset, unsigned int nob); +void lnet_copy_iov2kiov(unsigned int nkiov, lnet_kiov_t *kiov, + unsigned int kiovoffset, + unsigned int niov, struct kvec *iov, + unsigned int iovoffset, unsigned int nob); +void lnet_copy_kiov2kiov(unsigned int ndkiov, lnet_kiov_t *dkiov, + unsigned int doffset, + unsigned int nskiov, lnet_kiov_t *skiov, + unsigned int soffset, unsigned int nob); static inline void lnet_copy_iov2flat(int dlen, void *dest, unsigned int doffset, - unsigned int nsiov, struct iovec *siov, unsigned int soffset, - unsigned int nob) + unsigned int nsiov, struct kvec *siov, unsigned int soffset, + unsigned int nob) { - struct iovec diov = {/*.iov_base = */ dest, /*.iov_len = */ dlen}; + struct kvec diov = {/*.iov_base = */ dest, /*.iov_len = */ dlen}; - lnet_copy_iov2iov(1, &diov, doffset, - nsiov, siov, soffset, nob); + lnet_copy_iov2iov(1, &diov, doffset, + nsiov, siov, soffset, nob); } static inline void lnet_copy_kiov2flat(int dlen, void *dest, unsigned int doffset, - unsigned int nsiov, lnet_kiov_t *skiov, unsigned int soffset, - unsigned int nob) + unsigned int nsiov, lnet_kiov_t *skiov, + unsigned int soffset, unsigned int nob) { - struct iovec diov = {/* .iov_base = */ dest, /* .iov_len = */ dlen}; + struct kvec diov = {/* .iov_base = */ dest, /* .iov_len = */ dlen}; - lnet_copy_kiov2iov(1, &diov, doffset, - nsiov, skiov, soffset, nob); + lnet_copy_kiov2iov(1, &diov, doffset, + nsiov, skiov, soffset, nob); } static inline void -lnet_copy_flat2iov(unsigned int ndiov, struct iovec *diov, unsigned int doffset, - int slen, void *src, unsigned int soffset, unsigned int nob) +lnet_copy_flat2iov(unsigned int ndiov, struct kvec *diov, unsigned int doffset, + int slen, void *src, unsigned int soffset, + unsigned int nob) { - struct iovec siov = {/*.iov_base = */ src, /*.iov_len = */slen}; - lnet_copy_iov2iov(ndiov, diov, doffset, - 1, &siov, soffset, nob); + struct kvec siov = {/*.iov_base = */ src, /*.iov_len = */slen}; + lnet_copy_iov2iov(ndiov, diov, doffset, + 1, &siov, soffset, nob); } static inline void -lnet_copy_flat2kiov(unsigned int ndiov, lnet_kiov_t *dkiov, unsigned int doffset, - int slen, void *src, unsigned int soffset, unsigned int nob) +lnet_copy_flat2kiov(unsigned int ndiov, lnet_kiov_t *dkiov, + unsigned int doffset, int slen, void *src, + unsigned int soffset, unsigned int nob) { - struct iovec siov = {/* .iov_base = */ src, /* .iov_len = */ slen}; - lnet_copy_iov2kiov(ndiov, dkiov, doffset, - 1, &siov, soffset, nob); + struct kvec siov = {/* .iov_base = */ src, /* .iov_len = */ slen}; + lnet_copy_iov2kiov(ndiov, dkiov, doffset, + 1, &siov, soffset, nob); } void lnet_me_unlink(lnet_me_t *me); @@ -917,70 +746,140 @@ void lnet_md_deconstruct(lnet_libmd_t *lmd, lnet_md_t *umd); void lnet_register_lnd(lnd_t *lnd); void lnet_unregister_lnd(lnd_t *lnd); -int lnet_set_ip_niaddr (lnet_ni_t *ni); -#ifdef __KERNEL__ -int lnet_connect(cfs_socket_t **sockp, lnet_nid_t peer_nid, - __u32 local_ip, __u32 peer_ip, int peer_port); +int lnet_connect(struct socket **sockp, lnet_nid_t peer_nid, + __u32 local_ip, __u32 peer_ip, int peer_port); void lnet_connect_console_error(int rc, lnet_nid_t peer_nid, __u32 peer_ip, int port); -int lnet_count_acceptor_nis(void); +int lnet_count_acceptor_nets(void); int lnet_acceptor_timeout(void); int lnet_acceptor_port(void); -#else -void lnet_router_checker(void); -#endif - -#ifdef HAVE_LIBPTHREAD -int lnet_count_acceptor_nis(void); -int lnet_acceptor_port(void); -#endif - int lnet_acceptor_start(void); void lnet_acceptor_stop(void); -void lnet_get_tunables(void); +int lnet_ipif_query(char *name, int *up, __u32 *ip, __u32 *mask); +int lnet_ipif_enumerate(char ***names); +void lnet_ipif_free_enumeration(char **names, int n); +int lnet_sock_setbuf(struct socket *socket, int txbufsize, int rxbufsize); +int lnet_sock_getbuf(struct socket *socket, int *txbufsize, int *rxbufsize); +int lnet_sock_getaddr(struct socket *socket, bool remote, __u32 *ip, int *port); +int lnet_sock_write(struct socket *sock, void *buffer, int nob, int timeout); +int lnet_sock_read(struct socket *sock, void *buffer, int nob, int timeout); + +int lnet_sock_listen(struct socket **sockp, __u32 ip, int port, int backlog); +int lnet_sock_accept(struct socket **newsockp, struct socket *sock); +int lnet_sock_connect(struct socket **sockp, int *fatal, + __u32 local_ip, int local_port, + __u32 peer_ip, int peer_port); + int lnet_peers_start_down(void); -int lnet_peer_buffer_credits(lnet_ni_t *ni); +int lnet_peer_buffer_credits(struct lnet_net *net); int lnet_router_checker_start(void); void lnet_router_checker_stop(void); -void lnet_swap_pinginfo(lnet_ping_info_t *info); - -int lnet_ping_target_init(void); -void lnet_ping_target_fini(void); -int lnet_ping(lnet_process_id_t id, int timeout_ms, - lnet_process_id_t *ids, int n_ids); - -int lnet_parse_ip2nets (char **networksp, char *ip2nets); -int lnet_parse_routes (char *route_str, int *im_a_router); -int lnet_parse_networks (cfs_list_t *nilist, char *networks); - -int lnet_nid2peer_locked(lnet_peer_t **lpp, lnet_nid_t nid, int cpt); -lnet_peer_t *lnet_find_peer_locked(struct lnet_peer_table *ptable, - lnet_nid_t nid); -void lnet_peer_tables_cleanup(void); -void lnet_peer_tables_destroy(void); +void lnet_router_ni_update_locked(struct lnet_peer_ni *gw, __u32 net); +void lnet_swap_pinginfo(struct lnet_ping_info *info); + +int lnet_parse_ip2nets(char **networksp, char *ip2nets); +int lnet_parse_routes(char *route_str, int *im_a_router); +int lnet_parse_networks(struct list_head *nilist, char *networks, + bool use_tcp_bonding); +bool lnet_net_unique(__u32 net_id, struct list_head *nilist, + struct lnet_net **net); +bool lnet_ni_unique_net(struct list_head *nilist, char *iface); +void lnet_incr_dlc_seq(void); +__u32 lnet_get_dlc_seq_locked(void); +inline __u32 lnet_get_numa_range(void); + +struct lnet_peer_ni *lnet_get_next_peer_ni_locked(struct lnet_peer *peer, + struct lnet_peer_net *peer_net, + struct lnet_peer_ni *prev); +struct lnet_peer *lnet_find_or_create_peer_locked(lnet_nid_t dst_nid, int cpt); +struct lnet_peer_ni *lnet_nid2peerni_locked(lnet_nid_t nid, int cpt); +struct lnet_peer_ni *lnet_find_peer_ni_locked(lnet_nid_t nid); +void lnet_peer_net_added(struct lnet_net *net); +lnet_nid_t lnet_peer_primary_nid(lnet_nid_t nid); +void lnet_peer_tables_cleanup(struct lnet_net *net); +void lnet_peer_uninit(void); int lnet_peer_tables_create(void); void lnet_debug_peer(lnet_nid_t nid); +struct lnet_peer_net *lnet_peer_get_net_locked(struct lnet_peer *peer, + __u32 net_id); +bool lnet_peer_is_ni_pref_locked(struct lnet_peer_ni *lpni, + struct lnet_ni *ni); +int lnet_add_peer_ni_to_peer(lnet_nid_t key_nid, lnet_nid_t nid, bool mr); +int lnet_del_peer_ni_from_peer(lnet_nid_t key_nid, lnet_nid_t nid); +int lnet_get_peer_info(__u32 idx, lnet_nid_t *primary_nid, lnet_nid_t *nid, + bool *mr, struct lnet_peer_ni_credit_info *peer_ni_info, + struct lnet_ioctl_element_stats *peer_ni_stats); +int lnet_get_peer_ni_info(__u32 peer_index, __u64 *nid, + char alivness[LNET_MAX_STR_LEN], + __u32 *cpt_iter, __u32 *refcount, + __u32 *ni_peer_tx_credits, __u32 *peer_tx_credits, + __u32 *peer_rtr_credits, __u32 *peer_min_rtr_credtis, + __u32 *peer_tx_qnob); -#ifndef __KERNEL__ -static inline int -lnet_parse_int_tunable(int *value, char *name) + +static inline __u32 +lnet_get_num_peer_nis(struct lnet_peer *peer) { - char *env = getenv(name); - char *end; + struct lnet_peer_net *lpn; + struct lnet_peer_ni *lpni; + __u32 count = 0; - if (env == NULL) - return 0; + list_for_each_entry(lpn, &peer->lp_peer_nets, lpn_on_peer_list) + list_for_each_entry(lpni, &lpn->lpn_peer_nis, + lpni_on_peer_net_list) + count++; - *value = strtoull(env, &end, 0); - if (*end == 0) - return 0; + return count; +} - CERROR("Can't parse tunable %s=%s\n", name, env); - return -EINVAL; +static inline bool +lnet_is_peer_ni_healthy_locked(struct lnet_peer_ni *lpni) +{ + return lpni->lpni_healthy; +} + +static inline void +lnet_set_peer_ni_health_locked(struct lnet_peer_ni *lpni, bool health) +{ + lpni->lpni_healthy = health; +} + +static inline bool +lnet_is_peer_net_healthy_locked(struct lnet_peer_net *peer_net) +{ + struct lnet_peer_ni *lpni; + + list_for_each_entry(lpni, &peer_net->lpn_peer_nis, + lpni_on_peer_net_list) { + if (lnet_is_peer_ni_healthy_locked(lpni)) + return true; + } + + return false; +} + +static inline bool +lnet_is_peer_healthy_locked(struct lnet_peer *peer) +{ + struct lnet_peer_net *peer_net; + + list_for_each_entry(peer_net, &peer->lp_peer_nets, lpn_on_peer_list) { + if (lnet_is_peer_net_healthy_locked(peer_net)) + return true; + } + + return false; +} + +static inline void +lnet_peer_set_alive(struct lnet_peer_ni *lp) +{ + lp->lpni_last_alive = lp->lpni_last_query = cfs_time_current(); + if (!lp->lpni_alive) + lnet_notify_locked(lp, 0, 1, lp->lpni_last_alive); } -#endif #endif