diff options
| -rw-r--r-- | src/vcl/vcl_locked.c | 268 |
1 files changed, 167 insertions, 101 deletions
diff --git a/src/vcl/vcl_locked.c b/src/vcl/vcl_locked.c index 54850c6201b..b006d89391b 100644 --- a/src/vcl/vcl_locked.c +++ b/src/vcl/vcl_locked.c @@ -1,5 +1,5 @@ /* - * Copyright (c) 2019 Cisco and/or its affiliates. + * Copyright (c) 2021 Cisco and/or its affiliates. * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this * You may obtain a copy of the License at: @@ -13,46 +13,114 @@ * limitations under the License. */ +/** + * VCL Locked Sessions (VLS) is a wrapper that synchronizes access to VCL APIs + * which are, by construction, not thread safe. To this end, VLS uses + * configuration and heuristics to detect how applications use sessions in + * an attempt to optimize the locking strategy. The modes of operation + * currently supported are the following: + * + * 1) per-process workers + * + * +----------+ +----------+ + * | | | | + * | process0 | | process1 | + * | | | | + * +-----+----+ +-----+----+ + * | | + * | | + * +-----+----+ +-----+----+ + * | | | | + * | vls_wrk0 | | vls_wrk1 | + * | | | | + * +-----+----+ +-----+----+ + * | | + * | | + * +-----+----+ +-----+----+ + * | | | | + * | vcl_wrk0 | | vcl_wrk1 | + * | | | | + * +----------+ +----------+ + * + * 2) per-thread workers 3) single-worker multi-thread + * + * +---------+ +---------+ +---------+ +---------+ + * | | | | | | | | + * | thread0 | | thread1 | | thread0 | | thread1 | + * | | | | | | | | + * +--------++ +-+-------+ +--------++ +-+-------+ + * | | | | + * | | | | + * +-+------+-+ +-+------+-+ + * | | | | + * | vls_wrk0 | | vls_wrk0 | + * | | | | + * +-+------+-+ +----+-----+ + * | | | + * | | | + * +--------+-+ +-+--------+ +----+-----+ + * | | | | | | + * | vcl_wrk0 | | vcl_wrk1 | | vcl_wrk0 | + * | | | | | | + * +----------+ +----------+ +----------+ + * + * 1) per-process workers: intercept fork calls and assume all children + * processes are new workers that must be registered with vcl. VLS + * sessions are cloned and shared between workers. Only shared sessions + * are locked on use and thereby only one process can interact with + * them at a time (explicit sharing). + * + * 2) per-thread workers: each newly detected pthread is assumed to be a new + * worker and is registered with vcl. Enabled via configuration. + * When a thread tries to access a session it does not own, a clone and + * share rpc request is sent to the owning thread via vcl and vpp. + * Consequently, a vls session can map to multiple vcl sessions, one per + * vcl worker. VLS sessions are locked on use (implicit sharing). + * + * 3) single-worker multi-thread: vls does not make any assumptions about + * application threads and therefore implements an aggressive locking + * strategy that limits access to underlying vcl resources based on type + * of interaction and locks vls session on use (implicit sharing). + */ + #include <vcl/vcl_locked.h> #include <vcl/vcl_private.h> typedef struct vls_shared_data_ { - clib_spinlock_t lock; - u32 owner_wrk_index; - u32 *workers_subscribed; - clib_bitmap_t *listeners; + clib_spinlock_t lock; /**< shared data lock */ + u32 owner_wrk_index; /**< vcl wrk that owns session */ + u32 *workers_subscribed; /**< vec of wrks subscribed to session */ + clib_bitmap_t *listeners; /**< bitmap of wrks actively listening */ } vls_shared_data_t; typedef struct vcl_locked_session_ { - clib_spinlock_t lock; - u32 session_index; - u32 worker_index; - u32 vls_index; - u32 shared_data_index; - /** VCL session owned by different workers because of migration */ - u32 owner_vcl_wrk_index; - uword *vcl_wrk_index_to_session_index; + clib_spinlock_t lock; /**< vls lock when in use */ + u32 session_index; /**< vcl session index */ + u32 vcl_wrk_index; /**< vcl worker index */ + u32 vls_index; /**< index in vls pool */ + u32 shared_data_index; /**< shared data index if any */ + u32 owner_vcl_wrk_index; /**< vcl wrk of the vls wrk at alloc */ + uword *vcl_wrk_index_to_session_index; /**< map vcl wrk to session */ } vcl_locked_session_t; typedef struct vls_worker_ { - clib_rwlock_t sh_to_vlsh_table_lock; /** valid for multithread workers */ - vcl_locked_session_t *vls_pool; - uword *session_handle_to_vlsh_table; - u32 wrk_index; - /** Vector of child wrk to cleanup */ - u32 *pending_wrk_cleanup; + clib_rwlock_t sh_to_vlsh_table_lock; /**< ht rwlock with mt workers */ + vcl_locked_session_t *vls_pool; /**< pool of vls session */ + uword *sh_to_vlsh_table; /**< map from vcl sh to vls sh */ + u32 *pending_vcl_wrk_cleanup; /**< child vcl wrks to cleanup */ + u32 vcl_wrk_index; /**< if 1:1 map vls to vcl wrk */ } vls_worker_t; typedef struct vls_local_ { - int vls_wrk_index; - volatile int vls_mt_n_threads; - pthread_mutex_t vls_mt_mq_mlock; - pthread_mutex_t vls_mt_spool_mlock; - volatile u8 select_mp_check; + int vls_wrk_index; /**< vls wrk index, 1 per process */ + volatile int vls_mt_n_threads; /**< number of threads detected */ + pthread_mutex_t vls_mt_mq_mlock; /**< vcl mq lock */ + pthread_mutex_t vls_mt_spool_mlock; /**< vcl select or pool lock */ + volatile u8 select_mp_check; /**< flag set if select checks done */ } vls_process_local_t; static vls_process_local_t vls_local; @@ -60,13 +128,11 @@ static vls_process_local_t *vlsl = &vls_local; typedef struct vls_main_ { - vls_worker_t *workers; - clib_rwlock_t vls_table_lock; - /** Pool of data shared by sessions owned by different workers */ - vls_shared_data_t *shared_data_pool; - clib_rwlock_t shared_data_lock; - /** Lock to protect rpc among workers */ - clib_spinlock_t worker_rpc_lock; + vls_worker_t *workers; /**< pool of vls workers */ + clib_rwlock_t vls_pool_lock; /**< per process/wrk vls pool locks */ + vls_shared_data_t *shared_data_pool; /**< inter proc pool of shared data */ + clib_rwlock_t shared_data_lock; /**< shared data pool lock */ + clib_spinlock_t worker_rpc_lock; /**< lock for inter-worker rpcs */ } vls_main_t; vls_main_t *vlsm; @@ -178,31 +244,31 @@ vls_shared_data_pool_runlock (void) } static inline void -vls_mt_table_rlock (void) +vls_mt_pool_rlock (void) { if (vlsl->vls_mt_n_threads > 1) - clib_rwlock_reader_lock (&vlsm->vls_table_lock); + clib_rwlock_reader_lock (&vlsm->vls_pool_lock); } static inline void -vls_mt_table_runlock (void) +vls_mt_pool_runlock (void) { if (vlsl->vls_mt_n_threads > 1) - clib_rwlock_reader_unlock (&vlsm->vls_table_lock); + clib_rwlock_reader_unlock (&vlsm->vls_pool_lock); } static inline void -vls_mt_table_wlock (void) +vls_mt_pool_wlock (void) { if (vlsl->vls_mt_n_threads > 1) - clib_rwlock_writer_lock (&vlsm->vls_table_lock); + clib_rwlock_writer_lock (&vlsm->vls_pool_lock); } static inline void -vls_mt_table_wunlock (void) +vls_mt_pool_wunlock (void) { if (vlsl->vls_mt_n_threads > 1) - clib_rwlock_writer_unlock (&vlsm->vls_table_lock); + clib_rwlock_writer_unlock (&vlsm->vls_pool_lock); } typedef enum @@ -298,7 +364,7 @@ vls_to_sh_tu (vcl_locked_session_t * vls) { vcl_session_handle_t sh; sh = vls_to_sh (vls); - vls_mt_table_runlock (); + vls_mt_pool_runlock (); return sh; } @@ -316,15 +382,15 @@ vls_worker_alloc (void) pool_get_zero (vlsm->workers, wrk); if (vls_mt_wrk_supported ()) clib_rwlock_init (&wrk->sh_to_vlsh_table_lock); - wrk->wrk_index = vcl_get_worker_index (); - vec_validate (wrk->pending_wrk_cleanup, 16); - vec_reset_length (wrk->pending_wrk_cleanup); + wrk->vcl_wrk_index = vcl_get_worker_index (); + vec_validate (wrk->pending_vcl_wrk_cleanup, 16); + vec_reset_length (wrk->pending_vcl_wrk_cleanup); } static void vls_worker_free (vls_worker_t * wrk) { - hash_free (wrk->session_handle_to_vlsh_table); + hash_free (wrk->sh_to_vlsh_table); if (vls_mt_wrk_supported ()) clib_rwlock_free (&wrk->sh_to_vlsh_table_lock); pool_free (wrk->vls_pool); @@ -344,7 +410,7 @@ vls_sh_to_vlsh_table_add (vls_worker_t *wrk, vcl_session_handle_t sh, u32 vlsh) { if (vls_mt_wrk_supported ()) clib_rwlock_writer_lock (&wrk->sh_to_vlsh_table_lock); - hash_set (wrk->session_handle_to_vlsh_table, sh, vlsh); + hash_set (wrk->sh_to_vlsh_table, sh, vlsh); if (vls_mt_wrk_supported ()) clib_rwlock_writer_unlock (&wrk->sh_to_vlsh_table_lock); } @@ -354,7 +420,7 @@ vls_sh_to_vlsh_table_del (vls_worker_t *wrk, vcl_session_handle_t sh) { if (vls_mt_wrk_supported ()) clib_rwlock_writer_lock (&wrk->sh_to_vlsh_table_lock); - hash_unset (wrk->session_handle_to_vlsh_table, sh); + hash_unset (wrk->sh_to_vlsh_table, sh); if (vls_mt_wrk_supported ()) clib_rwlock_writer_unlock (&wrk->sh_to_vlsh_table_lock); } @@ -364,7 +430,7 @@ vls_sh_to_vlsh_table_get (vls_worker_t *wrk, vcl_session_handle_t sh) { if (vls_mt_wrk_supported ()) clib_rwlock_reader_lock (&wrk->sh_to_vlsh_table_lock); - uword *vlshp = hash_get (wrk->session_handle_to_vlsh_table, sh); + uword *vlshp = hash_get (wrk->sh_to_vlsh_table, sh); if (vls_mt_wrk_supported ()) clib_rwlock_reader_unlock (&wrk->sh_to_vlsh_table_lock); return vlshp; @@ -376,23 +442,23 @@ vls_alloc (vcl_session_handle_t sh) vls_worker_t *wrk = vls_worker_get_current (); vcl_locked_session_t *vls; - vls_mt_table_wlock (); + vls_mt_pool_wlock (); pool_get_zero (wrk->vls_pool, vls); vls->session_index = vppcom_session_index (sh); - vls->worker_index = vppcom_session_worker (sh); + vls->vcl_wrk_index = vppcom_session_worker (sh); vls->vls_index = vls - wrk->vls_pool; vls->shared_data_index = ~0; vls_sh_to_vlsh_table_add (wrk, sh, vls->vls_index); if (vls_mt_wrk_supported ()) { - hash_set (vls->vcl_wrk_index_to_session_index, vls->worker_index, + hash_set (vls->vcl_wrk_index_to_session_index, vls->vcl_wrk_index, vls->session_index); - vls->owner_vcl_wrk_index = vls->worker_index; + vls->owner_vcl_wrk_index = vls->vcl_wrk_index; } clib_spinlock_init (&vls->lock); - vls_mt_table_wunlock (); + vls_mt_pool_wunlock (); return vls->vls_index; } @@ -433,10 +499,10 @@ static vcl_locked_session_t * vls_get_w_dlock (vls_handle_t vlsh) { vcl_locked_session_t *vls; - vls_mt_table_rlock (); + vls_mt_pool_rlock (); vls = vls_get_and_lock (vlsh); if (!vls) - vls_mt_table_runlock (); + vls_mt_pool_runlock (); return vls; } @@ -444,17 +510,17 @@ static inline void vls_get_and_unlock (vls_handle_t vlsh) { vcl_locked_session_t *vls; - vls_mt_table_rlock (); + vls_mt_pool_rlock (); vls = vls_get (vlsh); vls_unlock (vls); - vls_mt_table_runlock (); + vls_mt_pool_runlock (); } static inline void vls_dunlock (vcl_locked_session_t * vls) { vls_unlock (vls); - vls_mt_table_runlock (); + vls_mt_pool_runlock (); } static vcl_locked_session_t * @@ -503,9 +569,9 @@ vls_session_index_to_vlsh (uint32_t session_index) { vls_handle_t vlsh; - vls_mt_table_rlock (); + vls_mt_pool_rlock (); vlsh = vls_si_wi_to_vlsh (session_index, vcl_get_worker_index ()); - vls_mt_table_runlock (); + vls_mt_pool_runlock (); return vlsh; } @@ -660,7 +726,7 @@ vls_unshare_session (vcl_locked_session_t * vls, vcl_worker_t * wrk) if (pos < 0) { clib_warning ("worker %u not subscribed for vls %u", wrk->wrk_index, - vls->worker_index); + vls->vcl_wrk_index); goto done; } @@ -732,9 +798,9 @@ vls_init_share_session (vls_worker_t * vls_wrk, vcl_locked_session_t * vls) vls_shared_data_pool_rlock (); vls_shd = vls_shared_data_get (vls_shd_index); - vls_shd->owner_wrk_index = vls_wrk->wrk_index; + vls_shd->owner_wrk_index = vls_wrk->vcl_wrk_index; vls->shared_data_index = vls_shd_index; - vec_add1 (vls_shd->workers_subscribed, vls_wrk->wrk_index); + vec_add1 (vls_shd->workers_subscribed, vls_wrk->vcl_wrk_index); vls_shared_data_pool_runlock (); } @@ -742,7 +808,7 @@ vls_init_share_session (vls_worker_t * vls_wrk, vcl_locked_session_t * vls) void vls_share_session (vls_worker_t * vls_wrk, vcl_locked_session_t * vls) { - vcl_worker_t *vcl_wrk = vcl_worker_get (vls_wrk->wrk_index); + vcl_worker_t *vcl_wrk = vcl_worker_get (vls_wrk->vcl_wrk_index); vls_shared_data_t *vls_shd; vcl_session_t *s; @@ -764,7 +830,7 @@ vls_share_session (vls_worker_t * vls_wrk, vcl_locked_session_t * vls) vls_shd = vls_shared_data_get (vls->shared_data_index); clib_spinlock_lock (&vls_shd->lock); - vec_add1 (vls_shd->workers_subscribed, vls_wrk->wrk_index); + vec_add1 (vls_shd->workers_subscribed, vls_wrk->vcl_wrk_index); clib_spinlock_unlock (&vls_shd->lock); vls_shared_data_pool_runlock (); @@ -830,7 +896,7 @@ void vls_worker_copy_on_fork (vcl_worker_t * parent_wrk) { vls_worker_t *vls_wrk = vls_worker_get_current (), *vls_parent_wrk; - vcl_worker_t *wrk = vcl_worker_get_current (); + vcl_worker_t *vcl_wrk = vcl_worker_get_current (); u32 vls_index, session_index, wrk_index; vcl_session_handle_t sh; vcl_locked_session_t *vls; @@ -838,22 +904,22 @@ vls_worker_copy_on_fork (vcl_worker_t * parent_wrk) /* * init vcl worker */ - wrk->sessions = pool_dup (parent_wrk->sessions); - wrk->session_index_by_vpp_handles = + vcl_wrk->sessions = pool_dup (parent_wrk->sessions); + vcl_wrk->session_index_by_vpp_handles = hash_dup (parent_wrk->session_index_by_vpp_handles); /* * init vls worker */ vls_parent_wrk = vls_worker_get (parent_wrk->wrk_index); - /* *INDENT-OFF* */ - hash_foreach (sh, vls_index, vls_parent_wrk->session_handle_to_vlsh_table, - ({ - vcl_session_handle_parse (sh, &wrk_index, &session_index); - hash_set (vls_wrk->session_handle_to_vlsh_table, - vcl_session_handle_from_index (session_index), vls_index); - })); - /* *INDENT-ON* */ + + /* clang-format off */ + hash_foreach (sh, vls_index, vls_parent_wrk->sh_to_vlsh_table, ({ + vcl_session_handle_parse (sh, &wrk_index, &session_index); + hash_set (vls_wrk->sh_to_vlsh_table, + vcl_session_handle_from_index (session_index), vls_index); + })); + /* clang-format on */ vls_wrk->vls_pool = pool_dup (vls_parent_wrk->vls_pool); /* @@ -861,11 +927,11 @@ vls_worker_copy_on_fork (vcl_worker_t * parent_wrk) */ pool_foreach (vls, vls_wrk->vls_pool) { - vls->worker_index = wrk->wrk_index; + vls->vcl_wrk_index = vcl_wrk->wrk_index; } /* Validate vep's handle */ - vls_validate_veps (wrk); + vls_validate_veps (vcl_wrk); vls_share_sessions (vls_parent_wrk, vls_wrk); } @@ -931,8 +997,8 @@ vls_mt_rel_locks (int locks_acq) static inline u8 vls_mt_session_should_migrate (vcl_locked_session_t * vls) { - return (vls_mt_wrk_supported () - && vls->worker_index != vcl_get_worker_index ()); + return (vls_mt_wrk_supported () && + vls->vcl_wrk_index != vcl_get_worker_index ()); } static vcl_locked_session_t * @@ -945,7 +1011,7 @@ vls_mt_session_migrate (vcl_locked_session_t *vls) vcl_session_t *session; uword *p; - ASSERT (vls_mt_wrk_supported () && vls->worker_index != wrk_index); + ASSERT (vls_mt_wrk_supported () && vls->vcl_wrk_index != wrk_index); /* * VCL session on current vcl worker already allocated. Update current @@ -953,7 +1019,7 @@ vls_mt_session_migrate (vcl_locked_session_t *vls) */ if ((p = hash_get (vls->vcl_wrk_index_to_session_index, wrk_index))) { - vls->worker_index = wrk_index; + vls->vcl_wrk_index = wrk_index; vls->session_index = (u32) p[0]; return vls; } @@ -969,7 +1035,7 @@ vls_mt_session_migrate (vcl_locked_session_t *vls) VERR ("session in owner worker(%u) is free", vls->owner_vcl_wrk_index); ASSERT (0); vls_unlock (vls); - vls_mt_table_runlock (); + vls_mt_pool_runlock (); return 0; } @@ -984,7 +1050,7 @@ vls_mt_session_migrate (vcl_locked_session_t *vls) vls_index = vls->vls_index; own_vcl_wrk_index = vls->owner_vcl_wrk_index; vls_unlock (vls); - vls_mt_table_runlock (); + vls_mt_pool_runlock (); vls_send_clone_and_share_rpc (wrk, vls_index, sid, vls_get_worker_index (), own_vcl_wrk_index, vls_index, src_sid); @@ -1022,7 +1088,7 @@ vls_mt_session_migrate (vcl_locked_session_t *vls) } session->session_index = sid; - vls->worker_index = wrk_index; + vls->vcl_wrk_index = wrk_index; vls->session_index = sid; hash_set (vls->vcl_wrk_index_to_session_index, wrk_index, sid); vls_sh_to_vlsh_table_add (vls_wrk, vcl_session_handle (session), @@ -1046,7 +1112,7 @@ vls_mt_detect (void) if (vls_mt_wrk_supported ()) \ { \ if (PREDICT_FALSE (_vls && \ - ((vcl_locked_session_t *) _vls)->worker_index != \ + ((vcl_locked_session_t *) _vls)->vcl_wrk_index != \ vcl_get_worker_index ())) \ { \ _vls = vls_mt_session_migrate (_vls); \ @@ -1224,24 +1290,24 @@ vls_mp_checks (vcl_locked_session_t * vls, int is_add) if (vls_mt_wrk_supported ()) return; - ASSERT (wrk->wrk_index == vls->worker_index); + ASSERT (wrk->wrk_index == vls->vcl_wrk_index); s = vcl_session_get (wrk, vls->session_index); switch (s->session_state) { case VCL_STATE_LISTEN: if (is_add) { - vls_listener_wrk_set (vls, vls->worker_index, 1 /* is_active */ ); + vls_listener_wrk_set (vls, vls->vcl_wrk_index, 1 /* is_active */); break; } - vls_listener_wrk_stop_listen (vls, vls->worker_index); + vls_listener_wrk_stop_listen (vls, vls->vcl_wrk_index); break; case VCL_STATE_LISTEN_NO_MQ: if (!is_add) break; /* Register worker as listener */ - vls_listener_wrk_start_listen (vls, vls->worker_index); + vls_listener_wrk_start_listen (vls, vls->vcl_wrk_index); /* If owner worker did not attempt to accept/xpoll on the session, * force a listen stop for it, since it may not be interested in @@ -1331,12 +1397,12 @@ vls_close (vls_handle_t vlsh) int rv; vls_mt_detect (); - vls_mt_table_wlock (); + vls_mt_pool_wlock (); vls = vls_get_and_lock (vlsh); if (!vls) { - vls_mt_table_wunlock (); + vls_mt_pool_wunlock (); return VPPCOM_EBADFD; } @@ -1353,7 +1419,7 @@ vls_close (vls_handle_t vlsh) vls_free (vls); vls_mt_unguard (); - vls_mt_table_wunlock (); + vls_mt_pool_wunlock (); return rv; } @@ -1413,7 +1479,7 @@ vls_epoll_ctl (vls_handle_t ep_vlsh, int op, vls_handle_t vlsh, int rv; vls_mt_detect (); - vls_mt_table_rlock (); + vls_mt_pool_rlock (); ep_vls = vls_get_and_lock (ep_vlsh); if (vls_mt_session_should_migrate (ep_vls)) @@ -1428,15 +1494,15 @@ vls_epoll_ctl (vls_handle_t ep_vlsh, int op, vls_handle_t vlsh, sh = vls_to_sh (vls); vls_epoll_ctl_mp_checks (vls, op); - vls_mt_table_runlock (); + vls_mt_pool_runlock (); rv = vppcom_epoll_ctl (ep_sh, op, sh, event); - vls_mt_table_rlock (); + vls_mt_pool_rlock (); ep_vls = vls_get (ep_vlsh); vls = vls_get (vlsh); vls_unlock (vls); vls_unlock (ep_vls); - vls_mt_table_runlock (); + vls_mt_pool_runlock (); return rv; } @@ -1577,18 +1643,18 @@ vls_handle_pending_wrk_cleanup (void) vcl_worker_t *child_wrk, *wrk; vls_worker_t *vls_wrk = vls_worker_get_current (); - if (PREDICT_TRUE (vec_len (vls_wrk->pending_wrk_cleanup) == 0)) + if (PREDICT_TRUE (vec_len (vls_wrk->pending_vcl_wrk_cleanup) == 0)) return; wrk = vcl_worker_get_current (); - vec_foreach (wip, vls_wrk->pending_wrk_cleanup) + vec_foreach (wip, vls_wrk->pending_vcl_wrk_cleanup) { child_wrk = vcl_worker_get_if_valid (*wip); if (!child_wrk) continue; vls_cleanup_forked_child (wrk, child_wrk); } - vec_reset_length (vls_wrk->pending_wrk_cleanup); + vec_reset_length (vls_wrk->pending_vcl_wrk_cleanup); } static struct sigaction old_sa; @@ -1628,7 +1694,7 @@ vls_intercept_sigchld_handler (int signum, siginfo_t * si, void *uc) * So move child wrk cleanup from sighandler to vls_epoll_wait/vls_select. */ vls_wrk = vls_worker_get_current (); - vec_add1 (vls_wrk->pending_wrk_cleanup, child_wrk->wrk_index); + vec_add1 (vls_wrk->pending_vcl_wrk_cleanup, child_wrk->wrk_index); done: if (old_sa.sa_flags & SA_SIGINFO) @@ -1874,7 +1940,7 @@ vls_app_create (char *app_name) vlsm = clib_mem_alloc (sizeof (vls_main_t)); clib_memset (vlsm, 0, sizeof (*vlsm)); - clib_rwlock_init (&vlsm->vls_table_lock); + clib_rwlock_init (&vlsm->vls_pool_lock); clib_rwlock_init (&vlsm->shared_data_lock); clib_spinlock_init (&vlsm->worker_rpc_lock); pool_alloc (vlsm->workers, vcm->cfg.max_workers); |