/* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright 2009 Sun Microsystems, Inc. All rights reserved. * Use is subject to license terms. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include extern u_longlong_t nfs4_srv_caller_id; extern time_t rfs4_start_time; extern uint_t nfs4_srv_vkey; stateid4 special0 = { 0, { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } }; stateid4 special1 = { 0xffffffff, { (char)0xff, (char)0xff, (char)0xff, (char)0xff, (char)0xff, (char)0xff, (char)0xff, (char)0xff, (char)0xff, (char)0xff, (char)0xff, (char)0xff } }; #define ISSPECIAL(id) (stateid4_cmp(id, &special0) || \ stateid4_cmp(id, &special1)) /* For embedding the cluster nodeid into our clientid */ #define CLUSTER_NODEID_SHIFT 24 #define CLUSTER_MAX_NODEID 255 #ifdef DEBUG int rfs4_debug; #endif static uint32_t rfs4_database_debug = 0x00; static void rfs4_ss_clid_write(rfs4_client_t *cp, char *leaf); static void rfs4_ss_clid_write_one(rfs4_client_t *cp, char *dir, char *leaf); static void rfs4_dss_clear_oldstate(rfs4_servinst_t *sip); static void rfs4_ss_chkclid_sip(rfs4_client_t *cp, rfs4_servinst_t *sip); /* * Couple of simple init/destroy functions for a general waiter */ void rfs4_sw_init(rfs4_state_wait_t *swp) { mutex_init(swp->sw_cv_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(swp->sw_cv, NULL, CV_DEFAULT, NULL); swp->sw_active = FALSE; swp->sw_wait_count = 0; } void rfs4_sw_destroy(rfs4_state_wait_t *swp) { mutex_destroy(swp->sw_cv_lock); cv_destroy(swp->sw_cv); } void rfs4_sw_enter(rfs4_state_wait_t *swp) { mutex_enter(swp->sw_cv_lock); while (swp->sw_active) { swp->sw_wait_count++; cv_wait(swp->sw_cv, swp->sw_cv_lock); swp->sw_wait_count--; } ASSERT(swp->sw_active == FALSE); swp->sw_active = TRUE; mutex_exit(swp->sw_cv_lock); } void rfs4_sw_exit(rfs4_state_wait_t *swp) { mutex_enter(swp->sw_cv_lock); ASSERT(swp->sw_active == TRUE); swp->sw_active = FALSE; if (swp->sw_wait_count != 0) cv_broadcast(swp->sw_cv); mutex_exit(swp->sw_cv_lock); } /* * CPR callback id -- not related to v4 callbacks */ static callb_id_t cpr_id = 0; static void deep_lock_copy(LOCK4res *dres, LOCK4res *sres) { lock_owner4 *slo = &sres->LOCK4res_u.denied.owner; lock_owner4 *dlo = &dres->LOCK4res_u.denied.owner; if (sres->status == NFS4ERR_DENIED) { dlo->owner_val = kmem_alloc(slo->owner_len, KM_SLEEP); bcopy(slo->owner_val, dlo->owner_val, slo->owner_len); } } static void deep_lock_free(LOCK4res *res) { lock_owner4 *lo = &res->LOCK4res_u.denied.owner; if (res->status == NFS4ERR_DENIED) kmem_free(lo->owner_val, lo->owner_len); } static void deep_open_copy(OPEN4res *dres, OPEN4res *sres) { nfsace4 *sacep, *dacep; if (sres->status != NFS4_OK) { return; } dres->attrset = sres->attrset; switch (sres->delegation.delegation_type) { case OPEN_DELEGATE_NONE: return; case OPEN_DELEGATE_READ: sacep = &sres->delegation.open_delegation4_u.read.permissions; dacep = &dres->delegation.open_delegation4_u.read.permissions; break; case OPEN_DELEGATE_WRITE: sacep = &sres->delegation.open_delegation4_u.write.permissions; dacep = &dres->delegation.open_delegation4_u.write.permissions; break; } dacep->who.utf8string_val = kmem_alloc(sacep->who.utf8string_len, KM_SLEEP); bcopy(sacep->who.utf8string_val, dacep->who.utf8string_val, sacep->who.utf8string_len); } static void deep_open_free(OPEN4res *res) { nfsace4 *acep; if (res->status != NFS4_OK) return; switch (res->delegation.delegation_type) { case OPEN_DELEGATE_NONE: return; case OPEN_DELEGATE_READ: acep = &res->delegation.open_delegation4_u.read.permissions; break; case OPEN_DELEGATE_WRITE: acep = &res->delegation.open_delegation4_u.write.permissions; break; } if (acep->who.utf8string_val) { kmem_free(acep->who.utf8string_val, acep->who.utf8string_len); acep->who.utf8string_val = NULL; } } void rfs4_free_reply(nfs_resop4 *rp) { switch (rp->resop) { case OP_LOCK: deep_lock_free(&rp->nfs_resop4_u.oplock); break; case OP_OPEN: deep_open_free(&rp->nfs_resop4_u.opopen); default: break; } } void rfs4_copy_reply(nfs_resop4 *dst, nfs_resop4 *src) { *dst = *src; /* Handle responses that need deep copy */ switch (src->resop) { case OP_LOCK: deep_lock_copy(&dst->nfs_resop4_u.oplock, &src->nfs_resop4_u.oplock); break; case OP_OPEN: deep_open_copy(&dst->nfs_resop4_u.opopen, &src->nfs_resop4_u.opopen); break; default: break; }; } /* * This is the implementation of the underlying state engine. The * public interface to this engine is described by * nfs4_state.h. Callers to the engine should hold no state engine * locks when they call in to it. If the protocol needs to lock data * structures it should do so after acquiring all references to them * first and then follow the following lock order: * * client > openowner > state > lo_state > lockowner > file. * * Internally we only allow a thread to hold one hash bucket lock at a * time and the lock is higher in the lock order (must be acquired * first) than the data structure that is on that hash list. * * If a new reference was acquired by the caller, that reference needs * to be released after releasing all acquired locks with the * corresponding rfs4_*_rele routine. */ /* * This code is some what prototypical for now. Its purpose currently is to * implement the interfaces sufficiently to finish the higher protocol * elements. This will be replaced by a dynamically resizeable tables * backed by kmem_cache allocator. However synchronization is handled * correctly (I hope) and will not change by much. The mutexes for * the hash buckets that can be used to create new instances of data * structures might be good candidates to evolve into reader writer * locks. If it has to do a creation, it would be holding the * mutex across a kmem_alloc with KM_SLEEP specified. */ #ifdef DEBUG #define TABSIZE 17 #else #define TABSIZE 2047 #endif #define ADDRHASH(key) ((unsigned long)(key) >> 3) /* Used to serialize create/destroy of rfs4_server_state database */ kmutex_t rfs4_state_lock; static rfs4_database_t *rfs4_server_state = NULL; /* Used to serialize lookups of clientids */ static krwlock_t rfs4_findclient_lock; /* * For now this "table" is exposed so that the CPR callback * function can tromp through it.. */ rfs4_table_t *rfs4_client_tab; static rfs4_index_t *rfs4_clientid_idx; static rfs4_index_t *rfs4_nfsclnt_idx; static rfs4_table_t *rfs4_openowner_tab; static rfs4_index_t *rfs4_openowner_idx; static rfs4_table_t *rfs4_state_tab; static rfs4_index_t *rfs4_state_idx; static rfs4_index_t *rfs4_state_owner_file_idx; static rfs4_index_t *rfs4_state_file_idx; static rfs4_table_t *rfs4_lo_state_tab; static rfs4_index_t *rfs4_lo_state_idx; static rfs4_index_t *rfs4_lo_state_owner_idx; static rfs4_table_t *rfs4_lockowner_tab; static rfs4_index_t *rfs4_lockowner_idx; static rfs4_index_t *rfs4_lockowner_pid_idx; static rfs4_table_t *rfs4_file_tab; static rfs4_index_t *rfs4_file_idx; static rfs4_table_t *rfs4_deleg_state_tab; static rfs4_index_t *rfs4_deleg_idx; static rfs4_index_t *rfs4_deleg_state_idx; #define MAXTABSZ 1024*1024 /* The values below are rfs4_lease_time units */ #ifdef DEBUG #define CLIENT_CACHE_TIME 1 #define OPENOWNER_CACHE_TIME 1 #define STATE_CACHE_TIME 1 #define LO_STATE_CACHE_TIME 1 #define LOCKOWNER_CACHE_TIME 1 #define FILE_CACHE_TIME 3 #define DELEG_STATE_CACHE_TIME 1 #else #define CLIENT_CACHE_TIME 10 #define OPENOWNER_CACHE_TIME 5 #define STATE_CACHE_TIME 1 #define LO_STATE_CACHE_TIME 1 #define LOCKOWNER_CACHE_TIME 3 #define FILE_CACHE_TIME 40 #define DELEG_STATE_CACHE_TIME 1 #endif static time_t rfs4_client_cache_time = 0; static time_t rfs4_openowner_cache_time = 0; static time_t rfs4_state_cache_time = 0; static time_t rfs4_lo_state_cache_time = 0; static time_t rfs4_lockowner_cache_time = 0; static time_t rfs4_file_cache_time = 0; static time_t rfs4_deleg_state_cache_time = 0; static bool_t rfs4_client_create(rfs4_entry_t, void *); static void rfs4_dss_remove_cpleaf(rfs4_client_t *); static void rfs4_dss_remove_leaf(rfs4_servinst_t *, char *, char *); static void rfs4_client_destroy(rfs4_entry_t); static bool_t rfs4_client_expiry(rfs4_entry_t); static uint32_t clientid_hash(void *); static bool_t clientid_compare(rfs4_entry_t, void *); static void *clientid_mkkey(rfs4_entry_t); static uint32_t nfsclnt_hash(void *); static bool_t nfsclnt_compare(rfs4_entry_t, void *); static void *nfsclnt_mkkey(rfs4_entry_t); static bool_t rfs4_openowner_create(rfs4_entry_t, void *); static void rfs4_openowner_destroy(rfs4_entry_t); static bool_t rfs4_openowner_expiry(rfs4_entry_t); static uint32_t openowner_hash(void *); static bool_t openowner_compare(rfs4_entry_t, void *); static void *openowner_mkkey(rfs4_entry_t); static bool_t rfs4_state_create(rfs4_entry_t, void *); static void rfs4_state_destroy(rfs4_entry_t); static bool_t rfs4_state_expiry(rfs4_entry_t); static uint32_t state_hash(void *); static bool_t state_compare(rfs4_entry_t, void *); static void *state_mkkey(rfs4_entry_t); static uint32_t state_owner_file_hash(void *); static bool_t state_owner_file_compare(rfs4_entry_t, void *); static void *state_owner_file_mkkey(rfs4_entry_t); static uint32_t state_file_hash(void *); static bool_t state_file_compare(rfs4_entry_t, void *); static void *state_file_mkkey(rfs4_entry_t); static bool_t rfs4_lo_state_create(rfs4_entry_t, void *); static void rfs4_lo_state_destroy(rfs4_entry_t); static bool_t rfs4_lo_state_expiry(rfs4_entry_t); static uint32_t lo_state_hash(void *); static bool_t lo_state_compare(rfs4_entry_t, void *); static void *lo_state_mkkey(rfs4_entry_t); static uint32_t lo_state_lo_hash(void *); static bool_t lo_state_lo_compare(rfs4_entry_t, void *); static void *lo_state_lo_mkkey(rfs4_entry_t); static bool_t rfs4_lockowner_create(rfs4_entry_t, void *); static void rfs4_lockowner_destroy(rfs4_entry_t); static bool_t rfs4_lockowner_expiry(rfs4_entry_t); static uint32_t lockowner_hash(void *); static bool_t lockowner_compare(rfs4_entry_t, void *); static void *lockowner_mkkey(rfs4_entry_t); static uint32_t pid_hash(void *); static bool_t pid_compare(rfs4_entry_t, void *); static void *pid_mkkey(rfs4_entry_t); static bool_t rfs4_file_create(rfs4_entry_t, void *); static void rfs4_file_destroy(rfs4_entry_t); static uint32_t file_hash(void *); static bool_t file_compare(rfs4_entry_t, void *); static void *file_mkkey(rfs4_entry_t); static bool_t rfs4_deleg_state_create(rfs4_entry_t, void *); static void rfs4_deleg_state_destroy(rfs4_entry_t); static bool_t rfs4_deleg_state_expiry(rfs4_entry_t); static uint32_t deleg_hash(void *); static bool_t deleg_compare(rfs4_entry_t, void *); static void *deleg_mkkey(rfs4_entry_t); static uint32_t deleg_state_hash(void *); static bool_t deleg_state_compare(rfs4_entry_t, void *); static void *deleg_state_mkkey(rfs4_entry_t); static void rfs4_state_rele_nounlock(rfs4_state_t *); static int rfs4_ss_enabled = 0; extern void (*rfs4_client_clrst)(struct nfs4clrst_args *); void rfs4_ss_pnfree(rfs4_ss_pn_t *ss_pn) { kmem_free(ss_pn, sizeof (rfs4_ss_pn_t)); } static rfs4_ss_pn_t * rfs4_ss_pnalloc(char *dir, char *leaf) { rfs4_ss_pn_t *ss_pn; int dir_len, leaf_len; /* * validate we have a resonable path * (account for the '/' and trailing null) */ if ((dir_len = strlen(dir)) > MAXPATHLEN || (leaf_len = strlen(leaf)) > MAXNAMELEN || (dir_len + leaf_len + 2) > MAXPATHLEN) { return (NULL); } ss_pn = kmem_alloc(sizeof (rfs4_ss_pn_t), KM_SLEEP); (void) snprintf(ss_pn->pn, MAXPATHLEN, "%s/%s", dir, leaf); /* Handy pointer to just the leaf name */ ss_pn->leaf = ss_pn->pn + dir_len + 1; return (ss_pn); } /* * Move the "leaf" filename from "sdir" directory * to the "ddir" directory. Return the pathname of * the destination unless the rename fails in which * case we need to return the source pathname. */ static rfs4_ss_pn_t * rfs4_ss_movestate(char *sdir, char *ddir, char *leaf) { rfs4_ss_pn_t *src, *dst; if ((src = rfs4_ss_pnalloc(sdir, leaf)) == NULL) return (NULL); if ((dst = rfs4_ss_pnalloc(ddir, leaf)) == NULL) { rfs4_ss_pnfree(src); return (NULL); } /* * If the rename fails we shall return the src * pathname and free the dst. Otherwise we need * to free the src and return the dst pathanme. */ if (vn_rename(src->pn, dst->pn, UIO_SYSSPACE)) { rfs4_ss_pnfree(dst); return (src); } rfs4_ss_pnfree(src); return (dst); } static rfs4_oldstate_t * rfs4_ss_getstate(vnode_t *dvp, rfs4_ss_pn_t *ss_pn) { struct uio uio; struct iovec iov[3]; rfs4_oldstate_t *cl_ss = NULL; vnode_t *vp; vattr_t va; uint_t id_len; int err, kill_file, file_vers; if (ss_pn == NULL) return (NULL); /* * open the state file. */ if (vn_open(ss_pn->pn, UIO_SYSSPACE, FREAD, 0, &vp, 0, 0) != 0) { return (NULL); } if (vp->v_type != VREG) { (void) VOP_CLOSE(vp, FREAD, 1, (offset_t)0, CRED(), NULL); VN_RELE(vp); return (NULL); } err = VOP_ACCESS(vp, VREAD, 0, CRED(), NULL); if (err) { /* * We don't have read access? better get the heck out. */ (void) VOP_CLOSE(vp, FREAD, 1, (offset_t)0, CRED(), NULL); VN_RELE(vp); return (NULL); } (void) VOP_RWLOCK(vp, V_WRITELOCK_FALSE, NULL); /* * get the file size to do some basic validation */ va.va_mask = AT_SIZE; err = VOP_GETATTR(vp, &va, 0, CRED(), NULL); kill_file = (va.va_size == 0 || va.va_size < (NFS4_VERIFIER_SIZE + sizeof (uint_t)+1)); if (err || kill_file) { VOP_RWUNLOCK(vp, V_WRITELOCK_FALSE, NULL); (void) VOP_CLOSE(vp, FREAD, 1, (offset_t)0, CRED(), NULL); VN_RELE(vp); if (kill_file) { (void) VOP_REMOVE(dvp, ss_pn->leaf, CRED(), NULL, 0); } return (NULL); } cl_ss = kmem_alloc(sizeof (rfs4_oldstate_t), KM_SLEEP); /* * build iovecs to read in the file_version, verifier and id_len */ iov[0].iov_base = (caddr_t)&file_vers; iov[0].iov_len = sizeof (int); iov[1].iov_base = (caddr_t)&cl_ss->cl_id4.verifier; iov[1].iov_len = NFS4_VERIFIER_SIZE; iov[2].iov_base = (caddr_t)&id_len; iov[2].iov_len = sizeof (uint_t); uio.uio_iov = iov; uio.uio_iovcnt = 3; uio.uio_segflg = UIO_SYSSPACE; uio.uio_loffset = 0; uio.uio_resid = sizeof (int) + NFS4_VERIFIER_SIZE + sizeof (uint_t); if (err = VOP_READ(vp, &uio, FREAD, CRED(), NULL)) { VOP_RWUNLOCK(vp, V_WRITELOCK_FALSE, NULL); (void) VOP_CLOSE(vp, FREAD, 1, (offset_t)0, CRED(), NULL); VN_RELE(vp); kmem_free(cl_ss, sizeof (rfs4_oldstate_t)); return (NULL); } /* * if the file_version doesn't match or if the * id_len is zero or the combination of the verifier, * id_len and id_val is bigger than the file we have * a problem. If so ditch the file. */ kill_file = (file_vers != NFS4_SS_VERSION || id_len == 0 || (id_len + NFS4_VERIFIER_SIZE + sizeof (uint_t)) > va.va_size); if (err || kill_file) { VOP_RWUNLOCK(vp, V_WRITELOCK_FALSE, NULL); (void) VOP_CLOSE(vp, FREAD, 1, (offset_t)0, CRED(), NULL); VN_RELE(vp); kmem_free(cl_ss, sizeof (rfs4_oldstate_t)); if (kill_file) { (void) VOP_REMOVE(dvp, ss_pn->leaf, CRED(), NULL, 0); } return (NULL); } /* * now get the client id value */ cl_ss->cl_id4.id_val = kmem_alloc(id_len, KM_SLEEP); iov[0].iov_base = cl_ss->cl_id4.id_val; iov[0].iov_len = id_len; uio.uio_iov = iov; uio.uio_iovcnt = 1; uio.uio_segflg = UIO_SYSSPACE; uio.uio_resid = cl_ss->cl_id4.id_len = id_len; if (err = VOP_READ(vp, &uio, FREAD, CRED(), NULL)) { VOP_RWUNLOCK(vp, V_WRITELOCK_FALSE, NULL); (void) VOP_CLOSE(vp, FREAD, 1, (offset_t)0, CRED(), NULL); VN_RELE(vp); kmem_free(cl_ss->cl_id4.id_val, id_len); kmem_free(cl_ss, sizeof (rfs4_oldstate_t)); return (NULL); } VOP_RWUNLOCK(vp, V_WRITELOCK_FALSE, NULL); (void) VOP_CLOSE(vp, FREAD, 1, (offset_t)0, CRED(), NULL); VN_RELE(vp); return (cl_ss); } #ifdef nextdp #undef nextdp #endif #define nextdp(dp) ((struct dirent64 *)((char *)(dp) + (dp)->d_reclen)) /* * Add entries from statedir to supplied oldstate list. * Optionally, move all entries from statedir -> destdir. */ void rfs4_ss_oldstate(rfs4_oldstate_t *oldstate, char *statedir, char *destdir) { rfs4_ss_pn_t *ss_pn; rfs4_oldstate_t *cl_ss = NULL; char *dirt = NULL; int err, dir_eof = 0, size = 0; vnode_t *dvp; struct iovec iov; struct uio uio; struct dirent64 *dep; offset_t dirchunk_offset = 0; /* * open the state directory */ if (vn_open(statedir, UIO_SYSSPACE, FREAD, 0, &dvp, 0, 0)) return; if (dvp->v_type != VDIR || VOP_ACCESS(dvp, VREAD, 0, CRED(), NULL)) goto out; dirt = kmem_alloc(RFS4_SS_DIRSIZE, KM_SLEEP); /* * Get and process the directory entries */ while (!dir_eof) { (void) VOP_RWLOCK(dvp, V_WRITELOCK_FALSE, NULL); iov.iov_base = dirt; iov.iov_len = RFS4_SS_DIRSIZE; uio.uio_iov = &iov; uio.uio_iovcnt = 1; uio.uio_segflg = UIO_SYSSPACE; uio.uio_loffset = dirchunk_offset; uio.uio_resid = RFS4_SS_DIRSIZE; err = VOP_READDIR(dvp, &uio, CRED(), &dir_eof, NULL, 0); VOP_RWUNLOCK(dvp, V_WRITELOCK_FALSE, NULL); if (err) goto out; size = RFS4_SS_DIRSIZE - uio.uio_resid; /* * Process all the directory entries in this * readdir chunk */ for (dep = (struct dirent64 *)dirt; size > 0; dep = nextdp(dep)) { size -= dep->d_reclen; dirchunk_offset = dep->d_off; /* * Skip '.' and '..' */ if (NFS_IS_DOTNAME(dep->d_name)) continue; ss_pn = rfs4_ss_pnalloc(statedir, dep->d_name); if (ss_pn == NULL) continue; if (cl_ss = rfs4_ss_getstate(dvp, ss_pn)) { if (destdir != NULL) { rfs4_ss_pnfree(ss_pn); cl_ss->ss_pn = rfs4_ss_movestate( statedir, destdir, dep->d_name); } else { cl_ss->ss_pn = ss_pn; } insque(cl_ss, oldstate); } else { rfs4_ss_pnfree(ss_pn); } } } out: (void) VOP_CLOSE(dvp, FREAD, 1, (offset_t)0, CRED(), NULL); VN_RELE(dvp); if (dirt) kmem_free((caddr_t)dirt, RFS4_SS_DIRSIZE); } static void rfs4_ss_init(void) { int npaths = 1; char *default_dss_path = NFS4_DSS_VAR_DIR; /* read the default stable storage state */ rfs4_dss_readstate(npaths, &default_dss_path); rfs4_ss_enabled = 1; } static void rfs4_ss_fini(void) { rfs4_servinst_t *sip; mutex_enter(&rfs4_servinst_lock); sip = rfs4_cur_servinst; while (sip != NULL) { rfs4_dss_clear_oldstate(sip); sip = sip->next; } mutex_exit(&rfs4_servinst_lock); } /* * Remove all oldstate files referenced by this servinst. */ static void rfs4_dss_clear_oldstate(rfs4_servinst_t *sip) { rfs4_oldstate_t *os_head, *osp; rw_enter(&sip->oldstate_lock, RW_WRITER); os_head = sip->oldstate; if (os_head == NULL) return; /* skip dummy entry */ osp = os_head->next; while (osp != os_head) { char *leaf = osp->ss_pn->leaf; rfs4_oldstate_t *os_next; rfs4_dss_remove_leaf(sip, NFS4_DSS_OLDSTATE_LEAF, leaf); if (osp->cl_id4.id_val) kmem_free(osp->cl_id4.id_val, osp->cl_id4.id_len); if (osp->ss_pn) kmem_free(osp->ss_pn, sizeof (rfs4_ss_pn_t)); os_next = osp->next; remque(osp); kmem_free(osp, sizeof (rfs4_oldstate_t)); osp = os_next; } /* free dummy entry */ kmem_free(osp, sizeof (rfs4_oldstate_t)); sip->oldstate = NULL; rw_exit(&sip->oldstate_lock); } /* * Form the state and oldstate paths, and read in the stable storage files. */ void rfs4_dss_readstate(int npaths, char **paths) { int i; char *state, *oldstate; state = kmem_alloc(MAXPATHLEN, KM_SLEEP); oldstate = kmem_alloc(MAXPATHLEN, KM_SLEEP); for (i = 0; i < npaths; i++) { char *path = paths[i]; (void) sprintf(state, "%s/%s", path, NFS4_DSS_STATE_LEAF); (void) sprintf(oldstate, "%s/%s", path, NFS4_DSS_OLDSTATE_LEAF); /* * Populate the current server instance's oldstate list. * * 1. Read stable storage data from old state directory, * leaving its contents alone. * * 2. Read stable storage data from state directory, * and move the latter's contents to old state * directory. */ rfs4_ss_oldstate(rfs4_cur_servinst->oldstate, oldstate, NULL); rfs4_ss_oldstate(rfs4_cur_servinst->oldstate, state, oldstate); } kmem_free(state, MAXPATHLEN); kmem_free(oldstate, MAXPATHLEN); } /* * Check if we are still in grace and if the client can be * granted permission to perform reclaims. */ void rfs4_ss_chkclid(rfs4_client_t *cp) { rfs4_servinst_t *sip; /* * It should be sufficient to check the oldstate data for just * this client's instance. However, since our per-instance * client grouping is solely temporal, HA-NFSv4 RG failover * might result in clients of the same RG being partitioned into * separate instances. * * Until the client grouping is improved, we must check the * oldstate data for all instances with an active grace period. * * This also serves as the mechanism to remove stale oldstate data. * The first time we check an instance after its grace period has * expired, the oldstate data should be cleared. * * Start at the current instance, and walk the list backwards * to the first. */ mutex_enter(&rfs4_servinst_lock); for (sip = rfs4_cur_servinst; sip != NULL; sip = sip->prev) { rfs4_ss_chkclid_sip(cp, sip); /* if the above check found this client, we're done */ if (cp->rc_can_reclaim) break; } mutex_exit(&rfs4_servinst_lock); } static void rfs4_ss_chkclid_sip(rfs4_client_t *cp, rfs4_servinst_t *sip) { rfs4_oldstate_t *osp, *os_head; /* short circuit everything if this server instance has no oldstate */ rw_enter(&sip->oldstate_lock, RW_READER); os_head = sip->oldstate; rw_exit(&sip->oldstate_lock); if (os_head == NULL) return; /* * If this server instance is no longer in a grace period then * the client won't be able to reclaim. No further need for this * instance's oldstate data, so it can be cleared. */ if (!rfs4_servinst_in_grace(sip)) return; /* this instance is still in grace; search for the clientid */ rw_enter(&sip->oldstate_lock, RW_READER); os_head = sip->oldstate; /* skip dummy entry */ osp = os_head->next; while (osp != os_head) { if (osp->cl_id4.id_len == cp->rc_nfs_client.id_len) { if (bcmp(osp->cl_id4.id_val, cp->rc_nfs_client.id_val, osp->cl_id4.id_len) == 0) { cp->rc_can_reclaim = 1; break; } } osp = osp->next; } rw_exit(&sip->oldstate_lock); } /* * Place client information into stable storage: 1/3. * First, generate the leaf filename, from the client's IP address and * the server-generated short-hand clientid. */ void rfs4_ss_clid(rfs4_client_t *cp, struct svc_req *req) { const char *kinet_ntop6(uchar_t *, char *, size_t); char leaf[MAXNAMELEN], buf[INET6_ADDRSTRLEN]; struct sockaddr *ca; uchar_t *b; if (rfs4_ss_enabled == 0) { return; } buf[0] = 0; ca = (struct sockaddr *)svc_getrpccaller(req->rq_xprt)->buf; if (ca == NULL) { return; } /* * Convert the caller's IP address to a dotted string */ if (ca->sa_family == AF_INET) { bcopy(svc_getrpccaller(req->rq_xprt)->buf, &cp->rc_addr, sizeof (struct sockaddr_in)); b = (uchar_t *)&((struct sockaddr_in *)ca)->sin_addr; (void) sprintf(buf, "%03d.%03d.%03d.%03d", b[0] & 0xFF, b[1] & 0xFF, b[2] & 0xFF, b[3] & 0xFF); } else if (ca->sa_family == AF_INET6) { struct sockaddr_in6 *sin6; sin6 = (struct sockaddr_in6 *)ca; bcopy(svc_getrpccaller(req->rq_xprt)->buf, &cp->rc_addr, sizeof (struct sockaddr_in6)); (void) kinet_ntop6((uchar_t *)&sin6->sin6_addr, buf, INET6_ADDRSTRLEN); } (void) snprintf(leaf, MAXNAMELEN, "%s-%llx", buf, (longlong_t)cp->rc_clientid); rfs4_ss_clid_write(cp, leaf); } /* * Place client information into stable storage: 2/3. * DSS: distributed stable storage: the file may need to be written to * multiple directories. */ static void rfs4_ss_clid_write(rfs4_client_t *cp, char *leaf) { rfs4_servinst_t *sip; /* * It should be sufficient to write the leaf file to (all) DSS paths * associated with just this client's instance. However, since our * per-instance client grouping is solely temporal, HA-NFSv4 RG * failover might result in us losing DSS data. * * Until the client grouping is improved, we must write the DSS data * to all instances' paths. Start at the current instance, and * walk the list backwards to the first. */ mutex_enter(&rfs4_servinst_lock); for (sip = rfs4_cur_servinst; sip != NULL; sip = sip->prev) { int i, npaths = sip->dss_npaths; /* write the leaf file to all DSS paths */ for (i = 0; i < npaths; i++) { rfs4_dss_path_t *dss_path = sip->dss_paths[i]; /* HA-NFSv4 path might have been failed-away from us */ if (dss_path == NULL) continue; rfs4_ss_clid_write_one(cp, dss_path->path, leaf); } } mutex_exit(&rfs4_servinst_lock); } /* * Place client information into stable storage: 3/3. * Write the stable storage data to the requested file. */ static void rfs4_ss_clid_write_one(rfs4_client_t *cp, char *dss_path, char *leaf) { int ioflag; int file_vers = NFS4_SS_VERSION; size_t dirlen; struct uio uio; struct iovec iov[4]; char *dir; rfs4_ss_pn_t *ss_pn; vnode_t *vp; nfs_client_id4 *cl_id4 = &(cp->rc_nfs_client); /* allow 2 extra bytes for '/' & NUL */ dirlen = strlen(dss_path) + strlen(NFS4_DSS_STATE_LEAF) + 2; dir = kmem_alloc(dirlen, KM_SLEEP); (void) sprintf(dir, "%s/%s", dss_path, NFS4_DSS_STATE_LEAF); ss_pn = rfs4_ss_pnalloc(dir, leaf); /* rfs4_ss_pnalloc takes its own copy */ kmem_free(dir, dirlen); if (ss_pn == NULL) return; if (vn_open(ss_pn->pn, UIO_SYSSPACE, FCREAT|FWRITE, 0600, &vp, CRCREAT, 0)) { rfs4_ss_pnfree(ss_pn); return; } /* * We need to record leaf - i.e. the filename - so that we know * what to remove, in the future. However, the dir part of cp->ss_pn * should never be referenced directly, since it's potentially only * one of several paths with this leaf in it. */ if (cp->rc_ss_pn != NULL) { if (strcmp(cp->rc_ss_pn->leaf, leaf) == 0) { /* we've already recorded *this* leaf */ rfs4_ss_pnfree(ss_pn); } else { /* replace with this leaf */ rfs4_ss_pnfree(cp->rc_ss_pn); cp->rc_ss_pn = ss_pn; } } else { cp->rc_ss_pn = ss_pn; } /* * Build a scatter list that points to the nfs_client_id4 */ iov[0].iov_base = (caddr_t)&file_vers; iov[0].iov_len = sizeof (int); iov[1].iov_base = (caddr_t)&(cl_id4->verifier); iov[1].iov_len = NFS4_VERIFIER_SIZE; iov[2].iov_base = (caddr_t)&(cl_id4->id_len); iov[2].iov_len = sizeof (uint_t); iov[3].iov_base = (caddr_t)cl_id4->id_val; iov[3].iov_len = cl_id4->id_len; uio.uio_iov = iov; uio.uio_iovcnt = 4; uio.uio_loffset = 0; uio.uio_segflg = UIO_SYSSPACE; uio.uio_llimit = (rlim64_t)MAXOFFSET_T; uio.uio_resid = cl_id4->id_len + sizeof (int) + NFS4_VERIFIER_SIZE + sizeof (uint_t); ioflag = uio.uio_fmode = (FWRITE|FSYNC); uio.uio_extflg = UIO_COPY_DEFAULT; (void) VOP_RWLOCK(vp, V_WRITELOCK_TRUE, NULL); /* write the full client id to the file. */ (void) VOP_WRITE(vp, &uio, ioflag, CRED(), NULL); VOP_RWUNLOCK(vp, V_WRITELOCK_TRUE, NULL); (void) VOP_CLOSE(vp, FWRITE, 1, (offset_t)0, CRED(), NULL); VN_RELE(vp); } /* * DSS: distributed stable storage. * Unpack the list of paths passed by nfsd. * Use nvlist_alloc(9F) to manage the data. * The caller is responsible for allocating and freeing the buffer. */ int rfs4_dss_setpaths(char *buf, size_t buflen) { int error; /* * If this is a "warm start", i.e. we previously had DSS paths, * preserve the old paths. */ if (rfs4_dss_paths != NULL) { /* * Before we lose the ptr, destroy the nvlist and pathnames * array from the warm start before this one. */ if (rfs4_dss_oldpaths) nvlist_free(rfs4_dss_oldpaths); rfs4_dss_oldpaths = rfs4_dss_paths; } /* unpack the buffer into a searchable nvlist */ error = nvlist_unpack(buf, buflen, &rfs4_dss_paths, KM_SLEEP); if (error) return (error); /* * Search the nvlist for the pathnames nvpair (which is the only nvpair * in the list, and record its location. */ error = nvlist_lookup_string_array(rfs4_dss_paths, NFS4_DSS_NVPAIR_NAME, &rfs4_dss_newpaths, &rfs4_dss_numnewpaths); return (error); } /* * Ultimately the nfssys() call NFS4_CLR_STATE endsup here * to find and mark the client for forced expire. */ static void rfs4_client_scrub(rfs4_entry_t ent, void *arg) { rfs4_client_t *cp = (rfs4_client_t *)ent; struct nfs4clrst_args *clr = arg; struct sockaddr_in6 *ent_sin6; struct in6_addr clr_in6; struct sockaddr_in *ent_sin; struct in_addr clr_in; if (clr->addr_type != cp->rc_addr.ss_family) { return; } switch (clr->addr_type) { case AF_INET6: /* copyin the address from user space */ if (copyin(clr->ap, &clr_in6, sizeof (clr_in6))) { break; } ent_sin6 = (struct sockaddr_in6 *)&cp->rc_addr; /* * now compare, and if equivalent mark entry * for forced expiration */ if (IN6_ARE_ADDR_EQUAL(&ent_sin6->sin6_addr, &clr_in6)) { cp->rc_forced_expire = 1; } break; case AF_INET: /* copyin the address from user space */ if (copyin(clr->ap, &clr_in, sizeof (clr_in))) { break; } ent_sin = (struct sockaddr_in *)&cp->rc_addr; /* * now compare, and if equivalent mark entry * for forced expiration */ if (ent_sin->sin_addr.s_addr == clr_in.s_addr) { cp->rc_forced_expire = 1; } break; default: /* force this assert to fail */ ASSERT(clr->addr_type != clr->addr_type); } } /* * This is called from nfssys() in order to clear server state * for the specified client IP Address. */ void rfs4_clear_client_state(struct nfs4clrst_args *clr) { (void) rfs4_dbe_walk(rfs4_client_tab, rfs4_client_scrub, clr); } /* * Used to initialize the NFSv4 server's state or database. All of * the tables are created and timers are set. Only called when NFSv4 * service is provided. */ void rfs4_state_init() { int start_grace; extern boolean_t rfs4_cpr_callb(void *, int); char *dss_path = NFS4_DSS_VAR_DIR; mutex_enter(&rfs4_state_lock); /* * If the server state database has already been initialized, * skip it */ if (rfs4_server_state != NULL) { mutex_exit(&rfs4_state_lock); return; } rw_init(&rfs4_findclient_lock, NULL, RW_DEFAULT, NULL); /* * Set the boot time. If the server * has been restarted quickly and has had the opportunity to * service clients, then the start_time needs to be bumped * regardless. A small window but it exists... */ if (rfs4_start_time != gethrestime_sec()) rfs4_start_time = gethrestime_sec(); else rfs4_start_time++; /* DSS: distributed stable storage: initialise served paths list */ rfs4_dss_pathlist = NULL; /* * Create the first server instance, or a new one if the server has * been restarted; see above comments on rfs4_start_time. Don't * start its grace period; that will be done later, to maximise the * clients' recovery window. */ start_grace = 0; rfs4_servinst_create(start_grace, 1, &dss_path); /* reset the "first NFSv4 request" status */ rfs4_seen_first_compound = 0; /* * Add a CPR callback so that we can update client * access times to extend the lease after a suspend * and resume (using the same class as rpcmod/connmgr) */ cpr_id = callb_add(rfs4_cpr_callb, 0, CB_CL_CPR_RPC, "rfs4"); /* set the various cache timers for table creation */ if (rfs4_client_cache_time == 0) rfs4_client_cache_time = CLIENT_CACHE_TIME; if (rfs4_openowner_cache_time == 0) rfs4_openowner_cache_time = OPENOWNER_CACHE_TIME; if (rfs4_state_cache_time == 0) rfs4_state_cache_time = STATE_CACHE_TIME; if (rfs4_lo_state_cache_time == 0) rfs4_lo_state_cache_time = LO_STATE_CACHE_TIME; if (rfs4_lockowner_cache_time == 0) rfs4_lockowner_cache_time = LOCKOWNER_CACHE_TIME; if (rfs4_file_cache_time == 0) rfs4_file_cache_time = FILE_CACHE_TIME; if (rfs4_deleg_state_cache_time == 0) rfs4_deleg_state_cache_time = DELEG_STATE_CACHE_TIME; /* Create the overall database to hold all server state */ rfs4_server_state = rfs4_database_create(rfs4_database_debug); /* Now create the individual tables */ rfs4_client_cache_time *= rfs4_lease_time; rfs4_client_tab = rfs4_table_create(rfs4_server_state, "Client", rfs4_client_cache_time, 2, rfs4_client_create, rfs4_client_destroy, rfs4_client_expiry, sizeof (rfs4_client_t), TABSIZE, MAXTABSZ/8, 100); rfs4_nfsclnt_idx = rfs4_index_create(rfs4_client_tab, "nfs_client_id4", nfsclnt_hash, nfsclnt_compare, nfsclnt_mkkey, TRUE); rfs4_clientid_idx = rfs4_index_create(rfs4_client_tab, "client_id", clientid_hash, clientid_compare, clientid_mkkey, FALSE); rfs4_openowner_cache_time *= rfs4_lease_time; rfs4_openowner_tab = rfs4_table_create(rfs4_server_state, "OpenOwner", rfs4_openowner_cache_time, 1, rfs4_openowner_create, rfs4_openowner_destroy, rfs4_openowner_expiry, sizeof (rfs4_openowner_t), TABSIZE, MAXTABSZ, 100); rfs4_openowner_idx = rfs4_index_create(rfs4_openowner_tab, "open_owner4", openowner_hash, openowner_compare, openowner_mkkey, TRUE); rfs4_state_cache_time *= rfs4_lease_time; rfs4_state_tab = rfs4_table_create(rfs4_server_state, "OpenStateID", rfs4_state_cache_time, 3, rfs4_state_create, rfs4_state_destroy, rfs4_state_expiry, sizeof (rfs4_state_t), TABSIZE, MAXTABSZ, 100); rfs4_state_owner_file_idx = rfs4_index_create(rfs4_state_tab, "Openowner-File", state_owner_file_hash, state_owner_file_compare, state_owner_file_mkkey, TRUE); rfs4_state_idx = rfs4_index_create(rfs4_state_tab, "State-id", state_hash, state_compare, state_mkkey, FALSE); rfs4_state_file_idx = rfs4_index_create(rfs4_state_tab, "File", state_file_hash, state_file_compare, state_file_mkkey, FALSE); rfs4_lo_state_cache_time *= rfs4_lease_time; rfs4_lo_state_tab = rfs4_table_create(rfs4_server_state, "LockStateID", rfs4_lo_state_cache_time, 2, rfs4_lo_state_create, rfs4_lo_state_destroy, rfs4_lo_state_expiry, sizeof (rfs4_lo_state_t), TABSIZE, MAXTABSZ, 100); rfs4_lo_state_owner_idx = rfs4_index_create(rfs4_lo_state_tab, "lockownerxstate", lo_state_lo_hash, lo_state_lo_compare, lo_state_lo_mkkey, TRUE); rfs4_lo_state_idx = rfs4_index_create(rfs4_lo_state_tab, "State-id", lo_state_hash, lo_state_compare, lo_state_mkkey, FALSE); rfs4_lockowner_cache_time *= rfs4_lease_time; rfs4_lockowner_tab = rfs4_table_create(rfs4_server_state, "Lockowner", rfs4_lockowner_cache_time, 2, rfs4_lockowner_create, rfs4_lockowner_destroy, rfs4_lockowner_expiry, sizeof (rfs4_lockowner_t), TABSIZE, MAXTABSZ, 100); rfs4_lockowner_idx = rfs4_index_create(rfs4_lockowner_tab, "lock_owner4", lockowner_hash, lockowner_compare, lockowner_mkkey, TRUE); rfs4_lockowner_pid_idx = rfs4_index_create(rfs4_lockowner_tab, "pid", pid_hash, pid_compare, pid_mkkey, FALSE); rfs4_file_cache_time *= rfs4_lease_time; rfs4_file_tab = rfs4_table_create(rfs4_server_state, "File", rfs4_file_cache_time, 1, rfs4_file_create, rfs4_file_destroy, NULL, sizeof (rfs4_file_t), TABSIZE, MAXTABSZ, -1); rfs4_file_idx = rfs4_index_create(rfs4_file_tab, "Filehandle", file_hash, file_compare, file_mkkey, TRUE); rfs4_deleg_state_cache_time *= rfs4_lease_time; rfs4_deleg_state_tab = rfs4_table_create(rfs4_server_state, "DelegStateID", rfs4_deleg_state_cache_time, 2, rfs4_deleg_state_create, rfs4_deleg_state_destroy, rfs4_deleg_state_expiry, sizeof (rfs4_deleg_state_t), TABSIZE, MAXTABSZ, 100); rfs4_deleg_idx = rfs4_index_create(rfs4_deleg_state_tab, "DelegByFileClient", deleg_hash, deleg_compare, deleg_mkkey, TRUE); rfs4_deleg_state_idx = rfs4_index_create(rfs4_deleg_state_tab, "DelegState", deleg_state_hash, deleg_state_compare, deleg_state_mkkey, FALSE); /* * Init the stable storage. */ rfs4_ss_init(); rfs4_client_clrst = rfs4_clear_client_state; mutex_exit(&rfs4_state_lock); } /* * Used at server shutdown to cleanup all of the NFSv4 server's structures * and other state. */ void rfs4_state_fini() { rfs4_database_t *dbp; mutex_enter(&rfs4_state_lock); if (rfs4_server_state == NULL) { mutex_exit(&rfs4_state_lock); return; } rfs4_client_clrst = NULL; rfs4_set_deleg_policy(SRV_NEVER_DELEGATE); dbp = rfs4_server_state; rfs4_server_state = NULL; /* * Cleanup the CPR callback. */ if (cpr_id) (void) callb_delete(cpr_id); rw_destroy(&rfs4_findclient_lock); /* First stop all of the reaper threads in the database */ rfs4_database_shutdown(dbp); /* clean up any dangling stable storage structures */ rfs4_ss_fini(); /* Now actually destroy/release the database and its tables */ rfs4_database_destroy(dbp); /* Reset the cache timers for next time */ rfs4_client_cache_time = 0; rfs4_openowner_cache_time = 0; rfs4_state_cache_time = 0; rfs4_lo_state_cache_time = 0; rfs4_lockowner_cache_time = 0; rfs4_file_cache_time = 0; rfs4_deleg_state_cache_time = 0; mutex_exit(&rfs4_state_lock); /* destroy server instances and current instance ptr */ rfs4_servinst_destroy_all(); /* reset the "first NFSv4 request" status */ rfs4_seen_first_compound = 0; /* DSS: distributed stable storage */ if (rfs4_dss_oldpaths) nvlist_free(rfs4_dss_oldpaths); if (rfs4_dss_paths) nvlist_free(rfs4_dss_paths); rfs4_dss_paths = rfs4_dss_oldpaths = NULL; } typedef union { struct { uint32_t start_time; uint32_t c_id; } impl_id; clientid4 id4; } cid; static int foreign_stateid(stateid_t *id); static int foreign_clientid(cid *cidp); static void embed_nodeid(cid *cidp); typedef union { struct { uint32_t c_id; uint32_t gen_num; } cv_impl; verifier4 confirm_verf; } scid_confirm_verf; static uint32_t clientid_hash(void *key) { cid *idp = key; return (idp->impl_id.c_id); } static bool_t clientid_compare(rfs4_entry_t entry, void *key) { rfs4_client_t *cp = (rfs4_client_t *)entry; clientid4 *idp = key; return (*idp == cp->rc_clientid); } static void * clientid_mkkey(rfs4_entry_t entry) { rfs4_client_t *cp = (rfs4_client_t *)entry; return (&cp->rc_clientid); } static uint32_t nfsclnt_hash(void *key) { nfs_client_id4 *client = key; int i; uint32_t hash = 0; for (i = 0; i < client->id_len; i++) { hash <<= 1; hash += (uint_t)client->id_val[i]; } return (hash); } static bool_t nfsclnt_compare(rfs4_entry_t entry, void *key) { rfs4_client_t *cp = (rfs4_client_t *)entry; nfs_client_id4 *nfs_client = key; if (cp->rc_nfs_client.id_len != nfs_client->id_len) return (FALSE); return (bcmp(cp->rc_nfs_client.id_val, nfs_client->id_val, nfs_client->id_len) == 0); } static void * nfsclnt_mkkey(rfs4_entry_t entry) { rfs4_client_t *cp = (rfs4_client_t *)entry; return (&cp->rc_nfs_client); } static bool_t rfs4_client_expiry(rfs4_entry_t u_entry) { rfs4_client_t *cp = (rfs4_client_t *)u_entry; bool_t cp_expired; if (rfs4_dbe_is_invalid(cp->rc_dbe)) return (TRUE); /* * If the sysadmin has used clear_locks for this * entry then forced_expire will be set and we * want this entry to be reaped. Or the entry * has exceeded its lease period. */ cp_expired = (cp->rc_forced_expire || (gethrestime_sec() - cp->rc_last_access > rfs4_lease_time)); if (!cp->rc_ss_remove && cp_expired) cp->rc_ss_remove = 1; return (cp_expired); } /* * Remove the leaf file from all distributed stable storage paths. */ static void rfs4_dss_remove_cpleaf(rfs4_client_t *cp) { char *leaf = cp->rc_ss_pn->leaf; rfs4_dss_remove_leaf(cp->rc_server_instance, NFS4_DSS_STATE_LEAF, leaf); } static void rfs4_dss_remove_leaf(rfs4_servinst_t *sip, char *dir_leaf, char *leaf) { int i, npaths = sip->dss_npaths; for (i = 0; i < npaths; i++) { rfs4_dss_path_t *dss_path = sip->dss_paths[i]; char *path, *dir; size_t pathlen; /* the HA-NFSv4 path might have been failed-over away from us */ if (dss_path == NULL) continue; dir = dss_path->path; /* allow 3 extra bytes for two '/' & a NUL */ pathlen = strlen(dir) + strlen(dir_leaf) + strlen(leaf) + 3; path = kmem_alloc(pathlen, KM_SLEEP); (void) sprintf(path, "%s/%s/%s", dir, dir_leaf, leaf); (void) vn_remove(path, UIO_SYSSPACE, RMFILE); kmem_free(path, pathlen); } } static void rfs4_client_destroy(rfs4_entry_t u_entry) { rfs4_client_t *cp = (rfs4_client_t *)u_entry; mutex_destroy(cp->rc_cbinfo.cb_lock); cv_destroy(cp->rc_cbinfo.cb_cv); cv_destroy(cp->rc_cbinfo.cb_cv_nullcaller); list_destroy(&cp->rc_openownerlist); /* free callback info */ rfs4_cbinfo_free(&cp->rc_cbinfo); if (cp->rc_cp_confirmed) rfs4_client_rele(cp->rc_cp_confirmed); if (cp->rc_ss_pn) { /* check if the stable storage files need to be removed */ if (cp->rc_ss_remove) rfs4_dss_remove_cpleaf(cp); rfs4_ss_pnfree(cp->rc_ss_pn); } /* Free the client supplied client id */ kmem_free(cp->rc_nfs_client.id_val, cp->rc_nfs_client.id_len); if (cp->rc_sysidt != LM_NOSYSID) lm_free_sysidt(cp->rc_sysidt); } static bool_t rfs4_client_create(rfs4_entry_t u_entry, void *arg) { rfs4_client_t *cp = (rfs4_client_t *)u_entry; nfs_client_id4 *client = (nfs_client_id4 *)arg; cid *cidp; scid_confirm_verf *scvp; /* Get a clientid to give to the client */ cidp = (cid *)&cp->rc_clientid; cidp->impl_id.start_time = rfs4_start_time; cidp->impl_id.c_id = (uint32_t)rfs4_dbe_getid(cp->rc_dbe); /* If we are booted as a cluster node, embed our nodeid */ if (cluster_bootflags & CLUSTER_BOOTED) embed_nodeid(cidp); /* Allocate and copy client's client id value */ cp->rc_nfs_client.id_val = kmem_alloc(client->id_len, KM_SLEEP); cp->rc_nfs_client.id_len = client->id_len; bcopy(client->id_val, cp->rc_nfs_client.id_val, client->id_len); cp->rc_nfs_client.verifier = client->verifier; /* Init the value for the SETCLIENTID_CONFIRM verifier */ scvp = (scid_confirm_verf *)&cp->rc_confirm_verf; scvp->cv_impl.c_id = cidp->impl_id.c_id; scvp->cv_impl.gen_num = 0; /* An F_UNLKSYS has been done for this client */ cp->rc_unlksys_completed = FALSE; /* We need the client to ack us */ cp->rc_need_confirm = TRUE; cp->rc_cp_confirmed = NULL; /* TRUE all the time until the callback path actually fails */ cp->rc_cbinfo.cb_notified_of_cb_path_down = TRUE; /* Initialize the access time to now */ cp->rc_last_access = gethrestime_sec(); cp->rc_cr_set = NULL; cp->rc_sysidt = LM_NOSYSID; list_create(&cp->rc_openownerlist, sizeof (rfs4_openowner_t), offsetof(rfs4_openowner_t, ro_node)); /* set up the callback control structure */ cp->rc_cbinfo.cb_state = CB_UNINIT; mutex_init(cp->rc_cbinfo.cb_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(cp->rc_cbinfo.cb_cv, NULL, CV_DEFAULT, NULL); cv_init(cp->rc_cbinfo.cb_cv_nullcaller, NULL, CV_DEFAULT, NULL); /* * Associate the client_t with the current server instance. * The hold is solely to satisfy the calling requirement of * rfs4_servinst_assign(). In this case it's not strictly necessary. */ rfs4_dbe_hold(cp->rc_dbe); rfs4_servinst_assign(cp, rfs4_cur_servinst); rfs4_dbe_rele(cp->rc_dbe); return (TRUE); } /* * Caller wants to generate/update the setclientid_confirm verifier * associated with a client. This is done during the SETCLIENTID * processing. */ void rfs4_client_scv_next(rfs4_client_t *cp) { scid_confirm_verf *scvp; /* Init the value for the SETCLIENTID_CONFIRM verifier */ scvp = (scid_confirm_verf *)&cp->rc_confirm_verf; scvp->cv_impl.gen_num++; } void rfs4_client_rele(rfs4_client_t *cp) { rfs4_dbe_rele(cp->rc_dbe); } rfs4_client_t * rfs4_findclient(nfs_client_id4 *client, bool_t *create, rfs4_client_t *oldcp) { rfs4_client_t *cp; if (oldcp) { rw_enter(&rfs4_findclient_lock, RW_WRITER); rfs4_dbe_hide(oldcp->rc_dbe); } else { rw_enter(&rfs4_findclient_lock, RW_READER); } cp = (rfs4_client_t *)rfs4_dbsearch(rfs4_nfsclnt_idx, client, create, (void *)client, RFS4_DBS_VALID); if (oldcp) rfs4_dbe_unhide(oldcp->rc_dbe); rw_exit(&rfs4_findclient_lock); return (cp); } rfs4_client_t * rfs4_findclient_by_id(clientid4 clientid, bool_t find_unconfirmed) { rfs4_client_t *cp; bool_t create = FALSE; cid *cidp = (cid *)&clientid; /* If we're a cluster and the nodeid isn't right, short-circuit */ if (cluster_bootflags & CLUSTER_BOOTED && foreign_clientid(cidp)) return (NULL); rw_enter(&rfs4_findclient_lock, RW_READER); cp = (rfs4_client_t *)rfs4_dbsearch(rfs4_clientid_idx, &clientid, &create, NULL, RFS4_DBS_VALID); rw_exit(&rfs4_findclient_lock); if (cp && cp->rc_need_confirm && find_unconfirmed == FALSE) { rfs4_client_rele(cp); return (NULL); } else { return (cp); } } bool_t rfs4_lease_expired(rfs4_client_t *cp) { bool_t rc; rfs4_dbe_lock(cp->rc_dbe); /* * If the admin has executed clear_locks for this * client id, force expire will be set, so no need * to calculate anything because it's "outa here". */ if (cp->rc_forced_expire) { rc = TRUE; } else { rc = (gethrestime_sec() - cp->rc_last_access > rfs4_lease_time); } /* * If the lease has expired we will also want * to remove any stable storage state data. So * mark the client id accordingly. */ if (!cp->rc_ss_remove) cp->rc_ss_remove = (rc == TRUE); rfs4_dbe_unlock(cp->rc_dbe); return (rc); } void rfs4_update_lease(rfs4_client_t *cp) { rfs4_dbe_lock(cp->rc_dbe); if (!cp->rc_forced_expire) cp->rc_last_access = gethrestime_sec(); rfs4_dbe_unlock(cp->rc_dbe); } static bool_t EQOPENOWNER(open_owner4 *a, open_owner4 *b) { bool_t rc; if (a->clientid != b->clientid) return (FALSE); if (a->owner_len != b->owner_len) return (FALSE); rc = (bcmp(a->owner_val, b->owner_val, a->owner_len) == 0); return (rc); } static uint_t openowner_hash(void *key) { int i; open_owner4 *openowner = key; uint_t hash = 0; for (i = 0; i < openowner->owner_len; i++) { hash <<= 4; hash += (uint_t)openowner->owner_val[i]; } hash += (uint_t)openowner->clientid; hash |= (openowner->clientid >> 32); return (hash); } static bool_t openowner_compare(rfs4_entry_t u_entry, void *key) { rfs4_openowner_t *oo = (rfs4_openowner_t *)u_entry; open_owner4 *arg = key; return (EQOPENOWNER(&oo->ro_owner, arg)); } void * openowner_mkkey(rfs4_entry_t u_entry) { rfs4_openowner_t *oo = (rfs4_openowner_t *)u_entry; return (&oo->ro_owner); } static bool_t rfs4_openowner_expiry(rfs4_entry_t u_entry) { rfs4_openowner_t *oo = (rfs4_openowner_t *)u_entry; if (rfs4_dbe_is_invalid(oo->ro_dbe)) return (TRUE); return ((gethrestime_sec() - oo->ro_client->rc_last_access > rfs4_lease_time)); } static void rfs4_openowner_destroy(rfs4_entry_t u_entry) { rfs4_openowner_t *oo = (rfs4_openowner_t *)u_entry; /* Remove open owner from client's lists of open owners */ rfs4_dbe_lock(oo->ro_client->rc_dbe); list_remove(&oo->ro_client->rc_openownerlist, oo); rfs4_dbe_unlock(oo->ro_client->rc_dbe); /* One less reference to the client */ rfs4_client_rele(oo->ro_client); oo->ro_client = NULL; /* Free the last reply for this lock owner */ rfs4_free_reply(&oo->ro_reply); if (oo->ro_reply_fh.nfs_fh4_val) { kmem_free(oo->ro_reply_fh.nfs_fh4_val, oo->ro_reply_fh.nfs_fh4_len); oo->ro_reply_fh.nfs_fh4_val = NULL; oo->ro_reply_fh.nfs_fh4_len = 0; } rfs4_sw_destroy(&oo->ro_sw); list_destroy(&oo->ro_statelist); /* Free the lock owner id */ kmem_free(oo->ro_owner.owner_val, oo->ro_owner.owner_len); } void rfs4_openowner_rele(rfs4_openowner_t *oo) { rfs4_dbe_rele(oo->ro_dbe); } static bool_t rfs4_openowner_create(rfs4_entry_t u_entry, void *arg) { rfs4_openowner_t *oo = (rfs4_openowner_t *)u_entry; rfs4_openowner_t *argp = (rfs4_openowner_t *)arg; open_owner4 *openowner = &argp->ro_owner; seqid4 seqid = argp->ro_open_seqid; rfs4_client_t *cp; bool_t create = FALSE; rw_enter(&rfs4_findclient_lock, RW_READER); cp = (rfs4_client_t *)rfs4_dbsearch(rfs4_clientid_idx, &openowner->clientid, &create, NULL, RFS4_DBS_VALID); rw_exit(&rfs4_findclient_lock); if (cp == NULL) return (FALSE); oo->ro_reply_fh.nfs_fh4_len = 0; oo->ro_reply_fh.nfs_fh4_val = NULL; oo->ro_owner.clientid = openowner->clientid; oo->ro_owner.owner_val = kmem_alloc(openowner->owner_len, KM_SLEEP); bcopy(openowner->owner_val, oo->ro_owner.owner_val, openowner->owner_len); oo->ro_owner.owner_len = openowner->owner_len; oo->ro_need_confirm = TRUE; rfs4_sw_init(&oo->ro_sw); oo->ro_open_seqid = seqid; bzero(&oo->ro_reply, sizeof (nfs_resop4)); oo->ro_client = cp; oo->ro_cr_set = NULL; list_create(&oo->ro_statelist, sizeof (rfs4_state_t), offsetof(rfs4_state_t, rs_node)); /* Insert openowner into client's open owner list */ rfs4_dbe_lock(cp->rc_dbe); list_insert_tail(&cp->rc_openownerlist, oo); rfs4_dbe_unlock(cp->rc_dbe); return (TRUE); } rfs4_openowner_t * rfs4_findopenowner(open_owner4 *openowner, bool_t *create, seqid4 seqid) { rfs4_openowner_t *oo; rfs4_openowner_t arg; arg.ro_owner = *openowner; arg.ro_open_seqid = seqid; oo = (rfs4_openowner_t *)rfs4_dbsearch(rfs4_openowner_idx, openowner, create, &arg, RFS4_DBS_VALID); return (oo); } void rfs4_update_open_sequence(rfs4_openowner_t *oo) { rfs4_dbe_lock(oo->ro_dbe); oo->ro_open_seqid++; rfs4_dbe_unlock(oo->ro_dbe); } void rfs4_update_open_resp(rfs4_openowner_t *oo, nfs_resop4 *resp, nfs_fh4 *fh) { rfs4_dbe_lock(oo->ro_dbe); rfs4_free_reply(&oo->ro_reply); rfs4_copy_reply(&oo->ro_reply, resp); /* Save the filehandle if provided and free if not used */ if (resp->nfs_resop4_u.opopen.status == NFS4_OK && fh && fh->nfs_fh4_len) { if (oo->ro_reply_fh.nfs_fh4_val == NULL) oo->ro_reply_fh.nfs_fh4_val = kmem_alloc(fh->nfs_fh4_len, KM_SLEEP); nfs_fh4_copy(fh, &oo->ro_reply_fh); } else { if (oo->ro_reply_fh.nfs_fh4_val) { kmem_free(oo->ro_reply_fh.nfs_fh4_val, oo->ro_reply_fh.nfs_fh4_len); oo->ro_reply_fh.nfs_fh4_val = NULL; oo->ro_reply_fh.nfs_fh4_len = 0; } } rfs4_dbe_unlock(oo->ro_dbe); } static bool_t lockowner_compare(rfs4_entry_t u_entry, void *key) { rfs4_lockowner_t *lo = (rfs4_lockowner_t *)u_entry; lock_owner4 *b = (lock_owner4 *)key; if (lo->rl_owner.clientid != b->clientid) return (FALSE); if (lo->rl_owner.owner_len != b->owner_len) return (FALSE); return (bcmp(lo->rl_owner.owner_val, b->owner_val, lo->rl_owner.owner_len) == 0); } void * lockowner_mkkey(rfs4_entry_t u_entry) { rfs4_lockowner_t *lo = (rfs4_lockowner_t *)u_entry; return (&lo->rl_owner); } static uint32_t lockowner_hash(void *key) { int i; lock_owner4 *lockowner = key; uint_t hash = 0; for (i = 0; i < lockowner->owner_len; i++) { hash <<= 4; hash += (uint_t)lockowner->owner_val[i]; } hash += (uint_t)lockowner->clientid; hash |= (lockowner->clientid >> 32); return (hash); } static uint32_t pid_hash(void *key) { return ((uint32_t)(uintptr_t)key); } static void * pid_mkkey(rfs4_entry_t u_entry) { rfs4_lockowner_t *lo = (rfs4_lockowner_t *)u_entry; return ((void *)(uintptr_t)lo->rl_pid); } static bool_t pid_compare(rfs4_entry_t u_entry, void *key) { rfs4_lockowner_t *lo = (rfs4_lockowner_t *)u_entry; return (lo->rl_pid == (pid_t)(uintptr_t)key); } static void rfs4_lockowner_destroy(rfs4_entry_t u_entry) { rfs4_lockowner_t *lo = (rfs4_lockowner_t *)u_entry; /* Free the lock owner id */ kmem_free(lo->rl_owner.owner_val, lo->rl_owner.owner_len); rfs4_client_rele(lo->rl_client); } void rfs4_lockowner_rele(rfs4_lockowner_t *lo) { rfs4_dbe_rele(lo->rl_dbe); } /* ARGSUSED */ static bool_t rfs4_lockowner_expiry(rfs4_entry_t u_entry) { /* * Since expiry is called with no other references on * this struct, go ahead and have it removed. */ return (TRUE); } static bool_t rfs4_lockowner_create(rfs4_entry_t u_entry, void *arg) { rfs4_lockowner_t *lo = (rfs4_lockowner_t *)u_entry; lock_owner4 *lockowner = (lock_owner4 *)arg; rfs4_client_t *cp; bool_t create = FALSE; rw_enter(&rfs4_findclient_lock, RW_READER); cp = (rfs4_client_t *)rfs4_dbsearch(rfs4_clientid_idx, &lockowner->clientid, &create, NULL, RFS4_DBS_VALID); rw_exit(&rfs4_findclient_lock); if (cp == NULL) return (FALSE); /* Reference client */ lo->rl_client = cp; lo->rl_owner.clientid = lockowner->clientid; lo->rl_owner.owner_val = kmem_alloc(lockowner->owner_len, KM_SLEEP); bcopy(lockowner->owner_val, lo->rl_owner.owner_val, lockowner->owner_len); lo->rl_owner.owner_len = lockowner->owner_len; lo->rl_pid = rfs4_dbe_getid(lo->rl_dbe); return (TRUE); } rfs4_lockowner_t * rfs4_findlockowner(lock_owner4 *lockowner, bool_t *create) { rfs4_lockowner_t *lo; lo = (rfs4_lockowner_t *)rfs4_dbsearch(rfs4_lockowner_idx, lockowner, create, lockowner, RFS4_DBS_VALID); return (lo); } rfs4_lockowner_t * rfs4_findlockowner_by_pid(pid_t pid) { rfs4_lockowner_t *lo; bool_t create = FALSE; lo = (rfs4_lockowner_t *)rfs4_dbsearch(rfs4_lockowner_pid_idx, (void *)(uintptr_t)pid, &create, NULL, RFS4_DBS_VALID); return (lo); } static uint32_t file_hash(void *key) { return (ADDRHASH(key)); } static void * file_mkkey(rfs4_entry_t u_entry) { rfs4_file_t *fp = (rfs4_file_t *)u_entry; return (fp->rf_vp); } static bool_t file_compare(rfs4_entry_t u_entry, void *key) { rfs4_file_t *fp = (rfs4_file_t *)u_entry; return (fp->rf_vp == (vnode_t *)key); } static void rfs4_file_destroy(rfs4_entry_t u_entry) { rfs4_file_t *fp = (rfs4_file_t *)u_entry; list_destroy(&fp->rf_delegstatelist); if (fp->rf_filehandle.nfs_fh4_val) kmem_free(fp->rf_filehandle.nfs_fh4_val, fp->rf_filehandle.nfs_fh4_len); cv_destroy(fp->rf_dinfo.rd_recall_cv); if (fp->rf_vp) { vnode_t *vp = fp->rf_vp; mutex_enter(&vp->v_vsd_lock); (void) vsd_set(vp, nfs4_srv_vkey, NULL); mutex_exit(&vp->v_vsd_lock); VN_RELE(vp); fp->rf_vp = NULL; } rw_destroy(&fp->rf_file_rwlock); } /* * Used to unlock the underlying dbe struct only */ void rfs4_file_rele(rfs4_file_t *fp) { rfs4_dbe_rele(fp->rf_dbe); } typedef struct { vnode_t *vp; nfs_fh4 *fh; } rfs4_fcreate_arg; static bool_t rfs4_file_create(rfs4_entry_t u_entry, void *arg) { rfs4_file_t *fp = (rfs4_file_t *)u_entry; rfs4_fcreate_arg *ap = (rfs4_fcreate_arg *)arg; vnode_t *vp = ap->vp; nfs_fh4 *fh = ap->fh; VN_HOLD(vp); fp->rf_filehandle.nfs_fh4_len = 0; fp->rf_filehandle.nfs_fh4_val = NULL; ASSERT(fh && fh->nfs_fh4_len); if (fh && fh->nfs_fh4_len) { fp->rf_filehandle.nfs_fh4_val = kmem_alloc(fh->nfs_fh4_len, KM_SLEEP); nfs_fh4_copy(fh, &fp->rf_filehandle); } fp->rf_vp = vp; list_create(&fp->rf_delegstatelist, sizeof (rfs4_deleg_state_t), offsetof(rfs4_deleg_state_t, rds_node)); fp->rf_share_deny = fp->rf_share_access = fp->rf_access_read = 0; fp->rf_access_write = fp->rf_deny_read = fp->rf_deny_write = 0; mutex_init(fp->rf_dinfo.rd_recall_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(fp->rf_dinfo.rd_recall_cv, NULL, CV_DEFAULT, NULL); fp->rf_dinfo.rd_dtype = OPEN_DELEGATE_NONE; rw_init(&fp->rf_file_rwlock, NULL, RW_DEFAULT, NULL); mutex_enter(&vp->v_vsd_lock); VERIFY(vsd_set(vp, nfs4_srv_vkey, (void *)fp) == 0); mutex_exit(&vp->v_vsd_lock); return (TRUE); } rfs4_file_t * rfs4_findfile(vnode_t *vp, nfs_fh4 *fh, bool_t *create) { rfs4_file_t *fp; rfs4_fcreate_arg arg; arg.vp = vp; arg.fh = fh; if (*create == TRUE) fp = (rfs4_file_t *)rfs4_dbsearch(rfs4_file_idx, vp, create, &arg, RFS4_DBS_VALID); else { mutex_enter(&vp->v_vsd_lock); fp = (rfs4_file_t *)vsd_get(vp, nfs4_srv_vkey); if (fp) { rfs4_dbe_lock(fp->rf_dbe); if (rfs4_dbe_is_invalid(fp->rf_dbe) || (rfs4_dbe_refcnt(fp->rf_dbe) == 0)) { rfs4_dbe_unlock(fp->rf_dbe); fp = NULL; } else { rfs4_dbe_hold(fp->rf_dbe); rfs4_dbe_unlock(fp->rf_dbe); } } mutex_exit(&vp->v_vsd_lock); } return (fp); } /* * Find a file in the db and once it is located, take the rw lock. * Need to check the vnode pointer and if it does not exist (it was * removed between the db location and check) redo the find. This * assumes that a file struct that has a NULL vnode pointer is marked * at 'invalid' and will not be found in the db the second time * around. */ rfs4_file_t * rfs4_findfile_withlock(vnode_t *vp, nfs_fh4 *fh, bool_t *create) { rfs4_file_t *fp; rfs4_fcreate_arg arg; bool_t screate = *create; if (screate == FALSE) { mutex_enter(&vp->v_vsd_lock); fp = (rfs4_file_t *)vsd_get(vp, nfs4_srv_vkey); if (fp) { rfs4_dbe_lock(fp->rf_dbe); if (rfs4_dbe_is_invalid(fp->rf_dbe) || (rfs4_dbe_refcnt(fp->rf_dbe) == 0)) { rfs4_dbe_unlock(fp->rf_dbe); mutex_exit(&vp->v_vsd_lock); fp = NULL; } else { rfs4_dbe_hold(fp->rf_dbe); rfs4_dbe_unlock(fp->rf_dbe); mutex_exit(&vp->v_vsd_lock); rw_enter(&fp->rf_file_rwlock, RW_WRITER); if (fp->rf_vp == NULL) { rw_exit(&fp->rf_file_rwlock); rfs4_file_rele(fp); fp = NULL; } } } else { mutex_exit(&vp->v_vsd_lock); } } else { retry: arg.vp = vp; arg.fh = fh; fp = (rfs4_file_t *)rfs4_dbsearch(rfs4_file_idx, vp, create, &arg, RFS4_DBS_VALID); if (fp != NULL) { rw_enter(&fp->rf_file_rwlock, RW_WRITER); if (fp->rf_vp == NULL) { rw_exit(&fp->rf_file_rwlock); rfs4_file_rele(fp); *create = screate; goto retry; } } } return (fp); } static uint32_t lo_state_hash(void *key) { stateid_t *id = key; return (id->bits.ident+id->bits.pid); } static bool_t lo_state_compare(rfs4_entry_t u_entry, void *key) { rfs4_lo_state_t *lsp = (rfs4_lo_state_t *)u_entry; stateid_t *id = key; bool_t rc; rc = (lsp->rls_lockid.bits.boottime == id->bits.boottime && lsp->rls_lockid.bits.type == id->bits.type && lsp->rls_lockid.bits.ident == id->bits.ident && lsp->rls_lockid.bits.pid == id->bits.pid); return (rc); } static void * lo_state_mkkey(rfs4_entry_t u_entry) { rfs4_lo_state_t *lsp = (rfs4_lo_state_t *)u_entry; return (&lsp->rls_lockid); } static bool_t rfs4_lo_state_expiry(rfs4_entry_t u_entry) { rfs4_lo_state_t *lsp = (rfs4_lo_state_t *)u_entry; if (rfs4_dbe_is_invalid(lsp->rls_dbe)) return (TRUE); if (lsp->rls_state->rs_closed) return (TRUE); return ((gethrestime_sec() - lsp->rls_state->rs_owner->ro_client->rc_last_access > rfs4_lease_time)); } static void rfs4_lo_state_destroy(rfs4_entry_t u_entry) { rfs4_lo_state_t *lsp = (rfs4_lo_state_t *)u_entry; rfs4_dbe_lock(lsp->rls_state->rs_dbe); list_remove(&lsp->rls_state->rs_lostatelist, lsp); rfs4_dbe_unlock(lsp->rls_state->rs_dbe); rfs4_sw_destroy(&lsp->rls_sw); /* Make sure to release the file locks */ if (lsp->rls_locks_cleaned == FALSE) { lsp->rls_locks_cleaned = TRUE; if (lsp->rls_locker->rl_client->rc_sysidt != LM_NOSYSID) { /* Is the PxFS kernel module loaded? */ if (lm_remove_file_locks != NULL) { int new_sysid; /* Encode the cluster nodeid in new sysid */ new_sysid = lsp->rls_locker->rl_client->rc_sysidt; lm_set_nlmid_flk(&new_sysid); /* * This PxFS routine removes file locks for a * client over all nodes of a cluster. */ DTRACE_PROBE1(nfss_i_clust_rm_lck, int, new_sysid); (*lm_remove_file_locks)(new_sysid); } else { (void) cleanlocks( lsp->rls_state->rs_finfo->rf_vp, lsp->rls_locker->rl_pid, lsp->rls_locker->rl_client->rc_sysidt); } } } /* Free the last reply for this state */ rfs4_free_reply(&lsp->rls_reply); rfs4_lockowner_rele(lsp->rls_locker); lsp->rls_locker = NULL; rfs4_state_rele_nounlock(lsp->rls_state); lsp->rls_state = NULL; } static bool_t rfs4_lo_state_create(rfs4_entry_t u_entry, void *arg) { rfs4_lo_state_t *lsp = (rfs4_lo_state_t *)u_entry; rfs4_lo_state_t *argp = (rfs4_lo_state_t *)arg; rfs4_lockowner_t *lo = argp->rls_locker; rfs4_state_t *sp = argp->rls_state; lsp->rls_state = sp; lsp->rls_lockid = sp->rs_stateid; lsp->rls_lockid.bits.type = LOCKID; lsp->rls_lockid.bits.chgseq = 0; lsp->rls_lockid.bits.pid = lo->rl_pid; lsp->rls_locks_cleaned = FALSE; lsp->rls_lock_completed = FALSE; rfs4_sw_init(&lsp->rls_sw); /* Attached the supplied lock owner */ rfs4_dbe_hold(lo->rl_dbe); lsp->rls_locker = lo; rfs4_dbe_lock(sp->rs_dbe); list_insert_tail(&sp->rs_lostatelist, lsp); rfs4_dbe_hold(sp->rs_dbe); rfs4_dbe_unlock(sp->rs_dbe); return (TRUE); } void rfs4_lo_state_rele(rfs4_lo_state_t *lsp, bool_t unlock_fp) { if (unlock_fp == TRUE) rw_exit(&lsp->rls_state->rs_finfo->rf_file_rwlock); rfs4_dbe_rele(lsp->rls_dbe); } static rfs4_lo_state_t * rfs4_findlo_state(stateid_t *id, bool_t lock_fp) { rfs4_lo_state_t *lsp; bool_t create = FALSE; lsp = (rfs4_lo_state_t *)rfs4_dbsearch(rfs4_lo_state_idx, id, &create, NULL, RFS4_DBS_VALID); if (lock_fp == TRUE && lsp != NULL) rw_enter(&lsp->rls_state->rs_finfo->rf_file_rwlock, RW_READER); return (lsp); } static uint32_t lo_state_lo_hash(void *key) { rfs4_lo_state_t *lsp = key; return (ADDRHASH(lsp->rls_locker) ^ ADDRHASH(lsp->rls_state)); } static bool_t lo_state_lo_compare(rfs4_entry_t u_entry, void *key) { rfs4_lo_state_t *lsp = (rfs4_lo_state_t *)u_entry; rfs4_lo_state_t *keyp = key; return (keyp->rls_locker == lsp->rls_locker && keyp->rls_state == lsp->rls_state); } static void * lo_state_lo_mkkey(rfs4_entry_t u_entry) { return (u_entry); } rfs4_lo_state_t * rfs4_findlo_state_by_owner(rfs4_lockowner_t *lo, rfs4_state_t *sp, bool_t *create) { rfs4_lo_state_t *lsp; rfs4_lo_state_t arg; arg.rls_locker = lo; arg.rls_state = sp; lsp = (rfs4_lo_state_t *)rfs4_dbsearch(rfs4_lo_state_owner_idx, &arg, create, &arg, RFS4_DBS_VALID); return (lsp); } static stateid_t get_stateid(id_t eid) { stateid_t id; id.bits.boottime = rfs4_start_time; id.bits.ident = eid; id.bits.chgseq = 0; id.bits.type = 0; id.bits.pid = 0; /* * If we are booted as a cluster node, embed our nodeid. * We've already done sanity checks in rfs4_client_create() so no * need to repeat them here. */ id.bits.clnodeid = (cluster_bootflags & CLUSTER_BOOTED) ? clconf_get_nodeid() : 0; return (id); } /* * For use only when booted as a cluster node. * Returns TRUE if the embedded nodeid indicates that this stateid was * generated on another node. */ static int foreign_stateid(stateid_t *id) { ASSERT(cluster_bootflags & CLUSTER_BOOTED); return (id->bits.clnodeid != (uint32_t)clconf_get_nodeid()); } /* * For use only when booted as a cluster node. * Returns TRUE if the embedded nodeid indicates that this clientid was * generated on another node. */ static int foreign_clientid(cid *cidp) { ASSERT(cluster_bootflags & CLUSTER_BOOTED); return (cidp->impl_id.c_id >> CLUSTER_NODEID_SHIFT != (uint32_t)clconf_get_nodeid()); } /* * For use only when booted as a cluster node. * Embed our cluster nodeid into the clientid. */ static void embed_nodeid(cid *cidp) { int clnodeid; /* * Currently, our state tables are small enough that their * ids will leave enough bits free for the nodeid. If the * tables become larger, we mustn't overwrite the id. * Equally, we only have room for so many bits of nodeid, so * must check that too. */ ASSERT(cluster_bootflags & CLUSTER_BOOTED); ASSERT(cidp->impl_id.c_id >> CLUSTER_NODEID_SHIFT == 0); clnodeid = clconf_get_nodeid(); ASSERT(clnodeid <= CLUSTER_MAX_NODEID); ASSERT(clnodeid != NODEID_UNKNOWN); cidp->impl_id.c_id |= (clnodeid << CLUSTER_NODEID_SHIFT); } static uint32_t state_hash(void *key) { stateid_t *ip = (stateid_t *)key; return (ip->bits.ident); } static bool_t state_compare(rfs4_entry_t u_entry, void *key) { rfs4_state_t *sp = (rfs4_state_t *)u_entry; stateid_t *id = (stateid_t *)key; bool_t rc; rc = (sp->rs_stateid.bits.boottime == id->bits.boottime && sp->rs_stateid.bits.ident == id->bits.ident); return (rc); } static void * state_mkkey(rfs4_entry_t u_entry) { rfs4_state_t *sp = (rfs4_state_t *)u_entry; return (&sp->rs_stateid); } static void rfs4_state_destroy(rfs4_entry_t u_entry) { rfs4_state_t *sp = (rfs4_state_t *)u_entry; /* remove from openowner list */ rfs4_dbe_lock(sp->rs_owner->ro_dbe); list_remove(&sp->rs_owner->ro_statelist, sp); rfs4_dbe_unlock(sp->rs_owner->ro_dbe); list_destroy(&sp->rs_lostatelist); /* release any share locks for this stateid if it's still open */ if (!sp->rs_closed) { rfs4_dbe_lock(sp->rs_dbe); (void) rfs4_unshare(sp); rfs4_dbe_unlock(sp->rs_dbe); } /* Were done with the file */ rfs4_file_rele(sp->rs_finfo); sp->rs_finfo = NULL; /* And now with the openowner */ rfs4_openowner_rele(sp->rs_owner); sp->rs_owner = NULL; } static void rfs4_state_rele_nounlock(rfs4_state_t *sp) { rfs4_dbe_rele(sp->rs_dbe); } void rfs4_state_rele(rfs4_state_t *sp) { rw_exit(&sp->rs_finfo->rf_file_rwlock); rfs4_dbe_rele(sp->rs_dbe); } static uint32_t deleg_hash(void *key) { rfs4_deleg_state_t *dsp = (rfs4_deleg_state_t *)key; return (ADDRHASH(dsp->rds_client) ^ ADDRHASH(dsp->rds_finfo)); } static bool_t deleg_compare(rfs4_entry_t u_entry, void *key) { rfs4_deleg_state_t *dsp = (rfs4_deleg_state_t *)u_entry; rfs4_deleg_state_t *kdsp = (rfs4_deleg_state_t *)key; return (dsp->rds_client == kdsp->rds_client && dsp->rds_finfo == kdsp->rds_finfo); } static void * deleg_mkkey(rfs4_entry_t u_entry) { return (u_entry); } static uint32_t deleg_state_hash(void *key) { stateid_t *ip = (stateid_t *)key; return (ip->bits.ident); } static bool_t deleg_state_compare(rfs4_entry_t u_entry, void *key) { rfs4_deleg_state_t *dsp = (rfs4_deleg_state_t *)u_entry; stateid_t *id = (stateid_t *)key; bool_t rc; if (id->bits.type != DELEGID) return (FALSE); rc = (dsp->rds_delegid.bits.boottime == id->bits.boottime && dsp->rds_delegid.bits.ident == id->bits.ident); return (rc); } static void * deleg_state_mkkey(rfs4_entry_t u_entry) { rfs4_deleg_state_t *dsp = (rfs4_deleg_state_t *)u_entry; return (&dsp->rds_delegid); } static bool_t rfs4_deleg_state_expiry(rfs4_entry_t u_entry) { rfs4_deleg_state_t *dsp = (rfs4_deleg_state_t *)u_entry; if (rfs4_dbe_is_invalid(dsp->rds_dbe)) return (TRUE); if ((gethrestime_sec() - dsp->rds_client->rc_last_access > rfs4_lease_time)) { rfs4_dbe_invalidate(dsp->rds_dbe); return (TRUE); } return (FALSE); } static bool_t rfs4_deleg_state_create(rfs4_entry_t u_entry, void *argp) { rfs4_deleg_state_t *dsp = (rfs4_deleg_state_t *)u_entry; rfs4_file_t *fp = ((rfs4_deleg_state_t *)argp)->rds_finfo; rfs4_client_t *cp = ((rfs4_deleg_state_t *)argp)->rds_client; rfs4_dbe_hold(fp->rf_dbe); rfs4_dbe_hold(cp->rc_dbe); dsp->rds_delegid = get_stateid(rfs4_dbe_getid(dsp->rds_dbe)); dsp->rds_delegid.bits.type = DELEGID; dsp->rds_finfo = fp; dsp->rds_client = cp; dsp->rds_dtype = OPEN_DELEGATE_NONE; dsp->rds_time_granted = gethrestime_sec(); /* observability */ dsp->rds_time_revoked = 0; list_link_init(&dsp->rds_node); return (TRUE); } static void rfs4_deleg_state_destroy(rfs4_entry_t u_entry) { rfs4_deleg_state_t *dsp = (rfs4_deleg_state_t *)u_entry; /* return delegation if necessary */ rfs4_return_deleg(dsp, FALSE); /* Were done with the file */ rfs4_file_rele(dsp->rds_finfo); dsp->rds_finfo = NULL; /* And now with the openowner */ rfs4_client_rele(dsp->rds_client); dsp->rds_client = NULL; } rfs4_deleg_state_t * rfs4_finddeleg(rfs4_state_t *sp, bool_t *create) { rfs4_deleg_state_t ds, *dsp; ds.rds_client = sp->rs_owner->ro_client; ds.rds_finfo = sp->rs_finfo; dsp = (rfs4_deleg_state_t *)rfs4_dbsearch(rfs4_deleg_idx, &ds, create, &ds, RFS4_DBS_VALID); return (dsp); } rfs4_deleg_state_t * rfs4_finddelegstate(stateid_t *id) { rfs4_deleg_state_t *dsp; bool_t create = FALSE; dsp = (rfs4_deleg_state_t *)rfs4_dbsearch(rfs4_deleg_state_idx, id, &create, NULL, RFS4_DBS_VALID); return (dsp); } void rfs4_deleg_state_rele(rfs4_deleg_state_t *dsp) { rfs4_dbe_rele(dsp->rds_dbe); } void rfs4_update_lock_sequence(rfs4_lo_state_t *lsp) { rfs4_dbe_lock(lsp->rls_dbe); /* * If we are skipping sequence id checking, this means that * this is the first lock request and therefore the sequence * id does not need to be updated. This only happens on the * first lock request for a lockowner */ if (!lsp->rls_skip_seqid_check) lsp->rls_seqid++; rfs4_dbe_unlock(lsp->rls_dbe); } void rfs4_update_lock_resp(rfs4_lo_state_t *lsp, nfs_resop4 *resp) { rfs4_dbe_lock(lsp->rls_dbe); rfs4_free_reply(&lsp->rls_reply); rfs4_copy_reply(&lsp->rls_reply, resp); rfs4_dbe_unlock(lsp->rls_dbe); } void rfs4_free_opens(rfs4_openowner_t *oo, bool_t invalidate, bool_t close_of_client) { rfs4_state_t *sp; rfs4_dbe_lock(oo->ro_dbe); for (sp = list_head(&oo->ro_statelist); sp != NULL; sp = list_next(&oo->ro_statelist, sp)) { rfs4_state_close(sp, FALSE, close_of_client, CRED()); if (invalidate == TRUE) rfs4_dbe_invalidate(sp->rs_dbe); } rfs4_dbe_invalidate(oo->ro_dbe); rfs4_dbe_unlock(oo->ro_dbe); } static uint32_t state_owner_file_hash(void *key) { rfs4_state_t *sp = key; return (ADDRHASH(sp->rs_owner) ^ ADDRHASH(sp->rs_finfo)); } static bool_t state_owner_file_compare(rfs4_entry_t u_entry, void *key) { rfs4_state_t *sp = (rfs4_state_t *)u_entry; rfs4_state_t *arg = key; if (sp->rs_closed == TRUE) return (FALSE); return (arg->rs_owner == sp->rs_owner && arg->rs_finfo == sp->rs_finfo); } static void * state_owner_file_mkkey(rfs4_entry_t u_entry) { return (u_entry); } static uint32_t state_file_hash(void *key) { return (ADDRHASH(key)); } static bool_t state_file_compare(rfs4_entry_t u_entry, void *key) { rfs4_state_t *sp = (rfs4_state_t *)u_entry; rfs4_file_t *fp = key; if (sp->rs_closed == TRUE) return (FALSE); return (fp == sp->rs_finfo); } static void * state_file_mkkey(rfs4_entry_t u_entry) { rfs4_state_t *sp = (rfs4_state_t *)u_entry; return (sp->rs_finfo); } rfs4_state_t * rfs4_findstate_by_owner_file(rfs4_openowner_t *oo, rfs4_file_t *fp, bool_t *create) { rfs4_state_t *sp; rfs4_state_t key; key.rs_owner = oo; key.rs_finfo = fp; sp = (rfs4_state_t *)rfs4_dbsearch(rfs4_state_owner_file_idx, &key, create, &key, RFS4_DBS_VALID); return (sp); } /* This returns ANY state struct that refers to this file */ static rfs4_state_t * rfs4_findstate_by_file(rfs4_file_t *fp) { bool_t create = FALSE; return ((rfs4_state_t *)rfs4_dbsearch(rfs4_state_file_idx, fp, &create, fp, RFS4_DBS_VALID)); } static bool_t rfs4_state_expiry(rfs4_entry_t u_entry) { rfs4_state_t *sp = (rfs4_state_t *)u_entry; if (rfs4_dbe_is_invalid(sp->rs_dbe)) return (TRUE); if (sp->rs_closed == TRUE && ((gethrestime_sec() - rfs4_dbe_get_timerele(sp->rs_dbe)) > rfs4_lease_time)) return (TRUE); return ((gethrestime_sec() - sp->rs_owner->ro_client->rc_last_access > rfs4_lease_time)); } static bool_t rfs4_state_create(rfs4_entry_t u_entry, void *argp) { rfs4_state_t *sp = (rfs4_state_t *)u_entry; rfs4_file_t *fp = ((rfs4_state_t *)argp)->rs_finfo; rfs4_openowner_t *oo = ((rfs4_state_t *)argp)->rs_owner; rfs4_dbe_hold(fp->rf_dbe); rfs4_dbe_hold(oo->ro_dbe); sp->rs_stateid = get_stateid(rfs4_dbe_getid(sp->rs_dbe)); sp->rs_stateid.bits.type = OPENID; sp->rs_owner = oo; sp->rs_finfo = fp; list_create(&sp->rs_lostatelist, sizeof (rfs4_lo_state_t), offsetof(rfs4_lo_state_t, rls_node)); /* Insert state on per open owner's list */ rfs4_dbe_lock(oo->ro_dbe); list_insert_tail(&oo->ro_statelist, sp); rfs4_dbe_unlock(oo->ro_dbe); return (TRUE); } static rfs4_state_t * rfs4_findstate(stateid_t *id, rfs4_dbsearch_type_t find_invalid, bool_t lock_fp) { rfs4_state_t *sp; bool_t create = FALSE; sp = (rfs4_state_t *)rfs4_dbsearch(rfs4_state_idx, id, &create, NULL, find_invalid); if (lock_fp == TRUE && sp != NULL) rw_enter(&sp->rs_finfo->rf_file_rwlock, RW_READER); return (sp); } void rfs4_state_close(rfs4_state_t *sp, bool_t lock_held, bool_t close_of_client, cred_t *cr) { /* Remove the associated lo_state owners */ if (!lock_held) rfs4_dbe_lock(sp->rs_dbe); /* * If refcnt == 0, the dbe is about to be destroyed. * lock state will be released by the reaper thread. */ if (rfs4_dbe_refcnt(sp->rs_dbe) > 0) { if (sp->rs_closed == FALSE) { rfs4_release_share_lock_state(sp, cr, close_of_client); sp->rs_closed = TRUE; } } if (!lock_held) rfs4_dbe_unlock(sp->rs_dbe); } /* * Remove all state associated with the given client. */ void rfs4_client_state_remove(rfs4_client_t *cp) { rfs4_openowner_t *oo; rfs4_dbe_lock(cp->rc_dbe); for (oo = list_head(&cp->rc_openownerlist); oo != NULL; oo = list_next(&cp->rc_openownerlist, oo)) { rfs4_free_opens(oo, TRUE, TRUE); } rfs4_dbe_unlock(cp->rc_dbe); } void rfs4_client_close(rfs4_client_t *cp) { /* Mark client as going away. */ rfs4_dbe_lock(cp->rc_dbe); rfs4_dbe_invalidate(cp->rc_dbe); rfs4_dbe_unlock(cp->rc_dbe); rfs4_client_state_remove(cp); /* Release the client */ rfs4_client_rele(cp); } nfsstat4 rfs4_check_clientid(clientid4 *cp, int setclid_confirm) { cid *cidp = (cid *) cp; /* * If we are booted as a cluster node, check the embedded nodeid. * If it indicates that this clientid was generated on another node, * inform the client accordingly. */ if (cluster_bootflags & CLUSTER_BOOTED && foreign_clientid(cidp)) return (NFS4ERR_STALE_CLIENTID); /* * If the server start time matches the time provided * by the client (via the clientid) and this is NOT a * setclientid_confirm then return EXPIRED. */ if (!setclid_confirm && cidp->impl_id.start_time == rfs4_start_time) return (NFS4ERR_EXPIRED); return (NFS4ERR_STALE_CLIENTID); } /* * This is used when a stateid has not been found amongst the * current server's state. Check the stateid to see if it * was from this server instantiation or not. */ static nfsstat4 what_stateid_error(stateid_t *id, stateid_type_t type) { /* If we are booted as a cluster node, was stateid locally generated? */ if ((cluster_bootflags & CLUSTER_BOOTED) && foreign_stateid(id)) return (NFS4ERR_STALE_STATEID); /* If types don't match then no use checking further */ if (type != id->bits.type) return (NFS4ERR_BAD_STATEID); /* From a previous server instantiation, return STALE */ if (id->bits.boottime < rfs4_start_time) return (NFS4ERR_STALE_STATEID); /* * From this server but the state is most likely beyond lease * timeout: return NFS4ERR_EXPIRED. However, there is the * case of a delegation stateid. For delegations, there is a * case where the state can be removed without the client's * knowledge/consent: revocation. In the case of delegation * revocation, the delegation state will be removed and will * not be found. If the client does something like a * DELEGRETURN or even a READ/WRITE with a delegatoin stateid * that has been revoked, the server should return BAD_STATEID * instead of the more common EXPIRED error. */ if (id->bits.boottime == rfs4_start_time) { if (type == DELEGID) return (NFS4ERR_BAD_STATEID); else return (NFS4ERR_EXPIRED); } return (NFS4ERR_BAD_STATEID); } /* * Used later on to find the various state structs. When called from * rfs4_check_stateid()->rfs4_get_all_state(), no file struct lock is * taken (it is not needed) and helps on the read/write path with * respect to performance. */ static nfsstat4 rfs4_get_state_lockit(stateid4 *stateid, rfs4_state_t **spp, rfs4_dbsearch_type_t find_invalid, bool_t lock_fp) { stateid_t *id = (stateid_t *)stateid; rfs4_state_t *sp; *spp = NULL; /* If we are booted as a cluster node, was stateid locally generated? */ if ((cluster_bootflags & CLUSTER_BOOTED) && foreign_stateid(id)) return (NFS4ERR_STALE_STATEID); sp = rfs4_findstate(id, find_invalid, lock_fp); if (sp == NULL) { return (what_stateid_error(id, OPENID)); } if (rfs4_lease_expired(sp->rs_owner->ro_client)) { if (lock_fp == TRUE) rfs4_state_rele(sp); else rfs4_state_rele_nounlock(sp); return (NFS4ERR_EXPIRED); } *spp = sp; return (NFS4_OK); } nfsstat4 rfs4_get_state(stateid4 *stateid, rfs4_state_t **spp, rfs4_dbsearch_type_t find_invalid) { return (rfs4_get_state_lockit(stateid, spp, find_invalid, TRUE)); } int rfs4_check_stateid_seqid(rfs4_state_t *sp, stateid4 *stateid) { stateid_t *id = (stateid_t *)stateid; if (rfs4_lease_expired(sp->rs_owner->ro_client)) return (NFS4_CHECK_STATEID_EXPIRED); /* Stateid is some time in the future - that's bad */ if (sp->rs_stateid.bits.chgseq < id->bits.chgseq) return (NFS4_CHECK_STATEID_BAD); if (sp->rs_stateid.bits.chgseq == id->bits.chgseq + 1) return (NFS4_CHECK_STATEID_REPLAY); /* Stateid is some time in the past - that's old */ if (sp->rs_stateid.bits.chgseq > id->bits.chgseq) return (NFS4_CHECK_STATEID_OLD); /* Caller needs to know about confirmation before closure */ if (sp->rs_owner->ro_need_confirm) return (NFS4_CHECK_STATEID_UNCONFIRMED); if (sp->rs_closed == TRUE) return (NFS4_CHECK_STATEID_CLOSED); return (NFS4_CHECK_STATEID_OKAY); } int rfs4_check_lo_stateid_seqid(rfs4_lo_state_t *lsp, stateid4 *stateid) { stateid_t *id = (stateid_t *)stateid; if (rfs4_lease_expired(lsp->rls_state->rs_owner->ro_client)) return (NFS4_CHECK_STATEID_EXPIRED); /* Stateid is some time in the future - that's bad */ if (lsp->rls_lockid.bits.chgseq < id->bits.chgseq) return (NFS4_CHECK_STATEID_BAD); if (lsp->rls_lockid.bits.chgseq == id->bits.chgseq + 1) return (NFS4_CHECK_STATEID_REPLAY); /* Stateid is some time in the past - that's old */ if (lsp->rls_lockid.bits.chgseq > id->bits.chgseq) return (NFS4_CHECK_STATEID_OLD); if (lsp->rls_state->rs_closed == TRUE) return (NFS4_CHECK_STATEID_CLOSED); return (NFS4_CHECK_STATEID_OKAY); } nfsstat4 rfs4_get_deleg_state(stateid4 *stateid, rfs4_deleg_state_t **dspp) { stateid_t *id = (stateid_t *)stateid; rfs4_deleg_state_t *dsp; *dspp = NULL; /* If we are booted as a cluster node, was stateid locally generated? */ if ((cluster_bootflags & CLUSTER_BOOTED) && foreign_stateid(id)) return (NFS4ERR_STALE_STATEID); dsp = rfs4_finddelegstate(id); if (dsp == NULL) { return (what_stateid_error(id, DELEGID)); } if (rfs4_lease_expired(dsp->rds_client)) { rfs4_deleg_state_rele(dsp); return (NFS4ERR_EXPIRED); } *dspp = dsp; return (NFS4_OK); } nfsstat4 rfs4_get_lo_state(stateid4 *stateid, rfs4_lo_state_t **lspp, bool_t lock_fp) { stateid_t *id = (stateid_t *)stateid; rfs4_lo_state_t *lsp; *lspp = NULL; /* If we are booted as a cluster node, was stateid locally generated? */ if ((cluster_bootflags & CLUSTER_BOOTED) && foreign_stateid(id)) return (NFS4ERR_STALE_STATEID); lsp = rfs4_findlo_state(id, lock_fp); if (lsp == NULL) { return (what_stateid_error(id, LOCKID)); } if (rfs4_lease_expired(lsp->rls_state->rs_owner->ro_client)) { rfs4_lo_state_rele(lsp, lock_fp); return (NFS4ERR_EXPIRED); } *lspp = lsp; return (NFS4_OK); } static nfsstat4 rfs4_get_all_state(stateid4 *sid, rfs4_state_t **spp, rfs4_deleg_state_t **dspp, rfs4_lo_state_t **lspp) { rfs4_state_t *sp = NULL; rfs4_deleg_state_t *dsp = NULL; rfs4_lo_state_t *lsp = NULL; stateid_t *id; nfsstat4 status; *spp = NULL; *dspp = NULL; *lspp = NULL; id = (stateid_t *)sid; switch (id->bits.type) { case OPENID: status = rfs4_get_state_lockit(sid, &sp, FALSE, FALSE); break; case DELEGID: status = rfs4_get_deleg_state(sid, &dsp); break; case LOCKID: status = rfs4_get_lo_state(sid, &lsp, FALSE); if (status == NFS4_OK) { sp = lsp->rls_state; rfs4_dbe_hold(sp->rs_dbe); } break; default: status = NFS4ERR_BAD_STATEID; } if (status == NFS4_OK) { *spp = sp; *dspp = dsp; *lspp = lsp; } return (status); } /* * Given the I/O mode (FREAD or FWRITE), this checks whether the * rfs4_state_t struct has access to do this operation and if so * return NFS4_OK; otherwise the proper NFSv4 error is returned. */ nfsstat4 rfs4_state_has_access(rfs4_state_t *sp, int mode, vnode_t *vp) { nfsstat4 stat = NFS4_OK; rfs4_file_t *fp; bool_t create = FALSE; rfs4_dbe_lock(sp->rs_dbe); if (mode == FWRITE) { if (!(sp->rs_share_access & OPEN4_SHARE_ACCESS_WRITE)) { stat = NFS4ERR_OPENMODE; } } else if (mode == FREAD) { if (!(sp->rs_share_access & OPEN4_SHARE_ACCESS_READ)) { /* * If we have OPENed the file with DENYing access * to both READ and WRITE then no one else could * have OPENed the file, hence no conflicting READ * deny. This check is merely an optimization. */ if (sp->rs_share_deny == OPEN4_SHARE_DENY_BOTH) goto out; /* Check against file struct's DENY mode */ fp = rfs4_findfile(vp, NULL, &create); if (fp != NULL) { int deny_read = 0; rfs4_dbe_lock(fp->rf_dbe); /* * Check if any other open owner has the file * OPENed with deny READ. */ if (sp->rs_share_deny & OPEN4_SHARE_DENY_READ) deny_read = 1; ASSERT(fp->rf_deny_read - deny_read >= 0); if (fp->rf_deny_read - deny_read > 0) stat = NFS4ERR_OPENMODE; rfs4_dbe_unlock(fp->rf_dbe); rfs4_file_rele(fp); } } } else { /* Illegal I/O mode */ stat = NFS4ERR_INVAL; } out: rfs4_dbe_unlock(sp->rs_dbe); return (stat); } /* * Given the I/O mode (FREAD or FWRITE), the vnode, the stateid and whether * the file is being truncated, return NFS4_OK if allowed or appropriate * V4 error if not. Note NFS4ERR_DELAY will be returned and a recall on * the associated file will be done if the I/O is not consistent with any * delegation in effect on the file. Should be holding VOP_RWLOCK, either * as reader or writer as appropriate. rfs4_op_open will acquire the * VOP_RWLOCK as writer when setting up delegation. If the stateid is bad * this routine will return NFS4ERR_BAD_STATEID. In addition, through the * deleg parameter, we will return whether a write delegation is held by * the client associated with this stateid. * If the server instance associated with the relevant client is in its * grace period, return NFS4ERR_GRACE. */ nfsstat4 rfs4_check_stateid(int mode, vnode_t *vp, stateid4 *stateid, bool_t trunc, bool_t *deleg, bool_t do_access, caller_context_t *ct) { rfs4_file_t *fp; bool_t create = FALSE; rfs4_state_t *sp; rfs4_deleg_state_t *dsp; rfs4_lo_state_t *lsp; stateid_t *id = (stateid_t *)stateid; nfsstat4 stat = NFS4_OK; if (ct != NULL) { ct->cc_sysid = 0; ct->cc_pid = 0; ct->cc_caller_id = nfs4_srv_caller_id; ct->cc_flags = CC_DONTBLOCK; } if (ISSPECIAL(stateid)) { fp = rfs4_findfile(vp, NULL, &create); if (fp == NULL) return (NFS4_OK); if (fp->rf_dinfo.rd_dtype == OPEN_DELEGATE_NONE) { rfs4_file_rele(fp); return (NFS4_OK); } if (mode == FWRITE || fp->rf_dinfo.rd_dtype == OPEN_DELEGATE_WRITE) { rfs4_recall_deleg(fp, trunc, NULL); rfs4_file_rele(fp); return (NFS4ERR_DELAY); } rfs4_file_rele(fp); return (NFS4_OK); } else { stat = rfs4_get_all_state(stateid, &sp, &dsp, &lsp); if (stat != NFS4_OK) return (stat); if (lsp != NULL) { /* Is associated server instance in its grace period? */ if (rfs4_clnt_in_grace(lsp->rls_locker->rl_client)) { rfs4_lo_state_rele(lsp, FALSE); if (sp != NULL) rfs4_state_rele_nounlock(sp); return (NFS4ERR_GRACE); } if (id->bits.type == LOCKID) { /* Seqid in the future? - that's bad */ if (lsp->rls_lockid.bits.chgseq < id->bits.chgseq) { rfs4_lo_state_rele(lsp, FALSE); if (sp != NULL) rfs4_state_rele_nounlock(sp); return (NFS4ERR_BAD_STATEID); } /* Seqid in the past? - that's old */ if (lsp->rls_lockid.bits.chgseq > id->bits.chgseq) { rfs4_lo_state_rele(lsp, FALSE); if (sp != NULL) rfs4_state_rele_nounlock(sp); return (NFS4ERR_OLD_STATEID); } /* Ensure specified filehandle matches */ if (lsp->rls_state->rs_finfo->rf_vp != vp) { rfs4_lo_state_rele(lsp, FALSE); if (sp != NULL) rfs4_state_rele_nounlock(sp); return (NFS4ERR_BAD_STATEID); } } if (ct != NULL) { ct->cc_sysid = lsp->rls_locker->rl_client->rc_sysidt; ct->cc_pid = lsp->rls_locker->rl_pid; } rfs4_lo_state_rele(lsp, FALSE); } /* Stateid provided was an "open" stateid */ if (sp != NULL) { /* Is associated server instance in its grace period? */ if (rfs4_clnt_in_grace(sp->rs_owner->ro_client)) { rfs4_state_rele_nounlock(sp); return (NFS4ERR_GRACE); } if (id->bits.type == OPENID) { /* Seqid in the future? - that's bad */ if (sp->rs_stateid.bits.chgseq < id->bits.chgseq) { rfs4_state_rele_nounlock(sp); return (NFS4ERR_BAD_STATEID); } /* Seqid in the past - that's old */ if (sp->rs_stateid.bits.chgseq > id->bits.chgseq) { rfs4_state_rele_nounlock(sp); return (NFS4ERR_OLD_STATEID); } } /* Ensure specified filehandle matches */ if (sp->rs_finfo->rf_vp != vp) { rfs4_state_rele_nounlock(sp); return (NFS4ERR_BAD_STATEID); } if (sp->rs_owner->ro_need_confirm) { rfs4_state_rele_nounlock(sp); return (NFS4ERR_BAD_STATEID); } if (sp->rs_closed == TRUE) { rfs4_state_rele_nounlock(sp); return (NFS4ERR_OLD_STATEID); } if (do_access) stat = rfs4_state_has_access(sp, mode, vp); else stat = NFS4_OK; /* * Return whether this state has write * delegation if desired */ if (deleg && (sp->rs_finfo->rf_dinfo.rd_dtype == OPEN_DELEGATE_WRITE)) *deleg = TRUE; /* * We got a valid stateid, so we update the * lease on the client. Ideally we would like * to do this after the calling op succeeds, * but for now this will be good * enough. Callers of this routine are * currently insulated from the state stuff. */ rfs4_update_lease(sp->rs_owner->ro_client); /* * If a delegation is present on this file and * this is a WRITE, then update the lastwrite * time to indicate that activity is present. */ if (sp->rs_finfo->rf_dinfo.rd_dtype == OPEN_DELEGATE_WRITE && mode == FWRITE) { sp->rs_finfo->rf_dinfo.rd_time_lastwrite = gethrestime_sec(); } rfs4_state_rele_nounlock(sp); return (stat); } if (dsp != NULL) { /* Is associated server instance in its grace period? */ if (rfs4_clnt_in_grace(dsp->rds_client)) { rfs4_deleg_state_rele(dsp); return (NFS4ERR_GRACE); } if (dsp->rds_delegid.bits.chgseq != id->bits.chgseq) { rfs4_deleg_state_rele(dsp); return (NFS4ERR_BAD_STATEID); } /* Ensure specified filehandle matches */ if (dsp->rds_finfo->rf_vp != vp) { rfs4_deleg_state_rele(dsp); return (NFS4ERR_BAD_STATEID); } /* * Return whether this state has write * delegation if desired */ if (deleg && (dsp->rds_finfo->rf_dinfo.rd_dtype == OPEN_DELEGATE_WRITE)) *deleg = TRUE; rfs4_update_lease(dsp->rds_client); /* * If a delegation is present on this file and * this is a WRITE, then update the lastwrite * time to indicate that activity is present. */ if (dsp->rds_finfo->rf_dinfo.rd_dtype == OPEN_DELEGATE_WRITE && mode == FWRITE) { dsp->rds_finfo->rf_dinfo.rd_time_lastwrite = gethrestime_sec(); } /* * XXX - what happens if this is a WRITE and the * delegation type of for READ. */ rfs4_deleg_state_rele(dsp); return (stat); } /* * If we got this far, something bad happened */ return (NFS4ERR_BAD_STATEID); } } /* * This is a special function in that for the file struct provided the * server wants to remove/close all current state associated with the * file. The prime use of this would be with OP_REMOVE to force the * release of state and particularly of file locks. * * There is an assumption that there is no delegations outstanding on * this file at this point. The caller should have waited for those * to be returned or revoked. */ void rfs4_close_all_state(rfs4_file_t *fp) { rfs4_state_t *sp; rfs4_dbe_lock(fp->rf_dbe); #ifdef DEBUG /* only applies when server is handing out delegations */ if (rfs4_deleg_policy != SRV_NEVER_DELEGATE) ASSERT(fp->rf_dinfo.rd_hold_grant > 0); #endif /* No delegations for this file */ ASSERT(list_is_empty(&fp->rf_delegstatelist)); /* Make sure that it can not be found */ rfs4_dbe_invalidate(fp->rf_dbe); if (fp->rf_vp == NULL) { rfs4_dbe_unlock(fp->rf_dbe); return; } rfs4_dbe_unlock(fp->rf_dbe); /* * Hold as writer to prevent other server threads from * processing requests related to the file while all state is * being removed. */ rw_enter(&fp->rf_file_rwlock, RW_WRITER); /* Remove ALL state from the file */ while (sp = rfs4_findstate_by_file(fp)) { rfs4_state_close(sp, FALSE, FALSE, CRED()); rfs4_state_rele_nounlock(sp); } /* * This is only safe since there are no further references to * the file. */ rfs4_dbe_lock(fp->rf_dbe); if (fp->rf_vp) { vnode_t *vp = fp->rf_vp; mutex_enter(&vp->v_vsd_lock); (void) vsd_set(vp, nfs4_srv_vkey, NULL); mutex_exit(&vp->v_vsd_lock); VN_RELE(vp); fp->rf_vp = NULL; } rfs4_dbe_unlock(fp->rf_dbe); /* Finally let other references to proceed */ rw_exit(&fp->rf_file_rwlock); } /* * This function is used as a target for the rfs4_dbe_walk() call * below. The purpose of this function is to see if the * lockowner_state refers to a file that resides within the exportinfo * export. If so, then remove the lock_owner state (file locks and * share "locks") for this object since the intent is the server is * unexporting the specified directory. Be sure to invalidate the * object after the state has been released */ static void rfs4_lo_state_walk_callout(rfs4_entry_t u_entry, void *e) { rfs4_lo_state_t *lsp = (rfs4_lo_state_t *)u_entry; struct exportinfo *exi = (struct exportinfo *)e; nfs_fh4_fmt_t fhfmt4, *exi_fhp, *finfo_fhp; fhandle_t *efhp; efhp = (fhandle_t *)&exi->exi_fh; exi_fhp = (nfs_fh4_fmt_t *)&fhfmt4; FH_TO_FMT4(efhp, exi_fhp); finfo_fhp = (nfs_fh4_fmt_t *)lsp->rls_state->rs_finfo-> rf_filehandle.nfs_fh4_val; if (EQFSID(&finfo_fhp->fh4_fsid, &exi_fhp->fh4_fsid) && bcmp(&finfo_fhp->fh4_xdata, &exi_fhp->fh4_xdata, exi_fhp->fh4_xlen) == 0) { rfs4_state_close(lsp->rls_state, FALSE, FALSE, CRED()); rfs4_dbe_invalidate(lsp->rls_dbe); rfs4_dbe_invalidate(lsp->rls_state->rs_dbe); } } /* * This function is used as a target for the rfs4_dbe_walk() call * below. The purpose of this function is to see if the state refers * to a file that resides within the exportinfo export. If so, then * remove the open state for this object since the intent is the * server is unexporting the specified directory. The main result for * this type of entry is to invalidate it such it will not be found in * the future. */ static void rfs4_state_walk_callout(rfs4_entry_t u_entry, void *e) { rfs4_state_t *sp = (rfs4_state_t *)u_entry; struct exportinfo *exi = (struct exportinfo *)e; nfs_fh4_fmt_t fhfmt4, *exi_fhp, *finfo_fhp; fhandle_t *efhp; efhp = (fhandle_t *)&exi->exi_fh; exi_fhp = (nfs_fh4_fmt_t *)&fhfmt4; FH_TO_FMT4(efhp, exi_fhp); finfo_fhp = (nfs_fh4_fmt_t *)sp->rs_finfo->rf_filehandle.nfs_fh4_val; if (EQFSID(&finfo_fhp->fh4_fsid, &exi_fhp->fh4_fsid) && bcmp(&finfo_fhp->fh4_xdata, &exi_fhp->fh4_xdata, exi_fhp->fh4_xlen) == 0) { rfs4_state_close(sp, TRUE, FALSE, CRED()); rfs4_dbe_invalidate(sp->rs_dbe); } } /* * This function is used as a target for the rfs4_dbe_walk() call * below. The purpose of this function is to see if the state refers * to a file that resides within the exportinfo export. If so, then * remove the deleg state for this object since the intent is the * server is unexporting the specified directory. The main result for * this type of entry is to invalidate it such it will not be found in * the future. */ static void rfs4_deleg_state_walk_callout(rfs4_entry_t u_entry, void *e) { rfs4_deleg_state_t *dsp = (rfs4_deleg_state_t *)u_entry; struct exportinfo *exi = (struct exportinfo *)e; nfs_fh4_fmt_t fhfmt4, *exi_fhp, *finfo_fhp; fhandle_t *efhp; efhp = (fhandle_t *)&exi->exi_fh; exi_fhp = (nfs_fh4_fmt_t *)&fhfmt4; FH_TO_FMT4(efhp, exi_fhp); finfo_fhp = (nfs_fh4_fmt_t *)dsp->rds_finfo->rf_filehandle.nfs_fh4_val; if (EQFSID(&finfo_fhp->fh4_fsid, &exi_fhp->fh4_fsid) && bcmp(&finfo_fhp->fh4_xdata, &exi_fhp->fh4_xdata, exi_fhp->fh4_xlen) == 0) { rfs4_dbe_invalidate(dsp->rds_dbe); } } /* * This function is used as a target for the rfs4_dbe_walk() call * below. The purpose of this function is to see if the state refers * to a file that resides within the exportinfo export. If so, then * release vnode hold for this object since the intent is the server * is unexporting the specified directory. Invalidation will prevent * this struct from being found in the future. */ static void rfs4_file_walk_callout(rfs4_entry_t u_entry, void *e) { rfs4_file_t *fp = (rfs4_file_t *)u_entry; struct exportinfo *exi = (struct exportinfo *)e; nfs_fh4_fmt_t fhfmt4, *exi_fhp, *finfo_fhp; fhandle_t *efhp; efhp = (fhandle_t *)&exi->exi_fh; exi_fhp = (nfs_fh4_fmt_t *)&fhfmt4; FH_TO_FMT4(efhp, exi_fhp); finfo_fhp = (nfs_fh4_fmt_t *)fp->rf_filehandle.nfs_fh4_val; if (EQFSID(&finfo_fhp->fh4_fsid, &exi_fhp->fh4_fsid) && bcmp(&finfo_fhp->fh4_xdata, &exi_fhp->fh4_xdata, exi_fhp->fh4_xlen) == 0) { if (fp->rf_vp) { vnode_t *vp = fp->rf_vp; /* * don't leak monitors and remove the reference * put on the vnode when the delegation was granted. */ if (fp->rf_dinfo.rd_dtype == OPEN_DELEGATE_READ) { (void) fem_uninstall(vp, deleg_rdops, (void *)fp); vn_open_downgrade(vp, FREAD); } else if (fp->rf_dinfo.rd_dtype == OPEN_DELEGATE_WRITE) { (void) fem_uninstall(vp, deleg_wrops, (void *)fp); vn_open_downgrade(vp, FREAD|FWRITE); } mutex_enter(&vp->v_vsd_lock); (void) vsd_set(vp, nfs4_srv_vkey, NULL); mutex_exit(&vp->v_vsd_lock); VN_RELE(vp); fp->rf_vp = NULL; } rfs4_dbe_invalidate(fp->rf_dbe); } } /* * Given a directory that is being unexported, cleanup/release all * state in the server that refers to objects residing underneath this * particular export. The ordering of the release is important. * Lock_owner, then state and then file. */ void rfs4_clean_state_exi(struct exportinfo *exi) { mutex_enter(&rfs4_state_lock); if (rfs4_server_state == NULL) { mutex_exit(&rfs4_state_lock); return; } rfs4_dbe_walk(rfs4_lo_state_tab, rfs4_lo_state_walk_callout, exi); rfs4_dbe_walk(rfs4_state_tab, rfs4_state_walk_callout, exi); rfs4_dbe_walk(rfs4_deleg_state_tab, rfs4_deleg_state_walk_callout, exi); rfs4_dbe_walk(rfs4_file_tab, rfs4_file_walk_callout, exi); mutex_exit(&rfs4_state_lock); }