1 /* $NetBSD: nfs_node.c,v 1.108 2009/01/02 12:57:29 ad Exp $ */ 2 3 /* 4 * Copyright (c) 1989, 1993 5 * The Regents of the University of California. All rights reserved. 6 * 7 * This code is derived from software contributed to Berkeley by 8 * Rick Macklem at The University of Guelph. 9 * 10 * Redistribution and use in source and binary forms, with or without 11 * modification, are permitted provided that the following conditions 12 * are met: 13 * 1. Redistributions of source code must retain the above copyright 14 * notice, this list of conditions and the following disclaimer. 15 * 2. Redistributions in binary form must reproduce the above copyright 16 * notice, this list of conditions and the following disclaimer in the 17 * documentation and/or other materials provided with the distribution. 18 * 3. Neither the name of the University nor the names of its contributors 19 * may be used to endorse or promote products derived from this software 20 * without specific prior written permission. 21 * 22 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 23 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 24 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 25 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 26 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 27 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 28 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 29 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 30 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 31 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 32 * SUCH DAMAGE. 33 * 34 * @(#)nfs_node.c 8.6 (Berkeley) 5/22/95 35 */ 36 37 #include <sys/cdefs.h> 38 __KERNEL_RCSID(0, "$NetBSD: nfs_node.c,v 1.108 2009/01/02 12:57:29 ad Exp $"); 39 40 #ifdef _KERNEL_OPT 41 #include "opt_nfs.h" 42 #endif 43 44 #include <sys/param.h> 45 #include <sys/systm.h> 46 #include <sys/proc.h> 47 #include <sys/mount.h> 48 #include <sys/namei.h> 49 #include <sys/vnode.h> 50 #include <sys/kernel.h> 51 #include <sys/pool.h> 52 #include <sys/lock.h> 53 #include <sys/hash.h> 54 #include <sys/kauth.h> 55 56 #include <nfs/rpcv2.h> 57 #include <nfs/nfsproto.h> 58 #include <nfs/nfs.h> 59 #include <nfs/nfsnode.h> 60 #include <nfs/nfsmount.h> 61 #include <nfs/nfs_var.h> 62 63 struct pool nfs_node_pool; 64 struct pool nfs_vattr_pool; 65 static struct workqueue *nfs_sillyworkq; 66 67 extern int prtactive; 68 69 static void nfs_gop_size(struct vnode *, off_t, off_t *, int); 70 static int nfs_gop_alloc(struct vnode *, off_t, off_t, int, kauth_cred_t); 71 static int nfs_gop_write(struct vnode *, struct vm_page **, int, int); 72 static void nfs_sillyworker(struct work *, void *); 73 74 static const struct genfs_ops nfs_genfsops = { 75 .gop_size = nfs_gop_size, 76 .gop_alloc = nfs_gop_alloc, 77 .gop_write = nfs_gop_write, 78 }; 79 80 /* 81 * Reinitialize inode hash table. 82 */ 83 void 84 nfs_node_init() 85 { 86 87 pool_init(&nfs_node_pool, sizeof(struct nfsnode), 0, 0, 0, "nfsnodepl", 88 &pool_allocator_nointr, IPL_NONE); 89 pool_init(&nfs_vattr_pool, sizeof(struct vattr), 0, 0, 0, "nfsvapl", 90 &pool_allocator_nointr, IPL_NONE); 91 if (workqueue_create(&nfs_sillyworkq, "nfssilly", nfs_sillyworker, 92 NULL, PRI_NONE, IPL_NONE, 0) != 0) { 93 panic("nfs_node_init"); 94 } 95 } 96 97 /* 98 * Free resources previously allocated in nfs_node_reinit(). 99 */ 100 void 101 nfs_node_done() 102 { 103 104 pool_destroy(&nfs_node_pool); 105 pool_destroy(&nfs_vattr_pool); 106 workqueue_destroy(nfs_sillyworkq); 107 } 108 109 #define RBTONFSNODE(node) \ 110 (void *)((uintptr_t)(node) - offsetof(struct nfsnode, n_rbnode)) 111 112 struct fh_match { 113 nfsfh_t *fhm_fhp; 114 size_t fhm_fhsize; 115 size_t fhm_fhoffset; 116 }; 117 118 static int 119 nfs_compare_nodes(const struct rb_node *parent, const struct rb_node *node) 120 { 121 const struct nfsnode * const pnp = RBTONFSNODE(parent); 122 const struct nfsnode * const np = RBTONFSNODE(node); 123 124 if (pnp->n_fhsize != np->n_fhsize) 125 return np->n_fhsize - pnp->n_fhsize; 126 127 return memcmp(np->n_fhp, pnp->n_fhp, np->n_fhsize); 128 } 129 130 static int 131 nfs_compare_node_fh(const struct rb_node *b, const void *key) 132 { 133 const struct nfsnode * const pnp = RBTONFSNODE(b); 134 const struct fh_match * const fhm = key; 135 136 if (pnp->n_fhsize != fhm->fhm_fhsize) 137 return fhm->fhm_fhsize - pnp->n_fhsize; 138 139 return memcmp(fhm->fhm_fhp, pnp->n_fhp, pnp->n_fhsize); 140 } 141 142 static const struct rb_tree_ops nfs_node_rbtree_ops = { 143 .rbto_compare_nodes = nfs_compare_nodes, 144 .rbto_compare_key = nfs_compare_node_fh, 145 }; 146 147 void 148 nfs_rbtinit(struct nfsmount *nmp) 149 { 150 rb_tree_init(&nmp->nm_rbtree, &nfs_node_rbtree_ops); 151 } 152 153 154 /* 155 * Look up a vnode/nfsnode by file handle. 156 * Callers must check for mount points!! 157 * In all cases, a pointer to a 158 * nfsnode structure is returned. 159 */ 160 int 161 nfs_nget1(mntp, fhp, fhsize, npp, lkflags) 162 struct mount *mntp; 163 nfsfh_t *fhp; 164 int fhsize; 165 struct nfsnode **npp; 166 int lkflags; 167 { 168 struct nfsnode *np; 169 struct vnode *vp; 170 struct nfsmount *nmp = VFSTONFS(mntp); 171 int error; 172 struct fh_match fhm; 173 struct rb_node *node; 174 175 fhm.fhm_fhp = fhp; 176 fhm.fhm_fhsize = fhsize; 177 178 loop: 179 rw_enter(&nmp->nm_rbtlock, RW_READER); 180 node = rb_tree_find_node(&nmp->nm_rbtree, &fhm); 181 if (node != NULL) { 182 np = RBTONFSNODE(node); 183 vp = NFSTOV(np); 184 mutex_enter(&vp->v_interlock); 185 rw_exit(&nmp->nm_rbtlock); 186 error = vget(vp, LK_EXCLUSIVE | LK_INTERLOCK | lkflags); 187 if (error == EBUSY) 188 return error; 189 if (error) 190 goto loop; 191 *npp = np; 192 return(0); 193 } 194 rw_exit(&nmp->nm_rbtlock); 195 196 error = getnewvnode(VT_NFS, mntp, nfsv2_vnodeop_p, &vp); 197 if (error) { 198 *npp = 0; 199 return (error); 200 } 201 np = pool_get(&nfs_node_pool, PR_WAITOK); 202 memset(np, 0, sizeof *np); 203 np->n_vnode = vp; 204 205 /* 206 * Insert the nfsnode in the hash queue for its new file handle 207 */ 208 209 if (fhsize > NFS_SMALLFH) { 210 np->n_fhp = kmem_alloc(fhsize, KM_SLEEP); 211 } else 212 np->n_fhp = &np->n_fh; 213 memcpy(np->n_fhp, fhp, fhsize); 214 np->n_fhsize = fhsize; 215 np->n_accstamp = -1; 216 np->n_vattr = pool_get(&nfs_vattr_pool, PR_WAITOK); 217 218 rw_enter(&nmp->nm_rbtlock, RW_WRITER); 219 if (NULL != rb_tree_find_node(&nmp->nm_rbtree, &fhm)) { 220 rw_exit(&nmp->nm_rbtlock); 221 if (fhsize > NFS_SMALLFH) { 222 kmem_free(np->n_fhp, fhsize); 223 } 224 pool_put(&nfs_vattr_pool, np->n_vattr); 225 pool_put(&nfs_node_pool, np); 226 ungetnewvnode(vp); 227 goto loop; 228 } 229 vp->v_data = np; 230 genfs_node_init(vp, &nfs_genfsops); 231 /* 232 * Initalize read/write creds to useful values. VOP_OPEN will 233 * overwrite these. 234 */ 235 np->n_rcred = curlwp->l_cred; 236 kauth_cred_hold(np->n_rcred); 237 np->n_wcred = curlwp->l_cred; 238 kauth_cred_hold(np->n_wcred); 239 vlockmgr(&vp->v_lock, LK_EXCLUSIVE); 240 NFS_INVALIDATE_ATTRCACHE(np); 241 uvm_vnp_setsize(vp, 0); 242 rb_tree_insert_node(&nmp->nm_rbtree, &np->n_rbnode); 243 rw_exit(&nmp->nm_rbtlock); 244 245 *npp = np; 246 return (0); 247 } 248 249 int 250 nfs_inactive(v) 251 void *v; 252 { 253 struct vop_inactive_args /* { 254 struct vnode *a_vp; 255 bool *a_recycle; 256 } */ *ap = v; 257 struct nfsnode *np; 258 struct sillyrename *sp; 259 struct vnode *vp = ap->a_vp; 260 261 np = VTONFS(vp); 262 if (vp->v_type != VDIR) { 263 sp = np->n_sillyrename; 264 np->n_sillyrename = (struct sillyrename *)0; 265 } else 266 sp = NULL; 267 if (sp != NULL) 268 nfs_vinvalbuf(vp, 0, sp->s_cred, curlwp, 1); 269 *ap->a_recycle = (np->n_flag & NREMOVED) != 0; 270 np->n_flag &= 271 (NMODIFIED | NFLUSHINPROG | NFLUSHWANT | NEOFVALID | NTRUNCDELAYED); 272 273 if (vp->v_type == VDIR && np->n_dircache) 274 nfs_invaldircache(vp, 275 NFS_INVALDIRCACHE_FORCE | NFS_INVALDIRCACHE_KEEPEOF); 276 277 VOP_UNLOCK(vp, 0); 278 279 if (sp != NULL) { 280 workqueue_enqueue(nfs_sillyworkq, &sp->s_work, NULL); 281 } 282 283 return (0); 284 } 285 286 /* 287 * Reclaim an nfsnode so that it can be used for other purposes. 288 */ 289 int 290 nfs_reclaim(v) 291 void *v; 292 { 293 struct vop_reclaim_args /* { 294 struct vnode *a_vp; 295 } */ *ap = v; 296 struct vnode *vp = ap->a_vp; 297 struct nfsnode *np = VTONFS(vp); 298 struct nfsmount *nmp = VFSTONFS(vp->v_mount); 299 300 if (prtactive && vp->v_usecount > 1) 301 vprint("nfs_reclaim: pushing active", vp); 302 303 rw_enter(&nmp->nm_rbtlock, RW_WRITER); 304 rb_tree_remove_node(&nmp->nm_rbtree, &np->n_rbnode); 305 rw_exit(&nmp->nm_rbtlock); 306 307 /* 308 * Free up any directory cookie structures and 309 * large file handle structures that might be associated with 310 * this nfs node. 311 */ 312 if (vp->v_type == VDIR && np->n_dircache != NULL) { 313 nfs_invaldircache(vp, NFS_INVALDIRCACHE_FORCE); 314 hashdone(np->n_dircache, HASH_LIST, nfsdirhashmask); 315 } 316 KASSERT(np->n_dirgens == NULL); 317 318 if (np->n_fhsize > NFS_SMALLFH) 319 kmem_free(np->n_fhp, np->n_fhsize); 320 321 pool_put(&nfs_vattr_pool, np->n_vattr); 322 if (np->n_rcred) 323 kauth_cred_free(np->n_rcred); 324 325 if (np->n_wcred) 326 kauth_cred_free(np->n_wcred); 327 328 cache_purge(vp); 329 if (vp->v_type == VREG) { 330 mutex_destroy(&np->n_commitlock); 331 } 332 genfs_node_destroy(vp); 333 pool_put(&nfs_node_pool, np); 334 vp->v_data = NULL; 335 return (0); 336 } 337 338 void 339 nfs_gop_size(struct vnode *vp, off_t size, off_t *eobp, int flags) 340 { 341 342 *eobp = MAX(size, vp->v_size); 343 } 344 345 int 346 nfs_gop_alloc(struct vnode *vp, off_t off, off_t len, int flags, 347 kauth_cred_t cred) 348 { 349 350 return 0; 351 } 352 353 int 354 nfs_gop_write(struct vnode *vp, struct vm_page **pgs, int npages, int flags) 355 { 356 int i; 357 358 for (i = 0; i < npages; i++) { 359 pmap_page_protect(pgs[i], VM_PROT_READ); 360 } 361 return genfs_gop_write(vp, pgs, npages, flags); 362 } 363 364 /* 365 * Remove a silly file that was rename'd earlier 366 */ 367 static void 368 nfs_sillyworker(struct work *work, void *arg) 369 { 370 struct sillyrename *sp; 371 int error; 372 373 sp = (struct sillyrename *)work; 374 error = vn_lock(sp->s_dvp, LK_EXCLUSIVE); 375 if (error || sp->s_dvp->v_data == NULL) { 376 /* XXX should recover */ 377 printf("%s: vp=%p error=%d\n", __func__, sp->s_dvp, error); 378 if (error == 0) { 379 vput(sp->s_dvp); 380 } else { 381 vrele(sp->s_dvp); 382 } 383 } else { 384 nfs_removeit(sp); 385 vput(sp->s_dvp); 386 } 387 kauth_cred_free(sp->s_cred); 388 kmem_free(sp, sizeof(*sp)); 389 } 390