sys/vm/vm_radix.c

   1 /*
   2  * Copyright (c) 2013 EMC Corp.
   3  * Copyright (c) 2011 Jeffrey Roberson <jeff@freebsd.org>
   4  * Copyright (c) 2008 Mayur Shardul <mayur.shardul@gmail.com>
   5  * All rights reserved.
   6  *
   7  * Redistribution and use in source and binary forms, with or without
   8  * modification, are permitted provided that the following conditions
   9  * are met:
  10  * 1. Redistributions of source code must retain the above copyright
  11  *    notice, this list of conditions and the following disclaimer.
  12  * 2. Redistributions in binary form must reproduce the above copyright
  13  *    notice, this list of conditions and the following disclaimer in the
  14  *    documentation and/or other materials provided with the distribution.
  15  *
  16  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
  17  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  18  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  19  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
  20  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  21  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  22  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  23  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  24  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  25  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  26  * SUCH DAMAGE.
  27  *
  28  */
  29
  30 /*
  31  * Path-compressed radix trie implementation.
  32  * The following code is not generalized into a general purpose library
  33  * because there are way too many parameters embedded that should really
  34  * be decided by the library consumers.  At the same time, consumers
  35  * of this code must achieve highest possible performance.
  36  *
  37  * The implementation takes into account the following rationale:
  38  * - Size of the nodes should be as small as possible but still big enough
  39  *   to avoid a large maximum depth for the trie.  This is a balance
  40  *   between the necessity to not wire too much physical memory for the nodes
  41  *   and the necessity to avoid too much cache pollution during the trie
  42  *   operations.
  43  * - There is not a huge bias toward the number of lookup operations over
  44  *   the number of insert and remove operations.  This basically implies
  45  *   that optimizations supposedly helping one operation but hurting the
  46  *   other might be carefully evaluated.
  47  * - On average not many nodes are expected to be fully populated, hence
  48  *   level compression may just complicate things.
  49  */
  50
  51 #include <sys/cdefs.h>
  52 __FBSDID("$FreeBSD$");
  53
  54 #include "opt_ddb.h"
  55
  56 #include <sys/param.h>
  57 #include <sys/systm.h>
  58 #include <sys/kernel.h>
  59 #include <sys/vmmeter.h>
  60
  61 #include <vm/uma.h>
  62 #include <vm/vm.h>
  63 #include <vm/vm_param.h>
  64 #include <vm/vm_page.h>
  65 #include <vm/vm_radix.h>
  66
  67 #ifdef DDB
  68 #include <ddb/ddb.h>
  69 #endif
  70
  71 /*
  72  * These widths should allow the pointers to a node's children to fit within
  73  * a single cache line.  The extra levels from a narrow width should not be
  74  * a problem thanks to path compression.
  75  */
  76 #ifdef __LP64__
  77 #define VM_RADIX_WIDTH  4
  78 #else
  79 #define VM_RADIX_WIDTH  3
  80 #endif
  81
  82 #define VM_RADIX_COUNT  (1 << VM_RADIX_WIDTH)
  83 #define VM_RADIX_MASK   (VM_RADIX_COUNT - 1)
  84 #define VM_RADIX_LIMIT                                                  \
  85         (howmany((sizeof(vm_pindex_t) * NBBY), VM_RADIX_WIDTH) - 1)
  86
  87 /* Flag bits stored in node pointers. */
  88 #define VM_RADIX_ISLEAF 0x1
  89 #define VM_RADIX_FLAGS  0x1
  90 #define VM_RADIX_PAD    VM_RADIX_FLAGS
  91
  92 /* Returns one unit associated with specified level. */
  93 #define VM_RADIX_UNITLEVEL(lev)                                         \
  94         ((vm_pindex_t)1 << ((lev) * VM_RADIX_WIDTH))
  95
  96 struct vm_radix_node {
  97         vm_pindex_t      rn_owner;                      /* Owner of record. */
  98         uint16_t         rn_count;                      /* Valid children. */
  99         uint16_t         rn_clev;                       /* Current level. */
 100         void            *rn_child[VM_RADIX_COUNT];      /* Child nodes. */
 101 };
 102
 103 static uma_zone_t vm_radix_node_zone;
 104
 105 /*
 106  * Allocate a radix node.  Pre-allocation should ensure that the request
 107  * will always be satisfied.
 108  */
 109 static __inline struct vm_radix_node *
 110 vm_radix_node_get(vm_pindex_t owner, uint16_t count, uint16_t clevel)
 111 {
 112         struct vm_radix_node *rnode;
 113
 114         rnode = uma_zalloc(vm_radix_node_zone, M_NOWAIT);
 115
 116         /*
 117          * The required number of nodes should already be pre-allocated
 118          * by vm_radix_prealloc().  However, UMA can hold a few nodes
 119          * in per-CPU buckets, which will not be accessible by the
 120          * current CPU.  Thus, the allocation could return NULL when
 121          * the pre-allocated pool is close to exhaustion.  Anyway,
 122          * in practice this should never occur because a new node
 123          * is not always required for insert.  Thus, the pre-allocated
 124          * pool should have some extra pages that prevent this from
 125          * becoming a problem.
 126          */
 127         if (rnode == NULL)
 128                 panic("%s: uma_zalloc() returned NULL for a new node",
 129                     __func__);
 130         rnode->rn_owner = owner;
 131         rnode->rn_count = count;
 132         rnode->rn_clev = clevel;
 133         return (rnode);
 134 }
 135
 136 /*
 137  * Free radix node.
 138  */
 139 static __inline void
 140 vm_radix_node_put(struct vm_radix_node *rnode)
 141 {
 142
 143         uma_zfree(vm_radix_node_zone, rnode);
 144 }
 145
 146 /*
 147  * Return the position in the array for a given level.
 148  */
 149 static __inline int
 150 vm_radix_slot(vm_pindex_t index, uint16_t level)
 151 {
 152
 153         return ((index >> (level * VM_RADIX_WIDTH)) & VM_RADIX_MASK);
 154 }
 155
 156 /* Trims the key after the specified level. */
 157 static __inline vm_pindex_t
 158 vm_radix_trimkey(vm_pindex_t index, uint16_t level)
 159 {
 160         vm_pindex_t ret;
 161
 162         ret = index;
 163         if (level > 0) {
 164                 ret >>= level * VM_RADIX_WIDTH;
 165                 ret <<= level * VM_RADIX_WIDTH;
 166         }
 167         return (ret);
 168 }
 169
 170 /*
 171  * Get the root node for a radix tree.
 172  */
 173 static __inline struct vm_radix_node *
 174 vm_radix_getroot(struct vm_radix *rtree)
 175 {
 176
 177         return ((struct vm_radix_node *)rtree->rt_root);
 178 }
 179
 180 /*
 181  * Set the root node for a radix tree.
 182  */
 183 static __inline void
 184 vm_radix_setroot(struct vm_radix *rtree, struct vm_radix_node *rnode)
 185 {
 186
 187         rtree->rt_root = (uintptr_t)rnode;
 188 }
 189
 190 /*
 191  * Returns TRUE if the specified radix node is a leaf and FALSE otherwise.
 192  */
 193 static __inline boolean_t
 194 vm_radix_isleaf(struct vm_radix_node *rnode)
 195 {
 196
 197         return (((uintptr_t)rnode & VM_RADIX_ISLEAF) != 0);
 198 }
 199
 200 /*
 201  * Returns the associated page extracted from rnode.
 202  */
 203 static __inline vm_page_t
 204 vm_radix_topage(struct vm_radix_node *rnode)
 205 {
 206
 207         return ((vm_page_t)((uintptr_t)rnode & ~VM_RADIX_FLAGS));
 208 }
 209
 210 /*
 211  * Adds the page as a child of the provided node.
 212  */
 213 static __inline void
 214 vm_radix_addpage(struct vm_radix_node *rnode, vm_pindex_t index, uint16_t clev,
 215     vm_page_t page)
 216 {
 217         int slot;
 218
 219         slot = vm_radix_slot(index, clev);
 220         rnode->rn_child[slot] = (void *)((uintptr_t)page | VM_RADIX_ISLEAF);
 221 }
 222
 223 /*
 224  * Returns the slot where two keys differ.
 225  * It cannot accept 2 equal keys.
 226  */
 227 static __inline uint16_t
 228 vm_radix_keydiff(vm_pindex_t index1, vm_pindex_t index2)
 229 {
 230         uint16_t clev;
 231
 232         KASSERT(index1 != index2, ("%s: passing the same key value %jx",
 233             __func__, (uintmax_t)index1));
 234
 235         index1 ^= index2;
 236         for (clev = VM_RADIX_LIMIT;; clev--)
 237                 if (vm_radix_slot(index1, clev) != 0)
 238                         return (clev);
 239 }
 240
 241 /*
 242  * Returns TRUE if it can be determined that key does not belong to the
 243  * specified rnode.  Otherwise, returns FALSE.
 244  */
 245 static __inline boolean_t
 246 vm_radix_keybarr(struct vm_radix_node *rnode, vm_pindex_t idx)
 247 {
 248
 249         if (rnode->rn_clev < VM_RADIX_LIMIT) {
 250                 idx = vm_radix_trimkey(idx, rnode->rn_clev + 1);
 251                 return (idx != rnode->rn_owner);
 252         }
 253         return (FALSE);
 254 }
 255
 256 /*
 257  * Internal helper for vm_radix_reclaim_allnodes().
 258  * This function is recursive.
 259  */
 260 static void
 261 vm_radix_reclaim_allnodes_int(struct vm_radix_node *rnode)
 262 {
 263         int slot;
 264
 265         KASSERT(rnode->rn_count <= VM_RADIX_COUNT,
 266             ("vm_radix_reclaim_allnodes_int: bad count in rnode %p", rnode));
 267         for (slot = 0; rnode->rn_count != 0; slot++) {
 268                 if (rnode->rn_child[slot] == NULL)
 269                         continue;
 270                 if (!vm_radix_isleaf(rnode->rn_child[slot]))
 271                         vm_radix_reclaim_allnodes_int(rnode->rn_child[slot]);
 272                 rnode->rn_child[slot] = NULL;
 273                 rnode->rn_count--;
 274         }
 275         vm_radix_node_put(rnode);
 276 }
 277
 278 #ifdef INVARIANTS
 279 /*
 280  * Radix node zone destructor.
 281  */
 282 static void
 283 vm_radix_node_zone_dtor(void *mem, int size __unused, void *arg __unused)
 284 {
 285         struct vm_radix_node *rnode;
 286         int slot;
 287
 288         rnode = mem;
 289         KASSERT(rnode->rn_count == 0,
 290             ("vm_radix_node_put: rnode %p has %d children", rnode,
 291             rnode->rn_count));
 292         for (slot = 0; slot < VM_RADIX_COUNT; slot++)
 293                 KASSERT(rnode->rn_child[slot] == NULL,
 294                     ("vm_radix_node_put: rnode %p has a child", rnode));
 295 }
 296 #endif
 297
 298 /*
 299  * Radix node zone initializer.
 300  */
 301 static int
 302 vm_radix_node_zone_init(void *mem, int size __unused, int flags __unused)
 303 {
 304         struct vm_radix_node *rnode;
 305
 306         rnode = mem;
 307         memset(rnode->rn_child, 0, sizeof(rnode->rn_child));
 308         return (0);
 309 }
 310
 311 /*
 312  * Pre-allocate intermediate nodes from the UMA slab zone.
 313  */
 314 static void
 315 vm_radix_prealloc(void *arg __unused)
 316 {
 317         int nodes;
 318
 319         /*
 320          * Calculate the number of reserved nodes, discounting the pages that
 321          * are needed to store them.
 322          */
 323         nodes = ((vm_paddr_t)cnt.v_page_count * PAGE_SIZE) / (PAGE_SIZE +
 324             sizeof(struct vm_radix_node));
 325         if (!uma_zone_reserve_kva(vm_radix_node_zone, nodes))
 326                 panic("%s: unable to create new zone", __func__);
 327         uma_prealloc(vm_radix_node_zone, nodes);
 328 }
 329 SYSINIT(vm_radix_prealloc, SI_SUB_KMEM, SI_ORDER_SECOND, vm_radix_prealloc,
 330     NULL);
 331
 332 /*
 333  * Initialize the UMA slab zone.
 334  * Until vm_radix_prealloc() is called, the zone will be served by the
 335  * UMA boot-time pre-allocated pool of pages.
 336  */
 337 void
 338 vm_radix_init(void)
 339 {
 340
 341         vm_radix_node_zone = uma_zcreate("RADIX NODE",
 342             sizeof(struct vm_radix_node), NULL,
 343 #ifdef INVARIANTS
 344             vm_radix_node_zone_dtor,
 345 #else
 346             NULL,
 347 #endif
 348             vm_radix_node_zone_init, NULL, VM_RADIX_PAD, UMA_ZONE_VM |
 349             UMA_ZONE_NOFREE);
 350 }
 351
 352 /*
 353  * Inserts the key-value pair into the trie.
 354  * Panics if the key already exists.
 355  */
 356 void
 357 vm_radix_insert(struct vm_radix *rtree, vm_page_t page)
 358 {
 359         vm_pindex_t index, newind;
 360         void **parentp;
 361         struct vm_radix_node *rnode, *tmp;
 362         vm_page_t m;
 363         int slot;
 364         uint16_t clev;
 365
 366         index = page->pindex;
 367
 368         /*
 369          * The owner of record for root is not really important because it
 370          * will never be used.
 371          */
 372         rnode = vm_radix_getroot(rtree);
 373         if (rnode == NULL) {
 374                 rtree->rt_root = (uintptr_t)page | VM_RADIX_ISLEAF;
 375                 return;
 376         }
 377         parentp = (void **)&rtree->rt_root;
 378         for (;;) {
 379                 if (vm_radix_isleaf(rnode)) {
 380                         m = vm_radix_topage(rnode);
 381                         if (m->pindex == index)
 382                                 panic("%s: key %jx is already present",
 383                                     __func__, (uintmax_t)index);
 384                         clev = vm_radix_keydiff(m->pindex, index);
 385                         tmp = vm_radix_node_get(vm_radix_trimkey(index,
 386                             clev + 1), 2, clev);
 387                         *parentp = tmp;
 388                         vm_radix_addpage(tmp, index, clev, page);
 389                         vm_radix_addpage(tmp, m->pindex, clev, m);
 390                         return;
 391                 } else if (vm_radix_keybarr(rnode, index))
 392                         break;
 393                 slot = vm_radix_slot(index, rnode->rn_clev);
 394                 if (rnode->rn_child[slot] == NULL) {
 395                         rnode->rn_count++;
 396                         vm_radix_addpage(rnode, index, rnode->rn_clev, page);
 397                         return;
 398                 }
 399                 parentp = &rnode->rn_child[slot];
 400                 rnode = rnode->rn_child[slot];
 401         }
 402
 403         /*
 404          * A new node is needed because the right insertion level is reached.
 405          * Setup the new intermediate node and add the 2 children: the
 406          * new object and the older edge.
 407          */
 408         newind = rnode->rn_owner;
 409         clev = vm_radix_keydiff(newind, index);
 410         tmp = vm_radix_node_get(vm_radix_trimkey(index, clev + 1), 2,
 411             clev);
 412         *parentp = tmp;
 413         vm_radix_addpage(tmp, index, clev, page);
 414         slot = vm_radix_slot(newind, clev);
 415         tmp->rn_child[slot] = rnode;
 416 }
 417
 418 /*
 419  * Returns the value stored at the index.  If the index is not present,
 420  * NULL is returned.
 421  */
 422 vm_page_t
 423 vm_radix_lookup(struct vm_radix *rtree, vm_pindex_t index)
 424 {
 425         struct vm_radix_node *rnode;
 426         vm_page_t m;
 427         int slot;
 428
 429         rnode = vm_radix_getroot(rtree);
 430         while (rnode != NULL) {
 431                 if (vm_radix_isleaf(rnode)) {
 432                         m = vm_radix_topage(rnode);
 433                         if (m->pindex == index)
 434                                 return (m);
 435                         else
 436                                 break;
 437                 } else if (vm_radix_keybarr(rnode, index))
 438                         break;
 439                 slot = vm_radix_slot(index, rnode->rn_clev);
 440                 rnode = rnode->rn_child[slot];
 441         }
 442         return (NULL);
 443 }
 444
 445 /*
 446  * Look up the nearest entry at a position bigger than or equal to index.
 447  */
 448 vm_page_t
 449 vm_radix_lookup_ge(struct vm_radix *rtree, vm_pindex_t index)
 450 {
 451         struct vm_radix_node *stack[VM_RADIX_LIMIT];
 452         vm_pindex_t inc;
 453         vm_page_t m;
 454         struct vm_radix_node *child, *rnode;
 455 #ifdef INVARIANTS
 456         int loops = 0;
 457 #endif
 458         int slot, tos;
 459
 460         rnode = vm_radix_getroot(rtree);
 461         if (rnode == NULL)
 462                 return (NULL);
 463         else if (vm_radix_isleaf(rnode)) {
 464                 m = vm_radix_topage(rnode);
 465                 if (m->pindex >= index)
 466                         return (m);
 467                 else
 468                         return (NULL);
 469         }
 470         tos = 0;
 471         for (;;) {
 472                 /*
 473                  * If the keys differ before the current bisection node,
 474                  * then the search key might rollback to the earliest
 475                  * available bisection node or to the smallest key
 476                  * in the current node (if the owner is bigger than the
 477                  * search key).
 478                  */
 479                 if (vm_radix_keybarr(rnode, index)) {
 480                         if (index > rnode->rn_owner) {
 481 ascend:
 482                                 KASSERT(++loops < 1000,
 483                                     ("vm_radix_lookup_ge: too many loops"));
 484
 485                                 /*
 486                                  * Pop nodes from the stack until either the
 487                                  * stack is empty or a node that could have a
 488                                  * matching descendant is found.
 489                                  */
 490                                 do {
 491                                         if (tos == 0)
 492                                                 return (NULL);
 493                                         rnode = stack[--tos];
 494                                 } while (vm_radix_slot(index,
 495                                     rnode->rn_clev) == (VM_RADIX_COUNT - 1));
 496
 497                                 /*
 498                                  * The following computation cannot overflow
 499                                  * because index's slot at the current level
 500                                  * is less than VM_RADIX_COUNT - 1.
 501                                  */
 502                                 index = vm_radix_trimkey(index,
 503                                     rnode->rn_clev);
 504                                 index += VM_RADIX_UNITLEVEL(rnode->rn_clev);
 505                         } else
 506                                 index = rnode->rn_owner;
 507                         KASSERT(!vm_radix_keybarr(rnode, index),
 508                             ("vm_radix_lookup_ge: keybarr failed"));
 509                 }
 510                 slot = vm_radix_slot(index, rnode->rn_clev);
 511                 child = rnode->rn_child[slot];
 512                 if (vm_radix_isleaf(child)) {
 513                         m = vm_radix_topage(child);
 514                         if (m->pindex >= index)
 515                                 return (m);
 516                 } else if (child != NULL)
 517                         goto descend;
 518
 519                 /*
 520                  * Look for an available edge or page within the current
 521                  * bisection node.
 522                  */
 523                 if (slot < (VM_RADIX_COUNT - 1)) {
 524                         inc = VM_RADIX_UNITLEVEL(rnode->rn_clev);
 525                         index = vm_radix_trimkey(index, rnode->rn_clev);
 526                         do {
 527                                 index += inc;
 528                                 slot++;
 529                                 child = rnode->rn_child[slot];
 530                                 if (vm_radix_isleaf(child)) {
 531                                         m = vm_radix_topage(child);
 532                                         if (m->pindex >= index)
 533                                                 return (m);
 534                                 } else if (child != NULL)
 535                                         goto descend;
 536                         } while (slot < (VM_RADIX_COUNT - 1));
 537                 }
 538                 KASSERT(child == NULL || vm_radix_isleaf(child),
 539                     ("vm_radix_lookup_ge: child is radix node"));
 540
 541                 /*
 542                  * If a page or edge bigger than the search slot is not found
 543                  * in the current node, ascend to the next higher-level node.
 544                  */
 545                 goto ascend;
 546 descend:
 547                 KASSERT(rnode->rn_clev > 0,
 548                     ("vm_radix_lookup_ge: pushing leaf's parent"));
 549                 KASSERT(tos < VM_RADIX_LIMIT,
 550                     ("vm_radix_lookup_ge: stack overflow"));
 551                 stack[tos++] = rnode;
 552                 rnode = child;
 553         }
 554 }
 555
 556 /*
 557  * Look up the nearest entry at a position less than or equal to index.
 558  */
 559 vm_page_t
 560 vm_radix_lookup_le(struct vm_radix *rtree, vm_pindex_t index)
 561 {
 562         struct vm_radix_node *stack[VM_RADIX_LIMIT];
 563         vm_pindex_t inc;
 564         vm_page_t m;
 565         struct vm_radix_node *child, *rnode;
 566 #ifdef INVARIANTS
 567         int loops = 0;
 568 #endif
 569         int slot, tos;
 570
 571         rnode = vm_radix_getroot(rtree);
 572         if (rnode == NULL)
 573                 return (NULL);
 574         else if (vm_radix_isleaf(rnode)) {
 575                 m = vm_radix_topage(rnode);
 576                 if (m->pindex <= index)
 577                         return (m);
 578                 else
 579                         return (NULL);
 580         }
 581         tos = 0;
 582         for (;;) {
 583                 /*
 584                  * If the keys differ before the current bisection node,
 585                  * then the search key might rollback to the earliest
 586                  * available bisection node or to the largest key
 587                  * in the current node (if the owner is smaller than the
 588                  * search key).
 589                  */
 590                 if (vm_radix_keybarr(rnode, index)) {
 591                         if (index > rnode->rn_owner) {
 592                                 index = rnode->rn_owner + VM_RADIX_COUNT *
 593                                     VM_RADIX_UNITLEVEL(rnode->rn_clev);
 594                         } else {
 595 ascend:
 596                                 KASSERT(++loops < 1000,
 597                                     ("vm_radix_lookup_le: too many loops"));
 598
 599                                 /*
 600                                  * Pop nodes from the stack until either the
 601                                  * stack is empty or a node that could have a
 602                                  * matching descendant is found.
 603                                  */
 604                                 do {
 605                                         if (tos == 0)
 606                                                 return (NULL);
 607                                         rnode = stack[--tos];
 608                                 } while (vm_radix_slot(index,
 609                                     rnode->rn_clev) == 0);
 610
 611                                 /*
 612                                  * The following computation cannot overflow
 613                                  * because index's slot at the current level
 614                                  * is greater than 0.
 615                                  */
 616                                 index = vm_radix_trimkey(index,
 617                                     rnode->rn_clev);
 618                         }
 619                         index--;
 620                         KASSERT(!vm_radix_keybarr(rnode, index),
 621                             ("vm_radix_lookup_le: keybarr failed"));
 622                 }
 623                 slot = vm_radix_slot(index, rnode->rn_clev);
 624                 child = rnode->rn_child[slot];
 625                 if (vm_radix_isleaf(child)) {
 626                         m = vm_radix_topage(child);
 627                         if (m->pindex <= index)
 628                                 return (m);
 629                 } else if (child != NULL)
 630                         goto descend;
 631
 632                 /*
 633                  * Look for an available edge or page within the current
 634                  * bisection node.
 635                  */
 636                 if (slot > 0) {
 637                         inc = VM_RADIX_UNITLEVEL(rnode->rn_clev);
 638                         index |= inc - 1;
 639                         do {
 640                                 index -= inc;
 641                                 slot--;
 642                                 child = rnode->rn_child[slot];
 643                                 if (vm_radix_isleaf(child)) {
 644                                         m = vm_radix_topage(child);
 645                                         if (m->pindex <= index)
 646                                                 return (m);
 647                                 } else if (child != NULL)
 648                                         goto descend;
 649                         } while (slot > 0);
 650                 }
 651                 KASSERT(child == NULL || vm_radix_isleaf(child),
 652                     ("vm_radix_lookup_le: child is radix node"));
 653
 654                 /*
 655                  * If a page or edge smaller than the search slot is not found
 656                  * in the current node, ascend to the next higher-level node.
 657                  */
 658                 goto ascend;
 659 descend:
 660                 KASSERT(rnode->rn_clev > 0,
 661                     ("vm_radix_lookup_le: pushing leaf's parent"));
 662                 KASSERT(tos < VM_RADIX_LIMIT,
 663                     ("vm_radix_lookup_le: stack overflow"));
 664                 stack[tos++] = rnode;
 665                 rnode = child;
 666         }
 667 }
 668
 669 /*
 670  * Remove the specified index from the tree.
 671  * Panics if the key is not present.
 672  */
 673 void
 674 vm_radix_remove(struct vm_radix *rtree, vm_pindex_t index)
 675 {
 676         struct vm_radix_node *rnode, *parent;
 677         vm_page_t m;
 678         int i, slot;
 679
 680         rnode = vm_radix_getroot(rtree);
 681         if (vm_radix_isleaf(rnode)) {
 682                 m = vm_radix_topage(rnode);
 683                 if (m->pindex != index)
 684                         panic("%s: invalid key found", __func__);
 685                 vm_radix_setroot(rtree, NULL);
 686                 return;
 687         }
 688         parent = NULL;
 689         for (;;) {
 690                 if (rnode == NULL)
 691                         panic("vm_radix_remove: impossible to locate the key");
 692                 slot = vm_radix_slot(index, rnode->rn_clev);
 693                 if (vm_radix_isleaf(rnode->rn_child[slot])) {
 694                         m = vm_radix_topage(rnode->rn_child[slot]);
 695                         if (m->pindex != index)
 696                                 panic("%s: invalid key found", __func__);
 697                         rnode->rn_child[slot] = NULL;
 698                         rnode->rn_count--;
 699                         if (rnode->rn_count > 1)
 700                                 break;
 701                         for (i = 0; i < VM_RADIX_COUNT; i++)
 702                                 if (rnode->rn_child[i] != NULL)
 703                                         break;
 704                         KASSERT(i != VM_RADIX_COUNT,
 705                             ("%s: invalid node configuration", __func__));
 706                         if (parent == NULL)
 707                                 vm_radix_setroot(rtree, rnode->rn_child[i]);
 708                         else {
 709                                 slot = vm_radix_slot(index, parent->rn_clev);
 710                                 KASSERT(parent->rn_child[slot] == rnode,
 711                                     ("%s: invalid child value", __func__));
 712                                 parent->rn_child[slot] = rnode->rn_child[i];
 713                         }
 714                         rnode->rn_count--;
 715                         rnode->rn_child[i] = NULL;
 716                         vm_radix_node_put(rnode);
 717                         break;
 718                 }
 719                 parent = rnode;
 720                 rnode = rnode->rn_child[slot];
 721         }
 722 }
 723
 724 /*
 725  * Remove and free all the nodes from the radix tree.
 726  * This function is recursive but there is a tight control on it as the
 727  * maximum depth of the tree is fixed.
 728  */
 729 void
 730 vm_radix_reclaim_allnodes(struct vm_radix *rtree)
 731 {
 732         struct vm_radix_node *root;
 733
 734         root = vm_radix_getroot(rtree);
 735         if (root == NULL)
 736                 return;
 737         vm_radix_setroot(rtree, NULL);
 738         if (!vm_radix_isleaf(root))
 739                 vm_radix_reclaim_allnodes_int(root);
 740 }
 741
 742 #ifdef DDB
 743 /*
 744  * Show details about the given radix node.
 745  */
 746 DB_SHOW_COMMAND(radixnode, db_show_radixnode)
 747 {
 748         struct vm_radix_node *rnode;
 749         int i;
 750
 751         if (!have_addr)
 752                 return;
 753         rnode = (struct vm_radix_node *)addr;
 754         db_printf("radixnode %p, owner %jx, children count %u, level %u:\n",
 755             (void *)rnode, (uintmax_t)rnode->rn_owner, rnode->rn_count,
 756             rnode->rn_clev);
 757         for (i = 0; i < VM_RADIX_COUNT; i++)
 758                 if (rnode->rn_child[i] != NULL)
 759                         db_printf("slot: %d, val: %p, page: %p, clev: %d\n",
 760                             i, (void *)rnode->rn_child[i],
 761                             vm_radix_isleaf(rnode->rn_child[i]) ?
 762                             vm_radix_topage(rnode->rn_child[i]) : NULL,
 763                             rnode->rn_clev);
 764 }
 765 #endif /* DDB */