include/linux/radix-tree.h

   1 /*
   2  * Copyright (C) 2001 Momchil Velikov
   3  * Portions Copyright (C) 2001 Christoph Hellwig
   4  * Copyright (C) 2006 Nick Piggin
   5  *
   6  * This program is free software; you can redistribute it and/or
   7  * modify it under the terms of the GNU General Public License as
   8  * published by the Free Software Foundation; either version 2, or (at
   9  * your option) any later version.
  10  *
  11  * This program is distributed in the hope that it will be useful, but
  12  * WITHOUT ANY WARRANTY; without even the implied warranty of
  13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  14  * General Public License for more details.
  15  *
  16  * You should have received a copy of the GNU General Public License
  17  * along with this program; if not, write to the Free Software
  18  * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
  19  */
  20 #ifndef _LINUX_RADIX_TREE_H
  21 #define _LINUX_RADIX_TREE_H
  22
  23 #include <linux/preempt.h>
  24 #include <linux/types.h>
  25 #include <linux/kernel.h>
  26 #include <linux/rcupdate.h>
  27
  28 /*
  29  * An indirect pointer (root->rnode pointing to a radix_tree_node, rather
  30  * than a data item) is signalled by the low bit set in the root->rnode
  31  * pointer.
  32  *
  33  * In this case root->height is > 0, but the indirect pointer tests are
  34  * needed for RCU lookups (because root->height is unreliable). The only
  35  * time callers need worry about this is when doing a lookup_slot under
  36  * RCU.
  37  *
  38  * Indirect pointer in fact is also used to tag the last pointer of a node
  39  * when it is shrunk, before we rcu free the node. See shrink code for
  40  * details.
  41  */
  42 #define RADIX_TREE_INDIRECT_PTR         1
  43 /*
  44  * A common use of the radix tree is to store pointers to struct pages;
  45  * but shmem/tmpfs needs also to store swap entries in the same tree:
  46  * those are marked as exceptional entries to distinguish them.
  47  * EXCEPTIONAL_ENTRY tests the bit, EXCEPTIONAL_SHIFT shifts content past it.
  48  */
  49 #define RADIX_TREE_EXCEPTIONAL_ENTRY    2
  50 #define RADIX_TREE_EXCEPTIONAL_SHIFT    2
  51
  52 static inline int radix_tree_is_indirect_ptr(void *ptr)
  53 {
  54         return (int)((unsigned long)ptr & RADIX_TREE_INDIRECT_PTR);
  55 }
  56
  57 /*** radix-tree API starts here ***/
  58
  59 #define RADIX_TREE_MAX_TAGS 3
  60
  61 /* root tags are stored in gfp_mask, shifted by __GFP_BITS_SHIFT */
  62 struct radix_tree_root {
  63         unsigned int            height;
  64         gfp_t                   gfp_mask;
  65         struct radix_tree_node  __rcu *rnode;
  66 };
  67
  68 #define RADIX_TREE_INIT(mask)   {                                       \
  69         .height = 0,                                                    \
  70         .gfp_mask = (mask),                                             \
  71         .rnode = NULL,                                                  \
  72 }
  73
  74 #define RADIX_TREE(name, mask) \
  75         struct radix_tree_root name = RADIX_TREE_INIT(mask)
  76
  77 #define INIT_RADIX_TREE(root, mask)                                     \
  78 do {                                                                    \
  79         (root)->height = 0;                                             \
  80         (root)->gfp_mask = (mask);                                      \
  81         (root)->rnode = NULL;                                           \
  82 } while (0)
  83
  84 /**
  85  * Radix-tree synchronization
  86  *
  87  * The radix-tree API requires that users provide all synchronisation (with
  88  * specific exceptions, noted below).
  89  *
  90  * Synchronization of access to the data items being stored in the tree, and
  91  * management of their lifetimes must be completely managed by API users.
  92  *
  93  * For API usage, in general,
  94  * - any function _modifying_ the tree or tags (inserting or deleting
  95  *   items, setting or clearing tags) must exclude other modifications, and
  96  *   exclude any functions reading the tree.
  97  * - any function _reading_ the tree or tags (looking up items or tags,
  98  *   gang lookups) must exclude modifications to the tree, but may occur
  99  *   concurrently with other readers.
 100  *
 101  * The notable exceptions to this rule are the following functions:
 102  * radix_tree_lookup
 103  * radix_tree_lookup_slot
 104  * radix_tree_tag_get
 105  * radix_tree_gang_lookup
 106  * radix_tree_gang_lookup_slot
 107  * radix_tree_gang_lookup_tag
 108  * radix_tree_gang_lookup_tag_slot
 109  * radix_tree_tagged
 110  *
 111  * The first 7 functions are able to be called locklessly, using RCU. The
 112  * caller must ensure calls to these functions are made within rcu_read_lock()
 113  * regions. Other readers (lock-free or otherwise) and modifications may be
 114  * running concurrently.
 115  *
 116  * It is still required that the caller manage the synchronization and lifetimes
 117  * of the items. So if RCU lock-free lookups are used, typically this would mean
 118  * that the items have their own locks, or are amenable to lock-free access; and
 119  * that the items are freed by RCU (or only freed after having been deleted from
 120  * the radix tree *and* a synchronize_rcu() grace period).
 121  *
 122  * (Note, rcu_assign_pointer and rcu_dereference are not needed to control
 123  * access to data items when inserting into or looking up from the radix tree)
 124  *
 125  * Note that the value returned by radix_tree_tag_get() may not be relied upon
 126  * if only the RCU read lock is held.  Functions to set/clear tags and to
 127  * delete nodes running concurrently with it may affect its result such that
 128  * two consecutive reads in the same locked section may return different
 129  * values.  If reliability is required, modification functions must also be
 130  * excluded from concurrency.
 131  *
 132  * radix_tree_tagged is able to be called without locking or RCU.
 133  */
 134
 135 /**
 136  * radix_tree_deref_slot        - dereference a slot
 137  * @pslot:      pointer to slot, returned by radix_tree_lookup_slot
 138  * Returns:     item that was stored in that slot with any direct pointer flag
 139  *              removed.
 140  *
 141  * For use with radix_tree_lookup_slot().  Caller must hold tree at least read
 142  * locked across slot lookup and dereference. Not required if write lock is
 143  * held (ie. items cannot be concurrently inserted).
 144  *
 145  * radix_tree_deref_retry must be used to confirm validity of the pointer if
 146  * only the read lock is held.
 147  */
 148 static inline void *radix_tree_deref_slot(void **pslot)
 149 {
 150         return rcu_dereference(*pslot);
 151 }
 152
 153 /**
 154  * radix_tree_deref_slot_protected      - dereference a slot without RCU lock but with tree lock held
 155  * @pslot:      pointer to slot, returned by radix_tree_lookup_slot
 156  * Returns:     item that was stored in that slot with any direct pointer flag
 157  *              removed.
 158  *
 159  * Similar to radix_tree_deref_slot but only used during migration when a pages
 160  * mapping is being moved. The caller does not hold the RCU read lock but it
 161  * must hold the tree lock to prevent parallel updates.
 162  */
 163 static inline void *radix_tree_deref_slot_protected(void **pslot,
 164                                                         spinlock_t *treelock)
 165 {
 166         return rcu_dereference_protected(*pslot, lockdep_is_held(treelock));
 167 }
 168
 169 /**
 170  * radix_tree_deref_retry       - check radix_tree_deref_slot
 171  * @arg:        pointer returned by radix_tree_deref_slot
 172  * Returns:     0 if retry is not required, otherwise retry is required
 173  *
 174  * radix_tree_deref_retry must be used with radix_tree_deref_slot.
 175  */
 176 static inline int radix_tree_deref_retry(void *arg)
 177 {
 178         return unlikely((unsigned long)arg & RADIX_TREE_INDIRECT_PTR);
 179 }
 180
 181 /**
 182  * radix_tree_exceptional_entry - radix_tree_deref_slot gave exceptional entry?
 183  * @arg:        value returned by radix_tree_deref_slot
 184  * Returns:     0 if well-aligned pointer, non-0 if exceptional entry.
 185  */
 186 static inline int radix_tree_exceptional_entry(void *arg)
 187 {
 188         /* Not unlikely because radix_tree_exception often tested first */
 189         return (unsigned long)arg & RADIX_TREE_EXCEPTIONAL_ENTRY;
 190 }
 191
 192 /**
 193  * radix_tree_exception - radix_tree_deref_slot returned either exception?
 194  * @arg:        value returned by radix_tree_deref_slot
 195  * Returns:     0 if well-aligned pointer, non-0 if either kind of exception.
 196  */
 197 static inline int radix_tree_exception(void *arg)
 198 {
 199         return unlikely((unsigned long)arg &
 200                 (RADIX_TREE_INDIRECT_PTR | RADIX_TREE_EXCEPTIONAL_ENTRY));
 201 }
 202
 203 /**
 204  * radix_tree_replace_slot      - replace item in a slot
 205  * @pslot:      pointer to slot, returned by radix_tree_lookup_slot
 206  * @item:       new item to store in the slot.
 207  *
 208  * For use with radix_tree_lookup_slot().  Caller must hold tree write locked
 209  * across slot lookup and replacement.
 210  */
 211 static inline void radix_tree_replace_slot(void **pslot, void *item)
 212 {
 213         BUG_ON(radix_tree_is_indirect_ptr(item));
 214         rcu_assign_pointer(*pslot, item);
 215 }
 216
 217 int radix_tree_insert(struct radix_tree_root *, unsigned long, void *);
 218 void *radix_tree_lookup(struct radix_tree_root *, unsigned long);
 219 void **radix_tree_lookup_slot(struct radix_tree_root *, unsigned long);
 220 void *radix_tree_delete(struct radix_tree_root *, unsigned long);
 221 unsigned int
 222 radix_tree_gang_lookup(struct radix_tree_root *root, void **results,
 223                         unsigned long first_index, unsigned int max_items);
 224 unsigned int radix_tree_gang_lookup_slot(struct radix_tree_root *root,
 225                         void ***results, unsigned long *indices,
 226                         unsigned long first_index, unsigned int max_items);
 227 unsigned long radix_tree_next_hole(struct radix_tree_root *root,
 228                                 unsigned long index, unsigned long max_scan);
 229 unsigned long radix_tree_prev_hole(struct radix_tree_root *root,
 230                                 unsigned long index, unsigned long max_scan);
 231 int radix_tree_preload(gfp_t gfp_mask);
 232 void radix_tree_init(void);
 233 void *radix_tree_tag_set(struct radix_tree_root *root,
 234                         unsigned long index, unsigned int tag);
 235 void *radix_tree_tag_clear(struct radix_tree_root *root,
 236                         unsigned long index, unsigned int tag);
 237 int radix_tree_tag_get(struct radix_tree_root *root,
 238                         unsigned long index, unsigned int tag);
 239 unsigned int
 240 radix_tree_gang_lookup_tag(struct radix_tree_root *root, void **results,
 241                 unsigned long first_index, unsigned int max_items,
 242                 unsigned int tag);
 243 unsigned int
 244 radix_tree_gang_lookup_tag_slot(struct radix_tree_root *root, void ***results,
 245                 unsigned long first_index, unsigned int max_items,
 246                 unsigned int tag);
 247 unsigned long radix_tree_range_tag_if_tagged(struct radix_tree_root *root,
 248                 unsigned long *first_indexp, unsigned long last_index,
 249                 unsigned long nr_to_tag,
 250                 unsigned int fromtag, unsigned int totag);
 251 int radix_tree_tagged(struct radix_tree_root *root, unsigned int tag);
 252 unsigned long radix_tree_locate_item(struct radix_tree_root *root, void *item);
 253
 254 static inline void radix_tree_preload_end(void)
 255 {
 256         preempt_enable();
 257 }
 258
 259 #endif /* _LINUX_RADIX_TREE_H */