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 #define radix_tree_indirect_to_ptr(ptr) \
  53         radix_tree_indirect_to_ptr((void __force *)(ptr))
  54
  55 static inline int radix_tree_is_indirect_ptr(void *ptr)
  56 {
  57         return (int)((unsigned long)ptr & RADIX_TREE_INDIRECT_PTR);
  58 }
  59
  60 /*** radix-tree API starts here ***/
  61
  62 #define RADIX_TREE_MAX_TAGS 3
  63
  64 /* root tags are stored in gfp_mask, shifted by __GFP_BITS_SHIFT */
  65 struct radix_tree_root {
  66         unsigned int            height;
  67         gfp_t                   gfp_mask;
  68         struct radix_tree_node  __rcu *rnode;
  69 };
  70
  71 #define RADIX_TREE_INIT(mask)   {                                       \
  72         .height = 0,                                                    \
  73         .gfp_mask = (mask),                                             \
  74         .rnode = NULL,                                                  \
  75 }
  76
  77 #define RADIX_TREE(name, mask) \
  78         struct radix_tree_root name = RADIX_TREE_INIT(mask)
  79
  80 #define INIT_RADIX_TREE(root, mask)                                     \
  81 do {                                                                    \
  82         (root)->height = 0;                                             \
  83         (root)->gfp_mask = (mask);                                      \
  84         (root)->rnode = NULL;                                           \
  85 } while (0)
  86
  87 /**
  88  * Radix-tree synchronization
  89  *
  90  * The radix-tree API requires that users provide all synchronisation (with
  91  * specific exceptions, noted below).
  92  *
  93  * Synchronization of access to the data items being stored in the tree, and
  94  * management of their lifetimes must be completely managed by API users.
  95  *
  96  * For API usage, in general,
  97  * - any function _modifying_ the tree or tags (inserting or deleting
  98  *   items, setting or clearing tags) must exclude other modifications, and
  99  *   exclude any functions reading the tree.
 100  * - any function _reading_ the tree or tags (looking up items or tags,
 101  *   gang lookups) must exclude modifications to the tree, but may occur
 102  *   concurrently with other readers.
 103  *
 104  * The notable exceptions to this rule are the following functions:
 105  * radix_tree_lookup
 106  * radix_tree_lookup_slot
 107  * radix_tree_tag_get
 108  * radix_tree_gang_lookup
 109  * radix_tree_gang_lookup_slot
 110  * radix_tree_gang_lookup_tag
 111  * radix_tree_gang_lookup_tag_slot
 112  * radix_tree_tagged
 113  *
 114  * The first 7 functions are able to be called locklessly, using RCU. The
 115  * caller must ensure calls to these functions are made within rcu_read_lock()
 116  * regions. Other readers (lock-free or otherwise) and modifications may be
 117  * running concurrently.
 118  *
 119  * It is still required that the caller manage the synchronization and lifetimes
 120  * of the items. So if RCU lock-free lookups are used, typically this would mean
 121  * that the items have their own locks, or are amenable to lock-free access; and
 122  * that the items are freed by RCU (or only freed after having been deleted from
 123  * the radix tree *and* a synchronize_rcu() grace period).
 124  *
 125  * (Note, rcu_assign_pointer and rcu_dereference are not needed to control
 126  * access to data items when inserting into or looking up from the radix tree)
 127  *
 128  * Note that the value returned by radix_tree_tag_get() may not be relied upon
 129  * if only the RCU read lock is held.  Functions to set/clear tags and to
 130  * delete nodes running concurrently with it may affect its result such that
 131  * two consecutive reads in the same locked section may return different
 132  * values.  If reliability is required, modification functions must also be
 133  * excluded from concurrency.
 134  *
 135  * radix_tree_tagged is able to be called without locking or RCU.
 136  */
 137
 138 /**
 139  * radix_tree_deref_slot        - dereference a slot
 140  * @pslot:      pointer to slot, returned by radix_tree_lookup_slot
 141  * Returns:     item that was stored in that slot with any direct pointer flag
 142  *              removed.
 143  *
 144  * For use with radix_tree_lookup_slot().  Caller must hold tree at least read
 145  * locked across slot lookup and dereference. Not required if write lock is
 146  * held (ie. items cannot be concurrently inserted).
 147  *
 148  * radix_tree_deref_retry must be used to confirm validity of the pointer if
 149  * only the read lock is held.
 150  */
 151 static inline void *radix_tree_deref_slot(void **pslot)
 152 {
 153         return rcu_dereference(*pslot);
 154 }
 155
 156 /**
 157  * radix_tree_deref_slot_protected      - dereference a slot without RCU lock but with tree lock held
 158  * @pslot:      pointer to slot, returned by radix_tree_lookup_slot
 159  * Returns:     item that was stored in that slot with any direct pointer flag
 160  *              removed.
 161  *
 162  * Similar to radix_tree_deref_slot but only used during migration when a pages
 163  * mapping is being moved. The caller does not hold the RCU read lock but it
 164  * must hold the tree lock to prevent parallel updates.
 165  */
 166 static inline void *radix_tree_deref_slot_protected(void **pslot,
 167                                                         spinlock_t *treelock)
 168 {
 169         return rcu_dereference_protected(*pslot, lockdep_is_held(treelock));
 170 }
 171
 172 /**
 173  * radix_tree_deref_retry       - check radix_tree_deref_slot
 174  * @arg:        pointer returned by radix_tree_deref_slot
 175  * Returns:     0 if retry is not required, otherwise retry is required
 176  *
 177  * radix_tree_deref_retry must be used with radix_tree_deref_slot.
 178  */
 179 static inline int radix_tree_deref_retry(void *arg)
 180 {
 181         return unlikely((unsigned long)arg & RADIX_TREE_INDIRECT_PTR);
 182 }
 183
 184 /**
 185  * radix_tree_exceptional_entry - radix_tree_deref_slot gave exceptional entry?
 186  * @arg:        value returned by radix_tree_deref_slot
 187  * Returns:     0 if well-aligned pointer, non-0 if exceptional entry.
 188  */
 189 static inline int radix_tree_exceptional_entry(void *arg)
 190 {
 191         /* Not unlikely because radix_tree_exception often tested first */
 192         return (unsigned long)arg & RADIX_TREE_EXCEPTIONAL_ENTRY;
 193 }
 194
 195 /**
 196  * radix_tree_exception - radix_tree_deref_slot returned either exception?
 197  * @arg:        value returned by radix_tree_deref_slot
 198  * Returns:     0 if well-aligned pointer, non-0 if either kind of exception.
 199  */
 200 static inline int radix_tree_exception(void *arg)
 201 {
 202         return unlikely((unsigned long)arg &
 203                 (RADIX_TREE_INDIRECT_PTR | RADIX_TREE_EXCEPTIONAL_ENTRY));
 204 }
 205
 206 /**
 207  * radix_tree_replace_slot      - replace item in a slot
 208  * @pslot:      pointer to slot, returned by radix_tree_lookup_slot
 209  * @item:       new item to store in the slot.
 210  *
 211  * For use with radix_tree_lookup_slot().  Caller must hold tree write locked
 212  * across slot lookup and replacement.
 213  */
 214 static inline void radix_tree_replace_slot(void **pslot, void *item)
 215 {
 216         BUG_ON(radix_tree_is_indirect_ptr(item));
 217         rcu_assign_pointer(*pslot, item);
 218 }
 219
 220 int radix_tree_insert(struct radix_tree_root *, unsigned long, void *);
 221 void *radix_tree_lookup(struct radix_tree_root *, unsigned long);
 222 void **radix_tree_lookup_slot(struct radix_tree_root *, unsigned long);
 223 void *radix_tree_delete(struct radix_tree_root *, unsigned long);
 224 unsigned int
 225 radix_tree_gang_lookup(struct radix_tree_root *root, void **results,
 226                         unsigned long first_index, unsigned int max_items);
 227 unsigned int radix_tree_gang_lookup_slot(struct radix_tree_root *root,
 228                         void ***results, unsigned long *indices,
 229                         unsigned long first_index, unsigned int max_items);
 230 unsigned long radix_tree_next_hole(struct radix_tree_root *root,
 231                                 unsigned long index, unsigned long max_scan);
 232 unsigned long radix_tree_prev_hole(struct radix_tree_root *root,
 233                                 unsigned long index, unsigned long max_scan);
 234 int radix_tree_preload(gfp_t gfp_mask);
 235 void radix_tree_init(void);
 236 void *radix_tree_tag_set(struct radix_tree_root *root,
 237                         unsigned long index, unsigned int tag);
 238 void *radix_tree_tag_clear(struct radix_tree_root *root,
 239                         unsigned long index, unsigned int tag);
 240 int radix_tree_tag_get(struct radix_tree_root *root,
 241                         unsigned long index, unsigned int tag);
 242 unsigned int
 243 radix_tree_gang_lookup_tag(struct radix_tree_root *root, void **results,
 244                 unsigned long first_index, unsigned int max_items,
 245                 unsigned int tag);
 246 unsigned int
 247 radix_tree_gang_lookup_tag_slot(struct radix_tree_root *root, void ***results,
 248                 unsigned long first_index, unsigned int max_items,
 249                 unsigned int tag);
 250 unsigned long radix_tree_range_tag_if_tagged(struct radix_tree_root *root,
 251                 unsigned long *first_indexp, unsigned long last_index,
 252                 unsigned long nr_to_tag,
 253                 unsigned int fromtag, unsigned int totag);
 254 int radix_tree_tagged(struct radix_tree_root *root, unsigned int tag);
 255 unsigned long radix_tree_locate_item(struct radix_tree_root *root, void *item);
 256
 257 static inline void radix_tree_preload_end(void)
 258 {
 259         preempt_enable();
 260 }
 261
 262 #endif /* _LINUX_RADIX_TREE_H */