497 lines · plain
1.. SPDX-License-Identifier: GPL-2.0+2 3======4XArray5======6 7:Author: Matthew Wilcox8 9Overview10========11 12The XArray is an abstract data type which behaves like a very large array13of pointers. It meets many of the same needs as a hash or a conventional14resizable array. Unlike a hash, it allows you to sensibly go to the15next or previous entry in a cache-efficient manner. In contrast to a16resizable array, there is no need to copy data or change MMU mappings in17order to grow the array. It is more memory-efficient, parallelisable18and cache friendly than a doubly-linked list. It takes advantage of19RCU to perform lookups without locking.20 21The XArray implementation is efficient when the indices used are densely22clustered; hashing the object and using the hash as the index will not23perform well. The XArray is optimised for small indices, but still has24good performance with large indices. If your index can be larger than25``ULONG_MAX`` then the XArray is not the data type for you. The most26important user of the XArray is the page cache.27 28Normal pointers may be stored in the XArray directly. They must be 4-byte29aligned, which is true for any pointer returned from kmalloc() and30alloc_page(). It isn't true for arbitrary user-space pointers,31nor for function pointers. You can store pointers to statically allocated32objects, as long as those objects have an alignment of at least 4.33 34You can also store integers between 0 and ``LONG_MAX`` in the XArray.35You must first convert it into an entry using xa_mk_value().36When you retrieve an entry from the XArray, you can check whether it is37a value entry by calling xa_is_value(), and convert it back to38an integer by calling xa_to_value().39 40Some users want to tag the pointers they store in the XArray. You can41call xa_tag_pointer() to create an entry with a tag, xa_untag_pointer()42to turn a tagged entry back into an untagged pointer and xa_pointer_tag()43to retrieve the tag of an entry. Tagged pointers use the same bits that44are used to distinguish value entries from normal pointers, so you must45decide whether they want to store value entries or tagged pointers in46any particular XArray.47 48The XArray does not support storing IS_ERR() pointers as some49conflict with value entries or internal entries.50 51An unusual feature of the XArray is the ability to create entries which52occupy a range of indices. Once stored to, looking up any index in53the range will return the same entry as looking up any other index in54the range. Storing to any index will store to all of them. Multi-index55entries can be explicitly split into smaller entries, or storing ``NULL``56into any entry will cause the XArray to forget about the range.57 58Normal API59==========60 61Start by initialising an XArray, either with DEFINE_XARRAY()62for statically allocated XArrays or xa_init() for dynamically63allocated ones. A freshly-initialised XArray contains a ``NULL``64pointer at every index.65 66You can then set entries using xa_store() and get entries67using xa_load(). xa_store will overwrite any entry with the68new entry and return the previous entry stored at that index. You can69use xa_erase() instead of calling xa_store() with a70``NULL`` entry. There is no difference between an entry that has never71been stored to, one that has been erased and one that has most recently72had ``NULL`` stored to it.73 74You can conditionally replace an entry at an index by using75xa_cmpxchg(). Like cmpxchg(), it will only succeed if76the entry at that index has the 'old' value. It also returns the entry77which was at that index; if it returns the same entry which was passed as78'old', then xa_cmpxchg() succeeded.79 80If you want to only store a new entry to an index if the current entry81at that index is ``NULL``, you can use xa_insert() which82returns ``-EBUSY`` if the entry is not empty.83 84You can copy entries out of the XArray into a plain array by calling85xa_extract(). Or you can iterate over the present entries in the XArray86by calling xa_for_each(), xa_for_each_start() or xa_for_each_range().87You may prefer to use xa_find() or xa_find_after() to move to the next88present entry in the XArray.89 90Calling xa_store_range() stores the same entry in a range91of indices. If you do this, some of the other operations will behave92in a slightly odd way. For example, marking the entry at one index93may result in the entry being marked at some, but not all of the other94indices. Storing into one index may result in the entry retrieved by95some, but not all of the other indices changing.96 97Sometimes you need to ensure that a subsequent call to xa_store()98will not need to allocate memory. The xa_reserve() function99will store a reserved entry at the indicated index. Users of the100normal API will see this entry as containing ``NULL``. If you do101not need to use the reserved entry, you can call xa_release()102to remove the unused entry. If another user has stored to the entry103in the meantime, xa_release() will do nothing; if instead you104want the entry to become ``NULL``, you should use xa_erase().105Using xa_insert() on a reserved entry will fail.106 107If all entries in the array are ``NULL``, the xa_empty() function108will return ``true``.109 110Finally, you can remove all entries from an XArray by calling111xa_destroy(). If the XArray entries are pointers, you may wish112to free the entries first. You can do this by iterating over all present113entries in the XArray using the xa_for_each() iterator.114 115Search Marks116------------117 118Each entry in the array has three bits associated with it called marks.119Each mark may be set or cleared independently of the others. You can120iterate over marked entries by using the xa_for_each_marked() iterator.121 122You can enquire whether a mark is set on an entry by using123xa_get_mark(). If the entry is not ``NULL``, you can set a mark on it124by using xa_set_mark() and remove the mark from an entry by calling125xa_clear_mark(). You can ask whether any entry in the XArray has a126particular mark set by calling xa_marked(). Erasing an entry from the127XArray causes all marks associated with that entry to be cleared.128 129Setting or clearing a mark on any index of a multi-index entry will130affect all indices covered by that entry. Querying the mark on any131index will return the same result.132 133There is no way to iterate over entries which are not marked; the data134structure does not allow this to be implemented efficiently. There are135not currently iterators to search for logical combinations of bits (eg136iterate over all entries which have both ``XA_MARK_1`` and ``XA_MARK_2``137set, or iterate over all entries which have ``XA_MARK_0`` or ``XA_MARK_2``138set). It would be possible to add these if a user arises.139 140Allocating XArrays141------------------142 143If you use DEFINE_XARRAY_ALLOC() to define the XArray, or144initialise it by passing ``XA_FLAGS_ALLOC`` to xa_init_flags(),145the XArray changes to track whether entries are in use or not.146 147You can call xa_alloc() to store the entry at an unused index148in the XArray. If you need to modify the array from interrupt context,149you can use xa_alloc_bh() or xa_alloc_irq() to disable150interrupts while allocating the ID.151 152Using xa_store(), xa_cmpxchg() or xa_insert() will153also mark the entry as being allocated. Unlike a normal XArray, storing154``NULL`` will mark the entry as being in use, like xa_reserve().155To free an entry, use xa_erase() (or xa_release() if156you only want to free the entry if it's ``NULL``).157 158By default, the lowest free entry is allocated starting from 0. If you159want to allocate entries starting at 1, it is more efficient to use160DEFINE_XARRAY_ALLOC1() or ``XA_FLAGS_ALLOC1``. If you want to161allocate IDs up to a maximum, then wrap back around to the lowest free162ID, you can use xa_alloc_cyclic().163 164You cannot use ``XA_MARK_0`` with an allocating XArray as this mark165is used to track whether an entry is free or not. The other marks are166available for your use.167 168Memory allocation169-----------------170 171The xa_store(), xa_cmpxchg(), xa_alloc(),172xa_reserve() and xa_insert() functions take a gfp_t173parameter in case the XArray needs to allocate memory to store this entry.174If the entry is being deleted, no memory allocation needs to be performed,175and the GFP flags specified will be ignored.176 177It is possible for no memory to be allocatable, particularly if you pass178a restrictive set of GFP flags. In that case, the functions return a179special value which can be turned into an errno using xa_err().180If you don't need to know exactly which error occurred, using181xa_is_err() is slightly more efficient.182 183Locking184-------185 186When using the Normal API, you do not have to worry about locking.187The XArray uses RCU and an internal spinlock to synchronise access:188 189No lock needed:190 * xa_empty()191 * xa_marked()192 193Takes RCU read lock:194 * xa_load()195 * xa_for_each()196 * xa_for_each_start()197 * xa_for_each_range()198 * xa_find()199 * xa_find_after()200 * xa_extract()201 * xa_get_mark()202 203Takes xa_lock internally:204 * xa_store()205 * xa_store_bh()206 * xa_store_irq()207 * xa_insert()208 * xa_insert_bh()209 * xa_insert_irq()210 * xa_erase()211 * xa_erase_bh()212 * xa_erase_irq()213 * xa_cmpxchg()214 * xa_cmpxchg_bh()215 * xa_cmpxchg_irq()216 * xa_store_range()217 * xa_alloc()218 * xa_alloc_bh()219 * xa_alloc_irq()220 * xa_reserve()221 * xa_reserve_bh()222 * xa_reserve_irq()223 * xa_destroy()224 * xa_set_mark()225 * xa_clear_mark()226 227Assumes xa_lock held on entry:228 * __xa_store()229 * __xa_insert()230 * __xa_erase()231 * __xa_cmpxchg()232 * __xa_alloc()233 * __xa_set_mark()234 * __xa_clear_mark()235 236If you want to take advantage of the lock to protect the data structures237that you are storing in the XArray, you can call xa_lock()238before calling xa_load(), then take a reference count on the239object you have found before calling xa_unlock(). This will240prevent stores from removing the object from the array between looking241up the object and incrementing the refcount. You can also use RCU to242avoid dereferencing freed memory, but an explanation of that is beyond243the scope of this document.244 245The XArray does not disable interrupts or softirqs while modifying246the array. It is safe to read the XArray from interrupt or softirq247context as the RCU lock provides enough protection.248 249If, for example, you want to store entries in the XArray in process250context and then erase them in softirq context, you can do that this way::251 252 void foo_init(struct foo *foo)253 {254 xa_init_flags(&foo->array, XA_FLAGS_LOCK_BH);255 }256 257 int foo_store(struct foo *foo, unsigned long index, void *entry)258 {259 int err;260 261 xa_lock_bh(&foo->array);262 err = xa_err(__xa_store(&foo->array, index, entry, GFP_KERNEL));263 if (!err)264 foo->count++;265 xa_unlock_bh(&foo->array);266 return err;267 }268 269 /* foo_erase() is only called from softirq context */270 void foo_erase(struct foo *foo, unsigned long index)271 {272 xa_lock(&foo->array);273 __xa_erase(&foo->array, index);274 foo->count--;275 xa_unlock(&foo->array);276 }277 278If you are going to modify the XArray from interrupt or softirq context,279you need to initialise the array using xa_init_flags(), passing280``XA_FLAGS_LOCK_IRQ`` or ``XA_FLAGS_LOCK_BH``.281 282The above example also shows a common pattern of wanting to extend the283coverage of the xa_lock on the store side to protect some statistics284associated with the array.285 286Sharing the XArray with interrupt context is also possible, either287using xa_lock_irqsave() in both the interrupt handler and process288context, or xa_lock_irq() in process context and xa_lock()289in the interrupt handler. Some of the more common patterns have helper290functions such as xa_store_bh(), xa_store_irq(),291xa_erase_bh(), xa_erase_irq(), xa_cmpxchg_bh()292and xa_cmpxchg_irq().293 294Sometimes you need to protect access to the XArray with a mutex because295that lock sits above another mutex in the locking hierarchy. That does296not entitle you to use functions like __xa_erase() without taking297the xa_lock; the xa_lock is used for lockdep validation and will be used298for other purposes in the future.299 300The __xa_set_mark() and __xa_clear_mark() functions are also301available for situations where you look up an entry and want to atomically302set or clear a mark. It may be more efficient to use the advanced API303in this case, as it will save you from walking the tree twice.304 305Advanced API306============307 308The advanced API offers more flexibility and better performance at the309cost of an interface which can be harder to use and has fewer safeguards.310No locking is done for you by the advanced API, and you are required311to use the xa_lock while modifying the array. You can choose whether312to use the xa_lock or the RCU lock while doing read-only operations on313the array. You can mix advanced and normal operations on the same array;314indeed the normal API is implemented in terms of the advanced API. The315advanced API is only available to modules with a GPL-compatible license.316 317The advanced API is based around the xa_state. This is an opaque data318structure which you declare on the stack using the XA_STATE() macro.319This macro initialises the xa_state ready to start walking around the320XArray. It is used as a cursor to maintain the position in the XArray321and let you compose various operations together without having to restart322from the top every time. The contents of the xa_state are protected by323the rcu_read_lock() or the xas_lock(). If you need to drop whichever of324those locks is protecting your state and tree, you must call xas_pause()325so that future calls do not rely on the parts of the state which were326left unprotected.327 328The xa_state is also used to store errors. You can call329xas_error() to retrieve the error. All operations check whether330the xa_state is in an error state before proceeding, so there's no need331for you to check for an error after each call; you can make multiple332calls in succession and only check at a convenient point. The only333errors currently generated by the XArray code itself are ``ENOMEM`` and334``EINVAL``, but it supports arbitrary errors in case you want to call335xas_set_err() yourself.336 337If the xa_state is holding an ``ENOMEM`` error, calling xas_nomem()338will attempt to allocate more memory using the specified gfp flags and339cache it in the xa_state for the next attempt. The idea is that you take340the xa_lock, attempt the operation and drop the lock. The operation341attempts to allocate memory while holding the lock, but it is more342likely to fail. Once you have dropped the lock, xas_nomem()343can try harder to allocate more memory. It will return ``true`` if it344is worth retrying the operation (i.e. that there was a memory error *and*345more memory was allocated). If it has previously allocated memory, and346that memory wasn't used, and there is no error (or some error that isn't347``ENOMEM``), then it will free the memory previously allocated.348 349Internal Entries350----------------351 352The XArray reserves some entries for its own purposes. These are never353exposed through the normal API, but when using the advanced API, it's354possible to see them. Usually the best way to handle them is to pass them355to xas_retry(), and retry the operation if it returns ``true``.356 357.. flat-table::358 :widths: 1 1 6359 360 * - Name361 - Test362 - Usage363 364 * - Node365 - xa_is_node()366 - An XArray node. May be visible when using a multi-index xa_state.367 368 * - Sibling369 - xa_is_sibling()370 - A non-canonical entry for a multi-index entry. The value indicates371 which slot in this node has the canonical entry.372 373 * - Retry374 - xa_is_retry()375 - This entry is currently being modified by a thread which has the376 xa_lock. The node containing this entry may be freed at the end377 of this RCU period. You should restart the lookup from the head378 of the array.379 380 * - Zero381 - xa_is_zero()382 - Zero entries appear as ``NULL`` through the Normal API, but occupy383 an entry in the XArray which can be used to reserve the index for384 future use. This is used by allocating XArrays for allocated entries385 which are ``NULL``.386 387Other internal entries may be added in the future. As far as possible, they388will be handled by xas_retry().389 390Additional functionality391------------------------392 393The xas_create_range() function allocates all the necessary memory394to store every entry in a range. It will set ENOMEM in the xa_state if395it cannot allocate memory.396 397You can use xas_init_marks() to reset the marks on an entry398to their default state. This is usually all marks clear, unless the399XArray is marked with ``XA_FLAGS_TRACK_FREE``, in which case mark 0 is set400and all other marks are clear. Replacing one entry with another using401xas_store() will not reset the marks on that entry; if you want402the marks reset, you should do that explicitly.403 404The xas_load() will walk the xa_state as close to the entry405as it can. If you know the xa_state has already been walked to the406entry and need to check that the entry hasn't changed, you can use407xas_reload() to save a function call.408 409If you need to move to a different index in the XArray, call410xas_set(). This resets the cursor to the top of the tree, which411will generally make the next operation walk the cursor to the desired412spot in the tree. If you want to move to the next or previous index,413call xas_next() or xas_prev(). Setting the index does414not walk the cursor around the array so does not require a lock to be415held, while moving to the next or previous index does.416 417You can search for the next present entry using xas_find(). This418is the equivalent of both xa_find() and xa_find_after();419if the cursor has been walked to an entry, then it will find the next420entry after the one currently referenced. If not, it will return the421entry at the index of the xa_state. Using xas_next_entry() to422move to the next present entry instead of xas_find() will save423a function call in the majority of cases at the expense of emitting more424inline code.425 426The xas_find_marked() function is similar. If the xa_state has427not been walked, it will return the entry at the index of the xa_state,428if it is marked. Otherwise, it will return the first marked entry after429the entry referenced by the xa_state. The xas_next_marked()430function is the equivalent of xas_next_entry().431 432When iterating over a range of the XArray using xas_for_each()433or xas_for_each_marked(), it may be necessary to temporarily stop434the iteration. The xas_pause() function exists for this purpose.435After you have done the necessary work and wish to resume, the xa_state436is in an appropriate state to continue the iteration after the entry437you last processed. If you have interrupts disabled while iterating,438then it is good manners to pause the iteration and reenable interrupts439every ``XA_CHECK_SCHED`` entries.440 441The xas_get_mark(), xas_set_mark() and xas_clear_mark() functions require442the xa_state cursor to have been moved to the appropriate location in the443XArray; they will do nothing if you have called xas_pause() or xas_set()444immediately before.445 446You can call xas_set_update() to have a callback function447called each time the XArray updates a node. This is used by the page448cache workingset code to maintain its list of nodes which contain only449shadow entries.450 451Multi-Index Entries452-------------------453 454The XArray has the ability to tie multiple indices together so that455operations on one index affect all indices. For example, storing into456any index will change the value of the entry retrieved from any index.457Setting or clearing a mark on any index will set or clear the mark458on every index that is tied together. The current implementation459only allows tying ranges which are aligned powers of two together;460eg indices 64-127 may be tied together, but 2-6 may not be. This may461save substantial quantities of memory; for example tying 512 entries462together will save over 4kB.463 464You can create a multi-index entry by using XA_STATE_ORDER()465or xas_set_order() followed by a call to xas_store().466Calling xas_load() with a multi-index xa_state will walk the467xa_state to the right location in the tree, but the return value is not468meaningful, potentially being an internal entry or ``NULL`` even when there469is an entry stored within the range. Calling xas_find_conflict()470will return the first entry within the range or ``NULL`` if there are no471entries in the range. The xas_for_each_conflict() iterator will472iterate over every entry which overlaps the specified range.473 474If xas_load() encounters a multi-index entry, the xa_index475in the xa_state will not be changed. When iterating over an XArray476or calling xas_find(), if the initial index is in the middle477of a multi-index entry, it will not be altered. Subsequent calls478or iterations will move the index to the first index in the range.479Each entry will only be returned once, no matter how many indices it480occupies.481 482Using xas_next() or xas_prev() with a multi-index xa_state is not483supported. Using either of these functions on a multi-index entry will484reveal sibling entries; these should be skipped over by the caller.485 486Storing ``NULL`` into any index of a multi-index entry will set the487entry at every index to ``NULL`` and dissolve the tie. A multi-index488entry can be split into entries occupying smaller ranges by calling489xas_split_alloc() without the xa_lock held, followed by taking the lock490and calling xas_split().491 492Functions and structures493========================494 495.. kernel-doc:: include/linux/xarray.h496.. kernel-doc:: lib/xarray.c497