src/pcl/cache.lisp

   1 ;;;; the basics of the PCL wrapper cache mechanism
   2
   3 ;;;; This software is part of the SBCL system. See the README file for
   4 ;;;; more information.
   5
   6 ;;;; This software is derived from software originally released by Xerox
   7 ;;;; Corporation. Copyright and release statements follow. Later modifications
   8 ;;;; to the software are in the public domain and are provided with
   9 ;;;; absolutely no warranty. See the COPYING and CREDITS files for more
  10 ;;;; information.
  11
  12 ;;;; copyright information from original PCL sources:
  13 ;;;;
  14 ;;;; Copyright (c) 1985, 1986, 1987, 1988, 1989, 1990 Xerox Corporation.
  15 ;;;; All rights reserved.
  16 ;;;;
  17 ;;;; Use and copying of this software and preparation of derivative works based
  18 ;;;; upon this software are permitted. Any distribution of this software or
  19 ;;;; derivative works must comply with all applicable United States export
  20 ;;;; control laws.
  21 ;;;;
  22 ;;;; This software is made available AS IS, and Xerox Corporation makes no
  23 ;;;; warranty about the software, its performance or its conformity to any
  24 ;;;; specification.
  25
  26 ;;;; Note: as of SBCL 1.0.6.3 it is questionable if cache.lisp can
  27 ;;;; anymore be considered to be "derived from software originally
  28 ;;;; released by Xerox Corporation", as at that time the whole cache
  29 ;;;; implementation was essentially redone from scratch.
  30
  31 (in-package "SB-PCL")
  32
  33 ;;;; Public API:
  34 ;;;;
  35 ;;;;   fill-cache
  36 ;;;;   probe-cache
  37 ;;;;   make-cache
  38 ;;;;   map-cache
  39 ;;;;   emit-cache-lookup
  40 ;;;;   copy-cache
  41 ;;;;   hash-table-to-cache
  42 ;;;;
  43 ;;;; This is a thread and interrupt safe reimplementation loosely
  44 ;;;; based on the original PCL cache by Kickzales and Rodrigues,
  45 ;;;; as described in "Efficient Method Dispatch in PCL".
  46 ;;;;
  47 ;;;; * Writes to cache are made atomic using compare-and-swap on
  48 ;;;;   wrappers. Wrappers are never moved or deleted after they have
  49 ;;;;   been written: to clean them out the cache need to be copied.
  50 ;;;;
  51 ;;;; * Copying or expanding the cache drops out incomplete and invalid
  52 ;;;;   lines.
  53 ;;;;
  54 ;;;; * Since the cache is used for memoization only we don't need to
  55 ;;;;   worry about which of simultaneous replacements (when expanding
  56 ;;;;   the cache) takes place: the loosing one will have its work
  57 ;;;;   redone later. This also allows us to drop entries when the
  58 ;;;;   cache is about to grow insanely huge.
  59 ;;;;
  60 ;;;; The cache is essentially a specialized hash-table for layouts, used
  61 ;;;; for memoization of effective methods, slot locations, and constant
  62 ;;;; return values.
  63 ;;;;
  64 ;;;; Subsequences of the cache vector are called cache lines.
  65 ;;;;
  66 ;;;; The cache vector uses the symbol SB-PCL::..EMPTY.. as a sentinel
  67 ;;;; value, to allow storing NILs in the vector as well.
  68
  69 (defstruct (cache (:constructor %make-cache)
  70                   (:copier %copy-cache))
  71   ;; Number of keys the cache uses.
  72   (key-count 1 :type (integer 1 (#.call-arguments-limit)))
  73   ;; True if we store values in the cache.
  74   (value)
  75   ;; Number of vector elements a single cache line uses in the vector.
  76   ;; This is always a power of two, so that the vector length can be both
  77   ;; an exact multiple of this and a power of two.
  78   (line-size 1 :type (integer 1 #.most-positive-fixnum))
  79   ;; Cache vector, its length is always both a multiple of line-size
  80   ;; and a power of two. This is so that we can calculate
  81   ;;   (mod index (length vector))
  82   ;; using a bitmask.
  83   (vector #() :type simple-vector)
  84   ;; The bitmask used to calculate
  85   ;;   (mod (* line-size line-hash) (length vector))).
  86   (mask 0 :type fixnum)
  87   ;; Current probe-depth needed in the cache.
  88   (depth 0 :type index)
  89   ;; Maximum allowed probe-depth before the cache needs to expand.
  90   (limit 0 :type index))
  91
  92 (defun compute-cache-mask (vector-length line-size)
  93   ;; Since both vector-length and line-size are powers of two, we
  94   ;; can compute a bitmask such that
  95   ;;
  96   ;;  (logand <mask> <combined-layout-hash>)
  97   ;;
  98   ;; is "morally equal" to
  99   ;;
 100   ;;  (mod (* <line-size> <combined-layout-hash>) <vector-length>)
 101   ;;
 102   ;; This is it: (1- vector-length) is #b111... of the approriate size
 103   ;; to get the MOD, and (- line-size) gives right the number of zero
 104   ;; bits at the low end.
 105   (logand (1- vector-length) (- line-size)))
 106
 107 ;;; The smallest power of two that is equal to or greater then X.
 108 (declaim (inline power-of-two-ceiling))
 109 (defun power-of-two-ceiling (x)
 110   (ash 1 (integer-length (1- x))))
 111
 112 (defun cache-statistics (cache)
 113   (let* ((vector (cache-vector cache))
 114          (size (length vector))
 115          (line-size (cache-line-size cache))
 116          (total-lines (/ size line-size))
 117          (free-lines (loop for i from 0 by line-size below size
 118                            unless (eq (svref vector i) '..empty..)
 119                            count t)))
 120     (values (- total-lines free-lines) total-lines
 121             (cache-depth cache) (cache-limit cache))))
 122
 123 ;;; Don't allocate insanely huge caches: this is 4096 lines for a
 124 ;;; value cache with 8-15 keys -- probably "big enough for anyone",
 125 ;;; and 16384 lines for a commonplace 2-key value cache.
 126 (defconstant +cache-vector-max-length+ (expt 2 16))
 127
 128 ;;; Compute the maximum allowed probe depth as a function of cache size.
 129 ;;; Cache size refers to number of cache lines, not the length of the
 130 ;;; cache vector.
 131 ;;;
 132 ;;; FIXME: It would be nice to take the generic function optimization
 133 ;;; policy into account here (speed vs. space.)
 134 (declaim (inline compute-limit))
 135 (defun compute-limit (size)
 136   (ceiling (sqrt (sqrt size))))
 137
 138 ;;; Returns VALUE if it is not ..EMPTY.., otherwise executes ELSE:
 139 (defmacro non-empty-or (value else)
 140   (with-unique-names (n-value)
 141     `(let ((,n-value ,value))
 142        (if (eq ,n-value '..empty..)
 143            ,else
 144            ,n-value))))
 145
 146 ;;; Fast way to check if a thing found at the position of a cache key is one:
 147 ;;; it is always either a wrapper, or the ..EMPTY.. symbol.
 148 (declaim (inline cache-key-p))
 149 (defun cache-key-p (thing)
 150   (not (symbolp thing)))
 151
 152 ;;; Atomically update the current probe depth of a cache.
 153 (defun note-cache-depth (cache depth)
 154   (loop for old = (cache-depth cache)
 155         while (and (< old depth)
 156                    (not (eq old (compare-and-swap (cache-depth cache)
 157                                                   old depth))))))
 158
 159 ;;; Compute the starting index of the next cache line in the cache vector.
 160 (declaim (inline next-cache-index))
 161 (defun next-cache-index (mask index line-size)
 162   (logand mask (+ index line-size)))
 163
 164 ;;; Returns the hash-value for layout, or executes ELSE if the layout
 165 ;;; is invalid.
 166 (defmacro hash-layout-or (layout else)
 167   (with-unique-names (n-hash)
 168     `(let ((,n-hash (layout-clos-hash ,layout)))
 169        (if (zerop ,n-hash)
 170            ,else
 171            ,n-hash))))
 172
 173 ;;; Compute cache index for the cache and a list of layouts.
 174 (declaim (inline compute-cache-index))
 175 (defun compute-cache-index (cache layouts)
 176   (let ((index (hash-layout-or (car layouts)
 177                                (return-from compute-cache-index nil))))
 178     (declare (fixnum index))
 179     (dolist (layout (cdr layouts))
 180       (mixf index (hash-layout-or layout (return-from compute-cache-index nil))))
 181     ;; align with cache lines
 182     (logand index (cache-mask cache))))
 183
 184 ;;; Emit code that does lookup in cache bound to CACHE-VAR using
 185 ;;; layouts bound to LAYOUT-VARS. Go to MISS-TAG on event of a miss or
 186 ;;; invalid layout. Otherwise, if VALUE-VAR is non-nil, set it to the
 187 ;;; value found. (VALUE-VAR is non-nil only when CACHE-VALUE is true.)
 188 ;;;
 189 ;;; In other words, produces inlined code for COMPUTE-CACHE-INDEX when
 190 ;;; number of keys and presence of values in the cache is known
 191 ;;; beforehand.
 192 (defun emit-cache-lookup (cache-var layout-vars miss-tag value-var)
 193   (let ((line-size (power-of-two-ceiling (+ (length layout-vars)
 194                                             (if value-var 1 0)))))
 195     (with-unique-names (n-index n-vector n-depth n-pointer n-mask
 196                        MATCH-WRAPPERS EXIT-WITH-HIT)
 197       `(let* ((,n-index (hash-layout-or ,(car layout-vars) (go ,miss-tag)))
 198               (,n-vector (cache-vector ,cache-var))
 199               (,n-mask (cache-mask ,cache-var)))
 200          (declare (index ,n-index))
 201          ,@(mapcar (lambda (layout-var)
 202                      `(mixf ,n-index (hash-layout-or ,layout-var (go ,miss-tag))))
 203                    (cdr layout-vars))
 204          ;; align with cache lines
 205          (setf ,n-index (logand ,n-index ,n-mask))
 206          (let ((,n-depth (cache-depth ,cache-var))
 207                (,n-pointer ,n-index))
 208            (declare (index ,n-depth ,n-pointer))
 209            (tagbody
 210             ,MATCH-WRAPPERS
 211               (when (and ,@(mapcar
 212                             (lambda (layout-var)
 213                               `(prog1
 214                                    (eq ,layout-var (svref ,n-vector ,n-pointer))
 215                                  (incf ,n-pointer)))
 216                             layout-vars))
 217                 ,@(when value-var
 218                     `((setf ,value-var (non-empty-or (svref ,n-vector ,n-pointer)
 219                                                      (go ,miss-tag)))))
 220                 (go ,EXIT-WITH-HIT))
 221               (if (zerop ,n-depth)
 222                   (go ,miss-tag)
 223                   (decf ,n-depth))
 224               (setf ,n-index (next-cache-index ,n-mask ,n-index ,line-size)
 225                     ,n-pointer ,n-index)
 226               (go ,MATCH-WRAPPERS)
 227             ,EXIT-WITH-HIT))))))
 228
 229 ;;; Probes CACHE for LAYOUTS.
 230 ;;;
 231 ;;; Returns two values: a boolean indicating a hit or a miss, and a secondary
 232 ;;; value that is the value that was stored in the cache if any.
 233 (defun probe-cache (cache layouts)
 234   (unless (consp layouts)
 235     (setf layouts (list layouts)))
 236   (let ((vector (cache-vector cache))
 237         (key-count (cache-key-count cache))
 238         (line-size (cache-line-size cache))
 239         (mask (cache-mask cache)))
 240     (flet ((probe-line (base)
 241              (tagbody
 242                 (loop for offset from 0 below key-count
 243                       for layout in layouts do
 244                       (unless (eq layout (svref vector (+ base offset)))
 245                         ;; missed
 246                         (go :miss)))
 247                 ;; all layouts match!
 248                 (let ((value (when (cache-value cache)
 249                                (non-empty-or (svref vector (+ base key-count))
 250                                              (go :miss)))))
 251                   (return-from probe-cache (values t value)))
 252               :miss
 253                 (return-from probe-line (next-cache-index mask base line-size)))))
 254       (let ((index (compute-cache-index cache layouts)))
 255         (when index
 256           (loop repeat (1+ (cache-depth cache)) do
 257                 (setf index (probe-line index)))))))
 258   (values nil nil))
 259
 260 ;;; Tries to write LAYOUTS and VALUE at the cache line starting at
 261 ;;; the index BASE. Returns true on success, and false on failure.
 262 (defun try-update-cache-line (cache base layouts value)
 263   (declare (index base))
 264   (let ((vector (cache-vector cache))
 265         (new (pop layouts)))
 266     ;; If we unwind from here, we will be left with an incomplete
 267     ;; cache line, but that is OK: next write using the same layouts
 268     ;; will fill it, and reads will treat an incomplete line as a
 269     ;; miss -- causing it to be filled.
 270     (loop for old = (compare-and-swap (svref vector base) '..empty.. new)  do
 271           (when (and (cache-key-p old) (not (eq old new)))
 272             ;; The place was already taken, and doesn't match our key.
 273             (return-from try-update-cache-line nil))
 274           (unless layouts
 275             ;; All keys match or succesfully saved, save our value --
 276             ;; just smash it in. Until the first time it is written
 277             ;; there is ..EMPTY.. here, which probes look for, so we
 278             ;; don't get bogus hits. This is necessary because we want
 279             ;; to be able store arbitrary values here for use with
 280             ;; constant-value dispatch functions.
 281             (when (cache-value cache)
 282               (setf (svref vector (1+ base)) value))
 283             (return-from try-update-cache-line t))
 284           (setf new (pop layouts))
 285           (incf base))))
 286
 287 ;;; Tries to write LAYOUTS and VALUE somewhere in the cache. Returns
 288 ;;; true on success and false on failure, meaning the cache is too
 289 ;;; full.
 290 (defun try-update-cache (cache layouts value)
 291   (let ((vector (cache-vector cache))
 292         (index (or (compute-cache-index cache layouts)
 293                    ;; At least one of the layouts was invalid: just
 294                    ;; pretend we updated the cache, and let the next
 295                    ;; read pick up the mess.
 296                    (return-from try-update-cache t)))
 297         (line-size (cache-line-size cache))
 298         (mask (cache-mask cache)))
 299     (declare (index index))
 300     (loop for depth from 0 upto (cache-limit cache) do
 301           (when (try-update-cache-line cache index layouts value)
 302             (note-cache-depth cache depth)
 303             (return-from try-update-cache t))
 304           (setf index (next-cache-index mask index line-size)))))
 305
 306 ;;; Constructs a new cache.
 307 (defun make-cache (&key (key-count (missing-arg)) (value (missing-arg))
 308                    (size 1))
 309   (let* ((line-size (power-of-two-ceiling (+ key-count (if value 1 0))))
 310          (adjusted-size (power-of-two-ceiling size))
 311          (length (* adjusted-size line-size)))
 312     (if (<= length +cache-vector-max-length+)
 313         (%make-cache :key-count key-count
 314                      :line-size line-size
 315                      :vector (make-array length :initial-element '..empty..)
 316                      :value value
 317                      :mask (compute-cache-mask length line-size)
 318                      :limit (compute-limit adjusted-size))
 319         ;; Make a smaller one, then
 320         (make-cache :key-count key-count :value value :size (ceiling size 2)))))
 321
 322 ;;;; Copies and expands the cache, dropping any invalidated or
 323 ;;;; incomplete lines.
 324 (defun copy-and-expand-cache (cache layouts value)
 325   (let ((copy (%copy-cache cache))
 326         (length (length (cache-vector cache))))
 327     (when (< length +cache-vector-max-length+)
 328       (setf length (* 2 length)))
 329     (tagbody
 330      :again
 331        ;; Blow way the old vector first, so a GC potentially triggered by
 332        ;; MAKE-ARRAY can collect it.
 333        (setf (cache-vector copy) #()
 334              (cache-vector copy) (make-array length :initial-element '..empty..)
 335              (cache-depth copy) 0
 336              (cache-mask copy) (compute-cache-mask length (cache-line-size cache))
 337              (cache-limit copy) (compute-limit (/ length (cache-line-size cache))))
 338        ;; First insert the new one -- if we don't do this first and
 339        ;; the cache has reached it's maximum size we may end up
 340        ;; looping in FILL-CACHE.
 341        (unless (try-update-cache copy layouts value)
 342          (bug "Could not insert ~S:~S to supposedly empty ~S." layouts value cache))
 343        (map-cache (lambda (layouts value)
 344                     (unless (try-update-cache copy layouts value)
 345                       ;; If the cache would grow too much we drop the
 346                       ;; remaining the entries that don't fit. FIXME:
 347                       ;; It would be better to drop random entries to
 348                       ;; avoid getting into a rut here (best done by
 349                       ;; making MAP-CACHE map in a random order?), and
 350                       ;; possibly to downsize the cache more
 351                       ;; aggressively (on the assumption that most
 352                       ;; entries aren't getting used at the moment.)
 353                       (when (< length +cache-vector-max-length+)
 354                         (setf length (* 2 length))
 355                         (go :again))))
 356                   cache))
 357     copy))
 358
 359 (defun cache-has-invalid-entries-p (cache)
 360   (let ((vector (cache-vector cache))
 361         (line-size (cache-line-size cache))
 362         (key-count (cache-key-count cache))
 363         (mask (cache-mask cache))
 364         (index 0))
 365     (loop
 366       ;; Check if the line is in use, and check validity of the keys.
 367       (let ((key1 (svref vector index)))
 368         (when (cache-key-p key1)
 369           (if (zerop (layout-clos-hash key1))
 370               ;; First key invalid.
 371               (return-from cache-has-invalid-entries-p t)
 372               ;; Line is in use and the first key is valid: check the rest.
 373               (loop for offset from 1 below key-count
 374                     do (let ((thing (svref vector (+ index offset))))
 375                          (when (or (not (cache-key-p thing))
 376                                    (zerop (layout-clos-hash thing)))
 377                            ;; Incomplete line or invalid layout.
 378                            (return-from cache-has-invalid-entries-p t)))))))
 379       ;; Line empty of valid, onwards.
 380       (setf index (next-cache-index mask index line-size))
 381       (when (zerop index)
 382         ;; wrapped around
 383         (return-from cache-has-invalid-entries-p nil)))))
 384
 385 (defun hash-table-to-cache (table &key value key-count)
 386   (let ((cache (make-cache :key-count key-count :value value
 387                            :size (hash-table-count table))))
 388     (maphash (lambda (class value)
 389                (setq cache (fill-cache cache (class-wrapper class) value)))
 390              table)
 391     cache))
 392
 393 ;;; Inserts VALUE to CACHE keyd by LAYOUTS. Expands the cache if
 394 ;;; necessary, and returns the new cache.
 395 (defun fill-cache (cache layouts value)
 396   (labels
 397       ((%fill-cache (cache layouts value)
 398          (cond ((try-update-cache cache layouts value)
 399                 cache)
 400                ((cache-has-invalid-entries-p cache)
 401                 ;; Don't expand yet: maybe there will be enough space if
 402                 ;; we just drop the invalid entries.
 403                 (%fill-cache (copy-cache cache) layouts value))
 404                (t
 405                 (copy-and-expand-cache cache layouts value)))))
 406     (if (listp layouts)
 407         (%fill-cache cache layouts value)
 408         (%fill-cache cache (list layouts) value))))
 409
 410 ;;; Calls FUNCTION with all layouts and values in cache.
 411 (defun map-cache (function cache)
 412   (let* ((vector (cache-vector cache))
 413          (key-count (cache-key-count cache))
 414          (valuep (cache-value cache))
 415          (line-size (cache-line-size cache))
 416          (mask (cache-mask cache))
 417          (fun (if (functionp function)
 418                   function
 419                   (fdefinition function)))
 420          (index 0)
 421          (key nil))
 422     (tagbody
 423      :map
 424        (let ((layouts
 425               (loop for offset from 0 below key-count
 426                     collect (non-empty-or (svref vector (+ offset index))
 427                                           (go :next)))))
 428          (let ((value (when valuep
 429                         (non-empty-or (svref vector (+ index key-count))
 430                                       (go :next)))))
 431            ;; Let the callee worry about invalid layouts
 432            (funcall fun layouts value)))
 433      :next
 434        (setf index (next-cache-index mask index line-size))
 435        (unless (zerop index)
 436          (go :map))))
 437   cache)
 438
 439 ;;; Copying a cache without expanding it is very much like mapping it:
 440 ;;; we need to be carefull because there may be updates while we are
 441 ;;; copying it, and we don't want to copy incomplete entries or invalid
 442 ;;; ones.
 443 (defun copy-cache (cache)
 444   (let* ((vector (cache-vector cache))
 445          (copy (make-array (length vector) :initial-element '..empty..))
 446          (line-size (cache-line-size cache))
 447          (key-count (cache-key-count cache))
 448          (valuep (cache-value cache))
 449          (mask (cache-mask cache))
 450          (size (/ (length vector) line-size))
 451          (index 0)
 452          (elt nil)
 453          (depth 0))
 454     (tagbody
 455      :copy
 456        (let ((layouts (loop for offset from 0 below key-count
 457                             collect (non-empty-or (svref vector (+ index offset))
 458                                                   (go :next)))))
 459          ;; Check validity & compute primary index.
 460          (let ((primary (or (compute-cache-index cache layouts)
 461                             (go :next))))
 462            ;; Check & copy value.
 463            (when valuep
 464              (setf (svref copy (+ index key-count))
 465                    (non-empty-or (svref vector (+ index key-count))
 466                                  (go :next))))
 467            ;; Copy layouts.
 468            (loop for offset from 0 below key-count do
 469                  (setf (svref copy (+ index offset)) (pop layouts)))
 470            ;; Update probe depth.
 471            (let ((distance (/ (- index primary) line-size)))
 472              (setf depth (max depth (if (minusp distance)
 473                                         ;; account for wrap-around
 474                                         (+ distance size)
 475                                         distance))))))
 476      :next
 477        (setf index (next-cache-index mask index line-size))
 478        (unless (zerop index)
 479          (go :copy)))
 480     (%make-cache :vector copy
 481                  :depth depth
 482                  :key-count (cache-key-count cache)
 483                  :line-size line-size
 484                  :value valuep
 485                  :mask mask
 486                  :limit (cache-limit cache))))
 487
 488 ;;;; For debugging & collecting statistics.
 489
 490 (defun map-all-caches (function)
 491   (dolist (p (list-all-packages))
 492     (do-symbols (s p)
 493       (when (eq p (symbol-package s))
 494         (dolist (name (list s
 495                             `(setf ,s)
 496                             (slot-reader-name s)
 497                             (slot-writer-name s)
 498                             (slot-boundp-name s)))
 499           (when (fboundp name)
 500             (let ((fun (fdefinition name)))
 501               (when (typep fun 'generic-function)
 502                 (let ((cache (gf-dfun-cache fun)))
 503                   (when cache
 504                     (funcall function name cache)))))))))))
 505
 506 (defun check-cache-consistency (cache)
 507   (let ((table (make-hash-table :test 'equal)))
 508     (map-cache (lambda (layouts value)
 509                  (declare (ignore value))
 510                  (if (gethash layouts table)
 511                      (cerror "Check futher."
 512                              "Multiple appearances of ~S." layouts)
 513                      (setf (gethash layouts table) t)))
 514                cache)))