2 ( ProcPointSet, Status(..)
3 , callProcPoints, minimalProcPointSet
4 , addProcPointProtocols, splitAtProcPoints, procPointAnalysis
8 import qualified Prelude as P
9 import Prelude hiding (zip, unzip, last)
13 import Cmm hiding (blockId)
24 import MkZipCfgCmm hiding (CmmBlock, CmmGraph, CmmTopZ)
34 -- Compute a minimal set of proc points for a control-flow graph.
36 -- Determine a protocol for each proc point (which live variables will
37 -- be passed as arguments and which will be on the stack).
40 A proc point is a basic block that, after CPS transformation, will
41 start a new function. The entry block of the original function is a
42 proc point, as is the continuation of each function call.
43 A third kind of proc point arises if we want to avoid copying code.
44 Suppose we have code like the following:
47 if (...) { ..1..; call foo(); ..2..}
48 else { ..3..; call bar(); ..4..}
53 The statement 'x = y + z' can be reached from two different proc
54 points: the continuations of foo() and bar(). We would prefer not to
55 put a copy in each continuation; instead we would like 'x = y + z' to
56 be the start of a new procedure to which the continuations can jump:
59 if (...) { ..1..; push k_foo; jump foo_cps(); }
60 else { ..3..; push k_bar; jump bar_cps(); }
62 k_foo() { ..2..; jump k_join(y, z); }
63 k_bar() { ..4..; jump k_join(y, z); }
64 k_join(y, z) { x = y + z; return x; }
66 You might think then that a criterion to make a node a proc point is
67 that it is directly reached by two distinct proc points. (Note
68 [Direct reachability].) But this criterion is a bit too simple; for
69 example, 'return x' is also reached by two proc points, yet there is
70 no point in pulling it out of k_join. A good criterion would be to
71 say that a node should be made a proc point if it is reached by a set
72 of proc points that is different than its immediate dominator. NR
73 believes this criterion can be shown to produce a minimum set of proc
74 points, and given a dominator tree, the proc points can be chosen in
75 time linear in the number of blocks. Lacking a dominator analysis,
76 however, we turn instead to an iterative solution, starting with no
77 proc points and adding them according to these rules:
79 1. The entry block is a proc point.
80 2. The continuation of a call is a proc point.
81 3. A node is a proc point if it is directly reached by more proc
82 points than one of its predecessors.
84 Because we don't understand the problem very well, we apply rule 3 at
85 most once per iteration, then recompute the reachability information.
86 (See Note [No simple dataflow].) The choice of the new proc point is
87 arbitrary, and I don't know if the choice affects the final solution,
88 so I don't know if the number of proc points chosen is the
89 minimum---but the set will be minimal.
92 type ProcPointSet = BlockSet
95 = ReachedBy ProcPointSet -- set of proc points that directly reach the block
96 | ProcPoint -- this block is itself a proc point
98 instance Outputable Status where
100 | isEmptyBlockSet ps = text "<not-reached>"
101 | otherwise = text "reached by" <+>
102 (hsep $ punctuate comma $ map ppr $ blockSetToList ps)
103 ppr ProcPoint = text "<procpt>"
106 lattice :: DataflowLattice Status
107 lattice = DataflowLattice "direct proc-point reachability" unreached add_to False
108 where unreached = ReachedBy emptyBlockSet
109 add_to _ ProcPoint = noTx ProcPoint
110 add_to ProcPoint _ = aTx ProcPoint -- aTx because of previous case again
111 add_to (ReachedBy p) (ReachedBy p') =
112 let union = unionBlockSets p p'
113 in if sizeBlockSet union > sizeBlockSet p' then
114 aTx (ReachedBy union)
117 --------------------------------------------------
118 -- transfer equations
120 forward :: ForwardTransfers Middle Last Status
121 forward = ForwardTransfers first middle last exit
122 where first id ProcPoint = ReachedBy $ unitBlockSet id
125 last (LastCall _ (Just id) _ _ _) _ = LastOutFacts [(id, ProcPoint)]
126 last l x = LastOutFacts $ map (\id -> (id, x)) (succs l)
129 -- It is worth distinguishing two sets of proc points:
130 -- those that are induced by calls in the original graph
131 -- and those that are introduced because they're reachable from multiple proc points.
132 callProcPoints :: CmmGraph -> ProcPointSet
133 minimalProcPointSet :: ProcPointSet -> CmmGraph -> FuelMonad ProcPointSet
135 callProcPoints g = fold_blocks add (unitBlockSet (lg_entry g)) g
136 where add b set = case last $ unzip b of
137 LastOther (LastCall _ (Just k) _ _ _) -> extendBlockSet set k
140 minimalProcPointSet callProcPoints g = extendPPSet g (postorder_dfs g) callProcPoints
142 type PPFix = FuelMonad (ForwardFixedPoint Middle Last Status ())
144 procPointAnalysis :: ProcPointSet -> CmmGraph -> FuelMonad (BlockEnv Status)
145 procPointAnalysis procPoints g =
146 let addPP env id = extendBlockEnv env id ProcPoint
147 initProcPoints = foldl addPP emptyBlockEnv (blockSetToList procPoints)
148 in liftM zdfFpFacts $
149 (zdfSolveFrom initProcPoints "proc-point reachability" lattice
150 forward (fact_bot lattice) $ graphOfLGraph g :: PPFix)
152 extendPPSet :: CmmGraph -> [CmmBlock] -> ProcPointSet -> FuelMonad ProcPointSet
153 extendPPSet g blocks procPoints =
154 do env <- procPointAnalysis procPoints g
155 let add block pps = let id = blockId block
156 in case lookupBlockEnv env id of
157 Just ProcPoint -> extendBlockSet pps id
159 procPoints' = fold_blocks add emptyBlockSet g
160 newPoints = mapMaybe ppSuccessor blocks
161 newPoint = listToMaybe newPoints
162 ppSuccessor b@(Block bid _) =
163 let nreached id = case lookupBlockEnv env id `orElse`
164 pprPanic "no ppt" (ppr id <+> ppr b) of
166 ReachedBy ps -> sizeBlockSet ps
167 block_procpoints = nreached bid
168 -- | Looking for a successor of b that is reached by
169 -- more proc points than b and is not already a proc
170 -- point. If found, it can become a proc point.
171 newId succ_id = not (elemBlockSet succ_id procPoints') &&
172 nreached succ_id > block_procpoints
173 in listToMaybe $ filter newId $ succs b
176 [] -> return procPoints'
177 pps -> extendPPSet g blocks
178 (foldl extendBlockSet procPoints' pps)
180 case newPoint of Just id ->
181 if elemBlockSet id procPoints' then panic "added old proc pt"
182 else extendPPSet g blocks (extendBlockSet procPoints' id)
183 Nothing -> return procPoints'
186 ------------------------------------------------------------------------
187 -- Computing Proc-Point Protocols --
188 ------------------------------------------------------------------------
192 There is one major trick, discovered by Michael Adams, which is that
193 we want to choose protocols in a way that enables us to optimize away
194 some continuations. The optimization is very much like branch-chain
195 elimination, except that it involves passing results as well as
196 control. The idea is that if a call's continuation k does nothing but
197 CopyIn its results and then goto proc point P, the call's continuation
198 may be changed to P, *provided* P's protocol is identical to the
199 protocol for the CopyIn. We choose protocols to make this so.
201 Here's an explanatory example; we begin with the source code (lines
202 separate basic blocks):
210 Zipperization converts this code as follows:
213 call g() returns to k;
220 What we'd like to do is assign P the same CopyIn protocol as k, so we
224 call g() returns to P;
226 P: CopyIn(x, y); ..2..;
228 Of course, P may be the target of more than one continuation, and
229 different continuations may have different protocols. Michael Adams
230 implemented a voting mechanism, but he thinks a simple greedy
231 algorithm would be just as good, so that's what we do.
235 data Protocol = Protocol Convention CmmFormals Area
237 instance Outputable Protocol where
238 ppr (Protocol c fs a) = text "Protocol" <+> ppr c <+> ppr fs <+> ppr a
240 -- | Function 'optimize_calls' chooses protocols only for those proc
241 -- points that are relevant to the optimization explained above.
242 -- The others are assigned by 'add_unassigned', which is not yet clever.
244 addProcPointProtocols :: ProcPointSet -> ProcPointSet -> CmmGraph -> FuelMonad CmmGraph
245 addProcPointProtocols callPPs procPoints g =
246 do liveness <- cmmLivenessZ g
247 (protos, g') <- optimize_calls liveness g
248 blocks'' <- add_CopyOuts protos procPoints g'
249 return $ LGraph (lg_entry g) blocks''
250 where optimize_calls liveness g = -- see Note [Separate Adams optimization]
251 do let (protos, blocks') =
252 fold_blocks maybe_add_call (init_protocols, emptyBlockEnv) g
253 protos' = add_unassigned liveness procPoints protos
254 blocks <- add_CopyIns callPPs protos' blocks'
255 let g' = LGraph (lg_entry g) (mkBlockEnv (map withKey (concat blocks)))
256 withKey b@(Block bid _) = (bid, b)
257 return (protos', runTx removeUnreachableBlocksZ g')
258 maybe_add_call :: CmmBlock -> (BlockEnv Protocol, BlockEnv CmmBlock)
259 -> (BlockEnv Protocol, BlockEnv CmmBlock)
260 -- ^ If the block is a call whose continuation goes to a proc point
261 -- whose protocol either matches the continuation's or is not yet set,
262 -- redirect the call (cf 'newblock') and set the protocol if necessary
263 maybe_add_call block (protos, blocks) =
264 case goto_end $ unzip block of
265 (h, LastOther (LastCall tgt (Just k) args res s))
266 | Just proto <- lookupBlockEnv protos k,
267 Just pee <- branchesToProcPoint k
268 -> let newblock = zipht h (tailOfLast (LastCall tgt (Just pee)
270 changed_blocks = insertBlock newblock blocks
271 unchanged_blocks = insertBlock block blocks
272 in case lookupBlockEnv protos pee of
273 Nothing -> (extendBlockEnv protos pee proto,changed_blocks)
275 if proto == proto' then (protos, changed_blocks)
276 else (protos, unchanged_blocks)
277 _ -> (protos, insertBlock block blocks)
279 branchesToProcPoint :: BlockId -> Maybe BlockId
280 -- ^ Tells whether the named block is just a branch to a proc point
281 branchesToProcPoint id =
282 let (Block _ t) = lookupBlockEnv (lg_blocks g) id `orElse`
283 panic "branch out of graph"
285 ZLast (LastOther (LastBranch pee))
286 | elemBlockSet pee procPoints -> Just pee
288 init_protocols = fold_blocks maybe_add_proto emptyBlockEnv g
289 maybe_add_proto :: CmmBlock -> BlockEnv Protocol -> BlockEnv Protocol
290 --maybe_add_proto (Block id (ZTail (CopyIn c _ fs _srt) _)) env =
291 -- extendBlockEnv env id (Protocol c fs $ toArea id fs)
292 maybe_add_proto _ env = env
293 -- JD: Is this proto stuff even necessary, now that we have
294 -- common blockification?
296 -- | For now, following a suggestion by Ben Lippmeier, we pass all
297 -- live variables as arguments, hoping that a clever register
298 -- allocator might help.
300 add_unassigned :: BlockEnv CmmLive -> ProcPointSet -> BlockEnv Protocol ->
302 add_unassigned = pass_live_vars_as_args
304 pass_live_vars_as_args :: BlockEnv CmmLive -> ProcPointSet ->
305 BlockEnv Protocol -> BlockEnv Protocol
306 pass_live_vars_as_args _liveness procPoints protos = protos'
307 where protos' = foldBlockSet addLiveVars protos procPoints
308 addLiveVars :: BlockId -> BlockEnv Protocol -> BlockEnv Protocol
309 addLiveVars id protos =
310 case lookupBlockEnv protos id of
312 Nothing -> let live = emptyRegSet
313 --lookupBlockEnv _liveness id `orElse`
314 --panic ("no liveness at block " ++ show id)
315 formals = uniqSetToList live
316 prot = Protocol Private formals $ CallArea $ Young id
317 in extendBlockEnv protos id prot
320 -- | Add copy-in instructions to each proc point that did not arise from a call
321 -- instruction. (Proc-points that arise from calls already have their copy-in instructions.)
323 add_CopyIns :: ProcPointSet -> BlockEnv Protocol -> BlockEnv CmmBlock ->
324 FuelMonad [[CmmBlock]]
325 add_CopyIns callPPs protos blocks =
326 liftUniq $ mapM maybe_insert_CopyIns (blockEnvToList blocks)
327 where maybe_insert_CopyIns (_, b@(Block id t))
328 | not $ elemBlockSet id callPPs
329 = case lookupBlockEnv protos id of
330 Just (Protocol c fs _area) ->
331 do LGraph _ blocks <-
332 lgraphOfAGraph (mkLabel id <*> copyInSlot c False fs <*> mkZTail t)
333 return (map snd $ blockEnvToList blocks)
334 Nothing -> return [b]
335 | otherwise = return [b]
337 -- | Add a CopyOut node before each procpoint.
338 -- If the predecessor is a call, then the copy outs should already be done by the callee.
339 -- Note: If we need to add copy-out instructions, they may require stack space,
340 -- so we accumulate a map from the successors to the necessary stack space,
341 -- then update the successors after we have finished inserting the copy-outs.
343 add_CopyOuts :: BlockEnv Protocol -> ProcPointSet -> CmmGraph ->
344 FuelMonad (BlockEnv CmmBlock)
345 add_CopyOuts protos procPoints g = fold_blocks mb_copy_out (return emptyBlockEnv) g
346 where mb_copy_out :: CmmBlock -> FuelMonad (BlockEnv CmmBlock) ->
347 FuelMonad (BlockEnv CmmBlock)
348 mb_copy_out b@(Block bid _) z | bid == lg_entry g = skip b z
350 case last $ unzip b of
351 LastOther (LastCall _ _ _ _ _) -> skip b z -- copy out done by callee
353 copy_out b z = fold_succs trySucc b init >>= finish
354 where init = z >>= (\bmap -> return (b, bmap))
356 if elemBlockSet succId procPoints then
357 case lookupBlockEnv protos succId of
359 Just (Protocol c fs _area) ->
360 insert z succId $ copyOutSlot c Jump fs
364 (b, bs) <- insertBetween b m succId
365 -- pprTrace "insert for succ" (ppr succId <> ppr m) $ do
366 return $ (b, foldl (flip insertBlock) bmap bs)
367 finish (b@(Block bid _), bmap) =
368 return $ (extendBlockEnv bmap bid b)
369 skip b@(Block bid _) bs =
370 bs >>= (\bmap -> return (extendBlockEnv bmap bid b))
372 -- At this point, we have found a set of procpoints, each of which should be
373 -- the entry point of a procedure.
374 -- Now, we create the procedure for each proc point,
375 -- which requires that we:
376 -- 1. build a map from proc points to the blocks reachable from the proc point
377 -- 2. turn each branch to a proc point into a jump
378 -- 3. turn calls and returns into jumps
379 -- 4. build info tables for the procedures -- and update the info table for
380 -- the SRTs in the entry procedure as well.
381 -- Input invariant: A block should only be reachable from a single ProcPoint.
382 splitAtProcPoints :: CLabel -> ProcPointSet-> ProcPointSet -> BlockEnv Status ->
383 CmmTopZ -> FuelMonad [CmmTopZ]
384 splitAtProcPoints entry_label callPPs procPoints procMap
385 (CmmProc (CmmInfo gc upd_fr info_tbl) top_l top_args
386 (stackInfo, g@(LGraph entry blocks))) =
387 do -- Build a map from procpoints to the blocks they reach
388 let addBlock b@(Block bid _) graphEnv =
389 case lookupBlockEnv procMap bid of
390 Just ProcPoint -> add graphEnv bid bid b
391 Just (ReachedBy set) ->
392 case blockSetToList set of
394 [id] -> add graphEnv id bid b
395 _ -> panic "Each block should be reachable from only one ProcPoint"
396 Nothing -> pprPanic "block not reached by a proc point?" (ppr bid)
397 add graphEnv procId bid b = extendBlockEnv graphEnv procId graph'
398 where graph = lookupBlockEnv graphEnv procId `orElse` emptyBlockEnv
399 graph' = extendBlockEnv graph bid b
400 graphEnv <- return $ fold_blocks addBlock emptyBlockEnv g
401 -- Build a map from proc point BlockId to labels for their new procedures
402 -- Due to common blockification, we may overestimate the set of procpoints.
403 let add_label map pp = return $ addToFM map pp lbl
404 where lbl = if pp == entry then entry_label else blockLbl pp
405 procLabels <- foldM add_label emptyFM
406 (filter (elemBlockEnv blocks) (blockSetToList procPoints))
407 -- For each procpoint, we need to know the SP offset on entry.
408 -- If the procpoint is:
409 -- - continuation of a call, the SP offset is in the call
410 -- - otherwise, 0 -- no overflow for passing those variables
411 let add_sp_off b env =
412 case last (unzip b) of
413 LastOther (LastCall {cml_cont = Just succ, cml_ret_args = off,
414 cml_ret_off = updfr_off}) ->
415 extendBlockEnv env succ (off, updfr_off)
417 spEntryMap = fold_blocks add_sp_off (mkBlockEnv [(entry, stackInfo)]) g
418 getStackInfo id = lookupBlockEnv spEntryMap id `orElse` (0, Nothing)
419 -- In each new graph, add blocks jumping off to the new procedures,
420 -- and replace branches to procpoints with branches to the jump-off blocks
421 let add_jump_block (env, bs) (pp, l) =
422 do bid <- liftM mkBlockId getUniqueM
423 let b = Block bid (ZLast (LastOther jump))
424 (argSpace, _) = getStackInfo pp
425 jump = LastCall (CmmLit (CmmLabel l')) Nothing argSpace 0 Nothing
426 l' = if elemBlockSet pp callPPs then entryLblToInfoLbl l else l
427 return (extendBlockEnv env pp bid, b : bs)
428 add_jumps (newGraphEnv) (ppId, blockEnv) =
429 do let needed_jumps = -- find which procpoints we currently branch to
430 foldBlockEnv' add_if_branch_to_pp [] blockEnv
431 add_if_branch_to_pp block rst =
432 case last (unzip block) of
433 LastOther (LastBranch id) -> add_if_pp id rst
434 LastOther (LastCondBranch _ ti fi) ->
435 add_if_pp ti (add_if_pp fi rst)
436 LastOther (LastSwitch _ tbl) -> foldr add_if_pp rst (catMaybes tbl)
438 add_if_pp id rst = case lookupFM procLabels id of
439 Just x -> (id, x) : rst
441 (jumpEnv, jumpBlocks) <-
442 foldM add_jump_block (emptyBlockEnv, []) needed_jumps
443 -- update the entry block
444 let b = expectJust "block in env" $ lookupBlockEnv blockEnv ppId
445 off = getStackInfo ppId
446 blockEnv' = extendBlockEnv blockEnv ppId b
447 -- replace branches to procpoints with branches to jumps
448 LGraph _ blockEnv'' = replaceBranches jumpEnv $ LGraph ppId blockEnv'
449 -- add the jump blocks to the graph
450 blockEnv''' = foldl (flip insertBlock) blockEnv'' jumpBlocks
451 let g' = (off, LGraph ppId blockEnv''')
452 -- pprTrace "g' pre jumps" (ppr g') $ do
453 return (extendBlockEnv newGraphEnv ppId g')
454 graphEnv <- foldM add_jumps emptyBlockEnv $ blockEnvToList graphEnv
455 let to_proc (bid, g) | elemBlockSet bid callPPs =
457 CmmProc (CmmInfo gc upd_fr info_tbl) top_l top_args g
459 CmmProc emptyContInfoTable lbl [] g
460 where lbl = expectJust "pp label" $ lookupFM procLabels bid
462 CmmProc (CmmInfo Nothing Nothing CmmNonInfoTable) lbl [] g
463 where lbl = expectJust "pp label" $ lookupFM procLabels bid
464 -- The C back end expects to see return continuations before the call sites.
465 -- Here, we sort them in reverse order -- it gets reversed later.
466 let (_, block_order) = foldl add_block_num (0::Int, emptyBlockEnv) (postorder_dfs g)
467 add_block_num (i, map) (Block bid _) = (i+1, extendBlockEnv map bid i)
468 sort_fn (bid, _) (bid', _) =
469 compare (expectJust "block_order" $ lookupBlockEnv block_order bid)
470 (expectJust "block_order" $ lookupBlockEnv block_order bid')
471 procs <- return $ map to_proc $ sortBy sort_fn $ blockEnvToList graphEnv
472 return -- pprTrace "procLabels" (ppr procLabels)
473 -- pprTrace "splitting graphs" (ppr procs)
475 splitAtProcPoints _ _ _ _ t@(CmmData _ _) = return [t]
477 ----------------------------------------------------------------
480 Note [Direct reachability]
482 Block B is directly reachable from proc point P iff control can flow
483 from P to B without passing through an intervening proc point.
486 ----------------------------------------------------------------
489 Note [No simple dataflow]
491 Sadly, it seems impossible to compute the proc points using a single
492 dataflow pass. One might attempt to use this simple lattice:
494 data Location = Unknown
495 | InProc BlockId -- node is in procedure headed by the named proc point
496 | ProcPoint -- node is itself a proc point
498 At a join, a node in two different blocks becomes a proc point.
499 The difficulty is that the change of information during iterative
500 computation may promote a node prematurely. Here's a program that
501 illustrates the difficulty:
510 L2: if (...) { g(); goto L1; }
514 The only proc-point needed (besides the entry) is L1. But in an
515 iterative analysis, consider what happens to L2. On the first pass
516 through, it rises from Unknown to 'InProc entry', but when L1 is
517 promoted to a proc point (because it's the successor of g()), L1's
518 successors will be promoted to 'InProc L1'. The problem hits when the
519 new fact 'InProc L1' flows into L2 which is already bound to 'InProc entry'.
520 The join operation makes it a proc point when in fact it needn't be,
521 because its immediate dominator L1 is already a proc point and there
522 are no other proc points that directly reach L2.
527 {- Note [Separate Adams optimization]
528 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
529 It may be worthwhile to attempt the Adams optimization by rewriting
530 the graph before the assignment of proc-point protocols. Here are a
533 g() returns to k; g() returns to L;
534 k: CopyIn c ress; goto L:
536 L: // no CopyIn node here L: CopyIn c ress;
539 And when c == c' and ress == ress', this also:
541 g() returns to k; g() returns to L;
542 k: CopyIn c ress; goto L:
544 L: CopyIn c' ress' L: CopyIn c' ress' ;
546 In both cases the goal is to eliminate k.