[llvm] r253876 - [WebAssembly] Use dominator information to improve BLOCK placement
Dan Gohman via llvm-commits
llvm-commits at lists.llvm.org
Mon Nov 23 08:19:56 PST 2015
Author: djg
Date: Mon Nov 23 10:19:56 2015
New Revision: 253876
URL: http://llvm.org/viewvc/llvm-project?rev=253876&view=rev
Log:
[WebAssembly] Use dominator information to improve BLOCK placement
Always starting blocks at the top of their containing loops works, but creates
unnecessarily deep nesting because it makes all blocks in a loop overlap.
Refine the BLOCK placement algorithm to start blocks at nearest common
dominating points instead, which significantly shrinks them and reduces
overlapping.
Modified:
llvm/trunk/lib/Target/WebAssembly/WebAssemblyCFGStackify.cpp
llvm/trunk/test/CodeGen/WebAssembly/cfg-stackify.ll
Modified: llvm/trunk/lib/Target/WebAssembly/WebAssemblyCFGStackify.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/WebAssembly/WebAssemblyCFGStackify.cpp?rev=253876&r1=253875&r2=253876&view=diff
==============================================================================
--- llvm/trunk/lib/Target/WebAssembly/WebAssemblyCFGStackify.cpp (original)
+++ llvm/trunk/lib/Target/WebAssembly/WebAssemblyCFGStackify.cpp Mon Nov 23 10:19:56 2015
@@ -29,6 +29,7 @@
#include "MCTargetDesc/WebAssemblyMCTargetDesc.h"
#include "WebAssemblySubtarget.h"
#include "llvm/ADT/SCCIterator.h"
+#include "llvm/CodeGen/MachineDominators.h"
#include "llvm/CodeGen/MachineFunction.h"
#include "llvm/CodeGen/MachineInstrBuilder.h"
#include "llvm/CodeGen/MachineLoopInfo.h"
@@ -47,6 +48,8 @@ class WebAssemblyCFGStackify final : pub
void getAnalysisUsage(AnalysisUsage &AU) const override {
AU.setPreservesCFG();
+ AU.addRequired<MachineDominatorTree>();
+ AU.addPreserved<MachineDominatorTree>();
AU.addRequired<MachineLoopInfo>();
AU.addPreserved<MachineLoopInfo>();
MachineFunctionPass::getAnalysisUsage(AU);
@@ -107,6 +110,11 @@ struct POStackEntry {
};
} // end anonymous namespace
+static bool LoopContains(const MachineLoop *Loop,
+ const MachineBasicBlock *MBB) {
+ return Loop ? Loop->contains(MBB) : true;
+}
+
POStackEntry::POStackEntry(MachineBasicBlock *MBB, MachineFunction &MF,
const MachineLoopInfo &MLI)
: MBB(MBB), Succs(MBB->successors()) {
@@ -124,11 +132,11 @@ POStackEntry::POStackEntry(MachineBasicB
// Keep loops contiguous by preferring the block that's in the same
// loop.
- MachineLoop *LoopA = MLI.getLoopFor(A);
- MachineLoop *LoopB = MLI.getLoopFor(B);
- if (LoopA == Loop && LoopB != Loop)
+ bool LoopContainsA = LoopContains(Loop, A);
+ bool LoopContainsB = LoopContains(Loop, B);
+ if (LoopContainsA && !LoopContainsB)
return true;
- if (LoopA != Loop && LoopB == Loop)
+ if (!LoopContainsA && LoopContainsB)
return false;
// Minimize perturbation by preferring the block which is the immediate
@@ -192,12 +200,17 @@ static void SortBlocks(MachineFunction &
#ifndef NDEBUG
for (auto &MBB : MF)
if (MachineLoop *Loop = MLI.getLoopFor(&MBB)) {
- // Assert that loops are contiguous.
- assert(Loop->getHeader() == Loop->getTopBlock());
- assert((Loop->getHeader() == &MBB ||
- Loop->contains(
- MLI.getLoopFor(&*prev(MachineFunction::iterator(&MBB))))) &&
- "Loop isn't contiguous");
+ // Assert that all containing loops are contiguous.
+ for (MachineLoop *L = Loop; L; L = L->getParentLoop()) {
+ if (&MBB == L->getHeader()) {
+ assert(&MBB == L->getTopBlock());
+ } else {
+ assert(&MBB != L->getTopBlock());
+ assert(L->contains(
+ MLI.getLoopFor(&*prev(MachineFunction::iterator(&MBB)))) &&
+ "Loop isn't contiguous");
+ }
+ }
} else {
// Assert that non-loops have no backedge predecessors.
for (auto Pred : MBB.predecessors())
@@ -207,45 +220,68 @@ static void SortBlocks(MachineFunction &
#endif
}
-/// Insert BLOCK markers at appropriate places.
-static void PlaceBlockMarkers(MachineBasicBlock &MBB, MachineBasicBlock &Succ,
- MachineFunction &MF, const MachineLoopInfo &MLI,
- const WebAssemblyInstrInfo &TII) {
- // Backward branches are loop backedges, and we place the LOOP markers
- // separately. So only consider forward branches here.
- if (Succ.getNumber() <= MBB.getNumber())
- return;
+static int GetLoopDepth(const MachineLoop *Loop) {
+ return Loop ? Loop->getLoopDepth() : 0;
+}
- // Place the BLOCK for a forward branch. For simplicity, we just insert
- // blocks immediately inside loop boundaries.
- MachineLoop *Loop = MLI.getLoopFor(&Succ);
- MachineBasicBlock &Header = *(Loop ? Loop->getHeader() : &MF.front());
- MachineBasicBlock::iterator InsertPos = Header.begin(), End = Header.end();
- if (InsertPos != End) {
- if (InsertPos->getOpcode() == WebAssembly::LOOP)
- ++InsertPos;
- int SuccNumber = Succ.getNumber();
- // Position the BLOCK in nesting order.
- for (; InsertPos != End && InsertPos->getOpcode() == WebAssembly::BLOCK;
- ++InsertPos) {
- int N = InsertPos->getOperand(0).getMBB()->getNumber();
- if (N < SuccNumber)
- break;
- // If there's already a BLOCK for Succ, we don't need another.
- if (N == SuccNumber)
- return;
+/// Insert a BLOCK marker for branches to MBB (if needed).
+static void PlaceBlockMarkers(MachineBasicBlock &MBB, MachineFunction &MF,
+ const WebAssemblyInstrInfo &TII,
+ MachineDominatorTree &MDT,
+ const MachineLoopInfo &MLI) {
+ // Place the BLOCK for forward non-fallthrough branches. Put it at the nearest
+ // common dominator of all preceding predecesors so that we minimize the time
+ // that it's on the stack, which reduces overall stack height.
+ MachineBasicBlock *Header = nullptr;
+ bool IsBranchedTo = false;
+ int MBBNumber = MBB.getNumber();
+ for (MachineBasicBlock *Pred : MBB.predecessors())
+ if (Pred->getNumber() < MBBNumber) {
+ Header = Header ? MDT.findNearestCommonDominator(Header, Pred) : Pred;
+ if (!Pred->isLayoutSuccessor(&MBB) ||
+ !(Pred->empty() || !Pred->back().isBarrier()))
+ IsBranchedTo = true;
}
+ if (!Header)
+ return;
+ if (!IsBranchedTo)
+ return;
+
+ MachineBasicBlock::iterator InsertPos;
+ MachineLoop *HeaderLoop = MLI.getLoopFor(Header);
+ int MBBLoopDepth = GetLoopDepth(MLI.getLoopFor(&MBB));
+ int HeaderLoopDepth = GetLoopDepth(HeaderLoop);
+ if (HeaderLoopDepth > MBBLoopDepth) {
+ // The nearest common dominating point is more deeply nested. Insert the
+ // BLOCK just above the LOOP.
+ for (int i = 0; i < HeaderLoopDepth - 1 - MBBLoopDepth; ++i)
+ HeaderLoop = HeaderLoop->getParentLoop();
+ Header = HeaderLoop->getHeader();
+ InsertPos = Header->begin();
+ // Don't insert a BLOCK if we can reuse a loop exit label though.
+ if (InsertPos != Header->end() &&
+ InsertPos->getOpcode() == WebAssembly::LOOP &&
+ InsertPos->getOperand(0).getMBB() == &MBB)
+ return;
+ } else {
+ // Insert the BLOCK as late in the block as we can, but before any existing
+ // BLOCKs.
+ InsertPos = Header->getFirstTerminator();
+ while (InsertPos != Header->begin() &&
+ std::prev(InsertPos)->getOpcode() == WebAssembly::BLOCK)
+ --InsertPos;
}
- BuildMI(Header, InsertPos, DebugLoc(), TII.get(WebAssembly::BLOCK))
- .addMBB(&Succ);
+ BuildMI(*Header, InsertPos, DebugLoc(), TII.get(WebAssembly::BLOCK))
+ .addMBB(&MBB);
}
/// Insert LOOP and BLOCK markers at appropriate places.
static void PlaceMarkers(MachineFunction &MF, const MachineLoopInfo &MLI,
- const WebAssemblyInstrInfo &TII) {
+ const WebAssemblyInstrInfo &TII,
+ MachineDominatorTree &MDT) {
for (auto &MBB : MF) {
- // Place the LOOP for loops.
+ // Place the LOOP for MBB if MBB is the header of a loop.
if (MachineLoop *Loop = MLI.getLoopFor(&MBB))
if (Loop->getHeader() == &MBB) {
// The operand of a LOOP is the first block after the loop. If the loop
@@ -260,20 +296,29 @@ static void PlaceMarkers(MachineFunction
.addMBB(&*Iter);
}
- // Check for forward branches and switches that need BLOCKS placed.
- for (auto &Term : MBB.terminators())
- for (auto &MO : Term.operands())
- if (MO.isMBB())
- PlaceBlockMarkers(MBB, *MO.getMBB(), MF, MLI, TII);
+ // Place the BLOCK for MBB if MBB is branched to from above.
+ PlaceBlockMarkers(MBB, MF, TII, MDT, MLI);
}
}
+#ifndef NDEBUG
+static bool
+IsOnStack(const SmallVectorImpl<std::pair<MachineBasicBlock *, bool>> &Stack,
+ const MachineBasicBlock *MBB) {
+ for (const auto &Pair : Stack)
+ if (Pair.first == MBB)
+ return true;
+ return false;
+}
+#endif
+
bool WebAssemblyCFGStackify::runOnMachineFunction(MachineFunction &MF) {
DEBUG(dbgs() << "********** CFG Stackifying **********\n"
"********** Function: "
<< MF.getName() << '\n');
const auto &MLI = getAnalysis<MachineLoopInfo>();
+ auto &MDT = getAnalysis<MachineDominatorTree>();
const auto &TII = *MF.getSubtarget<WebAssemblySubtarget>().getInstrInfo();
// RPO sorting needs all loops to be single-entry.
@@ -283,7 +328,41 @@ bool WebAssemblyCFGStackify::runOnMachin
SortBlocks(MF, MLI);
// Place the BLOCK and LOOP markers to indicate the beginnings of scopes.
- PlaceMarkers(MF, MLI, TII);
+ PlaceMarkers(MF, MLI, TII, MDT);
+
+#ifndef NDEBUG
+ // Verify that block and loop beginnings and endings are in FIFO order, and
+ // that all references to blocks are to blocks on the stack at the point of
+ // the reference.
+ SmallVector<std::pair<MachineBasicBlock *, bool>, 0> Stack;
+ for (auto &MBB : MF) {
+ while (!Stack.empty() && Stack.back().first == &MBB)
+ if (Stack.back().second) {
+ assert(Stack.size() >= 2);
+ Stack.pop_back();
+ Stack.pop_back();
+ } else {
+ assert(Stack.size() >= 1);
+ Stack.pop_back();
+ }
+ for (auto &MI : MBB)
+ switch (MI.getOpcode()) {
+ case WebAssembly::LOOP:
+ Stack.push_back(std::make_pair(&MBB, false));
+ Stack.push_back(std::make_pair(MI.getOperand(0).getMBB(), true));
+ break;
+ case WebAssembly::BLOCK:
+ Stack.push_back(std::make_pair(MI.getOperand(0).getMBB(), false));
+ break;
+ default:
+ for (const MachineOperand &MO : MI.explicit_operands())
+ if (MO.isMBB())
+ assert(IsOnStack(Stack, MO.getMBB()));
+ break;
+ }
+ }
+ assert(Stack.empty());
+#endif
return true;
}
Modified: llvm/trunk/test/CodeGen/WebAssembly/cfg-stackify.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/WebAssembly/cfg-stackify.ll?rev=253876&r1=253875&r2=253876&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/WebAssembly/cfg-stackify.ll (original)
+++ llvm/trunk/test/CodeGen/WebAssembly/cfg-stackify.ll Mon Nov 23 10:19:56 2015
@@ -99,9 +99,9 @@ for.end:
; CHECK-LABEL: doublediamond:
; CHECK: block BB3_5{{$}}
-; CHECK: block BB3_4{{$}}
; CHECK: block BB3_2{{$}}
; CHECK: br_if $pop{{[0-9]+}}, BB3_2{{$}}
+; CHECK: block BB3_4{{$}}
; CHECK: br_if $pop{{[0-9]+}}, BB3_4{{$}}
; CHECK: br BB3_5{{$}}
; CHECK: BB3_4:
@@ -216,8 +216,8 @@ exit:
; CHECK-LABEL: doubletriangle:
; CHECK: block BB9_4{{$}}
-; CHECK: block BB9_3{{$}}
; CHECK: br_if $pop{{[0-9]+}}, BB9_4{{$}}
+; CHECK: block BB9_3{{$}}
; CHECK: br_if $pop{{[0-9]+}}, BB9_3{{$}}
; CHECK: BB9_3:
; CHECK: BB9_4:
@@ -271,6 +271,45 @@ exit:
ret i32 0
}
+; CHECK-LABEL: doublediamond_in_a_loop:
+; CHECK: BB11_1:
+; CHECK: loop BB11_7{{$}}
+; CHECK: block BB11_6{{$}}
+; CHECK: block BB11_3{{$}}
+; CHECK: br_if $pop{{.*}}, BB11_3{{$}}
+; CHECK: br BB11_6{{$}}
+; CHECK: BB11_3:
+; CHECK: block BB11_5{{$}}
+; CHECK: br_if $pop{{.*}}, BB11_5{{$}}
+; CHECK: br BB11_6{{$}}
+; CHECK: BB11_5:
+; CHECK: BB11_6:
+; CHECK: br BB11_1{{$}}
+define i32 @doublediamond_in_a_loop(i32 %a, i32 %b, i32* %p) {
+entry:
+ br label %header
+header:
+ %c = icmp eq i32 %a, 0
+ %d = icmp eq i32 %b, 0
+ store volatile i32 0, i32* %p
+ br i1 %c, label %true, label %false
+true:
+ store volatile i32 1, i32* %p
+ br label %exit
+false:
+ store volatile i32 2, i32* %p
+ br i1 %d, label %ft, label %ff
+ft:
+ store volatile i32 3, i32* %p
+ br label %exit
+ff:
+ store volatile i32 4, i32* %p
+ br label %exit
+exit:
+ store volatile i32 5, i32* %p
+ br label %header
+}
+
; Test that nested loops are handled.
declare void @bar()
More information about the llvm-commits
mailing list