using namespace llvm::PatternMatch;
STATISTIC(NumBlocksElim, "Number of blocks eliminated");
-STATISTIC(NumPHIsElim, "Number of trivial PHIs eliminated");
-STATISTIC(NumGEPsElim, "Number of GEPs converted to casts");
+STATISTIC(NumPHIsElim, "Number of trivial PHIs eliminated");
+STATISTIC(NumGEPsElim, "Number of GEPs converted to casts");
STATISTIC(NumCmpUses, "Number of uses of Cmp expressions replaced with uses of "
"sunken Cmps");
STATISTIC(NumCastUses, "Number of uses of Cast expressions replaced with uses "
"of sunken Casts");
STATISTIC(NumMemoryInsts, "Number of memory instructions whose address "
"computations were sunk");
-STATISTIC(NumExtsMoved, "Number of [s|z]ext instructions combined with loads");
-STATISTIC(NumExtUses, "Number of uses of [s|z]ext instructions optimized");
-STATISTIC(NumRetsDup, "Number of return instructions duplicated");
+STATISTIC(NumExtsMoved, "Number of [s|z]ext instructions combined with loads");
+STATISTIC(NumExtUses, "Number of uses of [s|z]ext instructions optimized");
static cl::opt<bool> DisableBranchOpts(
"disable-cgp-branch-opts", cl::Hidden, cl::init(false),
bool OptimizeCallInst(CallInst *CI);
bool MoveExtToFormExtLoad(Instruction *I);
bool OptimizeExtUses(Instruction *I);
- bool DupRetToEnableTailCallOpts(ReturnInst *RI);
};
}
return Simplifier.fold(CI, TD);
}
-/// DupRetToEnableTailCallOpts - Look for opportunities to duplicate return
-/// instructions to the predecessor to enable tail call optimizations. The
-/// case it is currently looking for is:
-/// bb0:
-/// %tmp0 = tail call i32 @f0()
-/// br label %return
-/// bb1:
-/// %tmp1 = tail call i32 @f1()
-/// br label %return
-/// bb2:
-/// %tmp2 = tail call i32 @f2()
-/// br label %return
-/// return:
-/// %retval = phi i32 [ %tmp0, %bb0 ], [ %tmp1, %bb1 ], [ %tmp2, %bb2 ]
-/// ret i32 %retval
-///
-/// =>
-///
-/// bb0:
-/// %tmp0 = tail call i32 @f0()
-/// ret i32 %tmp0
-/// bb1:
-/// %tmp1 = tail call i32 @f1()
-/// ret i32 %tmp1
-/// bb2:
-/// %tmp2 = tail call i32 @f2()
-/// ret i32 %tmp2
-///
-bool CodeGenPrepare::DupRetToEnableTailCallOpts(ReturnInst *RI) {
- Value *V = RI->getReturnValue();
- if (!V)
- return false;
-
- if (PHINode *PN = dyn_cast<PHINode>(V)) {
- BasicBlock *BB = RI->getParent();
- if (PN->getParent() != BB)
- return false;
-
- // It's not safe to eliminate the sign / zero extension of the return value.
- // See llvm::isInTailCallPosition().
- const Function *F = BB->getParent();
- unsigned CallerRetAttr = F->getAttributes().getRetAttributes();
- if ((CallerRetAttr & Attribute::ZExt) || (CallerRetAttr & Attribute::SExt))
- return false;
-
- // Make sure there are no instructions between PHI and return.
- BasicBlock::iterator BI = PN;
- do { ++BI; } while (isa<DbgInfoIntrinsic>(BI));
- if (&*BI != RI)
- return false;
-
- /// Only dup the ReturnInst if the CallInst is likely to be emitted as a
- /// tail call.
- SmallVector<CallInst*, 4> TailCalls;
- for (unsigned I = 0, E = PN->getNumIncomingValues(); I != E; ++I) {
- CallInst *CI = dyn_cast<CallInst>(PN->getIncomingValue(I));
- if (CI && TLI->mayBeEmittedAsTailCall(CI))
- TailCalls.push_back(CI);
- }
-
- bool Changed = false;
- for (unsigned i = 0, e = TailCalls.size(); i != e; ++i) {
- CallInst *CI = TailCalls[i];
- CallSite CS(CI);
-
- // Conservatively require the attributes of the call to match those of
- // the return. Ignore noalias because it doesn't affect the call sequence.
- unsigned CalleeRetAttr = CS.getAttributes().getRetAttributes();
- if ((CalleeRetAttr ^ CallerRetAttr) & ~Attribute::NoAlias)
- continue;
-
- // Make sure the call instruction is followed by an unconditional branch
- // to the return block.
- BasicBlock *CallBB = CI->getParent();
- BranchInst *BI = dyn_cast<BranchInst>(CallBB->getTerminator());
- if (!BI || !BI->isUnconditional() || BI->getSuccessor(0) != BB)
- continue;
-
- // Duplicate the return into CallBB.
- (void)FoldReturnIntoUncondBranch(RI, BB, CallBB);
- Changed = true;
- ++NumRetsDup;
- }
-
- return Changed;
- }
-
- return false;
-}
-
//===----------------------------------------------------------------------===//
// Memory Optimization
//===----------------------------------------------------------------------===//
if (CallInst *CI = dyn_cast<CallInst>(I))
return OptimizeCallInst(CI);
- if (ReturnInst *RI = dyn_cast<ReturnInst>(I))
- return DupRetToEnableTailCallOpts(RI);
-
return false;
}
+++ /dev/null
-; RUN: llc < %s -mtriple=x86_64-apple-darwin | FileCheck %s
-
-; Teach CGP to dup returns to enable tail call optimization.
-; rdar://9147433
-
-define i32 @foo(i32 %x) nounwind ssp {
-; CHECK: foo:
-entry:
- switch i32 %x, label %return [
- i32 1, label %sw.bb
- i32 2, label %sw.bb1
- i32 3, label %sw.bb3
- i32 4, label %sw.bb5
- i32 5, label %sw.bb7
- i32 6, label %sw.bb9
- ]
-
-sw.bb: ; preds = %entry
-; CHECK: jmp _f1
- %call = tail call i32 @f1() nounwind
- br label %return
-
-sw.bb1: ; preds = %entry
-; CHECK: jmp _f2
- %call2 = tail call i32 @f2() nounwind
- br label %return
-
-sw.bb3: ; preds = %entry
-; CHECK: jmp _f3
- %call4 = tail call i32 @f3() nounwind
- br label %return
-
-sw.bb5: ; preds = %entry
-; CHECK: jmp _f4
- %call6 = tail call i32 @f4() nounwind
- br label %return
-
-sw.bb7: ; preds = %entry
-; CHECK: jmp _f5
- %call8 = tail call i32 @f5() nounwind
- br label %return
-
-sw.bb9: ; preds = %entry
-; CHECK: jmp _f6
- %call10 = tail call i32 @f6() nounwind
- br label %return
-
-return: ; preds = %entry, %sw.bb9, %sw.bb7, %sw.bb5, %sw.bb3, %sw.bb1, %sw.bb
- %retval.0 = phi i32 [ %call10, %sw.bb9 ], [ %call8, %sw.bb7 ], [ %call6, %sw.bb5 ], [ %call4, %sw.bb3 ], [ %call2, %sw.bb1 ], [ %call, %sw.bb ], [ 0, %entry ]
- ret i32 %retval.0
-}
-
-declare i32 @f1()
-
-declare i32 @f2()
-
-declare i32 @f3()
-
-declare i32 @f4()
-
-declare i32 @f5()
-
-declare i32 @f6()