diff options
author | Roman Lebedev <lebedev.ri@gmail.com> | 2022-02-17 12:07:17 +0300 |
---|---|---|
committer | Tom Stellard <tstellar@redhat.com> | 2022-03-08 19:22:12 -0800 |
commit | ba9ff030d302910e0d2155536558f71da98d5a5a (patch) | |
tree | 0a364f9a15811ffa7d0cad89e2176ac8cbc9f0f9 | |
parent | [NFC][PhaseOrdering] Add some tests from D119839 (diff) | |
download | llvm-project-ba9ff030d302910e0d2155536558f71da98d5a5a.tar.gz llvm-project-ba9ff030d302910e0d2155536558f71da98d5a5a.tar.bz2 llvm-project-ba9ff030d302910e0d2155536558f71da98d5a5a.zip |
[SimplifyCFG][PhaseOrdering] Defer lowering switch into an integer range comparison and branch until after at least the IPSCCP
That transformation is lossy, as discussed in
https://github.com/llvm/llvm-project/issues/53853
and https://github.com/rust-lang/rust/issues/85133#issuecomment-904185574
This is an alternative to D119839,
which would add a limited IPSCCP into SimplifyCFG.
Unlike lowering switch to lookup, we still want this transformation
to happen relatively early, but after giving a chance for the things
like CVP to do their thing. It seems like deferring it just until
the IPSCCP is enough for the tests at hand, but perhaps we need to
be more aggressive and disable it until CVP.
Fixes https://github.com/llvm/llvm-project/issues/53853
Refs. https://github.com/rust-lang/rust/issues/85133
Reviewed By: nikic
Differential Revision: https://reviews.llvm.org/D119854
23 files changed, 104 insertions, 60 deletions
diff --git a/llvm/include/llvm/Transforms/Utils/SimplifyCFGOptions.h b/llvm/include/llvm/Transforms/Utils/SimplifyCFGOptions.h index fb3a7490346f..7af879638a4d 100644 --- a/llvm/include/llvm/Transforms/Utils/SimplifyCFGOptions.h +++ b/llvm/include/llvm/Transforms/Utils/SimplifyCFGOptions.h @@ -23,6 +23,7 @@ class AssumptionCache; struct SimplifyCFGOptions { int BonusInstThreshold = 1; bool ForwardSwitchCondToPhi = false; + bool ConvertSwitchRangeToICmp = false; bool ConvertSwitchToLookupTable = false; bool NeedCanonicalLoop = true; bool HoistCommonInsts = false; @@ -41,6 +42,10 @@ struct SimplifyCFGOptions { ForwardSwitchCondToPhi = B; return *this; } + SimplifyCFGOptions &convertSwitchRangeToICmp(bool B) { + ConvertSwitchRangeToICmp = B; + return *this; + } SimplifyCFGOptions &convertSwitchToLookupTable(bool B) { ConvertSwitchToLookupTable = B; return *this; diff --git a/llvm/lib/Passes/PassBuilder.cpp b/llvm/lib/Passes/PassBuilder.cpp index 015ca1eec4df..dedfc81f11bb 100644 --- a/llvm/lib/Passes/PassBuilder.cpp +++ b/llvm/lib/Passes/PassBuilder.cpp @@ -679,6 +679,8 @@ Expected<SimplifyCFGOptions> parseSimplifyCFGOptions(StringRef Params) { bool Enable = !ParamName.consume_front("no-"); if (ParamName == "forward-switch-cond") { Result.forwardSwitchCondToPhi(Enable); + } else if (ParamName == "switch-range-to-icmp") { + Result.convertSwitchRangeToICmp(Enable); } else if (ParamName == "switch-to-lookup") { Result.convertSwitchToLookupTable(Enable); } else if (ParamName == "keep-loops") { diff --git a/llvm/lib/Passes/PassBuilderPipelines.cpp b/llvm/lib/Passes/PassBuilderPipelines.cpp index 194bff054fd7..e838665eb9ce 100644 --- a/llvm/lib/Passes/PassBuilderPipelines.cpp +++ b/llvm/lib/Passes/PassBuilderPipelines.cpp @@ -259,14 +259,16 @@ PassBuilder::buildO1FunctionSimplificationPipeline(OptimizationLevel Level, FPM.addPass(EarlyCSEPass(true /* Enable mem-ssa. */)); // Hoisting of scalars and load expressions. - FPM.addPass(SimplifyCFGPass()); + FPM.addPass( + SimplifyCFGPass(SimplifyCFGOptions().convertSwitchRangeToICmp(true))); FPM.addPass(InstCombinePass()); FPM.addPass(LibCallsShrinkWrapPass()); invokePeepholeEPCallbacks(FPM, Level); - FPM.addPass(SimplifyCFGPass()); + FPM.addPass( + SimplifyCFGPass(SimplifyCFGOptions().convertSwitchRangeToICmp(true))); // Form canonically associated expression trees, and simplify the trees using // basic mathematical properties. For example, this will form (nearly) @@ -340,7 +342,8 @@ PassBuilder::buildO1FunctionSimplificationPipeline(OptimizationLevel Level, FPM.addPass(createFunctionToLoopPassAdaptor(std::move(LPM1), /*UseMemorySSA=*/true, /*UseBlockFrequencyInfo=*/true)); - FPM.addPass(SimplifyCFGPass()); + FPM.addPass( + SimplifyCFGPass(SimplifyCFGOptions().convertSwitchRangeToICmp(true))); FPM.addPass(InstCombinePass()); // The loop passes in LPM2 (LoopFullUnrollPass) do not preserve MemorySSA. // *All* loop passes must preserve it, in order to be able to use it. @@ -378,7 +381,8 @@ PassBuilder::buildO1FunctionSimplificationPipeline(OptimizationLevel Level, // the simplifications and basic cleanup after all the simplifications. // TODO: Investigate if this is too expensive. FPM.addPass(ADCEPass()); - FPM.addPass(SimplifyCFGPass()); + FPM.addPass( + SimplifyCFGPass(SimplifyCFGOptions().convertSwitchRangeToICmp(true))); FPM.addPass(InstCombinePass()); invokePeepholeEPCallbacks(FPM, Level); @@ -413,7 +417,8 @@ PassBuilder::buildFunctionSimplificationPipeline(OptimizationLevel Level, // Global value numbering based sinking. if (EnableGVNSink) { FPM.addPass(GVNSinkPass()); - FPM.addPass(SimplifyCFGPass()); + FPM.addPass( + SimplifyCFGPass(SimplifyCFGOptions().convertSwitchRangeToICmp(true))); } if (EnableConstraintElimination) @@ -426,7 +431,8 @@ PassBuilder::buildFunctionSimplificationPipeline(OptimizationLevel Level, FPM.addPass(JumpThreadingPass()); FPM.addPass(CorrelatedValuePropagationPass()); - FPM.addPass(SimplifyCFGPass()); + FPM.addPass( + SimplifyCFGPass(SimplifyCFGOptions().convertSwitchRangeToICmp(true))); FPM.addPass(InstCombinePass()); if (Level == OptimizationLevel::O3) FPM.addPass(AggressiveInstCombinePass()); @@ -443,7 +449,8 @@ PassBuilder::buildFunctionSimplificationPipeline(OptimizationLevel Level, FPM.addPass(PGOMemOPSizeOpt()); FPM.addPass(TailCallElimPass()); - FPM.addPass(SimplifyCFGPass()); + FPM.addPass( + SimplifyCFGPass(SimplifyCFGOptions().convertSwitchRangeToICmp(true))); // Form canonically associated expression trees, and simplify the trees using // basic mathematical properties. For example, this will form (nearly) @@ -520,7 +527,8 @@ PassBuilder::buildFunctionSimplificationPipeline(OptimizationLevel Level, FPM.addPass(createFunctionToLoopPassAdaptor(std::move(LPM1), /*UseMemorySSA=*/true, /*UseBlockFrequencyInfo=*/true)); - FPM.addPass(SimplifyCFGPass()); + FPM.addPass( + SimplifyCFGPass(SimplifyCFGOptions().convertSwitchRangeToICmp(true))); FPM.addPass(InstCombinePass()); // The loop passes in LPM2 (LoopIdiomRecognizePass, IndVarSimplifyPass, // LoopDeletionPass and LoopFullUnrollPass) do not preserve MemorySSA. @@ -586,8 +594,10 @@ PassBuilder::buildFunctionSimplificationPipeline(OptimizationLevel Level, for (auto &C : ScalarOptimizerLateEPCallbacks) C(FPM, Level); - FPM.addPass(SimplifyCFGPass( - SimplifyCFGOptions().hoistCommonInsts(true).sinkCommonInsts(true))); + FPM.addPass(SimplifyCFGPass(SimplifyCFGOptions() + .convertSwitchRangeToICmp(true) + .hoistCommonInsts(true) + .sinkCommonInsts(true))); FPM.addPass(InstCombinePass()); invokePeepholeEPCallbacks(FPM, Level); @@ -625,7 +635,8 @@ void PassBuilder::addPGOInstrPasses(ModulePassManager &MPM, FunctionPassManager FPM; FPM.addPass(SROAPass()); FPM.addPass(EarlyCSEPass()); // Catch trivial redundancies. - FPM.addPass(SimplifyCFGPass()); // Merge & remove basic blocks. + FPM.addPass(SimplifyCFGPass(SimplifyCFGOptions().convertSwitchRangeToICmp( + true))); // Merge & remove basic blocks. FPM.addPass(InstCombinePass()); // Combine silly sequences. invokePeepholeEPCallbacks(FPM, Level); @@ -939,7 +950,8 @@ PassBuilder::buildModuleSimplificationPipeline(OptimizationLevel Level, GlobalCleanupPM.addPass(InstCombinePass()); invokePeepholeEPCallbacks(GlobalCleanupPM, Level); - GlobalCleanupPM.addPass(SimplifyCFGPass()); + GlobalCleanupPM.addPass( + SimplifyCFGPass(SimplifyCFGOptions().convertSwitchRangeToICmp(true))); MPM.addPass(createModuleToFunctionPassAdaptor(std::move(GlobalCleanupPM), PTO.EagerlyInvalidateAnalyses)); @@ -1027,7 +1039,8 @@ void PassBuilder::addVectorPasses(OptimizationLevel Level, ExtraPasses.addPass( createFunctionToLoopPassAdaptor(std::move(LPM), /*UseMemorySSA=*/true, /*UseBlockFrequencyInfo=*/true)); - ExtraPasses.addPass(SimplifyCFGPass()); + ExtraPasses.addPass( + SimplifyCFGPass(SimplifyCFGOptions().convertSwitchRangeToICmp(true))); ExtraPasses.addPass(InstCombinePass()); FPM.addPass(std::move(ExtraPasses)); } @@ -1043,6 +1056,7 @@ void PassBuilder::addVectorPasses(OptimizationLevel Level, // before SLP vectorization. FPM.addPass(SimplifyCFGPass(SimplifyCFGOptions() .forwardSwitchCondToPhi(true) + .convertSwitchRangeToICmp(true) .convertSwitchToLookupTable(true) .needCanonicalLoops(false) .hoistCommonInsts(true) @@ -1215,7 +1229,8 @@ PassBuilder::buildModuleOptimizationPipeline(OptimizationLevel Level, // LoopSink (and other loop passes since the last simplifyCFG) might have // resulted in single-entry-single-exit or empty blocks. Clean up the CFG. - OptimizePM.addPass(SimplifyCFGPass()); + OptimizePM.addPass( + SimplifyCFGPass(SimplifyCFGOptions().convertSwitchRangeToICmp(true))); OptimizePM.addPass(CoroCleanupPass()); @@ -1690,8 +1705,9 @@ PassBuilder::buildLTODefaultPipeline(OptimizationLevel Level, // Add late LTO optimization passes. // Delete basic blocks, which optimization passes may have killed. - MPM.addPass(createModuleToFunctionPassAdaptor( - SimplifyCFGPass(SimplifyCFGOptions().hoistCommonInsts(true)))); + MPM.addPass(createModuleToFunctionPassAdaptor(SimplifyCFGPass( + SimplifyCFGOptions().convertSwitchRangeToICmp(true).hoistCommonInsts( + true)))); // Drop bodies of available eternally objects to improve GlobalDCE. MPM.addPass(EliminateAvailableExternallyPass()); diff --git a/llvm/lib/Passes/PassRegistry.def b/llvm/lib/Passes/PassRegistry.def index 8e0af11b854d..69d8d8c43267 100644 --- a/llvm/lib/Passes/PassRegistry.def +++ b/llvm/lib/Passes/PassRegistry.def @@ -423,6 +423,7 @@ FUNCTION_PASS_WITH_PARAMS("simplifycfg", }, parseSimplifyCFGOptions, "no-forward-switch-cond;forward-switch-cond;" + "no-switch-range-to-icmp;switch-range-to-icmp;" "no-switch-to-lookup;switch-to-lookup;" "no-keep-loops;keep-loops;" "no-hoist-common-insts;hoist-common-insts;" diff --git a/llvm/lib/Target/AArch64/AArch64TargetMachine.cpp b/llvm/lib/Target/AArch64/AArch64TargetMachine.cpp index 4af28fc070dd..6a751da7ad55 100644 --- a/llvm/lib/Target/AArch64/AArch64TargetMachine.cpp +++ b/llvm/lib/Target/AArch64/AArch64TargetMachine.cpp @@ -531,6 +531,7 @@ void AArch64PassConfig::addIRPasses() { if (TM->getOptLevel() != CodeGenOpt::None && EnableAtomicTidy) addPass(createCFGSimplificationPass(SimplifyCFGOptions() .forwardSwitchCondToPhi(true) + .convertSwitchRangeToICmp(true) .convertSwitchToLookupTable(true) .needCanonicalLoops(false) .hoistCommonInsts(true) diff --git a/llvm/lib/Target/Hexagon/HexagonTargetMachine.cpp b/llvm/lib/Target/Hexagon/HexagonTargetMachine.cpp index c6703bb8a62a..08acf81961a3 100644 --- a/llvm/lib/Target/Hexagon/HexagonTargetMachine.cpp +++ b/llvm/lib/Target/Hexagon/HexagonTargetMachine.cpp @@ -345,6 +345,7 @@ void HexagonPassConfig::addIRPasses() { if (EnableInitialCFGCleanup) addPass(createCFGSimplificationPass(SimplifyCFGOptions() .forwardSwitchCondToPhi(true) + .convertSwitchRangeToICmp(true) .convertSwitchToLookupTable(true) .needCanonicalLoops(false) .hoistCommonInsts(true) diff --git a/llvm/lib/Transforms/IPO/PassManagerBuilder.cpp b/llvm/lib/Transforms/IPO/PassManagerBuilder.cpp index fcfc7a90cf8b..6e5aeb9c41f6 100644 --- a/llvm/lib/Transforms/IPO/PassManagerBuilder.cpp +++ b/llvm/lib/Transforms/IPO/PassManagerBuilder.cpp @@ -365,7 +365,9 @@ void PassManagerBuilder::addPGOInstrPasses(legacy::PassManagerBase &MPM, MPM.add(createFunctionInliningPass(IP)); MPM.add(createSROAPass()); MPM.add(createEarlyCSEPass()); // Catch trivial redundancies - MPM.add(createCFGSimplificationPass()); // Merge & remove BBs + MPM.add(createCFGSimplificationPass( + SimplifyCFGOptions().convertSwitchRangeToICmp( + true))); // Merge & remove BBs MPM.add(createInstructionCombiningPass()); // Combine silly seq's addExtensionsToPM(EP_Peephole, MPM); } @@ -404,7 +406,8 @@ void PassManagerBuilder::addFunctionSimplificationPasses( MPM.add(createGVNHoistPass()); if (EnableGVNSink) { MPM.add(createGVNSinkPass()); - MPM.add(createCFGSimplificationPass()); + MPM.add(createCFGSimplificationPass( + SimplifyCFGOptions().convertSwitchRangeToICmp(true))); } } @@ -418,7 +421,9 @@ void PassManagerBuilder::addFunctionSimplificationPasses( MPM.add(createJumpThreadingPass()); // Thread jumps. MPM.add(createCorrelatedValuePropagationPass()); // Propagate conditionals } - MPM.add(createCFGSimplificationPass()); // Merge & remove BBs + MPM.add( + createCFGSimplificationPass(SimplifyCFGOptions().convertSwitchRangeToICmp( + true))); // Merge & remove BBs // Combine silly seq's if (OptLevel > 2) MPM.add(createAggressiveInstCombinerPass()); @@ -434,7 +439,9 @@ void PassManagerBuilder::addFunctionSimplificationPasses( // TODO: Investigate the cost/benefit of tail call elimination on debugging. if (OptLevel > 1) MPM.add(createTailCallEliminationPass()); // Eliminate tail calls - MPM.add(createCFGSimplificationPass()); // Merge & remove BBs + MPM.add( + createCFGSimplificationPass(SimplifyCFGOptions().convertSwitchRangeToICmp( + true))); // Merge & remove BBs MPM.add(createReassociatePass()); // Reassociate expressions // The matrix extension can introduce large vector operations early, which can @@ -470,7 +477,8 @@ void PassManagerBuilder::addFunctionSimplificationPasses( // FIXME: We break the loop pass pipeline here in order to do full // simplifycfg. Eventually loop-simplifycfg should be enhanced to replace the // need for this. - MPM.add(createCFGSimplificationPass()); + MPM.add(createCFGSimplificationPass( + SimplifyCFGOptions().convertSwitchRangeToICmp(true))); MPM.add(createInstructionCombiningPass()); // We resume loop passes creating a second loop pipeline here. if (EnableLoopFlatten) { @@ -589,7 +597,8 @@ void PassManagerBuilder::addVectorPasses(legacy::PassManagerBase &PM, PM.add(createLICMPass(LicmMssaOptCap, LicmMssaNoAccForPromotionCap, /*AllowSpeculation=*/true)); PM.add(createLoopUnswitchPass(SizeLevel || OptLevel < 3, DivergentTarget)); - PM.add(createCFGSimplificationPass()); + PM.add(createCFGSimplificationPass( + SimplifyCFGOptions().convertSwitchRangeToICmp(true))); PM.add(createInstructionCombiningPass()); } @@ -604,6 +613,7 @@ void PassManagerBuilder::addVectorPasses(legacy::PassManagerBase &PM, // before SLP vectorization. PM.add(createCFGSimplificationPass(SimplifyCFGOptions() .forwardSwitchCondToPhi(true) + .convertSwitchRangeToICmp(true) .convertSwitchToLookupTable(true) .needCanonicalLoops(false) .hoistCommonInsts(true) @@ -780,7 +790,9 @@ void PassManagerBuilder::populateModulePassManager( MPM.add(createInstructionCombiningPass()); // Clean up after IPCP & DAE addExtensionsToPM(EP_Peephole, MPM); - MPM.add(createCFGSimplificationPass()); // Clean up after IPCP & DAE + MPM.add( + createCFGSimplificationPass(SimplifyCFGOptions().convertSwitchRangeToICmp( + true))); // Clean up after IPCP & DAE // For SamplePGO in ThinLTO compile phase, we do not want to do indirect // call promotion as it will change the CFG too much to make the 2nd @@ -981,7 +993,8 @@ void PassManagerBuilder::populateModulePassManager( // LoopSink (and other loop passes since the last simplifyCFG) might have // resulted in single-entry-single-exit or empty blocks. Clean up the CFG. - MPM.add(createCFGSimplificationPass()); + MPM.add(createCFGSimplificationPass( + SimplifyCFGOptions().convertSwitchRangeToICmp(true))); addExtensionsToPM(EP_OptimizerLast, MPM); diff --git a/llvm/lib/Transforms/Scalar/SimplifyCFGPass.cpp b/llvm/lib/Transforms/Scalar/SimplifyCFGPass.cpp index ee17da1875e5..b8972751066d 100644 --- a/llvm/lib/Transforms/Scalar/SimplifyCFGPass.cpp +++ b/llvm/lib/Transforms/Scalar/SimplifyCFGPass.cpp @@ -59,6 +59,11 @@ static cl::opt<bool> UserKeepLoops( "keep-loops", cl::Hidden, cl::init(true), cl::desc("Preserve canonical loop structure (default = true)")); +static cl::opt<bool> UserSwitchRangeToICmp( + "switch-range-to-icmp", cl::Hidden, cl::init(false), + cl::desc( + "Convert switches into an integer range comparison (default = false)")); + static cl::opt<bool> UserSwitchToLookup( "switch-to-lookup", cl::Hidden, cl::init(false), cl::desc("Convert switches to lookup tables (default = false)")); @@ -311,6 +316,8 @@ static void applyCommandLineOverridesToOptions(SimplifyCFGOptions &Options) { Options.BonusInstThreshold = UserBonusInstThreshold; if (UserForwardSwitchCond.getNumOccurrences()) Options.ForwardSwitchCondToPhi = UserForwardSwitchCond; + if (UserSwitchRangeToICmp.getNumOccurrences()) + Options.ConvertSwitchRangeToICmp = UserSwitchRangeToICmp; if (UserSwitchToLookup.getNumOccurrences()) Options.ConvertSwitchToLookupTable = UserSwitchToLookup; if (UserKeepLoops.getNumOccurrences()) @@ -337,6 +344,8 @@ void SimplifyCFGPass::printPipeline( OS << "<"; OS << "bonus-inst-threshold=" << Options.BonusInstThreshold << ";"; OS << (Options.ForwardSwitchCondToPhi ? "" : "no-") << "forward-switch-cond;"; + OS << (Options.ConvertSwitchRangeToICmp ? "" : "no-") + << "switch-range-to-icmp;"; OS << (Options.ConvertSwitchToLookupTable ? "" : "no-") << "switch-to-lookup;"; OS << (Options.NeedCanonicalLoop ? "" : "no-") << "keep-loops;"; diff --git a/llvm/lib/Transforms/Utils/SimplifyCFG.cpp b/llvm/lib/Transforms/Utils/SimplifyCFG.cpp index 335ac03ccb52..8c4e1b381b4d 100644 --- a/llvm/lib/Transforms/Utils/SimplifyCFG.cpp +++ b/llvm/lib/Transforms/Utils/SimplifyCFG.cpp @@ -6211,7 +6211,9 @@ bool SimplifyCFGOpt::simplifySwitch(SwitchInst *SI, IRBuilder<> &Builder) { } // Try to transform the switch into an icmp and a branch. - if (TurnSwitchRangeIntoICmp(SI, Builder)) + // The conversion from switch to comparison may lose information on + // impossible switch values, so disable it early in the pipeline. + if (Options.ConvertSwitchRangeToICmp && TurnSwitchRangeIntoICmp(SI, Builder)) return requestResimplify(); // Remove unreachable cases. diff --git a/llvm/test/Other/new-pm-print-pipeline.ll b/llvm/test/Other/new-pm-print-pipeline.ll index 20d624a96852..3abf54ff4b49 100644 --- a/llvm/test/Other/new-pm-print-pipeline.ll +++ b/llvm/test/Other/new-pm-print-pipeline.ll @@ -1,7 +1,7 @@ ;; Test that the -print-pipeline-passes option correctly prints some explicitly specified pipelines. ; RUN: opt -disable-output -disable-verify -print-pipeline-passes -passes='function(adce),function(simplifycfg<bonus-inst-threshold=123;no-forward-switch-cond;switch-to-lookup;keep-loops;no-hoist-common-insts;sink-common-insts>)' < %s | FileCheck %s --match-full-lines --check-prefixes=CHECK-0 -; CHECK-0: function(adce),function(simplifycfg<bonus-inst-threshold=123;no-forward-switch-cond;switch-to-lookup;keep-loops;no-hoist-common-insts;sink-common-insts>) +; CHECK-0: function(adce),function(simplifycfg<bonus-inst-threshold=123;no-forward-switch-cond;no-switch-range-to-icmp;switch-to-lookup;keep-loops;no-hoist-common-insts;sink-common-insts>) ; RUN: opt -disable-output -disable-verify -print-pipeline-passes -passes='module(rpo-function-attrs,require<globals-aa>,function(float2int,lower-constant-intrinsics,loop(loop-rotate)),invalidate<globals-aa>)' < %s | FileCheck %s --match-full-lines --check-prefixes=CHECK-1 ; CHECK-1: rpo-function-attrs,require<globals-aa>,function(float2int,lower-constant-intrinsics,loop(loop-rotate)),invalidate<globals-aa> @@ -56,7 +56,7 @@ ; CHECK-17: function(print<stack-lifetime><may>,print<stack-lifetime><must>) ; RUN: opt -disable-output -disable-verify -print-pipeline-passes -passes='function(simplifycfg<bonus-inst-threshold=5;forward-switch-cond;switch-to-lookup;keep-loops;hoist-common-insts;sink-common-insts>,simplifycfg<bonus-inst-threshold=7;no-forward-switch-cond;no-switch-to-lookup;no-keep-loops;no-hoist-common-insts;no-sink-common-insts>)' < %s | FileCheck %s --match-full-lines --check-prefixes=CHECK-18 -; CHECK-18: function(simplifycfg<bonus-inst-threshold=5;forward-switch-cond;switch-to-lookup;keep-loops;hoist-common-insts;sink-common-insts>,simplifycfg<bonus-inst-threshold=7;no-forward-switch-cond;no-switch-to-lookup;no-keep-loops;no-hoist-common-insts;no-sink-common-insts>) +; CHECK-18: function(simplifycfg<bonus-inst-threshold=5;forward-switch-cond;no-switch-range-to-icmp;switch-to-lookup;keep-loops;hoist-common-insts;sink-common-insts>,simplifycfg<bonus-inst-threshold=7;no-forward-switch-cond;no-switch-range-to-icmp;no-switch-to-lookup;no-keep-loops;no-hoist-common-insts;no-sink-common-insts>) ; RUN: opt -disable-output -disable-verify -print-pipeline-passes -passes='function(loop-vectorize<no-interleave-forced-only;no-vectorize-forced-only>,loop-vectorize<interleave-forced-only;vectorize-forced-only>)' < %s | FileCheck %s --match-full-lines --check-prefixes=CHECK-19 ; CHECK-19: function(loop-vectorize<no-interleave-forced-only;no-vectorize-forced-only;>,loop-vectorize<interleave-forced-only;vectorize-forced-only;>) diff --git a/llvm/test/Transforms/Coroutines/coro-catchswitch-cleanuppad.ll b/llvm/test/Transforms/Coroutines/coro-catchswitch-cleanuppad.ll index 4650820c84ea..c14094e5b48d 100644 --- a/llvm/test/Transforms/Coroutines/coro-catchswitch-cleanuppad.ll +++ b/llvm/test/Transforms/Coroutines/coro-catchswitch-cleanuppad.ll @@ -1,6 +1,6 @@ ; Tests the PHI nodes in cleanuppads for catchswitch instructions are correctly ; split up. -; RUN: opt < %s -passes='cgscc(coro-split),simplifycfg,early-cse' -S | FileCheck %s +; RUN: opt < %s -passes='cgscc(coro-split),simplifycfg<switch-range-to-icmp>,early-cse' -S | FileCheck %s declare i32 @__CxxFrameHandler3(...) define i8* @f2(i1 %val) "coroutine.presplit"="1" personality i32 (...)* @__CxxFrameHandler3 { diff --git a/llvm/test/Transforms/LoopUnroll/ARM/upperbound.ll b/llvm/test/Transforms/LoopUnroll/ARM/upperbound.ll index 33151c68b319..ffa474afcd50 100644 --- a/llvm/test/Transforms/LoopUnroll/ARM/upperbound.ll +++ b/llvm/test/Transforms/LoopUnroll/ARM/upperbound.ll @@ -1,5 +1,5 @@ ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py -; RUN: opt -loop-unroll -simplifycfg -instcombine -simplifycfg -S -mtriple arm-none-eabi -mcpu=cortex-m7 %s | FileCheck %s +; RUN: opt -loop-unroll -simplifycfg -switch-range-to-icmp -instcombine -simplifycfg -S -mtriple arm-none-eabi -mcpu=cortex-m7 %s | FileCheck %s ; This test is meant to check that this loop is unrolled into three iterations. define void @test(i32* %x, i32 %n) { diff --git a/llvm/test/Transforms/PhaseOrdering/simplifycfg-switch-lowering-vs-correlatedpropagation.ll b/llvm/test/Transforms/PhaseOrdering/simplifycfg-switch-lowering-vs-correlatedpropagation.ll index 4e1863304e6b..0ca042f9ad7f 100644 --- a/llvm/test/Transforms/PhaseOrdering/simplifycfg-switch-lowering-vs-correlatedpropagation.ll +++ b/llvm/test/Transforms/PhaseOrdering/simplifycfg-switch-lowering-vs-correlatedpropagation.ll @@ -14,11 +14,9 @@ define i64 @test1(i64 %x) { ; CHECK-LABEL: @test1( ; CHECK-NEXT: entry: -; CHECK-NEXT: [[SWITCH:%.*]] = icmp eq i64 [[X:%.*]], 0 -; CHECK-NEXT: [[TMP0:%.*]] = icmp eq i64 [[X]], 100 -; CHECK-NEXT: [[DOT:%.*]] = select i1 [[TMP0]], i64 200, i64 10 -; CHECK-NEXT: [[COMMON_RET_OP:%.*]] = select i1 [[SWITCH]], i64 0, i64 [[DOT]] -; CHECK-NEXT: ret i64 [[COMMON_RET_OP]] +; CHECK-NEXT: [[SWITCH_SELECTCMP:%.*]] = icmp eq i64 [[X:%.*]], 0 +; CHECK-NEXT: [[SWITCH_SELECT:%.*]] = select i1 [[SWITCH_SELECTCMP]], i64 0, i64 10 +; CHECK-NEXT: ret i64 [[SWITCH_SELECT]] ; entry: switch i64 %x, label %bb3 [ @@ -42,11 +40,9 @@ bb5: define i64 @test2(i64 %x) { ; CHECK-LABEL: @test2( ; CHECK-NEXT: entry: -; CHECK-NEXT: [[SWITCH_SELECTCMP:%.*]] = icmp eq i64 [[X:%.*]], 101 -; CHECK-NEXT: [[SWITCH_SELECT:%.*]] = select i1 [[SWITCH_SELECTCMP]], i64 200, i64 10 -; CHECK-NEXT: [[SWITCH_SELECTCMP1:%.*]] = icmp eq i64 [[X]], 1 -; CHECK-NEXT: [[SWITCH_SELECT2:%.*]] = select i1 [[SWITCH_SELECTCMP1]], i64 0, i64 [[SWITCH_SELECT]] -; CHECK-NEXT: ret i64 [[SWITCH_SELECT2]] +; CHECK-NEXT: [[SWITCH_SELECTCMP:%.*]] = icmp eq i64 [[X:%.*]], 1 +; CHECK-NEXT: [[SWITCH_SELECT:%.*]] = select i1 [[SWITCH_SELECTCMP]], i64 0, i64 10 +; CHECK-NEXT: ret i64 [[SWITCH_SELECT]] ; entry: switch i64 %x, label %bb3 [ @@ -96,10 +92,8 @@ define i64 @test_fail1(i64 %x) { ; CHECK-LABEL: @test_fail1( ; CHECK-NEXT: entry: ; CHECK-NEXT: [[SWITCH:%.*]] = icmp eq i64 [[X:%.*]], 0 -; CHECK-NEXT: [[TMP0:%.*]] = icmp eq i64 [[X]], 100 -; CHECK-NEXT: [[DOT:%.*]] = select i1 [[TMP0]], i64 200, i64 10 -; CHECK-NEXT: [[COMMON_RET_OP:%.*]] = select i1 [[SWITCH]], i64 0, i64 [[DOT]] -; CHECK-NEXT: ret i64 [[COMMON_RET_OP]] +; CHECK-NEXT: [[SPEC_SELECT:%.*]] = select i1 [[SWITCH]], i64 0, i64 10 +; CHECK-NEXT: ret i64 [[SPEC_SELECT]] ; entry: switch i64 %x, label %bb3 [ @@ -124,9 +118,9 @@ bb5: define i64 @test_fail2(i64 %x) { ; CHECK-LABEL: @test_fail2( ; CHECK-NEXT: entry: -; CHECK-NEXT: [[SWITCH:%.*]] = icmp eq i64 [[X:%.*]], 0 -; CHECK-NEXT: [[SPEC_SELECT:%.*]] = select i1 [[SWITCH]], i64 2, i64 [[X]] -; CHECK-NEXT: ret i64 [[SPEC_SELECT]] +; CHECK-NEXT: [[SWITCH_SELECTCMP:%.*]] = icmp eq i64 [[X:%.*]], 0 +; CHECK-NEXT: [[SWITCH_SELECT:%.*]] = select i1 [[SWITCH_SELECTCMP]], i64 2, i64 1 +; CHECK-NEXT: ret i64 [[SWITCH_SELECT]] ; entry: switch i64 %x, label %bb2 [ diff --git a/llvm/test/Transforms/SimplifyCFG/DeadSetCC.ll b/llvm/test/Transforms/SimplifyCFG/DeadSetCC.ll index fd322a938f8f..ff9c3be3397d 100644 --- a/llvm/test/Transforms/SimplifyCFG/DeadSetCC.ll +++ b/llvm/test/Transforms/SimplifyCFG/DeadSetCC.ll @@ -1,5 +1,5 @@ ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py -; RUN: opt < %s -simplifycfg -simplifycfg-require-and-preserve-domtree=1 -S | FileCheck %s +; RUN: opt < %s -simplifycfg -simplifycfg-require-and-preserve-domtree=1 -switch-range-to-icmp -S | FileCheck %s ; Check that simplifycfg deletes a dead 'seteq' instruction when it ; folds a conditional branch into a switch instruction. diff --git a/llvm/test/Transforms/SimplifyCFG/EqualPHIEdgeBlockMerge.ll b/llvm/test/Transforms/SimplifyCFG/EqualPHIEdgeBlockMerge.ll index 96cd0469a38d..c550abd9ad82 100644 --- a/llvm/test/Transforms/SimplifyCFG/EqualPHIEdgeBlockMerge.ll +++ b/llvm/test/Transforms/SimplifyCFG/EqualPHIEdgeBlockMerge.ll @@ -1,7 +1,7 @@ ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py ; Test merging of blocks with phi nodes. ; -; RUN: opt < %s -simplifycfg -simplifycfg-require-and-preserve-domtree=1 -S | FileCheck %s +; RUN: opt < %s -simplifycfg -simplifycfg-require-and-preserve-domtree=1 -switch-range-to-icmp -S | FileCheck %s ; ; ModuleID = '<stdin>' diff --git a/llvm/test/Transforms/SimplifyCFG/FoldValueComparisonIntoPredecessors-no-new-successors.ll b/llvm/test/Transforms/SimplifyCFG/FoldValueComparisonIntoPredecessors-no-new-successors.ll index 04d5d4d5a645..3fbd5bfe4574 100644 --- a/llvm/test/Transforms/SimplifyCFG/FoldValueComparisonIntoPredecessors-no-new-successors.ll +++ b/llvm/test/Transforms/SimplifyCFG/FoldValueComparisonIntoPredecessors-no-new-successors.ll @@ -1,5 +1,5 @@ ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py -; RUN: opt -S -simplifycfg -simplifycfg-require-and-preserve-domtree=1 < %s | FileCheck %s +; RUN: opt -S -simplifycfg -simplifycfg-require-and-preserve-domtree=1 -switch-range-to-icmp < %s | FileCheck %s define void @widget(i32 %arg) { ; CHECK-LABEL: @widget( diff --git a/llvm/test/Transforms/SimplifyCFG/ForwardSwitchConditionToPHI.ll b/llvm/test/Transforms/SimplifyCFG/ForwardSwitchConditionToPHI.ll index b57284d16688..bebe18abfb73 100644 --- a/llvm/test/Transforms/SimplifyCFG/ForwardSwitchConditionToPHI.ll +++ b/llvm/test/Transforms/SimplifyCFG/ForwardSwitchConditionToPHI.ll @@ -1,9 +1,9 @@ ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py -; RUN: opt < %s -simplifycfg -simplifycfg-require-and-preserve-domtree=1 -forward-switch-cond=false -S | FileCheck %s --check-prefix=NO_FWD -; RUN: opt < %s -simplifycfg -simplifycfg-require-and-preserve-domtree=1 -forward-switch-cond=true -S | FileCheck %s --check-prefix=FWD +; RUN: opt < %s -simplifycfg -simplifycfg-require-and-preserve-domtree=1 -forward-switch-cond=false -switch-range-to-icmp -S | FileCheck %s --check-prefix=NO_FWD +; RUN: opt < %s -simplifycfg -simplifycfg-require-and-preserve-domtree=1 -forward-switch-cond=true -switch-range-to-icmp -S | FileCheck %s --check-prefix=FWD -; RUN: opt < %s -passes='simplifycfg<no-forward-switch-cond>' -S | FileCheck %s --check-prefix=NO_FWD -; RUN: opt < %s -passes='simplifycfg<forward-switch-cond>' -S | FileCheck %s --check-prefix=FWD +; RUN: opt < %s -passes='simplifycfg<no-forward-switch-cond;switch-range-to-icmp>' -S | FileCheck %s --check-prefix=NO_FWD +; RUN: opt < %s -passes='simplifycfg<forward-switch-cond;switch-range-to-icmp>' -S | FileCheck %s --check-prefix=FWD ; PR10131 diff --git a/llvm/test/Transforms/SimplifyCFG/preserve-branchweights.ll b/llvm/test/Transforms/SimplifyCFG/preserve-branchweights.ll index f113ad84dc8f..fc200e041125 100644 --- a/llvm/test/Transforms/SimplifyCFG/preserve-branchweights.ll +++ b/llvm/test/Transforms/SimplifyCFG/preserve-branchweights.ll @@ -1,5 +1,5 @@ ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals -; RUN: opt -simplifycfg -simplifycfg-require-and-preserve-domtree=1 -S -o - < %s | FileCheck %s +; RUN: opt -simplifycfg -simplifycfg-require-and-preserve-domtree=1 -switch-range-to-icmp -S -o - < %s | FileCheck %s declare void @helper(i32) diff --git a/llvm/test/Transforms/SimplifyCFG/switch-dead-default.ll b/llvm/test/Transforms/SimplifyCFG/switch-dead-default.ll index 154ecb216631..1662bb99f27b 100644 --- a/llvm/test/Transforms/SimplifyCFG/switch-dead-default.ll +++ b/llvm/test/Transforms/SimplifyCFG/switch-dead-default.ll @@ -1,5 +1,5 @@ ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py -; RUN: opt %s -S -passes='simplifycfg<switch-to-lookup>' -simplifycfg-require-and-preserve-domtree=1 | FileCheck %s +; RUN: opt %s -S -passes='simplifycfg<switch-to-lookup>' -simplifycfg-require-and-preserve-domtree=1 -switch-range-to-icmp | FileCheck %s target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128" declare void @foo(i32) diff --git a/llvm/test/Transforms/SimplifyCFG/switch-range-to-icmp.ll b/llvm/test/Transforms/SimplifyCFG/switch-range-to-icmp.ll index 56a6895f29b8..5a6cb463bd6f 100644 --- a/llvm/test/Transforms/SimplifyCFG/switch-range-to-icmp.ll +++ b/llvm/test/Transforms/SimplifyCFG/switch-range-to-icmp.ll @@ -1,5 +1,5 @@ ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py -; RUN: opt %s -simplifycfg -simplifycfg-require-and-preserve-domtree=1 -S | FileCheck %s +; RUN: opt %s -simplifycfg -simplifycfg-require-and-preserve-domtree=1 -switch-range-to-icmp -S | FileCheck %s declare i32 @f(i32) diff --git a/llvm/test/Transforms/SimplifyCFG/switch-to-icmp.ll b/llvm/test/Transforms/SimplifyCFG/switch-to-icmp.ll index bfb27cd46114..c9ba839ec0a5 100644 --- a/llvm/test/Transforms/SimplifyCFG/switch-to-icmp.ll +++ b/llvm/test/Transforms/SimplifyCFG/switch-to-icmp.ll @@ -1,4 +1,4 @@ -; RUN: opt -S -simplifycfg -simplifycfg-require-and-preserve-domtree=1 < %s | FileCheck %s +; RUN: opt -S -simplifycfg -simplifycfg-require-and-preserve-domtree=1 -switch-range-to-icmp < %s | FileCheck %s define zeroext i1 @test1(i32 %x) nounwind readnone ssp noredzone { entry: diff --git a/llvm/test/Transforms/SimplifyCFG/switch_create-custom-dl.ll b/llvm/test/Transforms/SimplifyCFG/switch_create-custom-dl.ll index 499064e40a90..59064fbd8dc1 100644 --- a/llvm/test/Transforms/SimplifyCFG/switch_create-custom-dl.ll +++ b/llvm/test/Transforms/SimplifyCFG/switch_create-custom-dl.ll @@ -1,5 +1,5 @@ ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py -; RUN: opt -S -simplifycfg -simplifycfg-require-and-preserve-domtree=1 < %s | FileCheck %s +; RUN: opt -S -simplifycfg -simplifycfg-require-and-preserve-domtree=1 -switch-range-to-icmp < %s | FileCheck %s target datalayout="p:40:64:64:32" declare void @foo1() diff --git a/llvm/test/Transforms/SimplifyCFG/switch_create.ll b/llvm/test/Transforms/SimplifyCFG/switch_create.ll index 1ba0971fc4f0..d4ae9471a281 100644 --- a/llvm/test/Transforms/SimplifyCFG/switch_create.ll +++ b/llvm/test/Transforms/SimplifyCFG/switch_create.ll @@ -1,6 +1,6 @@ ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py -; RUN: opt -S -simplifycfg -simplifycfg-require-and-preserve-domtree=1 < %s | FileCheck %s -; RUN: opt -S -data-layout="p:32:32-p1:16:16" -simplifycfg -simplifycfg-require-and-preserve-domtree=1 < %s | FileCheck -check-prefix=CHECK -check-prefix=DL %s +; RUN: opt -S -simplifycfg -simplifycfg-require-and-preserve-domtree=1 -switch-range-to-icmp < %s | FileCheck %s +; RUN: opt -S -data-layout="p:32:32-p1:16:16" -simplifycfg -simplifycfg-require-and-preserve-domtree=1 -switch-range-to-icmp < %s | FileCheck -check-prefix=CHECK -check-prefix=DL %s declare void @foo1() |