From 4a16efa3e43e35f0cc9efe3a67f620f0017c3d36 Mon Sep 17 00:00:00 2001 From: Dimitry Andric Date: Mon, 8 Apr 2013 18:41:23 +0000 Subject: Vendor import of llvm trunk r178860: http://llvm.org/svn/llvm-project/llvm/trunk@178860 --- .../ArgumentPromotion/2008-02-01-ReturnAttrs.ll | 18 +- test/Transforms/ArgumentPromotion/crash.ll | 8 +- test/Transforms/BBVectorize/X86/pr15289.ll | 98 ++ test/Transforms/BBVectorize/X86/simple-int.ll | 79 ++ test/Transforms/BBVectorize/cycle.ll | 2 +- test/Transforms/BBVectorize/ld1.ll | 2 +- test/Transforms/BBVectorize/loop1.ll | 4 +- test/Transforms/BBVectorize/req-depth.ll | 4 +- test/Transforms/BBVectorize/search-limit.ll | 4 +- test/Transforms/BBVectorize/simple-int.ll | 38 +- test/Transforms/BBVectorize/simple-ldstr.ll | 4 +- test/Transforms/BBVectorize/simple-sel.ll | 4 +- test/Transforms/BBVectorize/simple.ll | 2 +- test/Transforms/BBVectorize/simple3.ll | 2 +- test/Transforms/CodeGenPrepare/basic.ll | 2 +- test/Transforms/ConstProp/2007-11-23-cttz.ll | 2 +- .../2003-10-28-MergeExternalConstants.ll | 2 +- .../ConstantMerge/2011-01-15-EitherOrder.ll | 2 +- test/Transforms/ConstantMerge/merge-both.ll | 2 +- test/Transforms/ConstantMerge/unnamed-addr.ll | 2 +- .../Transforms/CorrelatedValuePropagation/basic.ll | 22 +- .../DeadArgElim/2007-12-20-ParamAttrs.ll | 22 +- test/Transforms/DeadArgElim/2010-04-30-DbgInfo.ll | 9 +- test/Transforms/DeadArgElim/dbginfo.ll | 21 +- test/Transforms/DeadArgElim/deadexternal.ll | 2 +- test/Transforms/DeadArgElim/keepalive.ll | 7 +- .../DeadStoreElimination/const-pointers.ll | 2 +- test/Transforms/DeadStoreElimination/dominate.ll | 2 +- .../DeadStoreElimination/no-targetdata.ll | 2 +- test/Transforms/DeadStoreElimination/pr11390.ll | 2 +- test/Transforms/EarlyCSE/commute.ll | 4 +- test/Transforms/EarlyCSE/floatingpoint.ll | 14 + .../FunctionAttrs/2008-09-03-ReadNone.ll | 8 +- .../FunctionAttrs/2008-09-03-ReadOnly.ll | 10 +- .../FunctionAttrs/2009-01-04-Annotate.ll | 21 + test/Transforms/FunctionAttrs/annotate-1.ll | 18 + test/Transforms/FunctionAttrs/atomic.ll | 6 +- test/Transforms/FunctionAttrs/noreturn.ll | 18 + test/Transforms/GCOVProfiling/linkagename.ll | 27 + test/Transforms/GCOVProfiling/lit.local.cfg | 1 + test/Transforms/GCOVProfiling/version.ll | 29 + test/Transforms/GVN/2011-04-27-phioperands.ll | 2 +- test/Transforms/GVN/MemdepMiscompile.ll | 54 + test/Transforms/GVN/crash-no-aa.ll | 5 +- test/Transforms/GVN/crash.ll | 2 +- test/Transforms/GVN/edge.ll | 2 +- test/Transforms/GVN/fpmath.ll | 2 +- test/Transforms/GVN/lpre-call-wrap-2.ll | 2 +- test/Transforms/GVN/lpre-call-wrap.ll | 2 +- test/Transforms/GVN/nonescaping-malloc.ll | 1 + test/Transforms/GVN/null-aliases-nothing.ll | 2 +- test/Transforms/GVN/pr12979.ll | 2 +- test/Transforms/GVN/range.ll | 2 +- test/Transforms/GVN/rle.ll | 13 +- test/Transforms/GVN/tbaa.ll | 2 +- test/Transforms/GlobalOpt/2009-03-05-dbg.ll | 1 + .../GlobalOpt/2010-02-25-MallocPromote.ll | 2 +- test/Transforms/GlobalOpt/2010-02-26-MallocSROA.ll | 2 +- test/Transforms/GlobalOpt/crash-2.ll | 19 + test/Transforms/GlobalOpt/crash.ll | 2 +- .../GlobalOpt/ctor-list-opt-constexpr.ll | 2 +- .../GlobalOpt/externally-initialized-global-ctr.ll | 35 + test/Transforms/GlobalOpt/integer-bool.ll | 27 +- test/Transforms/GlobalOpt/memset-null.ll | 2 +- test/Transforms/GlobalOpt/unnamed-addr.ll | 2 +- .../IPConstantProp/user-with-multiple-uses.ll | 8 +- .../IndVarSimplify/2003-09-23-NotAtTop.ll | 2 +- test/Transforms/IndVarSimplify/crash.ll | 2 +- test/Transforms/IndVarSimplify/dont-recompute.ll | 69 ++ test/Transforms/IndVarSimplify/iv-zext.ll | 2 +- .../phi-uses-value-multiple-times.ll | 1 + .../Inline/2003-09-22-PHINodeInlineFail.ll | 10 +- .../2003-09-22-PHINodesInNormalInvokeDest.ll | 4 +- .../Inline/2006-11-09-InlineCGUpdate-2.ll | 1 - .../Transforms/Inline/2006-11-09-InlineCGUpdate.ll | 1 - test/Transforms/Inline/2010-05-12-ValueMap.ll | 2 +- test/Transforms/Inline/alloca_test.ll | 2 +- test/Transforms/Inline/basictest.ll | 45 + test/Transforms/Inline/crash2.ll | 2 +- test/Transforms/Inline/delete-call.ll | 8 +- test/Transforms/Inline/devirtualize-3.ll | 2 +- test/Transforms/Inline/devirtualize.ll | 2 +- test/Transforms/Inline/gvn-inline-iteration.ll | 2 +- test/Transforms/Inline/inline-optsize.ll | 4 +- test/Transforms/Inline/inline_constprop.ll | 76 ++ test/Transforms/Inline/inline_invoke.ll | 12 +- test/Transforms/Inline/inline_minisize.ll | 232 +++++ test/Transforms/Inline/inline_ssp.ll | 160 +++ test/Transforms/Inline/lifetime-no-datalayout.ll | 23 + test/Transforms/Inline/lifetime.ll | 46 +- test/Transforms/Inline/noinline-recursive-fn.ll | 2 +- test/Transforms/Inline/noinline.ll | 2 +- test/Transforms/Inline/recursive.ll | 2 +- .../InstCombine/2008-05-08-StrLenSink.ll | 2 +- .../InstCombine/2009-02-11-NotInitialized.ll | 14 + test/Transforms/InstCombine/2010-03-03-ExtElim.ll | 2 +- .../InstCombine/2010-05-30-memcpy-Struct.ll | 20 + .../Transforms/InstCombine/2010-11-01-lshr-mask.ll | 4 +- .../InstCombine/2012-04-23-Neon-Intrinsics.ll | 6 +- .../Transforms/InstCombine/2012-12-14-simp-vgep.ll | 10 + .../2013-03-05-Combine-BitcastTy-Into-Alloca.ll | 45 + test/Transforms/InstCombine/abs-1.ll | 41 + test/Transforms/InstCombine/align-external.ll | 2 +- test/Transforms/InstCombine/bitcast-bigendian.ll | 50 + test/Transforms/InstCombine/bitcast-vector-fold.ll | 5 + test/Transforms/InstCombine/bitcast.ll | 21 +- test/Transforms/InstCombine/cast.ll | 14 +- test/Transforms/InstCombine/compare-signs.ll | 2 +- .../InstCombine/constant-expr-datalayout.ll | 12 + test/Transforms/InstCombine/cos-1.ll | 38 + test/Transforms/InstCombine/cos-2.ll | 17 + test/Transforms/InstCombine/debug-line.ll | 24 + test/Transforms/InstCombine/debuginfo.ll | 32 +- test/Transforms/InstCombine/devirt.ll | 2 +- .../InstCombine/disable-simplify-libcalls.ll | 99 ++ .../InstCombine/double-float-shrink-1.ll | 333 ++++++ .../InstCombine/double-float-shrink-2.ll | 80 ++ test/Transforms/InstCombine/exact.ll | 7 +- test/Transforms/InstCombine/exp2-1.ll | 76 ++ test/Transforms/InstCombine/exp2-2.ll | 17 + test/Transforms/InstCombine/fast-math.ll | 467 +++++++++ test/Transforms/InstCombine/ffs-1.ll | 134 +++ test/Transforms/InstCombine/fmul.ll | 72 ++ test/Transforms/InstCombine/fold-phi.ll | 39 + test/Transforms/InstCombine/fpcast.ll | 19 + test/Transforms/InstCombine/fprintf-1.ll | 80 ++ test/Transforms/InstCombine/fputs-1.ll | 43 + test/Transforms/InstCombine/fwrite-1.ll | 57 ++ test/Transforms/InstCombine/getelementptr.ll | 20 +- test/Transforms/InstCombine/icmp.ll | 209 ++++ test/Transforms/InstCombine/idioms.ll | 2 +- test/Transforms/InstCombine/intrinsics.ll | 38 +- test/Transforms/InstCombine/isascii-1.ll | 32 + test/Transforms/InstCombine/isdigit-1.ll | 48 + test/Transforms/InstCombine/load-cmp.ll | 12 + test/Transforms/InstCombine/load3.ll | 25 +- test/Transforms/InstCombine/logical-select.ll | 20 +- test/Transforms/InstCombine/malloc-free-delete.ll | 29 + test/Transforms/InstCombine/memcmp-1.ll | 4 +- test/Transforms/InstCombine/memcpy-from-global.ll | 10 + test/Transforms/InstCombine/mul.ll | 8 +- test/Transforms/InstCombine/obfuscated_splat.ll | 2 +- test/Transforms/InstCombine/objsize.ll | 128 +++ test/Transforms/InstCombine/or.ll | 5 +- test/Transforms/InstCombine/osx-names.ll | 30 + test/Transforms/InstCombine/pow-1.ll | 154 +++ test/Transforms/InstCombine/pow-2.ll | 14 + test/Transforms/InstCombine/pr12338.ll | 42 +- test/Transforms/InstCombine/printf-1.ll | 119 +++ test/Transforms/InstCombine/printf-2.ll | 41 + test/Transforms/InstCombine/ptr-int-cast.ll | 31 + test/Transforms/InstCombine/puts-1.ll | 31 + test/Transforms/InstCombine/sdiv-1.ll | 4 +- test/Transforms/InstCombine/sext.ll | 9 + test/Transforms/InstCombine/shift.ll | 90 +- test/Transforms/InstCombine/signext.ll | 4 +- test/Transforms/InstCombine/sink_instruction.ll | 2 +- test/Transforms/InstCombine/sprintf-1.ll | 100 ++ test/Transforms/InstCombine/sqrt.ll | 2 +- test/Transforms/InstCombine/store.ll | 34 + test/Transforms/InstCombine/strto-1.ll | 16 +- test/Transforms/InstCombine/toascii-1.ll | 59 ++ test/Transforms/InstCombine/vec_extract_elt.ll | 10 + test/Transforms/InstCombine/vector-casts.ll | 3 +- test/Transforms/InstCombine/vector-type.ll | 15 + test/Transforms/InstCombine/vector_gep1.ll | 9 +- test/Transforms/InstCombine/xor2.ll | 31 + test/Transforms/InstCombine/zext-bool-add-sub.ll | 4 +- test/Transforms/InstSimplify/call-callconv.ll | 48 + test/Transforms/InstSimplify/call.ll | 103 ++ test/Transforms/InstSimplify/compare.ll | 86 ++ test/Transforms/InstSimplify/fast-math.ll | 107 ++ .../InstSimplify/floating-point-arithmetic.ll | 35 + test/Transforms/InstSimplify/past-the-end.ll | 77 ++ test/Transforms/InstSimplify/ptr_diff.ll | 30 + test/Transforms/InstSimplify/vector_gep.ll | 2 +- test/Transforms/JumpThreading/basic.ll | 40 +- test/Transforms/JumpThreading/degenerate-phi.ll | 2 +- test/Transforms/JumpThreading/or-undef.ll | 2 +- test/Transforms/LICM/2003-12-11-SinkingToPHI.ll | 2 +- test/Transforms/LICM/2011-07-06-Alignment.ll | 2 +- test/Transforms/LICM/crash.ll | 2 +- test/Transforms/LICM/hoist-invariant-load.ll | 1 + test/Transforms/LICM/hoisting.ll | 26 + test/Transforms/LICM/scalar_promote.ll | 125 ++- .../LoopDeletion/2011-06-21-phioperands.ll | 2 +- .../LoopDeletion/simplify-then-delete.ll | 4 +- test/Transforms/LoopIdiom/X86/lit.local.cfg | 6 + test/Transforms/LoopIdiom/X86/popcnt.ll | 140 +++ test/Transforms/LoopRotate/basic.ll | 28 +- test/Transforms/LoopRotate/crash.ll | 2 +- test/Transforms/LoopRotate/dbgvalue.ll | 2 +- test/Transforms/LoopRotate/phi-duplicate.ll | 2 +- .../LoopStrengthReduce/2008-08-14-ShadowIV.ll | 99 -- .../LoopStrengthReduce/2011-07-20-DoubleIV.ll | 43 - .../2012-07-18-LimitReassociate.ll | 14 +- .../LoopStrengthReduce/2013-01-05-IndBr.ll | 44 + .../LoopStrengthReduce/2013-01-14-ReuseCast.ll | 84 ++ .../LoopStrengthReduce/ARM/ivchain-ARM.ll | 101 +- .../LoopStrengthReduce/X86/2008-08-14-ShadowIV.ll | 99 ++ .../LoopStrengthReduce/X86/2011-07-20-DoubleIV.ll | 43 + .../LoopStrengthReduce/X86/2011-12-04-loserreg.ll | 13 +- .../LoopStrengthReduce/dominate-assert.ll | 2 +- .../LoopStrengthReduce/exit_compare_live_range.ll | 2 +- .../LoopStrengthReduce/post-inc-icmpzero.ll | 9 +- test/Transforms/LoopUnroll/basic.ll | 23 + test/Transforms/LoopUnroll/runtime-loop3.ll | 1 + .../LoopUnswitch/2008-11-03-Invariant.ll | 1 + .../LoopUnswitch/2011-11-18-SimpleSwitch.ll | 10 +- .../2011-11-18-TwoSwitches-Threshold.ll | 10 +- .../LoopUnswitch/2011-11-18-TwoSwitches.ll | 12 +- test/Transforms/LoopUnswitch/basictest.ll | 39 +- test/Transforms/LoopUnswitch/infinite-loop.ll | 9 +- test/Transforms/LoopUnswitch/preserve-analyses.ll | 2 +- test/Transforms/LoopVectorize/12-12-11-if-conv.ll | 44 + .../Transforms/LoopVectorize/2012-10-20-infloop.ll | 46 +- .../LoopVectorize/2012-10-22-isconsec.ll | 2 +- test/Transforms/LoopVectorize/ARM/arm-unroll.ll | 32 + test/Transforms/LoopVectorize/ARM/gcc-examples.ll | 60 ++ test/Transforms/LoopVectorize/ARM/lit.local.cfg | 6 + test/Transforms/LoopVectorize/ARM/mul-cast-vect.ll | 114 +++ test/Transforms/LoopVectorize/ARM/width-detect.ll | 52 + test/Transforms/LoopVectorize/X86/avx1.ll | 4 +- .../LoopVectorize/X86/constant-vector-operand.ll | 28 + .../LoopVectorize/X86/conversion-cost.ll | 11 +- test/Transforms/LoopVectorize/X86/cost-model.ll | 5 +- test/Transforms/LoopVectorize/X86/gcc-examples.ll | 27 +- .../LoopVectorize/X86/min-trip-count-switch.ll | 28 + test/Transforms/LoopVectorize/X86/no-vector.ll | 22 + .../X86/parallel-loops-after-reg2mem.ll | 52 + .../Transforms/LoopVectorize/X86/parallel-loops.ll | 114 +++ .../LoopVectorize/X86/reduction-crash.ll | 35 + test/Transforms/LoopVectorize/X86/small-size.ll | 170 +++ test/Transforms/LoopVectorize/X86/struct-store.ll | 27 + .../LoopVectorize/X86/unroll-small-loops.ll | 50 + .../LoopVectorize/X86/unroll_selection.ll | 71 ++ .../LoopVectorize/X86/vector-scalar-select-cost.ll | 66 ++ .../LoopVectorize/X86/vector_ptr_load_store.ll | 150 +++ .../Transforms/LoopVectorize/bzip_reverse_loops.ll | 71 ++ test/Transforms/LoopVectorize/calloc.ll | 53 + test/Transforms/LoopVectorize/cast-induction.ll | 30 + test/Transforms/LoopVectorize/cpp-new-array.ll | 4 +- test/Transforms/LoopVectorize/dbg.value.ll | 70 ++ test/Transforms/LoopVectorize/flags.ll | 2 +- test/Transforms/LoopVectorize/float-reduction.ll | 29 + test/Transforms/LoopVectorize/gcc-examples.ll | 57 +- test/Transforms/LoopVectorize/global_alias.ll | 1078 ++++++++++++++++++++ test/Transforms/LoopVectorize/i8-induction.ll | 35 + test/Transforms/LoopVectorize/if-conv-crash.ll | 39 + .../LoopVectorize/if-conversion-reduction.ll | 38 + test/Transforms/LoopVectorize/if-conversion.ll | 108 ++ test/Transforms/LoopVectorize/increment.ll | 2 +- test/Transforms/LoopVectorize/induction_plus.ll | 5 +- test/Transforms/LoopVectorize/intrinsic.ll | 935 +++++++++++++++++ test/Transforms/LoopVectorize/lcssa-crash.ll | 29 + test/Transforms/LoopVectorize/no_int_induction.ll | 33 + test/Transforms/LoopVectorize/nofloat.ll | 29 + test/Transforms/LoopVectorize/non-const-n.ll | 2 +- test/Transforms/LoopVectorize/nsw-crash.ll | 25 + test/Transforms/LoopVectorize/phi-hang.ll | 29 + test/Transforms/LoopVectorize/ptr_loops.ll | 74 ++ test/Transforms/LoopVectorize/read-only.ll | 2 +- test/Transforms/LoopVectorize/reduction.ll | 95 +- test/Transforms/LoopVectorize/runtime-check.ll | 6 +- test/Transforms/LoopVectorize/same-base-access.ll | 110 ++ test/Transforms/LoopVectorize/scalar-select.ll | 2 +- test/Transforms/LoopVectorize/simple-unroll.ll | 39 + test/Transforms/LoopVectorize/small-loop.ll | 2 +- test/Transforms/LoopVectorize/start-non-zero.ll | 2 +- test/Transforms/LoopVectorize/struct_access.ll | 50 + test/Transforms/LoopVectorize/vectorize-once.ll | 75 ++ test/Transforms/LoopVectorize/write-only.ll | 2 +- test/Transforms/Mem2Reg/ConvertDebugInfo.ll | 28 +- test/Transforms/Mem2Reg/ConvertDebugInfo2.ll | 27 +- test/Transforms/MemCpyOpt/memcpy.ll | 22 +- .../Transforms/MergeFunc/2011-02-08-RemoveEqual.ll | 2 +- .../MergeFunc/2013-01-10-MergeFuncAssert.ll | 36 + test/Transforms/MergeFunc/phi-speculation1.ll | 1 + test/Transforms/MergeFunc/phi-speculation2.ll | 1 + test/Transforms/MergeFunc/vector.ll | 1 + test/Transforms/MergeFunc/vectors-and-arrays.ll | 1 + test/Transforms/MetaRenamer/metarenamer.ll | 2 +- test/Transforms/ObjCARC/apelim.ll | 6 +- test/Transforms/ObjCARC/arc-annotations.ll | 307 ++++++ test/Transforms/ObjCARC/basic.ll | 55 +- test/Transforms/ObjCARC/cfg-hazards.ll | 42 +- ...clang-arc-used-intrinsic-removed-if-isolated.ll | 16 + test/Transforms/ObjCARC/contract-marker.ll | 4 +- test/Transforms/ObjCARC/contract-storestrong.ll | 30 +- test/Transforms/ObjCARC/contract-testcases.ll | 5 +- test/Transforms/ObjCARC/contract.ll | 37 +- ...ensure-that-exception-unwind-path-is-visited.ll | 174 ++++ test/Transforms/ObjCARC/escape.ll | 7 +- test/Transforms/ObjCARC/gvn.ll | 2 +- test/Transforms/ObjCARC/intrinsic-use.ll | 63 ++ test/Transforms/ObjCARC/invoke.ll | 16 +- .../ObjCARC/move-and-form-retain-autorelease.ll | 6 +- test/Transforms/ObjCARC/nested.ll | 9 +- test/Transforms/ObjCARC/no-objc-arc-exceptions.ll | 7 +- test/Transforms/ObjCARC/pr12270.ll | 2 +- test/Transforms/ObjCARC/retain-block-alloca.ll | 6 +- .../ObjCARC/retain-block-escape-analysis.ll | 127 +++ .../ObjCARC/retain-block-side-effects.ll | 5 +- test/Transforms/ObjCARC/retain-block.ll | 26 +- test/Transforms/ObjCARC/retain-not-declared.ll | 8 +- test/Transforms/ObjCARC/rle-s2l.ll | 7 +- test/Transforms/ObjCARC/rv.ll | 32 +- test/Transforms/ObjCARC/split-backedge.ll | 14 +- .../ObjCARC/tail-call-invariant-enforcement.ll | 74 ++ test/Transforms/ObjCARC/weak-copies.ll | 6 +- .../PhaseOrdering/2010-03-22-empty-baseclass.ll | 2 +- test/Transforms/PhaseOrdering/PR6627.ll | 2 +- test/Transforms/PhaseOrdering/basic.ll | 2 +- test/Transforms/PhaseOrdering/gdce.ll | 2 +- test/Transforms/PhaseOrdering/scev.ll | 2 +- test/Transforms/Reassociate/crash.ll | 2 +- test/Transforms/Reassociate/xor_reassoc.ll | 166 +++ test/Transforms/Reg2Mem/crash.ll | 88 ++ test/Transforms/Reg2Mem/lit.local.cfg | 1 + test/Transforms/SCCP/crash.ll | 2 +- test/Transforms/SCCP/ipsccp-addr-taken.ll | 2 +- test/Transforms/SCCP/retvalue-undef.ll | 2 +- test/Transforms/SCCP/undef-resolve.ll | 2 +- test/Transforms/SROA/basictest.ll | 97 +- test/Transforms/SROA/big-endian.ll | 16 +- test/Transforms/SROA/phi-and-select.ll | 12 +- test/Transforms/SROA/vector-promotion.ll | 164 ++- test/Transforms/SROA/vectors-of-pointers.ll | 25 + .../ScalarRepl/2003-09-12-IncorrectPromote.ll | 5 +- test/Transforms/ScalarRepl/crash.ll | 4 +- test/Transforms/ScalarRepl/debuginfo-preserved.ll | 21 +- test/Transforms/ScalarRepl/memcpy-align.ll | 2 +- test/Transforms/ScalarRepl/phi-cycle.ll | 5 +- test/Transforms/ScalarRepl/phi-select.ll | 2 +- test/Transforms/ScalarRepl/volatile.ll | 5 +- .../SimplifyCFG/2002-05-05-EmptyBlockMerge.ll | 22 - test/Transforms/SimplifyCFG/2002-06-24-PHINode.ll | 14 - .../SimplifyCFG/2010-03-30-InvokeCrash.ll | 2 +- test/Transforms/SimplifyCFG/EmptyBlockMerge.ll | 21 + test/Transforms/SimplifyCFG/PHINode.ll | 15 + test/Transforms/SimplifyCFG/PR9946.ll | 2 +- test/Transforms/SimplifyCFG/SpeculativeExec.ll | 41 + .../SimplifyCFG/X86/switch_to_lookup_table.ll | 26 + test/Transforms/SimplifyCFG/branch-fold-dbg.ll | 2 +- test/Transforms/SimplifyCFG/select-gep.ll | 2 +- .../SimplifyCFG/switch-on-const-select.ll | 9 +- test/Transforms/SimplifyCFG/trivial-throw.ll | 77 ++ test/Transforms/SimplifyCFG/volatile-phioper.ll | 48 + .../SimplifyLibCalls/2009-01-04-Annotate.ll | 12 - .../SimplifyLibCalls/2009-02-11-NotInitialized.ll | 13 - .../SimplifyLibCalls/2010-05-30-memcpy-Struct.ll | 20 - test/Transforms/SimplifyLibCalls/FFS.ll | 45 - test/Transforms/SimplifyLibCalls/FPrintF.ll | 28 - test/Transforms/SimplifyLibCalls/FPuts.ll | 29 - test/Transforms/SimplifyLibCalls/IsDigit.ll | 21 - test/Transforms/SimplifyLibCalls/Printf.ll | 37 - test/Transforms/SimplifyLibCalls/Puts.ll | 15 - test/Transforms/SimplifyLibCalls/SPrintF.ll | 40 - test/Transforms/SimplifyLibCalls/ToAscii.ll | 21 - test/Transforms/SimplifyLibCalls/abs.ll | 11 - test/Transforms/SimplifyLibCalls/cos.ll | 14 - test/Transforms/SimplifyLibCalls/debug-line.ll | 24 - .../SimplifyLibCalls/double-float-shrink.ll | 333 ------ test/Transforms/SimplifyLibCalls/exp2.ll | 38 - .../SimplifyLibCalls/float-shrink-compare.ll | 2 +- test/Transforms/SimplifyLibCalls/floor.ll | 85 -- test/Transforms/SimplifyLibCalls/fwrite.ll | 13 - test/Transforms/SimplifyLibCalls/iprintf.ll | 71 -- test/Transforms/SimplifyLibCalls/osx-names.ll | 30 - test/Transforms/SimplifyLibCalls/pow-to-sqrt.ll | 33 - test/Transforms/SimplifyLibCalls/pow2.ll | 37 - test/Transforms/StripSymbols/2010-08-25-crash.ll | 2 +- test/Transforms/StripSymbols/block-address.ll | 2 +- test/Transforms/TailCallElim/ackermann.ll | 1 + .../TailCallElim/dont-tce-tail-marked-call.ll | 4 +- test/Transforms/TailCallElim/dup_tail.ll | 5 +- test/Transforms/TailCallElim/intervening-inst.ll | 3 +- .../TailCallElim/move_alloca_for_tail_call.ll | 2 +- test/Transforms/TailCallElim/nocapture.ll | 2 +- test/Transforms/TailCallElim/reorder_load.ll | 6 +- test/Transforms/TailCallElim/return_constant.ll | 3 +- .../TailCallElim/trivial_codegen_tailcall.ll | 6 +- .../TailDup/2008-06-11-AvoidDupLoopHeader.ll | 1 + 383 files changed, 12014 insertions(+), 1808 deletions(-) create mode 100644 test/Transforms/BBVectorize/X86/pr15289.ll create mode 100644 test/Transforms/BBVectorize/X86/simple-int.ll create mode 100644 test/Transforms/EarlyCSE/floatingpoint.ll create mode 100644 test/Transforms/FunctionAttrs/2009-01-04-Annotate.ll create mode 100644 test/Transforms/FunctionAttrs/annotate-1.ll create mode 100644 test/Transforms/FunctionAttrs/noreturn.ll create mode 100644 test/Transforms/GCOVProfiling/linkagename.ll create mode 100644 test/Transforms/GCOVProfiling/lit.local.cfg create mode 100644 test/Transforms/GCOVProfiling/version.ll create mode 100644 test/Transforms/GVN/MemdepMiscompile.ll create mode 100644 test/Transforms/GlobalOpt/crash-2.ll create mode 100644 test/Transforms/GlobalOpt/externally-initialized-global-ctr.ll create mode 100644 test/Transforms/IndVarSimplify/dont-recompute.ll create mode 100644 test/Transforms/Inline/inline_minisize.ll create mode 100644 test/Transforms/Inline/inline_ssp.ll create mode 100644 test/Transforms/Inline/lifetime-no-datalayout.ll create mode 100644 test/Transforms/InstCombine/2009-02-11-NotInitialized.ll create mode 100644 test/Transforms/InstCombine/2010-05-30-memcpy-Struct.ll create mode 100644 test/Transforms/InstCombine/2012-12-14-simp-vgep.ll create mode 100644 test/Transforms/InstCombine/2013-03-05-Combine-BitcastTy-Into-Alloca.ll create mode 100644 test/Transforms/InstCombine/abs-1.ll create mode 100644 test/Transforms/InstCombine/bitcast-bigendian.ll create mode 100644 test/Transforms/InstCombine/constant-expr-datalayout.ll create mode 100644 test/Transforms/InstCombine/cos-1.ll create mode 100644 test/Transforms/InstCombine/cos-2.ll create mode 100644 test/Transforms/InstCombine/debug-line.ll create mode 100644 test/Transforms/InstCombine/double-float-shrink-1.ll create mode 100644 test/Transforms/InstCombine/double-float-shrink-2.ll create mode 100644 test/Transforms/InstCombine/exp2-1.ll create mode 100644 test/Transforms/InstCombine/exp2-2.ll create mode 100644 test/Transforms/InstCombine/fast-math.ll create mode 100644 test/Transforms/InstCombine/ffs-1.ll create mode 100644 test/Transforms/InstCombine/fmul.ll create mode 100644 test/Transforms/InstCombine/fold-phi.ll create mode 100644 test/Transforms/InstCombine/fprintf-1.ll create mode 100644 test/Transforms/InstCombine/fputs-1.ll create mode 100644 test/Transforms/InstCombine/fwrite-1.ll create mode 100644 test/Transforms/InstCombine/isascii-1.ll create mode 100644 test/Transforms/InstCombine/isdigit-1.ll create mode 100644 test/Transforms/InstCombine/osx-names.ll create mode 100644 test/Transforms/InstCombine/pow-1.ll create mode 100644 test/Transforms/InstCombine/pow-2.ll create mode 100644 test/Transforms/InstCombine/printf-1.ll create mode 100644 test/Transforms/InstCombine/printf-2.ll create mode 100644 test/Transforms/InstCombine/puts-1.ll create mode 100644 test/Transforms/InstCombine/sprintf-1.ll create mode 100644 test/Transforms/InstCombine/toascii-1.ll create mode 100644 test/Transforms/InstCombine/vector-type.ll create mode 100644 test/Transforms/InstSimplify/call-callconv.ll create mode 100644 test/Transforms/InstSimplify/call.ll create mode 100644 test/Transforms/InstSimplify/fast-math.ll create mode 100644 test/Transforms/InstSimplify/floating-point-arithmetic.ll create mode 100644 test/Transforms/InstSimplify/past-the-end.ll create mode 100644 test/Transforms/LoopIdiom/X86/lit.local.cfg create mode 100644 test/Transforms/LoopIdiom/X86/popcnt.ll delete mode 100644 test/Transforms/LoopStrengthReduce/2008-08-14-ShadowIV.ll delete mode 100644 test/Transforms/LoopStrengthReduce/2011-07-20-DoubleIV.ll create mode 100644 test/Transforms/LoopStrengthReduce/2013-01-05-IndBr.ll create mode 100644 test/Transforms/LoopStrengthReduce/2013-01-14-ReuseCast.ll create mode 100644 test/Transforms/LoopStrengthReduce/X86/2008-08-14-ShadowIV.ll create mode 100644 test/Transforms/LoopStrengthReduce/X86/2011-07-20-DoubleIV.ll create mode 100644 test/Transforms/LoopVectorize/12-12-11-if-conv.ll create mode 100644 test/Transforms/LoopVectorize/ARM/arm-unroll.ll create mode 100644 test/Transforms/LoopVectorize/ARM/gcc-examples.ll create mode 100644 test/Transforms/LoopVectorize/ARM/lit.local.cfg create mode 100644 test/Transforms/LoopVectorize/ARM/mul-cast-vect.ll create mode 100644 test/Transforms/LoopVectorize/ARM/width-detect.ll create mode 100644 test/Transforms/LoopVectorize/X86/constant-vector-operand.ll create mode 100644 test/Transforms/LoopVectorize/X86/min-trip-count-switch.ll create mode 100644 test/Transforms/LoopVectorize/X86/no-vector.ll create mode 100644 test/Transforms/LoopVectorize/X86/parallel-loops-after-reg2mem.ll create mode 100644 test/Transforms/LoopVectorize/X86/parallel-loops.ll create mode 100644 test/Transforms/LoopVectorize/X86/reduction-crash.ll create mode 100644 test/Transforms/LoopVectorize/X86/small-size.ll create mode 100644 test/Transforms/LoopVectorize/X86/struct-store.ll create mode 100644 test/Transforms/LoopVectorize/X86/unroll-small-loops.ll create mode 100644 test/Transforms/LoopVectorize/X86/unroll_selection.ll create mode 100644 test/Transforms/LoopVectorize/X86/vector-scalar-select-cost.ll create mode 100644 test/Transforms/LoopVectorize/X86/vector_ptr_load_store.ll create mode 100644 test/Transforms/LoopVectorize/bzip_reverse_loops.ll create mode 100644 test/Transforms/LoopVectorize/calloc.ll create mode 100644 test/Transforms/LoopVectorize/cast-induction.ll create mode 100644 test/Transforms/LoopVectorize/dbg.value.ll create mode 100644 test/Transforms/LoopVectorize/float-reduction.ll create mode 100644 test/Transforms/LoopVectorize/global_alias.ll create mode 100644 test/Transforms/LoopVectorize/i8-induction.ll create mode 100644 test/Transforms/LoopVectorize/if-conv-crash.ll create mode 100644 test/Transforms/LoopVectorize/if-conversion-reduction.ll create mode 100644 test/Transforms/LoopVectorize/if-conversion.ll create mode 100644 test/Transforms/LoopVectorize/intrinsic.ll create mode 100644 test/Transforms/LoopVectorize/lcssa-crash.ll create mode 100644 test/Transforms/LoopVectorize/no_int_induction.ll create mode 100644 test/Transforms/LoopVectorize/nofloat.ll create mode 100644 test/Transforms/LoopVectorize/nsw-crash.ll create mode 100644 test/Transforms/LoopVectorize/phi-hang.ll create mode 100644 test/Transforms/LoopVectorize/ptr_loops.ll create mode 100644 test/Transforms/LoopVectorize/same-base-access.ll create mode 100644 test/Transforms/LoopVectorize/simple-unroll.ll create mode 100644 test/Transforms/LoopVectorize/struct_access.ll create mode 100644 test/Transforms/LoopVectorize/vectorize-once.ll create mode 100644 test/Transforms/MergeFunc/2013-01-10-MergeFuncAssert.ll create mode 100644 test/Transforms/ObjCARC/arc-annotations.ll create mode 100644 test/Transforms/ObjCARC/clang-arc-used-intrinsic-removed-if-isolated.ll create mode 100644 test/Transforms/ObjCARC/ensure-that-exception-unwind-path-is-visited.ll create mode 100644 test/Transforms/ObjCARC/intrinsic-use.ll create mode 100644 test/Transforms/ObjCARC/retain-block-escape-analysis.ll create mode 100644 test/Transforms/ObjCARC/tail-call-invariant-enforcement.ll create mode 100644 test/Transforms/Reassociate/xor_reassoc.ll create mode 100644 test/Transforms/Reg2Mem/crash.ll create mode 100644 test/Transforms/Reg2Mem/lit.local.cfg create mode 100644 test/Transforms/SROA/vectors-of-pointers.ll delete mode 100644 test/Transforms/SimplifyCFG/2002-05-05-EmptyBlockMerge.ll delete mode 100644 test/Transforms/SimplifyCFG/2002-06-24-PHINode.ll create mode 100644 test/Transforms/SimplifyCFG/EmptyBlockMerge.ll create mode 100644 test/Transforms/SimplifyCFG/PHINode.ll create mode 100644 test/Transforms/SimplifyCFG/trivial-throw.ll create mode 100644 test/Transforms/SimplifyCFG/volatile-phioper.ll delete mode 100644 test/Transforms/SimplifyLibCalls/2009-01-04-Annotate.ll delete mode 100644 test/Transforms/SimplifyLibCalls/2009-02-11-NotInitialized.ll delete mode 100644 test/Transforms/SimplifyLibCalls/2010-05-30-memcpy-Struct.ll delete mode 100644 test/Transforms/SimplifyLibCalls/FFS.ll delete mode 100644 test/Transforms/SimplifyLibCalls/FPrintF.ll delete mode 100644 test/Transforms/SimplifyLibCalls/FPuts.ll delete mode 100644 test/Transforms/SimplifyLibCalls/IsDigit.ll delete mode 100644 test/Transforms/SimplifyLibCalls/Printf.ll delete mode 100644 test/Transforms/SimplifyLibCalls/Puts.ll delete mode 100644 test/Transforms/SimplifyLibCalls/SPrintF.ll delete mode 100644 test/Transforms/SimplifyLibCalls/ToAscii.ll delete mode 100644 test/Transforms/SimplifyLibCalls/abs.ll delete mode 100644 test/Transforms/SimplifyLibCalls/cos.ll delete mode 100644 test/Transforms/SimplifyLibCalls/debug-line.ll delete mode 100644 test/Transforms/SimplifyLibCalls/double-float-shrink.ll delete mode 100644 test/Transforms/SimplifyLibCalls/exp2.ll delete mode 100644 test/Transforms/SimplifyLibCalls/floor.ll delete mode 100644 test/Transforms/SimplifyLibCalls/fwrite.ll delete mode 100644 test/Transforms/SimplifyLibCalls/iprintf.ll delete mode 100644 test/Transforms/SimplifyLibCalls/osx-names.ll delete mode 100644 test/Transforms/SimplifyLibCalls/pow-to-sqrt.ll delete mode 100644 test/Transforms/SimplifyLibCalls/pow2.ll (limited to 'test/Transforms') diff --git a/test/Transforms/ArgumentPromotion/2008-02-01-ReturnAttrs.ll b/test/Transforms/ArgumentPromotion/2008-02-01-ReturnAttrs.ll index e740b29f9288b..1226b98a998ed 100644 --- a/test/Transforms/ArgumentPromotion/2008-02-01-ReturnAttrs.ll +++ b/test/Transforms/ArgumentPromotion/2008-02-01-ReturnAttrs.ll @@ -1,15 +1,19 @@ -; RUN: opt < %s -argpromotion -S | grep nounwind | count 2 +; RUN: opt < %s -argpromotion -S | FileCheck %s +; CHECK: define internal i32 @deref(i32 %x.val) #0 { define internal i32 @deref(i32* %x) nounwind { entry: - %tmp2 = load i32* %x, align 4 ; [#uses=1] - ret i32 %tmp2 + %tmp2 = load i32* %x, align 4 + ret i32 %tmp2 } define i32 @f(i32 %x) { entry: - %x_addr = alloca i32 ; [#uses=2] - store i32 %x, i32* %x_addr, align 4 - %tmp1 = call i32 @deref( i32* %x_addr ) nounwind ; [#uses=1] - ret i32 %tmp1 + %x_addr = alloca i32 + store i32 %x, i32* %x_addr, align 4 +; CHECK: %tmp1 = call i32 @deref(i32 %x_addr.val) [[NUW:#[0-9]+]] + %tmp1 = call i32 @deref( i32* %x_addr ) nounwind + ret i32 %tmp1 } + +; CHECK: attributes [[NUW]] = { nounwind } diff --git a/test/Transforms/ArgumentPromotion/crash.ll b/test/Transforms/ArgumentPromotion/crash.ll index fed002aa98a9f..5e1a0370dbb16 100644 --- a/test/Transforms/ArgumentPromotion/crash.ll +++ b/test/Transforms/ArgumentPromotion/crash.ll @@ -1,7 +1,5 @@ +; RUN: opt -inline -argpromotion < %s ; rdar://7879828 -; RUN: opt -inline -argpromotion %s -target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64" -target triple = "x86_64-apple-darwin10.0.0" define void @foo() { invoke void @foo2() @@ -11,6 +9,8 @@ if.end432: unreachable for.end520: + %exn = landingpad {i8*, i32} personality i32 (...)* @__gxx_personality_v0 + cleanup unreachable } @@ -57,3 +57,5 @@ init: %4 = call i32 @"clay_assign(Chain, Chain)"(%0* %3, %0* %1) ret i32 0 } + +declare i32 @__gxx_personality_v0(...) diff --git a/test/Transforms/BBVectorize/X86/pr15289.ll b/test/Transforms/BBVectorize/X86/pr15289.ll new file mode 100644 index 0000000000000..07cc5d8b96b7e --- /dev/null +++ b/test/Transforms/BBVectorize/X86/pr15289.ll @@ -0,0 +1,98 @@ +; RUN: opt < %s -basicaa -bb-vectorize -disable-output +; This is a bugpoint-reduced test case. It did not always assert, but does reproduce the bug +; and running under valgrind (or some similar tool) will catch the error. + +target datalayout = "e-p:64:64:64-S128-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f16:16:16-f32:32:32-f64:64:64-f128:128:128-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64" +target triple = "x86_64-apple-darwin12.2.0" + +%0 = type { [10 x { float, float }], [10 x { float, float }], [10 x { float, float }], [10 x { float, float }], [10 x { float, float }] } +%1 = type { [10 x [8 x i8]] } +%2 = type { i64, i64 } +%3 = type { [10 x i64], i64, i64, i64, i64, i64 } +%4 = type { i64, i64, i64, i64, i64, i64 } +%5 = type { [10 x i64] } +%6 = type { [10 x float], [10 x float], [10 x float], [10 x float] } +%struct.__st_parameter_dt.1.3.5.7 = type { %struct.__st_parameter_common.0.2.4.6, i64, i64*, i64*, i8*, i8*, i32, i32, i8*, i8*, i32, i32, i8*, [256 x i8], i32*, i64, i8*, i32, i32, i8*, i8*, i32, i32, i8*, i8*, i32, i32, i8*, i8*, i32, [4 x i8] } +%struct.__st_parameter_common.0.2.4.6 = type { i32, i32, i8*, i32, i32, i8*, i32* } + +@cctenso_ = external unnamed_addr global %0, align 32 +@ctenso_ = external unnamed_addr global %1, align 32 +@i_dim_ = external unnamed_addr global %2, align 16 +@itenso1_ = external unnamed_addr global %3, align 32 +@itenso2_ = external unnamed_addr global %4, align 32 +@ltenso_ = external unnamed_addr global %5, align 32 +@rtenso_ = external unnamed_addr global %6, align 32 +@.cst = external unnamed_addr constant [8 x i8], align 8 +@.cst1 = external unnamed_addr constant [3 x i8], align 8 +@.cst2 = external unnamed_addr constant [29 x i8], align 8 +@.cst3 = external unnamed_addr constant [32 x i8], align 64 + +define void @cart_to_dc2y_(double* noalias nocapture %xx, double* noalias nocapture %yy, double* noalias nocapture %zz, [5 x { double, double }]* noalias nocapture %c2ten) nounwind uwtable { +entry: + %0 = fmul double undef, undef + %1 = fmul double undef, undef + %2 = fadd double undef, undef + %3 = fmul double undef, 0x3FE8B8B76E3E9919 + %4 = fsub double %0, %1 + %5 = fsub double -0.000000e+00, undef + %6 = fmul double undef, undef + %7 = fmul double %4, %6 + %8 = fmul double undef, 2.000000e+00 + %9 = fmul double %8, undef + %10 = fmul double undef, %9 + %11 = fmul double %10, undef + %12 = fsub double undef, %7 + %13 = fmul double %3, %12 + %14 = fmul double %3, undef + %15 = getelementptr inbounds [5 x { double, double }]* %c2ten, i64 0, i64 0, i32 0 + store double %13, double* %15, align 8, !tbaa !0 + %16 = getelementptr inbounds [5 x { double, double }]* %c2ten, i64 0, i64 0, i32 1 + %17 = fmul double undef, %8 + %18 = fmul double %17, undef + %19 = fmul double undef, %18 + %20 = fadd double undef, undef + %21 = fmul double %3, %19 + %22 = fsub double -0.000000e+00, %21 + %23 = getelementptr inbounds [5 x { double, double }]* %c2ten, i64 0, i64 1, i32 0 + store double %22, double* %23, align 8, !tbaa !0 + %24 = getelementptr inbounds [5 x { double, double }]* %c2ten, i64 0, i64 1, i32 1 + %25 = fmul double undef, 0x3FE42F601A8C6794 + %26 = fmul double undef, 2.000000e+00 + %27 = fsub double %26, %0 + %28 = fmul double %6, undef + %29 = fsub double undef, %28 + %30 = getelementptr inbounds [5 x { double, double }]* %c2ten, i64 0, i64 2, i32 0 + store double undef, double* %30, align 8, !tbaa !0 + %31 = getelementptr inbounds [5 x { double, double }]* %c2ten, i64 0, i64 2, i32 1 + %32 = fmul double undef, %17 + %33 = fmul double undef, %17 + %34 = fmul double undef, %32 + %35 = fmul double undef, %33 + %36 = fsub double undef, %35 + %37 = fmul double %3, %34 + %38 = getelementptr inbounds [5 x { double, double }]* %c2ten, i64 0, i64 3, i32 0 + store double %37, double* %38, align 8, !tbaa !0 + %39 = getelementptr inbounds [5 x { double, double }]* %c2ten, i64 0, i64 3, i32 1 + %40 = fmul double undef, %8 + %41 = fmul double undef, %40 + %42 = fmul double undef, %41 + %43 = fsub double undef, %42 + %44 = fmul double %3, %43 + %45 = getelementptr inbounds [5 x { double, double }]* %c2ten, i64 0, i64 4, i32 0 + store double %13, double* %45, align 8, !tbaa !0 + %46 = getelementptr inbounds [5 x { double, double }]* %c2ten, i64 0, i64 4, i32 1 + %47 = fsub double -0.000000e+00, %14 + store double %47, double* %16, align 8, !tbaa !0 + store double undef, double* %24, align 8, !tbaa !0 + store double -0.000000e+00, double* %31, align 8, !tbaa !0 + store double undef, double* %39, align 8, !tbaa !0 + store double undef, double* %46, align 8, !tbaa !0 + ret void +} + +attributes #0 = { nounwind uwtable } +attributes #1 = { nounwind readnone } +attributes #2 = { nounwind } + +!0 = metadata !{metadata !"alias set 17: real(kind=8)", metadata !1} +!1 = metadata !{metadata !1} diff --git a/test/Transforms/BBVectorize/X86/simple-int.ll b/test/Transforms/BBVectorize/X86/simple-int.ll new file mode 100644 index 0000000000000..f5dbe46b1480b --- /dev/null +++ b/test/Transforms/BBVectorize/X86/simple-int.ll @@ -0,0 +1,79 @@ +target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128" +; RUN: opt < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=corei7 -bb-vectorize -bb-vectorize-req-chain-depth=3 -instcombine -gvn -S | FileCheck %s + +declare double @llvm.fma.f64(double, double, double) +declare double @llvm.fmuladd.f64(double, double, double) +declare double @llvm.cos.f64(double) +declare double @llvm.powi.f64(double, i32) + +; Basic depth-3 chain with fma +define double @test1(double %A1, double %A2, double %B1, double %B2, double %C1, double %C2) { + %X1 = fsub double %A1, %B1 + %X2 = fsub double %A2, %B2 + %Y1 = call double @llvm.fma.f64(double %X1, double %A1, double %C1) + %Y2 = call double @llvm.fma.f64(double %X2, double %A2, double %C2) + %Z1 = fadd double %Y1, %B1 + %Z2 = fadd double %Y2, %B2 + %R = fmul double %Z1, %Z2 + ret double %R +; CHECK: @test1 +; CHECK: ret double %R +} + +; Basic depth-3 chain with fmuladd +define double @test1a(double %A1, double %A2, double %B1, double %B2, double %C1, double %C2) { + %X1 = fsub double %A1, %B1 + %X2 = fsub double %A2, %B2 + %Y1 = call double @llvm.fmuladd.f64(double %X1, double %A1, double %C1) + %Y2 = call double @llvm.fmuladd.f64(double %X2, double %A2, double %C2) + %Z1 = fadd double %Y1, %B1 + %Z2 = fadd double %Y2, %B2 + %R = fmul double %Z1, %Z2 + ret double %R +; CHECK: @test1a +; CHECK: ret double %R +} + +; Basic depth-3 chain with cos +define double @test2(double %A1, double %A2, double %B1, double %B2) { + %X1 = fsub double %A1, %B1 + %X2 = fsub double %A2, %B2 + %Y1 = call double @llvm.cos.f64(double %X1) + %Y2 = call double @llvm.cos.f64(double %X2) + %Z1 = fadd double %Y1, %B1 + %Z2 = fadd double %Y2, %B2 + %R = fmul double %Z1, %Z2 + ret double %R +; CHECK: @test2 +; CHECK: ret double %R +} + +; Basic depth-3 chain with powi +define double @test3(double %A1, double %A2, double %B1, double %B2, i32 %P) { + %X1 = fsub double %A1, %B1 + %X2 = fsub double %A2, %B2 + %Y1 = call double @llvm.powi.f64(double %X1, i32 %P) + %Y2 = call double @llvm.powi.f64(double %X2, i32 %P) + %Z1 = fadd double %Y1, %B1 + %Z2 = fadd double %Y2, %B2 + %R = fmul double %Z1, %Z2 + ret double %R +; CHECK: @test3 +; CHECK: ret double %R +} + +; Basic depth-3 chain with powi (different powers: should not vectorize) +define double @test4(double %A1, double %A2, double %B1, double %B2, i32 %P) { + %X1 = fsub double %A1, %B1 + %X2 = fsub double %A2, %B2 + %P2 = add i32 %P, 1 + %Y1 = call double @llvm.powi.f64(double %X1, i32 %P) + %Y2 = call double @llvm.powi.f64(double %X2, i32 %P2) + %Z1 = fadd double %Y1, %B1 + %Z2 = fadd double %Y2, %B2 + %R = fmul double %Z1, %Z2 + ret double %R +; CHECK: @test4 +; CHECK: ret double %R +} + diff --git a/test/Transforms/BBVectorize/cycle.ll b/test/Transforms/BBVectorize/cycle.ll index e8e82ce02479f..bdcb30da887f0 100644 --- a/test/Transforms/BBVectorize/cycle.ll +++ b/test/Transforms/BBVectorize/cycle.ll @@ -1,5 +1,5 @@ target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128" -; RUN: opt < %s -bb-vectorize -bb-vectorize-req-chain-depth=3 -instcombine -gvn -S | FileCheck %s +; RUN: opt < %s -bb-vectorize -bb-vectorize-req-chain-depth=3 -bb-vectorize-ignore-target-info -instcombine -gvn -S | FileCheck %s ; This test checks the non-trivial pairing-induced cycle avoidance. Without this cycle avoidance, the algorithm would otherwise ; want to select the pairs: diff --git a/test/Transforms/BBVectorize/ld1.ll b/test/Transforms/BBVectorize/ld1.ll index cea225d076e1c..ea5cb5dd93f7a 100644 --- a/test/Transforms/BBVectorize/ld1.ll +++ b/test/Transforms/BBVectorize/ld1.ll @@ -1,5 +1,5 @@ target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128" -; RUN: opt < %s -bb-vectorize -bb-vectorize-req-chain-depth=3 -instcombine -gvn -S | FileCheck %s +; RUN: opt < %s -bb-vectorize -bb-vectorize-req-chain-depth=3 -bb-vectorize-ignore-target-info -instcombine -gvn -S | FileCheck %s define double @test1(double* %a, double* %b, double* %c) nounwind uwtable readonly { entry: diff --git a/test/Transforms/BBVectorize/loop1.ll b/test/Transforms/BBVectorize/loop1.ll index c22ea5852a1bd..e592edb44a02f 100644 --- a/test/Transforms/BBVectorize/loop1.ll +++ b/test/Transforms/BBVectorize/loop1.ll @@ -1,7 +1,7 @@ target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128" target triple = "x86_64-unknown-linux-gnu" -; RUN: opt < %s -bb-vectorize -bb-vectorize-req-chain-depth=3 -instcombine -gvn -S | FileCheck %s -; RUN: opt < %s -basicaa -loop-unroll -unroll-threshold=45 -unroll-allow-partial -bb-vectorize -bb-vectorize-req-chain-depth=3 -instcombine -gvn -S | FileCheck %s -check-prefix=CHECK-UNRL +; RUN: opt < %s -bb-vectorize -bb-vectorize-req-chain-depth=3 -bb-vectorize-ignore-target-info -instcombine -gvn -S | FileCheck %s +; RUN: opt < %s -basicaa -loop-unroll -unroll-threshold=45 -unroll-allow-partial -bb-vectorize -bb-vectorize-req-chain-depth=3 -bb-vectorize-ignore-target-info -instcombine -gvn -S | FileCheck %s -check-prefix=CHECK-UNRL ; The second check covers the use of alias analysis (with loop unrolling). define void @test1(double* noalias %out, double* noalias %in1, double* noalias %in2) nounwind uwtable { diff --git a/test/Transforms/BBVectorize/req-depth.ll b/test/Transforms/BBVectorize/req-depth.ll index 8c9cc3c188e39..e0120059b9529 100644 --- a/test/Transforms/BBVectorize/req-depth.ll +++ b/test/Transforms/BBVectorize/req-depth.ll @@ -1,6 +1,6 @@ target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128" -; RUN: opt < %s -bb-vectorize -bb-vectorize-req-chain-depth 3 -S | FileCheck %s -check-prefix=CHECK-RD3 -; RUN: opt < %s -bb-vectorize -bb-vectorize-req-chain-depth 2 -S | FileCheck %s -check-prefix=CHECK-RD2 +; RUN: opt < %s -bb-vectorize -bb-vectorize-req-chain-depth 3 -bb-vectorize-ignore-target-info -S | FileCheck %s -check-prefix=CHECK-RD3 +; RUN: opt < %s -bb-vectorize -bb-vectorize-req-chain-depth 2 -bb-vectorize-ignore-target-info -S | FileCheck %s -check-prefix=CHECK-RD2 define double @test1(double %A1, double %A2, double %B1, double %B2) { %X1 = fsub double %A1, %B1 diff --git a/test/Transforms/BBVectorize/search-limit.ll b/test/Transforms/BBVectorize/search-limit.ll index aeaf98865bc97..a694e45bc1817 100644 --- a/test/Transforms/BBVectorize/search-limit.ll +++ b/test/Transforms/BBVectorize/search-limit.ll @@ -1,6 +1,6 @@ target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128" -; RUN: opt < %s -bb-vectorize -bb-vectorize-req-chain-depth=3 -instcombine -gvn -S | FileCheck %s -; RUN: opt < %s -bb-vectorize -bb-vectorize-req-chain-depth=3 -bb-vectorize-search-limit=4 -instcombine -gvn -S | FileCheck %s -check-prefix=CHECK-SL4 +; RUN: opt < %s -bb-vectorize -bb-vectorize-req-chain-depth=3 -bb-vectorize-ignore-target-info -instcombine -gvn -S | FileCheck %s +; RUN: opt < %s -bb-vectorize -bb-vectorize-req-chain-depth=3 -bb-vectorize-search-limit=4 -bb-vectorize-ignore-target-info -instcombine -gvn -S | FileCheck %s -check-prefix=CHECK-SL4 define double @test1(double %A1, double %A2, double %B1, double %B2) { ; CHECK: @test1 diff --git a/test/Transforms/BBVectorize/simple-int.ll b/test/Transforms/BBVectorize/simple-int.ll index ae1d63bfd852b..e4d51526ca118 100644 --- a/test/Transforms/BBVectorize/simple-int.ll +++ b/test/Transforms/BBVectorize/simple-int.ll @@ -1,7 +1,8 @@ target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128" -; RUN: opt < %s -bb-vectorize -bb-vectorize-req-chain-depth=3 -instcombine -gvn -S | FileCheck %s +; RUN: opt < %s -bb-vectorize -bb-vectorize-req-chain-depth=3 -bb-vectorize-ignore-target-info -instcombine -gvn -S | FileCheck %s declare double @llvm.fma.f64(double, double, double) +declare double @llvm.fmuladd.f64(double, double, double) declare double @llvm.cos.f64(double) declare double @llvm.powi.f64(double, i32) @@ -31,6 +32,32 @@ define double @test1(double %A1, double %A2, double %B1, double %B2, double %C1, ; CHECK: ret double %R } +; Basic depth-3 chain with fmuladd +define double @test1a(double %A1, double %A2, double %B1, double %B2, double %C1, double %C2) { + %X1 = fsub double %A1, %B1 + %X2 = fsub double %A2, %B2 + %Y1 = call double @llvm.fmuladd.f64(double %X1, double %A1, double %C1) + %Y2 = call double @llvm.fmuladd.f64(double %X2, double %A2, double %C2) + %Z1 = fadd double %Y1, %B1 + %Z2 = fadd double %Y2, %B2 + %R = fmul double %Z1, %Z2 + ret double %R +; CHECK: @test1a +; CHECK: %X1.v.i1.1 = insertelement <2 x double> undef, double %B1, i32 0 +; CHECK: %X1.v.i1.2 = insertelement <2 x double> %X1.v.i1.1, double %B2, i32 1 +; CHECK: %X1.v.i0.1 = insertelement <2 x double> undef, double %A1, i32 0 +; CHECK: %X1.v.i0.2 = insertelement <2 x double> %X1.v.i0.1, double %A2, i32 1 +; CHECK: %X1 = fsub <2 x double> %X1.v.i0.2, %X1.v.i1.2 +; CHECK: %Y1.v.i2.1 = insertelement <2 x double> undef, double %C1, i32 0 +; CHECK: %Y1.v.i2.2 = insertelement <2 x double> %Y1.v.i2.1, double %C2, i32 1 +; CHECK: %Y1 = call <2 x double> @llvm.fmuladd.v2f64(<2 x double> %X1, <2 x double> %X1.v.i0.2, <2 x double> %Y1.v.i2.2) +; CHECK: %Z1 = fadd <2 x double> %Y1, %X1.v.i1.2 +; CHECK: %Z1.v.r1 = extractelement <2 x double> %Z1, i32 0 +; CHECK: %Z1.v.r2 = extractelement <2 x double> %Z1, i32 1 +; CHECK: %R = fmul double %Z1.v.r1, %Z1.v.r2 +; CHECK: ret double %R +} + ; Basic depth-3 chain with cos define double @test2(double %A1, double %A2, double %B1, double %B2) { %X1 = fsub double %A1, %B1 @@ -97,7 +124,10 @@ define double @test4(double %A1, double %A2, double %B1, double %B2, i32 %P) { ; CHECK: ret double %R } -; CHECK: declare <2 x double> @llvm.fma.v2f64(<2 x double>, <2 x double>, <2 x double>) nounwind readnone -; CHECK: declare <2 x double> @llvm.cos.v2f64(<2 x double>) nounwind readonly -; CHECK: declare <2 x double> @llvm.powi.v2f64(<2 x double>, i32) nounwind readonly +; CHECK: declare <2 x double> @llvm.fma.v2f64(<2 x double>, <2 x double>, <2 x double>) #0 +; CHECK: declare <2 x double> @llvm.fmuladd.v2f64(<2 x double>, <2 x double>, <2 x double>) #0 +; CHECK: declare <2 x double> @llvm.cos.v2f64(<2 x double>) #1 +; CHECK: declare <2 x double> @llvm.powi.v2f64(<2 x double>, i32) #1 +; CHECK: attributes #0 = { nounwind readnone } +; CHECK: attributes #1 = { nounwind readonly } diff --git a/test/Transforms/BBVectorize/simple-ldstr.ll b/test/Transforms/BBVectorize/simple-ldstr.ll index 7dd77c933f6dc..8e51d297e8ec8 100644 --- a/test/Transforms/BBVectorize/simple-ldstr.ll +++ b/test/Transforms/BBVectorize/simple-ldstr.ll @@ -1,6 +1,6 @@ target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128" -; RUN: opt < %s -bb-vectorize -bb-vectorize-req-chain-depth=3 -instcombine -gvn -S | FileCheck %s -; RUN: opt < %s -bb-vectorize -bb-vectorize-req-chain-depth=3 -bb-vectorize-aligned-only -instcombine -gvn -S | FileCheck %s -check-prefix=CHECK-AO +; RUN: opt < %s -bb-vectorize -bb-vectorize-req-chain-depth=3 -bb-vectorize-ignore-target-info -instcombine -gvn -S | FileCheck %s +; RUN: opt < %s -bb-vectorize -bb-vectorize-req-chain-depth=3 -bb-vectorize-aligned-only -bb-vectorize-ignore-target-info -instcombine -gvn -S | FileCheck %s -check-prefix=CHECK-AO ; Simple 3-pair chain with loads and stores define void @test1(double* %a, double* %b, double* %c) nounwind uwtable readonly { diff --git a/test/Transforms/BBVectorize/simple-sel.ll b/test/Transforms/BBVectorize/simple-sel.ll index 15ecb597025aa..8caccfd32c343 100644 --- a/test/Transforms/BBVectorize/simple-sel.ll +++ b/test/Transforms/BBVectorize/simple-sel.ll @@ -1,6 +1,6 @@ target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128" -; RUN: opt < %s -bb-vectorize -bb-vectorize-req-chain-depth=3 -instcombine -gvn -S | FileCheck %s -; RUN: opt < %s -bb-vectorize -bb-vectorize-req-chain-depth=3 -bb-vectorize-no-bools -instcombine -gvn -S | FileCheck %s -check-prefix=CHECK-NB +; RUN: opt < %s -bb-vectorize -bb-vectorize-req-chain-depth=3 -bb-vectorize-ignore-target-info -instcombine -gvn -S | FileCheck %s +; RUN: opt < %s -bb-vectorize -bb-vectorize-req-chain-depth=3 -bb-vectorize-no-bools -bb-vectorize-ignore-target-info -instcombine -gvn -S | FileCheck %s -check-prefix=CHECK-NB ; Basic depth-3 chain with select define double @test1(double %A1, double %A2, double %B1, double %B2, i1 %C1, i1 %C2) { diff --git a/test/Transforms/BBVectorize/simple.ll b/test/Transforms/BBVectorize/simple.ll index 3527ae75b457d..a447908d16cc7 100644 --- a/test/Transforms/BBVectorize/simple.ll +++ b/test/Transforms/BBVectorize/simple.ll @@ -1,5 +1,5 @@ target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128" -; RUN: opt < %s -bb-vectorize -bb-vectorize-req-chain-depth=3 -instcombine -gvn -S | FileCheck %s +; RUN: opt < %s -bb-vectorize -bb-vectorize-req-chain-depth=3 -bb-vectorize-ignore-target-info -instcombine -gvn -S | FileCheck %s ; Basic depth-3 chain define double @test1(double %A1, double %A2, double %B1, double %B2) { diff --git a/test/Transforms/BBVectorize/simple3.ll b/test/Transforms/BBVectorize/simple3.ll index 153be73f832f6..78bcc9f83080b 100644 --- a/test/Transforms/BBVectorize/simple3.ll +++ b/test/Transforms/BBVectorize/simple3.ll @@ -1,5 +1,5 @@ target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128" -; RUN: opt < %s -bb-vectorize -bb-vectorize-req-chain-depth=3 -bb-vectorize-vector-bits=192 -instcombine -gvn -S | FileCheck %s +; RUN: opt < %s -bb-vectorize -bb-vectorize-req-chain-depth=3 -bb-vectorize-vector-bits=192 -bb-vectorize-ignore-target-info -instcombine -gvn -S | FileCheck %s ; Basic depth-3 chain define double @test1(double %A1, double %A2, double %A3, double %B1, double %B2, double %B3) { diff --git a/test/Transforms/CodeGenPrepare/basic.ll b/test/Transforms/CodeGenPrepare/basic.ll index c68e77eb555a0..d617e43be8653 100644 --- a/test/Transforms/CodeGenPrepare/basic.ll +++ b/test/Transforms/CodeGenPrepare/basic.ll @@ -1,4 +1,4 @@ -; RUN: opt -codegenprepare %s -S -o - | FileCheck %s +; RUN: opt -codegenprepare -S < %s | FileCheck %s target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64" target triple = "x86_64-apple-darwin10.0.0" diff --git a/test/Transforms/ConstProp/2007-11-23-cttz.ll b/test/Transforms/ConstProp/2007-11-23-cttz.ll index 6d34cb17fffd9..c5ee70c2ff120 100644 --- a/test/Transforms/ConstProp/2007-11-23-cttz.ll +++ b/test/Transforms/ConstProp/2007-11-23-cttz.ll @@ -3,6 +3,6 @@ declare i13 @llvm.cttz.i13(i13, i1) define i13 @test() { - %X = call i13 @llvm.cttz.i13(i13 0, i1 true) + %X = call i13 @llvm.cttz.i13(i13 0, i1 false) ret i13 %X } diff --git a/test/Transforms/ConstantMerge/2003-10-28-MergeExternalConstants.ll b/test/Transforms/ConstantMerge/2003-10-28-MergeExternalConstants.ll index ce79e3b2964a5..a415995070e54 100644 --- a/test/Transforms/ConstantMerge/2003-10-28-MergeExternalConstants.ll +++ b/test/Transforms/ConstantMerge/2003-10-28-MergeExternalConstants.ll @@ -1,4 +1,4 @@ -; RUN: opt -S -constmerge %s | FileCheck %s +; RUN: opt -S -constmerge < %s | FileCheck %s ; CHECK: @foo = constant i32 6 ; CHECK: @bar = constant i32 6 diff --git a/test/Transforms/ConstantMerge/2011-01-15-EitherOrder.ll b/test/Transforms/ConstantMerge/2011-01-15-EitherOrder.ll index f561daf667815..5aafcfe3d4fb3 100644 --- a/test/Transforms/ConstantMerge/2011-01-15-EitherOrder.ll +++ b/test/Transforms/ConstantMerge/2011-01-15-EitherOrder.ll @@ -1,4 +1,4 @@ -; RUN: opt -constmerge %s -S -o - | FileCheck %s +; RUN: opt -constmerge -S < %s | FileCheck %s ; PR8978 declare i32 @zed(%struct.foobar*, %struct.foobar*) diff --git a/test/Transforms/ConstantMerge/merge-both.ll b/test/Transforms/ConstantMerge/merge-both.ll index b71eb437dbc3f..b00345557c83a 100644 --- a/test/Transforms/ConstantMerge/merge-both.ll +++ b/test/Transforms/ConstantMerge/merge-both.ll @@ -1,4 +1,4 @@ -; RUN: opt -constmerge %s -S -o - | FileCheck %s +; RUN: opt -constmerge -S < %s | FileCheck %s ; Test that in one run var3 is merged into var2 and var1 into var4. ; Test that we merge @var5 and @var6 into one with the higher alignment, and ; don't merge var7/var8 into var5/var6. diff --git a/test/Transforms/ConstantMerge/unnamed-addr.ll b/test/Transforms/ConstantMerge/unnamed-addr.ll index 24100837aabd6..aff8540f2cb11 100644 --- a/test/Transforms/ConstantMerge/unnamed-addr.ll +++ b/test/Transforms/ConstantMerge/unnamed-addr.ll @@ -1,4 +1,4 @@ -; RUN: opt -constmerge %s -S -o - | FileCheck %s +; RUN: opt -constmerge -S < %s | FileCheck %s ; Test which corresponding x and y are merged and that unnamed_addr ; is correctly set. diff --git a/test/Transforms/CorrelatedValuePropagation/basic.ll b/test/Transforms/CorrelatedValuePropagation/basic.ll index 475cd8d772e67..39c437ccfae98 100644 --- a/test/Transforms/CorrelatedValuePropagation/basic.ll +++ b/test/Transforms/CorrelatedValuePropagation/basic.ll @@ -81,6 +81,26 @@ LessThanOrEqualToTwo: ret i32 0 } +declare i32* @f(i32*) +define void @test5(i32* %x, i32* %y) { +; CHECK: @test5 +entry: + %pre = icmp eq i32* %x, null + br i1 %pre, label %return, label %loop + +loop: + %phi = phi i32* [ %sel, %loop ], [ %x, %entry ] +; CHECK: %phi = phi i32* [ %f, %loop ], [ %x, %entry ] + %f = tail call i32* @f(i32* %phi) + %cmp1 = icmp ne i32* %f, %y + %sel = select i1 %cmp1, i32* %f, i32* null + %cmp2 = icmp eq i32* %sel, null + br i1 %cmp2, label %return, label %loop + +return: + ret void +} + define i32 @switch1(i32 %s) { ; CHECK: @switch1 entry: @@ -105,7 +125,7 @@ negative: ] out: - %p = phi i32 [ 1, %entry ], [ -1, %negative ], [ -1, %negative ], [ -1, %negative ], [ -1, %negative ] + %p = phi i32 [ 1, %entry ], [ -1, %negative ], [ -1, %negative ], [ -1, %negative ], [ -1, %negative ], [ -1, %negative ] ret i32 %p next: diff --git a/test/Transforms/DeadArgElim/2007-12-20-ParamAttrs.ll b/test/Transforms/DeadArgElim/2007-12-20-ParamAttrs.ll index 7c6c575ea80f7..f049265ce4ea4 100644 --- a/test/Transforms/DeadArgElim/2007-12-20-ParamAttrs.ll +++ b/test/Transforms/DeadArgElim/2007-12-20-ParamAttrs.ll @@ -1,20 +1,20 @@ -; RUN: opt < %s -deadargelim -S > %t -; RUN: cat %t | grep nounwind | count 2 -; RUN: cat %t | grep signext | count 2 -; RUN: cat %t | not grep inreg -; RUN: cat %t | not grep zeroext -; RUN: cat %t | not grep byval +; RUN: opt < %s -deadargelim -S | FileCheck %s - %struct = type { } +%struct = type { } @g = global i8 0 +; CHECK: define internal void @foo(i8 signext %y) [[NUW:#[0-9]+]] + define internal zeroext i8 @foo(i8* inreg %p, i8 signext %y, ... ) nounwind { - store i8 %y, i8* @g - ret i8 0 + store i8 %y, i8* @g + ret i8 0 } define i32 @bar() { - %A = call zeroext i8(i8*, i8, ...)* @foo(i8* inreg null, i8 signext 1, %struct* byval null ) nounwind - ret i32 0 +; CHECK: call void @foo(i8 signext 1) [[NUW]] + %A = call zeroext i8(i8*, i8, ...)* @foo(i8* inreg null, i8 signext 1, %struct* byval null ) nounwind + ret i32 0 } + +; CHECK: attributes [[NUW]] = { nounwind } diff --git a/test/Transforms/DeadArgElim/2010-04-30-DbgInfo.ll b/test/Transforms/DeadArgElim/2010-04-30-DbgInfo.ll index 2f820bad84747..f5d2588dd0597 100644 --- a/test/Transforms/DeadArgElim/2010-04-30-DbgInfo.ll +++ b/test/Transforms/DeadArgElim/2010-04-30-DbgInfo.ll @@ -8,14 +8,14 @@ entry: call void @llvm.dbg.value(metadata !{i32 %len}, i64 0, metadata !10) call void @llvm.dbg.value(metadata !{i32 %hash}, i64 0, metadata !11) call void @llvm.dbg.value(metadata !{i32 %flags}, i64 0, metadata !12) -; CHECK: call fastcc i8* @add_name_internal(i8* %name, i32 %hash) nounwind, !dbg !13 +; CHECK: call fastcc i8* @add_name_internal(i8* %name, i32 %hash) [[NUW:#[0-9]+]], !dbg !13 %0 = call fastcc i8* @add_name_internal(i8* %name, i32 %len, i32 %hash, i8 zeroext 0, i32 %flags) nounwind, !dbg !13 ; [#uses=1] ret i8* %0, !dbg !13 } declare void @llvm.dbg.declare(metadata, metadata) nounwind readnone -define internal fastcc i8* @add_name_internal(i8* %name, i32 %len, i32 %hash, i8 zeroext %extra, i32 %flags) nounwind noinline ssp { +define internal fastcc i8* @add_name_internal(i8* %name, i32 %len, i32 %hash, i8 zeroext %extra, i32 %flags) noinline nounwind ssp { entry: call void @llvm.dbg.value(metadata !{i8* %name}, i64 0, metadata !15) call void @llvm.dbg.value(metadata !{i32 %len}, i64 0, metadata !20) @@ -38,6 +38,11 @@ bb2: ; preds = %bb1, %bb declare void @llvm.dbg.value(metadata, i64, metadata) nounwind readnone +; CHECK: attributes #0 = { nounwind ssp } +; CHECK: attributes #1 = { nounwind readnone } +; CHECK: attributes #2 = { noinline nounwind ssp } +; CHECK: attributes [[NUW]] = { nounwind } + !0 = metadata !{i32 524545, metadata !1, metadata !"name", metadata !2, i32 8, metadata !6} ; [ DW_TAG_arg_variable ] !1 = metadata !{i32 524334, i32 0, metadata !2, metadata !"vfs_addname", metadata !"vfs_addname", metadata !"vfs_addname", metadata !2, i32 12, metadata !4, i1 false, i1 true, i32 0, i32 0, null, i1 false} ; [ DW_TAG_subprogram ] !2 = metadata !{i32 524329, metadata !"tail.c", metadata !"/Users/echeng/LLVM/radars/r7927803/", metadata !3} ; [ DW_TAG_file_type ] diff --git a/test/Transforms/DeadArgElim/dbginfo.ll b/test/Transforms/DeadArgElim/dbginfo.ll index dcbfaaa3d77b8..24448b7009ed5 100644 --- a/test/Transforms/DeadArgElim/dbginfo.ll +++ b/test/Transforms/DeadArgElim/dbginfo.ll @@ -1,4 +1,4 @@ -; RUN: opt %s -deadargelim -S | FileCheck %s +; RUN: opt -deadargelim -S < %s | FileCheck %s ; PR14016 ; Check that debug info metadata for subprograms stores pointers to @@ -36,19 +36,17 @@ entry: !llvm.dbg.cu = !{!0} -!0 = metadata !{i32 786449, i32 0, i32 4, metadata !"test.cc", metadata !"/home/samsonov/tmp/clang-di", metadata !"clang version 3.2 (trunk 165305)", i1 true, i1 false, metadata !"", i32 0, metadata !1, metadata !1, metadata !3, metadata !1} ; [ DW_TAG_compile_unit ] [/home/samsonov/tmp/clang-di/test.cc] [DW_LANG_C_plus_plus] -!1 = metadata !{metadata !2} -!2 = metadata !{i32 0} -!3 = metadata !{metadata !4} -!4 = metadata !{metadata !5, metadata !8, metadata !9} -!5 = metadata !{i32 786478, i32 0, metadata !6, metadata !"run", metadata !"run", metadata !"", metadata !6, i32 8, metadata !7, i1 false, i1 true, i32 0, i32 0, null, i32 256, i1 false, void ()* @_Z3runv, null, null, metadata !1, i32 8} ; [ DW_TAG_subprogram ] [line 8] [def] [run] -!6 = metadata !{i32 786473, metadata !"test.cc", metadata !"/home/samsonov/tmp/clang-di", null} ; [ DW_TAG_file_type ] -!7 = metadata !{i32 786453, i32 0, metadata !"", i32 0, i32 0, i64 0, i64 0, i64 0, i32 0, null, metadata !2, i32 0, i32 0} ; [ DW_TAG_subroutine_type ] [line 0, size 0, align 0, offset 0] [from ] -!8 = metadata !{i32 786478, i32 0, metadata !6, metadata !"dead_vararg", metadata !"dead_vararg", metadata !"", metadata !6, i32 5, metadata !7, i1 true, i1 true, i32 0, i32 0, null, i32 256, i1 false, void (...)* @_ZN12_GLOBAL__N_111dead_varargEz, null, null, metadata !1, i32 5} ; [ DW_TAG_subprogram ] [line 5] [local] [def] [dead_vararg] +!0 = metadata !{i32 786449, i32 4, metadata !6, metadata !"clang version 3.2 (trunk 165305)", i1 false, metadata !"", i32 0, metadata !1, metadata !1, metadata !3, metadata !1, metadata !""} ; [ DW_TAG_compile_unit ] [/home/samsonov/tmp/clang-di/test.cc] [DW_LANG_C_plus_plus] +!1 = metadata !{i32 0} +!3 = metadata !{metadata !5, metadata !8, metadata !9} +!5 = metadata !{i32 786478, metadata !6, metadata !"run", metadata !"run", metadata !"", metadata !6, i32 8, metadata !7, i1 false, i1 true, i32 0, i32 0, null, i32 256, i1 false, void ()* @_Z3runv, null, null, metadata !1, i32 8} ; [ DW_TAG_subprogram ] [line 8] [def] [run] +!6 = metadata !{i32 786473, metadata !20} ; [ DW_TAG_file_type ] +!7 = metadata !{i32 786453, i32 0, metadata !"", i32 0, i32 0, i64 0, i64 0, i64 0, i32 0, null, metadata !1, i32 0, i32 0} ; [ DW_TAG_subroutine_type ] [line 0, size 0, align 0, offset 0] [from ] +!8 = metadata !{i32 786478, metadata !6, metadata !"dead_vararg", metadata !"dead_vararg", metadata !"", metadata !6, i32 5, metadata !7, i1 true, i1 true, i32 0, i32 0, null, i32 256, i1 false, void (...)* @_ZN12_GLOBAL__N_111dead_varargEz, null, null, metadata !1, i32 5} ; [ DW_TAG_subprogram ] [line 5] [local] [def] [dead_vararg] ; CHECK: metadata !"dead_vararg"{{.*}}void ()* @_ZN12_GLOBAL__N_111dead_varargEz -!9 = metadata !{i32 786478, i32 0, metadata !6, metadata !"dead_arg", metadata !"dead_arg", metadata !"", metadata !6, i32 4, metadata !7, i1 true, i1 true, i32 0, i32 0, null, i32 256, i1 false, void (i8*)* @_ZN12_GLOBAL__N_18dead_argEPv, null, null, metadata !1, i32 4} ; [ DW_TAG_subprogram ] [line 4] [local] [def] [dead_arg] +!9 = metadata !{i32 786478, metadata !6, metadata !"dead_arg", metadata !"dead_arg", metadata !"", metadata !6, i32 4, metadata !7, i1 true, i1 true, i32 0, i32 0, null, i32 256, i1 false, void (i8*)* @_ZN12_GLOBAL__N_18dead_argEPv, null, null, metadata !1, i32 4} ; [ DW_TAG_subprogram ] [line 4] [local] [def] [dead_arg] ; CHECK: metadata !"dead_arg"{{.*}}void ()* @_ZN12_GLOBAL__N_18dead_argEPv @@ -62,3 +60,4 @@ entry: !17 = metadata !{i32 5, i32 25, metadata !18, null} !18 = metadata !{i32 786443, metadata !8, i32 5, i32 23, metadata !6, i32 1} ; [ DW_TAG_lexical_block ] [/home/samsonov/tmp/clang-di/test.cc] !19 = metadata !{i32 5, i32 30, metadata !18, null} +!20 = metadata !{metadata !"test.cc", metadata !"/home/samsonov/tmp/clang-di"} diff --git a/test/Transforms/DeadArgElim/deadexternal.ll b/test/Transforms/DeadArgElim/deadexternal.ll index e3fe1bbb548b9..cca58721e5324 100644 --- a/test/Transforms/DeadArgElim/deadexternal.ll +++ b/test/Transforms/DeadArgElim/deadexternal.ll @@ -1,4 +1,4 @@ -; RUN: opt -deadargelim -S %s | FileCheck %s +; RUN: opt -deadargelim -S < %s | FileCheck %s define void @test(i32) { ret void diff --git a/test/Transforms/DeadArgElim/keepalive.ll b/test/Transforms/DeadArgElim/keepalive.ll index dc92dc9f171aa..e41110c96ef45 100644 --- a/test/Transforms/DeadArgElim/keepalive.ll +++ b/test/Transforms/DeadArgElim/keepalive.ll @@ -1,6 +1,4 @@ -; RUN: opt < %s -deadargelim -S > %t -; RUN: grep "define internal zeroext i32 @test1() nounwind" %t -; RUN: grep "define internal <{ i32, i32 }> @test2" %t +; RUN: opt < %s -deadargelim -S | FileCheck %s %Ty = type <{ i32, i32 }> @@ -9,11 +7,13 @@ ; the function and then changing too much. ; This checks if the return value attributes are not removed +; CHECK: define internal zeroext i32 @test1() #0 define internal zeroext i32 @test1(i32 %DEADARG1) nounwind { ret i32 1 } ; This checks if the struct doesn't get non-packed +; CHECK: define internal <{ i32, i32 }> @test2 define internal <{ i32, i32 }> @test2(i32 %DEADARG1) { ret <{ i32, i32 }> <{ i32 1, i32 2 }> } @@ -28,3 +28,4 @@ define void @caller() { ret void } +; CHECK: attributes #0 = { nounwind } diff --git a/test/Transforms/DeadStoreElimination/const-pointers.ll b/test/Transforms/DeadStoreElimination/const-pointers.ll index 7d57804631daa..15976f9f10d40 100644 --- a/test/Transforms/DeadStoreElimination/const-pointers.ll +++ b/test/Transforms/DeadStoreElimination/const-pointers.ll @@ -1,4 +1,4 @@ -; RUN: opt %s -basicaa -dse -S | FileCheck %s +; RUN: opt -basicaa -dse -S < %s | FileCheck %s %t = type { i32 } diff --git a/test/Transforms/DeadStoreElimination/dominate.ll b/test/Transforms/DeadStoreElimination/dominate.ll index 284fea4234fc7..38cf1a066dae1 100644 --- a/test/Transforms/DeadStoreElimination/dominate.ll +++ b/test/Transforms/DeadStoreElimination/dominate.ll @@ -1,4 +1,4 @@ -; RUN: opt %s -dse -disable-output +; RUN: opt -dse -disable-output < %s ; test that we don't crash declare void @bar() diff --git a/test/Transforms/DeadStoreElimination/no-targetdata.ll b/test/Transforms/DeadStoreElimination/no-targetdata.ll index 6c7f940316a00..4022d76dcb52d 100644 --- a/test/Transforms/DeadStoreElimination/no-targetdata.ll +++ b/test/Transforms/DeadStoreElimination/no-targetdata.ll @@ -1,4 +1,4 @@ -; RUN: opt %s -basicaa -dse -S | FileCheck %s +; RUN: opt -basicaa -dse -S < %s | FileCheck %s declare void @test1f() diff --git a/test/Transforms/DeadStoreElimination/pr11390.ll b/test/Transforms/DeadStoreElimination/pr11390.ll index 2ce6eea365aa7..f63aa1eb8aae3 100644 --- a/test/Transforms/DeadStoreElimination/pr11390.ll +++ b/test/Transforms/DeadStoreElimination/pr11390.ll @@ -1,4 +1,4 @@ -; RUN: opt -basicaa -dse -S -o - %s | FileCheck %s +; RUN: opt -basicaa -dse -S < %s | FileCheck %s ; PR11390 target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128" target triple = "x86_64-unknown-linux-gnu" diff --git a/test/Transforms/EarlyCSE/commute.ll b/test/Transforms/EarlyCSE/commute.ll index f84a7dd1aae95..8cf04d1765b96 100644 --- a/test/Transforms/EarlyCSE/commute.ll +++ b/test/Transforms/EarlyCSE/commute.ll @@ -19,9 +19,9 @@ define void @test2(float %A, float %B, i1* %PA, i1* %PB) { ; CHECK-NEXT: store ; CHECK-NEXT: store ; CHECK-NEXT: ret - %C = fcmp eq float %A, %B + %C = fcmp oeq float %A, %B store i1 %C, i1* %PA - %D = fcmp eq float %B, %A + %D = fcmp oeq float %B, %A store i1 %D, i1* %PB ret void } diff --git a/test/Transforms/EarlyCSE/floatingpoint.ll b/test/Transforms/EarlyCSE/floatingpoint.ll new file mode 100644 index 0000000000000..2abecd74b63af --- /dev/null +++ b/test/Transforms/EarlyCSE/floatingpoint.ll @@ -0,0 +1,14 @@ +; RUN: opt < %s -S -early-cse | FileCheck %s + +; Ensure we don't simplify away additions vectors of +0.0's (same as scalars). +define <4 x float> @fV( <4 x float> %a) { + ; CHECK: %b = fadd <4 x float> %a, zeroinitializer + %b = fadd <4 x float> %a, + ret <4 x float> %b +} + +define <4 x float> @fW( <4 x float> %a) { + ; CHECK: ret <4 x float> %a + %b = fadd <4 x float> %a, + ret <4 x float> %b +} diff --git a/test/Transforms/FunctionAttrs/2008-09-03-ReadNone.ll b/test/Transforms/FunctionAttrs/2008-09-03-ReadNone.ll index 946453f586ede..36a765873487d 100644 --- a/test/Transforms/FunctionAttrs/2008-09-03-ReadNone.ll +++ b/test/Transforms/FunctionAttrs/2008-09-03-ReadNone.ll @@ -1,18 +1,24 @@ -; RUN: opt < %s -basicaa -functionattrs -S | grep readnone | count 4 +; RUN: opt < %s -basicaa -functionattrs -S | FileCheck %s @x = global i32 0 +; CHECK: declare i32 @e() #0 declare i32 @e() readnone +; CHECK: define i32 @f() #0 define i32 @f() { %tmp = call i32 @e( ) ; [#uses=1] ret i32 %tmp } +; CHECK: define i32 @g() #0 define i32 @g() readonly { ret i32 0 } +; CHECK: define i32 @h() #0 define i32 @h() readnone { %tmp = load i32* @x ; [#uses=1] ret i32 %tmp } + +; CHECK: attributes #0 = { readnone } diff --git a/test/Transforms/FunctionAttrs/2008-09-03-ReadOnly.ll b/test/Transforms/FunctionAttrs/2008-09-03-ReadOnly.ll index 22eca13204152..d8256ae8e647c 100644 --- a/test/Transforms/FunctionAttrs/2008-09-03-ReadOnly.ll +++ b/test/Transforms/FunctionAttrs/2008-09-03-ReadOnly.ll @@ -1,9 +1,13 @@ -; RUN: opt < %s -basicaa -functionattrs -S | grep readonly | count 2 +; RUN: opt < %s -basicaa -functionattrs -S | FileCheck %s +; CHECK: define i32 @f() #0 define i32 @f() { entry: - %tmp = call i32 @e( ) ; [#uses=1] - ret i32 %tmp + %tmp = call i32 @e( ) + ret i32 %tmp } +; CHECK: declare i32 @e() #0 declare i32 @e() readonly + +; CHECK: attributes #0 = { readonly } diff --git a/test/Transforms/FunctionAttrs/2009-01-04-Annotate.ll b/test/Transforms/FunctionAttrs/2009-01-04-Annotate.ll new file mode 100644 index 0000000000000..d414b73524fdb --- /dev/null +++ b/test/Transforms/FunctionAttrs/2009-01-04-Annotate.ll @@ -0,0 +1,21 @@ +; RUN: opt < %s -functionattrs -S | FileCheck %s + +; CHECK: declare noalias i8* @fopen(i8* nocapture, i8* nocapture) #0 +declare i8* @fopen(i8*, i8*) + +; CHECK: declare i8 @strlen(i8* nocapture) #1 +declare i8 @strlen(i8*) + +; CHECK: declare noalias i32* @realloc(i32* nocapture, i32) #0 +declare i32* @realloc(i32*, i32) + +; Test deliberately wrong declaration +declare i32 @strcpy(...) + +; CHECK-NOT: strcpy{{.*}}noalias +; CHECK-NOT: strcpy{{.*}}nocapture +; CHECK-NOT: strcpy{{.*}}nounwind +; CHECK-NOT: strcpy{{.*}}readonly + +; CHECK: attributes #0 = { nounwind } +; CHECK: attributes #1 = { nounwind readonly } diff --git a/test/Transforms/FunctionAttrs/annotate-1.ll b/test/Transforms/FunctionAttrs/annotate-1.ll new file mode 100644 index 0000000000000..ae77380acc4a1 --- /dev/null +++ b/test/Transforms/FunctionAttrs/annotate-1.ll @@ -0,0 +1,18 @@ +; RUN: opt < %s -functionattrs -S | FileCheck %s + +declare i8* @fopen(i8*, i8*) +; CHECK: declare noalias i8* @fopen(i8* nocapture, i8* nocapture) [[G0:#[0-9]]] + +declare i8 @strlen(i8*) +; CHECK: declare i8 @strlen(i8* nocapture) [[G1:#[0-9]]] + +declare i32* @realloc(i32*, i32) +; CHECK: declare noalias i32* @realloc(i32* nocapture, i32) [[G0]] + +; Test deliberately wrong declaration + +declare i32 @strcpy(...) +; CHECK: declare i32 @strcpy(...) + +; CHECK: attributes [[G0]] = { nounwind } +; CHECK: attributes [[G1]] = { nounwind readonly } diff --git a/test/Transforms/FunctionAttrs/atomic.ll b/test/Transforms/FunctionAttrs/atomic.ll index 7c2bff7a05f71..027ee0fd06a26 100644 --- a/test/Transforms/FunctionAttrs/atomic.ll +++ b/test/Transforms/FunctionAttrs/atomic.ll @@ -3,7 +3,7 @@ ; Atomic load/store to local doesn't affect whether a function is ; readnone/readonly. define i32 @test1(i32 %x) uwtable ssp { -; CHECK: define i32 @test1(i32 %x) uwtable readnone ssp { +; CHECK: define i32 @test1(i32 %x) #0 { entry: %x.addr = alloca i32, align 4 store atomic i32 %x, i32* %x.addr seq_cst, align 4 @@ -13,9 +13,11 @@ entry: ; A function with an Acquire load is not readonly. define i32 @test2(i32* %x) uwtable ssp { -; CHECK: define i32 @test2(i32* nocapture %x) uwtable ssp { +; CHECK: define i32 @test2(i32* nocapture %x) #1 { entry: %r = load atomic i32* %x seq_cst, align 4 ret i32 %r } +; CHECK: attributes #0 = { readnone ssp uwtable } +; CHECK: attributes #1 = { ssp uwtable } diff --git a/test/Transforms/FunctionAttrs/noreturn.ll b/test/Transforms/FunctionAttrs/noreturn.ll new file mode 100644 index 0000000000000..470ebcb1d3cdd --- /dev/null +++ b/test/Transforms/FunctionAttrs/noreturn.ll @@ -0,0 +1,18 @@ +; RUN: opt < %s -functionattrs -instcombine -S | FileCheck %s + +define void @endless_loop() noreturn nounwind readnone ssp uwtable { +entry: + br label %while.body + +while.body: + br label %while.body +} +;CHECK: @main +;CHECK: endless_loop +;CHECK: ret +define i32 @main() noreturn nounwind ssp uwtable { +entry: + tail call void @endless_loop() + unreachable +} + diff --git a/test/Transforms/GCOVProfiling/linkagename.ll b/test/Transforms/GCOVProfiling/linkagename.ll new file mode 100644 index 0000000000000..d1bce728e08c4 --- /dev/null +++ b/test/Transforms/GCOVProfiling/linkagename.ll @@ -0,0 +1,27 @@ +; RUN: echo '!9 = metadata !{metadata !"%T/linkagename.ll", metadata !0}' > %t1 +; RUN: cat %s %t1 > %t2 +; RUN: opt -insert-gcov-profiling -disable-output < %t2 +; RUN: grep _Z3foov %T/linkagename.gcno +; RUN: rm %T/linkagename.gcno + +; REQUIRES: shell + +define void @_Z3foov() { +entry: + ret void, !dbg !8 +} + +!llvm.dbg.cu = !{!0} +!llvm.gcov = !{!9} + +!0 = metadata !{i32 786449, i32 4, metadata !1, metadata !"clang version 3.3 (trunk 177323)", i1 false, metadata !"", i32 0, metadata !3, metadata !3, metadata !4, metadata !3, metadata !""} ; [ DW_TAG_compile_unit ] [/home/nlewycky/hello.cc] [DW_LANG_C_plus_plus] +!1 = metadata !{i32 786473, metadata !2} ; [ DW_TAG_file_type ] [/home/nlewycky/hello.cc] +!2 = metadata !{metadata !"hello.cc", metadata !"/home/nlewycky"} +!3 = metadata !{i32 0} +!4 = metadata !{metadata !5} +!5 = metadata !{i32 786478, metadata !1, metadata !1, metadata !"foo", metadata !"foo", metadata !"_Z3foov", i32 1, metadata !6, i1 false, i1 true, i32 0, i32 0, null, i32 256, i1 false, void ()* @_Z3foov, null, null, metadata !3, i32 1} ; [ DW_TAG_subprogram ] [line 1] [def] [foo] +!6 = metadata !{i32 786453, i32 0, metadata !"", i32 0, i32 0, i64 0, i64 0, i64 0, i32 0, null, metadata !7, i32 0, i32 0} ; [ DW_TAG_subroutine_type ] [line 0, size 0, align 0, offset 0] [from ] +!7 = metadata !{null} +!8 = metadata !{i32 1, i32 0, metadata !5, null} + + diff --git a/test/Transforms/GCOVProfiling/lit.local.cfg b/test/Transforms/GCOVProfiling/lit.local.cfg new file mode 100644 index 0000000000000..19eebc0ac7ac3 --- /dev/null +++ b/test/Transforms/GCOVProfiling/lit.local.cfg @@ -0,0 +1 @@ +config.suffixes = ['.ll', '.c', '.cpp'] diff --git a/test/Transforms/GCOVProfiling/version.ll b/test/Transforms/GCOVProfiling/version.ll new file mode 100644 index 0000000000000..d6d0f3314c269 --- /dev/null +++ b/test/Transforms/GCOVProfiling/version.ll @@ -0,0 +1,29 @@ +; RUN: echo '!9 = metadata !{metadata !"%T/version.ll", metadata !0}' > %t1 +; RUN: cat %s %t1 > %t2 +; RUN: opt -insert-gcov-profiling -disable-output < %t2 +; RUN: head -c12 %T/version.gcno | grep '^oncg\*204MVLL$' +; RUN: rm %T/version.gcno +; RUN: not opt -insert-gcov-profiling -default-gcov-version=asdfasdf -disable-output < %t2 +; RUN: opt -insert-gcov-profiling -default-gcov-version=407* -disable-output < %t2 +; RUN: head -c12 %T/version.gcno | grep '^oncg\*704MVLL$' +; RUN: rm %T/version.gcno + +define void @test() { + ret void, !dbg !8 +} + +; REQUIRES: shell + +!llvm.gcov = !{!9} +!llvm.dbg.cu = !{!0} + +!0 = metadata !{metadata !"./version", metadata !1} +!1 = metadata !{i32 786449, i32 0, i32 4, metadata !2, metadata !"clang version 3.3 (trunk 176994)", i1 false, metadata !"", i32 0, metadata !3, metadata !3, metadata !4, metadata !3, metadata !""} ; [ DW_TAG_compile_unit ] [./version] [DW_LANG_C_plus_plus] +!2 = metadata !{i32 786473, metadata !"version", metadata !"/usr/local/google/home/nlewycky"} ; [ DW_TAG_file_type ] +!3 = metadata !{i32 0} +!4 = metadata !{metadata !5} +!5 = metadata !{i32 786478, i32 0, metadata !6, metadata !"test", metadata !"test", metadata !"", metadata !6, i32 1, metadata !7, i1 false, i1 true, i32 0, i32 0, null, i32 256, i1 false, void ()* @test, null, null, metadata !3, i32 1} ; [ DW_TAG_subprogram ] [line 1] [def] [test] +!6 = metadata !{i32 786473, metadata !"", metadata !"."} ; [ DW_TAG_file_type ] +!7 = metadata !{i32 786453, i32 0, metadata !"", i32 0, i32 0, i64 0, i64 0, i64 0, i32 0, null, metadata !3, i32 0, i32 0} ; [ DW_TAG_subroutine_type ] [line 0, size 0, align 0, offset 0] [from ] +!8 = metadata !{i32 1, i32 0, metadata !5, null} +;; !9 is added through the echo line at the top. diff --git a/test/Transforms/GVN/2011-04-27-phioperands.ll b/test/Transforms/GVN/2011-04-27-phioperands.ll index 6e5075db7c8ef..42c46500c4837 100644 --- a/test/Transforms/GVN/2011-04-27-phioperands.ll +++ b/test/Transforms/GVN/2011-04-27-phioperands.ll @@ -1,4 +1,4 @@ -; RUN: opt %s -gvn -disable-output +; RUN: opt -gvn -disable-output < %s target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-f128:128:128-n8:16:32:64" diff --git a/test/Transforms/GVN/MemdepMiscompile.ll b/test/Transforms/GVN/MemdepMiscompile.ll new file mode 100644 index 0000000000000..d42016961575c --- /dev/null +++ b/test/Transforms/GVN/MemdepMiscompile.ll @@ -0,0 +1,54 @@ +; RUN: opt < %s -basicaa -gvn -S | FileCheck %s +target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64" +target triple = "x86_64-apple-macosx10.7.0" + +; rdar://12801584 +; Value of %shouldExit can be changed by RunInMode. +; Make sure we do not replace load %shouldExit in while.cond.backedge +; with a phi node where the value from while.body is 0. +define i32 @test() nounwind ssp { +entry: +; CHECK: test() +; CHECK: while.body: +; CHECK: call void @RunInMode +; CHECK: br i1 %tobool, label %while.cond.backedge, label %if.then +; CHECK: while.cond.backedge: +; CHECK: load i32* %shouldExit +; CHECK: br i1 %cmp, label %while.body + %shouldExit = alloca i32, align 4 + %tasksIdle = alloca i32, align 4 + store i32 0, i32* %shouldExit, align 4 + store i32 0, i32* %tasksIdle, align 4 + call void @CTestInitialize(i32* %tasksIdle) nounwind + %0 = load i32* %shouldExit, align 4 + %cmp1 = icmp eq i32 %0, 0 + br i1 %cmp1, label %while.body.lr.ph, label %while.end + +while.body.lr.ph: + br label %while.body + +while.body: + call void @RunInMode(i32 100) nounwind + %1 = load i32* %tasksIdle, align 4 + %tobool = icmp eq i32 %1, 0 + br i1 %tobool, label %while.cond.backedge, label %if.then + +if.then: + store i32 0, i32* %tasksIdle, align 4 + call void @TimerCreate(i32* %shouldExit) nounwind + br label %while.cond.backedge + +while.cond.backedge: + %2 = load i32* %shouldExit, align 4 + %cmp = icmp eq i32 %2, 0 + br i1 %cmp, label %while.body, label %while.cond.while.end_crit_edge + +while.cond.while.end_crit_edge: + br label %while.end + +while.end: + ret i32 0 +} +declare void @CTestInitialize(i32*) +declare void @RunInMode(i32) +declare void @TimerCreate(i32*) diff --git a/test/Transforms/GVN/crash-no-aa.ll b/test/Transforms/GVN/crash-no-aa.ll index dae65ddb2fe7d..9ad63a7350c27 100644 --- a/test/Transforms/GVN/crash-no-aa.ll +++ b/test/Transforms/GVN/crash-no-aa.ll @@ -1,7 +1,6 @@ -; RUN: opt -no-aa -gvn -S %s +; RUN: opt -no-aa -gvn -S < %s -target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v1 -28:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64" +target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64" target triple = "x86_64-unknown-freebsd8.0" ; PR5744 diff --git a/test/Transforms/GVN/crash.ll b/test/Transforms/GVN/crash.ll index 4a8c8e4589c8f..9fb612fcae139 100644 --- a/test/Transforms/GVN/crash.ll +++ b/test/Transforms/GVN/crash.ll @@ -1,4 +1,4 @@ -; RUN: opt -gvn %s -disable-output +; RUN: opt -gvn -disable-output < %s ; PR5631 diff --git a/test/Transforms/GVN/edge.ll b/test/Transforms/GVN/edge.ll index 32392f3ab0c8b..3a102b6c35391 100644 --- a/test/Transforms/GVN/edge.ll +++ b/test/Transforms/GVN/edge.ll @@ -1,4 +1,4 @@ -; RUN: opt %s -gvn -S -o - | FileCheck %s +; RUN: opt -gvn -S < %s | FileCheck %s define i32 @f1(i32 %x) { ; CHECK: define i32 @f1( diff --git a/test/Transforms/GVN/fpmath.ll b/test/Transforms/GVN/fpmath.ll index 8ab285448fbbf..403df5c9008a9 100644 --- a/test/Transforms/GVN/fpmath.ll +++ b/test/Transforms/GVN/fpmath.ll @@ -1,4 +1,4 @@ -; RUN: opt %s -gvn -S -o - | FileCheck %s +; RUN: opt -gvn -S < %s | FileCheck %s define double @test1(double %x, double %y) { ; CHECK: @test1(double %x, double %y) diff --git a/test/Transforms/GVN/lpre-call-wrap-2.ll b/test/Transforms/GVN/lpre-call-wrap-2.ll index e39f3ed87d1c3..35e3534a9c892 100644 --- a/test/Transforms/GVN/lpre-call-wrap-2.ll +++ b/test/Transforms/GVN/lpre-call-wrap-2.ll @@ -1,4 +1,4 @@ -; RUN: opt -S -basicaa -gvn -enable-load-pre %s | FileCheck %s +; RUN: opt -S -basicaa -gvn -enable-load-pre < %s | FileCheck %s ; ; The partially redundant load in bb1 should be hoisted to "bb". This comes ; from this C code (GCC PR 23455): diff --git a/test/Transforms/GVN/lpre-call-wrap.ll b/test/Transforms/GVN/lpre-call-wrap.ll index 40462798b5341..0646f3fe0aad1 100644 --- a/test/Transforms/GVN/lpre-call-wrap.ll +++ b/test/Transforms/GVN/lpre-call-wrap.ll @@ -1,4 +1,4 @@ -; RUN: opt -S -gvn -enable-load-pre %s | FileCheck %s +; RUN: opt -S -gvn -enable-load-pre < %s | FileCheck %s ; ; Make sure the load in bb3.backedge is removed and moved into bb1 after the ; call. This makes the non-call case faster. diff --git a/test/Transforms/GVN/nonescaping-malloc.ll b/test/Transforms/GVN/nonescaping-malloc.ll index afcb7fe3bb0f9..c2eeed56ffc19 100644 --- a/test/Transforms/GVN/nonescaping-malloc.ll +++ b/test/Transforms/GVN/nonescaping-malloc.ll @@ -1,3 +1,4 @@ +; REQUIRES: asserts ; RUN: opt < %s -basicaa -gvn -stats -disable-output 2>&1 | grep "Number of loads deleted" ; rdar://7363102 diff --git a/test/Transforms/GVN/null-aliases-nothing.ll b/test/Transforms/GVN/null-aliases-nothing.ll index 9e4ae18c710cc..37bf09d7f3ffc 100644 --- a/test/Transforms/GVN/null-aliases-nothing.ll +++ b/test/Transforms/GVN/null-aliases-nothing.ll @@ -1,4 +1,4 @@ -; RUN: opt %s -basicaa -gvn -S | FileCheck %s +; RUN: opt < %s -basicaa -gvn -S | FileCheck %s %t = type { i32 } declare void @test1f(i8*) diff --git a/test/Transforms/GVN/pr12979.ll b/test/Transforms/GVN/pr12979.ll index 669da9127d0b1..0198a56513ea9 100644 --- a/test/Transforms/GVN/pr12979.ll +++ b/test/Transforms/GVN/pr12979.ll @@ -1,4 +1,4 @@ -; RUN: opt %s -gvn -S -o - | FileCheck %s +; RUN: opt -gvn -S < %s | FileCheck %s define i32 @test1(i32 %x, i32 %y) { ; CHECK: @test1(i32 %x, i32 %y) diff --git a/test/Transforms/GVN/range.ll b/test/Transforms/GVN/range.ll index 3759c415dabc4..2115fe85661e5 100644 --- a/test/Transforms/GVN/range.ll +++ b/test/Transforms/GVN/range.ll @@ -1,4 +1,4 @@ -; RUN: opt %s -basicaa -gvn -S -o - | FileCheck %s +; RUN: opt -basicaa -gvn -S < %s | FileCheck %s define i32 @test1(i32* %p) { ; CHECK: @test1(i32* %p) diff --git a/test/Transforms/GVN/rle.ll b/test/Transforms/GVN/rle.ll index 72fa819d1c73a..f470ed88bb9c4 100644 --- a/test/Transforms/GVN/rle.ll +++ b/test/Transforms/GVN/rle.ll @@ -254,14 +254,11 @@ Cont: %A = load i8* %P3 ret i8 %A -;; FIXME: This is disabled because this caused a miscompile in the llvm-gcc -;; bootstrap, see r82411 -; -; HECK: @coerce_mustalias_nonlocal1 -; HECK: Cont: -; HECK: %A = phi i8 [ -; HECK-NOT: load -; HECK: ret i8 %A +; CHECK: @coerce_mustalias_nonlocal1 +; CHECK: Cont: +; CHECK: %A = phi i8 [ +; CHECK-NOT: load +; CHECK: ret i8 %A } diff --git a/test/Transforms/GVN/tbaa.ll b/test/Transforms/GVN/tbaa.ll index 90661c62507b1..85fe39a93b016 100644 --- a/test/Transforms/GVN/tbaa.ll +++ b/test/Transforms/GVN/tbaa.ll @@ -1,4 +1,4 @@ -; RUN: opt %s -basicaa -gvn -S -o - | FileCheck %s +; RUN: opt -basicaa -gvn -S < %s | FileCheck %s define i32 @test1(i8* %p, i8* %q) { ; CHECK: @test1(i8* %p, i8* %q) diff --git a/test/Transforms/GlobalOpt/2009-03-05-dbg.ll b/test/Transforms/GlobalOpt/2009-03-05-dbg.ll index 0f3efa09a1daa..e71aed9e05ff9 100644 --- a/test/Transforms/GlobalOpt/2009-03-05-dbg.ll +++ b/test/Transforms/GlobalOpt/2009-03-05-dbg.ll @@ -1,3 +1,4 @@ +; REQUIRES: asserts ; RUN: opt < %s -globalopt -stats -disable-output 2>&1 | grep "1 globalopt - Number of global vars shrunk to booleans" @Stop = internal global i32 0 ; [#uses=3] diff --git a/test/Transforms/GlobalOpt/2010-02-25-MallocPromote.ll b/test/Transforms/GlobalOpt/2010-02-25-MallocPromote.ll index 27352fa29066f..629d57c884244 100644 --- a/test/Transforms/GlobalOpt/2010-02-25-MallocPromote.ll +++ b/test/Transforms/GlobalOpt/2010-02-25-MallocPromote.ll @@ -1,5 +1,5 @@ ; PR6422 -; RUN: opt -globalopt -S %s +; RUN: opt -globalopt -S < %s target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128" target triple = "x86_64-unknown-linux-gnu" diff --git a/test/Transforms/GlobalOpt/2010-02-26-MallocSROA.ll b/test/Transforms/GlobalOpt/2010-02-26-MallocSROA.ll index 6f1996a867e3f..ab7721fd97204 100644 --- a/test/Transforms/GlobalOpt/2010-02-26-MallocSROA.ll +++ b/test/Transforms/GlobalOpt/2010-02-26-MallocSROA.ll @@ -1,4 +1,4 @@ -; RUN: opt -globalopt -S %s +; RUN: opt -globalopt -S < %s ; PR6435 target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128" target triple = "x86_64-unknown-linux-gnu" diff --git a/test/Transforms/GlobalOpt/crash-2.ll b/test/Transforms/GlobalOpt/crash-2.ll new file mode 100644 index 0000000000000..684f6cee180ba --- /dev/null +++ b/test/Transforms/GlobalOpt/crash-2.ll @@ -0,0 +1,19 @@ +; RUN: llvm-as < %s | opt -globalopt -disable-output +; NOTE: This needs to run through 'llvm-as' first to reproduce the error! +; PR15440 + +%union.U5.0.6.12 = type { i32 } +%struct.S0.1.7.13 = type { i8, i8, i8, i8, i16, [2 x i8] } +%struct.S1.2.8.14 = type { i32, i16, i8, i8 } + +@.str = external unnamed_addr constant [2 x i8], align 1 +@g_25 = external global i8, align 1 +@g_71 = internal global %struct.S0.1.7.13 { i8 1, i8 -93, i8 58, i8 -1, i16 -5, [2 x i8] undef }, align 4 +@g_114 = external global i8, align 1 +@g_30 = external global { i32, i8, i32, i8, i8, i8, i8, i8, i8, i8, i8, i8, i8, i8, i8, i8, i8, i8, i8 }, align 4 +@g_271 = internal global [7 x [6 x [5 x i8*]]] [[6 x [5 x i8*]] [[5 x i8*] [i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* @g_25, i8* null], [5 x i8*] [i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* @g_114, i8* @g_114, i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0)], [5 x i8*] [i8* null, i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* @g_25, i8* null, i8* null], [5 x i8*] [i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* @g_25, i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0)], [5 x i8*] [i8* @g_25, i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* null, i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0)], [5 x i8*] [i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr (i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i64 1), i8* @g_25, i8* @g_114, i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0)]], [6 x [5 x i8*]] [[5 x i8*] [i8* @g_25, i8* null, i8* @g_25, i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0)], [5 x i8*] [i8* @g_25, i8* @g_114, i8* @g_25, i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* @g_114], [5 x i8*] [i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr (i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i64 1), i8* @g_25, i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* @g_25], [5 x i8*] [i8* getelementptr (i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i64 1), i8* getelementptr (i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i64 1), i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* @g_114, i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0)], [5 x i8*] [i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* @g_25, i8* @g_25, i8* @g_25, i8* @g_25], [5 x i8*] [i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* @g_25, i8* @g_25, i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0)]], [6 x [5 x i8*]] [[5 x i8*] [i8* null, i8* @g_25, i8* @g_25, i8* @g_25, i8* null], [5 x i8*] [i8* @g_25, i8* getelementptr (i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i64 1), i8* @g_25, i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr (i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i64 1)], [5 x i8*] [i8* null, i8* getelementptr (i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i64 1), i8* getelementptr (i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i64 1), i8* null, i8* @g_25], [5 x i8*] [i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* @g_114, i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr (i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i64 1), i8* getelementptr (i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i64 1)], [5 x i8*] [i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* null, i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* @g_25, i8* null], [5 x i8*] [i8* getelementptr (i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i64 1), i8* @g_25, i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr (i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i64 1), i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0)]], [6 x [5 x i8*]] [[5 x i8*] [i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* @g_25, i8* null, i8* @g_25], [5 x i8*] [i8* @g_25, i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0)], [5 x i8*] [i8* @g_25, i8* getelementptr (i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i64 1), i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* @g_25, i8* @g_25], [5 x i8*] [i8* @g_114, i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* @g_114], [5 x i8*] [i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr (i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i64 1), i8* @g_25, i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0)], [5 x i8*] [i8* @g_114, i8* @g_25, i8* @g_25, i8* @g_114, i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0)]], [6 x [5 x i8*]] [[5 x i8*] [i8* @g_25, i8* null, i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* @g_25, i8* @g_25], [5 x i8*] [i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr (i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i64 1), i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* @g_25, i8* getelementptr (i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i64 1)], [5 x i8*] [i8* @g_25, i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr (i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i64 1), i8* @g_25, i8* getelementptr (i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i64 1)], [5 x i8*] [i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr (i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i64 1), i8* @g_114], [5 x i8*] [i8* @g_25, i8* null, i8* getelementptr (i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i64 1), i8* getelementptr (i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i64 1), i8* null], [5 x i8*] [i8* @g_114, i8* @g_25, i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* @g_114, i8* @g_25]], [6 x [5 x i8*]] [[5 x i8*] [i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* null, i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* null, i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0)], [5 x i8*] [i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* @g_25, i8* @g_25, i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0)], [5 x i8*] [i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr (i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i64 1)], [5 x i8*] [i8* @g_114, i8* getelementptr (i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i64 1), i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0)], [5 x i8*] [i8* @g_25, i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* @g_25, i8* getelementptr (i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i64 1), i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0)], [5 x i8*] [i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* @g_25, i8* @g_25, i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* @g_25]], [6 x [5 x i8*]] [[5 x i8*] [i8* @g_25, i8* @g_25, i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* null], [5 x i8*] [i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* @g_114, i8* @g_25, i8* @g_25, i8* @g_114], [5 x i8*] [i8* null, i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* @g_25, i8* null, i8* getelementptr (i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i64 1)], [5 x i8*] [i8* getelementptr (i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i64 1), i8* @g_114, i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* @g_114, i8* getelementptr (i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i64 1)], [5 x i8*] [i8* @g_25, i8* @g_25, i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i8* getelementptr (i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i64 1), i8* @g_25], [5 x i8*] [i8* getelementptr (i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i64 1), i8* @g_25, i8* @g_25, i8* getelementptr (i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), i64 1), i8* @g_25]]], align 4 + +define i32 @func() { + %tmp = load i8* getelementptr inbounds (%struct.S0.1.7.13* @g_71, i32 0, i32 0), align 1 + ret i32 0 +} diff --git a/test/Transforms/GlobalOpt/crash.ll b/test/Transforms/GlobalOpt/crash.ll index 366a874f73527..80c777ccabc1a 100644 --- a/test/Transforms/GlobalOpt/crash.ll +++ b/test/Transforms/GlobalOpt/crash.ll @@ -1,4 +1,4 @@ -; RUN: opt -globalopt -disable-output %s +; RUN: opt -globalopt -disable-output < %s target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128-n8:16:32" target triple = "i386-apple-darwin9.8" diff --git a/test/Transforms/GlobalOpt/ctor-list-opt-constexpr.ll b/test/Transforms/GlobalOpt/ctor-list-opt-constexpr.ll index e3bc473f52adb..c9076109443de 100644 --- a/test/Transforms/GlobalOpt/ctor-list-opt-constexpr.ll +++ b/test/Transforms/GlobalOpt/ctor-list-opt-constexpr.ll @@ -1,4 +1,4 @@ -; RUN: opt -globalopt %s -S | FileCheck %s +; RUN: opt -globalopt -S < %s | FileCheck %s target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64" target triple = "x86_64-apple-darwin10.0.0" diff --git a/test/Transforms/GlobalOpt/externally-initialized-global-ctr.ll b/test/Transforms/GlobalOpt/externally-initialized-global-ctr.ll new file mode 100644 index 0000000000000..9295c2025a2a2 --- /dev/null +++ b/test/Transforms/GlobalOpt/externally-initialized-global-ctr.ll @@ -0,0 +1,35 @@ +; RUN: opt < %s -globalopt -S | FileCheck %s +; rdar://12580965. +; ObjC++ test case. + +%struct.ButtonInitData = type { i8* } + +@_ZL14buttonInitData = internal global [1 x %struct.ButtonInitData] zeroinitializer, align 4 + +@"\01L_OBJC_METH_VAR_NAME_40" = internal global [7 x i8] c"print:\00", section "__TEXT,__objc_methname,cstring_literals", align 1 +@"\01L_OBJC_SELECTOR_REFERENCES_41" = internal externally_initialized global i8* getelementptr inbounds ([7 x i8]* @"\01L_OBJC_METH_VAR_NAME_40", i32 0, i32 0), section "__DATA, __objc_selrefs, literal_pointers, no_dead_strip" + +@llvm.global_ctors = appending global [1 x { i32, void ()* }] [{ i32, void ()* } { i32 65535, void ()* @_GLOBAL__I_a }] +@llvm.used = appending global [2 x i8*] [i8* getelementptr inbounds ([7 x i8]* @"\01L_OBJC_METH_VAR_NAME_40", i32 0, i32 0), i8* bitcast (i8** @"\01L_OBJC_SELECTOR_REFERENCES_41" to i8*)] + +define internal void @__cxx_global_var_init() section "__TEXT,__StaticInit,regular,pure_instructions" { + %1 = load i8** @"\01L_OBJC_SELECTOR_REFERENCES_41", !invariant.load !2009 + store i8* %1, i8** getelementptr inbounds ([1 x %struct.ButtonInitData]* @_ZL14buttonInitData, i32 0, i32 0, i32 0), align 4 + ret void +} + +define internal void @_GLOBAL__I_a() section "__TEXT,__StaticInit,regular,pure_instructions" { + call void @__cxx_global_var_init() + ret void +} + +declare void @test(i8*) + +define void @print() { +; CHECK: %1 = load i8** getelementptr inbounds ([1 x %struct.ButtonInitData]* @_ZL14buttonInitData, i32 0, i32 0, i32 0), align 4 + %1 = load i8** getelementptr inbounds ([1 x %struct.ButtonInitData]* @_ZL14buttonInitData, i32 0, i32 0, i32 0), align 4 + call void @test(i8* %1) + ret void +} + +!2009 = metadata !{} diff --git a/test/Transforms/GlobalOpt/integer-bool.ll b/test/Transforms/GlobalOpt/integer-bool.ll index 5a34a9c4dabdb..51858069ac5b7 100644 --- a/test/Transforms/GlobalOpt/integer-bool.ll +++ b/test/Transforms/GlobalOpt/integer-bool.ll @@ -1,23 +1,28 @@ -; RUN: opt < %s -globalopt -instcombine | \ -; RUN: llvm-dis | grep "ret i1 true" - +; RUN: opt < %s -S -globalopt -instcombine | FileCheck %s ;; check that global opt turns integers that only hold 0 or 1 into bools. -@G = internal global i32 0 ; [#uses=3] +@G = internal addrspace(1) global i32 0 +; CHECK: @G +; CHECK: addrspace(1) +; CHECK: global i1 false define void @set1() { - store i32 0, i32* @G - ret void + store i32 0, i32 addrspace(1)* @G +; CHECK: store i1 false + ret void } define void @set2() { - store i32 1, i32* @G - ret void + store i32 1, i32 addrspace(1)* @G +; CHECK: store i1 true + ret void } define i1 @get() { - %A = load i32* @G ; [#uses=1] - %C = icmp slt i32 %A, 2 ; [#uses=1] - ret i1 %C +; CHECK: @get + %A = load i32 addrspace(1) * @G + %C = icmp slt i32 %A, 2 + ret i1 %C +; CHECK: ret i1 true } diff --git a/test/Transforms/GlobalOpt/memset-null.ll b/test/Transforms/GlobalOpt/memset-null.ll index 01534025faa38..53ec7551130e4 100644 --- a/test/Transforms/GlobalOpt/memset-null.ll +++ b/test/Transforms/GlobalOpt/memset-null.ll @@ -1,4 +1,4 @@ -; RUN: opt -globalopt %s -S -o - | FileCheck %s +; RUN: opt -globalopt -S < %s | FileCheck %s ; PR10047 %0 = type { i32, void ()* } diff --git a/test/Transforms/GlobalOpt/unnamed-addr.ll b/test/Transforms/GlobalOpt/unnamed-addr.ll index ee75058731268..2ca91e50da2a7 100644 --- a/test/Transforms/GlobalOpt/unnamed-addr.ll +++ b/test/Transforms/GlobalOpt/unnamed-addr.ll @@ -1,4 +1,4 @@ -; RUN: opt %s -globalopt -S | FileCheck %s +; RUN: opt -globalopt -S < %s | FileCheck %s @a = internal global i32 0, align 4 @b = internal global i32 0, align 4 diff --git a/test/Transforms/IPConstantProp/user-with-multiple-uses.ll b/test/Transforms/IPConstantProp/user-with-multiple-uses.ll index 402ea41167ce1..968718084e428 100644 --- a/test/Transforms/IPConstantProp/user-with-multiple-uses.ll +++ b/test/Transforms/IPConstantProp/user-with-multiple-uses.ll @@ -4,9 +4,9 @@ ; IPSCCP should propagate the 0 argument, eliminate the switch, and propagate ; the result. -; CHECK: define i32 @main() noreturn nounwind { +; CHECK: define i32 @main() #0 { ; CHECK-NEXT: entry: -; CHECK-NEXT: %call2 = tail call i32 @wwrite(i64 0) nounwind +; CHECK-NEXT: %call2 = tail call i32 @wwrite(i64 0) [[NUW:#[0-9]+]] ; CHECK-NEXT: ret i32 123 define i32 @main() noreturn nounwind { @@ -28,3 +28,7 @@ sw.default: return: ret i32 0 } + +; CHECK: attributes #0 = { noreturn nounwind } +; CHECK: attributes #1 = { nounwind readnone } +; CHECK: attributes [[NUW]] = { nounwind } diff --git a/test/Transforms/IndVarSimplify/2003-09-23-NotAtTop.ll b/test/Transforms/IndVarSimplify/2003-09-23-NotAtTop.ll index 150ae70a8262d..e3de75e36fd85 100644 --- a/test/Transforms/IndVarSimplify/2003-09-23-NotAtTop.ll +++ b/test/Transforms/IndVarSimplify/2003-09-23-NotAtTop.ll @@ -1,4 +1,4 @@ -; RUN: opt -S -indvars %s | FileCheck %s +; RUN: opt -S -indvars < %s | FileCheck %s ; The indvar simplification code should ensure that the first PHI in the block ; is the canonical one! diff --git a/test/Transforms/IndVarSimplify/crash.ll b/test/Transforms/IndVarSimplify/crash.ll index 1b702a3b1a3c9..aa6a2ee16521b 100644 --- a/test/Transforms/IndVarSimplify/crash.ll +++ b/test/Transforms/IndVarSimplify/crash.ll @@ -1,4 +1,4 @@ -; RUN: opt -indvars %s -disable-output +; RUN: opt -indvars -disable-output < %s target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64" declare i32 @putchar(i8) nounwind diff --git a/test/Transforms/IndVarSimplify/dont-recompute.ll b/test/Transforms/IndVarSimplify/dont-recompute.ll new file mode 100644 index 0000000000000..d37b0e21f8263 --- /dev/null +++ b/test/Transforms/IndVarSimplify/dont-recompute.ll @@ -0,0 +1,69 @@ +; RUN: opt < %s -indvars -S | FileCheck %s + +; This tests that the IV is not recomputed outside of the loop when it is known +; to be computed by the loop and used in the loop any way. In the example below +; although a's value can be computed outside of the loop, there is no benefit +; in doing so as it has to be computed by the loop anyway. +; +; extern void func(unsigned val); +; +; void test(unsigned m) +; { +; unsigned a = 0; +; +; for (int i=0; i<186; i++) { +; a += m; +; func(a); +; } +; +; func(a); +; } + +declare void @func(i32) + +; CHECK: @test +define void @test(i32 %m) nounwind uwtable { +entry: + br label %for.body + +for.body: ; preds = %for.body, %entry + %i.06 = phi i32 [ 0, %entry ], [ %inc, %for.body ] + %a.05 = phi i32 [ 0, %entry ], [ %add, %for.body ] + %add = add i32 %a.05, %m +; CHECK: tail call void @func(i32 %add) + tail call void @func(i32 %add) + %inc = add nsw i32 %i.06, 1 + %exitcond = icmp eq i32 %inc, 186 + br i1 %exitcond, label %for.end, label %for.body + +for.end: ; preds = %for.body +; CHECK: for.end: +; CHECK-NOT: mul i32 %m, 186 +; CHECK:%add.lcssa = phi i32 [ %add, %for.body ] +; CHECK-NEXT: tail call void @func(i32 %add.lcssa) + tail call void @func(i32 %add) + ret void +} + +; CHECK: @test2 +define i32 @test2(i32 %m) nounwind uwtable { +entry: + br label %for.body + +for.body: ; preds = %for.body, %entry + %i.06 = phi i32 [ 0, %entry ], [ %inc, %for.body ] + %a.05 = phi i32 [ 0, %entry ], [ %add, %for.body ] + %add = add i32 %a.05, %m +; CHECK: tail call void @func(i32 %add) + tail call void @func(i32 %add) + %inc = add nsw i32 %i.06, 1 + %exitcond = icmp eq i32 %inc, 186 + br i1 %exitcond, label %for.end, label %for.body + +for.end: ; preds = %for.body +; CHECK: for.end: +; CHECK-NOT: mul i32 %m, 186 +; CHECK:%add.lcssa = phi i32 [ %add, %for.body ] +; CHECK-NEXT: ret i32 %add.lcssa + ret i32 %add +} diff --git a/test/Transforms/IndVarSimplify/iv-zext.ll b/test/Transforms/IndVarSimplify/iv-zext.ll index 2e0f70ce461ae..ed0514b08e332 100644 --- a/test/Transforms/IndVarSimplify/iv-zext.ll +++ b/test/Transforms/IndVarSimplify/iv-zext.ll @@ -2,7 +2,7 @@ ; CHECK-NOT: and ; CHECK-NOT: zext -target datalayout = "-p:64:64:64-n32:64" +target datalayout = "p:64:64:64-n32:64" define void @foo(double* %d, i64 %n) nounwind { entry: diff --git a/test/Transforms/IndVarSimplify/phi-uses-value-multiple-times.ll b/test/Transforms/IndVarSimplify/phi-uses-value-multiple-times.ll index 52c9e5c3ffc96..dc36b99482545 100644 --- a/test/Transforms/IndVarSimplify/phi-uses-value-multiple-times.ll +++ b/test/Transforms/IndVarSimplify/phi-uses-value-multiple-times.ll @@ -1,3 +1,4 @@ +; REQUIRES: asserts ; RUN: opt < %s -indvars -disable-output -stats -info-output-file - | FileCheck %s ; Check that IndVarSimplify is not creating unnecessary canonical IVs ; that will never be used. diff --git a/test/Transforms/Inline/2003-09-22-PHINodeInlineFail.ll b/test/Transforms/Inline/2003-09-22-PHINodeInlineFail.ll index 5ced3b8e8da9c..b8ca56050dcab 100644 --- a/test/Transforms/Inline/2003-09-22-PHINodeInlineFail.ll +++ b/test/Transforms/Inline/2003-09-22-PHINodeInlineFail.ll @@ -3,10 +3,15 @@ define i32 @main() { entry: invoke void @__main( ) - to label %LongJmpBlkPre unwind label %LongJmpBlkPre + to label %LongJmpBlkPost unwind label %LongJmpBlkPre -LongJmpBlkPre: ; preds = %entry, %entry +LongJmpBlkPost: + ret i32 0 + +LongJmpBlkPre: %i.3 = phi i32 [ 0, %entry ], [ 0, %entry ] ; [#uses=0] + %exn = landingpad {i8*, i32} personality i32 (...)* @__gxx_personality_v0 + cleanup ret i32 0 } @@ -14,3 +19,4 @@ define void @__main() { ret void } +declare i32 @__gxx_personality_v0(...) diff --git a/test/Transforms/Inline/2003-09-22-PHINodesInNormalInvokeDest.ll b/test/Transforms/Inline/2003-09-22-PHINodesInNormalInvokeDest.ll index 1bd55299a9012..43bdd309c9877 100644 --- a/test/Transforms/Inline/2003-09-22-PHINodesInNormalInvokeDest.ll +++ b/test/Transforms/Inline/2003-09-22-PHINodesInNormalInvokeDest.ll @@ -13,6 +13,8 @@ LJDecisionBB: ; preds = %else br label %else RethrowExcept: ; preds = %entry + %exn = landingpad {i8*, i32} personality i32 (...)* @__gxx_personality_v0 + cleanup ret i32 0 } @@ -20,4 +22,4 @@ define void @__main() { ret void } - +declare i32 @__gxx_personality_v0(...) diff --git a/test/Transforms/Inline/2006-11-09-InlineCGUpdate-2.ll b/test/Transforms/Inline/2006-11-09-InlineCGUpdate-2.ll index b4380d01e483f..ee5a378b18764 100644 --- a/test/Transforms/Inline/2006-11-09-InlineCGUpdate-2.ll +++ b/test/Transforms/Inline/2006-11-09-InlineCGUpdate-2.ll @@ -2,7 +2,6 @@ ; PR993 target datalayout = "e-p:32:32" target triple = "i386-unknown-openbsd3.9" -deplibs = [ "stdc++", "c", "crtend" ] %"struct.__gnu_cxx::__normal_iterator, std::allocator > >" = type { i8* } %"struct.__gnu_cxx::char_producer" = type { i32 (...)** } %struct.__sFILE = type { i8*, i32, i32, i16, i16, %struct.__sbuf, i32, i8*, i32 (i8*)*, i32 (i8*, i8*, i32)*, i64 (i8*, i64, i32)*, i32 (i8*, i8*, i32)*, %struct.__sbuf, i8*, i32, [3 x i8], [1 x i8], %struct.__sbuf, i32, i64 } diff --git a/test/Transforms/Inline/2006-11-09-InlineCGUpdate.ll b/test/Transforms/Inline/2006-11-09-InlineCGUpdate.ll index b754d9f9f5d9c..fb5a4b512b9c6 100644 --- a/test/Transforms/Inline/2006-11-09-InlineCGUpdate.ll +++ b/test/Transforms/Inline/2006-11-09-InlineCGUpdate.ll @@ -2,7 +2,6 @@ ; PR992 target datalayout = "e-p:32:32" target triple = "i686-pc-linux-gnu" -deplibs = [ "stdc++", "c", "crtend" ] %struct._IO_FILE = type { i32, i8*, i8*, i8*, i8*, i8*, i8*, i8*, i8*, i8*, i8*, i8*, %struct._IO_marker*, %struct._IO_FILE*, i32, i32, i32, i16, i8, [1 x i8], i8*, i64, i8*, i8*, i32, [52 x i8] } %struct._IO_marker = type { %struct._IO_marker*, %struct._IO_FILE*, i32 } %"struct.__cxxabiv1::__array_type_info" = type { %"struct.std::type_info" } diff --git a/test/Transforms/Inline/2010-05-12-ValueMap.ll b/test/Transforms/Inline/2010-05-12-ValueMap.ll index f9cc13f499b30..f452907efd073 100644 --- a/test/Transforms/Inline/2010-05-12-ValueMap.ll +++ b/test/Transforms/Inline/2010-05-12-ValueMap.ll @@ -1,4 +1,4 @@ -; RUN: opt %s -inline -mergefunc -disable-output +; RUN: opt -inline -mergefunc -disable-output < %s ; This tests for a bug where the inliner kept the functions in a ValueMap after ; it had completed and a ModulePass started to run. LLVM would crash deleting diff --git a/test/Transforms/Inline/alloca_test.ll b/test/Transforms/Inline/alloca_test.ll index e5791d5d25538..8464259ce1f88 100644 --- a/test/Transforms/Inline/alloca_test.ll +++ b/test/Transforms/Inline/alloca_test.ll @@ -1,7 +1,7 @@ ; This test ensures that alloca instructions in the entry block for an inlined ; function are moved to the top of the function they are inlined into. ; -; RUN: opt -S -inline %s | FileCheck %s +; RUN: opt -S -inline < %s | FileCheck %s define i32 @func(i32 %i) { %X = alloca i32 ; [#uses=1] diff --git a/test/Transforms/Inline/basictest.ll b/test/Transforms/Inline/basictest.ll index 609a3d4e153e9..39e25cb5d627e 100644 --- a/test/Transforms/Inline/basictest.ll +++ b/test/Transforms/Inline/basictest.ll @@ -45,3 +45,48 @@ define i32 @test2(i1 %cond) { ; CHECK-NOT: = alloca ; CHECK: ret i32 } + +declare void @barrier() noduplicate + +define internal i32 @f() { + call void @barrier() noduplicate + ret i32 1 +} + +define i32 @g() { + call void @barrier() noduplicate + ret i32 2 +} + +define internal i32 @h() { + call void @barrier() noduplicate + ret i32 3 +} + +define i32 @test3() { + %b = call i32 @f() + ret i32 %b +} + +; The call to @f cannot be inlined as there is another callsite +; calling @f, and @f contains a noduplicate call. +; +; The call to @g cannot be inlined as it has external linkage. +; +; The call to @h *can* be inlined. + +; CHECK: @test +define i32 @test() { +; CHECK: call i32 @f() + %a = call i32 @f() +; CHECK: call i32 @g() + %b = call i32 @g() +; CHECK-NOT: call i32 @h() + %c = call i32 @h() + + %d = add i32 %a, %b + %e = add i32 %d, %c + + ret i32 %e +; CHECK: } +} diff --git a/test/Transforms/Inline/crash2.ll b/test/Transforms/Inline/crash2.ll index cb1f44d5cca7c..be634f6256339 100644 --- a/test/Transforms/Inline/crash2.ll +++ b/test/Transforms/Inline/crash2.ll @@ -1,4 +1,4 @@ -; RUN: opt -inline -scalarrepl -max-cg-scc-iterations=1 %s -disable-output +; RUN: opt -inline -scalarrepl -max-cg-scc-iterations=1 -disable-output < %s target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64" target triple = "x86_64-apple-darwin10.3" diff --git a/test/Transforms/Inline/delete-call.ll b/test/Transforms/Inline/delete-call.ll index 7716d6a47becb..97c52af9e0d5a 100644 --- a/test/Transforms/Inline/delete-call.ll +++ b/test/Transforms/Inline/delete-call.ll @@ -1,5 +1,9 @@ -; RUN: opt %s -S -inline -functionattrs -stats 2>&1 | grep "Number of call sites deleted, not inlined" -; RUN: opt %s -S -inline -stats 2>&1 | grep "Number of functions inlined" +; REQUIRES: asserts +; RUN: opt -S -inline -stats < %s 2>&1 | FileCheck %s +; CHECK: Number of functions inlined + +; RUN: opt -S -inline -functionattrs -stats < %s 2>&1 | FileCheck -check-prefix=FUNCTIONATTRS %s +; CHECK-FUNCTIONATTRS: Number of call sites deleted, not inlined target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128-n8:16:32" target triple = "i386-apple-darwin9.8" diff --git a/test/Transforms/Inline/devirtualize-3.ll b/test/Transforms/Inline/devirtualize-3.ll index c32be4e024a3f..3f019676e4a46 100644 --- a/test/Transforms/Inline/devirtualize-3.ll +++ b/test/Transforms/Inline/devirtualize-3.ll @@ -1,4 +1,4 @@ -; RUN: opt -basicaa -inline -S -scalarrepl -gvn -instcombine %s | FileCheck %s +; RUN: opt -basicaa -inline -S -scalarrepl -gvn -instcombine < %s | FileCheck %s ; PR5009 ; CHECK: define i32 @main() diff --git a/test/Transforms/Inline/devirtualize.ll b/test/Transforms/Inline/devirtualize.ll index 51ea4baa38666..d46154ef6a989 100644 --- a/test/Transforms/Inline/devirtualize.ll +++ b/test/Transforms/Inline/devirtualize.ll @@ -1,4 +1,4 @@ -; RUN: opt -S -basicaa -inline -scalarrepl -instcombine -simplifycfg -instcombine -gvn -globaldce %s | FileCheck %s +; RUN: opt -S -Os < %s | FileCheck %s target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64" target triple = "x86_64-apple-darwin10.0.0" diff --git a/test/Transforms/Inline/gvn-inline-iteration.ll b/test/Transforms/Inline/gvn-inline-iteration.ll index e502fd5777d5a..526ed79e7b484 100644 --- a/test/Transforms/Inline/gvn-inline-iteration.ll +++ b/test/Transforms/Inline/gvn-inline-iteration.ll @@ -1,4 +1,4 @@ -; RUN: opt -basicaa -inline -gvn %s -S -max-cg-scc-iterations=1 | FileCheck %s +; RUN: opt -basicaa -inline -gvn -S -max-cg-scc-iterations=1 < %s | FileCheck %s ; rdar://6295824 and PR6724 target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64" diff --git a/test/Transforms/Inline/inline-optsize.ll b/test/Transforms/Inline/inline-optsize.ll index 20d7426abd13a..3ad573a04e427 100644 --- a/test/Transforms/Inline/inline-optsize.ll +++ b/test/Transforms/Inline/inline-optsize.ll @@ -1,5 +1,5 @@ -; RUN: opt -S -Oz %s | FileCheck %s -check-prefix=OZ -; RUN: opt -S -O2 %s | FileCheck %s -check-prefix=O2 +; RUN: opt -S -Oz < %s | FileCheck %s -check-prefix=OZ +; RUN: opt -S -O2 < %s | FileCheck %s -check-prefix=O2 ; The inline threshold for a function with the optsize attribute is currently ; the same as the global inline threshold for -Os. Check that the optsize diff --git a/test/Transforms/Inline/inline_constprop.ll b/test/Transforms/Inline/inline_constprop.ll index 0b48a7282f451..77bc3784acb40 100644 --- a/test/Transforms/Inline/inline_constprop.ll +++ b/test/Transforms/Inline/inline_constprop.ll @@ -111,6 +111,82 @@ bb.false: ret i32 %sub } +declare {i8, i1} @llvm.uadd.with.overflow.i8(i8 %a, i8 %b) + +define i8 @caller4(i8 %z) { +; Check that we can constant fold through intrinsics such as the +; overflow-detecting arithmetic instrinsics. These are particularly important +; as they are used heavily in standard library code and generic C++ code where +; the arguments are oftent constant but complete generality is required. +; +; CHECK: @caller4 +; CHECK-NOT: call +; CHECK: ret i8 -1 + +entry: + %x = call i8 @callee4(i8 254, i8 14, i8 %z) + ret i8 %x +} + +define i8 @callee4(i8 %x, i8 %y, i8 %z) { + %uadd = call {i8, i1} @llvm.uadd.with.overflow.i8(i8 %x, i8 %y) + %o = extractvalue {i8, i1} %uadd, 1 + br i1 %o, label %bb.true, label %bb.false + +bb.true: + ret i8 -1 + +bb.false: + ; This block musn't be counted in the inline cost. + %z1 = add i8 %z, 1 + %z2 = add i8 %z1, 1 + %z3 = add i8 %z2, 1 + %z4 = add i8 %z3, 1 + %z5 = add i8 %z4, 1 + %z6 = add i8 %z5, 1 + %z7 = add i8 %z6, 1 + %z8 = add i8 %z7, 1 + ret i8 %z8 +} + +define i64 @caller5(i64 %y) { +; Check that we can round trip constants through various kinds of casts etc w/o +; losing track of the constant prop in the inline cost analysis. +; +; CHECK: @caller5 +; CHECK-NOT: call +; CHECK: ret i64 -1 + +entry: + %x = call i64 @callee5(i64 42, i64 %y) + ret i64 %x +} + +define i64 @callee5(i64 %x, i64 %y) { + %inttoptr = inttoptr i64 %x to i8* + %bitcast = bitcast i8* %inttoptr to i32* + %ptrtoint = ptrtoint i32* %bitcast to i64 + %trunc = trunc i64 %ptrtoint to i32 + %zext = zext i32 %trunc to i64 + %cmp = icmp eq i64 %zext, 42 + br i1 %cmp, label %bb.true, label %bb.false + +bb.true: + ret i64 -1 + +bb.false: + ; This block musn't be counted in the inline cost. + %y1 = add i64 %y, 1 + %y2 = add i64 %y1, 1 + %y3 = add i64 %y2, 1 + %y4 = add i64 %y3, 1 + %y5 = add i64 %y4, 1 + %y6 = add i64 %y5, 1 + %y7 = add i64 %y6, 1 + %y8 = add i64 %y7, 1 + ret i64 %y8 +} + define i32 @PR13412.main() { ; This is a somewhat complicated three layer subprogram that was reported to diff --git a/test/Transforms/Inline/inline_invoke.ll b/test/Transforms/Inline/inline_invoke.ll index 9f5f670b859bc..c3941388f9378 100644 --- a/test/Transforms/Inline/inline_invoke.ll +++ b/test/Transforms/Inline/inline_invoke.ll @@ -96,6 +96,7 @@ eh.resume: ; CHECK: landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0 ; CHECK-NEXT: cleanup ; CHECK-NEXT: catch i8* bitcast (i8** @_ZTIi to i8*) +; CHECK-NEXT: catch i8* bitcast (i8** @_ZTIi to i8*) ; CHECK-NEXT: invoke void @_ZN1AD1Ev(%struct.A* [[A]]) ; CHECK-NEXT: to label %[[LBL:[^\s]+]] unwind ; CHECK: [[LBL]]: @@ -166,6 +167,7 @@ eh.resume: ; CHECK-NEXT: [[LPADVAL1:%.*]] = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0 ; CHECK-NEXT: cleanup ; CHECK-NEXT: catch i8* bitcast (i8** @_ZTIi to i8*) +; CHECK-NEXT: catch i8* bitcast (i8** @_ZTIi to i8*) ; CHECK-NEXT: invoke void @_ZN1AD1Ev(%struct.A* [[A1]]) ; CHECK-NEXT: to label %[[RESUME1:[^\s]+]] unwind ; CHECK: [[RESUME1]]: @@ -185,6 +187,7 @@ eh.resume: ; CHECK-NEXT: [[LPADVAL2:%.*]] = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0 ; CHECK-NEXT: cleanup ; CHECK-NEXT: catch i8* bitcast (i8** @_ZTIi to i8*) +; CHECK-NEXT: catch i8* bitcast (i8** @_ZTIi to i8*) ; CHECK-NEXT: invoke void @_ZN1AD1Ev(%struct.A* [[A2]]) ; CHECK-NEXT: to label %[[RESUME2:[^\s]+]] unwind ; CHECK: [[RESUME2]]: @@ -272,6 +275,7 @@ lpad.cont: ; CHECK: landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0 ; CHECK-NEXT: cleanup ; CHECK-NEXT: catch i8* bitcast (i8** @_ZTIi to i8*) +; CHECK-NEXT: catch i8* bitcast (i8** @_ZTIi to i8*) ; CHECK-NEXT: invoke void @_ZN1AD1Ev( ; CHECK-NEXT: to label %[[L:[^\s]+]] unwind ; CHECK: [[L]]: @@ -318,6 +322,7 @@ terminate: ; CHECK: landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0 ; CHECK-NEXT: cleanup ; CHECK-NEXT: catch i8* bitcast (i8** @_ZTIi to i8*) +; CHECK-NEXT: catch i8* bitcast (i8** @_ZTIi to i8*) ; CHECK-NEXT: invoke void @_ZN1AD1Ev( ; CHECK-NEXT: to label %[[L:[^\s]+]] unwind ; CHECK: [[L]]: @@ -330,7 +335,7 @@ terminate: ; CHECK-NEXT: br label %[[JOIN]] ; CHECK: [[JOIN]]: ; CHECK-NEXT: phi { i8*, i32 } -; CHECK-NEXT: call void @opaque() nounwind +; CHECK-NEXT: call void @opaque() [[NUW:#[0-9]+]] ; CHECK-NEXT: br label %[[FIX:[^\s]+]] ; CHECK: lpad: ; CHECK-NEXT: landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0 @@ -340,3 +345,8 @@ terminate: ; CHECK-NEXT: [[T1:%.*]] = phi i32 [ 0, %[[JOIN]] ], [ 1, %lpad ] ; CHECK-NEXT: call void @use(i32 [[T1]]) ; CHECK-NEXT: call void @_ZSt9terminatev() + +; CHECK: attributes [[NUW]] = { nounwind } +; CHECK: attributes #1 = { nounwind readnone } +; CHECK: attributes #2 = { ssp uwtable } +; CHECK: attributes #3 = { noreturn nounwind } diff --git a/test/Transforms/Inline/inline_minisize.ll b/test/Transforms/Inline/inline_minisize.ll new file mode 100644 index 0000000000000..3dddbcf3303d5 --- /dev/null +++ b/test/Transforms/Inline/inline_minisize.ll @@ -0,0 +1,232 @@ +; RUN: opt -O2 -S < %s | FileCheck %s + +@data = common global i32* null, align 8 + +define i32 @fct1(i32 %a) nounwind uwtable ssp { +entry: + %a.addr = alloca i32, align 4 + %res = alloca i32, align 4 + %i = alloca i32, align 4 + store i32 %a, i32* %a.addr, align 4 + %tmp = load i32* %a.addr, align 4 + %idxprom = sext i32 %tmp to i64 + %tmp1 = load i32** @data, align 8 + %arrayidx = getelementptr inbounds i32* %tmp1, i64 %idxprom + %tmp2 = load i32* %arrayidx, align 4 + %tmp3 = load i32* %a.addr, align 4 + %add = add nsw i32 %tmp3, 1 + %idxprom1 = sext i32 %add to i64 + %tmp4 = load i32** @data, align 8 + %arrayidx2 = getelementptr inbounds i32* %tmp4, i64 %idxprom1 + %tmp5 = load i32* %arrayidx2, align 4 + %mul = mul nsw i32 %tmp2, %tmp5 + store i32 %mul, i32* %res, align 4 + store i32 0, i32* %i, align 4 + store i32 0, i32* %i, align 4 + br label %for.cond + +for.cond: ; preds = %for.inc, %entry + %tmp6 = load i32* %i, align 4 + %tmp7 = load i32* %res, align 4 + %cmp = icmp slt i32 %tmp6, %tmp7 + br i1 %cmp, label %for.body, label %for.end + +for.body: ; preds = %for.cond + %tmp8 = load i32* %i, align 4 + %idxprom3 = sext i32 %tmp8 to i64 + %tmp9 = load i32** @data, align 8 + %arrayidx4 = getelementptr inbounds i32* %tmp9, i64 %idxprom3 + call void @fct0(i32* %arrayidx4) + br label %for.inc + +for.inc: ; preds = %for.body + %tmp10 = load i32* %i, align 4 + %inc = add nsw i32 %tmp10, 1 + store i32 %inc, i32* %i, align 4 + br label %for.cond + +for.end: ; preds = %for.cond + store i32 0, i32* %i, align 4 + br label %for.cond5 + +for.cond5: ; preds = %for.inc10, %for.end + %tmp11 = load i32* %i, align 4 + %tmp12 = load i32* %res, align 4 + %cmp6 = icmp slt i32 %tmp11, %tmp12 + br i1 %cmp6, label %for.body7, label %for.end12 + +for.body7: ; preds = %for.cond5 + %tmp13 = load i32* %i, align 4 + %idxprom8 = sext i32 %tmp13 to i64 + %tmp14 = load i32** @data, align 8 + %arrayidx9 = getelementptr inbounds i32* %tmp14, i64 %idxprom8 + call void @fct0(i32* %arrayidx9) + br label %for.inc10 + +for.inc10: ; preds = %for.body7 + %tmp15 = load i32* %i, align 4 + %inc11 = add nsw i32 %tmp15, 1 + store i32 %inc11, i32* %i, align 4 + br label %for.cond5 + +for.end12: ; preds = %for.cond5 + store i32 0, i32* %i, align 4 + br label %for.cond13 + +for.cond13: ; preds = %for.inc18, %for.end12 + %tmp16 = load i32* %i, align 4 + %tmp17 = load i32* %res, align 4 + %cmp14 = icmp slt i32 %tmp16, %tmp17 + br i1 %cmp14, label %for.body15, label %for.end20 + +for.body15: ; preds = %for.cond13 + %tmp18 = load i32* %i, align 4 + %idxprom16 = sext i32 %tmp18 to i64 + %tmp19 = load i32** @data, align 8 + %arrayidx17 = getelementptr inbounds i32* %tmp19, i64 %idxprom16 + call void @fct0(i32* %arrayidx17) + br label %for.inc18 + +for.inc18: ; preds = %for.body15 + %tmp20 = load i32* %i, align 4 + %inc19 = add nsw i32 %tmp20, 1 + store i32 %inc19, i32* %i, align 4 + br label %for.cond13 + +for.end20: ; preds = %for.cond13 + %tmp21 = load i32* %res, align 4 + ret i32 %tmp21 +} + +declare void @fct0(i32*) + +define i32 @fct2(i32 %a) nounwind uwtable inlinehint ssp { +entry: + %a.addr = alloca i32, align 4 + %res = alloca i32, align 4 + %i = alloca i32, align 4 + store i32 %a, i32* %a.addr, align 4 + %tmp = load i32* %a.addr, align 4 + %shl = shl i32 %tmp, 1 + %idxprom = sext i32 %shl to i64 + %tmp1 = load i32** @data, align 8 + %arrayidx = getelementptr inbounds i32* %tmp1, i64 %idxprom + %tmp2 = load i32* %arrayidx, align 4 + %tmp3 = load i32* %a.addr, align 4 + %shl1 = shl i32 %tmp3, 1 + %add = add nsw i32 %shl1, 13 + %idxprom2 = sext i32 %add to i64 + %tmp4 = load i32** @data, align 8 + %arrayidx3 = getelementptr inbounds i32* %tmp4, i64 %idxprom2 + %tmp5 = load i32* %arrayidx3, align 4 + %mul = mul nsw i32 %tmp2, %tmp5 + store i32 %mul, i32* %res, align 4 + store i32 0, i32* %i, align 4 + store i32 0, i32* %i, align 4 + br label %for.cond + +for.cond: ; preds = %for.inc, %entry + %tmp6 = load i32* %i, align 4 + %tmp7 = load i32* %res, align 4 + %cmp = icmp slt i32 %tmp6, %tmp7 + br i1 %cmp, label %for.body, label %for.end + +for.body: ; preds = %for.cond + %tmp8 = load i32* %i, align 4 + %idxprom4 = sext i32 %tmp8 to i64 + %tmp9 = load i32** @data, align 8 + %arrayidx5 = getelementptr inbounds i32* %tmp9, i64 %idxprom4 + call void @fct0(i32* %arrayidx5) + br label %for.inc + +for.inc: ; preds = %for.body + %tmp10 = load i32* %i, align 4 + %inc = add nsw i32 %tmp10, 1 + store i32 %inc, i32* %i, align 4 + br label %for.cond + +for.end: ; preds = %for.cond + store i32 0, i32* %i, align 4 + br label %for.cond6 + +for.cond6: ; preds = %for.inc11, %for.end + %tmp11 = load i32* %i, align 4 + %tmp12 = load i32* %res, align 4 + %cmp7 = icmp slt i32 %tmp11, %tmp12 + br i1 %cmp7, label %for.body8, label %for.end13 + +for.body8: ; preds = %for.cond6 + %tmp13 = load i32* %i, align 4 + %idxprom9 = sext i32 %tmp13 to i64 + %tmp14 = load i32** @data, align 8 + %arrayidx10 = getelementptr inbounds i32* %tmp14, i64 %idxprom9 + call void @fct0(i32* %arrayidx10) + br label %for.inc11 + +for.inc11: ; preds = %for.body8 + %tmp15 = load i32* %i, align 4 + %inc12 = add nsw i32 %tmp15, 1 + store i32 %inc12, i32* %i, align 4 + br label %for.cond6 + +for.end13: ; preds = %for.cond6 + store i32 0, i32* %i, align 4 + br label %for.cond14 + +for.cond14: ; preds = %for.inc19, %for.end13 + %tmp16 = load i32* %i, align 4 + %tmp17 = load i32* %res, align 4 + %cmp15 = icmp slt i32 %tmp16, %tmp17 + br i1 %cmp15, label %for.body16, label %for.end21 + +for.body16: ; preds = %for.cond14 + %tmp18 = load i32* %i, align 4 + %idxprom17 = sext i32 %tmp18 to i64 + %tmp19 = load i32** @data, align 8 + %arrayidx18 = getelementptr inbounds i32* %tmp19, i64 %idxprom17 + call void @fct0(i32* %arrayidx18) + br label %for.inc19 + +for.inc19: ; preds = %for.body16 + %tmp20 = load i32* %i, align 4 + %inc20 = add nsw i32 %tmp20, 1 + store i32 %inc20, i32* %i, align 4 + br label %for.cond14 + +for.end21: ; preds = %for.cond14 + %tmp21 = load i32* %res, align 4 + ret i32 %tmp21 +} + +define i32 @fct3(i32 %c) nounwind uwtable ssp { +entry: + ;CHECK: @fct3 + ;CHECK: call i32 @fct1 + ; The inline keyword gives a sufficient benefits to inline fct2 + ;CHECK-NOT: call i32 @fct2 + %c.addr = alloca i32, align 4 + store i32 %c, i32* %c.addr, align 4 + %tmp = load i32* %c.addr, align 4 + %call = call i32 @fct1(i32 %tmp) + %tmp1 = load i32* %c.addr, align 4 + %call1 = call i32 @fct2(i32 %tmp1) + %add = add nsw i32 %call, %call1 + ret i32 %add +} + +define i32 @fct4(i32 %c) minsize nounwind uwtable ssp { +entry: + ;CHECK: @fct4 + ;CHECK: call i32 @fct1 + ; With Oz (minsize attribute), the benefit of inlining fct2 + ; is the same as fct1, thus no inlining for fct2 + ;CHECK: call i32 @fct2 + %c.addr = alloca i32, align 4 + store i32 %c, i32* %c.addr, align 4 + %tmp = load i32* %c.addr, align 4 + %call = call i32 @fct1(i32 %tmp) + %tmp1 = load i32* %c.addr, align 4 + %call1 = call i32 @fct2(i32 %tmp1) + %add = add nsw i32 %call, %call1 + ret i32 %add +} diff --git a/test/Transforms/Inline/inline_ssp.ll b/test/Transforms/Inline/inline_ssp.ll new file mode 100644 index 0000000000000..a4b43a77bad23 --- /dev/null +++ b/test/Transforms/Inline/inline_ssp.ll @@ -0,0 +1,160 @@ +; RUN: opt -inline %s -S | FileCheck %s +; Ensure SSP attributes are propagated correctly when inlining. + +@.str = private unnamed_addr constant [11 x i8] c"fun_nossp\0A\00", align 1 +@.str1 = private unnamed_addr constant [9 x i8] c"fun_ssp\0A\00", align 1 +@.str2 = private unnamed_addr constant [15 x i8] c"fun_sspstrong\0A\00", align 1 +@.str3 = private unnamed_addr constant [12 x i8] c"fun_sspreq\0A\00", align 1 + +; These first four functions (@fun_sspreq, @fun_sspstrong, @fun_ssp, @fun_nossp) +; are used by the remaining functions to ensure that the SSP attributes are +; propagated correctly. The caller should have its SSP attribute set as: +; strictest(caller-ssp-attr, callee-ssp-attr), where strictness is ordered as: +; sspreq > sspstrong > ssp > [no ssp] +define internal void @fun_sspreq() nounwind sspreq uwtable { +entry: + %call = call i32 (i8*, ...)* @printf(i8* getelementptr inbounds ([12 x i8]* @.str3, i32 0, i32 0)) + ret void +} + +define internal void @fun_sspstrong() nounwind sspstrong uwtable { +entry: + %call = call i32 (i8*, ...)* @printf(i8* getelementptr inbounds ([15 x i8]* @.str2, i32 0, i32 0)) + ret void +} + +define internal void @fun_ssp() nounwind ssp uwtable { +entry: + %call = call i32 (i8*, ...)* @printf(i8* getelementptr inbounds ([9 x i8]* @.str1, i32 0, i32 0)) + ret void +} + +define internal void @fun_nossp() nounwind uwtable { +entry: + %call = call i32 (i8*, ...)* @printf(i8* getelementptr inbounds ([11 x i8]* @.str, i32 0, i32 0)) + ret void +} + +; Tests start below + +define void @inline_req_req() nounwind sspreq uwtable { +entry: +; CHECK: @inline_req_req() #0 + call void @fun_sspreq() + ret void +} + +define void @inline_req_strong() nounwind sspstrong uwtable { +entry: +; CHECK: @inline_req_strong() #0 + call void @fun_sspreq() + ret void +} + +define void @inline_req_ssp() nounwind ssp uwtable { +entry: +; CHECK: @inline_req_ssp() #0 + call void @fun_sspreq() + ret void +} + +define void @inline_req_nossp() nounwind uwtable { +entry: +; CHECK: @inline_req_nossp() #0 + call void @fun_sspreq() + ret void +} + +define void @inline_strong_req() nounwind sspreq uwtable { +entry: +; CHECK: @inline_strong_req() #0 + call void @fun_sspstrong() + ret void +} + + +define void @inline_strong_strong() nounwind sspstrong uwtable { +entry: +; CHECK: @inline_strong_strong() #1 + call void @fun_sspstrong() + ret void +} + +define void @inline_strong_ssp() nounwind ssp uwtable { +entry: +; CHECK: @inline_strong_ssp() #1 + call void @fun_sspstrong() + ret void +} + +define void @inline_strong_nossp() nounwind uwtable { +entry: +; CHECK: @inline_strong_nossp() #1 + call void @fun_sspstrong() + ret void +} + +define void @inline_ssp_req() nounwind sspreq uwtable { +entry: +; CHECK: @inline_ssp_req() #0 + call void @fun_ssp() + ret void +} + + +define void @inline_ssp_strong() nounwind sspstrong uwtable { +entry: +; CHECK: @inline_ssp_strong() #1 + call void @fun_ssp() + ret void +} + +define void @inline_ssp_ssp() nounwind ssp uwtable { +entry: +; CHECK: @inline_ssp_ssp() #2 + call void @fun_ssp() + ret void +} + +define void @inline_ssp_nossp() nounwind uwtable { +entry: +; CHECK: @inline_ssp_nossp() #2 + call void @fun_ssp() + ret void +} + +define void @inline_nossp_req() nounwind uwtable sspreq { +entry: +; CHECK: @inline_nossp_req() #0 + call void @fun_nossp() + ret void +} + + +define void @inline_nossp_strong() nounwind sspstrong uwtable { +entry: +; CHECK: @inline_nossp_strong() #1 + call void @fun_nossp() + ret void +} + +define void @inline_nossp_ssp() nounwind ssp uwtable { +entry: +; CHECK: @inline_nossp_ssp() #2 + call void @fun_nossp() + ret void +} + +define void @inline_nossp_nossp() nounwind uwtable { +entry: +; CHECK: @inline_nossp_nossp() #3 + call void @fun_nossp() + ret void +} + +declare i32 @printf(i8*, ...) + +; CHECK: attributes #0 = { nounwind sspreq uwtable } +; CHECK: attributes #1 = { nounwind sspstrong uwtable } +; CHECK: attributes #2 = { nounwind ssp uwtable } +; CHECK: attributes #3 = { nounwind uwtable } diff --git a/test/Transforms/Inline/lifetime-no-datalayout.ll b/test/Transforms/Inline/lifetime-no-datalayout.ll new file mode 100644 index 0000000000000..f4ffef3850f1b --- /dev/null +++ b/test/Transforms/Inline/lifetime-no-datalayout.ll @@ -0,0 +1,23 @@ +; RUN: opt -inline -S < %s | FileCheck %s + +declare void @use(i8* %a) + +define void @helper() { + %a = alloca i8 + call void @use(i8* %a) + ret void +} + +; Size in llvm.lifetime.X should be -1 (unknown). +define void @test() { +; CHECK: @test +; CHECK-NOT: lifetime +; CHECK: llvm.lifetime.start(i64 -1 +; CHECK-NOT: lifetime +; CHECK: llvm.lifetime.end(i64 -1 + call void @helper() +; CHECK-NOT: lifetime +; CHECK: ret void + ret void +} + diff --git a/test/Transforms/Inline/lifetime.ll b/test/Transforms/Inline/lifetime.ll index a95c836b77de6..fc73385295ed8 100644 --- a/test/Transforms/Inline/lifetime.ll +++ b/test/Transforms/Inline/lifetime.ll @@ -1,22 +1,25 @@ -; RUN: opt -inline %s -S -o - | FileCheck %s +; RUN: opt -inline -S < %s | FileCheck %s +target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128" declare void @llvm.lifetime.start(i64, i8*) declare void @llvm.lifetime.end(i64, i8*) define void @helper_both_markers() { %a = alloca i8 - call void @llvm.lifetime.start(i64 1, i8* %a) - call void @llvm.lifetime.end(i64 1, i8* %a) + ; Size in llvm.lifetime.start / llvm.lifetime.end differs from + ; allocation size. We should use the former. + call void @llvm.lifetime.start(i64 2, i8* %a) + call void @llvm.lifetime.end(i64 2, i8* %a) ret void } define void @test_both_markers() { ; CHECK: @test_both_markers -; CHECK: llvm.lifetime.start(i64 1 -; CHECK-NEXT: llvm.lifetime.end(i64 1 +; CHECK: llvm.lifetime.start(i64 2 +; CHECK-NEXT: llvm.lifetime.end(i64 2 call void @helper_both_markers() -; CHECK-NEXT: llvm.lifetime.start(i64 1 -; CHECK-NEXT: llvm.lifetime.end(i64 1 +; CHECK-NEXT: llvm.lifetime.start(i64 2 +; CHECK-NEXT: llvm.lifetime.end(i64 2 call void @helper_both_markers() ; CHECK-NEXT: ret void ret void @@ -27,7 +30,7 @@ define void @test_both_markers() { declare void @use(i8* %a) define void @helper_no_markers() { - %a = alloca i8 + %a = alloca i8 ; Allocation size is 1 byte. call void @use(i8* %a) ret void } @@ -37,14 +40,14 @@ define void @helper_no_markers() { define void @test_no_marker() { ; CHECK: @test_no_marker ; CHECK-NOT: lifetime -; CHECK: llvm.lifetime.start(i64 -1 +; CHECK: llvm.lifetime.start(i64 1 ; CHECK-NOT: lifetime -; CHECK: llvm.lifetime.end(i64 -1 +; CHECK: llvm.lifetime.end(i64 1 call void @helper_no_markers() ; CHECK-NOT: lifetime -; CHECK: llvm.lifetime.start(i64 -1 +; CHECK: llvm.lifetime.start(i64 1 ; CHECK-NOT: lifetime -; CHECK: llvm.lifetime.end(i64 -1 +; CHECK: llvm.lifetime.end(i64 1 call void @helper_no_markers() ; CHECK-NOT: lifetime ; CHECK: ret void @@ -76,3 +79,22 @@ define void @test_two_casts() { ; CHECK: ret void ret void } + +define void @helper_arrays_alloca() { + %a = alloca [10 x i32], align 16 + %1 = bitcast [10 x i32]* %a to i8* + call void @use(i8* %1) + ret void +} + +define void @test_arrays_alloca() { +; CHECK: @test_arrays_alloca +; CHECK-NOT: lifetime +; CHECK: llvm.lifetime.start(i64 40, +; CHECK-NOT: lifetime +; CHECK: llvm.lifetime.end(i64 40, + call void @helper_arrays_alloca() +; CHECK-NOT: lifetime +; CHECK: ret void + ret void +} diff --git a/test/Transforms/Inline/noinline-recursive-fn.ll b/test/Transforms/Inline/noinline-recursive-fn.ll index 6cde0e27fd1eb..5520093ee4570 100644 --- a/test/Transforms/Inline/noinline-recursive-fn.ll +++ b/test/Transforms/Inline/noinline-recursive-fn.ll @@ -2,7 +2,7 @@ ; This effectively is just peeling off the first iteration of a loop, and the ; inliner heuristics are not set up for this. -; RUN: opt -inline %s -S | FileCheck %s +; RUN: opt -inline -S < %s | FileCheck %s target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64" target triple = "x86_64-apple-darwin10.3" diff --git a/test/Transforms/Inline/noinline.ll b/test/Transforms/Inline/noinline.ll index dc3f6e003094a..7667114b68e66 100644 --- a/test/Transforms/Inline/noinline.ll +++ b/test/Transforms/Inline/noinline.ll @@ -1,4 +1,4 @@ -; RUN: opt %s -inline -S | FileCheck %s +; RUN: opt -inline -S < %s | FileCheck %s ; PR6682 declare void @foo() nounwind diff --git a/test/Transforms/Inline/recursive.ll b/test/Transforms/Inline/recursive.ll index 5fe8d1639ca36..fe1c041af9a87 100644 --- a/test/Transforms/Inline/recursive.ll +++ b/test/Transforms/Inline/recursive.ll @@ -1,4 +1,4 @@ -; RUN: opt %s -inline -S | FileCheck %s +; RUN: opt -inline -S < %s | FileCheck %s target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128" target triple = "i386-apple-darwin10.0" diff --git a/test/Transforms/InstCombine/2008-05-08-StrLenSink.ll b/test/Transforms/InstCombine/2008-05-08-StrLenSink.ll index 1da28562aae40..d266164fd8707 100644 --- a/test/Transforms/InstCombine/2008-05-08-StrLenSink.ll +++ b/test/Transforms/InstCombine/2008-05-08-StrLenSink.ll @@ -1,4 +1,4 @@ -; RUN: opt -S -instcombine %s | FileCheck %s +; RUN: opt -S -instcombine < %s | FileCheck %s ; PR2297 target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128" target triple = "i386-apple-darwin8" diff --git a/test/Transforms/InstCombine/2009-02-11-NotInitialized.ll b/test/Transforms/InstCombine/2009-02-11-NotInitialized.ll new file mode 100644 index 0000000000000..b66495d9cbaa3 --- /dev/null +++ b/test/Transforms/InstCombine/2009-02-11-NotInitialized.ll @@ -0,0 +1,14 @@ +; RUN: opt < %s -inline -instcombine -functionattrs | llvm-dis +; +; Check that nocapture attributes are added when run after an SCC pass. +; PR3520 + +define i32 @use(i8* %x) nounwind readonly { +; CHECK: @use(i8* nocapture %x) + %1 = tail call i64 @strlen(i8* %x) nounwind readonly + %2 = trunc i64 %1 to i32 + ret i32 %2 +} + +declare i64 @strlen(i8*) nounwind readonly +; CHECK: declare i64 @strlen(i8* nocapture) nounwind readonly diff --git a/test/Transforms/InstCombine/2010-03-03-ExtElim.ll b/test/Transforms/InstCombine/2010-03-03-ExtElim.ll index 2df12d670adbe..bb3159e1e6fa1 100644 --- a/test/Transforms/InstCombine/2010-03-03-ExtElim.ll +++ b/test/Transforms/InstCombine/2010-03-03-ExtElim.ll @@ -1,4 +1,4 @@ -; RUN: opt -instcombine -S %s | FileCheck %s +; RUN: opt -instcombine -S < %s | FileCheck %s ; PR6486 target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:32:32-n8:16:32" diff --git a/test/Transforms/InstCombine/2010-05-30-memcpy-Struct.ll b/test/Transforms/InstCombine/2010-05-30-memcpy-Struct.ll new file mode 100644 index 0000000000000..09a96749f2606 --- /dev/null +++ b/test/Transforms/InstCombine/2010-05-30-memcpy-Struct.ll @@ -0,0 +1,20 @@ +; RUN: opt -instcombine -S < %s | FileCheck %s +; PR7265 + +target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64" +target triple = "x86_64-unknown-linux-gnu" + +%union.anon = type { i32, [4 x i8] } + +@.str = private constant [3 x i8] c"%s\00" + +define void @CopyEventArg(%union.anon* %ev) nounwind { +entry: + %call = call i32 (i8*, i8*, ...)* @sprintf(i8* undef, i8* getelementptr inbounds ([3 x i8]* @.str, i64 0, i64 0), %union.anon* %ev) nounwind +; CHECK: bitcast %union.anon* %ev to i8* +; CHECK: call void @llvm.memcpy.p0i8.p0i8.i64 + ret void +} + +declare i32 @sprintf(i8*, i8*, ...) + diff --git a/test/Transforms/InstCombine/2010-11-01-lshr-mask.ll b/test/Transforms/InstCombine/2010-11-01-lshr-mask.ll index eb28994756952..8001621979197 100644 --- a/test/Transforms/InstCombine/2010-11-01-lshr-mask.ll +++ b/test/Transforms/InstCombine/2010-11-01-lshr-mask.ll @@ -5,8 +5,8 @@ define i32 @main(i32 %argc) nounwind ssp { entry: %tmp3151 = trunc i32 %argc to i8 -; CHECK: %tmp3163 = shl i8 %tmp3162, 6 -; CHECK: and i8 %tmp3163, 64 +; CHECK: %0 = shl i8 %tmp3151, 5 +; CHECK: and i8 %0, 64 ; CHECK-NOT: shl ; CHECK-NOT: shr %tmp3161 = or i8 %tmp3151, -17 diff --git a/test/Transforms/InstCombine/2012-04-23-Neon-Intrinsics.ll b/test/Transforms/InstCombine/2012-04-23-Neon-Intrinsics.ll index 0907c490bb35d..2dedd44e2be10 100644 --- a/test/Transforms/InstCombine/2012-04-23-Neon-Intrinsics.ll +++ b/test/Transforms/InstCombine/2012-04-23-Neon-Intrinsics.ll @@ -50,7 +50,7 @@ entry: %b = add <4 x i32> zeroinitializer, %a ret <4 x i32> %b ; CHECK: entry: -; CHECK-NEXT: %a = tail call <4 x i32> @llvm.arm.neon.vmulls.v4i32(<4 x i16> , <4 x i16> %x) nounwind +; CHECK-NEXT: %a = tail call <4 x i32> @llvm.arm.neon.vmulls.v4i32(<4 x i16> , <4 x i16> %x) [[NUW:#[0-9]+]] ; CHECK-NEXT: ret <4 x i32> %a } @@ -66,3 +66,7 @@ entry: declare <4 x i32> @llvm.arm.neon.vmulls.v4i32(<4 x i16>, <4 x i16>) nounwind readnone declare <4 x i32> @llvm.arm.neon.vmullu.v4i32(<4 x i16>, <4 x i16>) nounwind readnone + +; CHECK: attributes #0 = { nounwind readnone ssp } +; CHECK: attributes #1 = { nounwind readnone } +; CHECK: attributes [[NUW]] = { nounwind } diff --git a/test/Transforms/InstCombine/2012-12-14-simp-vgep.ll b/test/Transforms/InstCombine/2012-12-14-simp-vgep.ll new file mode 100644 index 0000000000000..fc29b095e5ced --- /dev/null +++ b/test/Transforms/InstCombine/2012-12-14-simp-vgep.ll @@ -0,0 +1,10 @@ +; RUN: opt < %s -instcombine -S + +target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128" + +define <4 x i32> @foo(<4 x i32*>* %in) { + %t17 = load <4 x i32*>* %in, align 8 + %t18 = icmp eq <4 x i32*> %t17, zeroinitializer + %t19 = zext <4 x i1> %t18 to <4 x i32> + ret <4 x i32> %t19 +} diff --git a/test/Transforms/InstCombine/2013-03-05-Combine-BitcastTy-Into-Alloca.ll b/test/Transforms/InstCombine/2013-03-05-Combine-BitcastTy-Into-Alloca.ll new file mode 100644 index 0000000000000..b20c3a07c0ac6 --- /dev/null +++ b/test/Transforms/InstCombine/2013-03-05-Combine-BitcastTy-Into-Alloca.ll @@ -0,0 +1,45 @@ +; RUN: opt < %s -instcombine -S | FileCheck %s +target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128" +target triple = "x86_64-apple-macosx10.9.0" + +%struct._my_struct = type <{ [12 x i8], [4 x i8] }> + +@initval = common global %struct._my_struct zeroinitializer, align 1 + +; InstCombine will try to change the %struct._my_struct alloca into an +; allocation of an i96 because of the bitcast to create %2. That's not valid, +; as the other 32 bits of the structure still feed into the return value +define { i64, i64 } @function(i32 %x, i32 %y, i32 %z) nounwind { +; CHECK: @function +; CHECK-NEXT: entry: +; CHECK-NEXT: %retval = alloca %struct._my_struct, align 8 +; CHECK-NOT: bitcast i96* %retval to %struct._my_struct* +entry: + %retval = alloca %struct._my_struct, align 8 + %k.sroa.0.0.copyload = load i96* bitcast (%struct._my_struct* @initval to i96*), align 1 + %k.sroa.1.12.copyload = load i32* bitcast ([4 x i8]* getelementptr inbounds (%struct._my_struct* @initval, i64 0, i32 1) to i32*), align 1 + %0 = zext i32 %x to i96 + %bf.value = shl nuw nsw i96 %0, 6 + %bf.clear = and i96 %k.sroa.0.0.copyload, -288230376151711744 + %1 = zext i32 %y to i96 + %bf.value2 = shl nuw nsw i96 %1, 32 + %bf.shl3 = and i96 %bf.value2, 288230371856744448 + %bf.value.masked = and i96 %bf.value, 4294967232 + %2 = zext i32 %z to i96 + %bf.value8 = and i96 %2, 63 + %bf.clear4 = or i96 %bf.shl3, %bf.value.masked + %bf.set5 = or i96 %bf.clear4, %bf.value8 + %bf.set10 = or i96 %bf.set5, %bf.clear + %retval.0.cast7 = bitcast %struct._my_struct* %retval to i96* + store i96 %bf.set10, i96* %retval.0.cast7, align 8 + %retval.12.idx8 = getelementptr inbounds %struct._my_struct* %retval, i64 0, i32 1 + %retval.12.cast9 = bitcast [4 x i8]* %retval.12.idx8 to i32* + store i32 %k.sroa.1.12.copyload, i32* %retval.12.cast9, align 4 + %trunc = trunc i96 %bf.set10 to i64 + %.fca.0.insert = insertvalue { i64, i64 } undef, i64 %trunc, 0 + %retval.8.idx12 = getelementptr inbounds %struct._my_struct* %retval, i64 0, i32 0, i64 8 + %retval.8.cast13 = bitcast i8* %retval.8.idx12 to i64* + %retval.8.load14 = load i64* %retval.8.cast13, align 8 + %.fca.1.insert = insertvalue { i64, i64 } %.fca.0.insert, i64 %retval.8.load14, 1 + ret { i64, i64 } %.fca.1.insert +} diff --git a/test/Transforms/InstCombine/abs-1.ll b/test/Transforms/InstCombine/abs-1.ll new file mode 100644 index 0000000000000..807f238755b5d --- /dev/null +++ b/test/Transforms/InstCombine/abs-1.ll @@ -0,0 +1,41 @@ +; Test that the abs library call simplifier works correctly. +; +; RUN: opt < %s -instcombine -S | FileCheck %s + +target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128" + +declare i32 @abs(i32) +declare i64 @labs(i64) +declare i64 @llabs(i64) + +; Check abs(x) -> x >s -1 ? x : -x. + +define i32 @test_simplify1(i32 %x) { +; CHECK: @test_simplify1 + %ret = call i32 @abs(i32 %x) +; CHECK-NEXT: [[ISPOS:%[a-z0-9]+]] = icmp sgt i32 %x, -1 +; CHECK-NEXT: [[NEG:%[a-z0-9]+]] = sub i32 0, %x +; CHECK-NEXT: [[RET:%[a-z0-9]+]] = select i1 [[ISPOS]], i32 %x, i32 [[NEG]] + ret i32 %ret +; CHECK-NEXT: ret i32 [[RET]] +} + +define i64 @test_simplify2(i64 %x) { +; CHECK: @test_simplify2 + %ret = call i64 @labs(i64 %x) +; CHECK-NEXT: [[ISPOS:%[a-z0-9]+]] = icmp sgt i64 %x, -1 +; CHECK-NEXT: [[NEG:%[a-z0-9]+]] = sub i64 0, %x +; CHECK-NEXT: [[RET:%[a-z0-9]+]] = select i1 [[ISPOS]], i64 %x, i64 [[NEG]] + ret i64 %ret +; CHECK-NEXT: ret i64 [[RET]] +} + +define i64 @test_simplify3(i64 %x) { +; CHECK: @test_simplify3 + %ret = call i64 @llabs(i64 %x) +; CHECK-NEXT: [[ISPOS:%[a-z0-9]+]] = icmp sgt i64 %x, -1 +; CHECK-NEXT: [[NEG:%[a-z0-9]+]] = sub i64 0, %x +; CHECK-NEXT: [[RET:%[a-z0-9]+]] = select i1 [[ISPOS]], i64 %x, i64 [[NEG]] + ret i64 %ret +; CHECK-NEXT: ret i64 [[RET]] +} diff --git a/test/Transforms/InstCombine/align-external.ll b/test/Transforms/InstCombine/align-external.ll index d4a5d429912bf..c3ef2dbb70f03 100644 --- a/test/Transforms/InstCombine/align-external.ll +++ b/test/Transforms/InstCombine/align-external.ll @@ -8,7 +8,7 @@ ; CHECK: %q = add i64 %r, 1 ; CHECK: ret i64 %q -target datalayout = "-i32:8:32" +target datalayout = "i32:8:32" @A = external global i32 @B = weak_odr global i32 0 diff --git a/test/Transforms/InstCombine/bitcast-bigendian.ll b/test/Transforms/InstCombine/bitcast-bigendian.ll new file mode 100644 index 0000000000000..4ded581a14c6b --- /dev/null +++ b/test/Transforms/InstCombine/bitcast-bigendian.ll @@ -0,0 +1,50 @@ +; RUN: opt < %s -instcombine -S | FileCheck %s + +target datalayout = "E-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-f128:128:128-v128:128:128-n32:64" +target triple = "powerpc64-unknown-linux-gnu" + +; These tests are extracted from bitcast.ll. +; Verify that they also work correctly on big-endian targets. + +define float @test2(<2 x float> %A, <2 x i32> %B) { + %tmp28 = bitcast <2 x float> %A to i64 ; [#uses=2] + %tmp23 = trunc i64 %tmp28 to i32 ; [#uses=1] + %tmp24 = bitcast i32 %tmp23 to float ; [#uses=1] + + %tmp = bitcast <2 x i32> %B to i64 + %tmp2 = trunc i64 %tmp to i32 ; [#uses=1] + %tmp4 = bitcast i32 %tmp2 to float ; [#uses=1] + + %add = fadd float %tmp24, %tmp4 + ret float %add + +; CHECK: @test2 +; CHECK-NEXT: %tmp24 = extractelement <2 x float> %A, i32 1 +; CHECK-NEXT: bitcast <2 x i32> %B to <2 x float> +; CHECK-NEXT: %tmp4 = extractelement <2 x float> {{.*}}, i32 1 +; CHECK-NEXT: %add = fadd float %tmp24, %tmp4 +; CHECK-NEXT: ret float %add +} + +define float @test3(<2 x float> %A, <2 x i64> %B) { + %tmp28 = bitcast <2 x float> %A to i64 + %tmp29 = lshr i64 %tmp28, 32 + %tmp23 = trunc i64 %tmp29 to i32 + %tmp24 = bitcast i32 %tmp23 to float + + %tmp = bitcast <2 x i64> %B to i128 + %tmp1 = lshr i128 %tmp, 64 + %tmp2 = trunc i128 %tmp1 to i32 + %tmp4 = bitcast i32 %tmp2 to float + + %add = fadd float %tmp24, %tmp4 + ret float %add + +; CHECK: @test3 +; CHECK-NEXT: %tmp24 = extractelement <2 x float> %A, i32 0 +; CHECK-NEXT: bitcast <2 x i64> %B to <4 x float> +; CHECK-NEXT: %tmp4 = extractelement <4 x float> {{.*}}, i32 1 +; CHECK-NEXT: %add = fadd float %tmp24, %tmp4 +; CHECK-NEXT: ret float %add +} + diff --git a/test/Transforms/InstCombine/bitcast-vector-fold.ll b/test/Transforms/InstCombine/bitcast-vector-fold.ll index 8feec229171af..8fd7f35b7bb72 100644 --- a/test/Transforms/InstCombine/bitcast-vector-fold.ll +++ b/test/Transforms/InstCombine/bitcast-vector-fold.ll @@ -31,3 +31,8 @@ define <4 x i32> @test6() { %tmp3 = bitcast <2 x double> to <4 x i32> ret <4 x i32> %tmp3 } + +define i32 @test7() { + %tmp3 = bitcast <2 x half> to i32 + ret i32 %tmp3 +} \ No newline at end of file diff --git a/test/Transforms/InstCombine/bitcast.ll b/test/Transforms/InstCombine/bitcast.ll index 8f6ae7d83527e..1e6113256bf36 100644 --- a/test/Transforms/InstCombine/bitcast.ll +++ b/test/Transforms/InstCombine/bitcast.ll @@ -11,7 +11,7 @@ define i32 @test1(i64 %a) { %t3 = xor <2 x i32> %t1, %t2 %t4 = extractelement <2 x i32> %t3, i32 0 ret i32 %t4 - + ; CHECK: @test1 ; CHECK: ret i32 0 } @@ -30,7 +30,7 @@ define float @test2(<2 x float> %A, <2 x i32> %B) { %add = fadd float %tmp24, %tmp4 ret float %add - + ; CHECK: @test2 ; CHECK-NEXT: %tmp24 = extractelement <2 x float> %A, i32 0 ; CHECK-NEXT: bitcast <2 x i32> %B to <2 x float> @@ -55,7 +55,7 @@ define float @test3(<2 x float> %A, <2 x i64> %B) { %add = fadd float %tmp24, %tmp4 ret float %add - + ; CHECK: @test3 ; CHECK-NEXT: %tmp24 = extractelement <2 x float> %A, i32 1 ; CHECK-NEXT: bitcast <2 x i64> %B to <4 x float> @@ -75,7 +75,7 @@ define <2 x i32> @test4(i32 %A, i32 %B){ ; CHECK: @test4 ; CHECK-NEXT: insertelement <2 x i32> undef, i32 %A, i32 0 ; CHECK-NEXT: insertelement <2 x i32> {{.*}}, i32 %B, i32 1 - ; CHECK-NEXT: ret <2 x i32> + ; CHECK-NEXT: ret <2 x i32> } @@ -92,7 +92,7 @@ define <2 x float> @test5(float %A, float %B) { ; CHECK: @test5 ; CHECK-NEXT: insertelement <2 x float> undef, float %A, i32 0 ; CHECK-NEXT: insertelement <2 x float> {{.*}}, float %B, i32 1 - ; CHECK-NEXT: ret <2 x float> + ; CHECK-NEXT: ret <2 x float> } define <2 x float> @test6(float %A){ @@ -123,7 +123,7 @@ define i64 @Vec2(i64 %in) { } define i64 @All11(i64 %in) { - %out = and i64 %in, xor (i64 bitcast (<2 x float> bitcast (i64 -1 to <2 x float>) to i64), i64 -1) + %out = and i64 %in, xor (i64 bitcast (<2 x float> bitcast (i64 -1 to <2 x float>) to i64), i64 -1) ret i64 %out ; CHECK: @All11 ; CHECK: ret i64 0 @@ -131,9 +131,16 @@ define i64 @All11(i64 %in) { define i32 @All111(i32 %in) { - %out = and i32 %in, xor (i32 bitcast (<1 x float> bitcast (i32 -1 to <1 x float>) to i32), i32 -1) + %out = and i32 %in, xor (i32 bitcast (<1 x float> bitcast (i32 -1 to <1 x float>) to i32), i32 -1) ret i32 %out ; CHECK: @All111 ; CHECK: ret i32 0 } +define <2 x i16> @BitcastInsert(i32 %a) { + %v = insertelement <1 x i32> undef, i32 %a, i32 0 + %r = bitcast <1 x i32> %v to <2 x i16> + ret <2 x i16> %r +; CHECK: @BitcastInsert +; CHECK: bitcast i32 %a to <2 x i16> +} diff --git a/test/Transforms/InstCombine/cast.ll b/test/Transforms/InstCombine/cast.ll index b4eb69d4363dd..de738bb7c06d7 100644 --- a/test/Transforms/InstCombine/cast.ll +++ b/test/Transforms/InstCombine/cast.ll @@ -473,14 +473,12 @@ define i64 @test51(i64 %A, i1 %cond) { %F = sext i32 %E to i64 ret i64 %F ; CHECK: @test51 - -; FIXME: disabled, see PR5997 -; HECK-NEXT: %C = and i64 %A, 4294967294 -; HECK-NEXT: %D = or i64 %A, 1 -; HECK-NEXT: %E = select i1 %cond, i64 %C, i64 %D -; HECK-NEXT: %sext = shl i64 %E, 32 -; HECK-NEXT: %F = ashr i64 %sext, 32 -; HECK-NEXT: ret i64 %F +; CHECK-NEXT: %C = and i64 %A, 4294967294 +; CHECK-NEXT: %D = or i64 %A, 1 +; CHECK-NEXT: %E = select i1 %cond, i64 %C, i64 %D +; CHECK-NEXT: %sext = shl i64 %E, 32 +; CHECK-NEXT: %F = ashr exact i64 %sext, 32 +; CHECK-NEXT: ret i64 %F } define i32 @test52(i64 %A) { diff --git a/test/Transforms/InstCombine/compare-signs.ll b/test/Transforms/InstCombine/compare-signs.ll index f8e49110610a2..72db66e3ab0f6 100644 --- a/test/Transforms/InstCombine/compare-signs.ll +++ b/test/Transforms/InstCombine/compare-signs.ll @@ -1,4 +1,4 @@ -; RUN: opt %s -instcombine -S | FileCheck %s +; RUN: opt -instcombine -S < %s | FileCheck %s ; PR5438 ; TODO: This should also optimize down. diff --git a/test/Transforms/InstCombine/constant-expr-datalayout.ll b/test/Transforms/InstCombine/constant-expr-datalayout.ll new file mode 100644 index 0000000000000..9a72c77afdb0f --- /dev/null +++ b/test/Transforms/InstCombine/constant-expr-datalayout.ll @@ -0,0 +1,12 @@ +; RUN: opt -instcombine %s -S -o - | FileCheck %s + +target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128" +target triple = "x86_64-unknown-linux-gnu" + +%test1.struct = type { i32, i32 } +@test1.aligned_glbl = global %test1.struct zeroinitializer, align 4 +define void @test1(i64 *%ptr) { + store i64 and (i64 ptrtoint (i32* getelementptr (%test1.struct* @test1.aligned_glbl, i32 0, i32 1) to i64), i64 3), i64* %ptr +; CHECK: store i64 0, i64* %ptr + ret void +} diff --git a/test/Transforms/InstCombine/cos-1.ll b/test/Transforms/InstCombine/cos-1.ll new file mode 100644 index 0000000000000..b92e448abd9f9 --- /dev/null +++ b/test/Transforms/InstCombine/cos-1.ll @@ -0,0 +1,38 @@ +; Test that the cos library call simplifier works correctly. +; +; RUN: opt < %s -instcombine -S | FileCheck %s -check-prefix=NO-FLOAT-SHRINK +; RUN: opt < %s -instcombine -enable-double-float-shrink -S | FileCheck %s -check-prefix=DO-FLOAT-SHRINK + +target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128" + +declare double @cos(double) + +; Check cos(-x) -> cos(x); + +define double @test_simplify1(double %d) { +; NO-FLOAT-SHRINK: @test_simplify1 + %neg = fsub double -0.000000e+00, %d + %cos = call double @cos(double %neg) +; NO-FLOAT-SHRINK: call double @cos(double %d) + ret double %cos +} + +define float @test_simplify2(float %f) { +; DO-FLOAT-SHRINK: @test_simplify2 + %conv1 = fpext float %f to double + %neg = fsub double -0.000000e+00, %conv1 + %cos = call double @cos(double %neg) + %conv2 = fptrunc double %cos to float +; DO-FLOAT-SHRINK: call float @cosf(float %f) + ret float %conv2 +} + +define float @test_simplify3(float %f) { +; NO-FLOAT-SHRINK: @test_simplify3 + %conv1 = fpext float %f to double + %neg = fsub double -0.000000e+00, %conv1 + %cos = call double @cos(double %neg) +; NO-FLOAT-SHRINK: call double @cos(double %conv1) + %conv2 = fptrunc double %cos to float + ret float %conv2 +} diff --git a/test/Transforms/InstCombine/cos-2.ll b/test/Transforms/InstCombine/cos-2.ll new file mode 100644 index 0000000000000..2f2dfafe484d5 --- /dev/null +++ b/test/Transforms/InstCombine/cos-2.ll @@ -0,0 +1,17 @@ +; Test that the cos library call simplifier works correctly. +; +; RUN: opt < %s -instcombine -S | FileCheck %s + +target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128" + +declare float @cos(double) + +; Check that cos functions with the wrong prototype aren't simplified. + +define float @test_no_simplify1(double %d) { +; CHECK: @test_no_simplify1 + %neg = fsub double -0.000000e+00, %d + %cos = call float @cos(double %neg) +; CHECK: call float @cos(double %neg) + ret float %cos +} diff --git a/test/Transforms/InstCombine/debug-line.ll b/test/Transforms/InstCombine/debug-line.ll new file mode 100644 index 0000000000000..084efdc989f93 --- /dev/null +++ b/test/Transforms/InstCombine/debug-line.ll @@ -0,0 +1,24 @@ +; RUN: opt -instcombine -S < %s | FileCheck %s + + +@.str = private constant [3 x i8] c"%c\00" + +define void @foo() nounwind ssp { +;CHECK: call i32 @putchar{{.+}} !dbg + %1 = call i32 (i8*, ...)* @printf(i8* getelementptr inbounds ([3 x i8]* @.str, i32 0, i32 0), i32 97), !dbg !5 + ret void, !dbg !7 +} + +declare i32 @printf(i8*, ...) + +!llvm.dbg.sp = !{!0} + +!0 = metadata !{i32 589870, i32 0, metadata !1, metadata !"foo", metadata !"foo", metadata !"", metadata !1, i32 4, metadata !3, i1 false, i1 true, i32 0, i32 0, null, i32 0, i1 false, void ()* @foo} ; [ DW_TAG_subprogram ] +!1 = metadata !{i32 589865, metadata !"m.c", metadata !"/private/tmp", metadata !2} ; [ DW_TAG_file_type ] +!2 = metadata !{i32 589841, i32 0, i32 12, metadata !"m.c", metadata !"/private/tmp", metadata !"clang", i1 true, i1 false, metadata !"", i32 0} ; [ DW_TAG_compile_unit ] +!3 = metadata !{i32 589845, metadata !1, metadata !"", metadata !1, i32 0, i64 0, i64 0, i64 0, i32 0, null, metadata !4, i32 0, null} ; [ DW_TAG_subroutine_type ] +!4 = metadata !{null} +!5 = metadata !{i32 5, i32 2, metadata !6, null} +!6 = metadata !{i32 589835, metadata !0, i32 4, i32 12, metadata !1, i32 0} ; [ DW_TAG_lexical_block ] +!7 = metadata !{i32 6, i32 1, metadata !6, null} + diff --git a/test/Transforms/InstCombine/debuginfo.ll b/test/Transforms/InstCombine/debuginfo.ll index f6892fc3e1f9c..cdbcd865117cf 100644 --- a/test/Transforms/InstCombine/debuginfo.ll +++ b/test/Transforms/InstCombine/debuginfo.ll @@ -28,22 +28,21 @@ entry: ret i8* %call, !dbg !21 } -!llvm.dbg.lv.foobar = !{!0, !7, !9} -!llvm.dbg.sp = !{!1} +!llvm.dbg.cu = !{!3} -!0 = metadata !{i32 590081, metadata !1, metadata !"__dest", metadata !2, i32 16777294, metadata !6, i32 0} ; [ DW_TAG_arg_variable ] -!1 = metadata !{i32 589870, i32 0, metadata !2, metadata !"foobar", metadata !"foobar", metadata !"", metadata !2, i32 79, metadata !4, i1 true, i1 true, i32 0, i32 0, i32 0, i32 256, i1 true, i8* (i8*, i32, i64)* @foobar} ; [ DW_TAG_subprogram ] -!2 = metadata !{i32 589865, metadata !"string.h", metadata !"Game", metadata !3} ; [ DW_TAG_file_type ] -!3 = metadata !{i32 589841, i32 0, i32 12, metadata !"bits.c", metadata !"Game", metadata !"clang version 3.0 (trunk 127710)", i1 true, i1 true, metadata !"", i32 0} ; [ DW_TAG_compile_unit ] -!4 = metadata !{i32 589845, metadata !2, metadata !"", metadata !2, i32 0, i64 0, i64 0, i32 0, i32 0, i32 0, metadata !5, i32 0, i32 0} ; [ DW_TAG_subroutine_type ] +!0 = metadata !{i32 786689, metadata !1, metadata !"__dest", metadata !2, i32 16777294, metadata !6, i32 0, null} ; [ DW_TAG_arg_variable ] +!1 = metadata !{i32 786478, metadata !2, metadata !"foobar", metadata !"foobar", metadata !"", metadata !2, i32 79, metadata !4, i1 true, i1 true, i32 0, i32 0, i32 0, i32 256, i1 true, i8* (i8*, i32, i64)* @foobar, null, null, metadata !25, i32 79} ; [ DW_TAG_subprogram ] +!2 = metadata !{i32 786473, metadata !27} ; [ DW_TAG_file_type ] +!3 = metadata !{i32 786449, i32 0, i32 12, metadata !26, metadata !"clang version 3.0 (trunk 127710)", i1 true, metadata !"", i32 0, null, null, metadata !24, null, null} ; [ DW_TAG_compile_unit ] +!4 = metadata !{i32 786453, metadata !2, metadata !"", metadata !2, i32 0, i64 0, i64 0, i32 0, i32 0, i32 0, metadata !5, i32 0, i32 0} ; [ DW_TAG_subroutine_type ] !5 = metadata !{metadata !6} -!6 = metadata !{i32 589839, metadata !3, metadata !"", null, i32 0, i64 64, i64 64, i64 0, i32 0, null} ; [ DW_TAG_pointer_type ] -!7 = metadata !{i32 590081, metadata !1, metadata !"__val", metadata !2, i32 33554510, metadata !8, i32 0} ; [ DW_TAG_arg_variable ] -!8 = metadata !{i32 589860, metadata !3, metadata !"int", null, i32 0, i64 32, i64 32, i64 0, i32 0, i32 5} ; [ DW_TAG_base_type ] -!9 = metadata !{i32 590081, metadata !1, metadata !"__len", metadata !2, i32 50331726, metadata !10, i32 0} ; [ DW_TAG_arg_variable ] +!6 = metadata !{i32 786447, metadata !3, metadata !"", null, i32 0, i64 64, i64 64, i64 0, i32 0, null} ; [ DW_TAG_pointer_type ] +!7 = metadata !{i32 786689, metadata !1, metadata !"__val", metadata !2, i32 33554510, metadata !8, i32 0, null} ; [ DW_TAG_arg_variable ] +!8 = metadata !{i32 786468, metadata !3, metadata !"int", null, i32 0, i64 32, i64 32, i64 0, i32 0, i32 5} ; [ DW_TAG_base_type ] +!9 = metadata !{i32 786689, metadata !1, metadata !"__len", metadata !2, i32 50331726, metadata !10, i32 0, null} ; [ DW_TAG_arg_variable ] !10 = metadata !{i32 589846, metadata !3, metadata !"size_t", metadata !2, i32 80, i64 0, i64 0, i64 0, i32 0, metadata !11} ; [ DW_TAG_typedef ] !11 = metadata !{i32 589846, metadata !3, metadata !"__darwin_size_t", metadata !2, i32 90, i64 0, i64 0, i64 0, i32 0, metadata !12} ; [ DW_TAG_typedef ] -!12 = metadata !{i32 589860, metadata !3, metadata !"long unsigned int", null, i32 0, i64 64, i64 64, i64 0, i32 0, i32 7} ; [ DW_TAG_base_type ] +!12 = metadata !{i32 786468, metadata !3, metadata !"long unsigned int", null, i32 0, i64 64, i64 64, i64 0, i32 0, i32 7} ; [ DW_TAG_base_type ] !13 = metadata !{metadata !"any pointer", metadata !14} !14 = metadata !{metadata !"omnipotent char", metadata !15} !15 = metadata !{metadata !"Simple C/C++ TBAA", null} @@ -53,5 +52,10 @@ entry: !19 = metadata !{metadata !"long", metadata !14} !20 = metadata !{i32 78, i32 54, metadata !1, null} !21 = metadata !{i32 80, i32 3, metadata !22, null} -!22 = metadata !{i32 589835, metadata !23, i32 80, i32 3, metadata !2, i32 7} ; [ DW_TAG_lexical_block ] -!23 = metadata !{i32 589835, metadata !1, i32 79, i32 1, metadata !2, i32 6} ; [ DW_TAG_lexical_block ] +!22 = metadata !{i32 786443, metadata !23, i32 80, i32 3, metadata !2, i32 7} ; [ DW_TAG_lexical_block ] +!23 = metadata !{i32 786443, metadata !1, i32 79, i32 1, metadata !2, i32 6} ; [ DW_TAG_lexical_block ] +!24 = metadata !{metadata !1} +!25 = metadata !{metadata !0, metadata !7, metadata !9} +!26 = metadata !{i32 786473, metadata !28} ; [ DW_TAG_file_type ] +!27 = metadata !{metadata !"string.h", metadata !"Game"} +!28 = metadata !{metadata !"bits.c", metadata !"Game"} diff --git a/test/Transforms/InstCombine/devirt.ll b/test/Transforms/InstCombine/devirt.ll index 6189dc2af4f94..9c7cf5d697e84 100644 --- a/test/Transforms/InstCombine/devirt.ll +++ b/test/Transforms/InstCombine/devirt.ll @@ -1,4 +1,4 @@ -; RUN: opt -instcombine -S -o - %s | FileCheck %s +; RUN: opt -instcombine -S < %s | FileCheck %s ; CHECK-NOT: getelementptr ; CHECK-NOT: ptrtoint diff --git a/test/Transforms/InstCombine/disable-simplify-libcalls.ll b/test/Transforms/InstCombine/disable-simplify-libcalls.ll index d81e9ae5bd732..c2c29368b1a8a 100644 --- a/test/Transforms/InstCombine/disable-simplify-libcalls.ll +++ b/test/Transforms/InstCombine/disable-simplify-libcalls.ll @@ -37,6 +37,18 @@ declare i64 @strtoll(i8*, i8**, i32) declare i64 @strtoul(i8*, i8**, i32) declare i64 @strtoull(i8*, i8**, i32) declare i64 @strcspn(i8*, i8*) +declare i32 @abs(i32) +declare i32 @ffs(i32) +declare i32 @ffsl(i64) +declare i32 @ffsll(i64) +declare i32 @fprintf(i8*, i8*) +declare i32 @isascii(i32) +declare i32 @isdigit(i32) +declare i32 @toascii(i32) +declare i64 @labs(i64) +declare i64 @llabs(i64) +declare i32 @printf(i8*) +declare i32 @sprintf(i8*, i8*) define double @t1(double %x) { ; CHECK: @t1 @@ -234,3 +246,90 @@ define i64 @t25(i8* %y) { ret i64 %ret ; CHECK: call i64 @strcspn } + +define i32 @t26(i32 %y) { +; CHECK: @t26 + %ret = call i32 @abs(i32 %y) + ret i32 %ret +; CHECK: call i32 @abs +} + +define i32 @t27(i32 %y) { +; CHECK: @t27 + %ret = call i32 @ffs(i32 %y) + ret i32 %ret +; CHECK: call i32 @ffs +} + +define i32 @t28(i64 %y) { +; CHECK: @t28 + %ret = call i32 @ffsl(i64 %y) + ret i32 %ret +; CHECK: call i32 @ffsl +} + +define i32 @t29(i64 %y) { +; CHECK: @t29 + %ret = call i32 @ffsll(i64 %y) + ret i32 %ret +; CHECK: call i32 @ffsll +} + +define void @t30() { +; CHECK: @t30 + %x = getelementptr inbounds [13 x i8]* @.str1, i32 0, i32 0 + call i32 @fprintf(i8* null, i8* %x) + ret void +; CHECK: call i32 @fprintf +} + +define i32 @t31(i32 %y) { +; CHECK: @t31 + %ret = call i32 @isascii(i32 %y) + ret i32 %ret +; CHECK: call i32 @isascii +} + +define i32 @t32(i32 %y) { +; CHECK: @t32 + %ret = call i32 @isdigit(i32 %y) + ret i32 %ret +; CHECK: call i32 @isdigit +} + +define i32 @t33(i32 %y) { +; CHECK: @t33 + %ret = call i32 @toascii(i32 %y) + ret i32 %ret +; CHECK: call i32 @toascii +} + +define i64 @t34(i64 %y) { +; CHECK: @t34 + %ret = call i64 @labs(i64 %y) + ret i64 %ret +; CHECK: call i64 @labs +} + +define i64 @t35(i64 %y) { +; CHECK: @t35 + %ret = call i64 @llabs(i64 %y) + ret i64 %ret +; CHECK: call i64 @llabs +} + +define void @t36() { +; CHECK: @t36 + %x = getelementptr inbounds [1 x i8]* @empty, i32 0, i32 0 + call i32 @printf(i8* %x) + ret void +; CHECK: call i32 @printf +} + +define void @t37(i8* %x) { +; CHECK: @t37 + %y = getelementptr inbounds [13 x i8]* @.str1, i32 0, i32 0 + call i32 @sprintf(i8* %x, i8* %y) + ret void +; CHECK: call i32 @sprintf +} diff --git a/test/Transforms/InstCombine/double-float-shrink-1.ll b/test/Transforms/InstCombine/double-float-shrink-1.ll new file mode 100644 index 0000000000000..e5448ee007654 --- /dev/null +++ b/test/Transforms/InstCombine/double-float-shrink-1.ll @@ -0,0 +1,333 @@ +; RUN: opt < %s -instcombine -enable-double-float-shrink -S | FileCheck %s + +target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128" +target triple = "x86_64-unknown-linux-gnu" + +define float @acos_test(float %f) nounwind readnone { +; CHECK: acos_test + %conv = fpext float %f to double + %call = call double @acos(double %conv) + %conv1 = fptrunc double %call to float + ret float %conv1 +; CHECK: call float @acosf(float %f) +} + +define double @acos_test2(float %f) nounwind readnone { +; CHECK: acos_test2 + %conv = fpext float %f to double + %call = call double @acos(double %conv) + ret double %call +; CHECK: call double @acos(double %conv) +} + +define float @acosh_test(float %f) nounwind readnone { +; CHECK: acosh_test + %conv = fpext float %f to double + %call = call double @acosh(double %conv) + %conv1 = fptrunc double %call to float + ret float %conv1 +; CHECK: call float @acoshf(float %f) +} + +define double @acosh_test2(float %f) nounwind readnone { +; CHECK: acosh_test2 + %conv = fpext float %f to double + %call = call double @acosh(double %conv) + ret double %call +; CHECK: call double @acosh(double %conv) +} + +define float @asin_test(float %f) nounwind readnone { +; CHECK: asin_test + %conv = fpext float %f to double + %call = call double @asin(double %conv) + %conv1 = fptrunc double %call to float + ret float %conv1 +; CHECK: call float @asinf(float %f) +} + +define double @asin_test2(float %f) nounwind readnone { +; CHECK: asin_test2 + %conv = fpext float %f to double + %call = call double @asin(double %conv) + ret double %call +; CHECK: call double @asin(double %conv) +} + +define float @asinh_test(float %f) nounwind readnone { +; CHECK: asinh_test + %conv = fpext float %f to double + %call = call double @asinh(double %conv) + %conv1 = fptrunc double %call to float + ret float %conv1 +; CHECK: call float @asinhf(float %f) +} + +define double @asinh_test2(float %f) nounwind readnone { +; CHECK: asinh_test2 + %conv = fpext float %f to double + %call = call double @asinh(double %conv) + ret double %call +; CHECK: call double @asinh(double %conv) +} + +define float @atan_test(float %f) nounwind readnone { +; CHECK: atan_test + %conv = fpext float %f to double + %call = call double @atan(double %conv) + %conv1 = fptrunc double %call to float + ret float %conv1 +; CHECK: call float @atanf(float %f) +} + +define double @atan_test2(float %f) nounwind readnone { +; CHECK: atan_test2 + %conv = fpext float %f to double + %call = call double @atan(double %conv) + ret double %call +; CHECK: call double @atan(double %conv) +} +define float @atanh_test(float %f) nounwind readnone { +; CHECK: atanh_test + %conv = fpext float %f to double + %call = call double @atanh(double %conv) + %conv1 = fptrunc double %call to float + ret float %conv1 +; CHECK: call float @atanhf(float %f) +} + +define double @atanh_test2(float %f) nounwind readnone { +; CHECK: atanh_test2 + %conv = fpext float %f to double + %call = call double @atanh(double %conv) + ret double %call +; CHECK: call double @atanh(double %conv) +} +define float @cbrt_test(float %f) nounwind readnone { +; CHECK: cbrt_test + %conv = fpext float %f to double + %call = call double @cbrt(double %conv) + %conv1 = fptrunc double %call to float + ret float %conv1 +; CHECK: call float @cbrtf(float %f) +} + +define double @cbrt_test2(float %f) nounwind readnone { +; CHECK: cbrt_test2 + %conv = fpext float %f to double + %call = call double @cbrt(double %conv) + ret double %call +; CHECK: call double @cbrt(double %conv) +} +define float @exp_test(float %f) nounwind readnone { +; CHECK: exp_test + %conv = fpext float %f to double + %call = call double @exp(double %conv) + %conv1 = fptrunc double %call to float + ret float %conv1 +; CHECK: call float @expf(float %f) +} + +define double @exp_test2(float %f) nounwind readnone { +; CHECK: exp_test2 + %conv = fpext float %f to double + %call = call double @exp(double %conv) + ret double %call +; CHECK: call double @exp(double %conv) +} +define float @expm1_test(float %f) nounwind readnone { +; CHECK: expm1_test + %conv = fpext float %f to double + %call = call double @expm1(double %conv) + %conv1 = fptrunc double %call to float + ret float %conv1 +; CHECK: call float @expm1f(float %f) +} + +define double @expm1_test2(float %f) nounwind readnone { +; CHECK: expm1_test2 + %conv = fpext float %f to double + %call = call double @expm1(double %conv) + ret double %call +; CHECK: call double @expm1(double %conv) +} +define float @exp10_test(float %f) nounwind readnone { +; CHECK: exp10_test + %conv = fpext float %f to double + %call = call double @exp10(double %conv) + %conv1 = fptrunc double %call to float + ret float %conv1 +; CHECK: call float @exp10f(float %f) +} + +define double @exp10_test2(float %f) nounwind readnone { +; CHECK: exp10_test2 + %conv = fpext float %f to double + %call = call double @exp10(double %conv) + ret double %call +; CHECK: call double @exp10(double %conv) +} +define float @log_test(float %f) nounwind readnone { +; CHECK: log_test + %conv = fpext float %f to double + %call = call double @log(double %conv) + %conv1 = fptrunc double %call to float + ret float %conv1 +; CHECK: call float @logf(float %f) +} + +define double @log_test2(float %f) nounwind readnone { +; CHECK: log_test2 + %conv = fpext float %f to double + %call = call double @log(double %conv) + ret double %call +; CHECK: call double @log(double %conv) +} +define float @log10_test(float %f) nounwind readnone { +; CHECK: log10_test + %conv = fpext float %f to double + %call = call double @log10(double %conv) + %conv1 = fptrunc double %call to float + ret float %conv1 +; CHECK: call float @log10f(float %f) +} + +define double @log10_test2(float %f) nounwind readnone { +; CHECK: log10_test2 + %conv = fpext float %f to double + %call = call double @log10(double %conv) + ret double %call +; CHECK: call double @log10(double %conv) +} +define float @log1p_test(float %f) nounwind readnone { +; CHECK: log1p_test + %conv = fpext float %f to double + %call = call double @log1p(double %conv) + %conv1 = fptrunc double %call to float + ret float %conv1 +; CHECK: call float @log1pf(float %f) +} + +define double @log1p_test2(float %f) nounwind readnone { +; CHECK: log1p_test2 + %conv = fpext float %f to double + %call = call double @log1p(double %conv) + ret double %call +; CHECK: call double @log1p(double %conv) +} +define float @log2_test(float %f) nounwind readnone { +; CHECK: log2_test + %conv = fpext float %f to double + %call = call double @log2(double %conv) + %conv1 = fptrunc double %call to float + ret float %conv1 +; CHECK: call float @log2f(float %f) +} + +define double @log2_test2(float %f) nounwind readnone { +; CHECK: log2_test2 + %conv = fpext float %f to double + %call = call double @log2(double %conv) + ret double %call +; CHECK: call double @log2(double %conv) +} +define float @logb_test(float %f) nounwind readnone { +; CHECK: logb_test + %conv = fpext float %f to double + %call = call double @logb(double %conv) + %conv1 = fptrunc double %call to float + ret float %conv1 +; CHECK: call float @logbf(float %f) +} + +define double @logb_test2(float %f) nounwind readnone { +; CHECK: logb_test2 + %conv = fpext float %f to double + %call = call double @logb(double %conv) + ret double %call +; CHECK: call double @logb(double %conv) +} +define float @sin_test(float %f) nounwind readnone { +; CHECK: sin_test + %conv = fpext float %f to double + %call = call double @sin(double %conv) + %conv1 = fptrunc double %call to float + ret float %conv1 +; CHECK: call float @sinf(float %f) +} + +define double @sin_test2(float %f) nounwind readnone { +; CHECK: sin_test2 + %conv = fpext float %f to double + %call = call double @sin(double %conv) + ret double %call +; CHECK: call double @sin(double %conv) +} +define float @sqrt_test(float %f) nounwind readnone { +; CHECK: sqrt_test + %conv = fpext float %f to double + %call = call double @sqrt(double %conv) + %conv1 = fptrunc double %call to float + ret float %conv1 +; CHECK: call float @sqrtf(float %f) +} + +define double @sqrt_test2(float %f) nounwind readnone { +; CHECK: sqrt_test2 + %conv = fpext float %f to double + %call = call double @sqrt(double %conv) + ret double %call +; CHECK: call double @sqrt(double %conv) +} +define float @tan_test(float %f) nounwind readnone { +; CHECK: tan_test + %conv = fpext float %f to double + %call = call double @tan(double %conv) + %conv1 = fptrunc double %call to float + ret float %conv1 +; CHECK: call float @tanf(float %f) +} + +define double @tan_test2(float %f) nounwind readnone { +; CHECK: tan_test2 + %conv = fpext float %f to double + %call = call double @tan(double %conv) + ret double %call +; CHECK: call double @tan(double %conv) +} +define float @tanh_test(float %f) nounwind readnone { +; CHECK: tanh_test + %conv = fpext float %f to double + %call = call double @tanh(double %conv) + %conv1 = fptrunc double %call to float + ret float %conv1 +; CHECK: call float @tanhf(float %f) +} + +define double @tanh_test2(float %f) nounwind readnone { +; CHECK: tanh_test2 + %conv = fpext float %f to double + %call = call double @tanh(double %conv) + ret double %call +; CHECK: call double @tanh(double %conv) +} + +declare double @tanh(double) nounwind readnone +declare double @tan(double) nounwind readnone +declare double @sqrt(double) nounwind readnone +declare double @sin(double) nounwind readnone +declare double @log2(double) nounwind readnone +declare double @log1p(double) nounwind readnone +declare double @log10(double) nounwind readnone +declare double @log(double) nounwind readnone +declare double @logb(double) nounwind readnone +declare double @exp10(double) nounwind readnone +declare double @expm1(double) nounwind readnone +declare double @exp(double) nounwind readnone +declare double @cbrt(double) nounwind readnone +declare double @atanh(double) nounwind readnone +declare double @atan(double) nounwind readnone +declare double @acos(double) nounwind readnone +declare double @acosh(double) nounwind readnone +declare double @asin(double) nounwind readnone +declare double @asinh(double) nounwind readnone diff --git a/test/Transforms/InstCombine/double-float-shrink-2.ll b/test/Transforms/InstCombine/double-float-shrink-2.ll new file mode 100644 index 0000000000000..7f6df92c96c5a --- /dev/null +++ b/test/Transforms/InstCombine/double-float-shrink-2.ll @@ -0,0 +1,80 @@ +; RUN: opt < %s -instcombine -S -mtriple "i386-pc-linux" | FileCheck -check-prefix=DO-SIMPLIFY %s +; RUN: opt < %s -instcombine -S -mtriple "i386-pc-win32" | FileCheck -check-prefix=DONT-SIMPLIFY %s +; RUN: opt < %s -instcombine -S -mtriple "x86_64-pc-win32" | FileCheck -check-prefix=C89-SIMPLIFY %s +; RUN: opt < %s -instcombine -S -mtriple "i386-pc-mingw32" | FileCheck -check-prefix=DO-SIMPLIFY %s +; RUN: opt < %s -instcombine -S -mtriple "x86_64-pc-mingw32" | FileCheck -check-prefix=DO-SIMPLIFY %s +; RUN: opt < %s -instcombine -S -mtriple "sparc-sun-solaris" | FileCheck -check-prefix=DO-SIMPLIFY %s + +; DO-SIMPLIFY: call float @floorf( +; DO-SIMPLIFY: call float @ceilf( +; DO-SIMPLIFY: call float @roundf( +; DO-SIMPLIFY: call float @nearbyintf( +; DO-SIMPLIFY: call float @truncf( +; DO-SIMPLIFY: call float @fabsf( + +; C89-SIMPLIFY: call float @floorf( +; C89-SIMPLIFY: call float @ceilf( +; C89-SIMPLIFY: call double @round( +; C89-SIMPLIFY: call double @nearbyint( + +; DONT-SIMPLIFY: call double @floor( +; DONT-SIMPLIFY: call double @ceil( +; DONT-SIMPLIFY: call double @round( +; DONT-SIMPLIFY: call double @nearbyint( +; DONT-SIMPLIFY: call double @trunc( +; DONT-SIMPLIFY: call double @fabs( + +declare double @floor(double) +declare double @ceil(double) +declare double @round(double) +declare double @nearbyint(double) +declare double @trunc(double) +declare double @fabs(double) + +define float @test_floor(float %C) { + %D = fpext float %C to double + ; --> floorf + %E = call double @floor(double %D) + %F = fptrunc double %E to float + ret float %F +} + +define float @test_ceil(float %C) { + %D = fpext float %C to double + ; --> ceilf + %E = call double @ceil(double %D) + %F = fptrunc double %E to float + ret float %F +} + +define float @test_round(float %C) { + %D = fpext float %C to double + ; --> roundf + %E = call double @round(double %D) + %F = fptrunc double %E to float + ret float %F +} + +define float @test_nearbyint(float %C) { + %D = fpext float %C to double + ; --> nearbyintf + %E = call double @nearbyint(double %D) + %F = fptrunc double %E to float + ret float %F +} + +define float @test_trunc(float %C) { + %D = fpext float %C to double + ; --> truncf + %E = call double @trunc(double %D) + %F = fptrunc double %E to float + ret float %F +} + +define float @test_fabs(float %C) { + %D = fpext float %C to double + ; --> fabsf + %E = call double @fabs(double %D) + %F = fptrunc double %E to float + ret float %F +} diff --git a/test/Transforms/InstCombine/exact.ll b/test/Transforms/InstCombine/exact.ll index 14741e3c1c336..88ca88c3b9275 100644 --- a/test/Transforms/InstCombine/exact.ll +++ b/test/Transforms/InstCombine/exact.ll @@ -99,9 +99,9 @@ define i1 @ashr_icmp2(i64 %X) nounwind { ; PR9998 ; Make sure we don't transform the ashr here into an sdiv ; CHECK: @pr9998 -; CHECK: = and i32 %V, 1 -; CHECK: %Z = icmp ne -; CHECK: ret i1 %Z +; CHECK: [[BIT:%[A-Za-z0-9.]+]] = and i32 %V, 1 +; CHECK-NEXT: [[CMP:%[A-Za-z0-9.]+]] = icmp ne i32 [[BIT]], 0 +; CHECK-NEXT: ret i1 [[CMP]] define i1 @pr9998(i32 %V) nounwind { entry: %W = shl i32 %V, 31 @@ -112,6 +112,7 @@ entry: } + ; CHECK: @udiv_icmp1 ; CHECK: icmp ne i64 %X, 0 define i1 @udiv_icmp1(i64 %X) nounwind { diff --git a/test/Transforms/InstCombine/exp2-1.ll b/test/Transforms/InstCombine/exp2-1.ll new file mode 100644 index 0000000000000..1b0ad5000412d --- /dev/null +++ b/test/Transforms/InstCombine/exp2-1.ll @@ -0,0 +1,76 @@ +; Test that the exp2 library call simplifier works correctly. +; +; RUN: opt < %s -instcombine -S | FileCheck %s + +target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128" + +declare double @exp2(double) +declare float @exp2f(float) + +; Check exp2(sitofp(x)) -> ldexp(1.0, sext(x)). + +define double @test_simplify1(i32 %x) { +; CHECK: @test_simplify1 + %conv = sitofp i32 %x to double + %ret = call double @exp2(double %conv) +; CHECK: call double @ldexp + ret double %ret +} + +define double @test_simplify2(i16 signext %x) { +; CHECK: @test_simplify2 + %conv = sitofp i16 %x to double + %ret = call double @exp2(double %conv) +; CHECK: call double @ldexp + ret double %ret +} + +define double @test_simplify3(i8 signext %x) { +; CHECK: @test_simplify3 + %conv = sitofp i8 %x to double + %ret = call double @exp2(double %conv) +; CHECK: call double @ldexp + ret double %ret +} + +define float @test_simplify4(i32 %x) { +; CHECK: @test_simplify4 + %conv = sitofp i32 %x to float + %ret = call float @exp2f(float %conv) +; CHECK: call float @ldexpf + ret float %ret +} + +; Check exp2(uitofp(x)) -> ldexp(1.0, zext(x)). + +define double @test_no_simplify1(i32 %x) { +; CHECK: @test_no_simplify1 + %conv = uitofp i32 %x to double + %ret = call double @exp2(double %conv) +; CHECK: call double @exp2 + ret double %ret +} + +define double @test_simplify6(i16 zeroext %x) { +; CHECK: @test_simplify6 + %conv = uitofp i16 %x to double + %ret = call double @exp2(double %conv) +; CHECK: call double @ldexp + ret double %ret +} + +define double @test_simplify7(i8 zeroext %x) { +; CHECK: @test_simplify7 + %conv = uitofp i8 %x to double + %ret = call double @exp2(double %conv) +; CHECK: call double @ldexp + ret double %ret +} + +define float @test_simplify8(i8 zeroext %x) { +; CHECK: @test_simplify8 + %conv = uitofp i8 %x to float + %ret = call float @exp2f(float %conv) +; CHECK: call float @ldexpf + ret float %ret +} diff --git a/test/Transforms/InstCombine/exp2-2.ll b/test/Transforms/InstCombine/exp2-2.ll new file mode 100644 index 0000000000000..bed063798e293 --- /dev/null +++ b/test/Transforms/InstCombine/exp2-2.ll @@ -0,0 +1,17 @@ +; Test that the exp2 library call simplifier works correctly. +; +; RUN: opt < %s -instcombine -S | FileCheck %s + +target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128" + +declare float @exp2(double) + +; Check that exp2 functions with the wrong prototype aren't simplified. + +define float @test_no_simplify1(i32 %x) { +; CHECK: @test_no_simplify1 + %conv = sitofp i32 %x to double + %ret = call float @exp2(double %conv) +; CHECK: call float @exp2(double %conv) + ret float %ret +} diff --git a/test/Transforms/InstCombine/fast-math.ll b/test/Transforms/InstCombine/fast-math.ll new file mode 100644 index 0000000000000..edcbcc71dfb48 --- /dev/null +++ b/test/Transforms/InstCombine/fast-math.ll @@ -0,0 +1,467 @@ +; RUN: opt < %s -instcombine -S | FileCheck %s + +; testing-case "float fold(float a) { return 1.2f * a * 2.3f; }" +; 1.2f and 2.3f is supposed to be fold. +define float @fold(float %a) { + %mul = fmul fast float %a, 0x3FF3333340000000 + %mul1 = fmul fast float %mul, 0x4002666660000000 + ret float %mul1 +; CHECK: @fold +; CHECK: fmul fast float %a, 0x4006147AE0000000 +} + +; Same testing-case as the one used in fold() except that the operators have +; fixed FP mode. +define float @notfold(float %a) { +; CHECK: @notfold +; CHECK: %mul = fmul fast float %a, 0x3FF3333340000000 + %mul = fmul fast float %a, 0x3FF3333340000000 + %mul1 = fmul float %mul, 0x4002666660000000 + ret float %mul1 +} + +define float @fold2(float %a) { +; CHECK: @fold2 +; CHECK: fmul fast float %a, 0x4006147AE0000000 + %mul = fmul float %a, 0x3FF3333340000000 + %mul1 = fmul fast float %mul, 0x4002666660000000 + ret float %mul1 +} + +; C * f1 + f1 = (C+1) * f1 +define double @fold3(double %f1) { + %t1 = fmul fast double 2.000000e+00, %f1 + %t2 = fadd fast double %f1, %t1 + ret double %t2 +; CHECK: @fold3 +; CHECK: fmul fast double %f1, 3.000000e+00 +} + +; (C1 - X) + (C2 - Y) => (C1+C2) - (X + Y) +define float @fold4(float %f1, float %f2) { + %sub = fsub float 4.000000e+00, %f1 + %sub1 = fsub float 5.000000e+00, %f2 + %add = fadd fast float %sub, %sub1 + ret float %add +; CHECK: @fold4 +; CHECK: %1 = fadd fast float %f1, %f2 +; CHECK: fsub fast float 9.000000e+00, %1 +} + +; (X + C1) + C2 => X + (C1 + C2) +define float @fold5(float %f1, float %f2) { + %add = fadd float %f1, 4.000000e+00 + %add1 = fadd fast float %add, 5.000000e+00 + ret float %add1 +; CHECK: @fold5 +; CHECK: fadd fast float %f1, 9.000000e+00 +} + +; (X + X) + X => 3.0 * X +define float @fold6(float %f1) { + %t1 = fadd fast float %f1, %f1 + %t2 = fadd fast float %f1, %t1 + ret float %t2 +; CHECK: @fold6 +; CHECK: fmul fast float %f1, 3.000000e+00 +} + +; C1 * X + (X + X) = (C1 + 2) * X +define float @fold7(float %f1) { + %t1 = fmul fast float %f1, 5.000000e+00 + %t2 = fadd fast float %f1, %f1 + %t3 = fadd fast float %t1, %t2 + ret float %t3 +; CHECK: @fold7 +; CHECK: fmul fast float %f1, 7.000000e+00 +} + +; (X + X) + (X + X) => 4.0 * X +define float @fold8(float %f1) { + %t1 = fadd fast float %f1, %f1 + %t2 = fadd fast float %f1, %f1 + %t3 = fadd fast float %t1, %t2 + ret float %t3 +; CHECK: fold8 +; CHECK: fmul fast float %f1, 4.000000e+00 +} + +; X - (X + Y) => 0 - Y +define float @fold9(float %f1, float %f2) { + %t1 = fadd float %f1, %f2 + %t3 = fsub fast float %f1, %t1 + ret float %t3 + +; CHECK: @fold9 +; CHECK: fsub fast float 0.000000e+00, %f2 +} + +; Let C3 = C1 + C2. (f1 + C1) + (f2 + C2) => (f1 + f2) + C3 instead of +; "(f1 + C3) + f2" or "(f2 + C3) + f1". Placing constant-addend at the +; top of resulting simplified expression tree may potentially reveal some +; optimization opportunities in the super-expression trees. +; +define float @fold10(float %f1, float %f2) { + %t1 = fadd fast float 2.000000e+00, %f1 + %t2 = fsub fast float %f2, 3.000000e+00 + %t3 = fadd fast float %t1, %t2 + ret float %t3 +; CHECK: @fold10 +; CHECK: %t3 = fadd fast float %t2, -1.000000e+00 +; CHECK: ret float %t3 +} + +; once cause Crash/miscompilation +define float @fail1(float %f1, float %f2) { + %conv3 = fadd fast float %f1, -1.000000e+00 + %add = fadd fast float %conv3, %conv3 + %add2 = fadd fast float %add, %conv3 + ret float %add2 +; CHECK: @fail1 +; CHECK: ret +} + +define double @fail2(double %f1, double %f2) { + %t1 = fsub fast double %f1, %f2 + %t2 = fadd fast double %f1, %f2 + %t3 = fsub fast double %t1, %t2 + ret double %t3 +; CHECK: @fail2 +; CHECK: ret +} + +; c1 * x - x => (c1 - 1.0) * x +define float @fold13(float %x) { + %mul = fmul fast float %x, 7.000000e+00 + %sub = fsub fast float %mul, %x + ret float %sub +; CHECK: fold13 +; CHECK: fmul fast float %x, 6.000000e+00 +; CHECK: ret +} + +; ========================================================================= +; +; Testing-cases about fmul begin +; +; ========================================================================= + +; ((X*C1) + C2) * C3 => (X * (C1*C3)) + (C2*C3) (i.e. distribution) +define float @fmul_distribute1(float %f1) { + %t1 = fmul float %f1, 6.0e+3 + %t2 = fadd float %t1, 2.0e+3 + %t3 = fmul fast float %t2, 5.0e+3 + ret float %t3 +; CHECK: @fmul_distribute1 +; CHECK: %1 = fmul fast float %f1, 3.000000e+07 +; CHECK: %t3 = fadd fast float %1, 1.000000e+07 +} + +; (X/C1 + C2) * C3 => X/(C1/C3) + C2*C3 +define double @fmul_distribute2(double %f1, double %f2) { + %t1 = fdiv double %f1, 3.0e+0 + %t2 = fadd double %t1, 5.0e+1 + ; 0x10000000000000 = DBL_MIN + %t3 = fmul fast double %t2, 0x10000000000000 + ret double %t3 + +; CHECK: @fmul_distribute2 +; CHECK: %1 = fdiv fast double %f1, 0x7FE8000000000000 +; CHECK: fadd fast double %1, 0x69000000000000 +} + +; 5.0e-1 * DBL_MIN yields denormal, so "(f1*3.0 + 5.0e-1) * DBL_MIN" cannot +; be simplified into f1 * (3.0*DBL_MIN) + (5.0e-1*DBL_MIN) +define double @fmul_distribute3(double %f1) { + %t1 = fdiv double %f1, 3.0e+0 + %t2 = fadd double %t1, 5.0e-1 + %t3 = fmul fast double %t2, 0x10000000000000 + ret double %t3 + +; CHECK: @fmul_distribute3 +; CHECK: fmul fast double %t2, 0x10000000000000 +} + +; ((X*C1) + C2) * C3 => (X * (C1*C3)) + (C2*C3) (i.e. distribution) +define float @fmul_distribute4(float %f1) { + %t1 = fmul float %f1, 6.0e+3 + %t2 = fsub float 2.0e+3, %t1 + %t3 = fmul fast float %t2, 5.0e+3 + ret float %t3 +; CHECK: @fmul_distribute4 +; CHECK: %1 = fmul fast float %f1, 3.000000e+07 +; CHECK: %t3 = fsub fast float 1.000000e+07, %1 +} + +; C1/X * C2 => (C1*C2) / X +define float @fmul2(float %f1) { + %t1 = fdiv float 2.0e+3, %f1 + %t3 = fmul fast float %t1, 6.0e+3 + ret float %t3 +; CHECK: @fmul2 +; CHECK: fdiv fast float 1.200000e+07, %f1 +} + +; X/C1 * C2 => X * (C2/C1) (if C2/C1 is normal Fp) +define float @fmul3(float %f1, float %f2) { + %t1 = fdiv float %f1, 2.0e+3 + %t3 = fmul fast float %t1, 6.0e+3 + ret float %t3 +; CHECK: @fmul3 +; CHECK: fmul fast float %f1, 3.000000e+00 +} + +; Rule "X/C1 * C2 => X * (C2/C1) is not applicable if C2/C1 is either a special +; value of a denormal. The 0x3810000000000000 here take value FLT_MIN +; +define float @fmul4(float %f1, float %f2) { + %t1 = fdiv float %f1, 2.0e+3 + %t3 = fmul fast float %t1, 0x3810000000000000 + ret float %t3 +; CHECK: @fmul4 +; CHECK: fmul fast float %t1, 0x3810000000000000 +} + +; X / C1 * C2 => X / (C2/C1) if C1/C2 is either a special value of a denormal, +; and C2/C1 is a normal value. +; +define float @fmul5(float %f1, float %f2) { + %t1 = fdiv float %f1, 3.0e+0 + %t3 = fmul fast float %t1, 0x3810000000000000 + ret float %t3 +; CHECK: @fmul5 +; CHECK: fdiv fast float %f1, 0x47E8000000000000 +} + +; (X*Y) * X => (X*X) * Y +define float @fmul6(float %f1, float %f2) { + %mul = fmul float %f1, %f2 + %mul1 = fmul fast float %mul, %f1 + ret float %mul1 +; CHECK: @fmul6 +; CHECK: fmul fast float %f1, %f1 +} + +; "(X*Y) * X => (X*X) * Y" is disabled if "X*Y" has multiple uses +define float @fmul7(float %f1, float %f2) { + %mul = fmul float %f1, %f2 + %mul1 = fmul fast float %mul, %f1 + %add = fadd float %mul1, %mul + ret float %add +; CHECK: @fmul7 +; CHECK: fmul fast float %mul, %f1 +} + +; ========================================================================= +; +; Testing-cases about negation +; +; ========================================================================= +define float @fneg1(float %f1, float %f2) { + %sub = fsub float -0.000000e+00, %f1 + %sub1 = fsub nsz float 0.000000e+00, %f2 + %mul = fmul float %sub, %sub1 + ret float %mul +; CHECK: @fneg1 +; CHECK: fmul float %f1, %f2 +} + +; ========================================================================= +; +; Testing-cases about div +; +; ========================================================================= + +; X/C1 / C2 => X * (1/(C2*C1)) +define float @fdiv1(float %x) { + %div = fdiv float %x, 0x3FF3333340000000 + %div1 = fdiv fast float %div, 0x4002666660000000 + ret float %div1 +; 0x3FF3333340000000 = 1.2f +; 0x4002666660000000 = 2.3f +; 0x3FD7303B60000000 = 0.36231884057971014492 +; CHECK: @fdiv1 +; CHECK: fmul fast float %x, 0x3FD7303B60000000 +} + +; X*C1 / C2 => X * (C1/C2) +define float @fdiv2(float %x) { + %mul = fmul float %x, 0x3FF3333340000000 + %div1 = fdiv fast float %mul, 0x4002666660000000 + ret float %div1 + +; 0x3FF3333340000000 = 1.2f +; 0x4002666660000000 = 2.3f +; 0x3FE0B21660000000 = 0.52173918485641479492 +; CHECK: @fdiv2 +; CHECK: fmul fast float %x, 0x3FE0B21660000000 +} + +; "X/C1 / C2 => X * (1/(C2*C1))" is disabled (for now) is C2/C1 is a denormal +; +define float @fdiv3(float %x) { + %div = fdiv float %x, 0x47EFFFFFE0000000 + %div1 = fdiv fast float %div, 0x4002666660000000 + ret float %div1 +; CHECK: @fdiv3 +; CHECK: fdiv float %x, 0x47EFFFFFE0000000 +} + +; "X*C1 / C2 => X * (C1/C2)" is disabled if C1/C2 is a denormal +define float @fdiv4(float %x) { + %mul = fmul float %x, 0x47EFFFFFE0000000 + %div = fdiv float %mul, 0x3FC99999A0000000 + ret float %div +; CHECK: @fdiv4 +; CHECK: fmul float %x, 0x47EFFFFFE0000000 +} + +; (X/Y)/Z = > X/(Y*Z) +define float @fdiv5(float %f1, float %f2, float %f3) { + %t1 = fdiv float %f1, %f2 + %t2 = fdiv fast float %t1, %f3 + ret float %t2 +; CHECK: @fdiv5 +; CHECK: fmul float %f2, %f3 +} + +; Z/(X/Y) = > (Z*Y)/X +define float @fdiv6(float %f1, float %f2, float %f3) { + %t1 = fdiv float %f1, %f2 + %t2 = fdiv fast float %f3, %t1 + ret float %t2 +; CHECK: @fdiv6 +; CHECK: fmul float %f3, %f2 +} + +; C1/(X*C2) => (C1/C2) / X +define float @fdiv7(float %x) { + %t1 = fmul float %x, 3.0e0 + %t2 = fdiv fast float 15.0e0, %t1 + ret float %t2 +; CHECK: @fdiv7 +; CHECK: fdiv fast float 5.000000e+00, %x +} + +; C1/(X/C2) => (C1*C2) / X +define float @fdiv8(float %x) { + %t1 = fdiv float %x, 3.0e0 + %t2 = fdiv fast float 15.0e0, %t1 + ret float %t2 +; CHECK: @fdiv8 +; CHECK: fdiv fast float 4.500000e+01, %x +} + +; C1/(C2/X) => (C1/C2) * X +define float @fdiv9(float %x) { + %t1 = fdiv float 3.0e0, %x + %t2 = fdiv fast float 15.0e0, %t1 + ret float %t2 +; CHECK: @fdiv9 +; CHECK: fmul fast float %x, 5.000000e+00 +} + +; ========================================================================= +; +; Testing-cases about factorization +; +; ========================================================================= +; x*z + y*z => (x+y) * z +define float @fact_mul1(float %x, float %y, float %z) { + %t1 = fmul fast float %x, %z + %t2 = fmul fast float %y, %z + %t3 = fadd fast float %t1, %t2 + ret float %t3 +; CHECK: @fact_mul1 +; CHECK: fmul fast float %1, %z +} + +; z*x + y*z => (x+y) * z +define float @fact_mul2(float %x, float %y, float %z) { + %t1 = fmul fast float %z, %x + %t2 = fmul fast float %y, %z + %t3 = fsub fast float %t1, %t2 + ret float %t3 +; CHECK: @fact_mul2 +; CHECK: fmul fast float %1, %z +} + +; z*x - z*y => (x-y) * z +define float @fact_mul3(float %x, float %y, float %z) { + %t2 = fmul fast float %z, %y + %t1 = fmul fast float %z, %x + %t3 = fsub fast float %t1, %t2 + ret float %t3 +; CHECK: @fact_mul3 +; CHECK: fmul fast float %1, %z +} + +; x*z - z*y => (x-y) * z +define float @fact_mul4(float %x, float %y, float %z) { + %t1 = fmul fast float %x, %z + %t2 = fmul fast float %z, %y + %t3 = fsub fast float %t1, %t2 + ret float %t3 +; CHECK: @fact_mul4 +; CHECK: fmul fast float %1, %z +} + +; x/y + x/z, no xform +define float @fact_div1(float %x, float %y, float %z) { + %t1 = fdiv fast float %x, %y + %t2 = fdiv fast float %x, %z + %t3 = fadd fast float %t1, %t2 + ret float %t3 +; CHECK: fact_div1 +; CHECK: fadd fast float %t1, %t2 +} + +; x/y + z/x; no xform +define float @fact_div2(float %x, float %y, float %z) { + %t1 = fdiv fast float %x, %y + %t2 = fdiv fast float %z, %x + %t3 = fadd fast float %t1, %t2 + ret float %t3 +; CHECK: fact_div2 +; CHECK: fadd fast float %t1, %t2 +} + +; y/x + z/x => (y+z)/x +define float @fact_div3(float %x, float %y, float %z) { + %t1 = fdiv fast float %y, %x + %t2 = fdiv fast float %z, %x + %t3 = fadd fast float %t1, %t2 + ret float %t3 +; CHECK: fact_div3 +; CHECK: fdiv fast float %1, %x +} + +; y/x - z/x => (y-z)/x +define float @fact_div4(float %x, float %y, float %z) { + %t1 = fdiv fast float %y, %x + %t2 = fdiv fast float %z, %x + %t3 = fsub fast float %t1, %t2 + ret float %t3 +; CHECK: fact_div4 +; CHECK: fdiv fast float %1, %x +} + +; y/x - z/x => (y-z)/x is disabled if y-z is denormal. +define float @fact_div5(float %x) { + %t1 = fdiv fast float 0x3810000000000000, %x + %t2 = fdiv fast float 0x3800000000000000, %x + %t3 = fadd fast float %t1, %t2 + ret float %t3 +; CHECK: fact_div5 +; CHECK: fdiv fast float 0x3818000000000000, %x +} + +; y/x - z/x => (y-z)/x is disabled if y-z is denormal. +define float @fact_div6(float %x) { + %t1 = fdiv fast float 0x3810000000000000, %x + %t2 = fdiv fast float 0x3800000000000000, %x + %t3 = fsub fast float %t1, %t2 + ret float %t3 +; CHECK: fact_div6 +; CHECK: %t3 = fsub fast float %t1, %t2 +} diff --git a/test/Transforms/InstCombine/ffs-1.ll b/test/Transforms/InstCombine/ffs-1.ll new file mode 100644 index 0000000000000..0510df3d24b9a --- /dev/null +++ b/test/Transforms/InstCombine/ffs-1.ll @@ -0,0 +1,134 @@ +; Test that the ffs* library call simplifier works correctly. +; +; RUN: opt < %s -instcombine -S | FileCheck %s +; RUN: opt < %s -mtriple i386-pc-linux -instcombine -S | FileCheck %s -check-prefix=LINUX + +target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128" + +declare i32 @ffs(i32) +declare i32 @ffsl(i32) +declare i32 @ffsll(i64) + +; Check ffs(0) -> 0. + +define i32 @test_simplify1() { +; CHECK: @test_simplify1 + %ret = call i32 @ffs(i32 0) + ret i32 %ret +; CHECK-NEXT: ret i32 0 +} + +define i32 @test_simplify2() { +; CHECK-LINUX: @test_simplify2 + %ret = call i32 @ffsl(i32 0) + ret i32 %ret +; CHECK-LINUX-NEXT: ret i32 0 +} + +define i32 @test_simplify3() { +; CHECK-LINUX: @test_simplify3 + %ret = call i32 @ffsll(i64 0) + ret i32 %ret +; CHECK-LINUX-NEXT: ret i32 0 +} + +; Check ffs(c) -> cttz(c) + 1, where 'c' is a constant. + +define i32 @test_simplify4() { +; CHECK: @test_simplify4 + %ret = call i32 @ffs(i32 1) + ret i32 %ret +; CHECK-NEXT: ret i32 1 +} + +define i32 @test_simplify5() { +; CHECK: @test_simplify5 + %ret = call i32 @ffs(i32 2048) + ret i32 %ret +; CHECK-NEXT: ret i32 12 +} + +define i32 @test_simplify6() { +; CHECK: @test_simplify6 + %ret = call i32 @ffs(i32 65536) + ret i32 %ret +; CHECK-NEXT: ret i32 17 +} + +define i32 @test_simplify7() { +; CHECK-LINUX: @test_simplify7 + %ret = call i32 @ffsl(i32 65536) + ret i32 %ret +; CHECK-LINUX-NEXT: ret i32 17 +} + +define i32 @test_simplify8() { +; CHECK-LINUX: @test_simplify8 + %ret = call i32 @ffsll(i64 1024) + ret i32 %ret +; CHECK-LINUX-NEXT: ret i32 11 +} + +define i32 @test_simplify9() { +; CHECK-LINUX: @test_simplify9 + %ret = call i32 @ffsll(i64 65536) + ret i32 %ret +; CHECK-LINUX-NEXT: ret i32 17 +} + +define i32 @test_simplify10() { +; CHECK-LINUX: @test_simplify10 + %ret = call i32 @ffsll(i64 17179869184) + ret i32 %ret +; CHECK-LINUX-NEXT: ret i32 35 +} + +define i32 @test_simplify11() { +; CHECK-LINUX: @test_simplify11 + %ret = call i32 @ffsll(i64 281474976710656) + ret i32 %ret +; CHECK-LINUX-NEXT: ret i32 49 +} + +define i32 @test_simplify12() { +; CHECK-LINUX: @test_simplify12 + %ret = call i32 @ffsll(i64 1152921504606846976) + ret i32 %ret +; CHECK-LINUX-NEXT: ret i32 61 +} + +; Check ffs(x) -> x != 0 ? (i32)llvm.cttz(x) + 1 : 0. + +define i32 @test_simplify13(i32 %x) { +; CHECK: @test_simplify13 + %ret = call i32 @ffs(i32 %x) +; CHECK-NEXT: [[CTTZ:%[a-z0-9]+]] = call i32 @llvm.cttz.i32(i32 %x, i1 false) +; CHECK-NEXT: [[INC:%[a-z0-9]+]] = add i32 [[CTTZ]], 1 +; CHECK-NEXT: [[CMP:%[a-z0-9]+]] = icmp ne i32 %x, 0 +; CHECK-NEXT: [[RET:%[a-z0-9]+]] = select i1 [[CMP]], i32 [[INC]], i32 0 + ret i32 %ret +; CHECK-NEXT: ret i32 [[RET]] +} + +define i32 @test_simplify14(i32 %x) { +; CHECK-LINUX: @test_simplify14 + %ret = call i32 @ffsl(i32 %x) +; CHECK-LINUX-NEXT: [[CTTZ:%[a-z0-9]+]] = call i32 @llvm.cttz.i32(i32 %x, i1 false) +; CHECK-LINUX-NEXT: [[INC:%[a-z0-9]+]] = add i32 [[CTTZ]], 1 +; CHECK-LINUX-NEXT: [[CMP:%[a-z0-9]+]] = icmp ne i32 %x, 0 +; CHECK-LINUX-NEXT: [[RET:%[a-z0-9]+]] = select i1 [[CMP]], i32 [[INC]], i32 0 + ret i32 %ret +; CHECK-LINUX-NEXT: ret i32 [[RET]] +} + +define i32 @test_simplify15(i64 %x) { +; CHECK-LINUX: @test_simplify15 + %ret = call i32 @ffsll(i64 %x) +; CHECK-LINUX-NEXT: [[CTTZ:%[a-z0-9]+]] = call i64 @llvm.cttz.i64(i64 %x, i1 false) +; CHECK-LINUX-NEXT: [[INC:%[a-z0-9]+]] = add i64 [[CTTZ]], 1 +; CHECK-LINUX-NEXT: [[TRUNC:%[a-z0-9]+]] = trunc i64 [[INC]] to i32 +; CHECK-LINUX-NEXT: [[CMP:%[a-z0-9]+]] = icmp ne i64 %x, 0 +; CHECK-LINUX-NEXT: [[RET:%[a-z0-9]+]] = select i1 [[CMP]], i32 [[TRUNC]], i32 0 + ret i32 %ret +; CHECK-LINUX-NEXT: ret i32 [[RET]] +} diff --git a/test/Transforms/InstCombine/fmul.ll b/test/Transforms/InstCombine/fmul.ll new file mode 100644 index 0000000000000..3671b4c6991cb --- /dev/null +++ b/test/Transforms/InstCombine/fmul.ll @@ -0,0 +1,72 @@ +; RUN: opt -S -instcombine < %s | FileCheck %s + +; (-0.0 - X) * C => X * -C +define float @test1(float %x) { + %sub = fsub float -0.000000e+00, %x + %mul = fmul float %sub, 2.0e+1 + ret float %mul + +; CHECK: @test1 +; CHECK: fmul float %x, -2.000000e+01 +} + +; (0.0 - X) * C => X * -C +define float @test2(float %x) { + %sub = fsub nsz float 0.000000e+00, %x + %mul = fmul float %sub, 2.0e+1 + ret float %mul + +; CHECK: @test2 +; CHECK: fmul float %x, -2.000000e+01 +} + +; (-0.0 - X) * (-0.0 - Y) => X * Y +define float @test3(float %x, float %y) { + %sub1 = fsub float -0.000000e+00, %x + %sub2 = fsub float -0.000000e+00, %y + %mul = fmul float %sub1, %sub2 + ret float %mul +; CHECK: @test3 +; CHECK: fmul float %x, %y +} + +; (0.0 - X) * (0.0 - Y) => X * Y +define float @test4(float %x, float %y) { + %sub1 = fsub nsz float 0.000000e+00, %x + %sub2 = fsub nsz float 0.000000e+00, %y + %mul = fmul float %sub1, %sub2 + ret float %mul +; CHECK: @test4 +; CHECK: fmul float %x, %y +} + +; (-0.0 - X) * Y => -0.0 - (X * Y) +define float @test5(float %x, float %y) { + %sub1 = fsub float -0.000000e+00, %x + %mul = fmul float %sub1, %y + ret float %mul +; CHECK: @test5 +; CHECK: %1 = fmul float %x, %y +; CHECK: %mul = fsub float -0.000000e+00, %1 +} + +; (0.0 - X) * Y => 0.0 - (X * Y) +define float @test6(float %x, float %y) { + %sub1 = fsub nsz float 0.000000e+00, %x + %mul = fmul float %sub1, %y + ret float %mul +; CHECK: @test6 +; CHECK: %1 = fmul float %x, %y +; CHECK: %mul = fsub float -0.000000e+00, %1 +} + +; "(-0.0 - X) * Y => -0.0 - (X * Y)" is disabled if expression "-0.0 - X" +; has multiple uses. +define float @test7(float %x, float %y) { + %sub1 = fsub float -0.000000e+00, %x + %mul = fmul float %sub1, %y + %mul2 = fmul float %mul, %sub1 + ret float %mul2 +; CHECK: @test7 +; CHECK: fsub float -0.000000e+00, %x +} diff --git a/test/Transforms/InstCombine/fold-phi.ll b/test/Transforms/InstCombine/fold-phi.ll new file mode 100644 index 0000000000000..bd01d58aa5860 --- /dev/null +++ b/test/Transforms/InstCombine/fold-phi.ll @@ -0,0 +1,39 @@ +; RUN: opt < %s -instcombine -S | FileCheck %s + +; CHECK: no_crash +define float @no_crash(float %a) nounwind { +entry: + br label %for.body + +for.body: + %sum.057 = phi float [ 0.000000e+00, %entry ], [ %add5, %bb0 ] + %add5 = fadd float %sum.057, %a ; PR14592 + br i1 undef, label %bb0, label %end + +bb0: + br label %for.body + +end: + ret float %add5 +} + +; CHECK: fold_phi +define float @fold_phi(float %a) nounwind { +entry: + br label %for.body + +for.body: +; CHECK: phi float +; CHECK-NEXT: br i1 undef + %sum.057 = phi float [ 0.000000e+00, %entry ], [ %add5, %bb0 ] + %add5 = fadd float %sum.057, 1.0 ;; Should be moved to the latch! + br i1 undef, label %bb0, label %end + +; CHECK: bb0: +bb0: +; CHECK: fadd float + br label %for.body + +end: + ret float %add5 +} diff --git a/test/Transforms/InstCombine/fpcast.ll b/test/Transforms/InstCombine/fpcast.ll index bc6aa0a6891f6..09f053289dc17 100644 --- a/test/Transforms/InstCombine/fpcast.ll +++ b/test/Transforms/InstCombine/fpcast.ll @@ -13,3 +13,22 @@ define i8 @test2() { ; CHECK: ret i8 -1 } +; CHECK: test3 +define half @test3(float %a) { +; CHECK: fptrunc +; CHECK: llvm.fabs.f16 + %b = call float @llvm.fabs.f32(float %a) + %c = fptrunc float %b to half + ret half %c +} + +; CHECK: test4 +define half @test4(float %a) { +; CHECK: fptrunc +; CHECK: fsub + %b = fsub float -0.0, %a + %c = fptrunc float %b to half + ret half %c +} + +declare float @llvm.fabs.f32(float) nounwind readonly diff --git a/test/Transforms/InstCombine/fprintf-1.ll b/test/Transforms/InstCombine/fprintf-1.ll new file mode 100644 index 0000000000000..39d86b4588cc8 --- /dev/null +++ b/test/Transforms/InstCombine/fprintf-1.ll @@ -0,0 +1,80 @@ +; Test that the fprintf library call simplifier works correctly. +; +; RUN: opt < %s -instcombine -S | FileCheck %s +; RUN: opt < %s -mtriple xcore-xmos-elf -instcombine -S | FileCheck %s -check-prefix=IPRINTF + +target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128" + +%FILE = type { } + +@hello_world = constant [13 x i8] c"hello world\0A\00" +@percent_c = constant [3 x i8] c"%c\00" +@percent_d = constant [3 x i8] c"%d\00" +@percent_f = constant [3 x i8] c"%f\00" +@percent_s = constant [3 x i8] c"%s\00" + +declare i32 @fprintf(%FILE*, i8*, ...) + +; Check fprintf(fp, "foo") -> fwrite("foo", 3, 1, fp). + +define void @test_simplify1(%FILE* %fp) { +; CHECK: @test_simplify1 + %fmt = getelementptr [13 x i8]* @hello_world, i32 0, i32 0 + call i32 (%FILE*, i8*, ...)* @fprintf(%FILE* %fp, i8* %fmt) +; CHECK-NEXT: call i32 @fwrite(i8* getelementptr inbounds ([13 x i8]* @hello_world, i32 0, i32 0), i32 12, i32 1, %FILE* %fp) + ret void +; CHECK-NEXT: ret void +} + +; Check fprintf(fp, "%c", chr) -> fputc(chr, fp). + +define void @test_simplify2(%FILE* %fp) { +; CHECK: @test_simplify2 + %fmt = getelementptr [3 x i8]* @percent_c, i32 0, i32 0 + call i32 (%FILE*, i8*, ...)* @fprintf(%FILE* %fp, i8* %fmt, i8 104) +; CHECK-NEXT: call i32 @fputc(i32 104, %FILE* %fp) + ret void +; CHECK-NEXT: ret void +} + +; Check fprintf(fp, "%s", str) -> fputs(str, fp). +; NOTE: The fputs simplifier simplifies this further to fwrite. + +define void @test_simplify3(%FILE* %fp) { +; CHECK: @test_simplify3 + %fmt = getelementptr [3 x i8]* @percent_s, i32 0, i32 0 + %str = getelementptr [13 x i8]* @hello_world, i32 0, i32 0 + call i32 (%FILE*, i8*, ...)* @fprintf(%FILE* %fp, i8* %fmt, i8* %str) +; CHECK-NEXT: call i32 @fwrite(i8* getelementptr inbounds ([13 x i8]* @hello_world, i32 0, i32 0), i32 12, i32 1, %FILE* %fp) + ret void +; CHECK-NEXT: ret void +} + +; Check fprintf(fp, fmt, ...) -> fiprintf(fp, fmt, ...) if no floating point. + +define void @test_simplify4(%FILE* %fp) { +; CHECK-IPRINTF: @test_simplify4 + %fmt = getelementptr [3 x i8]* @percent_d, i32 0, i32 0 + call i32 (%FILE*, i8*, ...)* @fprintf(%FILE* %fp, i8* %fmt, i32 187) +; CHECK-NEXT-IPRINTF: call i32 (%FILE*, i8*, ...)* @fiprintf(%FILE* %fp, i8* getelementptr inbounds ([3 x i8]* @percent_d, i32 0, i32 0), i32 187) + ret void +; CHECK-NEXT-IPRINTF: ret void +} + +define void @test_no_simplify1(%FILE* %fp) { +; CHECK-IPRINTF: @test_no_simplify1 + %fmt = getelementptr [3 x i8]* @percent_f, i32 0, i32 0 + call i32 (%FILE*, i8*, ...)* @fprintf(%FILE* %fp, i8* %fmt, double 1.87) +; CHECK-NEXT-IPRINTF: call i32 (%FILE*, i8*, ...)* @fprintf(%FILE* %fp, i8* getelementptr inbounds ([3 x i8]* @percent_f, i32 0, i32 0), double 1.870000e+00) + ret void +; CHECK-NEXT-IPRINTF: ret void +} + +define void @test_no_simplify2(%FILE* %fp, double %d) { +; CHECK: @test_no_simplify2 + %fmt = getelementptr [3 x i8]* @percent_f, i32 0, i32 0 + call i32 (%FILE*, i8*, ...)* @fprintf(%FILE* %fp, i8* %fmt, double %d) +; CHECK-NEXT: call i32 (%FILE*, i8*, ...)* @fprintf(%FILE* %fp, i8* getelementptr inbounds ([3 x i8]* @percent_f, i32 0, i32 0), double %d) + ret void +; CHECK-NEXT: ret void +} diff --git a/test/Transforms/InstCombine/fputs-1.ll b/test/Transforms/InstCombine/fputs-1.ll new file mode 100644 index 0000000000000..c7c5becfd038c --- /dev/null +++ b/test/Transforms/InstCombine/fputs-1.ll @@ -0,0 +1,43 @@ +; Test that the fputs library call simplifier works correctly. +; +; RUN: opt < %s -instcombine -S | FileCheck %s + +target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128" + +%FILE = type { } + +@empty = constant [1 x i8] zeroinitializer +@A = constant [2 x i8] c"A\00" +@hello = constant [7 x i8] c"hello\0A\00" + +declare i32 @fputs(i8*, %FILE*) + +; Check fputs(str, fp) --> fwrite(str, 1, strlen(s), fp). + +define void @test_simplify1(%FILE* %fp) { +; CHECK: @test_simplify1 + %str = getelementptr [1 x i8]* @empty, i32 0, i32 0 + call i32 @fputs(i8* %str, %FILE* %fp) + ret void +; CHECK-NEXT: ret void +} + +; NOTE: The fwrite simplifier simplifies this further to fputc. + +define void @test_simplify2(%FILE* %fp) { +; CHECK: @test_simplify2 + %str = getelementptr [2 x i8]* @A, i32 0, i32 0 + call i32 @fputs(i8* %str, %FILE* %fp) +; CHECK-NEXT: call i32 @fputc(i32 65, %FILE* %fp) + ret void +; CHECK-NEXT: ret void +} + +define void @test_simplify3(%FILE* %fp) { +; CHECK: @test_simplify3 + %str = getelementptr [7 x i8]* @hello, i32 0, i32 0 + call i32 @fputs(i8* %str, %FILE* %fp) +; CHECK-NEXT: call i32 @fwrite(i8* getelementptr inbounds ([7 x i8]* @hello, i32 0, i32 0), i32 6, i32 1, %FILE* %fp) + ret void +; CHECK-NEXT: ret void +} diff --git a/test/Transforms/InstCombine/fwrite-1.ll b/test/Transforms/InstCombine/fwrite-1.ll new file mode 100644 index 0000000000000..528cdec217f7e --- /dev/null +++ b/test/Transforms/InstCombine/fwrite-1.ll @@ -0,0 +1,57 @@ +; Test that the fwrite library call simplifier works correctly. +; +; RUN: opt < %s -instcombine -S | FileCheck %s + +target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128" + +%FILE = type { } + +@str = constant [1 x i8] zeroinitializer +@empty = constant [0 x i8] zeroinitializer + +declare i64 @fwrite(i8*, i64, i64, %FILE *) + +; Check fwrite(S, 1, 1, fp) -> fputc(S[0], fp). + +define void @test_simplify1(%FILE* %fp) { +; CHECK: @test_simplify1 + %str = getelementptr inbounds [1 x i8]* @str, i64 0, i64 0 + call i64 @fwrite(i8* %str, i64 1, i64 1, %FILE* %fp) +; CHECK-NEXT: call i32 @fputc(i32 0, %FILE* %fp) + ret void +; CHECK-NEXT: ret void +} + +define void @test_simplify2(%FILE* %fp) { +; CHECK: @test_simplify2 + %str = getelementptr inbounds [0 x i8]* @empty, i64 0, i64 0 + call i64 @fwrite(i8* %str, i64 1, i64 0, %FILE* %fp) + ret void +; CHECK-NEXT: ret void +} + +define void @test_simplify3(%FILE* %fp) { +; CHECK: @test_simplify3 + %str = getelementptr inbounds [0 x i8]* @empty, i64 0, i64 0 + call i64 @fwrite(i8* %str, i64 0, i64 1, %FILE* %fp) + ret void +; CHECK-NEXT: ret void +} + +define i64 @test_no_simplify1(%FILE* %fp) { +; CHECK: @test_no_simplify1 + %str = getelementptr inbounds [1 x i8]* @str, i64 0, i64 0 + %ret = call i64 @fwrite(i8* %str, i64 1, i64 1, %FILE* %fp) +; CHECK-NEXT: call i64 @fwrite + ret i64 %ret +; CHECK-NEXT: ret i64 %ret +} + +define void @test_no_simplify2(%FILE* %fp, i64 %size) { +; CHECK: @test_no_simplify2 + %str = getelementptr inbounds [1 x i8]* @str, i64 0, i64 0 + call i64 @fwrite(i8* %str, i64 %size, i64 1, %FILE* %fp) +; CHECK-NEXT: call i64 @fwrite + ret void +; CHECK-NEXT: ret void +} diff --git a/test/Transforms/InstCombine/getelementptr.ll b/test/Transforms/InstCombine/getelementptr.ll index 1c120ecbe9eb7..bb07736ef8037 100644 --- a/test/Transforms/InstCombine/getelementptr.ll +++ b/test/Transforms/InstCombine/getelementptr.ll @@ -424,7 +424,7 @@ define i32 @test35() nounwind { i8* getelementptr (%t1* bitcast (%t0* @s to %t1*), i32 0, i32 1, i32 0)) nounwind ret i32 0 ; CHECK: @test35 -; CHECK: call i32 (i8*, ...)* @printf(i8* getelementptr inbounds ([17 x i8]* @"\01LC8", i64 0, i64 0), i8* getelementptr inbounds (%t0* @s, i64 0, i32 1, i64 0)) nounwind +; CHECK: call i32 (i8*, ...)* @printf(i8* getelementptr inbounds ([17 x i8]* @"\01LC8", i64 0, i64 0), i8* getelementptr inbounds (%t0* @s, i64 0, i32 1, i64 0)) [[NUW:#[0-9]+]] } ; Instcombine should constant-fold the GEP so that indices that have @@ -492,3 +492,21 @@ define void @three_gep_f(%three_gep_t2* %x) { declare void @three_gep_g(i32*) declare void @three_gep_h(%three_gep_t2*) + +%struct.ham = type { i32, %struct.zot*, %struct.zot*, %struct.zot* } +%struct.zot = type { i64, i8 } + +define void @test39(%struct.ham* %arg, i8 %arg1) nounwind { + %tmp = getelementptr inbounds %struct.ham* %arg, i64 0, i32 2 + %tmp2 = load %struct.zot** %tmp, align 8 + %tmp3 = bitcast %struct.zot* %tmp2 to i8* + %tmp4 = getelementptr inbounds i8* %tmp3, i64 -8 + store i8 %arg1, i8* %tmp4, align 8 + ret void + +; CHECK: @test39 +; CHECK: getelementptr inbounds %struct.ham* %arg, i64 0, i32 2 +; CHECK: getelementptr inbounds i8* %tmp3, i64 -8 +} + +; CHECK: attributes [[NUW]] = { nounwind } diff --git a/test/Transforms/InstCombine/icmp.ll b/test/Transforms/InstCombine/icmp.ll index 8e064a4f2fc94..446c0e01dcaa5 100644 --- a/test/Transforms/InstCombine/icmp.ll +++ b/test/Transforms/InstCombine/icmp.ll @@ -677,3 +677,212 @@ define i1 @test66(i64 %A, i64 %B) { ; CHECK-NEXT: ret i1 true ret i1 %cmp } + +; CHECK: @test67 +; CHECK: %and = and i32 %x, 96 +; CHECK: %cmp = icmp ne i32 %and, 0 +define i1 @test67(i32 %x) nounwind uwtable { + %and = and i32 %x, 127 + %cmp = icmp sgt i32 %and, 31 + ret i1 %cmp +} + +; CHECK: @test68 +; CHECK: %cmp = icmp ugt i32 %and, 30 +define i1 @test68(i32 %x) nounwind uwtable { + %and = and i32 %x, 127 + %cmp = icmp sgt i32 %and, 30 + ret i1 %cmp +} + +; PR14708 +; CHECK: @test69 +; CHECK: %1 = and i32 %c, -33 +; CHECK: %2 = icmp eq i32 %1, 65 +; CHECK: ret i1 %2 +define i1 @test69(i32 %c) nounwind uwtable { + %1 = icmp eq i32 %c, 97 + %2 = icmp eq i32 %c, 65 + %3 = or i1 %1, %2 + ret i1 %3 +} + +; CHECK: @icmp_sext16trunc +; CHECK-NEXT: %1 = trunc i32 %x to i16 +; CHECK-NEXT: %cmp = icmp slt i16 %1, 36 +define i1 @icmp_sext16trunc(i32 %x) { + %trunc = trunc i32 %x to i16 + %sext = sext i16 %trunc to i32 + %cmp = icmp slt i32 %sext, 36 + ret i1 %cmp +} + +; CHECK: @icmp_sext8trunc +; CHECK-NEXT: %1 = trunc i32 %x to i8 +; CHECK-NEXT: %cmp = icmp slt i8 %1, 36 +define i1 @icmp_sext8trunc(i32 %x) { + %trunc = trunc i32 %x to i8 + %sext = sext i8 %trunc to i32 + %cmp = icmp slt i32 %sext, 36 + ret i1 %cmp +} + +; CHECK: @icmp_shl16 +; CHECK-NEXT: %1 = trunc i32 %x to i16 +; CHECK-NEXT: %cmp = icmp slt i16 %1, 36 +define i1 @icmp_shl16(i32 %x) { + %shl = shl i32 %x, 16 + %cmp = icmp slt i32 %shl, 2359296 + ret i1 %cmp +} + +; CHECK: @icmp_shl24 +; CHECK-NEXT: %1 = trunc i32 %x to i8 +; CHECK-NEXT: %cmp = icmp slt i8 %1, 36 +define i1 @icmp_shl24(i32 %x) { + %shl = shl i32 %x, 24 + %cmp = icmp slt i32 %shl, 603979776 + ret i1 %cmp +} + +; If the (shl x, C) preserved the sign and this is a sign test, +; compare the LHS operand instead +; CHECK: @icmp_shl_nsw_sgt +; CHECK-NEXT: icmp sgt i32 %x, 0 +define i1 @icmp_shl_nsw_sgt(i32 %x) { + %shl = shl nsw i32 %x, 21 + %cmp = icmp sgt i32 %shl, 0 + ret i1 %cmp +} + +; CHECK: @icmp_shl_nsw_sge0 +; CHECK-NEXT: icmp sgt i32 %x, -1 +define i1 @icmp_shl_nsw_sge0(i32 %x) { + %shl = shl nsw i32 %x, 21 + %cmp = icmp sge i32 %shl, 0 + ret i1 %cmp +} + +; CHECK: @icmp_shl_nsw_sge1 +; CHECK-NEXT: icmp sgt i32 %x, 0 +define i1 @icmp_shl_nsw_sge1(i32 %x) { + %shl = shl nsw i32 %x, 21 + %cmp = icmp sge i32 %shl, 1 + ret i1 %cmp +} + +; Checks for icmp (eq|ne) (shl x, C), 0 +; CHECK: @icmp_shl_nsw_eq +; CHECK-NEXT: icmp eq i32 %x, 0 +define i1 @icmp_shl_nsw_eq(i32 %x) { + %mul = shl nsw i32 %x, 5 + %cmp = icmp eq i32 %mul, 0 + ret i1 %cmp +} + +; CHECK: @icmp_shl_eq +; CHECK-NOT: icmp eq i32 %mul, 0 +define i1 @icmp_shl_eq(i32 %x) { + %mul = shl i32 %x, 5 + %cmp = icmp eq i32 %mul, 0 + ret i1 %cmp +} + +; CHECK: @icmp_shl_nsw_ne +; CHECK-NEXT: icmp ne i32 %x, 0 +define i1 @icmp_shl_nsw_ne(i32 %x) { + %mul = shl nsw i32 %x, 7 + %cmp = icmp ne i32 %mul, 0 + ret i1 %cmp +} + +; CHECK: @icmp_shl_ne +; CHECK-NOT: icmp ne i32 %x, 0 +define i1 @icmp_shl_ne(i32 %x) { + %mul = shl i32 %x, 7 + %cmp = icmp ne i32 %mul, 0 + ret i1 %cmp +} + +; If the (mul x, C) preserved the sign and this is sign test, +; compare the LHS operand instead +; CHECK: @icmp_mul_nsw +; CHECK-NEXT: icmp sgt i32 %x, 0 +define i1 @icmp_mul_nsw(i32 %x) { + %mul = mul nsw i32 %x, 12 + %cmp = icmp sgt i32 %mul, 0 + ret i1 %cmp +} + +; CHECK: @icmp_mul_nsw1 +; CHECK-NEXT: icmp slt i32 %x, 0 +define i1 @icmp_mul_nsw1(i32 %x) { + %mul = mul nsw i32 %x, 12 + %cmp = icmp sle i32 %mul, -1 + ret i1 %cmp +} + +; CHECK: @icmp_mul_nsw_neg +; CHECK-NEXT: icmp slt i32 %x, 1 +define i1 @icmp_mul_nsw_neg(i32 %x) { + %mul = mul nsw i32 %x, -12 + %cmp = icmp sge i32 %mul, 0 + ret i1 %cmp +} + +; CHECK: @icmp_mul_nsw_neg1 +; CHECK-NEXT: icmp slt i32 %x, 0 +define i1 @icmp_mul_nsw_neg1(i32 %x) { + %mul = mul nsw i32 %x, -12 + %cmp = icmp sge i32 %mul, 1 + ret i1 %cmp +} + +; CHECK: @icmp_mul_nsw_0 +; CHECK-NOT: icmp sgt i32 %x, 0 +define i1 @icmp_mul_nsw_0(i32 %x) { + %mul = mul nsw i32 %x, 0 + %cmp = icmp sgt i32 %mul, 0 + ret i1 %cmp +} + +; CHECK: @icmp_mul +; CHECK-NEXT: %mul = mul i32 %x, -12 +define i1 @icmp_mul(i32 %x) { + %mul = mul i32 %x, -12 + %cmp = icmp sge i32 %mul, 0 + ret i1 %cmp +} + +; Checks for icmp (eq|ne) (mul x, C), 0 +; CHECK: @icmp_mul_neq0 +; CHECK-NEXT: icmp ne i32 %x, 0 +define i1 @icmp_mul_neq0(i32 %x) { + %mul = mul nsw i32 %x, -12 + %cmp = icmp ne i32 %mul, 0 + ret i1 %cmp +} + +; CHECK: @icmp_mul_eq0 +; CHECK-NEXT: icmp eq i32 %x, 0 +define i1 @icmp_mul_eq0(i32 %x) { + %mul = mul nsw i32 %x, 12 + %cmp = icmp eq i32 %mul, 0 + ret i1 %cmp +} + +; CHECK: @icmp_mul0_eq0 +; CHECK-NEXT: ret i1 true +define i1 @icmp_mul0_eq0(i32 %x) { + %mul = mul i32 %x, 0 + %cmp = icmp eq i32 %mul, 0 + ret i1 %cmp +} + +; CHECK: @icmp_mul0_ne0 +; CHECK-NEXT: ret i1 false +define i1 @icmp_mul0_ne0(i32 %x) { + %mul = mul i32 %x, 0 + %cmp = icmp ne i32 %mul, 0 + ret i1 %cmp +} diff --git a/test/Transforms/InstCombine/idioms.ll b/test/Transforms/InstCombine/idioms.ll index 6b3567fc6e8d6..1a211668c3bf5 100644 --- a/test/Transforms/InstCombine/idioms.ll +++ b/test/Transforms/InstCombine/idioms.ll @@ -1,4 +1,4 @@ -; RUN: opt -instcombine %s -S | FileCheck %s +; RUN: opt -instcombine -S < %s | FileCheck %s ; Check that code corresponding to the following C function is ; simplified into a single ASR operation: diff --git a/test/Transforms/InstCombine/intrinsics.ll b/test/Transforms/InstCombine/intrinsics.ll index 382e6b38574d0..f334b3b1e9353 100644 --- a/test/Transforms/InstCombine/intrinsics.ll +++ b/test/Transforms/InstCombine/intrinsics.ll @@ -1,4 +1,4 @@ -; RUN: opt %s -instcombine -S | FileCheck %s +; RUN: opt -instcombine -S < %s | FileCheck %s %overflow.result = type {i8, i1} @@ -220,3 +220,39 @@ define i32 @cttz_simplify1b(i32 %x) nounwind readnone ssp { ; CHECK: @cttz_simplify1b ; CHECK-NEXT: ret i32 0 } + +define i32 @ctlz_undef(i32 %Value) nounwind { + %ctlz = call i32 @llvm.ctlz.i32(i32 0, i1 true) + ret i32 %ctlz + +; CHECK: @ctlz_undef +; CHECK-NEXT: ret i32 undef +} + +define i32 @cttz_undef(i32 %Value) nounwind { + %cttz = call i32 @llvm.cttz.i32(i32 0, i1 true) + ret i32 %cttz + +; CHECK: @cttz_undef +; CHECK-NEXT: ret i32 undef +} + +define i32 @ctlz_select(i32 %Value) nounwind { + %tobool = icmp ne i32 %Value, 0 + %ctlz = call i32 @llvm.ctlz.i32(i32 %Value, i1 true) + %s = select i1 %tobool, i32 %ctlz, i32 32 + ret i32 %s + +; CHECK: @ctlz_select +; CHECK: select i1 %tobool, i32 %ctlz, i32 32 +} + +define i32 @cttz_select(i32 %Value) nounwind { + %tobool = icmp ne i32 %Value, 0 + %cttz = call i32 @llvm.cttz.i32(i32 %Value, i1 true) + %s = select i1 %tobool, i32 %cttz, i32 32 + ret i32 %s + +; CHECK: @cttz_select +; CHECK: select i1 %tobool, i32 %cttz, i32 32 +} diff --git a/test/Transforms/InstCombine/isascii-1.ll b/test/Transforms/InstCombine/isascii-1.ll new file mode 100644 index 0000000000000..2a413d89b4923 --- /dev/null +++ b/test/Transforms/InstCombine/isascii-1.ll @@ -0,0 +1,32 @@ +; Test that the isascii library call simplifier works correctly. +; +; RUN: opt < %s -instcombine -S | FileCheck %s + +target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128" + +declare i32 @isascii(i32) + +; Check isascii(c) -> c (c - '0') [#uses=1] ret i32 %j -; CHECJ: sub i32 0, %i +; CHECK: sub i32 0, %i } define i32 @test10(i32 %a, i32 %b) { @@ -138,10 +138,8 @@ define i32 @test16(i32 %b, i1 %c) { ; e = b & (a >> 31) %e = mul i32 %d, %b ; [#uses=1] ret i32 %e -; CHECK: [[TEST16:%.*]] = zext i1 %c to i32 -; CHECK-NEXT: %1 = sub i32 0, [[TEST16]] -; CHECK-NEXT: %e = and i32 %1, %b -; CHECK-NEXT: ret i32 %e +; CHECK: [[TEST16:%.*]] = select i1 %c, i32 %b, i32 0 +; CHECK-NEXT: ret i32 [[TEST16]] } ; X * Y (when Y is 0 or 1) --> x & (0-Y) diff --git a/test/Transforms/InstCombine/obfuscated_splat.ll b/test/Transforms/InstCombine/obfuscated_splat.ll index c25dade168a42..fa9cb423d02c8 100644 --- a/test/Transforms/InstCombine/obfuscated_splat.ll +++ b/test/Transforms/InstCombine/obfuscated_splat.ll @@ -1,4 +1,4 @@ -; RUN: opt -instcombine -S %s | FileCheck %s +; RUN: opt -instcombine -S < %s | FileCheck %s define void @test(<4 x float> *%in_ptr, <4 x float> *%out_ptr) { %A = load <4 x float>* %in_ptr, align 16 diff --git a/test/Transforms/InstCombine/objsize.ll b/test/Transforms/InstCombine/objsize.ll index 31a3cb46e4595..0ead9d123749c 100644 --- a/test/Transforms/InstCombine/objsize.ll +++ b/test/Transforms/InstCombine/objsize.ll @@ -256,3 +256,131 @@ xpto: return: ret i32 7 } + +declare noalias i8* @valloc(i32) nounwind + +; CHECK: @test14 +; CHECK: ret i32 6 +define i32 @test14(i32 %a) nounwind { + switch i32 %a, label %sw.default [ + i32 1, label %sw.bb + i32 2, label %sw.bb1 + ] + +sw.bb: + %call = tail call noalias i8* @malloc(i32 6) nounwind + br label %sw.epilog + +sw.bb1: + %call2 = tail call noalias i8* @calloc(i32 3, i32 2) nounwind + br label %sw.epilog + +sw.default: + %call3 = tail call noalias i8* @valloc(i32 6) nounwind + br label %sw.epilog + +sw.epilog: + %b.0 = phi i8* [ %call3, %sw.default ], [ %call2, %sw.bb1 ], [ %call, %sw.bb ] + %1 = tail call i32 @llvm.objectsize.i32(i8* %b.0, i1 false) + ret i32 %1 +} + +; CHECK: @test15 +; CHECK: llvm.objectsize +define i32 @test15(i32 %a) nounwind { + switch i32 %a, label %sw.default [ + i32 1, label %sw.bb + i32 2, label %sw.bb1 + ] + +sw.bb: + %call = tail call noalias i8* @malloc(i32 3) nounwind + br label %sw.epilog + +sw.bb1: + %call2 = tail call noalias i8* @calloc(i32 2, i32 1) nounwind + br label %sw.epilog + +sw.default: + %call3 = tail call noalias i8* @valloc(i32 3) nounwind + br label %sw.epilog + +sw.epilog: + %b.0 = phi i8* [ %call3, %sw.default ], [ %call2, %sw.bb1 ], [ %call, %sw.bb ] + %1 = tail call i32 @llvm.objectsize.i32(i8* %b.0, i1 false) + ret i32 %1 +} + +; CHECK: @test16 +; CHECK: llvm.objectsize +define i32 @test16(i8* %a, i32 %n) nounwind { + %b = alloca [5 x i8], align 1 + %c = alloca [5 x i8], align 1 + switch i32 %n, label %sw.default [ + i32 1, label %sw.bb + i32 2, label %sw.bb1 + ] + +sw.bb: + %bp = bitcast [5 x i8]* %b to i8* + br label %sw.epilog + +sw.bb1: + %cp = bitcast [5 x i8]* %c to i8* + br label %sw.epilog + +sw.default: + br label %sw.epilog + +sw.epilog: + %phi = phi i8* [ %a, %sw.default ], [ %cp, %sw.bb1 ], [ %bp, %sw.bb ] + %sz = call i32 @llvm.objectsize.i32(i8* %phi, i1 false) + ret i32 %sz +} + +; CHECK: @test17 +; CHECK: ret i32 5 +define i32 @test17(i32 %n) nounwind { + %b = alloca [5 x i8], align 1 + %c = alloca [5 x i8], align 1 + %bp = bitcast [5 x i8]* %b to i8* + switch i32 %n, label %sw.default [ + i32 1, label %sw.bb + i32 2, label %sw.bb1 + ] + +sw.bb: + br label %sw.epilog + +sw.bb1: + %cp = bitcast [5 x i8]* %c to i8* + br label %sw.epilog + +sw.default: + br label %sw.epilog + +sw.epilog: + %phi = phi i8* [ %bp, %sw.default ], [ %cp, %sw.bb1 ], [ %bp, %sw.bb ] + %sz = call i32 @llvm.objectsize.i32(i8* %phi, i1 false) + ret i32 %sz +} + +@globalalias = alias internal [60 x i8]* @a + +; CHECK: @test18 +; CHECK-NEXT: ret i32 60 +define i32 @test18() { + %bc = bitcast [60 x i8]* @globalalias to i8* + %1 = call i32 @llvm.objectsize.i32(i8* %bc, i1 false) + ret i32 %1 +} + +@globalalias2 = alias weak [60 x i8]* @a + +; CHECK: @test19 +; CHECK: llvm.objectsize +define i32 @test19() { + %bc = bitcast [60 x i8]* @globalalias2 to i8* + %1 = call i32 @llvm.objectsize.i32(i8* %bc, i1 false) + ret i32 %1 +} diff --git a/test/Transforms/InstCombine/or.ll b/test/Transforms/InstCombine/or.ll index c0bb28d15ccf4..bde2a54048add 100644 --- a/test/Transforms/InstCombine/or.ll +++ b/test/Transforms/InstCombine/or.ll @@ -344,10 +344,9 @@ define <4 x i32> @test32(<4 x i1> %and.i1352, <4 x i32> %vecinit6.i176, <4 x i32 %and.i = and <4 x i32> %vecinit6.i191, %neg.i ; <<4 x i32>> [#uses=1] %or.i = or <4 x i32> %and.i, %and.i129 ; <<4 x i32>> [#uses=1] ret <4 x i32> %or.i -; Don't turn this into a vector select until codegen matures to handle them -; better. +; codegen is mature enough to handle vector selects. ; CHECK: @test32 -; CHECK: or <4 x i32> %and.i, %and.i129 +; CHECK: select <4 x i1> %and.i1352, <4 x i32> %vecinit6.i176, <4 x i32> %vecinit6.i191 } define i1 @test33(i1 %X, i1 %Y) { diff --git a/test/Transforms/InstCombine/osx-names.ll b/test/Transforms/InstCombine/osx-names.ll new file mode 100644 index 0000000000000..7b83526aceb50 --- /dev/null +++ b/test/Transforms/InstCombine/osx-names.ll @@ -0,0 +1,30 @@ +; RUN: opt < %s -instcombine -S | FileCheck %s +; +; On OSX x86-32, fwrite and fputs aren't called fwrite and fputs. +; Make sure we use the correct names. + +target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128-n8:16:32-S128" +target triple = "i386-apple-macosx10.7.2" + +%struct.__sFILE = type { i8*, i32, i32, i16, i16, %struct.__sbuf, i32, i8*, i32 (i8*)*, i32 (i8*, i8*, i32)*, i64 (i8*, i64, i32)*, i32 (i8*, i8*, i32)*, %struct.__sbuf, %struct.__sFILEX*, i32, [3 x i8], [1 x i8], %struct.__sbuf, i32, i64 } +%struct.__sbuf = type { i8*, i32 } +%struct.__sFILEX = type opaque + +@.str = private unnamed_addr constant [13 x i8] c"Hello world\0A\00", align 1 +@.str2 = private unnamed_addr constant [3 x i8] c"%s\00", align 1 + +define void @test1(%struct.__sFILE* %stream) nounwind { +; CHECK: define void @test1 +; CHECK: call i32 @"fwrite$UNIX2003" + %call = tail call i32 (%struct.__sFILE*, i8*, ...)* @fprintf(%struct.__sFILE* %stream, i8* getelementptr inbounds ([13 x i8]* @.str, i32 0, i32 0)) nounwind + ret void +} + +define void @test2(%struct.__sFILE* %stream, i8* %str) nounwind ssp { +; CHECK: define void @test2 +; CHECK: call i32 @"fputs$UNIX2003" + %call = tail call i32 (%struct.__sFILE*, i8*, ...)* @fprintf(%struct.__sFILE* %stream, i8* getelementptr inbounds ([3 x i8]* @.str2, i32 0, i32 0), i8* %str) nounwind + ret void +} + +declare i32 @fprintf(%struct.__sFILE*, i8*, ...) nounwind diff --git a/test/Transforms/InstCombine/pow-1.ll b/test/Transforms/InstCombine/pow-1.ll new file mode 100644 index 0000000000000..8a311f0b74c63 --- /dev/null +++ b/test/Transforms/InstCombine/pow-1.ll @@ -0,0 +1,154 @@ +; Test that the pow library call simplifier works correctly. +; +; RUN: opt < %s -instcombine -S | FileCheck %s +; rdar://7251832 + +; NOTE: The readonly attribute on the pow call should be preserved +; in the cases below where pow is transformed into another function call. + +declare float @powf(float, float) nounwind readonly +declare double @pow(double, double) nounwind readonly + +; Check pow(1.0, x) -> 1.0. + +define float @test_simplify1(float %x) { +; CHECK: @test_simplify1 + %retval = call float @powf(float 1.0, float %x) + ret float %retval +; CHECK-NEXT: ret float 1.000000e+00 +} + +define double @test_simplify2(double %x) { +; CHECK: @test_simplify2 + %retval = call double @pow(double 1.0, double %x) + ret double %retval +; CHECK-NEXT: ret double 1.000000e+00 +} + +; Check pow(2.0, x) -> exp2(x). + +define float @test_simplify3(float %x) { +; CHECK: @test_simplify3 + %retval = call float @powf(float 2.0, float %x) +; CHECK-NEXT: [[EXP2F:%[a-z0-9]+]] = call float @exp2f(float %x) [[NUW_RO:#[0-9]+]] + ret float %retval +; CHECK-NEXT: ret float [[EXP2F]] +} + +define double @test_simplify4(double %x) { +; CHECK: @test_simplify4 + %retval = call double @pow(double 2.0, double %x) +; CHECK-NEXT: [[EXP2:%[a-z0-9]+]] = call double @exp2(double %x) [[NUW_RO]] + ret double %retval +; CHECK-NEXT: ret double [[EXP2]] +} + +; Check pow(x, 0.0) -> 1.0. + +define float @test_simplify5(float %x) { +; CHECK: @test_simplify5 + %retval = call float @powf(float %x, float 0.0) + ret float %retval +; CHECK-NEXT: ret float 1.000000e+00 +} + +define double @test_simplify6(double %x) { +; CHECK: @test_simplify6 + %retval = call double @pow(double %x, double 0.0) + ret double %retval +; CHECK-NEXT: ret double 1.000000e+00 +} + +; Check pow(x, 0.5) -> fabs(sqrt(x)), where x != -infinity. + +define float @test_simplify7(float %x) { +; CHECK: @test_simplify7 + %retval = call float @powf(float %x, float 0.5) +; CHECK-NEXT: [[SQRTF:%[a-z0-9]+]] = call float @sqrtf(float %x) [[NUW_RO]] +; CHECK-NEXT: [[FABSF:%[a-z0-9]+]] = call float @fabsf(float [[SQRTF]]) [[NUW_RO]] +; CHECK-NEXT: [[FCMP:%[a-z0-9]+]] = fcmp oeq float %x, 0xFFF0000000000000 +; CHECK-NEXT: [[SELECT:%[a-z0-9]+]] = select i1 [[FCMP]], float 0x7FF0000000000000, float [[FABSF]] + ret float %retval +; CHECK-NEXT: ret float [[SELECT]] +} + +define double @test_simplify8(double %x) { +; CHECK: @test_simplify8 + %retval = call double @pow(double %x, double 0.5) +; CHECK-NEXT: [[SQRT:%[a-z0-9]+]] = call double @sqrt(double %x) [[NUW_RO]] +; CHECK-NEXT: [[FABS:%[a-z0-9]+]] = call double @fabs(double [[SQRT]]) [[NUW_RO]] +; CHECK-NEXT: [[FCMP:%[a-z0-9]+]] = fcmp oeq double %x, 0xFFF0000000000000 +; CHECK-NEXT: [[SELECT:%[a-z0-9]+]] = select i1 [[FCMP]], double 0x7FF0000000000000, double [[FABS]] + ret double %retval +; CHECK-NEXT: ret double [[SELECT]] +} + +; Check pow(-infinity, 0.5) -> +infinity. + +define float @test_simplify9(float %x) { +; CHECK: @test_simplify9 + %retval = call float @powf(float 0xFFF0000000000000, float 0.5) + ret float %retval +; CHECK-NEXT: ret float 0x7FF0000000000000 +} + +define double @test_simplify10(double %x) { +; CHECK: @test_simplify10 + %retval = call double @pow(double 0xFFF0000000000000, double 0.5) + ret double %retval +; CHECK-NEXT: ret double 0x7FF0000000000000 +} + +; Check pow(x, 1.0) -> x. + +define float @test_simplify11(float %x) { +; CHECK: @test_simplify11 + %retval = call float @powf(float %x, float 1.0) + ret float %retval +; CHECK-NEXT: ret float %x +} + +define double @test_simplify12(double %x) { +; CHECK: @test_simplify12 + %retval = call double @pow(double %x, double 1.0) + ret double %retval +; CHECK-NEXT: ret double %x +} + +; Check pow(x, 2.0) -> x*x. + +define float @test_simplify13(float %x) { +; CHECK: @test_simplify13 + %retval = call float @powf(float %x, float 2.0) +; CHECK-NEXT: [[SQUARE:%[a-z0-9]+]] = fmul float %x, %x + ret float %retval +; CHECK-NEXT: ret float [[SQUARE]] +} + +define double @test_simplify14(double %x) { +; CHECK: @test_simplify14 + %retval = call double @pow(double %x, double 2.0) +; CHECK-NEXT: [[SQUARE:%[a-z0-9]+]] = fmul double %x, %x + ret double %retval +; CHECK-NEXT: ret double [[SQUARE]] +} + +; Check pow(x, -1.0) -> 1.0/x. + +define float @test_simplify15(float %x) { +; CHECK: @test_simplify15 + %retval = call float @powf(float %x, float -1.0) +; CHECK-NEXT: [[RECIPROCAL:%[a-z0-9]+]] = fdiv float 1.000000e+00, %x + ret float %retval +; CHECK-NEXT: ret float [[RECIPROCAL]] +} + +define double @test_simplify16(double %x) { +; CHECK: @test_simplify16 + %retval = call double @pow(double %x, double -1.0) +; CHECK-NEXT: [[RECIPROCAL:%[a-z0-9]+]] = fdiv double 1.000000e+00, %x + ret double %retval +; CHECK-NEXT: ret double [[RECIPROCAL]] +} + +; CHECK: attributes [[NUW_RO]] = { nounwind readonly } diff --git a/test/Transforms/InstCombine/pow-2.ll b/test/Transforms/InstCombine/pow-2.ll new file mode 100644 index 0000000000000..af64cda0904a8 --- /dev/null +++ b/test/Transforms/InstCombine/pow-2.ll @@ -0,0 +1,14 @@ +; Test that the pow library call simplifier works correctly. +; +; RUN: opt < %s -instcombine -S | FileCheck %s + +declare float @pow(double, double) + +; Check that pow functions with the wrong prototype aren't simplified. + +define float @test_no_simplify1(double %x) { +; CHECK: @test_no_simplify1 + %retval = call float @pow(double 1.0, double %x) +; CHECK-NEXT: call float @pow(double 1.000000e+00, double %x) + ret float %retval +} diff --git a/test/Transforms/InstCombine/pr12338.ll b/test/Transforms/InstCombine/pr12338.ll index 2b5c8f8a74edb..d34600f0fa587 100644 --- a/test/Transforms/InstCombine/pr12338.ll +++ b/test/Transforms/InstCombine/pr12338.ll @@ -1,24 +1,24 @@ ; RUN: opt < %s -instcombine -S | FileCheck %s -define void @entry() nounwind { -entry: - br label %for.cond - -for.cond: +define void @entry() nounwind { +entry: + br label %for.cond + +for.cond: %local = phi <1 x i32> [ , %entry ], [ %phi2, %cond.end47 ] -; CHECK: sub <1 x i32> , %local - %phi3 = sub <1 x i32> zeroinitializer, %local - br label %cond.end - -cond.false: - br label %cond.end - -cond.end: - %cond = phi <1 x i32> [ %phi3, %for.cond ], [ undef, %cond.false ] - br label %cond.end47 - -cond.end47: - %sum = add <1 x i32> %cond, - %phi2 = sub <1 x i32> zeroinitializer, %sum - br label %for.cond -} +; CHECK: sub <1 x i32> , %local + %phi3 = sub <1 x i32> zeroinitializer, %local + br label %cond.end + +cond.false: + br label %cond.end + +cond.end: + %cond = phi <1 x i32> [ %phi3, %for.cond ], [ undef, %cond.false ] + br label %cond.end47 + +cond.end47: + %sum = add <1 x i32> %cond, + %phi2 = sub <1 x i32> zeroinitializer, %sum + br label %for.cond +} diff --git a/test/Transforms/InstCombine/printf-1.ll b/test/Transforms/InstCombine/printf-1.ll new file mode 100644 index 0000000000000..3a910ea437b72 --- /dev/null +++ b/test/Transforms/InstCombine/printf-1.ll @@ -0,0 +1,119 @@ +; Test that the printf library call simplifier works correctly. +; +; RUN: opt < %s -instcombine -S | FileCheck %s +; RUN: opt < %s -mtriple xcore-xmos-elf -instcombine -S | FileCheck %s -check-prefix=IPRINTF + +target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128" + +@hello_world = constant [13 x i8] c"hello world\0A\00" +@h = constant [2 x i8] c"h\00" +@percent = constant [2 x i8] c"%\00" +@percent_c = constant [3 x i8] c"%c\00" +@percent_d = constant [3 x i8] c"%d\00" +@percent_f = constant [3 x i8] c"%f\00" +@percent_s = constant [4 x i8] c"%s\0A\00" +@empty = constant [1 x i8] c"\00" +; CHECK: [[STR:@[a-z0-9]+]] = private unnamed_addr constant [12 x i8] c"hello world\00" + +declare i32 @printf(i8*, ...) + +; Check printf("") -> noop. + +define void @test_simplify1() { +; CHECK: @test_simplify1 + %fmt = getelementptr [1 x i8]* @empty, i32 0, i32 0 + call i32 (i8*, ...)* @printf(i8* %fmt) + ret void +; CHECK-NEXT: ret void +} + +; Check printf("x") -> putchar('x'), even for '%'. + +define void @test_simplify2() { +; CHECK: @test_simplify2 + %fmt = getelementptr [2 x i8]* @h, i32 0, i32 0 + call i32 (i8*, ...)* @printf(i8* %fmt) +; CHECK-NEXT: call i32 @putchar(i32 104) + ret void +; CHECK-NEXT: ret void +} + +define void @test_simplify3() { +; CHECK: @test_simplify3 + %fmt = getelementptr [2 x i8]* @percent, i32 0, i32 0 + call i32 (i8*, ...)* @printf(i8* %fmt) +; CHECK-NEXT: call i32 @putchar(i32 37) + ret void +; CHECK-NEXT: ret void +} + +; Check printf("foo\n") -> puts("foo"). + +define void @test_simplify4() { +; CHECK: @test_simplify4 + %fmt = getelementptr [13 x i8]* @hello_world, i32 0, i32 0 + call i32 (i8*, ...)* @printf(i8* %fmt) +; CHECK-NEXT: call i32 @puts(i8* getelementptr inbounds ([12 x i8]* [[STR]], i32 0, i32 0)) + ret void +; CHECK-NEXT: ret void +} + +; Check printf("%c", chr) -> putchar(chr). + +define void @test_simplify5() { +; CHECK: @test_simplify5 + %fmt = getelementptr [3 x i8]* @percent_c, i32 0, i32 0 + call i32 (i8*, ...)* @printf(i8* %fmt, i8 104) +; CHECK-NEXT: call i32 @putchar(i32 104) + ret void +; CHECK-NEXT: ret void +} + +; Check printf("%s\n", str) -> puts(str). + +define void @test_simplify6() { +; CHECK: @test_simplify6 + %fmt = getelementptr [4 x i8]* @percent_s, i32 0, i32 0 + %str = getelementptr [13 x i8]* @hello_world, i32 0, i32 0 + call i32 (i8*, ...)* @printf(i8* %fmt, i8* %str) +; CHECK-NEXT: call i32 @puts(i8* getelementptr inbounds ([13 x i8]* @hello_world, i32 0, i32 0)) + ret void +; CHECK-NEXT: ret void +} + +; Check printf(format, ...) -> iprintf(format, ...) if no floating point. + +define void @test_simplify7() { +; CHECK-IPRINTF: @test_simplify7 + %fmt = getelementptr [3 x i8]* @percent_d, i32 0, i32 0 + call i32 (i8*, ...)* @printf(i8* %fmt, i32 187) +; CHECK-NEXT-IPRINTF: call i32 (i8*, ...)* @iprintf(i8* getelementptr inbounds ([3 x i8]* @percent_d, i32 0, i32 0), i32 187) + ret void +; CHECK-NEXT-IPRINTF: ret void +} + +define void @test_no_simplify1() { +; CHECK-IPRINTF: @test_no_simplify1 + %fmt = getelementptr [3 x i8]* @percent_f, i32 0, i32 0 + call i32 (i8*, ...)* @printf(i8* %fmt, double 1.87) +; CHECK-NEXT-IPRINTF: call i32 (i8*, ...)* @printf(i8* getelementptr inbounds ([3 x i8]* @percent_f, i32 0, i32 0), double 1.870000e+00) + ret void +; CHECK-NEXT-IPRINTF: ret void +} + +define void @test_no_simplify2(i8* %fmt, double %d) { +; CHECK: @test_no_simplify2 + call i32 (i8*, ...)* @printf(i8* %fmt, double %d) +; CHECK-NEXT: call i32 (i8*, ...)* @printf(i8* %fmt, double %d) + ret void +; CHECK-NEXT: ret void +} + +define i32 @test_no_simplify3() { +; CHECK: @test_no_simplify3 + %fmt = getelementptr [2 x i8]* @h, i32 0, i32 0 + %ret = call i32 (i8*, ...)* @printf(i8* %fmt) +; CHECK-NEXT: call i32 (i8*, ...)* @printf(i8* getelementptr inbounds ([2 x i8]* @h, i32 0, i32 0)) + ret i32 %ret +; CHECK-NEXT: ret i32 %ret +} diff --git a/test/Transforms/InstCombine/printf-2.ll b/test/Transforms/InstCombine/printf-2.ll new file mode 100644 index 0000000000000..466ee1c757705 --- /dev/null +++ b/test/Transforms/InstCombine/printf-2.ll @@ -0,0 +1,41 @@ +; Test that the printf library call simplifier works correctly. +; +; RUN: opt < %s -instcombine -S | FileCheck %s + +target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128" + +@hello_world = constant [13 x i8] c"hello world\0A\00" +@h = constant [2 x i8] c"h\00" +@percent_s = constant [4 x i8] c"%s\0A\00" + +declare void @printf(i8*, ...) + +; Check simplification of printf with void return type. + +define void @test_simplify1() { +; CHECK: @test_simplify1 + %fmt = getelementptr [2 x i8]* @h, i32 0, i32 0 + call void (i8*, ...)* @printf(i8* %fmt) +; CHECK-NEXT: call i32 @putchar(i32 104) + ret void +; CHECK-NEXT: ret void +} + +define void @test_simplify2() { +; CHECK: @test_simplify2 + %fmt = getelementptr [13 x i8]* @hello_world, i32 0, i32 0 + call void (i8*, ...)* @printf(i8* %fmt) +; CHECK-NEXT: call i32 @puts(i8* getelementptr inbounds ([12 x i8]* @str, i32 0, i32 0)) + ret void +; CHECK-NEXT: ret void +} + +define void @test_simplify6() { +; CHECK: @test_simplify6 + %fmt = getelementptr [4 x i8]* @percent_s, i32 0, i32 0 + %str = getelementptr [13 x i8]* @hello_world, i32 0, i32 0 + call void (i8*, ...)* @printf(i8* %fmt, i8* %str) +; CHECK-NEXT: call i32 @puts(i8* getelementptr inbounds ([13 x i8]* @hello_world, i32 0, i32 0)) + ret void +; CHECK-NEXT: ret void +} diff --git a/test/Transforms/InstCombine/ptr-int-cast.ll b/test/Transforms/InstCombine/ptr-int-cast.ll index 9524d449dd8b0..7a6ecff9c0be0 100644 --- a/test/Transforms/InstCombine/ptr-int-cast.ll +++ b/test/Transforms/InstCombine/ptr-int-cast.ll @@ -27,3 +27,34 @@ define i64 @f0(i32 %a0) nounwind { ret i64 %t1 } +define <4 x i32> @test4(<4 x i8*> %arg) nounwind { +; CHECK: @test4 +; CHECK: ptrtoint <4 x i8*> %arg to <4 x i64> +; CHECK: trunc <4 x i64> %1 to <4 x i32> + %p1 = ptrtoint <4 x i8*> %arg to <4 x i32> + ret <4 x i32> %p1 +} + +define <4 x i128> @test5(<4 x i8*> %arg) nounwind { +; CHECK: @test5 +; CHECK: ptrtoint <4 x i8*> %arg to <4 x i64> +; CHECK: zext <4 x i64> %1 to <4 x i128> + %p1 = ptrtoint <4 x i8*> %arg to <4 x i128> + ret <4 x i128> %p1 +} + +define <4 x i8*> @test6(<4 x i32> %arg) nounwind { +; CHECK: @test6 +; CHECK: zext <4 x i32> %arg to <4 x i64> +; CHECK: inttoptr <4 x i64> %1 to <4 x i8*> + %p1 = inttoptr <4 x i32> %arg to <4 x i8*> + ret <4 x i8*> %p1 +} + +define <4 x i8*> @test7(<4 x i128> %arg) nounwind { +; CHECK: @test7 +; CHECK: trunc <4 x i128> %arg to <4 x i64> +; CHECK: inttoptr <4 x i64> %1 to <4 x i8*> + %p1 = inttoptr <4 x i128> %arg to <4 x i8*> + ret <4 x i8*> %p1 +} diff --git a/test/Transforms/InstCombine/puts-1.ll b/test/Transforms/InstCombine/puts-1.ll new file mode 100644 index 0000000000000..ef4e1bbd824c0 --- /dev/null +++ b/test/Transforms/InstCombine/puts-1.ll @@ -0,0 +1,31 @@ +; Test that the puts library call simplifier works correctly. +; +; RUN: opt < %s -instcombine -S | FileCheck %s + +target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128" + +@empty = constant [1 x i8] zeroinitializer + +declare i32 @puts(i8*) + +; Check puts("") -> putchar('\n'). + +define void @test_simplify1() { +; CHECK: @test_simplify1 + %str = getelementptr [1 x i8]* @empty, i32 0, i32 0 + call i32 @puts(i8* %str) +; CHECK-NEXT: call i32 @putchar(i32 10) + ret void +; CHECK-NEXT: ret void +} + +; Don't simplify if the return value is used. + +define i32 @test_no_simplify1() { +; CHECK: @test_no_simplify1 + %str = getelementptr [1 x i8]* @empty, i32 0, i32 0 + %ret = call i32 @puts(i8* %str) +; CHECK-NEXT: call i32 @puts(i8* getelementptr inbounds ([1 x i8]* @empty, i32 0, i32 0)) + ret i32 %ret +; CHECK-NEXT: ret i32 %ret +} diff --git a/test/Transforms/InstCombine/sdiv-1.ll b/test/Transforms/InstCombine/sdiv-1.ll index c46b5eaef4a87..6ab18ac7f8449 100644 --- a/test/Transforms/InstCombine/sdiv-1.ll +++ b/test/Transforms/InstCombine/sdiv-1.ll @@ -1,6 +1,8 @@ -; RUN: opt < %s -instcombine -inline -S | not grep '-715827882' +; RUN: opt < %s -instcombine -inline -S | FileCheck %s ; PR3142 +; CHECK-NOT: -715827882 + define i32 @a(i32 %X) nounwind readnone { entry: %0 = sub i32 0, %X diff --git a/test/Transforms/InstCombine/sext.ll b/test/Transforms/InstCombine/sext.ll index f1987973f4622..968f37c9c129e 100644 --- a/test/Transforms/InstCombine/sext.ll +++ b/test/Transforms/InstCombine/sext.ll @@ -184,3 +184,12 @@ define i32 @test16(i16 %x) nounwind { ; CHECK-NEXT: %ext = sext i16 %sext to i32 ; CHECK-NEXT: ret i32 %ext } + +define i32 @test17(i1 %x) nounwind { + %c1 = sext i1 %x to i32 + %c2 = sub i32 0, %c1 + ret i32 %c2 +; CHECK: @test17 +; CHECK-NEXT: [[TEST17:%.*]] = zext i1 %x to i32 +; CHECK-NEXT: ret i32 [[TEST17]] +} diff --git a/test/Transforms/InstCombine/shift.ll b/test/Transforms/InstCombine/shift.ll index 25e708b7f51d9..41f8aa9ee8124 100644 --- a/test/Transforms/InstCombine/shift.ll +++ b/test/Transforms/InstCombine/shift.ll @@ -523,9 +523,9 @@ entry: %tmp51 = xor i8 %tmp50, %tmp5 %tmp52 = and i8 %tmp51, -128 %tmp53 = lshr i8 %tmp52, 7 -; CHECK: lshr i8 %tmp51, 7 %tmp54 = mul i8 %tmp53, 16 -; CHECK: shl nuw nsw i8 %tmp53, 4 +; CHECK: %0 = shl i8 %tmp4, 2 +; CHECK: %tmp54 = and i8 %0, 16 %tmp55 = xor i8 %tmp54, %tmp51 ; CHECK: ret i8 %tmp551 ret i8 %tmp55 @@ -659,3 +659,89 @@ define i32 @test53(i32 %x) { ; CHECK-NEXT: %B = shl nuw i32 %x, 2 ; CHECK-NEXT: ret i32 %B } + +define i32 @test54(i32 %x) { + %shr2 = lshr i32 %x, 1 + %shl = shl i32 %shr2, 4 + %and = and i32 %shl, 16 + ret i32 %and +; CHECK: @test54 +; CHECK: shl i32 %x, 3 +} + + +define i32 @test55(i32 %x) { + %shr2 = lshr i32 %x, 1 + %shl = shl i32 %shr2, 4 + %or = or i32 %shl, 8 + ret i32 %or +; CHECK: @test55 +; CHECK: shl i32 %x, 3 +} + +define i32 @test56(i32 %x) { + %shr2 = lshr i32 %x, 1 + %shl = shl i32 %shr2, 4 + %or = or i32 %shl, 7 + ret i32 %or +; CHECK: @test56 +; CHECK: shl i32 %shr2, 4 +} + + +define i32 @test57(i32 %x) { + %shr = lshr i32 %x, 1 + %shl = shl i32 %shr, 4 + %and = and i32 %shl, 16 + ret i32 %and +; CHECK: @test57 +; CHECK: shl i32 %x, 3 +} + +define i32 @test58(i32 %x) { + %shr = lshr i32 %x, 1 + %shl = shl i32 %shr, 4 + %or = or i32 %shl, 8 + ret i32 %or +; CHECK: @test58 +; CHECK: shl i32 %x, 3 +} + +define i32 @test59(i32 %x) { + %shr = ashr i32 %x, 1 + %shl = shl i32 %shr, 4 + %or = or i32 %shl, 7 + ret i32 %or +; CHECK: @test59 +; CHECK: %shl = shl i32 %shr1, 4 +} + + +define i32 @test60(i32 %x) { + %shr = ashr i32 %x, 4 + %shl = shl i32 %shr, 1 + %or = or i32 %shl, 1 + ret i32 %or +; CHECK: @test60 +; CHECK: ashr i32 %x, 3 +} + + +define i32 @test61(i32 %x) { + %shr = ashr i32 %x, 4 + %shl = shl i32 %shr, 1 + %or = or i32 %shl, 2 + ret i32 %or +; CHECK: @test61 +; CHECK: ashr i32 %x, 4 +} + +; propagate "exact" trait +define i32 @test62(i32 %x) { + %shr = ashr exact i32 %x, 4 + %shl = shl i32 %shr, 1 + %or = or i32 %shl, 1 + ret i32 %or +; CHECK: @test62 +; CHECK: ashr exact i32 %x, 3 +} diff --git a/test/Transforms/InstCombine/signext.ll b/test/Transforms/InstCombine/signext.ll index ecee9830cd576..5ed1cd5590ae9 100644 --- a/test/Transforms/InstCombine/signext.ll +++ b/test/Transforms/InstCombine/signext.ll @@ -82,6 +82,6 @@ entry: %sub = add i32 %xor, -67108864 ; [#uses=1] ret i32 %sub ; CHECK: @test8 -; CHECK: %shr = ashr i32 %x, 5 -; CHECK: ret i32 %shr +; CHECK: %sub = ashr i32 %x, 5 +; CHECK: ret i32 %sub } diff --git a/test/Transforms/InstCombine/sink_instruction.ll b/test/Transforms/InstCombine/sink_instruction.ll index e521de208f216..5c4019a98df55 100644 --- a/test/Transforms/InstCombine/sink_instruction.ll +++ b/test/Transforms/InstCombine/sink_instruction.ll @@ -1,4 +1,4 @@ -; RUN: opt -instcombine %s -S | FileCheck %s +; RUN: opt -instcombine -S < %s | FileCheck %s ;; This tests that the instructions in the entry blocks are sunk into each ;; arm of the 'if'. diff --git a/test/Transforms/InstCombine/sprintf-1.ll b/test/Transforms/InstCombine/sprintf-1.ll new file mode 100644 index 0000000000000..9b8c8b1b12c7b --- /dev/null +++ b/test/Transforms/InstCombine/sprintf-1.ll @@ -0,0 +1,100 @@ +; Test that the sprintf library call simplifier works correctly. +; +; RUN: opt < %s -instcombine -S | FileCheck %s +; RUN: opt < %s -mtriple xcore-xmos-elf -instcombine -S | FileCheck %s -check-prefix=IPRINTF + +target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128" + +@hello_world = constant [13 x i8] c"hello world\0A\00" +@null = constant [1 x i8] zeroinitializer +@null_hello = constant [7 x i8] c"\00hello\00" +@h = constant [2 x i8] c"h\00" +@percent_c = constant [3 x i8] c"%c\00" +@percent_d = constant [3 x i8] c"%d\00" +@percent_f = constant [3 x i8] c"%f\00" +@percent_s = constant [3 x i8] c"%s\00" + +declare i32 @sprintf(i8*, i8*, ...) + +; Check sprintf(dst, fmt) -> llvm.memcpy(str, fmt, strlen(fmt) + 1, 1). + +define void @test_simplify1(i8* %dst) { +; CHECK: @test_simplify1 + %fmt = getelementptr [13 x i8]* @hello_world, i32 0, i32 0 + call i32 (i8*, i8*, ...)* @sprintf(i8* %dst, i8* %fmt) +; CHECK-NEXT: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dst, i8* getelementptr inbounds ([13 x i8]* @hello_world, i32 0, i32 0), i32 13, i32 1, i1 false) + ret void +; CHECK-NEXT: ret void +} + +define void @test_simplify2(i8* %dst) { +; CHECK: @test_simplify2 + %fmt = getelementptr [1 x i8]* @null, i32 0, i32 0 + call i32 (i8*, i8*, ...)* @sprintf(i8* %dst, i8* %fmt) +; CHECK-NEXT: store i8 0, i8* %dst, align 1 + ret void +; CHECK-NEXT: ret void +} + +define void @test_simplify3(i8* %dst) { +; CHECK: @test_simplify3 + %fmt = getelementptr [7 x i8]* @null_hello, i32 0, i32 0 + call i32 (i8*, i8*, ...)* @sprintf(i8* %dst, i8* %fmt) +; CHECK-NEXT: store i8 0, i8* %dst, align 1 + ret void +; CHECK-NEXT: ret void +} + +; Check sprintf(dst, "%c", chr) -> *(i8*)dst = chr; *((i8*)dst + 1) = 0. + +define void @test_simplify4(i8* %dst) { +; CHECK: @test_simplify4 + %fmt = getelementptr [3 x i8]* @percent_c, i32 0, i32 0 + call i32 (i8*, i8*, ...)* @sprintf(i8* %dst, i8* %fmt, i8 104) +; CHECK-NEXT: store i8 104, i8* %dst, align 1 +; CHECK-NEXT: [[NUL:%[a-z0-9]+]] = getelementptr i8* %dst, i32 1 +; CHECK-NEXT: store i8 0, i8* [[NUL]], align 1 + ret void +; CHECK-NEXT: ret void +} + +; Check sprintf(dst, "%s", str) -> llvm.memcpy(dest, str, strlen(str) + 1, 1). + +define void @test_simplify5(i8* %dst, i8* %str) { +; CHECK: @test_simplify5 + %fmt = getelementptr [3 x i8]* @percent_s, i32 0, i32 0 + call i32 (i8*, i8*, ...)* @sprintf(i8* %dst, i8* %fmt, i8* %str) +; CHECK-NEXT: [[STRLEN:%[a-z0-9]+]] = call i32 @strlen(i8* %str) +; CHECK-NEXT: [[LENINC:%[a-z0-9]+]] = add i32 [[STRLEN]], 1 +; CHECK-NEXT: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dst, i8* %str, i32 [[LENINC]], i32 1, i1 false) + ret void +; CHECK-NEXT: ret void +} + +; Check sprintf(dst, format, ...) -> siprintf(str, format, ...) if no floating. + +define void @test_simplify6(i8* %dst) { +; CHECK-IPRINTF: @test_simplify6 + %fmt = getelementptr [3 x i8]* @percent_d, i32 0, i32 0 + call i32 (i8*, i8*, ...)* @sprintf(i8* %dst, i8* %fmt, i32 187) +; CHECK-NEXT-IPRINTF: call i32 (i8*, i8*, ...)* @siprintf(i8* %dst, i8* getelementptr inbounds ([3 x i8]* @percent_d, i32 0, i32 0), i32 187) + ret void +; CHECK-NEXT-IPRINTF: ret void +} + +define void @test_no_simplify1(i8* %dst) { +; CHECK-IPRINTF: @test_no_simplify1 + %fmt = getelementptr [3 x i8]* @percent_f, i32 0, i32 0 + call i32 (i8*, i8*, ...)* @sprintf(i8* %dst, i8* %fmt, double 1.87) +; CHECK-NEXT-IPRINTF: call i32 (i8*, i8*, ...)* @sprintf(i8* %dst, i8* getelementptr inbounds ([3 x i8]* @percent_f, i32 0, i32 0), double 1.870000e+00) + ret void +; CHECK-NEXT-IPRINTF: ret void +} + +define void @test_no_simplify2(i8* %dst, i8* %fmt, double %d) { +; CHECK: @test_no_simplify2 + call i32 (i8*, i8*, ...)* @sprintf(i8* %dst, i8* %fmt, double %d) +; CHECK-NEXT: call i32 (i8*, i8*, ...)* @sprintf(i8* %dst, i8* %fmt, double %d) + ret void +; CHECK-NEXT: ret void +} diff --git a/test/Transforms/InstCombine/sqrt.ll b/test/Transforms/InstCombine/sqrt.ll index cc78417ebbd68..440b9748518dd 100644 --- a/test/Transforms/InstCombine/sqrt.ll +++ b/test/Transforms/InstCombine/sqrt.ll @@ -1,4 +1,4 @@ -; RUN: opt -S -instcombine %s | FileCheck %s +; RUN: opt -S -instcombine < %s | FileCheck %s define float @test1(float %x) nounwind readnone ssp { entry: diff --git a/test/Transforms/InstCombine/store.ll b/test/Transforms/InstCombine/store.ll index 64460d7a6d610..164ba7632684b 100644 --- a/test/Transforms/InstCombine/store.ll +++ b/test/Transforms/InstCombine/store.ll @@ -83,3 +83,37 @@ Cont: ; CHECK-NEXT: ret void } + +; PR14753 - merging two stores should preserve the TBAA tag. +define void @test6(i32 %n, float* %a, i32* %gi) nounwind uwtable ssp { +entry: + store i32 42, i32* %gi, align 4, !tbaa !0 + br label %for.cond + +for.cond: ; preds = %for.body, %entry + %storemerge = phi i32 [ 0, %entry ], [ %inc, %for.body ] + %0 = load i32* %gi, align 4, !tbaa !0 + %cmp = icmp slt i32 %0, %n + br i1 %cmp, label %for.body, label %for.end + +for.body: ; preds = %for.cond + %idxprom = sext i32 %0 to i64 + %arrayidx = getelementptr inbounds float* %a, i64 %idxprom + store float 0.000000e+00, float* %arrayidx, align 4, !tbaa !3 + %1 = load i32* %gi, align 4, !tbaa !0 + %inc = add nsw i32 %1, 1 + store i32 %inc, i32* %gi, align 4, !tbaa !0 + br label %for.cond + +for.end: ; preds = %for.cond + ret void +; CHECK: @test6 +; CHECK: for.cond: +; CHECK-NEXT: phi i32 [ 42 +; CHECK-NEXT: store i32 %storemerge, i32* %gi, align 4, !tbaa !0 +} + +!0 = metadata !{metadata !"int", metadata !1} +!1 = metadata !{metadata !"omnipotent char", metadata !2} +!2 = metadata !{metadata !"Simple C/C++ TBAA"} +!3 = metadata !{metadata !"float", metadata !1} diff --git a/test/Transforms/InstCombine/strto-1.ll b/test/Transforms/InstCombine/strto-1.ll index 16c0c67970db8..7139972fe0432 100644 --- a/test/Transforms/InstCombine/strto-1.ll +++ b/test/Transforms/InstCombine/strto-1.ll @@ -1,29 +1,29 @@ ; Test that the strto* library call simplifiers works correctly. ; -; RUN: opt < %s -instcombine -S | FileCheck %s +; RUN: opt < %s -instcombine -functionattrs -S | FileCheck %s target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128" declare i64 @strtol(i8* %s, i8** %endptr, i32 %base) -; CHECK: declare i64 @strtol(i8*, i8**, i32) +; CHECK: declare i64 @strtol(i8*, i8** nocapture, i32) declare double @strtod(i8* %s, i8** %endptr, i32 %base) -; CHECK: declare double @strtod(i8*, i8**, i32) +; CHECK: declare double @strtod(i8*, i8** nocapture, i32) declare float @strtof(i8* %s, i8** %endptr, i32 %base) -; CHECK: declare float @strtof(i8*, i8**, i32) +; CHECK: declare float @strtof(i8*, i8** nocapture, i32) declare i64 @strtoul(i8* %s, i8** %endptr, i32 %base) -; CHECK: declare i64 @strtoul(i8*, i8**, i32) +; CHECK: declare i64 @strtoul(i8*, i8** nocapture, i32) declare i64 @strtoll(i8* %s, i8** %endptr, i32 %base) -; CHECK: declare i64 @strtoll(i8*, i8**, i32) +; CHECK: declare i64 @strtoll(i8*, i8** nocapture, i32) declare double @strtold(i8* %s, i8** %endptr) -; CHECK: declare double @strtold(i8*, i8**) +; CHECK: declare double @strtold(i8*, i8** nocapture) declare i64 @strtoull(i8* %s, i8** %endptr, i32 %base) -; CHECK: declare i64 @strtoull(i8*, i8**, i32) +; CHECK: declare i64 @strtoull(i8*, i8** nocapture, i32) define void @test_simplify1(i8* %x, i8** %endptr) { ; CHECK: @test_simplify1 diff --git a/test/Transforms/InstCombine/toascii-1.ll b/test/Transforms/InstCombine/toascii-1.ll new file mode 100644 index 0000000000000..c4a13e2293937 --- /dev/null +++ b/test/Transforms/InstCombine/toascii-1.ll @@ -0,0 +1,59 @@ +; Test that the toascii library call simplifier works correctly. +; +; RUN: opt < %s -instcombine -S | FileCheck %s + +target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128" + +declare i32 @toascii(i32) + +; Check isascii(c) -> c & 0x7f. + +define i32 @test_simplify1() { +; CHECK: @test_simplify1 + %ret = call i32 @toascii(i32 0) + ret i32 %ret +; CHECK-NEXT: ret i32 0 +} + +define i32 @test_simplify2() { +; CHECK: @test_simplify2 + %ret = call i32 @toascii(i32 1) + ret i32 %ret +; CHECK-NEXT: ret i32 1 +} + +define i32 @test_simplify3() { +; CHECK: @test_simplify3 + %ret = call i32 @toascii(i32 127) + ret i32 %ret +; CHECK-NEXT: ret i32 127 +} + +define i32 @test_simplify4() { +; CHECK: @test_simplify4 + %ret = call i32 @toascii(i32 128) + ret i32 %ret +; CHECK-NEXT: ret i32 0 +} + +define i32 @test_simplify5() { +; CHECK: @test_simplify5 + %ret = call i32 @toascii(i32 255) + ret i32 %ret +; CHECK-NEXT: ret i32 127 +} + +define i32 @test_simplify6() { +; CHECK: @test_simplify6 + %ret = call i32 @toascii(i32 256) + ret i32 %ret +; CHECK-NEXT: ret i32 0 +} + +define i32 @test_simplify7(i32 %x) { +; CHECK: @test_simplify7 + %ret = call i32 @toascii(i32 %x) +; CHECK-NEXT: [[AND:%[a-z0-9]+]] = and i32 %x, 127 + ret i32 %ret +; CHECK-NEXT: ret i32 [[AND]] +} diff --git a/test/Transforms/InstCombine/vec_extract_elt.ll b/test/Transforms/InstCombine/vec_extract_elt.ll index 63e4ee2112d81..166066a201bf6 100644 --- a/test/Transforms/InstCombine/vec_extract_elt.ll +++ b/test/Transforms/InstCombine/vec_extract_elt.ll @@ -7,3 +7,13 @@ define i32 @test(float %f) { ret i32 %tmp19 } +define i64 @test2(i64 %in) { + %vec = insertelement <8 x i64> undef, i64 %in, i32 0 + %splat = shufflevector <8 x i64> %vec, <8 x i64> undef, <8 x i32> zeroinitializer + %add = add <8 x i64> %splat, + %scl1 = extractelement <8 x i64> %add, i32 0 + %scl2 = extractelement <8 x i64> %add, i32 0 + %r = add i64 %scl1, %scl2 + ret i64 %r +} + diff --git a/test/Transforms/InstCombine/vector-casts.ll b/test/Transforms/InstCombine/vector-casts.ll index 7bbf53c270f48..2f2990b7b0555 100644 --- a/test/Transforms/InstCombine/vector-casts.ll +++ b/test/Transforms/InstCombine/vector-casts.ll @@ -64,7 +64,8 @@ entry: ; CHECK: @test5 ; CHECK: sext <4 x i1> %cmp to <4 x i32> -; CHECK: sext <4 x i1> %cmp4 to <4 x i32> +; The sext-and pair is canonicalized to a select. +; CHECK: select <4 x i1> %cmp4, <4 x i32> %sext, <4 x i32> zeroinitializer } diff --git a/test/Transforms/InstCombine/vector-type.ll b/test/Transforms/InstCombine/vector-type.ll new file mode 100644 index 0000000000000..59a4bdd19e70a --- /dev/null +++ b/test/Transforms/InstCombine/vector-type.ll @@ -0,0 +1,15 @@ +; The code in InstCombiner::FoldSelectOpOp was calling +; Type::getVectorNumElements without checking first if the type was a vector. + +; RUN: opt < %s -instcombine -S + +define i32 @vselect1(i32 %a.coerce, i32 %b.coerce, i32 %c.coerce) { +entry: + %0 = bitcast i32 %a.coerce to <2 x i16> + %1 = bitcast i32 %b.coerce to <2 x i16> + %2 = bitcast i32 %c.coerce to <2 x i16> + %cmp = icmp sge <2 x i16> %2, zeroinitializer + %or = select <2 x i1> %cmp, <2 x i16> %0, <2 x i16> %1 + %3 = bitcast <2 x i16> %or to i32 + ret i32 %3 +} diff --git a/test/Transforms/InstCombine/vector_gep1.ll b/test/Transforms/InstCombine/vector_gep1.ll index 6523622995622..90ca26212f2a6 100644 --- a/test/Transforms/InstCombine/vector_gep1.ll +++ b/test/Transforms/InstCombine/vector_gep1.ll @@ -1,5 +1,5 @@ -; RUN: opt -instcombine %s -disable-output -; RUN: opt -instsimplify %s -disable-output +; RUN: opt -instcombine -disable-output < %s +; RUN: opt -instsimplify -disable-output < %s target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128" target triple = "x86_64-unknown-linux-gnu" @@ -35,3 +35,8 @@ define <2 x i1> @test5(<2 x i8*> %a) { %B = icmp ult <2 x i8*> %g, zeroinitializer ret <2 x i1> %B } + +define <2 x i32*> @test7(<2 x {i32, i32}*> %a) { + %w = getelementptr <2 x {i32, i32}*> %a, <2 x i32> , <2 x i32> zeroinitializer + ret <2 x i32*> %w +} diff --git a/test/Transforms/InstCombine/xor2.ll b/test/Transforms/InstCombine/xor2.ll index 89f00bd684757..be06d7999d841 100644 --- a/test/Transforms/InstCombine/xor2.ll +++ b/test/Transforms/InstCombine/xor2.ll @@ -51,3 +51,34 @@ define i32 @test4(i32 %A, i32 %B) { ; CHECK: %1 = ashr i32 %A, %B ; CHECK: ret i32 %1 } + +; defect-2 in rdar://12329730 +; (X^C1) >> C2) ^ C3 -> (X>>C2) ^ ((C1>>C2)^C3) +; where the "X" has more than one use +define i32 @test5(i32 %val1) { +test5: + %xor = xor i32 %val1, 1234 + %shr = lshr i32 %xor, 8 + %xor1 = xor i32 %shr, 1 + %add = add i32 %xor1, %xor + ret i32 %add +; CHECK: @test5 +; CHECK: lshr i32 %val1, 8 +; CHECK: ret +} + +; defect-1 in rdar://12329730 +; Simplify (X^Y) -> X or Y in the user's context if we know that +; only bits from X or Y are demanded. +; e.g. the "x ^ 1234" can be optimized into x in the context of "t >> 16". +; Put in other word, t >> 16 -> x >> 16. +; unsigned foo(unsigned x) { unsigned t = x ^ 1234; ; return (t >> 16) + t;} +define i32 @test6(i32 %x) { + %xor = xor i32 %x, 1234 + %shr = lshr i32 %xor, 16 + %add = add i32 %shr, %xor + ret i32 %add +; CHECK: @test6 +; CHECK: lshr i32 %x, 16 +; CHECK: ret +} diff --git a/test/Transforms/InstCombine/zext-bool-add-sub.ll b/test/Transforms/InstCombine/zext-bool-add-sub.ll index 78bcedbbc2e1a..b5310575502bd 100644 --- a/test/Transforms/InstCombine/zext-bool-add-sub.ll +++ b/test/Transforms/InstCombine/zext-bool-add-sub.ll @@ -4,9 +4,9 @@ define i32 @a(i1 zeroext %x, i1 zeroext %y) { entry: ; CHECK: @a -; CHECK: [[TMP1:%.*]] = zext i1 %y to i32 +; CHECK: [[TMP1:%.*]] = sext i1 %y to i32 ; CHECK: [[TMP2:%.*]] = select i1 %x, i32 2, i32 1 -; CHECK-NEXT: sub i32 [[TMP2]], [[TMP1]] +; CHECK-NEXT: add i32 [[TMP2]], [[TMP1]] %conv = zext i1 %x to i32 %conv3 = zext i1 %y to i32 %conv3.neg = sub i32 0, %conv3 diff --git a/test/Transforms/InstSimplify/call-callconv.ll b/test/Transforms/InstSimplify/call-callconv.ll new file mode 100644 index 0000000000000..e475be781db91 --- /dev/null +++ b/test/Transforms/InstSimplify/call-callconv.ll @@ -0,0 +1,48 @@ +; RUN: opt < %s -instcombine -S | FileCheck %s +; Verify that the non-default calling conv doesn't prevent the libcall simplification + +@.str = private unnamed_addr constant [4 x i8] c"abc\00", align 1 + +define arm_aapcscc i32 @_abs(i32 %i) nounwind readnone { +; CHECK: _abs + %call = tail call arm_aapcscc i32 @abs(i32 %i) nounwind readnone + ret i32 %call +; CHECK: %[[ISPOS:.*]] = icmp sgt i32 %i, -1 +; CHECK: %[[NEG:.*]] = sub i32 0, %i +; CHECK: %[[RET:.*]] = select i1 %[[ISPOS]], i32 %i, i32 %[[NEG]] +; CHECK: ret i32 %[[RET]] +} + +declare arm_aapcscc i32 @abs(i32) nounwind readnone + +define arm_aapcscc i32 @_labs(i32 %i) nounwind readnone { +; CHECK: _labs + %call = tail call arm_aapcscc i32 @labs(i32 %i) nounwind readnone + ret i32 %call +; CHECK: %[[ISPOS:.*]] = icmp sgt i32 %i, -1 +; CHECK: %[[NEG:.*]] = sub i32 0, %i +; CHECK: %[[RET:.*]] = select i1 %[[ISPOS]], i32 %i, i32 %[[NEG]] +; CHECK: ret i32 %[[RET]] +} + +declare arm_aapcscc i32 @labs(i32) nounwind readnone + +define arm_aapcscc i32 @_strlen1() { +; CHECK: _strlen1 + %call = tail call arm_aapcscc i32 @strlen(i8* getelementptr inbounds ([4 x i8]* @.str, i32 0, i32 0)) + ret i32 %call +; CHECK: ret i32 3 +} + +declare arm_aapcscc i32 @strlen(i8*) + +define arm_aapcscc zeroext i1 @_strlen2(i8* %str) { +; CHECK: _strlen2 + %call = tail call arm_aapcscc i32 @strlen(i8* %str) + %cmp = icmp ne i32 %call, 0 + ret i1 %cmp + +; CHECK: %[[STRLENFIRST:.*]] = load i8* %str +; CHECK: %[[CMP:.*]] = icmp ne i8 %[[STRLENFIRST]], 0 +; CHECK: ret i1 %[[CMP]] +} diff --git a/test/Transforms/InstSimplify/call.ll b/test/Transforms/InstSimplify/call.ll new file mode 100644 index 0000000000000..cf2f8476763fd --- /dev/null +++ b/test/Transforms/InstSimplify/call.ll @@ -0,0 +1,103 @@ +; RUN: opt < %s -instsimplify -S | FileCheck %s + +declare {i8, i1} @llvm.uadd.with.overflow.i8(i8 %a, i8 %b) + +define i1 @test_uadd1() { +; CHECK: @test_uadd1 + %x = call {i8, i1} @llvm.uadd.with.overflow.i8(i8 254, i8 3) + %overflow = extractvalue {i8, i1} %x, 1 + ret i1 %overflow +; CHECK-NEXT: ret i1 true +} + +define i8 @test_uadd2() { +; CHECK: @test_uadd2 + %x = call {i8, i1} @llvm.uadd.with.overflow.i8(i8 254, i8 44) + %result = extractvalue {i8, i1} %x, 0 + ret i8 %result +; CHECK-NEXT: ret i8 42 +} + +declare i256 @llvm.cttz.i256(i256 %src, i1 %is_zero_undef) + +define i256 @test_cttz() { +; CHECK: @test_cttz + %x = call i256 @llvm.cttz.i256(i256 10, i1 false) + ret i256 %x +; CHECK-NEXT: ret i256 1 +} + +declare i256 @llvm.ctpop.i256(i256 %src) + +define i256 @test_ctpop() { +; CHECK: @test_ctpop + %x = call i256 @llvm.ctpop.i256(i256 10) + ret i256 %x +; CHECK-NEXT: ret i256 2 +} + +; Test a non-intrinsic that we know about as a library call. +declare float @fabs(float %x) + +define float @test_fabs_libcall() { +; CHECK: @test_fabs_libcall + + %x = call float @fabs(float -42.0) +; This is still a real function call, so instsimplify won't nuke it -- other +; passes have to do that. +; CHECK-NEXT: call float @fabs + + ret float %x +; CHECK-NEXT: ret float 4.2{{0+}}e+01 +} + + +declare float @llvm.fabs.f32(float) nounwind readnone +declare float @llvm.floor.f32(float) nounwind readnone +declare float @llvm.ceil.f32(float) nounwind readnone +declare float @llvm.trunc.f32(float) nounwind readnone +declare float @llvm.rint.f32(float) nounwind readnone +declare float @llvm.nearbyint.f32(float) nounwind readnone + +; Test idempotent intrinsics +define float @test_idempotence(float %a) { +; CHECK: @test_idempotence + +; CHECK: fabs +; CHECK-NOT: fabs + %a0 = call float @llvm.fabs.f32(float %a) + %a1 = call float @llvm.fabs.f32(float %a0) + +; CHECK: floor +; CHECK-NOT: floor + %b0 = call float @llvm.floor.f32(float %a) + %b1 = call float @llvm.floor.f32(float %b0) + +; CHECK: ceil +; CHECK-NOT: ceil + %c0 = call float @llvm.ceil.f32(float %a) + %c1 = call float @llvm.ceil.f32(float %c0) + +; CHECK: trunc +; CHECK-NOT: trunc + %d0 = call float @llvm.trunc.f32(float %a) + %d1 = call float @llvm.trunc.f32(float %d0) + +; CHECK: rint +; CHECK-NOT: rint + %e0 = call float @llvm.rint.f32(float %a) + %e1 = call float @llvm.rint.f32(float %e0) + +; CHECK: nearbyint +; CHECK-NOT: nearbyint + %f0 = call float @llvm.nearbyint.f32(float %a) + %f1 = call float @llvm.nearbyint.f32(float %f0) + + %r0 = fadd float %a1, %b1 + %r1 = fadd float %r0, %c1 + %r2 = fadd float %r1, %d1 + %r3 = fadd float %r2, %e1 + %r4 = fadd float %r3, %f1 + + ret float %r4 +} diff --git a/test/Transforms/InstSimplify/compare.ll b/test/Transforms/InstSimplify/compare.ll index ce2bb799c813a..b764c761cfb20 100644 --- a/test/Transforms/InstSimplify/compare.ll +++ b/test/Transforms/InstSimplify/compare.ll @@ -165,6 +165,46 @@ entry: ret i1 %cmp } +define i1 @gep13(i8* %ptr) { +; CHECK: @gep13 +; We can prove this GEP is non-null because it is inbounds. + %x = getelementptr inbounds i8* %ptr, i32 1 + %cmp = icmp eq i8* %x, null + ret i1 %cmp +; CHECK-NEXT: ret i1 false +} + +define i1 @gep14({ {}, i8 }* %ptr) { +; CHECK: @gep14 +; We can't simplify this because the offset of one in the GEP actually doesn't +; move the pointer. + %x = getelementptr inbounds { {}, i8 }* %ptr, i32 0, i32 1 + %cmp = icmp eq i8* %x, null + ret i1 %cmp +; CHECK-NOT: ret i1 false +} + +define i1 @gep15({ {}, [4 x {i8, i8}]}* %ptr, i32 %y) { +; CHECK: @gep15 +; We can prove this GEP is non-null even though there is a user value, as we +; would necessarily violate inbounds on one side or the other. + %x = getelementptr inbounds { {}, [4 x {i8, i8}]}* %ptr, i32 0, i32 1, i32 %y, i32 1 + %cmp = icmp eq i8* %x, null + ret i1 %cmp +; CHECK-NEXT: ret i1 false +} + +define i1 @gep16(i8* %ptr, i32 %a) { +; CHECK: @gep16 +; We can prove this GEP is non-null because it is inbounds and because we know +; %b is non-zero even though we don't know its value. + %b = or i32 %a, 1 + %x = getelementptr inbounds i8* %ptr, i32 %b + %cmp = icmp eq i8* %x, null + ret i1 %cmp +; CHECK-NEXT: ret i1 false +} + define i1 @zext(i32 %x) { ; CHECK: @zext %e1 = zext i32 %x to i64 @@ -607,3 +647,49 @@ unreachableblock: %Y = icmp eq i32* %X, null ret i1 %Y } + +; It's not valid to fold a comparison of an argument with an alloca, even though +; that's tempting. An argument can't *alias* an alloca, however the aliasing rule +; relies on restrictions against guessing an object's address and dereferencing. +; There are no restrictions against guessing an object's address and comparing. + +define i1 @alloca_argument_compare(i64* %arg) { + %alloc = alloca i64 + %cmp = icmp eq i64* %arg, %alloc + ret i1 %cmp + ; CHECK: alloca_argument_compare + ; CHECK: ret i1 %cmp +} + +; As above, but with the operands reversed. + +define i1 @alloca_argument_compare_swapped(i64* %arg) { + %alloc = alloca i64 + %cmp = icmp eq i64* %alloc, %arg + ret i1 %cmp + ; CHECK: alloca_argument_compare_swapped + ; CHECK: ret i1 %cmp +} + +; Don't assume that a noalias argument isn't equal to a global variable's +; address. This is an example where AliasAnalysis' NoAlias concept is +; different from actual pointer inequality. + +@y = external global i32 +define zeroext i1 @external_compare(i32* noalias %x) { + %cmp = icmp eq i32* %x, @y + ret i1 %cmp + ; CHECK: external_compare + ; CHECK: ret i1 %cmp +} + +define i1 @alloca_gep(i64 %a, i64 %b) { +; CHECK: @alloca_gep +; We can prove this GEP is non-null because it is inbounds and the pointer +; is non-null. + %strs = alloca [1000 x [1001 x i8]], align 16 + %x = getelementptr inbounds [1000 x [1001 x i8]]* %strs, i64 0, i64 %a, i64 %b + %cmp = icmp eq i8* %x, null + ret i1 %cmp +; CHECK-NEXT: ret i1 false +} diff --git a/test/Transforms/InstSimplify/fast-math.ll b/test/Transforms/InstSimplify/fast-math.ll new file mode 100644 index 0000000000000..154b96739791c --- /dev/null +++ b/test/Transforms/InstSimplify/fast-math.ll @@ -0,0 +1,107 @@ +; RUN: opt < %s -instsimplify -S | FileCheck %s + +;; x * 0 ==> 0 when no-nans and no-signed-zero +; CHECK: mul_zero_1 +define float @mul_zero_1(float %a) { + %b = fmul nsz nnan float %a, 0.0 +; CHECK: ret float 0.0 + ret float %b +} +; CHECK: mul_zero_2 +define float @mul_zero_2(float %a) { + %b = fmul fast float 0.0, %a +; CHECK: ret float 0.0 + ret float %b +} + +;; x * 0 =/=> 0 when there could be nans or -0 +; CHECK: no_mul_zero_1 +define float @no_mul_zero_1(float %a) { + %b = fmul nsz float %a, 0.0 +; CHECK: ret float %b + ret float %b +} +; CHECK: no_mul_zero_2 +define float @no_mul_zero_2(float %a) { + %b = fmul nnan float %a, 0.0 +; CHECK: ret float %b + ret float %b +} +; CHECK: no_mul_zero_3 +define float @no_mul_zero_3(float %a) { + %b = fmul float %a, 0.0 +; CHECK: ret float %b + ret float %b +} + +; fadd [nnan ninf] X, (fsub [nnan ninf] 0, X) ==> 0 +; where nnan and ninf have to occur at least once somewhere in this +; expression +; CHECK: fadd_fsub_0 +define float @fadd_fsub_0(float %a) { +; X + -X ==> 0 + %t1 = fsub nnan ninf float 0.0, %a + %zero1 = fadd nnan ninf float %t1, %a + + %t2 = fsub nnan float 0.0, %a + %zero2 = fadd ninf float %t2, %a + + %t3 = fsub nnan ninf float 0.0, %a + %zero3 = fadd float %t3, %a + + %t4 = fsub float 0.0, %a + %zero4 = fadd nnan ninf float %t4, %a + +; Dont fold this +; CHECK: %nofold = fsub float 0.0 + %nofold = fsub float 0.0, %a +; CHECK: %no_zero = fadd nnan float %nofold, %a + %no_zero = fadd nnan float %nofold, %a + +; Coalesce the folded zeros + %zero5 = fadd float %zero1, %zero2 + %zero6 = fadd float %zero3, %zero4 + %zero7 = fadd float %zero5, %zero6 + +; Should get folded + %ret = fadd nsz float %no_zero, %zero7 + +; CHECK: ret float %no_zero + ret float %ret +} + +; fsub nnan ninf x, x ==> 0.0 +; CHECK: @fsub_x_x +define float @fsub_x_x(float %a) { +; X - X ==> 0 + %zero1 = fsub nnan ninf float %a, %a + +; Dont fold +; CHECK: %no_zero1 = fsub + %no_zero1 = fsub ninf float %a, %a +; CHECK: %no_zero2 = fsub + %no_zero2 = fsub nnan float %a, %a +; CHECK: %no_zero = fadd + %no_zero = fadd float %no_zero1, %no_zero2 + +; Should get folded + %ret = fadd nsz float %no_zero, %zero1 + +; CHECK: ret float %no_zero + ret float %ret +} + +; fadd nsz X, 0 ==> X +; CHECK: @nofold_fadd_x_0 +define float @nofold_fadd_x_0(float %a) { +; Dont fold +; CHECK: %no_zero1 = fadd + %no_zero1 = fadd ninf float %a, 0.0 +; CHECK: %no_zero2 = fadd + %no_zero2 = fadd nnan float %a, 0.0 +; CHECK: %no_zero = fadd + %no_zero = fadd float %no_zero1, %no_zero2 + +; CHECK: ret float %no_zero + ret float %no_zero +} diff --git a/test/Transforms/InstSimplify/floating-point-arithmetic.ll b/test/Transforms/InstSimplify/floating-point-arithmetic.ll new file mode 100644 index 0000000000000..f9c364cade36f --- /dev/null +++ b/test/Transforms/InstSimplify/floating-point-arithmetic.ll @@ -0,0 +1,35 @@ +; RUN: opt < %s -instsimplify -S | FileCheck %s + +; fsub 0, (fsub 0, X) ==> X +; CHECK: @fsub_0_0_x +define float @fsub_0_0_x(float %a) { + %t1 = fsub float -0.0, %a + %ret = fsub float -0.0, %t1 + +; CHECK: ret float %a + ret float %ret +} + +; fsub X, 0 ==> X +; CHECK: @fsub_x_0 +define float @fsub_x_0(float %a) { + %ret = fsub float %a, 0.0 +; CHECK ret float %a + ret float %ret +} + +; fadd X, -0 ==> X +; CHECK: @fadd_x_n0 +define float @fadd_x_n0(float %a) { + %ret = fadd float %a, -0.0 +; CHECK ret float %a + ret float %ret +} + +; fmul X, 1.0 ==> X +; CHECK: @fmul_X_1 +define double @fmul_X_1(double %a) { + %b = fmul double 1.000000e+00, %a ; [#uses=1] + ; CHECK: ret double %a + ret double %b +} diff --git a/test/Transforms/InstSimplify/past-the-end.ll b/test/Transforms/InstSimplify/past-the-end.ll new file mode 100644 index 0000000000000..075da4a24be08 --- /dev/null +++ b/test/Transforms/InstSimplify/past-the-end.ll @@ -0,0 +1,77 @@ +; RUN: opt < %s -instsimplify -S | FileCheck %s +target datalayout = "p:32:32" + +; Check some past-the-end subtleties. + +@opte_a = global i32 0 +@opte_b = global i32 0 + +; Comparing base addresses of two distinct globals. Never equal. + +define zeroext i1 @no_offsets() { + %t = icmp eq i32* @opte_a, @opte_b + ret i1 %t + ; CHECK: no_offsets( + ; CHECK: ret i1 false +} + +; Comparing past-the-end addresses of two distinct globals. Never equal. + +define zeroext i1 @both_past_the_end() { + %x = getelementptr i32* @opte_a, i32 1 + %y = getelementptr i32* @opte_b, i32 1 + %t = icmp eq i32* %x, %y + ret i1 %t + ; CHECK: both_past_the_end( + ; CHECK-NOT: ret i1 true + ; TODO: refine this +} + +; Comparing past-the-end addresses of one global to the base address +; of another. Can't fold this. + +define zeroext i1 @just_one_past_the_end() { + %x = getelementptr i32* @opte_a, i32 1 + %t = icmp eq i32* %x, @opte_b + ret i1 %t + ; CHECK: just_one_past_the_end( + ; CHECK: ret i1 icmp eq (i32* getelementptr inbounds (i32* @opte_a, i32 1), i32* @opte_b) +} + +; Comparing base addresses of two distinct allocas. Never equal. + +define zeroext i1 @no_alloca_offsets() { + %m = alloca i32 + %n = alloca i32 + %t = icmp eq i32* %m, %n + ret i1 %t + ; CHECK: no_alloca_offsets( + ; CHECK: ret i1 false +} + +; Comparing past-the-end addresses of two distinct allocas. Never equal. + +define zeroext i1 @both_past_the_end_alloca() { + %m = alloca i32 + %n = alloca i32 + %x = getelementptr i32* %m, i32 1 + %y = getelementptr i32* %n, i32 1 + %t = icmp eq i32* %x, %y + ret i1 %t + ; CHECK: both_past_the_end_alloca( + ; CHECK-NOT: ret i1 true + ; TODO: refine this +} + +; Comparing past-the-end addresses of one alloca to the base address +; of another. Can't fold this. + +define zeroext i1 @just_one_past_the_end_alloca() { + %m = alloca i32 + %n = alloca i32 + %x = getelementptr i32* %m, i32 1 + %t = icmp eq i32* %x, %n + ret i1 %t + ; CHECK: just_one_past_the_end_alloca( + ; CHECK: ret i1 %t +} diff --git a/test/Transforms/InstSimplify/ptr_diff.ll b/test/Transforms/InstSimplify/ptr_diff.ll index 1eb1fd4c097e3..8b4aa796013bc 100644 --- a/test/Transforms/InstSimplify/ptr_diff.ll +++ b/test/Transforms/InstSimplify/ptr_diff.ll @@ -46,3 +46,33 @@ define i64 @ptrdiff3(i8* %ptr) { %diff = sub i64 %last.int, %first.int ret i64 %diff } + +define <4 x i32> @ptrdiff4(<4 x i8*> %arg) nounwind { +; Handle simple cases of vectors of pointers. +; CHECK: @ptrdiff4 +; CHECK: ret <4 x i32> zeroinitializer + %p1 = ptrtoint <4 x i8*> %arg to <4 x i32> + %bc = bitcast <4 x i8*> %arg to <4 x i32*> + %p2 = ptrtoint <4 x i32*> %bc to <4 x i32> + %sub = sub <4 x i32> %p1, %p2 + ret <4 x i32> %sub +} + +%struct.ham = type { i32, [2 x [2 x i32]] } + +@global = internal global %struct.ham zeroinitializer, align 4 + +define i32 @ptrdiff5() nounwind { +bb: + %tmp = getelementptr inbounds %struct.ham* @global, i32 0, i32 1 + %tmp1 = getelementptr inbounds [2 x [2 x i32]]* %tmp, i32 0, i32 0 + %tmp2 = bitcast [2 x i32]* %tmp1 to i32* + %tmp3 = ptrtoint i32* %tmp2 to i32 + %tmp4 = getelementptr inbounds %struct.ham* @global, i32 0, i32 1 + %tmp5 = getelementptr inbounds [2 x [2 x i32]]* %tmp4, i32 0, i32 0 + %tmp6 = ptrtoint [2 x i32]* %tmp5 to i32 + %tmp7 = sub i32 %tmp3, %tmp6 + ret i32 %tmp7 +; CHECK: @ptrdiff5 +; CHECK: ret i32 0 +} diff --git a/test/Transforms/InstSimplify/vector_gep.ll b/test/Transforms/InstSimplify/vector_gep.ll index f65260e00f546..5ac1ddef64f8a 100644 --- a/test/Transforms/InstSimplify/vector_gep.ll +++ b/test/Transforms/InstSimplify/vector_gep.ll @@ -1,4 +1,4 @@ -;RUN: opt -instsimplify %s -disable-output +;RUN: opt -instsimplify -disable-output < %s declare void @helper(<2 x i8*>) define void @test(<2 x i8*> %a) { %A = getelementptr <2 x i8*> %a, <2 x i32> diff --git a/test/Transforms/JumpThreading/basic.ll b/test/Transforms/JumpThreading/basic.ll index 46271379bd0d4..fe3dc77c9c13f 100644 --- a/test/Transforms/JumpThreading/basic.ll +++ b/test/Transforms/JumpThreading/basic.ll @@ -1,4 +1,4 @@ -; RUN: opt %s -jump-threading -S | FileCheck %s +; RUN: opt -jump-threading -S < %s | FileCheck %s declare i32 @f1() declare i32 @f2() @@ -476,3 +476,41 @@ exit1: ; CHECK: } } +; In this test we check that block duplication is inhibited by the presence +; of a function with the 'noduplicate' attribute. + +declare void @g() +declare void @j() +declare void @k() + +; CHECK: define void @h(i32 %p) { +define void @h(i32 %p) { + %x = icmp ult i32 %p, 5 + br i1 %x, label %l1, label %l2 + +l1: + call void @j() + br label %l3 + +l2: + call void @k() + br label %l3 + +l3: +; CHECK: call void @g() [[NOD:#[0-9]+]] +; CHECK-NOT: call void @g() [[NOD]] + call void @g() noduplicate + %y = icmp ult i32 %p, 5 + br i1 %y, label %l4, label %l5 + +l4: + call void @j() + ret void + +l5: + call void @k() + ret void +; CHECK: } +} + +; CHECK: attributes [[NOD]] = { noduplicate } diff --git a/test/Transforms/JumpThreading/degenerate-phi.ll b/test/Transforms/JumpThreading/degenerate-phi.ll index 35d9fdec42816..2905b43af72cc 100644 --- a/test/Transforms/JumpThreading/degenerate-phi.ll +++ b/test/Transforms/JumpThreading/degenerate-phi.ll @@ -1,4 +1,4 @@ -; RUN: opt -jump-threading -disable-output %s +; RUN: opt -jump-threading -disable-output < %s ; PR9112 ; This is actually a test for value tracking. Jump threading produces diff --git a/test/Transforms/JumpThreading/or-undef.ll b/test/Transforms/JumpThreading/or-undef.ll index 6e359925b6c6a..6311b6df43734 100644 --- a/test/Transforms/JumpThreading/or-undef.ll +++ b/test/Transforms/JumpThreading/or-undef.ll @@ -1,4 +1,4 @@ -; RUN: opt -jump-threading -S %s | FileCheck %s +; RUN: opt -jump-threading -S < %s | FileCheck %s ; rdar://7620633 target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64" diff --git a/test/Transforms/LICM/2003-12-11-SinkingToPHI.ll b/test/Transforms/LICM/2003-12-11-SinkingToPHI.ll index fe8d445313228..2bf26041626c4 100644 --- a/test/Transforms/LICM/2003-12-11-SinkingToPHI.ll +++ b/test/Transforms/LICM/2003-12-11-SinkingToPHI.ll @@ -1,4 +1,4 @@ -; RUN: opt < %s -licm | lli %defaultjit +; RUN: opt < %s -licm | lli -force-interpreter define i32 @main() { entry: diff --git a/test/Transforms/LICM/2011-07-06-Alignment.ll b/test/Transforms/LICM/2011-07-06-Alignment.ll index f97b7010bc029..569231489fecf 100644 --- a/test/Transforms/LICM/2011-07-06-Alignment.ll +++ b/test/Transforms/LICM/2011-07-06-Alignment.ll @@ -1,4 +1,4 @@ -; RUN: opt -licm -S %s | FileCheck %s +; RUN: opt -licm -S < %s | FileCheck %s @A = common global [1024 x float] zeroinitializer, align 4 diff --git a/test/Transforms/LICM/crash.ll b/test/Transforms/LICM/crash.ll index de41d008a746b..b43477a56df57 100644 --- a/test/Transforms/LICM/crash.ll +++ b/test/Transforms/LICM/crash.ll @@ -1,4 +1,4 @@ -; RUN: opt -licm %s -disable-output +; RUN: opt -licm -disable-output < %s target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64" target triple = "x86_64-apple-darwin10.0.0" diff --git a/test/Transforms/LICM/hoist-invariant-load.ll b/test/Transforms/LICM/hoist-invariant-load.ll index f9fc551df358c..1ba94d6b489ce 100644 --- a/test/Transforms/LICM/hoist-invariant-load.ll +++ b/test/Transforms/LICM/hoist-invariant-load.ll @@ -1,3 +1,4 @@ +; REQUIRES: asserts ; RUN: opt < %s -licm -stats -S 2>&1 | grep "1 licm" @"\01L_OBJC_METH_VAR_NAME_" = internal global [4 x i8] c"foo\00", section "__TEXT,__objc_methname,cstring_literals", align 1 diff --git a/test/Transforms/LICM/hoisting.ll b/test/Transforms/LICM/hoisting.ll index 98f93345e3c3a..1ca377eb4a99c 100644 --- a/test/Transforms/LICM/hoisting.ll +++ b/test/Transforms/LICM/hoisting.ll @@ -90,3 +90,29 @@ for.end: ; preds = %for.body declare void @foo_may_call_exit(i32) +; PR14854 +; CHECK: @test5 +; CHECK: extractvalue +; CHECK: br label %tailrecurse +; CHECK: tailrecurse: +; CHECK: ifend: +; CHECK: insertvalue +define { i32*, i32 } @test5(i32 %i, { i32*, i32 } %e) { +entry: + br label %tailrecurse + +tailrecurse: ; preds = %then, %entry + %i.tr = phi i32 [ %i, %entry ], [ %cmp2, %then ] + %out = extractvalue { i32*, i32 } %e, 1 + %d = insertvalue { i32*, i32 } %e, i32* null, 0 + %cmp1 = icmp sgt i32 %out, %i.tr + br i1 %cmp1, label %then, label %ifend + +then: ; preds = %tailrecurse + call void @foo() + %cmp2 = add i32 %i.tr, 1 + br label %tailrecurse + +ifend: ; preds = %tailrecurse + ret { i32*, i32 } %d +} diff --git a/test/Transforms/LICM/scalar_promote.ll b/test/Transforms/LICM/scalar_promote.ll index 05a64d6322741..e7eab92aa8d7f 100644 --- a/test/Transforms/LICM/scalar_promote.ll +++ b/test/Transforms/LICM/scalar_promote.ll @@ -1,28 +1,28 @@ -; RUN: opt < %s -basicaa -licm -S | FileCheck %s +; RUN: opt < %s -basicaa -tbaa -licm -S | FileCheck %s target datalayout = "E-p:64:64:64-a0:0:8-f32:32:32-f64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-v64:64:64-v128:128:128" -@X = global i32 7 ; [#uses=4] +@X = global i32 7 ; [#uses=4] define void @test1(i32 %i) { Entry: - br label %Loop + br label %Loop ; CHECK: @test1 ; CHECK: Entry: ; CHECK-NEXT: load i32* @X ; CHECK-NEXT: br label %Loop -Loop: ; preds = %Loop, %0 - %j = phi i32 [ 0, %Entry ], [ %Next, %Loop ] ; [#uses=1] - %x = load i32* @X ; [#uses=1] - %x2 = add i32 %x, 1 ; [#uses=1] - store i32 %x2, i32* @X - %Next = add i32 %j, 1 ; [#uses=2] - %cond = icmp eq i32 %Next, 0 ; [#uses=1] - br i1 %cond, label %Out, label %Loop +Loop: ; preds = %Loop, %0 + %j = phi i32 [ 0, %Entry ], [ %Next, %Loop ] ; [#uses=1] + %x = load i32* @X ; [#uses=1] + %x2 = add i32 %x, 1 ; [#uses=1] + store i32 %x2, i32* @X + %Next = add i32 %j, 1 ; [#uses=2] + %cond = icmp eq i32 %Next, 0 ; [#uses=1] + br i1 %cond, label %Out, label %Loop -Out: - ret void +Out: + ret void ; CHECK: Out: ; CHECK-NEXT: store i32 %x2, i32* @X ; CHECK-NEXT: ret void @@ -31,22 +31,22 @@ Out: define void @test2(i32 %i) { Entry: - br label %Loop + br label %Loop ; CHECK: @test2 ; CHECK: Entry: ; CHECK-NEXT: %.promoted = load i32* getelementptr inbounds (i32* @X, i64 1) ; CHECK-NEXT: br label %Loop -Loop: ; preds = %Loop, %0 - %X1 = getelementptr i32* @X, i64 1 ; [#uses=1] - %A = load i32* %X1 ; [#uses=1] - %V = add i32 %A, 1 ; [#uses=1] - %X2 = getelementptr i32* @X, i64 1 ; [#uses=1] - store i32 %V, i32* %X2 - br i1 false, label %Loop, label %Exit +Loop: ; preds = %Loop, %0 + %X1 = getelementptr i32* @X, i64 1 ; [#uses=1] + %A = load i32* %X1 ; [#uses=1] + %V = add i32 %A, 1 ; [#uses=1] + %X2 = getelementptr i32* @X, i64 1 ; [#uses=1] + store i32 %V, i32* %X2 + br i1 false, label %Loop, label %Exit -Exit: ; preds = %Loop - ret void +Exit: ; preds = %Loop + ret void ; CHECK: Exit: ; CHECK-NEXT: store i32 %V, i32* getelementptr inbounds (i32* @X, i64 1) ; CHECK-NEXT: ret void @@ -56,19 +56,19 @@ Exit: ; preds = %Loop define void @test3(i32 %i) { ; CHECK: @test3 - br label %Loop + br label %Loop Loop: ; Should not promote this to a register - %x = load volatile i32* @X - %x2 = add i32 %x, 1 - store i32 %x2, i32* @X - br i1 true, label %Out, label %Loop - + %x = load volatile i32* @X + %x2 = add i32 %x, 1 + store i32 %x2, i32* @X + br i1 true, label %Out, label %Loop + ; CHECK: Loop: ; CHECK-NEXT: load volatile -Out: ; preds = %Loop - ret void +Out: ; preds = %Loop + ret void } ; PR8041 @@ -120,27 +120,27 @@ exit: define void @test5(i32 %i, i32** noalias %P2) { Entry: - br label %Loop + br label %Loop ; CHECK: @test5 ; CHECK: Entry: ; CHECK-NEXT: load i32* @X ; CHECK-NEXT: br label %Loop -Loop: ; preds = %Loop, %0 - %j = phi i32 [ 0, %Entry ], [ %Next, %Loop ] ; [#uses=1] - %x = load i32* @X ; [#uses=1] - %x2 = add i32 %x, 1 ; [#uses=1] - store i32 %x2, i32* @X - +Loop: ; preds = %Loop, %0 + %j = phi i32 [ 0, %Entry ], [ %Next, %Loop ] ; [#uses=1] + %x = load i32* @X ; [#uses=1] + %x2 = add i32 %x, 1 ; [#uses=1] + store i32 %x2, i32* @X + store volatile i32* @X, i32** %P2 - - %Next = add i32 %j, 1 ; [#uses=2] - %cond = icmp eq i32 %Next, 0 ; [#uses=1] - br i1 %cond, label %Out, label %Loop -Out: - ret void + %Next = add i32 %j, 1 ; [#uses=2] + %cond = icmp eq i32 %Next, 0 ; [#uses=1] + br i1 %cond, label %Out, label %Loop + +Out: + ret void ; CHECK: Out: ; CHECK-NEXT: store i32 %x2, i32* @X ; CHECK-NEXT: ret void @@ -148,3 +148,40 @@ Out: } +; PR14753 - Preserve TBAA tags when promoting values in a loop. +define void @test6(i32 %n, float* nocapture %a, i32* %gi) { +entry: + store i32 0, i32* %gi, align 4, !tbaa !0 + %cmp1 = icmp slt i32 0, %n + br i1 %cmp1, label %for.body.lr.ph, label %for.end + +for.body.lr.ph: ; preds = %entry + br label %for.body + +for.body: ; preds = %for.body.lr.ph, %for.body + %storemerge2 = phi i32 [ 0, %for.body.lr.ph ], [ %inc, %for.body ] + %idxprom = sext i32 %storemerge2 to i64 + %arrayidx = getelementptr inbounds float* %a, i64 %idxprom + store float 0.000000e+00, float* %arrayidx, align 4, !tbaa !3 + %0 = load i32* %gi, align 4, !tbaa !0 + %inc = add nsw i32 %0, 1 + store i32 %inc, i32* %gi, align 4, !tbaa !0 + %cmp = icmp slt i32 %inc, %n + br i1 %cmp, label %for.body, label %for.cond.for.end_crit_edge + +for.cond.for.end_crit_edge: ; preds = %for.body + br label %for.end + +for.end: ; preds = %for.cond.for.end_crit_edge, %entry + ret void + +; CHECK: for.body.lr.ph: +; CHECK-NEXT: %gi.promoted = load i32* %gi, align 4, !tbaa !0 +; CHECK: for.cond.for.end_crit_edge: +; CHECK-NEXT: store i32 %inc, i32* %gi, align 4, !tbaa !0 +} + +!0 = metadata !{metadata !"int", metadata !1} +!1 = metadata !{metadata !"omnipotent char", metadata !2} +!2 = metadata !{metadata !"Simple C/C++ TBAA"} +!3 = metadata !{metadata !"float", metadata !1} diff --git a/test/Transforms/LoopDeletion/2011-06-21-phioperands.ll b/test/Transforms/LoopDeletion/2011-06-21-phioperands.ll index 40c6629e6f4f2..cf9d8ce923ba3 100644 --- a/test/Transforms/LoopDeletion/2011-06-21-phioperands.ll +++ b/test/Transforms/LoopDeletion/2011-06-21-phioperands.ll @@ -1,4 +1,4 @@ -; RUN: opt %s -loop-deletion -disable-output +; RUN: opt -loop-deletion -disable-output < %s target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64" diff --git a/test/Transforms/LoopDeletion/simplify-then-delete.ll b/test/Transforms/LoopDeletion/simplify-then-delete.ll index 5a21672a59603..4278ef16d2145 100644 --- a/test/Transforms/LoopDeletion/simplify-then-delete.ll +++ b/test/Transforms/LoopDeletion/simplify-then-delete.ll @@ -4,7 +4,7 @@ ; Indvars and loop deletion should be able to eliminate all looping ; in this testcase. -; CHECK: define i32 @pmat(i32 %m, i32 %n, double* %y) nounwind { +; CHECK: define i32 @pmat(i32 %m, i32 %n, double* %y) #0 { ; CHECK-NEXT: entry: ; CHECK-NEXT: ret i32 0 ; CHECK-NEXT: } @@ -63,3 +63,5 @@ w.e: w.e12: ret i32 0 } + +; CHECK: attributes #0 = { nounwind } diff --git a/test/Transforms/LoopIdiom/X86/lit.local.cfg b/test/Transforms/LoopIdiom/X86/lit.local.cfg new file mode 100644 index 0000000000000..a8ad0f1a28b23 --- /dev/null +++ b/test/Transforms/LoopIdiom/X86/lit.local.cfg @@ -0,0 +1,6 @@ +config.suffixes = ['.ll', '.c', '.cpp'] + +targets = set(config.root.targets_to_build.split()) +if not 'X86' in targets: + config.unsupported = True + diff --git a/test/Transforms/LoopIdiom/X86/popcnt.ll b/test/Transforms/LoopIdiom/X86/popcnt.ll new file mode 100644 index 0000000000000..25df93d3a0821 --- /dev/null +++ b/test/Transforms/LoopIdiom/X86/popcnt.ll @@ -0,0 +1,140 @@ +; RUN: opt -loop-idiom < %s -mtriple=x86_64-apple-darwin -mcpu=corei7 -S | FileCheck %s + +;To recognize this pattern: +;int popcount(unsigned long long a) { +; int c = 0; +; while (a) { +; c++; +; a &= a - 1; +; } +; return c; +;} +; +; CHECK: entry +; CHECK: llvm.ctpop.i64 +; CHECK: ret +define i32 @popcount(i64 %a) nounwind uwtable readnone ssp { +entry: + %tobool3 = icmp eq i64 %a, 0 + br i1 %tobool3, label %while.end, label %while.body + +while.body: ; preds = %entry, %while.body + %c.05 = phi i32 [ %inc, %while.body ], [ 0, %entry ] + %a.addr.04 = phi i64 [ %and, %while.body ], [ %a, %entry ] + %inc = add nsw i32 %c.05, 1 + %sub = add i64 %a.addr.04, -1 + %and = and i64 %sub, %a.addr.04 + %tobool = icmp eq i64 %and, 0 + br i1 %tobool, label %while.end, label %while.body + +while.end: ; preds = %while.body, %entry + %c.0.lcssa = phi i32 [ 0, %entry ], [ %inc, %while.body ] + ret i32 %c.0.lcssa +} + +; To recognize this pattern: +;int popcount(unsigned long long a, int mydata1, int mydata2) { +; int c = 0; +; while (a) { +; c++; +; a &= a - 1; +; mydata1 *= c; +; mydata2 *= (int)a; +; } +; return c + mydata1 + mydata2; +;} +; CHECK: entry +; CHECK: llvm.ctpop.i64 +; CHECK: ret +define i32 @popcount2(i64 %a, i32 %mydata1, i32 %mydata2) nounwind uwtable readnone ssp { +entry: + %tobool9 = icmp eq i64 %a, 0 + br i1 %tobool9, label %while.end, label %while.body + +while.body: ; preds = %entry, %while.body + %c.013 = phi i32 [ %inc, %while.body ], [ 0, %entry ] + %mydata2.addr.012 = phi i32 [ %mul1, %while.body ], [ %mydata2, %entry ] + %mydata1.addr.011 = phi i32 [ %mul, %while.body ], [ %mydata1, %entry ] + %a.addr.010 = phi i64 [ %and, %while.body ], [ %a, %entry ] + %inc = add nsw i32 %c.013, 1 + %sub = add i64 %a.addr.010, -1 + %and = and i64 %sub, %a.addr.010 + %mul = mul nsw i32 %inc, %mydata1.addr.011 + %conv = trunc i64 %and to i32 + %mul1 = mul nsw i32 %conv, %mydata2.addr.012 + %tobool = icmp eq i64 %and, 0 + br i1 %tobool, label %while.end, label %while.body + +while.end: ; preds = %while.body, %entry + %c.0.lcssa = phi i32 [ 0, %entry ], [ %inc, %while.body ] + %mydata2.addr.0.lcssa = phi i32 [ %mydata2, %entry ], [ %mul1, %while.body ] + %mydata1.addr.0.lcssa = phi i32 [ %mydata1, %entry ], [ %mul, %while.body ] + %add = add i32 %mydata2.addr.0.lcssa, %mydata1.addr.0.lcssa + %add2 = add i32 %add, %c.0.lcssa + ret i32 %add2 +} + +; Some variants once cause crash +target triple = "x86_64-apple-macosx10.8.0" + +define i32 @PopCntCrash1(i64 %a) nounwind uwtable readnone ssp { +entry: + %tobool3 = icmp eq i64 %a, 0 + br i1 %tobool3, label %while.end, label %while.body + +while.body: ; preds = %entry, %while.body + %c.05 = phi i32 [ %inc, %while.body ], [ 0, %entry ] + %a.addr.04 = phi i64 [ %and, %while.body ], [ %a, %entry ] + %t = add i32 %c.05, %c.05 + %inc = add nsw i32 %t, 1 + %sub = add i64 %a.addr.04, -1 + %and = and i64 %sub, %a.addr.04 + %tobool = icmp eq i64 %and, 0 + br i1 %tobool, label %while.end, label %while.body + +while.end: ; preds = %while.body, %entry + %c.0.lcssa = phi i32 [ 0, %entry ], [ %inc, %while.body ] + ret i32 %c.0.lcssa + +; CHECK: entry +; CHECK: ret +} + +define i32 @PopCntCrash2(i64 %a, i32 %b) nounwind uwtable readnone ssp { +entry: + %tobool3 = icmp eq i64 %a, 0 + br i1 %tobool3, label %while.end, label %while.body + +while.body: ; preds = %entry, %while.body + %c.05 = phi i32 [ %inc, %while.body ], [ %b, %entry ] + %a.addr.04 = phi i64 [ %and, %while.body ], [ %a, %entry ] + %inc = add nsw i32 %c.05, 1 + %sub = add i64 %a.addr.04, -1 + %and = and i64 %sub, %a.addr.04 + %tobool = icmp eq i64 %and, 0 + br i1 %tobool, label %while.end, label %while.body + +while.end: ; preds = %while.body, %entry + %c.0.lcssa = phi i32 [ 0, %entry ], [ %inc, %while.body ] + ret i32 %c.0.lcssa +} + +define i32 @PopCntCrash3(i64 %a, i32 %x) { +entry: + %tobool3 = icmp eq i64 %a, 0 + %cmp = icmp eq i32 %x, 0 + br i1 %tobool3, label %while.end, label %while.body + +while.body: ; preds = %entry, %while.body + %c.05 = phi i32 [ %inc, %while.body ], [ 0, %entry ] + %a.addr.04 = phi i64 [ %and, %while.body ], [ %a, %entry ] + %inc = add nsw i32 %c.05, 1 + %sub = add i64 %a.addr.04, -1 + %and = and i64 %sub, %a.addr.04 + %tobool = icmp eq i64 %and, 0 + br i1 %cmp, label %while.end, label %while.body + +while.end: ; preds = %while.body, %entry + %c.0.lcssa = phi i32 [ 0, %entry ], [ %inc, %while.body ] + ret i32 %c.0.lcssa +} diff --git a/test/Transforms/LoopRotate/basic.ll b/test/Transforms/LoopRotate/basic.ll index b7bcb21d56f8b..78878f9fa663d 100644 --- a/test/Transforms/LoopRotate/basic.ll +++ b/test/Transforms/LoopRotate/basic.ll @@ -1,4 +1,4 @@ -; RUN: opt -S -loop-rotate %s | FileCheck %s +; RUN: opt -S -loop-rotate < %s | FileCheck %s target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64" target triple = "x86_64-apple-darwin10.0.0" @@ -33,3 +33,29 @@ for.end: ; preds = %for.cond declare void @g(i32*) +; CHECK: @test2 +define void @test2() nounwind ssp { +entry: + %array = alloca [20 x i32], align 16 + br label %for.cond + +for.cond: ; preds = %for.body, %entry + %i.0 = phi i32 [ 0, %entry ], [ %inc, %for.body ] + %cmp = icmp slt i32 %i.0, 100 +; CHECK: call void @f +; CHECK-NOT: call void @f + call void @f() noduplicate + br i1 %cmp, label %for.body, label %for.end + +for.body: ; preds = %for.cond + %inc = add nsw i32 %i.0, 1 + call void @h() + br label %for.cond + +for.end: ; preds = %for.cond + ret void +; CHECK: } +} + +declare void @f() noduplicate +declare void @h() diff --git a/test/Transforms/LoopRotate/crash.ll b/test/Transforms/LoopRotate/crash.ll index 954b834765517..fd922cb5569eb 100644 --- a/test/Transforms/LoopRotate/crash.ll +++ b/test/Transforms/LoopRotate/crash.ll @@ -1,4 +1,4 @@ -; RUN: opt -loop-rotate %s -disable-output -verify-dom-info -verify-loop-info +; RUN: opt -loop-rotate -disable-output -verify-dom-info -verify-loop-info < %s target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64" target triple = "x86_64-apple-darwin10.0.0" diff --git a/test/Transforms/LoopRotate/dbgvalue.ll b/test/Transforms/LoopRotate/dbgvalue.ll index b32ee82d3a57e..6a8d30820f6e2 100644 --- a/test/Transforms/LoopRotate/dbgvalue.ll +++ b/test/Transforms/LoopRotate/dbgvalue.ll @@ -1,4 +1,4 @@ -; RUN: opt -S -loop-rotate %s | FileCheck %s +; RUN: opt -S -loop-rotate < %s | FileCheck %s declare void @llvm.dbg.declare(metadata, metadata) nounwind readnone declare void @llvm.dbg.value(metadata, i64, metadata) nounwind readnone diff --git a/test/Transforms/LoopRotate/phi-duplicate.ll b/test/Transforms/LoopRotate/phi-duplicate.ll index 7372830922508..8ad2dce71a659 100644 --- a/test/Transforms/LoopRotate/phi-duplicate.ll +++ b/test/Transforms/LoopRotate/phi-duplicate.ll @@ -1,4 +1,4 @@ -; RUN: opt -S %s -loop-rotate | FileCheck %s +; RUN: opt -S -loop-rotate < %s | FileCheck %s target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64" target triple = "x86_64-apple-darwin10.0" diff --git a/test/Transforms/LoopStrengthReduce/2008-08-14-ShadowIV.ll b/test/Transforms/LoopStrengthReduce/2008-08-14-ShadowIV.ll deleted file mode 100644 index c650d8cf76d8d..0000000000000 --- a/test/Transforms/LoopStrengthReduce/2008-08-14-ShadowIV.ll +++ /dev/null @@ -1,99 +0,0 @@ -; RUN: opt < %s -loop-reduce -S | grep "phi double" | count 1 - -define void @foobar(i32 %n) nounwind { -entry: - icmp eq i32 %n, 0 ; :0 [#uses=2] - br i1 %0, label %return, label %bb.nph - -bb.nph: ; preds = %entry - %umax = select i1 %0, i32 1, i32 %n ; [#uses=1] - br label %bb - -bb: ; preds = %bb, %bb.nph - %i.03 = phi i32 [ 0, %bb.nph ], [ %indvar.next, %bb ] ; [#uses=3] - tail call void @bar( i32 %i.03 ) nounwind - uitofp i32 %i.03 to double ; :1 [#uses=1] - tail call void @foo( double %1 ) nounwind - %indvar.next = add i32 %i.03, 1 ; [#uses=2] - %exitcond = icmp eq i32 %indvar.next, %umax ; [#uses=1] - br i1 %exitcond, label %return, label %bb - -return: ; preds = %bb, %entry - ret void -} - -; Unable to eliminate cast because the mantissa bits for double are not enough -; to hold all of i64 IV bits. -define void @foobar2(i64 %n) nounwind { -entry: - icmp eq i64 %n, 0 ; :0 [#uses=2] - br i1 %0, label %return, label %bb.nph - -bb.nph: ; preds = %entry - %umax = select i1 %0, i64 1, i64 %n ; [#uses=1] - br label %bb - -bb: ; preds = %bb, %bb.nph - %i.03 = phi i64 [ 0, %bb.nph ], [ %indvar.next, %bb ] ; [#uses=3] - trunc i64 %i.03 to i32 ; :1 [#uses=1] - tail call void @bar( i32 %1 ) nounwind - uitofp i64 %i.03 to double ; :2 [#uses=1] - tail call void @foo( double %2 ) nounwind - %indvar.next = add i64 %i.03, 1 ; [#uses=2] - %exitcond = icmp eq i64 %indvar.next, %umax ; [#uses=1] - br i1 %exitcond, label %return, label %bb - -return: ; preds = %bb, %entry - ret void -} - -; Unable to eliminate cast due to potentional overflow. -define void @foobar3() nounwind { -entry: - tail call i32 (...)* @nn( ) nounwind ; :0 [#uses=1] - icmp eq i32 %0, 0 ; :1 [#uses=1] - br i1 %1, label %return, label %bb - -bb: ; preds = %bb, %entry - %i.03 = phi i32 [ 0, %entry ], [ %3, %bb ] ; [#uses=3] - tail call void @bar( i32 %i.03 ) nounwind - uitofp i32 %i.03 to double ; :2 [#uses=1] - tail call void @foo( double %2 ) nounwind - add i32 %i.03, 1 ; :3 [#uses=2] - tail call i32 (...)* @nn( ) nounwind ; :4 [#uses=1] - icmp ugt i32 %4, %3 ; :5 [#uses=1] - br i1 %5, label %bb, label %return - -return: ; preds = %bb, %entry - ret void -} - -; Unable to eliminate cast due to overflow. -define void @foobar4() nounwind { -entry: - br label %bb.nph - -bb.nph: ; preds = %entry - br label %bb - -bb: ; preds = %bb, %bb.nph - %i.03 = phi i8 [ 0, %bb.nph ], [ %indvar.next, %bb ] ; [#uses=3] - %tmp2 = sext i8 %i.03 to i32 ; :0 [#uses=1] - tail call void @bar( i32 %tmp2 ) nounwind - %tmp3 = uitofp i8 %i.03 to double ; :1 [#uses=1] - tail call void @foo( double %tmp3 ) nounwind - %indvar.next = add i8 %i.03, 1 ; [#uses=2] - %tmp = sext i8 %indvar.next to i32 - %exitcond = icmp eq i32 %tmp, 32767 ; [#uses=1] - br i1 %exitcond, label %return, label %bb - -return: ; preds = %bb, %entry - ret void -} - -declare void @bar(i32) - -declare void @foo(double) - -declare i32 @nn(...) - diff --git a/test/Transforms/LoopStrengthReduce/2011-07-20-DoubleIV.ll b/test/Transforms/LoopStrengthReduce/2011-07-20-DoubleIV.ll deleted file mode 100644 index 5d9ed64ef4223..0000000000000 --- a/test/Transforms/LoopStrengthReduce/2011-07-20-DoubleIV.ll +++ /dev/null @@ -1,43 +0,0 @@ -; RUN: opt < %s -loop-reduce -S | FileCheck %s -; -; Test LSR's OptimizeShadowIV. Handle a floating-point IV with a -; nonzero initial value. -; rdar://9786536 - -; First, make sure LSR doesn't crash on an empty IVUsers list. -; CHECK: @dummyIV -; CHECK-NOT: phi -; CHECK-NOT: sitofp -; CHECK: br -define void @dummyIV() nounwind { -entry: - br label %loop - -loop: - %i.01 = phi i32 [ -39, %entry ], [ %inc, %loop ] - %conv = sitofp i32 %i.01 to double - %inc = add nsw i32 %i.01, 1 - br i1 undef, label %loop, label %for.end - -for.end: - unreachable -} - -; Now check that the computed double constant is correct. -; CHECK: @doubleIV -; CHECK: phi double [ -3.900000e+01, %entry ] -; CHECK: br -define void @doubleIV() nounwind { -entry: - br label %loop - -loop: - %i.01 = phi i32 [ -39, %entry ], [ %inc, %loop ] - %conv = sitofp i32 %i.01 to double - %div = fdiv double %conv, 4.000000e+01 - %inc = add nsw i32 %i.01, 1 - br i1 undef, label %loop, label %for.end - -for.end: - unreachable -} diff --git a/test/Transforms/LoopStrengthReduce/2012-07-18-LimitReassociate.ll b/test/Transforms/LoopStrengthReduce/2012-07-18-LimitReassociate.ll index 3793baccbbc12..9524be3ceee03 100644 --- a/test/Transforms/LoopStrengthReduce/2012-07-18-LimitReassociate.ll +++ b/test/Transforms/LoopStrengthReduce/2012-07-18-LimitReassociate.ll @@ -1,4 +1,4 @@ -; RUN: opt -loop-reduce -disable-output -debug-only=loop-reduce %s 2> %t +; RUN: opt -loop-reduce -disable-output -debug-only=loop-reduce < %s 2> %t ; RUN: FileCheck %s < %t ; REQUIRES: asserts ; @@ -10,15 +10,13 @@ ; CHECK: After generating reuse formulae: ; CHECK: LSR is examining the following uses: ; CHECK: LSR Use: Kind=Special -; CHECK: {{.*reg\(\{\{\{\{\{\{\{\{\{}} -; CHECK: {{.*reg\(\{\{\{\{\{\{\{\{\{}} -; CHECK: {{.*reg\(\{\{\{\{\{\{\{\{\{}} -; CHECK: {{.*reg\(\{\{\{\{\{\{\{\{\{}} -; CHECK: {{.*reg\(\{\{\{\{\{\{\{\{\{}} +; CHECK: {{.*reg\(\{.*\{.*\{.*\{.*\{.*\{.*\{.*\{.*\{}} +; CHECK: {{.*reg\(\{.*\{.*\{.*\{.*\{.*\{.*\{.*\{.*\{}} +; CHECK: {{.*reg\(\{.*\{.*\{.*\{.*\{.*\{.*\{.*\{.*\{}} +; CHECK: {{.*reg\(\{.*\{.*\{.*\{.*\{.*\{.*\{.*\{.*\{}} +; CHECK: {{.*reg\(\{.*\{.*\{.*\{.*\{.*\{.*\{.*\{.*\{}} ; CHECK-NOT:reg ; CHECK: Filtering for use -target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128" -target triple = "x86_64-unknown-freebsd9" %struct.snork = type { %struct.fuga, i32, i32, i32, i32, i32, i32 } %struct.fuga = type { %struct.gork, i64 } diff --git a/test/Transforms/LoopStrengthReduce/2013-01-05-IndBr.ll b/test/Transforms/LoopStrengthReduce/2013-01-05-IndBr.ll new file mode 100644 index 0000000000000..bce234cd40664 --- /dev/null +++ b/test/Transforms/LoopStrengthReduce/2013-01-05-IndBr.ll @@ -0,0 +1,44 @@ +; RUN: opt -loop-reduce -S < %s | FileCheck %s +; +; Indirect branch in the preheader crashes replaceCongruentIVs. +; rdar://12910141 + +target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128-n8:16:32-S128" + +; CHECK: @test +; CHECK: bb8: +; CHECK-NEXT: phi i8 +; CHECK-NEXT: phi i8 +; CHECK: ret void +define void @test() nounwind ssp { +bb: + br label %bb190 + +bb8: ; preds = %bb190, %bb11 + %tmp = phi i8 [ %tmp14, %bb11 ], [ 25, %bb190 ] + %tmp9 = phi i8 [ %tmp12, %bb11 ], [ 25, %bb190 ] + %tmp10 = add i8 %tmp, -5 + indirectbr i8* undef, [label %bb11, label %bb15] + +bb11: ; preds = %bb8 + %tmp12 = add i8 %tmp9, 1 + %tmp13 = add i8 %tmp9, -19 + %tmp14 = add i8 %tmp, 1 + indirectbr i8* undef, [label %bb8] + +bb15: ; preds = %bb8 + indirectbr i8* undef, [label %bb16] + +bb16: ; preds = %bb16, %bb15 + indirectbr i8* undef, [label %bb37, label %bb190] + + +bb37: ; preds = %bb190 + indirectbr i8* undef, [label %bb38] + +bb38: ; preds = %bb37, %bb5 + ret void + +bb190: ; preds = %bb189, %bb187 + indirectbr i8* undef, [label %bb37, label %bb8] +} diff --git a/test/Transforms/LoopStrengthReduce/2013-01-14-ReuseCast.ll b/test/Transforms/LoopStrengthReduce/2013-01-14-ReuseCast.ll new file mode 100644 index 0000000000000..8fbddf8ae4c82 --- /dev/null +++ b/test/Transforms/LoopStrengthReduce/2013-01-14-ReuseCast.ll @@ -0,0 +1,84 @@ +; RUN: opt -loop-reduce -S < %s | FileCheck %s +; +; LTO of clang, which mistakenly uses no TargetLoweringInfo, causes a +; miscompile. ReuseOrCreateCast replace ptrtoint operand with undef. +; Reproducing the miscompile requires no triple, hence no "TTI". +; rdar://13007381 + +target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128" + +; Verify that nothing uses the "dead" ptrtoint from "undef". +; CHECK: @VerifyDiagnosticConsumerTest +; CHECK: bb: +; CHECK: %0 = ptrtoint i8* undef to i64 +; CHECK-NOT: %0 +; CHECK: .lr.ph +; CHECK-NOT: %0 +; CHECK: sub i64 %7, %tmp6 +; CHECK-NOT: %0 +; CHECK: ret void +define void @VerifyDiagnosticConsumerTest() unnamed_addr nounwind uwtable align 2 { +bb: + %tmp3 = call i8* @getCharData() nounwind + %tmp4 = call i8* @getCharData() nounwind + %tmp5 = ptrtoint i8* %tmp4 to i64 + %tmp6 = ptrtoint i8* %tmp3 to i64 + %tmp7 = sub i64 %tmp5, %tmp6 + br i1 undef, label %bb87, label %.preheader + +.preheader: ; preds = %bb10, %bb + br i1 undef, label %_ZNK4llvm9StringRef4findEcm.exit42.thread, label %bb10 + +bb10: ; preds = %.preheader + br i1 undef, label %_ZNK4llvm9StringRef4findEcm.exit42, label %.preheader + +_ZNK4llvm9StringRef4findEcm.exit42: ; preds = %bb10 + br i1 undef, label %_ZNK4llvm9StringRef4findEcm.exit42.thread, label %.lr.ph + +_ZNK4llvm9StringRef4findEcm.exit42.thread: ; preds = %_ZNK4llvm9StringRef4findEcm.exit42, %.preheader + unreachable + +.lr.ph: ; preds = %_ZNK4llvm9StringRef4findEcm.exit42 + br label %bb36 + +_ZNK4llvm9StringRef4findEcm.exit.loopexit: ; preds = %bb63 + %tmp21 = icmp eq i64 %i.0.i, -1 + br i1 %tmp21, label %_ZNK4llvm9StringRef4findEcm.exit._crit_edge, label %bb36 + +_ZNK4llvm9StringRef4findEcm.exit._crit_edge: ; preds = %bb61, %_ZNK4llvm9StringRef4findEcm.exit.loopexit + unreachable + +bb36: ; preds = %_ZNK4llvm9StringRef4findEcm.exit.loopexit, %.lr.ph + %loc.063 = phi i64 [ undef, %.lr.ph ], [ %i.0.i, %_ZNK4llvm9StringRef4findEcm.exit.loopexit ] + switch i8 undef, label %bb57 [ + i8 10, label %bb48 + i8 13, label %bb48 + ] + +bb48: ; preds = %bb36, %bb36 + br label %bb58 + +bb57: ; preds = %bb36 + br label %bb58 + +bb58: ; preds = %bb57, %bb48 + %tmp59 = icmp ugt i64 %tmp7, undef + %tmp60 = select i1 %tmp59, i64 undef, i64 %tmp7 + br label %bb61 + +bb61: ; preds = %bb63, %bb58 + %i.0.i = phi i64 [ %tmp60, %bb58 ], [ %tmp67, %bb63 ] + %tmp62 = icmp eq i64 %i.0.i, %tmp7 + br i1 %tmp62, label %_ZNK4llvm9StringRef4findEcm.exit._crit_edge, label %bb63 + +bb63: ; preds = %bb61 + %tmp64 = getelementptr inbounds i8* %tmp3, i64 %i.0.i + %tmp65 = load i8* %tmp64, align 1 + %tmp67 = add i64 %i.0.i, 1 + br i1 undef, label %_ZNK4llvm9StringRef4findEcm.exit.loopexit, label %bb61 + +bb87: ; preds = %bb + ret void +} + +declare i8* @getCharData() diff --git a/test/Transforms/LoopStrengthReduce/ARM/ivchain-ARM.ll b/test/Transforms/LoopStrengthReduce/ARM/ivchain-ARM.ll index 9189d79e2fb6a..ee3cc4dd78fc6 100644 --- a/test/Transforms/LoopStrengthReduce/ARM/ivchain-ARM.ll +++ b/test/Transforms/LoopStrengthReduce/ARM/ivchain-ARM.ll @@ -205,18 +205,18 @@ for.end: ; preds = %for.body ; post-increment addressing, no add's or add.w's beyond the three ; mentioned. Most importantly, there should be no spills or reloads! ; -; CHECK: testNeon: -; CHECK: %.lr.ph -; CHECK-NOT: lsl.w -; CHECK-NOT: {{ldr|str|adds|add r}} -; CHECK: add.w r -; CHECK-NOT: {{ldr|str|adds|add r}} -; CHECK: add.w r -; CHECK-NOT: {{ldr|str|adds|add r}} -; CHECK: add.w r -; CHECK-NOT: {{ldr|str|adds|add r}} -; CHECK-NOT: add.w r -; CHECK: bne +; A9: testNeon: +; A9: %.lr.ph +; A9-NOT: lsl.w +; A9-NOT: {{ldr|str|adds|add r}} +; A9: add.w r +; A9-NOT: {{ldr|str|adds|add r}} +; A9: add.w r +; A9-NOT: {{ldr|str|adds|add r}} +; A9: add.w r +; A9-NOT: {{ldr|str|adds|add r}} +; A9-NOT: add.w r +; A9: bne define hidden void @testNeon(i8* %ref_data, i32 %ref_stride, i32 %limit, <16 x i8>* nocapture %data) nounwind optsize { %1 = icmp sgt i32 %limit, 0 br i1 %1, label %.lr.ph, label %45 @@ -290,3 +290,80 @@ define hidden void @testNeon(i8* %ref_data, i32 %ref_stride, i32 %limit, <16 x i } declare <1 x i64> @llvm.arm.neon.vld1.v1i64(i8*, i32) nounwind readonly + +; Handle chains in which the same offset is used for both loads and +; stores to the same array. +; rdar://11410078. +; +; A9: @testReuse +; A9: %for.body +; A9: vld1.8 {d{{[0-9]+}}}, [[BASE:[r[0-9]+]]], [[INC:r[0-9]]] +; A9: vld1.8 {d{{[0-9]+}}}, [[BASE]], [[INC]] +; A9: vld1.8 {d{{[0-9]+}}}, [[BASE]], [[INC]] +; A9: vld1.8 {d{{[0-9]+}}}, [[BASE]], [[INC]] +; A9: vld1.8 {d{{[0-9]+}}}, [[BASE]], [[INC]] +; A9: vld1.8 {d{{[0-9]+}}}, [[BASE]], [[INC]] +; A9: vld1.8 {d{{[0-9]+}}}, [[BASE]], [[INC]] +; A9: vld1.8 {d{{[0-9]+}}}, [[BASE]], {{r[0-9]}} +; A9: vst1.8 {d{{[0-9]+}}}, [[BASE]], [[INC]] +; A9: vst1.8 {d{{[0-9]+}}}, [[BASE]], [[INC]] +; A9: vst1.8 {d{{[0-9]+}}}, [[BASE]], [[INC]] +; A9: vst1.8 {d{{[0-9]+}}}, [[BASE]], [[INC]] +; A9: vst1.8 {d{{[0-9]+}}}, [[BASE]], [[INC]] +; A9: vst1.8 {d{{[0-9]+}}}, [[BASE]] +; A9: bne +define void @testReuse(i8* %src, i32 %stride) nounwind ssp { +entry: + %mul = shl nsw i32 %stride, 2 + %idx.neg = sub i32 0, %mul + %mul1 = mul nsw i32 %stride, 3 + %idx.neg2 = sub i32 0, %mul1 + %mul5 = shl nsw i32 %stride, 1 + %idx.neg6 = sub i32 0, %mul5 + %idx.neg10 = sub i32 0, %stride + br label %for.body + +for.body: ; preds = %for.body, %entry + %i.0110 = phi i32 [ 0, %entry ], [ %inc, %for.body ] + %src.addr = phi i8* [ %src, %entry ], [ %add.ptr45, %for.body ] + %add.ptr = getelementptr inbounds i8* %src.addr, i32 %idx.neg + %vld1 = tail call <8 x i8> @llvm.arm.neon.vld1.v8i8(i8* %add.ptr, i32 1) + %add.ptr3 = getelementptr inbounds i8* %src.addr, i32 %idx.neg2 + %vld2 = tail call <8 x i8> @llvm.arm.neon.vld1.v8i8(i8* %add.ptr3, i32 1) + %add.ptr7 = getelementptr inbounds i8* %src.addr, i32 %idx.neg6 + %vld3 = tail call <8 x i8> @llvm.arm.neon.vld1.v8i8(i8* %add.ptr7, i32 1) + %add.ptr11 = getelementptr inbounds i8* %src.addr, i32 %idx.neg10 + %vld4 = tail call <8 x i8> @llvm.arm.neon.vld1.v8i8(i8* %add.ptr11, i32 1) + %vld5 = tail call <8 x i8> @llvm.arm.neon.vld1.v8i8(i8* %src.addr, i32 1) + %add.ptr17 = getelementptr inbounds i8* %src.addr, i32 %stride + %vld6 = tail call <8 x i8> @llvm.arm.neon.vld1.v8i8(i8* %add.ptr17, i32 1) + %add.ptr20 = getelementptr inbounds i8* %src.addr, i32 %mul5 + %vld7 = tail call <8 x i8> @llvm.arm.neon.vld1.v8i8(i8* %add.ptr20, i32 1) + %add.ptr23 = getelementptr inbounds i8* %src.addr, i32 %mul1 + %vld8 = tail call <8 x i8> @llvm.arm.neon.vld1.v8i8(i8* %add.ptr23, i32 1) + %vadd1 = tail call <8 x i8> @llvm.arm.neon.vhaddu.v8i8(<8 x i8> %vld1, <8 x i8> %vld2) nounwind + %vadd2 = tail call <8 x i8> @llvm.arm.neon.vhaddu.v8i8(<8 x i8> %vld2, <8 x i8> %vld3) nounwind + %vadd3 = tail call <8 x i8> @llvm.arm.neon.vhaddu.v8i8(<8 x i8> %vld3, <8 x i8> %vld4) nounwind + %vadd4 = tail call <8 x i8> @llvm.arm.neon.vhaddu.v8i8(<8 x i8> %vld4, <8 x i8> %vld5) nounwind + %vadd5 = tail call <8 x i8> @llvm.arm.neon.vhaddu.v8i8(<8 x i8> %vld5, <8 x i8> %vld6) nounwind + %vadd6 = tail call <8 x i8> @llvm.arm.neon.vhaddu.v8i8(<8 x i8> %vld6, <8 x i8> %vld7) nounwind + tail call void @llvm.arm.neon.vst1.v8i8(i8* %add.ptr3, <8 x i8> %vadd1, i32 1) + tail call void @llvm.arm.neon.vst1.v8i8(i8* %add.ptr7, <8 x i8> %vadd2, i32 1) + tail call void @llvm.arm.neon.vst1.v8i8(i8* %add.ptr11, <8 x i8> %vadd3, i32 1) + tail call void @llvm.arm.neon.vst1.v8i8(i8* %src.addr, <8 x i8> %vadd4, i32 1) + tail call void @llvm.arm.neon.vst1.v8i8(i8* %add.ptr17, <8 x i8> %vadd5, i32 1) + tail call void @llvm.arm.neon.vst1.v8i8(i8* %add.ptr20, <8 x i8> %vadd6, i32 1) + %inc = add nsw i32 %i.0110, 1 + %add.ptr45 = getelementptr inbounds i8* %src.addr, i32 8 + %exitcond = icmp eq i32 %inc, 4 + br i1 %exitcond, label %for.end, label %for.body + +for.end: ; preds = %for.body + ret void +} + +declare <8 x i8> @llvm.arm.neon.vld1.v8i8(i8*, i32) nounwind readonly + +declare void @llvm.arm.neon.vst1.v8i8(i8*, <8 x i8>, i32) nounwind + +declare <8 x i8> @llvm.arm.neon.vhaddu.v8i8(<8 x i8>, <8 x i8>) nounwind readnone diff --git a/test/Transforms/LoopStrengthReduce/X86/2008-08-14-ShadowIV.ll b/test/Transforms/LoopStrengthReduce/X86/2008-08-14-ShadowIV.ll new file mode 100644 index 0000000000000..9a7f4865c5913 --- /dev/null +++ b/test/Transforms/LoopStrengthReduce/X86/2008-08-14-ShadowIV.ll @@ -0,0 +1,99 @@ +; RUN: opt < %s -loop-reduce -S -mtriple=x86_64-unknown-unknown | grep "phi double" | count 1 + +define void @foobar(i32 %n) nounwind { +entry: + icmp eq i32 %n, 0 ; :0 [#uses=2] + br i1 %0, label %return, label %bb.nph + +bb.nph: ; preds = %entry + %umax = select i1 %0, i32 1, i32 %n ; [#uses=1] + br label %bb + +bb: ; preds = %bb, %bb.nph + %i.03 = phi i32 [ 0, %bb.nph ], [ %indvar.next, %bb ] ; [#uses=3] + tail call void @bar( i32 %i.03 ) nounwind + uitofp i32 %i.03 to double ; :1 [#uses=1] + tail call void @foo( double %1 ) nounwind + %indvar.next = add i32 %i.03, 1 ; [#uses=2] + %exitcond = icmp eq i32 %indvar.next, %umax ; [#uses=1] + br i1 %exitcond, label %return, label %bb + +return: ; preds = %bb, %entry + ret void +} + +; Unable to eliminate cast because the mantissa bits for double are not enough +; to hold all of i64 IV bits. +define void @foobar2(i64 %n) nounwind { +entry: + icmp eq i64 %n, 0 ; :0 [#uses=2] + br i1 %0, label %return, label %bb.nph + +bb.nph: ; preds = %entry + %umax = select i1 %0, i64 1, i64 %n ; [#uses=1] + br label %bb + +bb: ; preds = %bb, %bb.nph + %i.03 = phi i64 [ 0, %bb.nph ], [ %indvar.next, %bb ] ; [#uses=3] + trunc i64 %i.03 to i32 ; :1 [#uses=1] + tail call void @bar( i32 %1 ) nounwind + uitofp i64 %i.03 to double ; :2 [#uses=1] + tail call void @foo( double %2 ) nounwind + %indvar.next = add i64 %i.03, 1 ; [#uses=2] + %exitcond = icmp eq i64 %indvar.next, %umax ; [#uses=1] + br i1 %exitcond, label %return, label %bb + +return: ; preds = %bb, %entry + ret void +} + +; Unable to eliminate cast due to potentional overflow. +define void @foobar3() nounwind { +entry: + tail call i32 (...)* @nn( ) nounwind ; :0 [#uses=1] + icmp eq i32 %0, 0 ; :1 [#uses=1] + br i1 %1, label %return, label %bb + +bb: ; preds = %bb, %entry + %i.03 = phi i32 [ 0, %entry ], [ %3, %bb ] ; [#uses=3] + tail call void @bar( i32 %i.03 ) nounwind + uitofp i32 %i.03 to double ; :2 [#uses=1] + tail call void @foo( double %2 ) nounwind + add i32 %i.03, 1 ; :3 [#uses=2] + tail call i32 (...)* @nn( ) nounwind ; :4 [#uses=1] + icmp ugt i32 %4, %3 ; :5 [#uses=1] + br i1 %5, label %bb, label %return + +return: ; preds = %bb, %entry + ret void +} + +; Unable to eliminate cast due to overflow. +define void @foobar4() nounwind { +entry: + br label %bb.nph + +bb.nph: ; preds = %entry + br label %bb + +bb: ; preds = %bb, %bb.nph + %i.03 = phi i8 [ 0, %bb.nph ], [ %indvar.next, %bb ] ; [#uses=3] + %tmp2 = sext i8 %i.03 to i32 ; :0 [#uses=1] + tail call void @bar( i32 %tmp2 ) nounwind + %tmp3 = uitofp i8 %i.03 to double ; :1 [#uses=1] + tail call void @foo( double %tmp3 ) nounwind + %indvar.next = add i8 %i.03, 1 ; [#uses=2] + %tmp = sext i8 %indvar.next to i32 + %exitcond = icmp eq i32 %tmp, 32767 ; [#uses=1] + br i1 %exitcond, label %return, label %bb + +return: ; preds = %bb, %entry + ret void +} + +declare void @bar(i32) + +declare void @foo(double) + +declare i32 @nn(...) + diff --git a/test/Transforms/LoopStrengthReduce/X86/2011-07-20-DoubleIV.ll b/test/Transforms/LoopStrengthReduce/X86/2011-07-20-DoubleIV.ll new file mode 100644 index 0000000000000..a932b47925865 --- /dev/null +++ b/test/Transforms/LoopStrengthReduce/X86/2011-07-20-DoubleIV.ll @@ -0,0 +1,43 @@ +; RUN: opt < %s -loop-reduce -S -mtriple=x86_64-unknown-unknown | FileCheck %s +; +; Test LSR's OptimizeShadowIV. Handle a floating-point IV with a +; nonzero initial value. +; rdar://9786536 + +; First, make sure LSR doesn't crash on an empty IVUsers list. +; CHECK: @dummyIV +; CHECK-NOT: phi +; CHECK-NOT: sitofp +; CHECK: br +define void @dummyIV() nounwind { +entry: + br label %loop + +loop: + %i.01 = phi i32 [ -39, %entry ], [ %inc, %loop ] + %conv = sitofp i32 %i.01 to double + %inc = add nsw i32 %i.01, 1 + br i1 undef, label %loop, label %for.end + +for.end: + unreachable +} + +; Now check that the computed double constant is correct. +; CHECK: @doubleIV +; CHECK: phi double [ -3.900000e+01, %entry ] +; CHECK: br +define void @doubleIV() nounwind { +entry: + br label %loop + +loop: + %i.01 = phi i32 [ -39, %entry ], [ %inc, %loop ] + %conv = sitofp i32 %i.01 to double + %div = fdiv double %conv, 4.000000e+01 + %inc = add nsw i32 %i.01, 1 + br i1 undef, label %loop, label %for.end + +for.end: + unreachable +} diff --git a/test/Transforms/LoopStrengthReduce/X86/2011-12-04-loserreg.ll b/test/Transforms/LoopStrengthReduce/X86/2011-12-04-loserreg.ll index 5108650962728..eedfc200f48ba 100644 --- a/test/Transforms/LoopStrengthReduce/X86/2011-12-04-loserreg.ll +++ b/test/Transforms/LoopStrengthReduce/X86/2011-12-04-loserreg.ll @@ -1,4 +1,4 @@ -; RUN: llc < %s | FileCheck %s +; RUN: opt < %s -loop-reduce -S | FileCheck %s ; ; Test LSR's ability to prune formulae that refer to nonexistant ; AddRecs in other loops. @@ -15,13 +15,10 @@ target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f3 target triple = "x86_64-apple-darwin" ; CHECK: @test -; CHECK: # %for.body{{$}} -; dummyiv copy should be removed -; CHECK-NOT: movq -; CHECK: # %for.cond19.preheader -; dummycnt should be removed -; CHECK-NOT: incq -; CHECK: # %for.body23{{$}} +; CHECK: for.body: +; CHECK: %lsr.iv +; CHECK-NOT: %dummyout +; CHECK: ret define i64 @test(i64 %count, float* nocapture %srcrow, i32* nocapture %destrow) nounwind uwtable ssp { entry: %cmp34 = icmp eq i64 %count, 0 diff --git a/test/Transforms/LoopStrengthReduce/dominate-assert.ll b/test/Transforms/LoopStrengthReduce/dominate-assert.ll index b87bf620decf1..ff8cab83137bd 100644 --- a/test/Transforms/LoopStrengthReduce/dominate-assert.ll +++ b/test/Transforms/LoopStrengthReduce/dominate-assert.ll @@ -1,4 +1,4 @@ -; RUN: opt -loop-reduce %s +; RUN: opt -loop-reduce < %s ; we used to crash on this one declare i8* @_Znwm() diff --git a/test/Transforms/LoopStrengthReduce/exit_compare_live_range.ll b/test/Transforms/LoopStrengthReduce/exit_compare_live_range.ll index ad4959be340e9..498be1a9a1a28 100644 --- a/test/Transforms/LoopStrengthReduce/exit_compare_live_range.ll +++ b/test/Transforms/LoopStrengthReduce/exit_compare_live_range.ll @@ -2,7 +2,7 @@ ; having overlapping live ranges that result in copies. We want the setcc ; instruction immediately before the conditional branch. ; -; RUN: opt -S -loop-reduce %s | FileCheck %s +; RUN: opt -S -loop-reduce < %s | FileCheck %s define void @foo(float* %D, i32 %E) { entry: diff --git a/test/Transforms/LoopStrengthReduce/post-inc-icmpzero.ll b/test/Transforms/LoopStrengthReduce/post-inc-icmpzero.ll index 96904c66e640a..45aeb4e691a0c 100644 --- a/test/Transforms/LoopStrengthReduce/post-inc-icmpzero.ll +++ b/test/Transforms/LoopStrengthReduce/post-inc-icmpzero.ll @@ -4,18 +4,17 @@ ; LSR should properly handle the post-inc offset when folding the ; non-IV operand of an icmp into the IV. -; CHECK: %4 = sub i64 %sub.ptr.lhs.cast, %sub.ptr.rhs.cast -; CHECK: %5 = lshr i64 %4, 1 -; CHECK: %6 = mul i64 %5, 2 +; CHECK: [[r1:%[a-z0-9]+]] = sub i64 %sub.ptr.lhs.cast, %sub.ptr.rhs.cast +; CHECK: [[r2:%[a-z0-9]+]] = lshr i64 [[r1]], 1 +; CHECK: [[r3:%[a-z0-9]+]] = mul i64 [[r2]], 2 ; CHECK: br label %for.body ; CHECK: for.body: -; CHECK: %lsr.iv2 = phi i64 [ %lsr.iv.next, %for.body ], [ %6, %for.body.lr.ph ] +; CHECK: %lsr.iv2 = phi i64 [ %lsr.iv.next, %for.body ], [ [[r3]], %for.body.lr.ph ] ; CHECK: %lsr.iv.next = add i64 %lsr.iv2, -2 ; CHECK: %lsr.iv.next3 = inttoptr i64 %lsr.iv.next to i16* ; CHECK: %cmp27 = icmp eq i16* %lsr.iv.next3, null target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64" -target triple = "x86_64-unknown-linux-gnu" %struct.Vector2 = type { i16*, [64 x i16], i32 } diff --git a/test/Transforms/LoopUnroll/basic.ll b/test/Transforms/LoopUnroll/basic.ll index eeb3e9a57b06c..ab5bc568ede4a 100644 --- a/test/Transforms/LoopUnroll/basic.ll +++ b/test/Transforms/LoopUnroll/basic.ll @@ -22,3 +22,26 @@ l1: ; preds = %l1, %entry l2: ; preds = %l1 ret i32 0 } + +; This should not unroll since the call is 'noduplicate'. + +; CHECK: @test2 +define i32 @test2(i8** %P) nounwind ssp { +entry: + br label %l1 + +l1: ; preds = %l1, %entry + %x.0 = phi i32 [ 0, %entry ], [ %inc, %l1 ] +; CHECK: call void @f() +; CHECK-NOT: call void @f() + call void @f() noduplicate + %inc = add nsw i32 %x.0, 1 + %exitcond = icmp eq i32 %inc, 3 + br i1 %exitcond, label %l2, label %l1 + +l2: ; preds = %l1 + ret i32 0 +; CHECK: } +} + +declare void @f() diff --git a/test/Transforms/LoopUnroll/runtime-loop3.ll b/test/Transforms/LoopUnroll/runtime-loop3.ll index 55cf22373ecea..aa928ccc60c11 100644 --- a/test/Transforms/LoopUnroll/runtime-loop3.ll +++ b/test/Transforms/LoopUnroll/runtime-loop3.ll @@ -1,3 +1,4 @@ +; REQUIRES: asserts ; RUN: opt < %s -disable-output -stats -loop-unroll -unroll-runtime -unroll-threshold=400 -info-output-file - | FileCheck %s --check-prefix=STATS ; Test that nested loops can be unrolled. We need to increase threshold to do it diff --git a/test/Transforms/LoopUnswitch/2008-11-03-Invariant.ll b/test/Transforms/LoopUnswitch/2008-11-03-Invariant.ll index 9d73d31d50449..31dba79be1f89 100644 --- a/test/Transforms/LoopUnswitch/2008-11-03-Invariant.ll +++ b/test/Transforms/LoopUnswitch/2008-11-03-Invariant.ll @@ -1,3 +1,4 @@ +; REQUIRES: asserts ; RUN: opt < %s -loop-unswitch -stats -disable-output 2>&1 | grep "1 loop-unswitch - Number of branches unswitched" | count 1 ; PR 3170 define i32 @a(i32 %x, i32 %y) nounwind { diff --git a/test/Transforms/LoopUnswitch/2011-11-18-SimpleSwitch.ll b/test/Transforms/LoopUnswitch/2011-11-18-SimpleSwitch.ll index c1fd588106603..a8608b8772051 100644 --- a/test/Transforms/LoopUnswitch/2011-11-18-SimpleSwitch.ll +++ b/test/Transforms/LoopUnswitch/2011-11-18-SimpleSwitch.ll @@ -1,5 +1,6 @@ +; REQUIRES: asserts ; RUN: opt -loop-unswitch -disable-output -stats -info-output-file - < %s | FileCheck --check-prefix=STATS %s -; RUN: opt -S -loop-unswitch -verify-loop-info -verify-dom-info %s | FileCheck %s +; RUN: opt -S -loop-unswitch -verify-loop-info -verify-dom-info < %s | FileCheck %s ; STATS: 1 loop-simplify - Number of pre-header or exit blocks inserted ; STATS: 2 loop-unswitch - Number of switches unswitched @@ -19,7 +20,7 @@ ; CHECK-NEXT: i32 1, label %inc.us ; CHECK: inc.us: ; preds = %loop_begin.us -; CHECK-NEXT: call void @incf() noreturn nounwind +; CHECK-NEXT: call void @incf() [[NOR_NUW:#[0-9]+]] ; CHECK-NEXT: br label %loop_begin.backedge.us ; CHECK: .split: ; preds = %..split_crit_edge @@ -40,7 +41,7 @@ ; CHECK-NEXT: ] ; CHECK: dec.us3: ; preds = %loop_begin.us1 -; CHECK-NEXT: call void @decf() noreturn nounwind +; CHECK-NEXT: call void @decf() [[NOR_NUW]] ; CHECK-NEXT: br label %loop_begin.backedge.us5 ; CHECK: .split.split: ; preds = %.split..split.split_crit_edge @@ -89,3 +90,6 @@ loop_exit: declare void @incf() noreturn declare void @decf() noreturn + +; CHECK: attributes #0 = { noreturn } +; CHECK: attributes [[NOR_NUW]] = { noreturn nounwind } diff --git a/test/Transforms/LoopUnswitch/2011-11-18-TwoSwitches-Threshold.ll b/test/Transforms/LoopUnswitch/2011-11-18-TwoSwitches-Threshold.ll index f3db47119958e..686cedbbc51af 100644 --- a/test/Transforms/LoopUnswitch/2011-11-18-TwoSwitches-Threshold.ll +++ b/test/Transforms/LoopUnswitch/2011-11-18-TwoSwitches-Threshold.ll @@ -1,5 +1,6 @@ +; REQUIRES: asserts ; RUN: opt -loop-unswitch -loop-unswitch-threshold 13 -disable-output -stats -info-output-file - < %s | FileCheck --check-prefix=STATS %s -; RUN: opt -S -loop-unswitch -loop-unswitch-threshold 13 -verify-loop-info -verify-dom-info %s | FileCheck %s +; RUN: opt -S -loop-unswitch -loop-unswitch-threshold 13 -verify-loop-info -verify-dom-info < %s | FileCheck %s ; STATS: 1 loop-simplify - Number of pre-header or exit blocks inserted ; STATS: 1 loop-unswitch - Number of switches unswitched @@ -25,7 +26,7 @@ ; CHECK-NEXT: ] ; CHECK: inc.us: ; preds = %second_switch.us, %loop_begin.us -; CHECK-NEXT: call void @incf() noreturn nounwind +; CHECK-NEXT: call void @incf() [[NOR_NUW:#[0-9]+]] ; CHECK-NEXT: br label %loop_begin.backedge.us ; CHECK: .split: ; preds = %..split_crit_edge @@ -45,7 +46,7 @@ ; CHECK-NEXT: ] ; CHECK: inc: ; preds = %loop_begin.inc_crit_edge, %second_switch -; CHECK-NEXT: call void @incf() noreturn nounwind +; CHECK-NEXT: call void @incf() [[NOR_NUW]] ; CHECK-NEXT: br label %loop_begin.backedge define i32 @test(i32* %var) { @@ -82,3 +83,6 @@ loop_exit: declare void @incf() noreturn declare void @decf() noreturn + +; CHECK: attributes #0 = { noreturn } +; CHECK: attributes [[NOR_NUW]] = { noreturn nounwind } diff --git a/test/Transforms/LoopUnswitch/2011-11-18-TwoSwitches.ll b/test/Transforms/LoopUnswitch/2011-11-18-TwoSwitches.ll index 270899642ffa0..3ba9fc2f5cf15 100644 --- a/test/Transforms/LoopUnswitch/2011-11-18-TwoSwitches.ll +++ b/test/Transforms/LoopUnswitch/2011-11-18-TwoSwitches.ll @@ -1,5 +1,6 @@ +; REQUIRES: asserts ; RUN: opt -loop-unswitch -loop-unswitch-threshold 1000 -disable-output -stats -info-output-file - < %s | FileCheck --check-prefix=STATS %s -; RUN: opt -S -loop-unswitch -loop-unswitch-threshold 1000 -verify-loop-info -verify-dom-info %s | FileCheck %s +; RUN: opt -S -loop-unswitch -loop-unswitch-threshold 1000 -verify-loop-info -verify-dom-info < %s | FileCheck %s ; STATS: 1 loop-simplify - Number of pre-header or exit blocks inserted ; STATS: 3 loop-unswitch - Number of switches unswitched @@ -30,7 +31,7 @@ ; CHECK-NEXT: i32 1, label %inc.us.us ; CHECK: inc.us.us: ; preds = %second_switch.us.us, %loop_begin.us.us -; CHECK-NEXT: call void @incf() noreturn nounwind +; CHECK-NEXT: call void @incf() [[NOR_NUW:#[0-9]+]] ; CHECK-NEXT: br label %loop_begin.backedge.us.us ; CHECK: .split.us.split: ; preds = %.split.us..split.us.split_crit_edge @@ -50,7 +51,7 @@ ; CHECK-NEXT: br i1 true, label %us-unreachable8, label %inc.us ; CHECK: inc.us: ; preds = %second_switch.us.inc.us_crit_edge, %loop_begin.us -; CHECK-NEXT: call void @incf() noreturn nounwind +; CHECK-NEXT: call void @incf() [[NOR_NUW]] ; CHECK-NEXT: br label %loop_begin.backedge.us ; CHECK: .split: ; preds = %..split_crit_edge @@ -75,7 +76,7 @@ ; CHECK-NEXT: ] ; CHECK: inc.us4: ; preds = %loop_begin.inc_crit_edge.us, %second_switch.us3 -; CHECK-NEXT: call void @incf() noreturn nounwind +; CHECK-NEXT: call void @incf() [[NOR_NUW]] ; CHECK-NEXT: br label %loop_begin.backedge.us6 ; CHECK: loop_begin.inc_crit_edge.us: ; preds = %loop_begin.us1 @@ -136,3 +137,6 @@ loop_exit: declare void @incf() noreturn declare void @decf() noreturn + +; CHECK: attributes #0 = { noreturn } +; CHECK: attributes [[NOR_NUW]] = { noreturn nounwind } diff --git a/test/Transforms/LoopUnswitch/basictest.ll b/test/Transforms/LoopUnswitch/basictest.ll index 1e6f2cf15ee14..e98d82b6522db 100644 --- a/test/Transforms/LoopUnswitch/basictest.ll +++ b/test/Transforms/LoopUnswitch/basictest.ll @@ -1,4 +1,4 @@ -; RUN: opt < %s -loop-unswitch -disable-output +; RUN: opt < %s -loop-unswitch -verify-loop-info -S < %s 2>&1 | FileCheck %s define i32 @test(i32* %A, i1 %C) { entry: @@ -29,3 +29,40 @@ return: ; preds = %endif, %then ret i32 %tmp.13 } +; This simple test would normally unswitch, but should be inhibited by the presence of +; the noduplicate call. + +; CHECK: @test2 +define i32 @test2(i32* %var) { + %mem = alloca i32 + store i32 2, i32* %mem + %c = load i32* %mem + + br label %loop_begin + +loop_begin: + + %var_val = load i32* %var + + switch i32 %c, label %default [ + i32 1, label %inc + i32 2, label %dec + ] + +inc: + call void @incf() noreturn nounwind + br label %loop_begin +dec: +; CHECK: call void @decf() +; CHECK-NOT: call void @decf() + call void @decf() noreturn nounwind noduplicate + br label %loop_begin +default: + br label %loop_exit +loop_exit: + ret i32 0 +; CHECK: } +} + +declare void @incf() noreturn +declare void @decf() noreturn diff --git a/test/Transforms/LoopUnswitch/infinite-loop.ll b/test/Transforms/LoopUnswitch/infinite-loop.ll index 73391ca8d19d3..8261e389370a2 100644 --- a/test/Transforms/LoopUnswitch/infinite-loop.ll +++ b/test/Transforms/LoopUnswitch/infinite-loop.ll @@ -1,3 +1,4 @@ +; REQUIRES: asserts ; RUN: opt -loop-unswitch -disable-output -stats -info-output-file - < %s | FileCheck --check-prefix=STATS %s ; RUN: opt -loop-unswitch -simplifycfg -S < %s | FileCheck %s ; PR5373 @@ -21,11 +22,11 @@ ; CHECK-NEXT: br label %cond.end.us ; CHECK: abort0.split: -; CHECK-NEXT: call void @end0() noreturn nounwind +; CHECK-NEXT: call void @end0() [[NOR_NUW:#[0-9]+]] ; CHECK-NEXT: unreachable ; CHECK: abort1: -; CHECK-NEXT: call void @end1() noreturn nounwind +; CHECK-NEXT: call void @end1() [[NOR_NUW]] ; CHECK-NEXT: unreachable ; CHECK: } @@ -51,3 +52,7 @@ abort1: declare void @end0() noreturn declare void @end1() noreturn + +; CHECK: attributes #0 = { nounwind } +; CHECK: attributes #1 = { noreturn } +; CHECK: attributes [[NOR_NUW]] = { noreturn nounwind } diff --git a/test/Transforms/LoopUnswitch/preserve-analyses.ll b/test/Transforms/LoopUnswitch/preserve-analyses.ll index 668f8ecaf8a5a..f79612bef51e6 100644 --- a/test/Transforms/LoopUnswitch/preserve-analyses.ll +++ b/test/Transforms/LoopUnswitch/preserve-analyses.ll @@ -1,4 +1,4 @@ -; RUN: opt -loop-unswitch -verify-loop-info -verify-dom-info %s -disable-output +; RUN: opt -loop-unswitch -verify-loop-info -verify-dom-info -disable-output < %s ; Loop unswitch should be able to unswitch these loops and ; preserve LCSSA and LoopSimplify forms. diff --git a/test/Transforms/LoopVectorize/12-12-11-if-conv.ll b/test/Transforms/LoopVectorize/12-12-11-if-conv.ll new file mode 100644 index 0000000000000..2dd7fe34a70bf --- /dev/null +++ b/test/Transforms/LoopVectorize/12-12-11-if-conv.ll @@ -0,0 +1,44 @@ +; RUN: opt < %s -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -enable-if-conversion -dce -instcombine -S | FileCheck %s + +target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128" +target triple = "x86_64-apple-macosx10.8.0" + +;CHECK: @foo +;CHECK: icmp eq <4 x i32> +;CHECK: select <4 x i1> +;CHECK: ret i32 +define i32 @foo(i32 %x, i32 %t, i32* nocapture %A) nounwind uwtable ssp { +entry: + %cmp10 = icmp sgt i32 %x, 0 + br i1 %cmp10, label %for.body, label %for.end + +for.body: ; preds = %entry, %if.end + %indvars.iv = phi i64 [ %indvars.iv.next, %if.end ], [ 0, %entry ] + %arrayidx = getelementptr inbounds i32* %A, i64 %indvars.iv + %0 = load i32* %arrayidx, align 4, !tbaa !0 + %tobool = icmp eq i32 %0, 0 + br i1 %tobool, label %if.end, label %if.then + +if.then: ; preds = %for.body + %1 = add nsw i64 %indvars.iv, 45 + %2 = trunc i64 %indvars.iv to i32 + %mul = mul nsw i32 %2, %t + %3 = trunc i64 %1 to i32 + %add1 = add nsw i32 %3, %mul + br label %if.end + +if.end: ; preds = %for.body, %if.then + %z.0 = phi i32 [ %add1, %if.then ], [ 9, %for.body ] + store i32 %z.0, i32* %arrayidx, align 4, !tbaa !0 + %indvars.iv.next = add i64 %indvars.iv, 1 + %lftr.wideiv = trunc i64 %indvars.iv.next to i32 + %exitcond = icmp eq i32 %lftr.wideiv, %x + br i1 %exitcond, label %for.end, label %for.body + +for.end: ; preds = %if.end, %entry + ret i32 undef +} + +!0 = metadata !{metadata !"int", metadata !1} +!1 = metadata !{metadata !"omnipotent char", metadata !2} +!2 = metadata !{metadata !"Simple C/C++ TBAA"} diff --git a/test/Transforms/LoopVectorize/2012-10-20-infloop.ll b/test/Transforms/LoopVectorize/2012-10-20-infloop.ll index 0176c9a189666..aa7cc0ee325d4 100644 --- a/test/Transforms/LoopVectorize/2012-10-20-infloop.ll +++ b/test/Transforms/LoopVectorize/2012-10-20-infloop.ll @@ -1,4 +1,4 @@ -; RUN: opt < %s -loop-vectorize -force-vector-width=4 -dce +; RUN: opt < %s -loop-vectorize -force-vector-unroll=1 -force-vector-width=4 -dce ; Check that we don't fall into an infinite loop. define void @test() nounwind { @@ -25,3 +25,47 @@ for.body: ; preds = %for.body, %entry for.end: ; preds = %for.body unreachable } + +;PR14701 +define void @start_model_rare() nounwind uwtable ssp { +entry: + br i1 undef, label %return, label %if.end + +if.end: ; preds = %entry + br i1 undef, label %cond.false, label %cond.true + +cond.true: ; preds = %if.end + unreachable + +cond.false: ; preds = %if.end + br i1 undef, label %cond.false28, label %cond.true20 + +cond.true20: ; preds = %cond.false + unreachable + +cond.false28: ; preds = %cond.false + br label %for.body40 + +for.body40: ; preds = %for.inc50, %cond.false28 + %indvars.iv123 = phi i64 [ 3, %cond.false28 ], [ %indvars.iv.next124, %for.inc50 ] + %step.0121 = phi i32 [ 1, %cond.false28 ], [ %step.1, %for.inc50 ] + br i1 undef, label %if.then46, label %for.inc50 + +if.then46: ; preds = %for.body40 + %inc47 = add nsw i32 %step.0121, 1 + br label %for.inc50 + +for.inc50: ; preds = %if.then46, %for.body40 + %k.1 = phi i32 [ undef, %for.body40 ], [ %inc47, %if.then46 ] + %step.1 = phi i32 [ %step.0121, %for.body40 ], [ %inc47, %if.then46 ] + %indvars.iv.next124 = add i64 %indvars.iv123, 1 + %lftr.wideiv = trunc i64 %indvars.iv.next124 to i32 + %exitcond = icmp eq i32 %lftr.wideiv, 256 + br i1 %exitcond, label %for.end52, label %for.body40 + +for.end52: ; preds = %for.inc50 + unreachable + +return: ; preds = %entry + ret void +} diff --git a/test/Transforms/LoopVectorize/2012-10-22-isconsec.ll b/test/Transforms/LoopVectorize/2012-10-22-isconsec.ll index 2516e248bc964..405582c408999 100644 --- a/test/Transforms/LoopVectorize/2012-10-22-isconsec.ll +++ b/test/Transforms/LoopVectorize/2012-10-22-isconsec.ll @@ -1,4 +1,4 @@ -; RUN: opt < %s -loop-vectorize -dce -force-vector-width=4 +; RUN: opt < %s -loop-vectorize -dce -force-vector-unroll=1 -force-vector-width=4 ; Check that we don't crash. diff --git a/test/Transforms/LoopVectorize/ARM/arm-unroll.ll b/test/Transforms/LoopVectorize/ARM/arm-unroll.ll new file mode 100644 index 0000000000000..c8d307f5d4438 --- /dev/null +++ b/test/Transforms/LoopVectorize/ARM/arm-unroll.ll @@ -0,0 +1,32 @@ +; RUN: opt < %s -loop-vectorize -mtriple=thumbv7-apple-ios3.0.0 -S | FileCheck %s +; RUN: opt < %s -loop-vectorize -mtriple=thumbv7-apple-ios3.0.0 -mcpu=swift -S | FileCheck %s --check-prefix=SWIFT + +target datalayout = "e-p:32:32:32-i1:8:32-i8:8:32-i16:16:32-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:32:64-v128:32:128-a0:0:32-n32-S32" +target triple = "thumbv7-apple-ios3.0.0" + +;CHECK: @foo +;CHECK: load <4 x i32> +;CHECK-NOT: load <4 x i32> +;CHECK: ret +;SWIFT: @foo +;SWIFT: load <4 x i32> +;SWIFT: load <4 x i32> +;SWIFT: ret +define i32 @foo(i32* nocapture %A, i32 %n) nounwind readonly ssp { + %1 = icmp sgt i32 %n, 0 + br i1 %1, label %.lr.ph, label %._crit_edge + +.lr.ph: ; preds = %0, %.lr.ph + %i.02 = phi i32 [ %5, %.lr.ph ], [ 0, %0 ] + %sum.01 = phi i32 [ %4, %.lr.ph ], [ 0, %0 ] + %2 = getelementptr inbounds i32* %A, i32 %i.02 + %3 = load i32* %2, align 4 + %4 = add nsw i32 %3, %sum.01 + %5 = add nsw i32 %i.02, 1 + %exitcond = icmp eq i32 %5, %n + br i1 %exitcond, label %._crit_edge, label %.lr.ph + +._crit_edge: ; preds = %.lr.ph, %0 + %sum.0.lcssa = phi i32 [ 0, %0 ], [ %4, %.lr.ph ] + ret i32 %sum.0.lcssa +} diff --git a/test/Transforms/LoopVectorize/ARM/gcc-examples.ll b/test/Transforms/LoopVectorize/ARM/gcc-examples.ll new file mode 100644 index 0000000000000..6a68e81bcae0c --- /dev/null +++ b/test/Transforms/LoopVectorize/ARM/gcc-examples.ll @@ -0,0 +1,60 @@ +; RUN: opt < %s -loop-vectorize -mtriple=thumbv7-apple-ios3.0.0 -mcpu=swift -S -dce | FileCheck %s + +target datalayout = "e-p:32:32:32-i1:8:32-i8:8:32-i16:16:32-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:32:64-v128:32:128-a0:0:32-n32-S32" +target triple = "thumbv7-apple-ios3.0.0" + +@b = common global [2048 x i32] zeroinitializer, align 16 +@c = common global [2048 x i32] zeroinitializer, align 16 +@a = common global [2048 x i32] zeroinitializer, align 16 + +; Select VF = 8; +;CHECK: @example1 +;CHECK: load <4 x i32> +;CHECK: add nsw <4 x i32> +;CHECK: store <4 x i32> +;CHECK: ret void +define void @example1() nounwind uwtable ssp { + br label %1 + +;