diff options
| -rw-r--r-- | polly/lib/Analysis/ScopDetection.cpp | 20 | ||||
| -rw-r--r-- | polly/lib/Analysis/ScopInfo.cpp | 15 | ||||
| -rw-r--r-- | polly/test/ScopInfo/memset_null.ll | 37 |
3 files changed, 64 insertions, 8 deletions
diff --git a/polly/lib/Analysis/ScopDetection.cpp b/polly/lib/Analysis/ScopDetection.cpp index 2ceb3fa62dc..7810b94c4e5 100644 --- a/polly/lib/Analysis/ScopDetection.cpp +++ b/polly/lib/Analysis/ScopDetection.cpp @@ -533,17 +533,21 @@ bool ScopDetection::isValidIntrinsicInst(IntrinsicInst &II, case llvm::Intrinsic::memmove: case llvm::Intrinsic::memcpy: AF = SE->getSCEVAtScope(cast<MemTransferInst>(II).getSource(), L); - BP = dyn_cast<SCEVUnknown>(SE->getPointerBase(AF)); - // Bail if the source pointer is not valid. - if (!isValidAccess(&II, AF, BP, Context)) - return false; + if (!AF->isZero()) { + BP = dyn_cast<SCEVUnknown>(SE->getPointerBase(AF)); + // Bail if the source pointer is not valid. + if (!isValidAccess(&II, AF, BP, Context)) + return false; + } // Fall through case llvm::Intrinsic::memset: AF = SE->getSCEVAtScope(cast<MemIntrinsic>(II).getDest(), L); - BP = dyn_cast<SCEVUnknown>(SE->getPointerBase(AF)); - // Bail if the destination pointer is not valid. - if (!isValidAccess(&II, AF, BP, Context)) - return false; + if (!AF->isZero()) { + BP = dyn_cast<SCEVUnknown>(SE->getPointerBase(AF)); + // Bail if the destination pointer is not valid. + if (!isValidAccess(&II, AF, BP, Context)) + return false; + } // Bail if the length is not affine. if (!isAffine(SE->getSCEVAtScope(cast<MemIntrinsic>(II).getLength(), L), L, diff --git a/polly/lib/Analysis/ScopInfo.cpp b/polly/lib/Analysis/ScopInfo.cpp index af56aac719b..471a6669f7a 100644 --- a/polly/lib/Analysis/ScopInfo.cpp +++ b/polly/lib/Analysis/ScopInfo.cpp @@ -3942,8 +3942,17 @@ bool ScopInfo::buildAccessMemIntrinsic( auto *DestPtrVal = MemIntr->getDest(); assert(DestPtrVal); + auto *DestAccFunc = SE->getSCEVAtScope(DestPtrVal, L); assert(DestAccFunc); + // Ignore accesses to "NULL". + // TODO: We could use this to optimize the region further, e.g., intersect + // the context with + // isl_set_complement(isl_set_params(getDomain())) + // as we know it would be undefined to execute this instruction anyway. + if (DestAccFunc->isZero()) + return true; + auto *DestPtrSCEV = dyn_cast<SCEVUnknown>(SE->getPointerBase(DestAccFunc)); assert(DestPtrSCEV); DestAccFunc = SE->getMinusSCEV(DestAccFunc, DestPtrSCEV); @@ -3957,8 +3966,14 @@ bool ScopInfo::buildAccessMemIntrinsic( auto *SrcPtrVal = MemTrans->getSource(); assert(SrcPtrVal); + auto *SrcAccFunc = SE->getSCEVAtScope(SrcPtrVal, L); assert(SrcAccFunc); + // Ignore accesses to "NULL". + // TODO: See above TODO + if (SrcAccFunc->isZero()) + return true; + auto *SrcPtrSCEV = dyn_cast<SCEVUnknown>(SE->getPointerBase(SrcAccFunc)); assert(SrcPtrSCEV); SrcAccFunc = SE->getMinusSCEV(SrcAccFunc, SrcPtrSCEV); diff --git a/polly/test/ScopInfo/memset_null.ll b/polly/test/ScopInfo/memset_null.ll new file mode 100644 index 00000000000..b0d0be0d3b8 --- /dev/null +++ b/polly/test/ScopInfo/memset_null.ll @@ -0,0 +1,37 @@ +; RUN: opt %loadPolly -polly-allow-modref-calls -polly-scops -analyze < %s | FileCheck %s +; RUN: opt %loadPolly -polly-allow-modref-calls -S -polly-codegen < %s +; +; Verify we can handle a memset to "null" and that we do not model it. +; TODO: FIXME: We could use the undefined memset to optimize the code further, +; see the TODOs in the ScopInfo.cpp. +; +; CHECK: Statements { +; CHECK-NEXT: Stmt_for_cond5_preheader_us221 +; CHECK-NEXT: Domain := +; CHECK-NEXT: { Stmt_for_cond5_preheader_us221[0] }; +; CHECK-NEXT: Schedule := +; CHECK-NEXT: { Stmt_for_cond5_preheader_us221[i0] -> [0] }; +; CHECK-NEXT: MustWriteAccess := [Reduction Type: NONE] [Scalar: 0] +; CHECK-NEXT: { Stmt_for_cond5_preheader_us221[i0] -> MemRef_A[0] }; +; CHECK-NEXT: } + +; +target datalayout = "e-m:e-i64:64-i128:128-n8:16:32:64-S128" + +define void @test(i32* %A) { +entry: + br i1 undef, label %for.end68, label %for.cond5.preheader.lr.ph + +for.cond5.preheader.lr.ph: ; preds = %entry + br label %for.cond5.preheader.us221 + +for.cond5.preheader.us221: ; preds = %for.cond5.preheader.us221, %for.cond5.preheader.lr.ph + store i32 0, i32* %A + call void @llvm.memset.p0i8.i64(i8* null, i8 0, i64 undef, i32 1, i1 false) + br i1 true, label %for.end68, label %for.cond5.preheader.us221 + +for.end68: ; preds = %for.cond5.preheader.us221, %entry + ret void +} + +declare void @llvm.memset.p0i8.i64(i8* nocapture, i8, i64, i32, i1) |
