From a277dc299ed5f07fafc138f91891d94904287bad Mon Sep 17 00:00:00 2001 From: Igor Laevsky Date: Tue, 10 Nov 2015 14:10:31 +0000 Subject: [PATCH] Strip metadata when speculatively hoisting instructions This is fix for PR24059. When we are hoisting instruction above some condition it may turn out that metadata on this instruction was control dependant on the condition. This metadata becomes invalid and we need to drop it. This patch should cover most obvious places of speculative execution (which I have found by greping isSafeToSpeculativelyExecute). I think there are more cases but at least this change covers the severe ones. Differential Revision: http://reviews.llvm.org/D14398 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@252604 91177308-0d34-0410-b5e6-96231b3b80d8 --- lib/Analysis/LoopInfo.cpp | 7 +++ lib/Transforms/Scalar/LICM.cpp | 4 ++ lib/Transforms/Utils/SimplifyCFG.cpp | 5 +++ test/Analysis/TypeBasedAliasAnalysis/licm.ll | 2 +- test/Transforms/LICM/hoist-deref-load.ll | 44 +++++++++++++++++++ .../Transforms/SimplifyCFG/SpeculativeExec.ll | 26 +++++++++++ 6 files changed, 87 insertions(+), 1 deletion(-) diff --git a/lib/Analysis/LoopInfo.cpp b/lib/Analysis/LoopInfo.cpp index 9ee72361c92..e679b7ad7b8 100644 --- a/lib/Analysis/LoopInfo.cpp +++ b/lib/Analysis/LoopInfo.cpp @@ -120,6 +120,13 @@ bool Loop::makeLoopInvariant(Instruction *I, bool &Changed, // Hoist. I->moveBefore(InsertPt); + + // There is possibility of hoisting this instruction above some arbitrary + // condition. Any metadata defined on it can be control dependent on this + // condition. Conservatively strip it here so that we don't give any wrong + // information to the optimizer. + I->dropUnknownNonDebugMetadata(); + Changed = true; return true; } diff --git a/lib/Transforms/Scalar/LICM.cpp b/lib/Transforms/Scalar/LICM.cpp index ba39afd4287..6d70cdc3ade 100644 --- a/lib/Transforms/Scalar/LICM.cpp +++ b/lib/Transforms/Scalar/LICM.cpp @@ -672,6 +672,10 @@ static bool hoist(Instruction &I, BasicBlock *Preheader) { // Move the new node to the Preheader, before its terminator. I.moveBefore(Preheader->getTerminator()); + // Metadata can be dependent on the condition we are hoisting above. + // Conservatively strip all metadata on the instruction. + I.dropUnknownNonDebugMetadata(); + if (isa(I)) ++NumMovedLoads; else if (isa(I)) ++NumMovedCalls; ++NumHoisted; diff --git a/lib/Transforms/Utils/SimplifyCFG.cpp b/lib/Transforms/Utils/SimplifyCFG.cpp index e0c598f92e2..ff81e7d5acf 100644 --- a/lib/Transforms/Utils/SimplifyCFG.cpp +++ b/lib/Transforms/Utils/SimplifyCFG.cpp @@ -1618,6 +1618,11 @@ static bool SpeculativelyExecuteBB(BranchInst *BI, BasicBlock *ThenBB, SpeculatedStore->setOperand(0, S); } + // Metadata can be dependent on the condition we are hoisting above. + // Conservatively strip all metadata on the instruction. + for (auto &I: *ThenBB) + I.dropUnknownNonDebugMetadata(); + // Hoist the instructions. BB->getInstList().splice(BI->getIterator(), ThenBB->getInstList(), ThenBB->begin(), std::prev(ThenBB->end())); diff --git a/test/Analysis/TypeBasedAliasAnalysis/licm.ll b/test/Analysis/TypeBasedAliasAnalysis/licm.ll index fe07730577e..d2aee58204d 100644 --- a/test/Analysis/TypeBasedAliasAnalysis/licm.ll +++ b/test/Analysis/TypeBasedAliasAnalysis/licm.ll @@ -5,7 +5,7 @@ ; CHECK: @foo ; CHECK: entry: -; CHECK-NEXT: %tmp3 = load double*, double** @P, !tbaa !0 +; CHECK-NEXT: %tmp3 = load double*, double** @P ; CHECK-NEXT: br label %for.body @P = common global double* null diff --git a/test/Transforms/LICM/hoist-deref-load.ll b/test/Transforms/LICM/hoist-deref-load.ll index c384a858d1e..fd10c5d7503 100644 --- a/test/Transforms/LICM/hoist-deref-load.ll +++ b/test/Transforms/LICM/hoist-deref-load.ll @@ -388,5 +388,49 @@ for.end: ; preds = %for.inc, %entry ret void } +; In this test we should be able to only hoist load from %cptr. We can't hoist +; load from %c because it's dereferenceability can depend on %cmp1 condition. +; By moving it out of the loop we break this dependency and can not rely +; on the dereferenceability anymore. +; In other words this test checks that we strip dereferenceability metadata +; after hoisting an instruction. + +; CHECK-LABEL: @test10 +; CHECK: %c = load i32*, i32** %cptr +; CHECK-NOT: dereferenceable +; CHECK: if.then: +; CHECK: load i32, i32* %c, align 4 + +define void @test10(i32* noalias %a, i32* %b, i32** dereferenceable(8) %cptr, i32 %n) #0 { +entry: + %cmp11 = icmp sgt i32 %n, 0 + br i1 %cmp11, label %for.body, label %for.end + +for.body: ; preds = %entry, %for.inc + %indvars.iv = phi i64 [ %indvars.iv.next, %for.inc ], [ 0, %entry ] + %arrayidx = getelementptr inbounds i32, i32* %a, i64 %indvars.iv + %0 = load i32, i32* %arrayidx, align 4 + %cmp1 = icmp sgt i32 %0, 0 + br i1 %cmp1, label %if.then, label %for.inc + +if.then: ; preds = %for.body + %c = load i32*, i32** %cptr, !dereferenceable !0 + %1 = load i32, i32* %c, align 4 + %arrayidx3 = getelementptr inbounds i32, i32* %b, i64 %indvars.iv + %2 = load i32, i32* %arrayidx3, align 4 + %mul = mul nsw i32 %2, %1 + store i32 %mul, i32* %arrayidx, align 4 + br label %for.inc + +for.inc: ; preds = %for.body, %if.then + %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1 + %lftr.wideiv = trunc i64 %indvars.iv.next to i32 + %exitcond = icmp eq i32 %lftr.wideiv, %n + br i1 %exitcond, label %for.end, label %for.body + +for.end: ; preds = %for.inc, %entry + ret void +} + attributes #0 = { nounwind uwtable } !0 = !{i64 4} diff --git a/test/Transforms/SimplifyCFG/SpeculativeExec.ll b/test/Transforms/SimplifyCFG/SpeculativeExec.ll index c23a96df52e..73f9a0f88ac 100644 --- a/test/Transforms/SimplifyCFG/SpeculativeExec.ll +++ b/test/Transforms/SimplifyCFG/SpeculativeExec.ll @@ -69,3 +69,29 @@ end: ret i8* %x10 } + +define i32* @test5(i32 %a, i32 %b, i32 %c, i32* dereferenceable(10) %ptr1, + i32* dereferenceable(10) %ptr2, i32** dereferenceable(10) %ptr3) nounwind { +; CHECK-LABEL: @test5( +entry: + %tmp1 = icmp eq i32 %b, 0 + br i1 %tmp1, label %bb1, label %bb3 + +bb1: ; preds = %entry + %tmp2 = icmp sgt i32 %c, 1 + br i1 %tmp2, label %bb2, label %bb3 +; CHECK: bb1: +; CHECK-NEXT: icmp sgt i32 %c, 1 +; CHECK-NEXT: load i32*, i32** %ptr3 +; CHECK-NOT: dereferenceable +; CHECK-NEXT: select i1 %tmp2, i32* %tmp3, i32* %ptr2 +; CHECK-NEXT: ret i32* %tmp3.ptr2 + +bb2: ; preds = bb1 + %tmp3 = load i32*, i32** %ptr3, !dereferenceable !{i64 10} + br label %bb3 + +bb3: ; preds = %bb2, %entry + %tmp4 = phi i32* [ %ptr1, %entry ], [ %ptr2, %bb1 ], [ %tmp3, %bb2 ] + ret i32* %tmp4 +} -- 2.34.1