[llvm] r335675 - [JumpThreading] Don't try to rewrite a use if it's already valid.

Mikael Holmén via llvm-commits llvm-commits at lists.llvm.org
Tue Jul 3 00:04:10 PDT 2018


Hi,

On 06/29/2018 05:28 PM, Michael Zolotukhin wrote:
> Thanks for the reproducer, I’ll take a look!
> 

Sounds good! Did you get anywhere with this?

Regards,
Mikael

> Michael
> 
>> On Jun 29, 2018, at 2:36 AM, Mikael Holmén <mikael.holmen at ericsson.com> wrote:
>>
>> Aaaand here comes the reproducer as well.
>>
>> /Mikael
>>
>> On 06/29/2018 11:36 AM, Mikael Holmén wrote:
>>> Hi Michael,
>>> I get a miscompile from -jump-threading with this patch:
>>> Reproduce with:
>>>   opt -jump-threading -S -o - bbi-15872.ll
>>> %check should be 0 first time we get to %if.end, then we should jump to %label, then %if.end, set %check to 1, and jump to %if.end2 where we should pass %check (which is now 1) to @g.
>>> But after jump-threading the whole thing is replaced with just
>>>    call void @g(i16 0)
>>> If I revert this commit I instead get the expected
>>>    call void @g(i16 1)
>>> Regards,
>>> Mikael
>>> On 06/27/2018 12:19 AM, Michael Zolotukhin via llvm-commits wrote:
>>>> Author: mzolotukhin
>>>> Date: Tue Jun 26 15:19:48 2018
>>>> New Revision: 335675
>>>>
>>>> URL: http://llvm.org/viewvc/llvm-project?rev=335675&view=rev
>>>> Log:
>>>> [JumpThreading] Don't try to rewrite a use if it's already valid.
>>>>
>>>> Summary:
>>>> When recording uses we need to rewrite after cloning a loop we need to
>>>> check if the use is not dominated by the original def. The initial
>>>> assumption was that the cloned basic block will introduce a new path and
>>>> thus the original def will only dominate the use if they are in the same
>>>> BB, but as the reproducer from PR37745 shows it's not always the case.
>>>>
>>>> This fixes PR37745.
>>>>
>>>> Reviewers: haicheng, Ka-Ka
>>>>
>>>> Subscribers: hiraditya, llvm-commits
>>>>
>>>> Differential Revision: https://reviews.llvm.org/D48111
>>>>
>>>> Added:
>>>>       llvm/trunk/test/Transforms/JumpThreading/PR37745.ll
>>>> Modified:
>>>>       llvm/trunk/lib/Transforms/Scalar/JumpThreading.cpp
>>>>
>>>> Modified: llvm/trunk/lib/Transforms/Scalar/JumpThreading.cpp
>>>> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Scalar/JumpThreading.cpp?rev=335675&r1=335674&r2=335675&view=diff
>>>> ==============================================================================
>>>> --- llvm/trunk/lib/Transforms/Scalar/JumpThreading.cpp (original)
>>>> +++ llvm/trunk/lib/Transforms/Scalar/JumpThreading.cpp Tue Jun 26 15:19:48 2018
>>>> @@ -2009,6 +2009,14 @@ bool JumpThreadingPass::ThreadEdge(Basic
>>>>          PredTerm->setSuccessor(i, NewBB);
>>>>        }
>>>> +  // Enqueue required DT updates.
>>>> +  DDT->applyUpdates({{DominatorTree::Insert, NewBB, SuccBB},
>>>> +                     {DominatorTree::Insert, PredBB, NewBB},
>>>> +                     {DominatorTree::Delete, PredBB, BB}});
>>>> +
>>>> +  // Apply all updates we queued with DDT and get the updated Dominator Tree.
>>>> +  DominatorTree *DT = &DDT->flush();
>>>> +
>>>>      // If there were values defined in BB that are used outside the block, then we
>>>>      // now have to update all uses of the value to use either the original value,
>>>>      // the cloned value, or some PHI derived value.  This can require arbitrary
>>>> @@ -2018,16 +2026,16 @@ bool JumpThreadingPass::ThreadEdge(Basic
>>>>      for (Instruction &I : *BB) {
>>>>        SmallVector<Use*, 16> UsesToRename;
>>>> -    // Scan all uses of this instruction to see if it is used outside of its
>>>> -    // block, and if so, record them in UsesToRename. Also, skip phi operands
>>>> -    // from PredBB - we'll remove them anyway.
>>>> +    // Scan all uses of this instruction to see if their uses are no longer
>>>> +    // dominated by the previous def and if so, record them in UsesToRename.
>>>> +    // Also, skip phi operands from PredBB - we'll remove them anyway.
>>>>        for (Use &U : I.uses()) {
>>>>          Instruction *User = cast<Instruction>(U.getUser());
>>>>          if (PHINode *UserPN = dyn_cast<PHINode>(User)) {
>>>>            if (UserPN->getIncomingBlock(U) == BB ||
>>>>                UserPN->getIncomingBlock(U) == PredBB)
>>>>              continue;
>>>> -      } else if (User->getParent() == BB)
>>>> +      } else if (DT->dominates(&I, U))
>>>>            continue;
>>>>          UsesToRename.push_back(&U);
>>>> @@ -2046,12 +2054,6 @@ bool JumpThreadingPass::ThreadEdge(Basic
>>>>          SSAUpdate.AddUse(VarNum, U);
>>>>      }
>>>> -  DDT->applyUpdates({{DominatorTree::Insert, NewBB, SuccBB},
>>>> -                     {DominatorTree::Insert, PredBB, NewBB},
>>>> -                     {DominatorTree::Delete, PredBB, BB}});
>>>> -
>>>> -  // Apply all updates we queued with DDT and get the updated Dominator Tree.
>>>> -  DominatorTree *DT = &DDT->flush();
>>>>      SSAUpdate.RewriteAllUses(DT);
>>>>      // At this point, the IR is fully up to date and consistent.  Do a quick scan
>>>>
>>>> Added: llvm/trunk/test/Transforms/JumpThreading/PR37745.ll
>>>> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/JumpThreading/PR37745.ll?rev=335675&view=auto
>>>> ==============================================================================
>>>> --- llvm/trunk/test/Transforms/JumpThreading/PR37745.ll (added)
>>>> +++ llvm/trunk/test/Transforms/JumpThreading/PR37745.ll Tue Jun 26 15:19:48 2018
>>>> @@ -0,0 +1,19 @@
>>>> +; RUN: opt -jump-threading -verify-each -S -mtriple=x86_64-- -o - %s
>>>> +
>>>> +define void @foo() {
>>>> +entry:
>>>> +  br i1 false, label %A, label %B
>>>> +
>>>> +A:
>>>> +  %x = phi i32 [ undef, %entry ], [ %z, %B ]
>>>> +  br label %B
>>>> +
>>>> +B:
>>>> +  %y = phi i32 [ undef, %entry ], [ %x, %A ]
>>>> +  %z = add i32 %y, 1
>>>> +  %cmp = icmp ne i32 %z, 0
>>>> +  br i1 %cmp, label %exit, label %A
>>>> +
>>>> +exit:
>>>> +  ret void
>>>> +}
>>>>
>>>>
>>>> _______________________________________________
>>>> llvm-commits mailing list
>>>> llvm-commits at lists.llvm.org
>>>> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits
>>>>
>> <bbi-15872.ll>
> 



More information about the llvm-commits mailing list