[llvm-commits] [llvm] r127766 - in /llvm/trunk: docs/LangRef.html include/llvm/Target/TargetLowering.h lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp lib/Target/X86/X86ISelLowering.cpp lib/Target/X86/X86ISelLowering.h test/CodeGen/X86/bool-zext.ll
Cameron Zwarich
zwarich at apple.com
Wed Mar 16 17:26:50 PDT 2011
I am not too crazy about the name either. I originally had it named something return-specific (since my patch is return-specific), but remembered it also applies to the argument case. Anyone have any suggestions? I will try to think about it and change it to something better...
Cameron
On Mar 16, 2011, at 5:17 PM, Evan Cheng wrote:
> Thanks. But I am not crazy about getTypeForExtendedInteger(). Can you name it better so it's clear it's used for argument passing and return values?
>
> Evan
>
> On Mar 16, 2011, at 3:20 PM, Cameron Zwarich wrote:
>
>> Author: zwarich
>> Date: Wed Mar 16 17:20:18 2011
>> New Revision: 127766
>>
>> URL: http://llvm.org/viewvc/llvm-project?rev=127766&view=rev
>> Log:
>> The x86-64 ABI says that a bool is only guaranteed to be sign-extended to a byte
>> rather than an int. Thankfully, this only causes LLVM to miss optimizations, not
>> generate incorrect code.
>>
>> This just fixes the zext at the return. We still insert an i32 ZextAssert when
>> reading a function's arguments, but it is followed by a truncate and another i8
>> ZextAssert so it is not optimized.
>>
>> Modified:
>> llvm/trunk/docs/LangRef.html
>> llvm/trunk/include/llvm/Target/TargetLowering.h
>> llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp
>> llvm/trunk/lib/Target/X86/X86ISelLowering.cpp
>> llvm/trunk/lib/Target/X86/X86ISelLowering.h
>> llvm/trunk/test/CodeGen/X86/bool-zext.ll
>>
>> Modified: llvm/trunk/docs/LangRef.html
>> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/docs/LangRef.html?rev=127766&r1=127765&r2=127766&view=diff
>> ==============================================================================
>> --- llvm/trunk/docs/LangRef.html (original)
>> +++ llvm/trunk/docs/LangRef.html Wed Mar 16 17:20:18 2011
>> @@ -1017,8 +1017,9 @@
>> <dl>
>> <dt><tt><b>zeroext</b></tt></dt>
>> <dd>This indicates to the code generator that the parameter or return value
>> - should be zero-extended to a 32-bit value by the caller (for a parameter)
>> - or the callee (for a return value).</dd>
>> + should be zero-extended to the extent required by the target's ABI (which
>> + is usually 32-bits, but is 8-bits for a i1 on x86-64) by the caller (for a
>> + parameter) or the callee (for a return value).</dd>
>>
>> <dt><tt><b>signext</b></tt></dt>
>> <dd>This indicates to the code generator that the parameter or return value
>>
>> Modified: llvm/trunk/include/llvm/Target/TargetLowering.h
>> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/Target/TargetLowering.h?rev=127766&r1=127765&r2=127766&view=diff
>> ==============================================================================
>> --- llvm/trunk/include/llvm/Target/TargetLowering.h (original)
>> +++ llvm/trunk/include/llvm/Target/TargetLowering.h Wed Mar 16 17:20:18 2011
>> @@ -1287,6 +1287,17 @@
>> return false;
>> }
>>
>> + /// getTypeForExtendedInteger - Return the type that should be used to zero or
>> + /// sign extend a zeroext/signext integer argument or return value.
>> + /// FIXME: Most C calling convention requires the return type to be promoted,
>> + /// but this is not true all the time, e.g. i1 on x86-64. It is also not
>> + /// necessary for non-C calling conventions. The frontend should handle this
>> + /// and include all of the necessary information.
>> + virtual MVT
>> + getTypeForExtendedInteger(EVT VT, ISD::NodeType ExtendKind) const {
>> + return MVT::i32;
>> + }
>> +
>> /// LowerOperationWrapper - This callback is invoked by the type legalizer
>> /// to legalize nodes with an illegal operand type but legal result types.
>> /// It replaces the LowerOperation callback in the type Legalizer.
>>
>> Modified: llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp
>> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp?rev=127766&r1=127765&r2=127766&view=diff
>> ==============================================================================
>> --- llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp (original)
>> +++ llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp Wed Mar 16 17:20:18 2011
>> @@ -1128,12 +1128,9 @@
>> else if (F->paramHasAttr(0, Attribute::ZExt))
>> ExtendKind = ISD::ZERO_EXTEND;
>>
>> - // FIXME: C calling convention requires the return type to be promoted
>> - // to at least 32-bit. But this is not necessary for non-C calling
>> - // conventions. The frontend should mark functions whose return values
>> - // require promoting with signext or zeroext attributes.
>> if (ExtendKind != ISD::ANY_EXTEND && VT.isInteger()) {
>> - EVT MinVT = TLI.getRegisterType(*DAG.getContext(), MVT::i32);
>> + MVT ReturnMVT = TLI.getTypeForExtendedInteger(VT, ExtendKind);
>> + EVT MinVT = TLI.getRegisterType(*DAG.getContext(), ReturnMVT);
>> if (VT.bitsLT(MinVT))
>> VT = MinVT;
>> }
>>
>> Modified: llvm/trunk/lib/Target/X86/X86ISelLowering.cpp
>> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86ISelLowering.cpp?rev=127766&r1=127765&r2=127766&view=diff
>> ==============================================================================
>> --- llvm/trunk/lib/Target/X86/X86ISelLowering.cpp (original)
>> +++ llvm/trunk/lib/Target/X86/X86ISelLowering.cpp Wed Mar 16 17:20:18 2011
>> @@ -1448,6 +1448,15 @@
>> return HasRet;
>> }
>>
>> +MVT
>> +X86TargetLowering::getTypeForExtendedInteger(EVT VT,
>> + ISD::NodeType ExtendKind) const {
>> + // TODO: Is this also valid on 32-bit?
>> + if (Subtarget->is64Bit() && VT == MVT::i1 && ExtendKind == ISD::ZERO_EXTEND)
>> + return MVT::i8;
>> + return MVT::i32;
>> +}
>> +
>> /// LowerCallResult - Lower the result values of a call into the
>> /// appropriate copies out of appropriate physical registers.
>> ///
>>
>> Modified: llvm/trunk/lib/Target/X86/X86ISelLowering.h
>> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86ISelLowering.h?rev=127766&r1=127765&r2=127766&view=diff
>> ==============================================================================
>> --- llvm/trunk/lib/Target/X86/X86ISelLowering.h (original)
>> +++ llvm/trunk/lib/Target/X86/X86ISelLowering.h Wed Mar 16 17:20:18 2011
>> @@ -843,6 +843,9 @@
>>
>> virtual bool isUsedByReturnOnly(SDNode *N) const;
>>
>> + virtual MVT
>> + getTypeForExtendedInteger(EVT VT, ISD::NodeType ExtendKind) const;
>> +
>> virtual bool
>> CanLowerReturn(CallingConv::ID CallConv, bool isVarArg,
>> const SmallVectorImpl<ISD::OutputArg> &Outs,
>>
>> Modified: llvm/trunk/test/CodeGen/X86/bool-zext.ll
>> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/bool-zext.ll?rev=127766&r1=127765&r2=127766&view=diff
>> ==============================================================================
>> --- llvm/trunk/test/CodeGen/X86/bool-zext.ll (original)
>> +++ llvm/trunk/test/CodeGen/X86/bool-zext.ll Wed Mar 16 17:20:18 2011
>> @@ -6,7 +6,7 @@
>> define void @bar1(i1 zeroext %v1) nounwind ssp {
>> entry:
>> %conv = zext i1 %v1 to i32
>> - %call = tail call i32 (...)* @foo(i32 %conv) nounwind
>> + %call = tail call i32 (...)* @foo1(i32 %conv) nounwind
>> ret void
>> }
>>
>> @@ -16,8 +16,20 @@
>> define void @bar2(i8 zeroext %v1) nounwind ssp {
>> entry:
>> %conv = zext i8 %v1 to i32
>> - %call = tail call i32 (...)* @foo(i32 %conv) nounwind
>> + %call = tail call i32 (...)* @foo1(i32 %conv) nounwind
>> ret void
>> }
>>
>> -declare i32 @foo(...)
>> +; CHECK: @bar3
>> +; CHECK: callq
>> +; CHECK-NOT: movzbl
>> +; CHECK-NOT: and
>> +; CHECK: ret
>> +define zeroext i1 @bar3() nounwind ssp {
>> +entry:
>> + %call = call i1 @foo2() nounwind
>> + ret i1 %call
>> +}
>> +
>> +declare i32 @foo1(...)
>> +declare zeroext i1 @foo2()
>>
>>
>> _______________________________________________
>> llvm-commits mailing list
>> llvm-commits at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>
More information about the llvm-commits
mailing list