6502/llvm-6502
1
0
Fork 0
mirror of https://github.com/c64scene-ar/llvm-6502.git synced 2024-12-20 09:30:43 +00:00
Code Issues Projects Releases Wiki Activity
bae20a6353
llvm-6502/test/CodeGen/X86/2008-10-16-SpillerBug.ll
Evan Cheng 206d1856ad Added a linearscan register allocation optimization. When the register allocator spill an interval with multiple uses in the same basic block, it creates a different virtual register for each of the reloads. e.g. 
%reg1498<def> = MOV32rm %reg1024, 1, %reg0, 12, %reg0, Mem:LD(4,4) [sunkaddr39 + 0]
        %reg1506<def> = MOV32rm %reg1024, 1, %reg0, 8, %reg0, Mem:LD(4,4) [sunkaddr42 + 0]
        %reg1486<def> = MOV32rr %reg1506
        %reg1486<def> = XOR32rr %reg1486, %reg1498, %EFLAGS<imp-def,dead>
        %reg1510<def> = MOV32rm %reg1024, 1, %reg0, 4, %reg0, Mem:LD(4,4) [sunkaddr45 + 0]

=>

        %reg1498<def> = MOV32rm %reg2036, 1, %reg0, 12, %reg0, Mem:LD(4,4) [sunkaddr39 + 0]
        %reg1506<def> = MOV32rm %reg2037, 1, %reg0, 8, %reg0, Mem:LD(4,4) [sunkaddr42 + 0]
        %reg1486<def> = MOV32rr %reg1506
        %reg1486<def> = XOR32rr %reg1486, %reg1498, %EFLAGS<imp-def,dead>
        %reg1510<def> = MOV32rm %reg2038, 1, %reg0, 4, %reg0, Mem:LD(4,4) [sunkaddr45 + 0]

From linearscan's point of view, each of reg2036, 2037, and 2038 are separate registers, each is "killed" after a single use. The reloaded register is available and it's often clobbered right away. e.g. In thise case reg1498 is allocated EAX while reg2036 is allocated RAX. This means we end up with multiple reloads from the same stack slot in the same basic block.

Now linearscan recognize there are other reloads from same SS in the same BB. So it'll "downgrade" RAX (and its aliases) after reg2036 is allocated until the next reload (reg2037) is done. This greatly increase the likihood reloads from SS are reused.

This speeds up sha1 from OpenSSL by 5.8%. It is also an across the board win for SPEC2000 and 2006.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@69585 91177308-0d34-0410-b5e6-96231b3b80d8
2009-04-20 08:01:12 +00:00

156 lines
11 KiB
LLVM
Raw Blame History

; RUN: llvm-as < %s | llc -relocation-model=pic -disable-fp-elim -mtriple=i386-apple-darwin | grep {andl.*7.*edx}
%struct.XXDActiveTextureTargets = type { i64, i64, i64, i64, i64, i64 }
%struct.XXDAlphaTest = type { float, i16, i8, i8 }
%struct.XXDArrayRange = type { i8, i8, i8, i8 }
%struct.XXDBlendMode = type { i16, i16, i16, i16, %struct.XXTColor4, i16, i16, i8, i8, i8, i8 }
%struct.XXDClearColor = type { double, %struct.XXTColor4, %struct.XXTColor4, float, i32 }
%struct.XXDClipPlane = type { i32, [6 x %struct.XXTColor4] }
%struct.XXDColorBuffer = type { i16, i8, i8, [8 x i16], i8, i8, i8, i8 }
%struct.XXDColorMatrix = type { [16 x float]*, %struct.XXDImagingCC }
%struct.XXDConvolution = type { %struct.XXTColor4, %struct.XXDImagingCC, i16, i16, [0 x i32], float*, i32, i32 }
%struct.XXDDepthTest = type { i16, i16, i8, i8, i8, i8, double, double }
%struct.XXDFixedFunction = type { %struct.YYToken* }
%struct.XXDFogMode = type { %struct.XXTColor4, float, float, float, float, float, i16, i16, i16, i8, i8 }
%struct.XXDHintMode = type { i16, i16, i16, i16, i16, i16, i16, i16, i16, i16 }
%struct.XXDHistogram = type { %struct.XXTFixedColor4*, i32, i16, i8, i8 }
%struct.XXDImagingCC = type { { float, float }, { float, float }, { float, float }, { float, float } }
%struct.XXDImagingSubset = type { %struct.XXDConvolution, %struct.XXDConvolution, %struct.XXDConvolution, %struct.XXDColorMatrix, %struct.XXDMinmax, %struct.XXDHistogram, %struct.XXDImagingCC, %struct.XXDImagingCC, %struct.XXDImagingCC, %struct.XXDImagingCC, i32, [0 x i32] }
%struct.XXDLight = type { %struct.XXTColor4, %struct.XXTColor4, %struct.XXTColor4, %struct.XXTColor4, %struct.XXTCoord3, float, float, float, float, float, %struct.XXTCoord3, float, %struct.XXTCoord3, float, %struct.XXTCoord3, float, float, float, float, float }
%struct.XXDLightModel = type { %struct.XXTColor4, [8 x %struct.XXDLight], [2 x %struct.XXDMaterial], i32, i16, i16, i16, i8, i8, i8, i8, i8, i8 }
%struct.XXDLightProduct = type { %struct.XXTColor4, %struct.XXTColor4, %struct.XXTColor4 }
%struct.XXDLineMode = type { float, i32, i16, i16, i8, i8, i8, i8 }
%struct.XXDLogicOp = type { i16, i8, i8 }
%struct.XXDMaskMode = type { i32, [3 x i32], i8, i8, i8, i8, i8, i8, i8, i8 }
%struct.XXDMaterial = type { %struct.XXTColor4, %struct.XXTColor4, %struct.XXTColor4, %struct.XXTColor4, float, float, float, float, [8 x %struct.XXDLightProduct], %struct.XXTColor4, [8 x i32] }
%struct.XXDMinmax = type { %struct.XXDMinmaxTable*, i16, i8, i8, [0 x i32] }
%struct.XXDMinmaxTable = type { %struct.XXTColor4, %struct.XXTColor4 }
%struct.XXDMultisample = type { float, i8, i8, i8, i8, i8, i8, i8, i8 }
%struct.XXDPipelineProgramState = type { i8, i8, i8, i8, [0 x i32], %struct.XXTColor4* }
%struct.XXDPixelMap = type { i32*, float*, float*, float*, float*, float*, float*, float*, float*, i32*, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32 }
%struct.XXDPixelMode = type { float, float, %struct.XXDPixelStore, %struct.XXDPixelTransfer, %struct.XXDPixelMap, %struct.XXDImagingSubset, i32, i32 }
%struct.XXDPixelPack = type { i32, i32, i32, i32, i32, i32, i32, i32, i8, i8, i8, i8 }
%struct.XXDPixelStore = type { %struct.XXDPixelPack, %struct.XXDPixelPack }
%struct.XXDPixelTransfer = type { float, float, float, float, float, float, float, float, float, float, i32, i32, float, float, float, float, float, float, float, float, float, float, float, float }
%struct.XXDPointMode = type { float, float, float, float, %struct.XXTCoord3, float, i8, i8, i8, i8, i16, i16, i32, i16, i16 }
%struct.XXDPolygonMode = type { [128 x i8], float, float, i16, i16, i16, i16, i8, i8, i8, i8, i8, i8, i8, i8 }
%struct.XXDRegisterCombiners = type { i8, i8, i8, i8, i32, [2 x %struct.XXTColor4], [8 x %struct.XXDRegisterCombinersPerStageState], %struct.XXDRegisterCombinersFinalStageState }
%struct.XXDRegisterCombinersFinalStageState = type { i8, i8, i8, i8, [7 x %struct.XXDRegisterCombinersPerVariableState] }
%struct.XXDRegisterCombinersPerPortionState = type { [4 x %struct.XXDRegisterCombinersPerVariableState], i8, i8, i8, i8, i16, i16, i16, i16, i16, i16 }
%struct.XXDRegisterCombinersPerStageState = type { [2 x %struct.XXDRegisterCombinersPerPortionState], [2 x %struct.XXTColor4] }
%struct.XXDRegisterCombinersPerVariableState = type { i16, i16, i16, i16 }
%struct.XXDScissorTest = type { %struct.XXTFixedColor4, i8, i8, i8, i8 }
%struct.XXDState = type <{ i16, i16, i16, i16, i32, i32, [256 x %struct.XXTColor4], [128 x %struct.XXTColor4], %struct.XXDViewport, %struct.XXDTransform, %struct.XXDLightModel, %struct.XXDActiveTextureTargets, %struct.XXDAlphaTest, %struct.XXDBlendMode, %struct.XXDClearColor, %struct.XXDColorBuffer, %struct.XXDDepthTest, %struct.XXDArrayRange, %struct.XXDFogMode, %struct.XXDHintMode, %struct.XXDLineMode, %struct.XXDLogicOp, %struct.XXDMaskMode, %struct.XXDPixelMode, %struct.XXDPointMode, %struct.XXDPolygonMode, %struct.XXDScissorTest, i32, %struct.XXDStencilTest, [8 x %struct.XXDTextureMode], [16 x %struct.XXDTextureImageMode], %struct.XXDArrayRange, [8 x %struct.XXDTextureCoordGen], %struct.XXDClipPlane, %struct.XXDMultisample, %struct.XXDRegisterCombiners, %struct.XXDArrayRange, %struct.XXDArrayRange, [3 x %struct.XXDPipelineProgramState], %struct.XXDArrayRange, %struct.XXDTransformFeedback, i32*, %struct.XXDFixedFunction, [3 x i32], [2 x i32] }>
%struct.XXDStencilTest = type { [3 x { i32, i32, i16, i16, i16, i16 }], i32, [4 x i8] }
%struct.XXDTextureCoordGen = type { { i16, i16, %struct.XXTColor4, %struct.XXTColor4 }, { i16, i16, %struct.XXTColor4, %struct.XXTColor4 }, { i16, i16, %struct.XXTColor4, %struct.XXTColor4 }, { i16, i16, %struct.XXTColor4, %struct.XXTColor4 }, i8, i8, i8, i8 }
%struct.XXDTextureImageMode = type { float }
%struct.XXDTextureMode = type { %struct.XXTColor4, i32, i16, i16, i16, i16, i16, i16, i16, i16, i16, i16, i16, i16, i16, i16, i16, i16, float, float, i16, i16, i16, i16, i16, i16, [4 x i16], i8, i8, i8, i8, [3 x float], [4 x float], float, float }
%struct.XXDTextureRec = type opaque
%struct.XXDTransform = type <{ [24 x [16 x float]], [24 x [16 x float]], [16 x float], float, float, float, float, float, i8, i8, i8, i8, i32, i32, i32, i16, i16, i8, i8, i8, i8, i32 }>
%struct.XXDTransformFeedback = type { i8, i8, i8, i8, [0 x i32], [16 x i32], [16 x i32] }
%struct.XXDViewport = type { float, float, float, float, float, float, float, float, float, float, float, float, float, float, float, float, double, double, i32, i32, i32, i32, float, float, float, float }
%struct.XXTColor4 = type { float, float, float, float }
%struct.XXTCoord3 = type { float, float, float }
%struct.XXTFixedColor4 = type { i32, i32, i32, i32 }
%struct.XXVMTextures = type { [16 x %struct.XXDTextureRec*] }
%struct.XXVMVPContext = type { i32 }
%struct.XXVMVPStack = type { i32, i32 }
%struct.YYToken = type { { i16, i16, i32 } }
%struct._XXVMConstants = type { <4 x float>, <4 x float>, <4 x float>, <4 x float>, <4 x float>, <4 x float>, <4 x float>, <4 x float>, <4 x float>, <4 x float>, float, float, float, float, float, float, float, float, float, float, float, float, [256 x float], [4096 x i8], [8 x float], [48 x float], [128 x float], [528 x i8], { void (i8*, i8*, i32, i8*)*, float (float)*, float (float)*, float (float)*, i32 (float)* } }
@llvm.used = appending global [1 x i8*] [ i8* bitcast (void (%struct.XXDState*, <4 x float>*, <4 x float>**, %struct._XXVMConstants*, %struct.YYToken*, %struct.XXVMVPContext*, %struct.XXVMTextures*, %struct.XXVMVPStack*, <4 x float>*, <4 x float>*, <4 x float>*, <4 x float>*, <4 x float>*, <4 x float>*, <4 x float>*, <4 x float>*, [4 x <4 x float>]*, i32*, <4 x i32>*, i64)* @t to i8*) ], section "llvm.metadata" ; <[1 x i8*]*> [#uses=0]
define void @t(%struct.XXDState* %gldst, <4 x float>* %prgrm, <4 x float>** %buffs, %struct._XXVMConstants* %cnstn, %struct.YYToken* %pstrm, %struct.XXVMVPContext* %vmctx, %struct.XXVMTextures* %txtrs, %struct.XXVMVPStack* %vpstk, <4 x float>* %atr0, <4 x float>* %atr1, <4 x float>* %atr2, <4 x float>* %atr3, <4 x float>* %vtx0, <4 x float>* %vtx1, <4 x float>* %vtx2, <4 x float>* %vtx3, [4 x <4 x float>]* %tmpGbl, i32* %oldMsk, <4 x i32>* %adrGbl, i64 %key_token) nounwind {
entry:
%0 = trunc i64 %key_token to i32 ; <i32> [#uses=1]
%1 = getelementptr %struct.YYToken* %pstrm, i32 %0 ; <%struct.YYToken*> [#uses=5]
br label %bb1132
bb51: ; preds = %bb1132
%2 = getelementptr %struct.YYToken* %1, i32 %operation.0.rec, i32 0, i32 0 ; <i16*> [#uses=1]
%3 = load i16* %2, align 1 ; <i16> [#uses=3]
%4 = lshr i16 %3, 6 ; <i16> [#uses=1]
%5 = trunc i16 %4 to i8 ; <i8> [#uses=1]
%6 = zext i8 %5 to i32 ; <i32> [#uses=1]
%7 = trunc i16 %3 to i8 ; <i8> [#uses=1]
%8 = and i8 %7, 7 ; <i8> [#uses=1]
%mask5556 = zext i8 %8 to i32 ; <i32> [#uses=3]
%.sum1324 = add i32 %mask5556, 2 ; <i32> [#uses=1]
%.rec = add i32 %operation.0.rec, %.sum1324 ; <i32> [#uses=1]
%9 = bitcast %struct.YYToken* %operation.0 to i32* ; <i32*> [#uses=1]
%10 = load i32* %9, align 1 ; <i32> [#uses=1]
%11 = lshr i32 %10, 16 ; <i32> [#uses=2]
%12 = trunc i32 %11 to i8 ; <i8> [#uses=1]
%13 = and i8 %12, 1 ; <i8> [#uses=1]
%14 = lshr i16 %3, 15 ; <i16> [#uses=1]
%15 = trunc i16 %14 to i8 ; <i8> [#uses=1]
%16 = or i8 %13, %15 ; <i8> [#uses=1]
%17 = icmp eq i8 %16, 0 ; <i1> [#uses=1]
br i1 %17, label %bb94, label %bb75
bb75: ; preds = %bb51
%18 = getelementptr %struct.YYToken* %1, i32 0, i32 0, i32 0 ; <i16*> [#uses=1]
%19 = load i16* %18, align 4 ; <i16> [#uses=1]
%20 = load i16* null, align 2 ; <i16> [#uses=1]
%21 = zext i16 %19 to i64 ; <i64> [#uses=1]
%22 = zext i16 %20 to i64 ; <i64> [#uses=1]
%23 = shl i64 %22, 16 ; <i64> [#uses=1]
%.ins1177 = or i64 %23, %21 ; <i64> [#uses=1]
%.ins1175 = or i64 %.ins1177, 0 ; <i64> [#uses=1]
%24 = and i32 %11, 1 ; <i32> [#uses=1]
%.neg1333 = sub i32 %mask5556, %24 ; <i32> [#uses=1]
%.neg1335 = sub i32 %.neg1333, 0 ; <i32> [#uses=1]
%25 = sub i32 %.neg1335, 0 ; <i32> [#uses=1]
br label %bb94
bb94: ; preds = %bb75, %bb51
%extraToken.0 = phi i64 [ %.ins1175, %bb75 ], [ %extraToken.1, %bb51 ] ; <i64> [#uses=1]
%argCount.0 = phi i32 [ %25, %bb75 ], [ %mask5556, %bb51 ] ; <i32> [#uses=1]
%operation.0.sum1392 = add i32 %operation.0.rec, 1 ; <i32> [#uses=2]
%26 = getelementptr %struct.YYToken* %1, i32 %operation.0.sum1392, i32 0, i32 0 ; <i16*> [#uses=1]
%27 = load i16* %26, align 4 ; <i16> [#uses=1]
%28 = getelementptr %struct.YYToken* %1, i32 %operation.0.sum1392, i32 0, i32 1 ; <i16*> [#uses=1]
%29 = load i16* %28, align 2 ; <i16> [#uses=1]
store i16 %27, i16* null, align 8
store i16 %29, i16* null, align 2
br i1 false, label %bb1132, label %bb110
bb110: ; preds = %bb94
switch i32 %6, label %bb1078 [
i32 30, label %bb960
i32 32, label %bb801
i32 38, label %bb809
i32 78, label %bb1066
]
bb801: ; preds = %bb110
unreachable
bb809: ; preds = %bb110
unreachable
bb960: ; preds = %bb110
%30 = icmp eq i32 %argCount.0, 1 ; <i1> [#uses=1]
br i1 %30, label %bb962, label %bb965
bb962: ; preds = %bb960
unreachable
bb965: ; preds = %bb960
unreachable
bb1066: ; preds = %bb110
unreachable
bb1078: ; preds = %bb110
unreachable
bb1132: ; preds = %bb94, %entry
%extraToken.1 = phi i64 [ undef, %entry ], [ %extraToken.0, %bb94 ] ; <i64> [#uses=1]
%operation.0.rec = phi i32 [ 0, %entry ], [ %.rec, %bb94 ] ; <i32> [#uses=4]
%operation.0 = getelementptr %struct.YYToken* %1, i32 %operation.0.rec ; <%struct.YYToken*> [#uses=1]
br i1 false, label %bb1134, label %bb51
bb1134: ; preds = %bb1132
ret void
}
Reference in New Issue View Git Blame Copy Permalink