From: Vincent Lejeune Date: Thu, 4 Apr 2013 13:59:59 +0000 (+0000) Subject: R600: Take export into account when computing cf address X-Git-Url: http://demsky.eecs.uci.edu/git/?a=commitdiff_plain;h=39cd6fae34e3cc525bc98b6b0bd24fb8e6202cc5;p=oota-llvm.git R600: Take export into account when computing cf address git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@178761 91177308-0d34-0410-b5e6-96231b3b80d8 --- diff --git a/lib/Target/R600/R600ControlFlowFinalizer.cpp b/lib/Target/R600/R600ControlFlowFinalizer.cpp index 01dd22718ca..850349ee255 100644 --- a/lib/Target/R600/R600ControlFlowFinalizer.cpp +++ b/lib/Target/R600/R600ControlFlowFinalizer.cpp @@ -143,6 +143,10 @@ public: CurrentStack++; MaxStack = std::max(MaxStack, CurrentStack); case AMDGPU::CF_ALU: + case AMDGPU::EG_ExportBuf: + case AMDGPU::EG_ExportSwz: + case AMDGPU::R600_ExportBuf: + case AMDGPU::R600_ExportSwz: DEBUG(dbgs() << CfCount << ":"; MI->dump();); CfCount++; break; diff --git a/test/CodeGen/R600/jump_address.ll b/test/CodeGen/R600/jump_address.ll new file mode 100644 index 00000000000..cd35bffb130 --- /dev/null +++ b/test/CodeGen/R600/jump_address.ll @@ -0,0 +1,50 @@ +;RUN: llc < %s -march=r600 -mcpu=redwood | FileCheck %s + +; CHECK: JUMP @4 + +define void @main() #0 { +main_body: + %0 = load <4 x float> addrspace(8)* getelementptr ([1024 x <4 x float>] addrspace(8)* null, i64 0, i32 1) + %1 = extractelement <4 x float> %0, i32 0 + %2 = bitcast float %1 to i32 + %3 = icmp eq i32 %2, 0 + %4 = sext i1 %3 to i32 + %5 = bitcast i32 %4 to float + %6 = bitcast float %5 to i32 + %7 = icmp ne i32 %6, 0 + br i1 %7, label %ENDIF, label %ELSE + +ELSE: ; preds = %main_body + %8 = load <4 x float> addrspace(8)* getelementptr ([1024 x <4 x float>] addrspace(8)* null, i64 0, i32 1) + %9 = extractelement <4 x float> %8, i32 0 + %10 = bitcast float %9 to i32 + %11 = icmp eq i32 %10, 1 + %12 = sext i1 %11 to i32 + %13 = bitcast i32 %12 to float + %14 = bitcast float %13 to i32 + %15 = icmp ne i32 %14, 0 + br i1 %15, label %IF13, label %ENDIF + +ENDIF: ; preds = %IF13, %ELSE, %main_body + %temp.0 = phi float [ 0xFFF8000000000000, %main_body ], [ 0.000000e+00, %ELSE ], [ 0.000000e+00, %IF13 ] + %temp1.0 = phi float [ 0.000000e+00, %main_body ], [ %23, %IF13 ], [ 0.000000e+00, %ELSE ] + %temp2.0 = phi float [ 1.000000e+00, %main_body ], [ 0.000000e+00, %ELSE ], [ 0.000000e+00, %IF13 ] + %temp3.0 = phi float [ 5.000000e-01, %main_body ], [ 0.000000e+00, %ELSE ], [ 0.000000e+00, %IF13 ] + %16 = insertelement <4 x float> undef, float %temp.0, i32 0 + %17 = insertelement <4 x float> %16, float %temp1.0, i32 1 + %18 = insertelement <4 x float> %17, float %temp2.0, i32 2 + %19 = insertelement <4 x float> %18, float %temp3.0, i32 3 + call void @llvm.R600.store.swizzle(<4 x float> %19, i32 0, i32 0) + ret void + +IF13: ; preds = %ELSE + %20 = load <4 x float> addrspace(8)* null + %21 = extractelement <4 x float> %20, i32 0 + %22 = fsub float -0.000000e+00, %21 + %23 = fadd float 0xFFF8000000000000, %22 + br label %ENDIF +} + +declare void @llvm.R600.store.swizzle(<4 x float>, i32, i32) + +attributes #0 = { "ShaderType"="0" }