From debf7d345aa297ebedf2d3b1db234feb44c71483 Mon Sep 17 00:00:00 2001 From: Bill Schmidt Date: Tue, 2 Apr 2013 18:37:08 +0000 Subject: [PATCH] Fix PR15630: Replace faulty stdcx. with stwcx. When doing a partword atomic operation, a lwarx was being paired with a stdcx. instead of a stwcx. when compiling for a 64-bit target. The target has nothing to do with it in this case; we always need a stwcx. Thanks to Kai Nacke for reporting the problem. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@178559 91177308-0d34-0410-b5e6-96231b3b80d8 --- lib/Target/PowerPC/PPCISelLowering.cpp | 2 +- test/CodeGen/PowerPC/pr15630.ll | 16 ++++++++++++++++ 2 files changed, 17 insertions(+), 1 deletion(-) create mode 100644 test/CodeGen/PowerPC/pr15630.ll diff --git a/lib/Target/PowerPC/PPCISelLowering.cpp b/lib/Target/PowerPC/PPCISelLowering.cpp index 0d8bfb98159..ee43fc668e7 100644 --- a/lib/Target/PowerPC/PPCISelLowering.cpp +++ b/lib/Target/PowerPC/PPCISelLowering.cpp @@ -5960,7 +5960,7 @@ PPCTargetLowering::EmitPartwordAtomicBinary(MachineInstr *MI, .addReg(TmpReg).addReg(MaskReg); BuildMI(BB, dl, TII->get(is64bit ? PPC::OR8 : PPC::OR), Tmp4Reg) .addReg(Tmp3Reg).addReg(Tmp2Reg); - BuildMI(BB, dl, TII->get(is64bit ? PPC::STDCX : PPC::STWCX)) + BuildMI(BB, dl, TII->get(PPC::STWCX)) .addReg(Tmp4Reg).addReg(ZeroReg).addReg(PtrReg); BuildMI(BB, dl, TII->get(PPC::BCC)) .addImm(PPC::PRED_NE).addReg(PPC::CR0).addMBB(loopMBB); diff --git a/test/CodeGen/PowerPC/pr15630.ll b/test/CodeGen/PowerPC/pr15630.ll new file mode 100644 index 00000000000..c5ba8a4d4f0 --- /dev/null +++ b/test/CodeGen/PowerPC/pr15630.ll @@ -0,0 +1,16 @@ +; RUN: llc -mcpu=pwr7 -O0 < %s | FileCheck %s + +target datalayout = "E-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v128:128:128-n32:64" +target triple = "powerpc64-unknown-linux-gnu" + +define weak_odr void @_D4core6atomic49__T11atomicStoreVE4core6atomic11MemoryOrder3ThThZ11atomicStoreFNaNbKOhhZv(i8* %val_arg, i8 zeroext %newval_arg) { +entry: + %newval = alloca i8 + %ordering = alloca i32, align 4 + store i8 %newval_arg, i8* %newval + %tmp = load i8* %newval + store atomic volatile i8 %tmp, i8* %val_arg seq_cst, align 1 + ret void +} + +; CHECK: stwcx. -- 2.34.1