From: Puyan Lotfi Date: Fri, 31 May 2019 17:34:25 +0000 (+0000) Subject: [MIR-Canon] Don't do vreg skip for independent instructions if there are none. X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=e58eb5f72632d72f94fc354ec288d8a3b4fb559a;p=llvm [MIR-Canon] Don't do vreg skip for independent instructions if there are none. We don't want to create vregs if there is nothing to use them for. That causes verifier errors. Differential Revision: https://reviews.llvm.org/D62740 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@362247 91177308-0d34-0410-b5e6-96231b3b80d8 --- diff --git a/lib/CodeGen/MIRCanonicalizerPass.cpp b/lib/CodeGen/MIRCanonicalizerPass.cpp index a4097232d7d..c7d1131d7b8 100644 --- a/lib/CodeGen/MIRCanonicalizerPass.cpp +++ b/lib/CodeGen/MIRCanonicalizerPass.cpp @@ -743,7 +743,8 @@ static bool runOnBasicBlock(MachineBasicBlock *MBB, // of the MachineBasicBlock so that they are named in the order that we sorted // them alphabetically. Eventually we wont need SkipVRegs because we will use // named vregs instead. - NVC.SkipVRegs(); + if (IdempotentInstCount) + NVC.SkipVRegs(); auto MII = MBB->begin(); for (unsigned i = 0; i < IdempotentInstCount && MII != MBB->end(); ++i) { diff --git a/test/CodeGen/MIR/AMDGPU/parse-order-reserved-regs.mir b/test/CodeGen/MIR/AMDGPU/parse-order-reserved-regs.mir index abb2dde4d9e..629f7aefd6a 100644 --- a/test/CodeGen/MIR/AMDGPU/parse-order-reserved-regs.mir +++ b/test/CodeGen/MIR/AMDGPU/parse-order-reserved-regs.mir @@ -1,4 +1,5 @@ # RUN: llc -march=amdgcn -run-pass=none -verify-machineinstrs -o - %s | FileCheck %s +# RUN: llc -march=amdgcn -run-pass mir-canonicalizer -verify-machineinstrs -o - %s # Previously getReservedRegs was called before parsing # machineFunctionInfo, but the AMDGPU implementation depends on