From eabb8dd01547c6fcccb7dcbba7a4a3b8ba67694c Mon Sep 17 00:00:00 2001
From: Matt Arsenault <Matthew.Arsenault@amd.com>
Date: Fri, 16 Nov 2018 05:03:02 +0000
Subject: [PATCH] AMDGPU: Fix analyzeBranch failing with pseudoterminators

If a block had one of the _term instructions used for gluing
exec modifying instructions to the end of the block,
analyzeBranch would fail, preventing the verifier from catching
a broken successor list.

llvm-svn: 347027
---
 llvm/lib/Target/AMDGPU/SIInsertSkips.cpp           |  2 +-
 llvm/lib/Target/AMDGPU/SIInstrInfo.cpp             | 29 +++++++++++++++++++++-
 llvm/lib/Target/AMDGPU/SIInstructions.td           |  3 ++-
 .../CodeGen/AMDGPU/verifier-pseudo-terminators.mir | 23 +++++++++++++++++
 4 files changed, 54 insertions(+), 3 deletions(-)
 create mode 100644 llvm/test/CodeGen/AMDGPU/verifier-pseudo-terminators.mir

diff --git a/llvm/lib/Target/AMDGPU/SIInsertSkips.cpp b/llvm/lib/Target/AMDGPU/SIInsertSkips.cpp
index f23fa02..ba21a5c 100644
--- a/llvm/lib/Target/AMDGPU/SIInsertSkips.cpp
+++ b/llvm/lib/Target/AMDGPU/SIInsertSkips.cpp
@@ -476,7 +476,7 @@ bool SIInsertSkips::runOnMachineFunction(MachineFunction &MF) {
         kill(MI);
 
         if (ExecBranchStack.empty()) {
-          if (skipIfDead(MI, *NextBB)) {
+          if (NextBB != BE && skipIfDead(MI, *NextBB)) {
             HaveSkipBlock = true;
             NextBB = std::next(BI);
             BE = MF.end();
diff --git a/llvm/lib/Target/AMDGPU/SIInstrInfo.cpp b/llvm/lib/Target/AMDGPU/SIInstrInfo.cpp
index eb04338..7aef9fd 100644
--- a/llvm/lib/Target/AMDGPU/SIInstrInfo.cpp
+++ b/llvm/lib/Target/AMDGPU/SIInstrInfo.cpp
@@ -1632,7 +1632,34 @@ bool SIInstrInfo::analyzeBranch(MachineBasicBlock &MBB, MachineBasicBlock *&TBB,
                                 SmallVectorImpl<MachineOperand> &Cond,
                                 bool AllowModify) const {
   MachineBasicBlock::iterator I = MBB.getFirstTerminator();
-  if (I == MBB.end())
+  auto E = MBB.end();
+  if (I == E)
+    return false;
+
+  // Skip over the instructions that are artificially terminators for special
+  // exec management.
+  while (I != E && !I->isBranch() && !I->isReturn() &&
+         I->getOpcode() != AMDGPU::SI_MASK_BRANCH) {
+    switch (I->getOpcode()) {
+    case AMDGPU::SI_MASK_BRANCH:
+    case AMDGPU::S_MOV_B64_term:
+    case AMDGPU::S_XOR_B64_term:
+    case AMDGPU::S_ANDN2_B64_term:
+      break;
+    case AMDGPU::SI_IF:
+    case AMDGPU::SI_ELSE:
+    case AMDGPU::SI_KILL_I1_TERMINATOR:
+    case AMDGPU::SI_KILL_F32_COND_IMM_TERMINATOR:
+      // FIXME: It's messy that these need to be considered here at all.
+      return true;
+    default:
+      llvm_unreachable("unexpected non-branch terminator inst");
+    }
+
+    ++I;
+  }
+
+  if (I == E)
     return false;
 
   if (I->getOpcode() != AMDGPU::SI_MASK_BRANCH)
diff --git a/llvm/lib/Target/AMDGPU/SIInstructions.td b/llvm/lib/Target/AMDGPU/SIInstructions.td
index 11136e0..9b0edf4 100644
--- a/llvm/lib/Target/AMDGPU/SIInstructions.td
+++ b/llvm/lib/Target/AMDGPU/SIInstructions.td
@@ -247,7 +247,7 @@ def SI_LOOP : CFPseudoInstSI <
   (outs), (ins SReg_64:$saved, brtarget:$target),
   [(AMDGPUloop i64:$saved, bb:$target)], 1, 1> {
   let Size = 8;
-  let isBranch = 0;
+  let isBranch = 1;
   let hasSideEffects = 1;
 }
 
@@ -307,6 +307,7 @@ def SI_ILLEGAL_COPY : SPseudoInstSI <
 def SI_BR_UNDEF : SPseudoInstSI <(outs), (ins sopp_brtarget:$simm16)> {
   let isTerminator = 1;
   let usesCustomInserter = 1;
+  let isBranch = 1;
 }
 
 def SI_PS_LIVE : PseudoInstSI <
diff --git a/llvm/test/CodeGen/AMDGPU/verifier-pseudo-terminators.mir b/llvm/test/CodeGen/AMDGPU/verifier-pseudo-terminators.mir
new file mode 100644
index 0000000..d9e8aa7
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/verifier-pseudo-terminators.mir
@@ -0,0 +1,23 @@
+# RUN: not llc -march=amdgcn -run-pass=verify %s 2>&1 | FileCheck %s
+# Make sure that mismatched successors are caught when a _term
+# instruction is used
+
+# CHECK: *** Bad machine code: MBB exits via unconditional branch but the CFG successor doesn't match the actual successor! ***
+
+---
+name: verifier_pseudo_terminators
+body:             |
+  bb.0:
+    successors: %bb.1
+
+    %0:sreg_64 = S_XOR_B64_term undef %1:sreg_64, undef %2:sreg_64, implicit-def $scc
+    $exec = S_MOV_B64_term %0
+    S_BRANCH %bb.2
+
+  bb.1:
+    S_SETPC_B64_return undef $sgpr30_sgpr31
+
+  bb.2:
+    S_SETPC_B64_return undef $sgpr30_sgpr31
+
+...
-- 
2.7.4