bpf: generate better lowering code for certain select/setcc instructions

Currently, for code like below,
===
  inner_map = bpf_map_lookup_elem(outer_map, &port_key);
  if (!inner_map) {
    inner_map = &fallback_map;
  }
===
the compiler generates (pseudo) code like the below:
===
  I1: r1 = bpf_map_lookup_elem(outer_map, &port_key);
  I2: r2 = 0
  I3: if (r1 == r2)
  I4:   r6 = &fallback_map
  I5: ...
===

During kernel verification process, After I1, r1 holds a state
map_ptr_or_null. If I3 condition is not taken
(path [I1, I2, I3, I5]), supposedly r1 should become map_ptr.
Unfortunately, kernel does not recognize this pattern
and r1 remains map_ptr_or_null at insn I5. This will cause
verificaiton failure later on.

Kernel, however, is able to recognize pattern "if (r1 == 0)"
properly and give a map_ptr state to r1 in the above case.

LLVM here generates suboptimal code which causes kernel verification
failure. This patch fixes the issue by changing BPF insn pattern
matching and lowering to generate proper codes if the righthand
parameter of the above condition is a constant. A test case
is also added.

Signed-off-by: Yonghong Song <yhs@fb.com>
llvm-svn: 308080

1 files changed, 20 insertions, 26 deletions

diff --git a/llvm/lib/Target/BPF/BPFISelLowering.cpp b/llvm/lib/Target/BPF/BPFISelLowering.cpp
index cc7a7c3849b..c6e4f07aed7 100644
--- a/llvm/lib/Target/BPF/BPFISelLowering.cpp
+++ b/llvm/lib/Target/BPF/BPFISelLowering.cpp
@@ -515,8 +515,10 @@ BPFTargetLowering::EmitInstrWithCustomInserter(MachineInstr &MI,
                                                MachineBasicBlock *BB) const {
   const TargetInstrInfo &TII = *BB->getParent()->getSubtarget().getInstrInfo();
   DebugLoc DL = MI.getDebugLoc();
+  bool isSelectOp = MI.getOpcode() == BPF::Select;
+  bool isSelectRiOp = MI.getOpcode() == BPF::Select_Ri;
 
-  assert(MI.getOpcode() == BPF::Select && "Unexpected instr type to insert");
+  assert((isSelectOp || isSelectRiOp) && "Unexpected instr type to insert");
 
   // To "insert" a SELECT instruction, we actually have to insert the diamond
   // control-flow pattern.  The incoming instruction knows the destination vreg
@@ -548,48 +550,40 @@ BPFTargetLowering::EmitInstrWithCustomInserter(MachineInstr &MI,
 
   // Insert Branch if Flag
   unsigned LHS = MI.getOperand(1).getReg();
-  unsigned RHS = MI.getOperand(2).getReg();
   int CC = MI.getOperand(3).getImm();
+  int NewCC;
   switch (CC) {
   case ISD::SETGT:
-    BuildMI(BB, DL, TII.get(BPF::JSGT_rr))
-        .addReg(LHS)
-        .addReg(RHS)
-        .addMBB(Copy1MBB);
+    NewCC = isSelectOp ? BPF::JSGT_rr : BPF::JSGT_ri;
     break;
   case ISD::SETUGT:
-    BuildMI(BB, DL, TII.get(BPF::JUGT_rr))
-        .addReg(LHS)
-        .addReg(RHS)
-        .addMBB(Copy1MBB);
+    NewCC = isSelectOp ? BPF::JUGT_rr : BPF::JUGT_ri;
     break;
   case ISD::SETGE:
-    BuildMI(BB, DL, TII.get(BPF::JSGE_rr))
-        .addReg(LHS)
-        .addReg(RHS)
-        .addMBB(Copy1MBB);
+    NewCC = isSelectOp ? BPF::JSGE_rr : BPF::JSGE_ri;
     break;
   case ISD::SETUGE:
-    BuildMI(BB, DL, TII.get(BPF::JUGE_rr))
-        .addReg(LHS)
-        .addReg(RHS)
-        .addMBB(Copy1MBB);
+    NewCC = isSelectOp ? BPF::JUGE_rr : BPF::JUGE_ri;
     break;
   case ISD::SETEQ:
-    BuildMI(BB, DL, TII.get(BPF::JEQ_rr))
-        .addReg(LHS)
-        .addReg(RHS)
-        .addMBB(Copy1MBB);
+    NewCC = isSelectOp ? BPF::JEQ_rr : BPF::JEQ_ri;
     break;
   case ISD::SETNE:
-    BuildMI(BB, DL, TII.get(BPF::JNE_rr))
-        .addReg(LHS)
-        .addReg(RHS)
-        .addMBB(Copy1MBB);
+    NewCC = isSelectOp ? BPF::JNE_rr : BPF::JNE_ri;
     break;
   default:
     report_fatal_error("unimplemented select CondCode " + Twine(CC));
   }
+  if (isSelectOp)
+    BuildMI(BB, DL, TII.get(NewCC))
+        .addReg(LHS)
+        .addReg(MI.getOperand(2).getReg())
+        .addMBB(Copy1MBB);
+  else
+    BuildMI(BB, DL, TII.get(NewCC))
+        .addReg(LHS)
+        .addImm(MI.getOperand(2).getImm())
+        .addMBB(Copy1MBB);
 
   // Copy0MBB:
   //  %FalseValue = ...