[AMDGPU] Handle mandatory literals in isOperandLegal

[foad]

Summary

Extend SIInstrInfo::isOperandLegal to enforce a limit on the number of
literal operands for all VALU instructions, not just VOP3. In particular
it now handles VOP2 instructions with a mandatory literal operand like
V_FMAAK_F32.

[piotr]

Just verified this patch fixes the rendering corruption caused by D114643 - thanks!

llvm/lib/Target/AMDGPU/SIInstrInfo.cpp

bool SIInstrInfo::isOperandLegal(const MachineInstr &MI, unsigned OpIdx,
   const MCInstrDesc &InstDesc = MI.getDesc();
   const MCOperandInfo &OpInfo = InstDesc.OpInfo[OpIdx];
   const TargetRegisterClass *DefinedRC =
       OpInfo.RegClass != -1 ? RI.getRegClass(OpInfo.RegClass) : nullptr;
   if (!MO)
     MO = &MI.getOperand(OpIdx);
 
   int ConstantBusLimit = ST.getConstantBusLimit(MI.getOpcode());
-  int VOP3LiteralLimit = ST.hasVOP3Literal() ? 1 : 0;
+  int LiteralLimit = !isVOP3(MI) || ST.hasVOP3Literal() ? 1 : 0;
   if (isVALU(MI) && usesConstantBus(MRI, *MO, OpInfo)) {
-    if (isVOP3(MI) && isLiteralConstantLike(*MO, OpInfo) && !VOP3LiteralLimit--)
+    if (isLiteralConstantLike(*MO, OpInfo) && !LiteralLimit--)
       return false;
 
     SmallDenseSet<RegSubRegPair> SGPRsUsed;
     if (MO->isReg())
       SGPRsUsed.insert(RegSubRegPair(MO->getReg(), MO->getSubReg()));
 
     for (unsigned i = 0, e = MI.getNumOperands(); i != e; ++i) {
       if (i == OpIdx)
         continue;
       const MachineOperand &Op = MI.getOperand(i);
       if (Op.isReg()) {
         RegSubRegPair SGPR(Op.getReg(), Op.getSubReg());
         if (!SGPRsUsed.count(SGPR) &&
             usesConstantBus(MRI, Op, InstDesc.OpInfo[i])) {
           if (--ConstantBusLimit <= 0)
             return false;
           SGPRsUsed.insert(SGPR);
         }
-      } else if (InstDesc.OpInfo[i].OperandType == AMDGPU::OPERAND_KIMM32) {
-        if (--ConstantBusLimit <= 0)
-          return false;
-      } else if (isVOP3(MI) && AMDGPU::isSISrcOperand(InstDesc, i) &&
-                 isLiteralConstantLike(Op, InstDesc.OpInfo[i])) {
-        if (!VOP3LiteralLimit--)
+      } else if (InstDesc.OpInfo[i].OperandType == AMDGPU::OPERAND_KIMM32 ||
+                 (AMDGPU::isSISrcOperand(InstDesc, i) &&
+                  isLiteralConstantLike(Op, InstDesc.OpInfo[i]))) {
+        if (!LiteralLimit--)
           return false;
         if (--ConstantBusLimit <= 0)
           return false;
       }
     }
   }
 
   if (MO->isReg()) {

llvm/test/CodeGen/AMDGPU/fold-fmaak-bug.ll

-; RUN: not --crash llc -march=amdgcn -mcpu=gfx1010 -verify-machineinstrs < %s 2>&1 | FileCheck %s -check-prefix GFX10
+; RUN: llc -march=amdgcn -mcpu=gfx1010 -verify-machineinstrs < %s | FileCheck %s -check-prefix GFX10
 
 declare <2 x half> @llvm.amdgcn.cvt.pkrtz(float, float)
 declare void @llvm.amdgcn.exp.compr.v2f16(i32 immarg, i32 immarg, <2 x half>, <2 x half>, i1 immarg, i1 immarg)
 
-; FIXME: This instruction uses two different literal constants which is not
-; allowed.
+; Check that this constant is not folded into the v_fmaak_f32 instruction.
 ; GFX10-LABEL: _amdgpu_ps_main:
-; GFX10: Bad machine code: VOP2/VOP3 instruction uses more than one literal
-; GFX10: instruction: %4:vgpr_32 = nnan nsz arcp contract afn reassoc nofpexcept V_FMAAK_F32 1078530011, %0:vgpr_32, -1077342245, implicit $mode, implicit $exec
+; GFX10: v_mov_b32_e32 v1, 0x40490fdb
+; GFX10: v_fmaak_f32 v1, v0, v1, 0xbfc90fdb
 define amdgpu_ps void @_amdgpu_ps_main(float %arg) {
 bb:
   %i = fmul reassoc nnan nsz arcp contract afn float %arg, 0x400921FB60000000
   %i1 = fadd reassoc nnan nsz arcp contract afn float %i, 0xBFF921FB60000000
   %i2 = fmul reassoc nnan nsz arcp contract afn float %i1, %arg
   br label %bb3
 
 bb3: