[llvm] r325130 - [ARM] f16 stack spill/reloads

[Sjoerd Meijer via llvm-commits]

Author: sjoerdmeijer
Date: Wed Feb 14 07:09:09 2018
New Revision: 325130

URL: http://llvm.org/viewvc/llvm-project?rev=325130&view=rev
Log:
[ARM] f16 stack spill/reloads

This adds support for handling f16 stack spills/reloads.

Differential Revision: https://reviews.llvm.org/D43280

Modified:
    llvm/trunk/lib/Target/ARM/ARMBaseInstrInfo.cpp
    llvm/trunk/test/CodeGen/ARM/fp16-instructions.ll

Modified: llvm/trunk/lib/Target/ARM/ARMBaseInstrInfo.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/ARM/ARMBaseInstrInfo.cpp?rev=325130&r1=325129&r2=325130&view=diff
==============================================================================
--- llvm/trunk/lib/Target/ARM/ARMBaseInstrInfo.cpp (original)
+++ llvm/trunk/lib/Target/ARM/ARMBaseInstrInfo.cpp Wed Feb 14 07:09:09 2018
@@ -963,6 +963,17 @@ storeRegToStackSlot(MachineBasicBlock &M
       MFI.getObjectSize(FI), Align);
 
   switch (TRI->getSpillSize(*RC)) {
+    case 2:
+      if (ARM::HPRRegClass.hasSubClassEq(RC)) {
+        BuildMI(MBB, I, DL, get(ARM::VSTRH))
+            .addReg(SrcReg, getKillRegState(isKill))
+            .addFrameIndex(FI)
+            .addImm(0)
+            .addMemOperand(MMO)
+            .add(predOps(ARMCC::AL));
+      } else
+        llvm_unreachable("Unknown reg class!");
+      break;
     case 4:
       if (ARM::GPRRegClass.hasSubClassEq(RC)) {
         BuildMI(MBB, I, DL, get(ARM::STRi12))
@@ -1161,6 +1172,16 @@ loadRegFromStackSlot(MachineBasicBlock &
       MFI.getObjectSize(FI), Align);
 
   switch (TRI->getSpillSize(*RC)) {
+  case 2:
+    if (ARM::HPRRegClass.hasSubClassEq(RC)) {
+      BuildMI(MBB, I, DL, get(ARM::VLDRH), DestReg)
+          .addFrameIndex(FI)
+          .addImm(0)
+          .addMemOperand(MMO)
+          .add(predOps(ARMCC::AL));
+    } else
+      llvm_unreachable("Unknown reg class!");
+    break;
   case 4:
     if (ARM::GPRRegClass.hasSubClassEq(RC)) {
       BuildMI(MBB, I, DL, get(ARM::LDRi12), DestReg)
@@ -1168,7 +1189,6 @@ loadRegFromStackSlot(MachineBasicBlock &
           .addImm(0)
           .addMemOperand(MMO)
           .add(predOps(ARMCC::AL));
-
     } else if (ARM::SPRRegClass.hasSubClassEq(RC)) {
       BuildMI(MBB, I, DL, get(ARM::VLDRS), DestReg)
           .addFrameIndex(FI)

Modified: llvm/trunk/test/CodeGen/ARM/fp16-instructions.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/ARM/fp16-instructions.ll?rev=325130&r1=325129&r2=325130&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/ARM/fp16-instructions.ll (original)
+++ llvm/trunk/test/CodeGen/ARM/fp16-instructions.ll Wed Feb 14 07:09:09 2018
@@ -11,6 +11,10 @@
 ; RUN: llc < %s -mtriple=thumbv7-none-eabi -mattr=+vfp4        | FileCheck %s --check-prefixes=CHECK,CHECK-SOFTFP-FP16
 ; RUN: llc < %s -mtriple=thumbv7-none-eabi -mattr=+fullfp16    | FileCheck %s --check-prefixes=CHECK,CHECK-SOFTFP-FULLFP16
 
+; Test fast-isel
+; RUN: llc < %s -mtriple=arm-none-eabi -mattr=+fullfp16 -O0 | FileCheck %s --check-prefixes=CHECK-SPILL-RELOAD
+; RUN: llc < %s -mtriple=thumbv7-none-eabi -mattr=+fullfp16 -O0 | FileCheck %s --check-prefixes=CHECK-SPILL-RELOAD
+
 ; HARD:
 ; RUN: llc < %s -mtriple=arm-none-eabihf -mattr=+vfp3      | FileCheck %s --check-prefixes=CHECK,CHECK-HARDFP-VFP3
 ; RUN: llc < %s -mtriple=arm-none-eabihf -mattr=+vfp4      | FileCheck %s --check-prefixes=CHECK,CHECK-HARDFP-FP16
@@ -719,3 +723,25 @@ entry:
 ; CHECK-SOFTFP-FULLFP16:  vldr.16     [[S2:s[0-9]]], [sp, #{{.}}]
 ; CHECK-SOFTFP-FULLFP16:  vadd.f16    s{{.}}, [[S2]], [[S0_2]]
 }
+
+; Test function calls to check store/load reg to/from stack
+define i32 @fn1() {
+entry:
+  %coerce = alloca half, align 2
+  %tmp2 = alloca i32, align 4
+  store half 0xH7C00, half* %coerce, align 2
+  %0 = load i32, i32* %tmp2, align 4
+  %call = call i32 bitcast (i32 (...)* @fn2 to i32 (i32)*)(i32 %0)
+  store half 0xH7C00, half* %coerce, align 2
+  %1 = load i32, i32* %tmp2, align 4
+  %call3 = call i32 bitcast (i32 (...)* @fn3 to i32 (i32)*)(i32 %1)
+  ret i32 %call3
+
+; CHECK-SPILL-RELOAD-LABEL: fn1:
+; CHECK-SPILL-RELOAD:       vstr.16 s0, [sp, #{{.}}]  @ 2-byte Spill
+; CHECK-SPILL-RELOAD-NEXT:  bl  fn2
+; CHECK-SPILL-RELOAD-NEXT:  vldr.16 s0, [sp, #{{.}}]  @ 2-byte Reload
+}
+
+declare dso_local i32 @fn2(...)
+declare dso_local i32 @fn3(...)