[PATCH] D43280: [ARM] f16 stack spill/reloads

[Sjoerd Meijer via Phabricator via llvm-commits]

SjoerdMeijer created this revision.
SjoerdMeijer added reviewers: olista01, samparker.
Herald added subscribers: kristof.beyls, javed.absar.

This adds support for handling f16 stack spills/reloads.


https://reviews.llvm.org/D43280

Files:
  lib/Target/ARM/ARMBaseInstrInfo.cpp
  test/CodeGen/ARM/fp16-instructions.ll


Index: test/CodeGen/ARM/fp16-instructions.ll
===================================================================
--- test/CodeGen/ARM/fp16-instructions.ll
+++ test/CodeGen/ARM/fp16-instructions.ll
@@ -7,6 +7,8 @@
 ; RUN: llc < %s -mtriple=arm-none-eabi -mattr=+vfp4        | FileCheck %s --check-prefixes=CHECK,CHECK-SOFTFP-FP16
 ; RUN: llc < %s -mtriple=arm-none-eabi -mattr=+fullfp16    | FileCheck %s --check-prefixes=CHECK,CHECK-SOFTFP-FULLFP16
 
+; 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=+vfp3        | FileCheck %s --check-prefixes=CHECK,CHECK-SOFTFP-VFP3
 ; 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
@@ -714,3 +716,36 @@
 ; CHECK-LABEL:            ThumbAddrMode5FP16
 ; CHECK-SOFTFP-FULLFP16:  vstr.16 s0, [sp, #2]
 }
+
+; Test function calls to check store/load reg to/from stack
+define i32 @fn1() {
+entry:
+  %retval = alloca i32, align 4
+  %coerce = alloca half, align 2
+  %tmp = alloca i32, align 4
+  %coerce1 = alloca half, align 2
+  %tmp2 = alloca i32, align 4
+  store half 0xH7C00, half* %coerce, align 2
+  %0 = bitcast i32* %tmp to i8*
+  %1 = bitcast half* %coerce to i8*
+  call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %0, i8* align 2 %1, i32 2, i1 false)
+  %2 = load i32, i32* %tmp, align 4
+  %call = call i32 bitcast (i32 (...)* @fn2 to i32 (i32)*)(i32 %2)
+  store half 0xH7C00, half* %coerce1, align 2
+  %3 = bitcast i32* %tmp2 to i8*
+  %4 = bitcast half* %coerce1 to i8*
+  call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %3, i8* align 2 %4, i32 2, i1 false)
+  %5 = load i32, i32* %tmp2, align 4
+  %call3 = call i32 bitcast (i32 (...)* @fn3 to i32 (i32)*)(i32 %5)
+  %6 = load i32, i32* %retval, align 4
+  ret i32 %6
+
+; 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, #8]      @ 2-byte Reload
+}
+
+declare dso_local i32 @fn2(...) #1
+declare void @llvm.memcpy.p0i8.p0i8.i32(i8* nocapture writeonly, i8* nocapture readonly, i32, i1) #2
+declare dso_local i32 @fn3(...) #1
Index: lib/Target/ARM/ARMBaseInstrInfo.cpp
===================================================================
--- lib/Target/ARM/ARMBaseInstrInfo.cpp
+++ lib/Target/ARM/ARMBaseInstrInfo.cpp
@@ -963,6 +963,17 @@
       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))
@@ -1161,14 +1172,23 @@
       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)
           .addFrameIndex(FI)
           .addImm(0)
           .addMemOperand(MMO)
           .add(predOps(ARMCC::AL));
-
     } else if (ARM::SPRRegClass.hasSubClassEq(RC)) {
       BuildMI(MBB, I, DL, get(ARM::VLDRS), DestReg)
           .addFrameIndex(FI)


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D43280.134171.patch
Type: text/x-patch
Size: 3913 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20180214/6733be5f/attachment.bin>