[AArch64] Properly handle dllimport of variables when using fast-isel

[compnerd]

LGTM, but, I think that @aemerson should give the final approval on this.

[mstorsjo]

Ping @aemerson - can you give this an ack from a generic aarch64 target perspective - the windows side of it was already ok'd.

[aemerson]

Looks fine.

[mstorsjo]

@hans This would be good to have backported.

[hans]

In D42567#992230, @mstorsjo wrote:

@hans This would be good to have backported.

Merged in r323852.

lib/Target/AArch64/AArch64FastISel.cpp

unsigned AArch64FastISel::materializeGV(const GlobalValue *GV) {
 
   unsigned ADRPReg = createResultReg(&AArch64::GPR64commonRegClass);
   unsigned ResultReg;
 
   if (OpFlags & AArch64II::MO_GOT) {
     // ADRP + LDRX
     BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, TII.get(AArch64::ADRP),
             ADRPReg)
-      .addGlobalAddress(GV, 0, AArch64II::MO_GOT | AArch64II::MO_PAGE);
+        .addGlobalAddress(GV, 0, AArch64II::MO_PAGE | OpFlags);
 
     ResultReg = createResultReg(&AArch64::GPR64RegClass);
     BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, TII.get(AArch64::LDRXui),
             ResultReg)
         .addReg(ADRPReg)
-      .addGlobalAddress(GV, 0, AArch64II::MO_GOT | AArch64II::MO_PAGEOFF |
-                        AArch64II::MO_NC);
+        .addGlobalAddress(GV, 0,
+                          AArch64II::MO_PAGEOFF | AArch64II::MO_NC | OpFlags);
   } else {
     // ADRP + ADDX
     BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, TII.get(AArch64::ADRP),
             ADRPReg)
-      .addGlobalAddress(GV, 0, AArch64II::MO_PAGE);
+        .addGlobalAddress(GV, 0, AArch64II::MO_PAGE | OpFlags);
 
     ResultReg = createResultReg(&AArch64::GPR64spRegClass);
     BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc, TII.get(AArch64::ADDXri),
             ResultReg)
         .addReg(ADRPReg)
-      .addGlobalAddress(GV, 0, AArch64II::MO_PAGEOFF | AArch64II::MO_NC)
+        .addGlobalAddress(GV, 0,
+                          AArch64II::MO_PAGEOFF | AArch64II::MO_NC | OpFlags)
         .addImm(0);
   }
   return ResultReg;
 }
 
 unsigned AArch64FastISel::fastMaterializeConstant(const Constant *C) {
   EVT CEVT = TLI.getValueType(DL, C->getType(), true);
 
   // Only handle simple types.

lib/Target/AArch64/AArch64Subtarget.cpp

 unsigned char
 AArch64Subtarget::ClassifyGlobalReference(const GlobalValue *GV,
                                           const TargetMachine &TM) const {
   // MachO large model always goes via a GOT, simply to get a single 8-byte
   // absolute relocation on all global addresses.
   if (TM.getCodeModel() == CodeModel::Large && isTargetMachO())
     return AArch64II::MO_GOT;
 
+  unsigned Flags = GV->hasDLLImportStorageClass() ? AArch64II::MO_DLLIMPORT
+                                                  : AArch64II::MO_NO_FLAG;
+
   if (!TM.shouldAssumeDSOLocal(*GV->getParent(), GV))
-    return AArch64II::MO_GOT;
+    return AArch64II::MO_GOT | Flags;
 
   // The small code model's direct accesses use ADRP, which cannot
   // necessarily produce the value 0 (if the code is above 4GB).
   if (useSmallAddressing() && GV->hasExternalWeakLinkage())
-    return AArch64II::MO_GOT;
+    return AArch64II::MO_GOT | Flags;
 
-  return AArch64II::MO_NO_FLAG;
+  return Flags;
 }
 
 unsigned char AArch64Subtarget::classifyGlobalFunctionReference(
     const GlobalValue *GV, const TargetMachine &TM) const {
   // MachO large model always goes via a GOT, because we don't have the
   // relocations available to do anything else..
   if (TM.getCodeModel() == CodeModel::Large && isTargetMachO() &&
       !GV->hasInternalLinkage())

test/CodeGen/AArch64/dllimport.ll

-; RUN: llc -mtriple aarch64-unknown-windows-msvc -filetype asm -o - %s | FileCheck %s
+; RUN: llc -mtriple aarch64-unknown-windows-msvc -filetype asm -o - %s | FileCheck %s -check-prefixes=CHECK,DAG-ISEL
+; RUN: llc -mtriple aarch64-unknown-windows-msvc -fast-isel -filetype asm -o - %s | FileCheck %s -check-prefixes=CHECK,FAST-ISEL
 
 @var = external dllimport global i32
 @ext = external global i32
 declare dllimport i32 @external()
 declare i32 @internal()
 
 define i32 @get_var() {
   %1 = load i32, i32* @var, align 4
   ret i32 %1
 }
 
 ; CHECK-LABEL: get_var
 ; CHECK: adrp x8, __imp_var
 ; CHECK: ldr x8, [x8, __imp_var]
 ; CHECK: ldr w0, [x8]
 ; CHECK: ret
 
 define i32 @get_ext() {
   %1 = load i32, i32* @ext, align 4
   ret i32 %1
 }
 
 ; CHECK-LABEL: get_ext
 ; CHECK: adrp x8, ext
-; CHECK: ldr w0, [x8, ext]
+; DAG-ISEL: ldr w0, [x8, ext]
+; FAST-ISEL: add x8, x8, ext
+; FAST-ISEL: ldr w0, [x8]
 ; CHECK: ret
 
 define i32* @get_var_pointer() {
   ret i32* @var
 }
 
 ; CHECK-LABEL: get_var_pointer
-; CHECK: adrp x0, __imp_var
-; CHECK: ldr x0, [x0, __imp_var]
+; CHECK: adrp [[REG1:x[0-9]+]], __imp_var
+; CHECK: ldr {{x[0-9]+}}, {{\[}}[[REG1]], __imp_var]
 ; CHECK: ret
 
 define i32 @call_external() {
   %call = tail call i32 @external()
   ret i32 %call
 }
 
 ; CHECK-LABEL: call_external