Implement initial-exec TLS model for 32-bit PIC x86

This fixes a TODO from 2007 :) Previously, LLVM would emit the wrong
code here (see the update to test/CodeGen/X86/tls-pie.ll).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@156611 91177308-0d34-0410-b5e6-96231b3b80d8

5 files changed, 48 insertions, 20 deletions

diff --git a/lib/Target/X86/MCTargetDesc/X86BaseInfo.h b/lib/Target/X86/MCTargetDesc/X86BaseInfo.h
index 82ec6fb20f..0c7f14d01c 100644
--- a/lib/Target/X86/MCTargetDesc/X86BaseInfo.h
+++ b/lib/Target/X86/MCTargetDesc/X86BaseInfo.h
@@ -128,6 +128,13 @@ namespace X86II {
     ///    SYMBOL_LABEL @NTPOFF
     MO_NTPOFF,
 
+    /// MO_GOTNTPOFF - On a symbol operand this indicates that the immediate is
+    /// some TLS offset.
+    ///
+    /// See 'ELF Handling for Thread-Local Storage' for more details.
+    ///    SYMBOL_LABEL @GOTNTPOFF
+    MO_GOTNTPOFF,
+
     /// MO_DLLIMPORT - On a symbol operand "FOO", this indicates that the
     /// reference is actually to the "__imp_FOO" symbol.  This is used for
     /// dllimport linkage on windows.
diff --git a/lib/Target/X86/X86AsmPrinter.cpp b/lib/Target/X86/X86AsmPrinter.cpp
index 7db7ccbedc..6d97f46e9a 100644
--- a/lib/Target/X86/X86AsmPrinter.cpp
+++ b/lib/Target/X86/X86AsmPrinter.cpp
@@ -190,6 +190,7 @@ void X86AsmPrinter::printSymbolOperand(const MachineOperand &MO,
   case X86II::MO_INDNTPOFF: O << "@INDNTPOFF"; break;
   case X86II::MO_TPOFF:     O << "@TPOFF";     break;
   case X86II::MO_NTPOFF:    O << "@NTPOFF";    break;
+  case X86II::MO_GOTNTPOFF: O << "@GOTNTPOFF"; break;
   case X86II::MO_GOTPCREL:  O << "@GOTPCREL";  break;
   case X86II::MO_GOT:       O << "@GOT";       break;
   case X86II::MO_GOTOFF:    O << "@GOTOFF";    break;
diff --git a/lib/Target/X86/X86ISelLowering.cpp b/lib/Target/X86/X86ISelLowering.cpp
index 688c25c6a2..75e05883b6 100644
--- a/lib/Target/X86/X86ISelLowering.cpp
+++ b/lib/Target/X86/X86ISelLowering.cpp
@@ -7286,11 +7286,10 @@ LowerToTLSGeneralDynamicModel64(GlobalAddressSDNode *GA, SelectionDAG &DAG,
                     X86::RAX, X86II::MO_TLSGD);
 }
 
-// Lower ISD::GlobalTLSAddress using the "initial exec" (for no-pic) or
-// "local exec" model.
+// Lower ISD::GlobalTLSAddress using the "initial exec" or "local exec" model.
 static SDValue LowerToTLSExecModel(GlobalAddressSDNode *GA, SelectionDAG &DAG,
                                    const EVT PtrVT, TLSModel::Model model,
-                                   bool is64Bit) {
+                                   bool is64Bit, bool isPIC) {
   DebugLoc dl = GA->getDebugLoc();
 
   // Get the Thread Pointer, which is %gs:0 (32-bit) or %fs:0 (64-bit).
@@ -7308,25 +7307,36 @@ static SDValue LowerToTLSExecModel(GlobalAddressSDNode *GA, SelectionDAG &DAG,
   unsigned WrapperKind = X86ISD::Wrapper;
   if (model == TLSModel::LocalExec) {
     OperandFlags = is64Bit ? X86II::MO_TPOFF : X86II::MO_NTPOFF;
-  } else if (is64Bit) {
-    assert(model == TLSModel::InitialExec);
-    OperandFlags = X86II::MO_GOTTPOFF;
-    WrapperKind = X86ISD::WrapperRIP;
+  } else if (model == TLSModel::InitialExec) {
+    if (is64Bit) {
+      OperandFlags = X86II::MO_GOTTPOFF;
+      WrapperKind = X86ISD::WrapperRIP;
+    } else {
+      OperandFlags = isPIC ? X86II::MO_GOTNTPOFF : X86II::MO_INDNTPOFF;
+    }
   } else {
-    assert(model == TLSModel::InitialExec);
-    OperandFlags = X86II::MO_INDNTPOFF;
+    llvm_unreachable("Unexpected model");
   }
 
-  // emit "addl x@ntpoff,%eax" (local exec) or "addl x@indntpoff,%eax" (initial
-  // exec)
+  // emit "addl x@ntpoff,%eax" (local exec)
+  // or "addl x@indntpoff,%eax" (initial exec)
+  // or "addl x@gotntpoff(%ebx) ,%eax" (initial exec, 32-bit pic)
   SDValue TGA = DAG.getTargetGlobalAddress(GA->getGlobal(), dl,
                                            GA->getValueType(0),
                                            GA->getOffset(), OperandFlags);
   SDValue Offset = DAG.getNode(WrapperKind, dl, PtrVT, TGA);
 
-  if (model == TLSModel::InitialExec)
-    Offset = DAG.getLoad(PtrVT, dl, DAG.getEntryNode(), Offset,
-                         MachinePointerInfo::getGOT(), false, false, false, 0);
+  if (model == TLSModel::InitialExec) {
+    if (isPIC && !is64Bit) {
+      Offset = DAG.getNode(ISD::ADD, dl, PtrVT,
+                          DAG.getNode(X86ISD::GlobalBaseReg, DebugLoc(), PtrVT),
+                           Offset);
+    } else {
+      Offset = DAG.getLoad(PtrVT, dl, DAG.getEntryNode(), Offset,
+                           MachinePointerInfo::getGOT(), false, false, false,
+                           0);
+    }
+  }
 
   // The address of the thread local variable is the add of the thread
   // pointer with the offset of the variable.
@@ -7341,7 +7351,6 @@ X86TargetLowering::LowerGlobalTLSAddress(SDValue Op, SelectionDAG &DAG) const {
 
   if (Subtarget->isTargetELF()) {
     // TODO: implement the "local dynamic" model
-    // TODO: implement the "initial exec"model for pic executables
 
     // If GV is an alias then use the aliasee for determining
     // thread-localness.
@@ -7360,7 +7369,8 @@ X86TargetLowering::LowerGlobalTLSAddress(SDValue Op, SelectionDAG &DAG) const {
       case TLSModel::InitialExec:
       case TLSModel::LocalExec:
         return LowerToTLSExecModel(GA, DAG, getPointerTy(), model,
-                                   Subtarget->is64Bit());
+                                   Subtarget->is64Bit(),
+                         getTargetMachine().getRelocationModel() == Reloc::PIC_);
     }
     llvm_unreachable("Unknown TLS model.");
   }
diff --git a/lib/Target/X86/X86MCInstLower.cpp b/lib/Target/X86/X86MCInstLower.cpp
index b578e8d928..54bfce1298 100644
--- a/lib/Target/X86/X86MCInstLower.cpp
+++ b/lib/Target/X86/X86MCInstLower.cpp
@@ -160,6 +160,7 @@ MCOperand X86MCInstLower::LowerSymbolOperand(const MachineOperand &MO,
   case X86II::MO_INDNTPOFF: RefKind = MCSymbolRefExpr::VK_INDNTPOFF; break;
   case X86II::MO_TPOFF:     RefKind = MCSymbolRefExpr::VK_TPOFF; break;
   case X86II::MO_NTPOFF:    RefKind = MCSymbolRefExpr::VK_NTPOFF; break;
+  case X86II::MO_GOTNTPOFF: RefKind = MCSymbolRefExpr::VK_GOTNTPOFF; break;
   case X86II::MO_GOTPCREL:  RefKind = MCSymbolRefExpr::VK_GOTPCREL; break;
   case X86II::MO_GOT:       RefKind = MCSymbolRefExpr::VK_GOT; break;
   case X86II::MO_GOTOFF:    RefKind = MCSymbolRefExpr::VK_GOTOFF; break;
diff --git a/test/CodeGen/X86/tls-pie.ll b/test/CodeGen/X86/tls-pie.ll
index e2e58a541a..374cd36ed3 100644
--- a/test/CodeGen/X86/tls-pie.ll
+++ b/test/CodeGen/X86/tls-pie.ll
@@ -35,8 +35,12 @@ entry:
 
 define i32 @f3() {
 ; X32: f3:
-; X32:      movl i2@INDNTPOFF, %eax
-; X32-NEXT: movl %gs:(%eax), %eax
+; X32:      calll .L{{[0-9]+}}$pb
+; X32-NEXT: .L{{[0-9]+}}$pb:
+; X32-NEXT: popl %eax
+; X32-NEXT: .Ltmp{{[0-9]+}}:
+; X32-NEXT: addl $_GLOBAL_OFFSET_TABLE_+(.Ltmp{{[0-9]+}}-.L{{[0-9]+}}$pb), %eax
+; X32-NEXT: leal %gs:i2@GOTNTPOFF(%eax), %eax
 ; X32-NEXT: ret
 ; X64: f3:
 ; X64:      movq i2@GOTTPOFF(%rip), %rax
@@ -50,8 +54,13 @@ entry:
 
 define i32* @f4() {
 ; X32: f4:
-; X32:      movl %gs:0, %eax
-; X32-NEXT: addl i2@INDNTPOFF, %eax
+; X32:      calll .L{{[0-9]+}}$pb
+; X32-NEXT: .L{{[0-9]+}}$pb:
+; X32-NEXT: popl %eax
+; X32-NEXT: .Ltmp{{[0-9]+}}:
+; X32-NEXT: addl $_GLOBAL_OFFSET_TABLE_+(.Ltmp{{[0-9]+}}-.L{{[0-9]+}}$pb), %eax
+; X32-NEXT: leal i2@GOTNTPOFF(%eax), %eax
+; X32-NEXT: addl %gs:0, %eax
 ; X32-NEXT: ret
 ; X64: f4:
 ; X64:      movq %fs:0, %rax