Fix PR 4004 by including the call to __tls_get_addr in X86tlsaddr. This is not

very elegant, but neither is the tls specification :-(



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@69968 91177308-0d34-0410-b5e6-96231b3b80d8

lib/Target/X86/X86ISelLowering.cpp

@@ -4745,7 +4745,7 @@ X86TargetLowering::LowerGlobalAddress(SDValue Op, SelectionDAG &DAG) {
 
 static SDValue
 GetTLSADDR(SelectionDAG &DAG, SDValue Chain, GlobalAddressSDNode *GA,
-           SDValue *InFlag) {
+           SDValue *InFlag, const MVT PtrVT, unsigned ReturnReg) {
   SDVTList NodeTys = DAG.getVTList(MVT::Other, MVT::Flag);
   DebugLoc dl = GA->getDebugLoc();
   SDValue TGA = DAG.getTargetGlobalAddress(GA->getGlobal(),
@@ -4753,11 +4753,13 @@ GetTLSADDR(SelectionDAG &DAG, SDValue Chain, GlobalAddressSDNode *GA,
                                            GA->getOffset());
   if (InFlag) {
     SDValue Ops[] = { Chain,  TGA, *InFlag };
-    return DAG.getNode(X86ISD::TLSADDR, dl, NodeTys, Ops, 3);
+    Chain = DAG.getNode(X86ISD::TLSADDR, dl, NodeTys, Ops, 3);
   } else {
     SDValue Ops[]  = { Chain, TGA };
-    return DAG.getNode(X86ISD::TLSADDR, dl, NodeTys, Ops, 2);
+    Chain = DAG.getNode(X86ISD::TLSADDR, dl, NodeTys, Ops, 2);
   }
+  SDValue Flag = Chain.getValue(1);
+  return DAG.getCopyFromReg(Chain, dl, ReturnReg, PtrVT, Flag);
 }
 
 // Lower ISD::GlobalTLSAddress using the "general dynamic" model, 32 bit
@@ -4772,42 +4774,14 @@ LowerToTLSGeneralDynamicModel32(GlobalAddressSDNode *GA, SelectionDAG &DAG,
                                                  PtrVT), InFlag);
   InFlag = Chain.getValue(1);
 
-  Chain = GetTLSADDR(DAG, Chain, GA, &InFlag);
+  return GetTLSADDR(DAG, Chain, GA, &InFlag, PtrVT, X86::EAX);
-  InFlag = Chain.getValue(1);
-
-  SDVTList NodeTys = DAG.getVTList(MVT::Other, MVT::Flag);
-  SDValue Ops1[] = { Chain,
-                      DAG.getTargetExternalSymbol("___tls_get_addr",
-                                                  PtrVT),
-                      DAG.getRegister(X86::EAX, PtrVT),
-                      DAG.getRegister(X86::EBX, PtrVT),
-                      InFlag };
-  Chain = DAG.getNode(X86ISD::CALL, dl, NodeTys, Ops1, 5);
-  InFlag = Chain.getValue(1);
-
-  return DAG.getCopyFromReg(Chain, dl, X86::EAX, PtrVT, InFlag);
 }
 
 // Lower ISD::GlobalTLSAddress using the "general dynamic" model, 64 bit
 static SDValue
 LowerToTLSGeneralDynamicModel64(GlobalAddressSDNode *GA, SelectionDAG &DAG,
                                 const MVT PtrVT) {
-  SDValue InFlag, Chain;
+  return GetTLSADDR(DAG, DAG.getEntryNode(), GA, NULL, PtrVT, X86::RAX);
-  DebugLoc dl = GA->getDebugLoc();  // ? function entry point might be better
-
-  Chain = GetTLSADDR(DAG, DAG.getEntryNode(), GA, NULL);
-  InFlag = Chain.getValue(1);
-
-  SDVTList NodeTys = DAG.getVTList(MVT::Other, MVT::Flag);
-  SDValue Ops1[] = { Chain,
-                      DAG.getTargetExternalSymbol("__tls_get_addr",
-                                                  PtrVT),
-                      DAG.getRegister(X86::RDI, PtrVT),
-                      InFlag };
-  Chain = DAG.getNode(X86ISD::CALL, dl, NodeTys, Ops1, 4);
-  InFlag = Chain.getValue(1);
-
-  return DAG.getCopyFromReg(Chain, dl, X86::RAX, PtrVT, InFlag);
 }
 
 // Lower ISD::GlobalTLSAddress using the "initial exec" (for no-pic) or

lib/Target/X86/X86Instr64bit.td

@@ -1303,9 +1303,17 @@ def MOV64ri64i32 : Ii32<0xB8, AddRegFrm, (outs GR64:$dst), (ins i64i32imm:$src),
 // Thread Local Storage Instructions
 //===----------------------------------------------------------------------===//
 
-let hasSideEffects = 1, Defs = [RDI] in
+// All calls clobber the non-callee saved registers. RSP is marked as
+// a use to prevent stack-pointer assignments that appear immediately
+// before calls from potentially appearing dead.
+let Defs = [RAX, RCX, RDX, RSI, RDI, R8, R9, R10, R11,
+            FP0, FP1, FP2, FP3, FP4, FP5, FP6, ST0, ST1,
+            MM0, MM1, MM2, MM3, MM4, MM5, MM6, MM7,
+            XMM0, XMM1, XMM2, XMM3, XMM4, XMM5, XMM6, XMM7,
+            XMM8, XMM9, XMM10, XMM11, XMM12, XMM13, XMM14, XMM15, EFLAGS],
+    Uses = [RSP] in
 def TLS_addr64 : I<0, Pseudo, (outs), (ins i64imm:$sym),
-              ".byte\t0x66; leaq\t${sym:mem}(%rip), %rdi; .word\t0x6666; rex64",
+    ".byte\t0x66; leaq\t${sym:mem}(%rip), %rdi; .word\t0x6666; rex64;call\t__tls_get_addr@PLT",
                   [(X86tlsaddr tglobaltlsaddr:$sym)]>,
                   Requires<[In64BitMode]>;
 

lib/Target/X86/X86InstrInfo.td

@@ -2977,9 +2977,16 @@ def MOV32r0  : I<0x31, MRMInitReg,  (outs GR32:$dst), (ins),
 // Thread Local Storage Instructions
 //
 
-let hasSideEffects = 1, Uses = [EBX], Defs = [EAX] in
+// All calls clobber the non-callee saved registers. ESP is marked as
+// a use to prevent stack-pointer assignments that appear immediately
+// before calls from potentially appearing dead.
+let Defs = [EAX, ECX, EDX, FP0, FP1, FP2, FP3, FP4, FP5, FP6, ST0,
+            MM0, MM1, MM2, MM3, MM4, MM5, MM6, MM7,
+            XMM0, XMM1, XMM2, XMM3, XMM4, XMM5, XMM6, XMM7,
+            XMM8, XMM9, XMM10, XMM11, XMM12, XMM13, XMM14, XMM15, EFLAGS],
+    Uses = [ESP, EBX] in
 def TLS_addr32 : I<0, Pseudo, (outs), (ins i32imm:$sym),
-                  "leal\t${sym:mem}(,%ebx,1), %eax",
+                  "leal\t${sym:mem}(,%ebx,1), %eax; call\t___tls_get_addr@PLT",
                   [(X86tlsaddr tglobaltlsaddr:$sym)]>,
                   Requires<[In32BitMode]>;
 

test/CodeGen/X86/2009-04-24.ll

@@ -0,0 +1,13 @@
+; RUN: llvm-as < %s | llc -march=x86 -mtriple=i386-linux-gnu -regalloc=local -relocation-model=pic > %t
+; RUN: grep {leal.*TLSGD.*___tls_get_addr} %t
+; RUN: llvm-as < %s | llc -march=x86-64 -mtriple=x86_64-linux-gnu -regalloc=local -relocation-model=pic > %t2
+; RUN: grep {leaq.*TLSGD.*__tls_get_addr} %t2
+; PR/4004
+
+@i = thread_local global i32 15
+
+define i32 @f() {
+entry:
+	%tmp1 = load i32* @i
+	ret i32 %tmp1
+}