[PowerPC] Fix large code model with the ELFv2 ABI

The global entry point prologue currently assumes that the TOC
associated with a function is less than 2GB away from the function
entry point.  This is always true when using the medium or small
code model, but may not be the case when using the large code model.

This patch adds a new variant of the ELFv2 global entry point prologue
that lifts the 2GB restriction when building with -mcmodel=large.
This works by emitting a quadword containing the distance from the
function entry point to its associated TOC immediately before the
entry point, and then using a prologue like:

ld r2,-8(r12)
add r2,r2,r12

Since creation of the entry point prologue is now split across two
separate routines (PPCLinuxAsmPrinter::EmitFunctionEntryLabel emits
the data word, PPCLinuxAsmPrinter::EmitFunctionBodyStart the prolog
code), I've switched to using named labels instead of just temporaries
to indicate the locations of the global and local entry points and the
new TOC offset data word.

These names are provided by new routines in PPCFunctionInfo modeled
after the existing PPCFunctionInfo::getPICOffsetSymbol.

Note that a corresponding change was committed to GCC here:
https://gcc.gnu.org/ml/gcc-patches/2015-12/msg00355.html

Reviewers: hfinkel

Differential Revision: http://reviews.llvm.org/D15500

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@257597 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Target/PowerPC/PPCAsmPrinter.cpp b/lib/Target/PowerPC/PPCAsmPrinter.cpp
index 9a63c14b5053418766e9306aa90f319f958d1002..ec354c209ca0425e22fddd21802a591c20c211f5 100644 (file)
--- a/lib/Target/PowerPC/PPCAsmPrinter.cpp
+++ b/lib/Target/PowerPC/PPCAsmPrinter.cpp
@@ -1092,8 +1092,28 @@ void PPCLinuxAsmPrinter::EmitFunctionEntryLabel() {
   }
 
   // ELFv2 ABI - Normal entry label.
-  if (Subtarget->isELFv2ABI())
+  if (Subtarget->isELFv2ABI()) {
+    // In the Large code model, we allow arbitrary displacements between
+    // the text section and its associated TOC section.  We place the
+    // full 8-byte offset to the TOC in memory immediatedly preceding
+    // the function global entry point.
+    if (TM.getCodeModel() == CodeModel::Large
+        && !MF->getRegInfo().use_empty(PPC::X2)) {
+      const PPCFunctionInfo *PPCFI = MF->getInfo<PPCFunctionInfo>();
+
+      MCSymbol *TOCSymbol = OutContext.getOrCreateSymbol(StringRef(".TOC."));
+      MCSymbol *GlobalEPSymbol = PPCFI->getGlobalEPSymbol();
+      const MCExpr *TOCDeltaExpr =
+        MCBinaryExpr::createSub(MCSymbolRefExpr::create(TOCSymbol, OutContext),
+                                MCSymbolRefExpr::create(GlobalEPSymbol,
+                                                        OutContext),
+                                OutContext);
+
+      OutStreamer->EmitLabel(PPCFI->getTOCOffsetSymbol());
+      OutStreamer->EmitValue(TOCDeltaExpr, 8);
+    }
     return AsmPrinter::EmitFunctionEntryLabel();
+  }
 
   // Emit an official procedure descriptor.
   MCSectionSubPair Current = OutStreamer->getCurrentSection();
@@ -1160,10 +1180,25 @@ void PPCLinuxAsmPrinter::EmitFunctionBodyStart() {
   // thus emit a prefix sequence along the following lines:
   //
   // func:
+  // .Lfunc_gepNN:
+  //         # global entry point
+  //         addis r2,r12,(.TOC.-.Lfunc_gepNN)@ha
+  //         addi  r2,r2,(.TOC.-.Lfunc_gepNN)@l
+  // .Lfunc_lepNN:
+  //         .localentry func, .Lfunc_lepNN-.Lfunc_gepNN
+  //         # local entry point, followed by function body
+  //
+  // For the Large code model, we create
+  //
+  // .Lfunc_tocNN:
+  //         .quad .TOC.-.Lfunc_gepNN      # done by EmitFunctionEntryLabel
+  // func:
+  // .Lfunc_gepNN:
   //         # global entry point
-  //         addis r2,r12,(.TOC.-func)@ha
-  //         addi  r2,r2,(.TOC.-func)@l
-  //         .localentry func, .-func
+  //         ld    r2,.Lfunc_tocNN-.Lfunc_gepNN(r12)
+  //         add   r2,r2,r12
+  // .Lfunc_lepNN:
+  //         .localentry func, .Lfunc_lepNN-.Lfunc_gepNN
   //         # local entry point, followed by function body
   //
   // This ensures we have r2 set up correctly while executing the function
@@ -1171,32 +1206,49 @@ void PPCLinuxAsmPrinter::EmitFunctionBodyStart() {
   if (Subtarget->isELFv2ABI()
       // Only do all that if the function uses r2 in the first place.
       && !MF->getRegInfo().use_empty(PPC::X2)) {
+    const PPCFunctionInfo *PPCFI = MF->getInfo<PPCFunctionInfo>();
 
-    MCSymbol *GlobalEntryLabel = OutContext.createTempSymbol();
+    MCSymbol *GlobalEntryLabel = PPCFI->getGlobalEPSymbol();
     OutStreamer->EmitLabel(GlobalEntryLabel);
     const MCSymbolRefExpr *GlobalEntryLabelExp =
       MCSymbolRefExpr::create(GlobalEntryLabel, OutContext);
 
-    MCSymbol *TOCSymbol = OutContext.getOrCreateSymbol(StringRef(".TOC."));
-    const MCExpr *TOCDeltaExpr =
-      MCBinaryExpr::createSub(MCSymbolRefExpr::create(TOCSymbol, OutContext),
-                              GlobalEntryLabelExp, OutContext);
+    if (TM.getCodeModel() != CodeModel::Large) {
+      MCSymbol *TOCSymbol = OutContext.getOrCreateSymbol(StringRef(".TOC."));
+      const MCExpr *TOCDeltaExpr =
+        MCBinaryExpr::createSub(MCSymbolRefExpr::create(TOCSymbol, OutContext),
+                                GlobalEntryLabelExp, OutContext);
 
-    const MCExpr *TOCDeltaHi =
-      PPCMCExpr::createHa(TOCDeltaExpr, false, OutContext);
-    EmitToStreamer(*OutStreamer, MCInstBuilder(PPC::ADDIS)
-                                 .addReg(PPC::X2)
-                                 .addReg(PPC::X12)
-                                 .addExpr(TOCDeltaHi));
-
-    const MCExpr *TOCDeltaLo =
-      PPCMCExpr::createLo(TOCDeltaExpr, false, OutContext);
-    EmitToStreamer(*OutStreamer, MCInstBuilder(PPC::ADDI)
-                                 .addReg(PPC::X2)
-                                 .addReg(PPC::X2)
-                                 .addExpr(TOCDeltaLo));
-
-    MCSymbol *LocalEntryLabel = OutContext.createTempSymbol();
+      const MCExpr *TOCDeltaHi =
+        PPCMCExpr::createHa(TOCDeltaExpr, false, OutContext);
+      EmitToStreamer(*OutStreamer, MCInstBuilder(PPC::ADDIS)
+                                   .addReg(PPC::X2)
+                                   .addReg(PPC::X12)
+                                   .addExpr(TOCDeltaHi));
+
+      const MCExpr *TOCDeltaLo =
+        PPCMCExpr::createLo(TOCDeltaExpr, false, OutContext);
+      EmitToStreamer(*OutStreamer, MCInstBuilder(PPC::ADDI)
+                                   .addReg(PPC::X2)
+                                   .addReg(PPC::X2)
+                                   .addExpr(TOCDeltaLo));
+    } else {
+      MCSymbol *TOCOffset = PPCFI->getTOCOffsetSymbol();
+      const MCExpr *TOCOffsetDeltaExpr =
+        MCBinaryExpr::createSub(MCSymbolRefExpr::create(TOCOffset, OutContext),
+                                GlobalEntryLabelExp, OutContext);
+
+      EmitToStreamer(*OutStreamer, MCInstBuilder(PPC::LD)
+                                   .addReg(PPC::X2)
+                                   .addExpr(TOCOffsetDeltaExpr)
+                                   .addReg(PPC::X12));
+      EmitToStreamer(*OutStreamer, MCInstBuilder(PPC::ADD8)
+                                   .addReg(PPC::X2)
+                                   .addReg(PPC::X2)
+                                   .addReg(PPC::X12));
+    }
+
+    MCSymbol *LocalEntryLabel = PPCFI->getLocalEPSymbol();
     OutStreamer->EmitLabel(LocalEntryLabel);
     const MCSymbolRefExpr *LocalEntryLabelExp =
        MCSymbolRefExpr::create(LocalEntryLabel, OutContext);

diff --git a/lib/Target/PowerPC/PPCMachineFunctionInfo.cpp b/lib/Target/PowerPC/PPCMachineFunctionInfo.cpp
index 95f163153c744ce98f07b0286e9c83a3e4d79ee8..9d91e31165de6f10e45daa38c727bbc9fad3eee5 100644 (file)
--- a/lib/Target/PowerPC/PPCMachineFunctionInfo.cpp
+++ b/lib/Target/PowerPC/PPCMachineFunctionInfo.cpp
@@ -23,3 +23,24 @@ MCSymbol *PPCFunctionInfo::getPICOffsetSymbol() const {
                                            Twine(MF.getFunctionNumber()) +
                                            "$poff");
 }
+
+MCSymbol *PPCFunctionInfo::getGlobalEPSymbol() const {
+  const DataLayout &DL = MF.getDataLayout();
+  return MF.getContext().getOrCreateSymbol(Twine(DL.getPrivateGlobalPrefix()) +
+                                           "func_gep" +
+                                           Twine(MF.getFunctionNumber()));
+}
+
+MCSymbol *PPCFunctionInfo::getLocalEPSymbol() const {
+  const DataLayout &DL = MF.getDataLayout();
+  return MF.getContext().getOrCreateSymbol(Twine(DL.getPrivateGlobalPrefix()) +
+                                           "func_lep" +
+                                           Twine(MF.getFunctionNumber()));
+}
+
+MCSymbol *PPCFunctionInfo::getTOCOffsetSymbol() const {
+  const DataLayout &DL = MF.getDataLayout();
+  return MF.getContext().getOrCreateSymbol(Twine(DL.getPrivateGlobalPrefix()) +
+                                           "func_toc" +
+                                           Twine(MF.getFunctionNumber()));
+}

diff --git a/lib/Target/PowerPC/PPCMachineFunctionInfo.h b/lib/Target/PowerPC/PPCMachineFunctionInfo.h
index 607cdf612eef60a6c764ab87639543b7f950d9de..10a8ce068d40aa0b2e332c9f6c223d1ff7cbb143 100644 (file)
--- a/lib/Target/PowerPC/PPCMachineFunctionInfo.h
+++ b/lib/Target/PowerPC/PPCMachineFunctionInfo.h
@@ -197,6 +197,10 @@ public:
   bool usesPICBase() const { return UsesPICBase; }
 
   MCSymbol *getPICOffsetSymbol() const;
+
+  MCSymbol *getGlobalEPSymbol() const;
+  MCSymbol *getLocalEPSymbol() const;
+  MCSymbol *getTOCOffsetSymbol() const;
 };
 
 } // end of namespace llvm

diff --git a/test/CodeGen/PowerPC/ppc64le-localentry-large.ll b/test/CodeGen/PowerPC/ppc64le-localentry-large.ll
new file mode 100644 (file)
index 0000000..4cf1b5a
--- /dev/null
+++ b/test/CodeGen/PowerPC/ppc64le-localentry-large.ll
@@ -0,0 +1,27 @@
+; RUN: llc -march=ppc64le -code-model=large < %s | FileCheck %s
+
+target datalayout = "e-m:e-i64:64-n32:64"
+target triple = "powerpc64le-unknown-linux-gnu"
+
+@number64 = global i64 10, align 8
+
+; CHECK: .abiversion 2
+
+define i64 @use_toc(i64 %a) nounwind {
+entry:
+; CHECK: .Lfunc_toc[[FN:[0-9]+]]:
+; CHECK-NEXT: .quad .TOC.-.Lfunc_gep[[FN]]
+; CHECK: use_toc:
+; CHECK-NEXT: .L{{.*}}:
+; CHECK-NEXT: .Lfunc_gep[[FN]]:
+; CHECK-NEXT: ld 2, .Lfunc_toc[[FN]]-.Lfunc_gep[[FN]](12)
+; CHECK-NEXT: add 2, 2, 12
+; CHECK-NEXT: .Lfunc_lep[[FN]]:
+; CHECK-NEXT: .localentry use_toc, .Lfunc_lep[[FN]]-.Lfunc_gep[[FN]]
+; CHECK-NEXT: %entry
+  %0 = load i64, i64* @number64, align 8
+  %cmp = icmp eq i64 %0, %a
+  %conv1 = zext i1 %cmp to i64
+  ret i64 %conv1
+}
+

diff --git a/test/CodeGen/PowerPC/ppc64le-localentry.ll b/test/CodeGen/PowerPC/ppc64le-localentry.ll
index be64f1151769d0dc7e0f94724e84ed7d1e7b93d7..45cae6b17bc2282ae328685675bfdc12ad1ed773 100644 (file)
--- a/test/CodeGen/PowerPC/ppc64le-localentry.ll
+++ b/test/CodeGen/PowerPC/ppc64le-localentry.ll
@@ -17,11 +17,11 @@ define i64 @use_toc(i64 %a) nounwind {
 entry:
 ; CHECK-LABEL: @use_toc
 ; CHECK-NEXT: .L{{.*}}:
-; CHECK-NEXT: .Ltmp[[TMP1:[0-9]+]]:
-; CHECK-NEXT: addis 2, 12, .TOC.-.Ltmp[[TMP1]]@ha
-; CHECK-NEXT: addi 2, 2, .TOC.-.Ltmp[[TMP1]]@l
-; CHECK-NEXT: .Ltmp[[TMP2:[0-9]+]]:
-; CHECK-NEXT: .localentry use_toc, .Ltmp[[TMP2]]-.Ltmp[[TMP1]]
+; CHECK-NEXT: .Lfunc_gep[[FN:[0-9]+]]:
+; CHECK-NEXT: addis 2, 12, .TOC.-.Lfunc_gep[[FN]]@ha
+; CHECK-NEXT: addi 2, 2, .TOC.-.Lfunc_gep[[FN]]@l
+; CHECK-NEXT: .Lfunc_lep[[FN]]:
+; CHECK-NEXT: .localentry use_toc, .Lfunc_lep[[FN]]-.Lfunc_gep[[FN]]
 ; CHECK-NEXT: %entry
   %0 = load i64, i64* @number64, align 8
   %cmp = icmp eq i64 %0, %a
@@ -34,11 +34,11 @@ define void @use_toc_implicit() nounwind {
 entry:
 ; CHECK-LABEL: @use_toc_implicit
 ; CHECK-NEXT: .L{{.*}}:
-; CHECK-NEXT: .Ltmp[[TMP1:[0-9]+]]:
-; CHECK-NEXT: addis 2, 12, .TOC.-.Ltmp[[TMP1]]@ha
-; CHECK-NEXT: addi 2, 2, .TOC.-.Ltmp[[TMP1]]@l
-; CHECK-NEXT: .Ltmp[[TMP2:[0-9]+]]:
-; CHECK-NEXT: .localentry use_toc_implicit, .Ltmp[[TMP2]]-.Ltmp[[TMP1]]
+; CHECK-NEXT: .Lfunc_gep[[FN:[0-9]+]]:
+; CHECK-NEXT: addis 2, 12, .TOC.-.Lfunc_gep[[FN]]@ha
+; CHECK-NEXT: addi 2, 2, .TOC.-.Lfunc_gep[[FN]]@l
+; CHECK-NEXT: .Lfunc_lep[[FN]]:
+; CHECK-NEXT: .localentry use_toc_implicit, .Lfunc_lep[[FN]]-.Lfunc_gep[[FN]]
 ; CHECK-NEXT: %entry
   call void @callee()
   ret void