[ARM64] When compiling for ELF in PIC mode, local symbols shouldn't go through the GOT

There's no need for local symbols to go through the GOT, in fact it seems GNU ld is not even emitting GOT entries for local symbols and will error out when trying to resolve a GOT relocation for a local symbol.

This bug triggers when bootstrapping clang on AArch64 Linux with -fPIC and the ARM64 backend. The AArch64 backend is not affected.

With this commit it's now possible to bootstrap clang on AArch64 Linux with the ARM64 backend (-fPIC, -O3).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@207226 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Target/ARM64/ARM64Subtarget.cpp b/lib/Target/ARM64/ARM64Subtarget.cpp
index e7cafbb3047052d580a18becf4ae52b6bea19fbd..62a489fd86f8eb92ea384afc01b037640a066488 100644 (file)
--- a/lib/Target/ARM64/ARM64Subtarget.cpp
+++ b/lib/Target/ARM64/ARM64Subtarget.cpp
@@ -77,7 +77,8 @@ ARM64Subtarget::ClassifyGlobalReference(const GlobalValue *GV,
       return (isDecl || GV->isWeakForLinker()) ? ARM64II::MO_GOT
                                                : ARM64II::MO_NO_FLAG;
     else
-      return ARM64II::MO_GOT;
+      // No need to go through the GOT for local symbols on ELF.
+      return GV->hasLocalLinkage() ? ARM64II::MO_NO_FLAG : ARM64II::MO_GOT;
   }
 
   return ARM64II::MO_NO_FLAG;

diff --git a/test/CodeGen/ARM64/pic-local-symbol.ll b/test/CodeGen/ARM64/pic-local-symbol.ll
new file mode 100644 (file)
index 0000000..627e741
--- /dev/null
+++ b/test/CodeGen/ARM64/pic-local-symbol.ll
@@ -0,0 +1,22 @@
+; RUN: llc -mtriple=arm64-unknown-linux-gnu -relocation-model=pic < %s | FileCheck %s
+
+@a = internal unnamed_addr global i32 0, align 4
+@.str = private unnamed_addr constant [6 x i8] c"test\0A\00", align 1
+
+define i32 @get() {
+; CHECK: get:
+; CHECK: adrp x{{[0-9]+}}, a
+; CHECK-NEXT: ldr w{{[0-9]+}}, [x{{[0-9]}}, :lo12:a]
+  %res = load i32* @a, align 4
+  ret i32 %res
+}
+
+define void @foo() nounwind {
+; CHECK: foo:
+; CHECK: adrp x{{[0-9]}}, .L.str
+; CHECK-NEXT: add x{{[0-9]}}, x{{[0-9]}}, :lo12:.L.str
+  tail call void @bar(i8* getelementptr inbounds ([6 x i8]* @.str, i64 0, i64 0))
+  ret void
+}
+
+declare void @bar(i8*)