[ELF] Define __dso_handle symbol if needed

[phosek]

Summary

Traditionally, it has been defined in crtbegin.o, which is typically provided by libgcc or as part of the C library on some systems. However, but there's no principled reason for it to be there. We optionally define this symbol, which can be used on platforms that don't provide __dso_handle in crtbegin.o or which don't use crtbegin.o at all.

[phosek]

There is a related discussion in http://lists.llvm.org/pipermail/llvm-dev/2017-June/113648.html thread.

[ruiu]

Overall looks fine. I think this is the right approach to define __dso_handle, as it is very easy to do, compared to other solutions.

ELF/Writer.cpp

910

I'd describe a bit more about the __dso_handle symbol so that the code is more self-contained for the first-time reader of the source code. Specifically, I'd like to mention

• dso_handle is passed to cxa_finalize as a marker to identify each DSO
• the address of the symbol doesn't matter as long as they are different in different DSOs, so we chose the start address of the DSO

In particular, "... with a value which is an address in one of the object's segment" sounds too formal specification-ish. I'd make it more concrete.

913

Don't you want to do this only when Config->Shared?

[phosek]

ELF/Writer.cpp

913

The frontend generates the __dso_handle reference in any case because it doesn't whether the code is going to be linked into a shared library or not so we need to generate this symbol even in the non-shared case.

[ruiu]

LGTM

[ruiu]

ELF/Writer.cpp

910

Remove an extra space.

ELF/Writer.cpp

template <class ELFT> void Writer<ELFT>::addReservedSymbols() {
 
   // __tls_get_addr is defined by the dynamic linker for dynamic ELFs. For
   // static linking the linker is required to optimize away any references to
   // __tls_get_addr, so it's not defined anywhere. Create a hidden definition
   // to avoid the undefined symbol error.
   if (!InX::DynSymTab)
     Symtab<ELFT>::X->addIgnored("__tls_get_addr");
 
+  // __dso_handle symbol  is passed to cxa_finalize as a marker to identify
+  // each DSO. The address of the symbol doesn't matter as long as they are
+  // different in different DSOs, so we chose the start address of the DSO.
+  addOptionalRegular<ELFT>("__dso_handle", Out::ElfHeader, 0, STV_HIDDEN);
+
   // __ehdr_start is the location of ELF file headers. Note that we define
   // this symbol unconditionally even when using a linker script, which
   // differs from the behavior implemented by GNU linker which only define
   // this symbol if ELF headers are in the memory mapped segment.
   addOptionalRegular<ELFT>("__ehdr_start", Out::ElfHeader, 0, STV_HIDDEN);
 
   // If linker script do layout we do not need to create any standart symbols.
   if (Script->Opt.HasSections)

test/ELF/dso_handle.s

+# REQUIRES: x86
+
+# RUN: llvm-mc -filetype=obj -triple=x86_64-unknown-linux %s -o %t.o
+# RUN: ld.lld -shared %t.o -o %t
+# RUN: llvm-readobj -symbols %t | FileCheck %s
+# CHECK:    Name: __dso_handle
+# CHECK-NEXT:    Value: 0x0
+# CHECK-NEXT:    Size: 0
+# CHECK-NEXT:    Binding: Local
+# CHECK-NEXT:    Type: None
+# CHECK-NEXT:    Other [
+# CHECK-NEXT:      STV_HIDDEN
+# CHECK-NEXT:    ]
+# CHECK-NEXT:    Section: .dynsym
+
+.text
+.global foo, __dso_handle
+foo:
+  lea __dso_handle(%rip),%rax