Drop __real_ symbol from the output symbol table.

[ruiu]

Summary

In D34993, we discussed and concluded that we should drop real_
symbol from the symbol table, but I did the opposite in D50569.
This patch is to drop real_ symbol.

[chrisjackson]

@ruiu

Hello, LGTM.

[grimar]

My apologies here, I did not see this patch for some reason. I believe it is ok, so LGTM.

[MaskRay]

@thakis This fixed the --wrap bug mentioned in https://bugs.chromium.org/p/chromium/issues/detail?id=1086636

It seems that this patch had never been committed. I commited it with a few adjustment to the comment and wrap-no-real.s

Note that omitting __real_ is a correctness matter if __real_ is undefined:

// Now renaming is complete, and no one refers to real. We drop real from
// .symtab and .dynsym. If real is undefined, it is important that we don't
// leave it in .dynsym, because otherwise it might lead to an undefined symbol
// error in a subsequent link. If real is defined, we could emit real as an
// alias for sym, but that could degrade the user experience of some tools
// that can print out only one symbol for each location: sym is a preferred
// name than real, but they might print out real instead.

lld/ELF/SymbolTable.cpp

void SymbolTable::wrap(Symbol *sym, Symbol *real, Symbol *wrap) {
   // Swap symbols as instructed by -wrap.
   int &idx1 = symMap[CachedHashStringRef(sym->getName())];
   int &idx2 = symMap[CachedHashStringRef(real->getName())];
   int &idx3 = symMap[CachedHashStringRef(wrap->getName())];
 
   idx2 = idx1;
   idx1 = idx3;
 
-  // Now renaming is complete. No one refers Real symbol. We could leave
-  // Real as-is, but if Real is written to the symbol table, that may
-  // contain irrelevant values. So, we copy all values from Sym to Real.
-  StringRef s = real->getName();
-  memcpy(real, sym, sizeof(SymbolUnion));
-  real->setName(s);
+  // Now renaming is complete, and no one refers to real. We drop real from
+  // .symtab and .dynsym. If real is undefined, it is important that we don't
+  // leave it in .dynsym, because otherwise it might lead to an undefined symbol
+  // error in a subsequent link. If real is defined, we could emit real as an
+  // alias for sym, but that could degrade the user experience of some tools
+  // that can print out only one symbol for each location: sym is a preferred
+  // name than real, but they might print out real instead.
+  real->isUsedInRegularObj = false;
 }
 
 // Find an existing symbol or create a new one.
 Symbol *SymbolTable::insert(StringRef name) {
   // <name>@@<version> means the symbol is the default version. In that
   // case <name>@@<version> will be used to resolve references to <name>.
   //
   // Since this is a hot path, the following string search code is

lld/test/ELF/lto/wrap-2.ll

 ; CHECK-NEXT: popq  %rax
 ; CHECK-NEXT: jmp{{.*}}<bar>
 
 ; Check that bar and __wrap_bar retain their original binding.
 ; BIND:      Name: bar
 ; BIND-NEXT: Value:
 ; BIND-NEXT: Size:
 ; BIND-NEXT: Binding: Local
-; BIND:      Name: __real_bar
-; BIND-NEXT: Value:
-; BIND-NEXT: Size:
-; BIND-NEXT: Binding: Local
 ; BIND:      Name: __wrap_bar
 ; BIND-NEXT: Value:
 ; BIND-NEXT: Size:
 ; BIND-NEXT: Binding: Local
 
 target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
 target triple = "x86_64-unknown-linux-gnu"
 
 declare void @bar()
 declare void @__real_bar()
 
 define void @foo() {
   call void @bar()
   call void @__real_bar()
   ret void
 }

lld/test/ELF/wrap-no-real.s

 // REQUIRES: x86
 // RUN: llvm-mc -filetype=obj -triple=x86_64-pc-linux %s -o %t1.o
 // RUN: llvm-mc -filetype=obj -triple=x86_64-pc-linux %p/Inputs/wrap-no-real.s -o %t2.o
 // RUN: llvm-mc -filetype=obj -triple=x86_64-pc-linux %p/Inputs/wrap-no-real2.s -o %t3.o
 // RUN: ld.lld -o %t3.so -shared %t3.o
 
 // RUN: ld.lld -o %t %t1.o %t2.o -wrap foo
-// RUN: llvm-objdump -d --print-imm-hex %t | FileCheck %s
-
-// RUN: ld.lld -o %t %t1.o %t2.o %t3.so -wrap foo
-// RUN: llvm-objdump -d --print-imm-hex %t | FileCheck %s
+// RUN: llvm-objdump -d %t | FileCheck %s
+// RUN: llvm-readelf -s -x .got %t | FileCheck --check-prefix=READELF --implicit-check-not=__real_ %s
 
 // CHECK: <_start>:
-// CHECK-NEXT: movl $0x11010, %edx
-// CHECK-NEXT: movl $0x11010, %edx
-// CHECK-NEXT: movl $0x11000, %edx
-
-// RUN: llvm-objdump -t %t | FileCheck --check-prefix=SYM %s
-
-
-// SYM:      {{.*}}           l .dynamic 0000000000000000 .hidden _DYNAMIC
-// SYM-NEXT: 0000000000011000 g *ABS*    0000000000000000 __real_foo
-// SYM-NEXT: 0000000000011010 g *ABS*    0000000000000000 __wrap_foo
-// SYM-NEXT: {{.*}}           g .text    0000000000000000 _start
-// SYM-NEXT: 0000000000011000 g *ABS*    0000000000000000 foo
+// CHECK-NEXT: movq {{.*}}(%rip), %rax  # 2021a8
+// CHECK-NEXT: movq {{.*}}(%rip), %rbx  # 2021a8
+// CHECK-NEXT: movq {{.*}}(%rip), %rcx  # 2021b0
+
+// READELF:      0000000000011010  0 NOTYPE GLOBAL DEFAULT ABS __wrap_foo
+// READELF:      0000000000011000  0 NOTYPE GLOBAL DEFAULT ABS foo
+// READELF:      Hex dump of section '.got':
+// READELF-NEXT: 0x[[#%x,ADDR:]] 10100100 00000000 00100100 00000000
+
+// RUN: ld.lld -o %t2 %t1.o %t2.o %t3.so --wrap foo
+// RUN: llvm-objdump -d %t2 | FileCheck --check-prefix=CHECK2 %s
+// RUN: llvm-readelf -s -x .got %t2 | FileCheck --check-prefix=READELF --implicit-check-not=__real_ %s
+
+// CHECK2: <_start>:
+// CHECK2-NEXT: movq {{.*}}(%rip), %rax  # 2022f8
+// CHECK2-NEXT: movq {{.*}}(%rip), %rbx  # 2022f8
+// CHECK2-NEXT: movq {{.*}}(%rip), %rcx  # 202300
 
 .global _start
 _start:
-  movl $foo, %edx
-  movl $__wrap_foo, %edx
-  movl $__real_foo, %edx
+  mov foo@gotpcrel(%rip), %rax
+  mov __wrap_foo@gotpcrel(%rip), %rbx
+  mov __real_foo@gotpcrel(%rip), %rcx

lld/test/ELF/wrap.s

 // SYM2:      Name: __wrap_foo
 // SYM2-NEXT: Value: 0x11010
 // SYM2-NEXT: Size:
 // SYM2-NEXT: Binding: Weak
 // SYM2-NEXT: Type:    None
 // SYM2-NEXT: Other [
 // SYM2-NEXT:   STV_PROTECTED
 // SYM2-NEXT: ]
-// SYM3:      Name: __real_foo
-// SYM3-NEXT: Value: 0x11000
-// SYM3-NEXT: Size:
-// SYM3-NEXT: Binding: Global
-// SYM3-NEXT: Type:    None
-// SYM3-NEXT: Other:   0
+// SYM3-NOT:  Name: __real_foo
 
 .global _start
 _start:
   movl $foo, %edx
   movl $__wrap_foo, %edx
   movl $__real_foo, %edx