[llvm] [Bolt] fix a wrong relocation update issue with weak references (PR #69136)

[Vladislav Khmelevsky via llvm-commits]

================
@@ -0,0 +1,34 @@
+// This test checks whether BOLT can correctly handle relocations against weak symbols.
+
+// RUN: %clang %cflags -Wl,-z,notext -shared -Wl,-q %s -o %t.so
+// RUN: llvm-bolt %t.so -o %t.so.bolt
+// RUN: llvm-nm -n %t.so.bolt > %t.out.txt
+// RUN: llvm-objdump -dj .rodata %t.so.bolt >> %t.out.txt
+// RUN: FileCheck %s --input-file=%t.out.txt
+
+# CHECK: w func_1
+# CHECK: {{0+}}[[#%x,ADDR:]] W func_2
+
+# CHECK: {{.*}} <.rodata>:
+# CHECK-NEXT: {{.*}} .word 0x00000000
+# CHECK-NEXT: {{.*}} .word 0x00000000
+# CHECK-NEXT: {{.*}} .word 0x{{[0]+}}[[#ADDR]]
----------------
yota9 wrote:

> > It's true that before it is 0 here. But it is up to the linker to set value here or not. and for example to support RELR we need to set values here. I don't think it is a problem and would rather set the value, than not.
> 
> Can we rely on the linker's decision in such cases? I.e., if the linker decided to put 0, keep it at that. If it was the symbol value, then we can update it.

Theoretically we can. But is there a reason behind it? It doesn't affect runtime. And to be honest I prefer to set the values, we can easily see the value during objdump, sometimes it is useful. 

https://github.com/llvm/llvm-project/pull/69136