[llvm-bugs] [Bug 32425] Regression with Musl after adding support for outputting to /dev/null

[via llvm-bugs]

https://bugs.llvm.org/show_bug.cgi?id=32425

Rafael Ávila de Espíndola <rafael.espindola at gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |WONTFIX
             Status|NEW                         |RESOLVED

--- Comment #5 from Rafael Ávila de Espíndola <rafael.espindola at gmail.com> ---

> > Also, ld.lld --reproduce should leave a tar file that you can upload here
> > for further analysis.
> 
> Attached it also (had to compress, so it is under 1Mb).

The problem is that crt1.o has

000000000000001b  0000000300000002 R_X86_64_PC32          0000000000000000
__libc_start_main - 4

Which requires knowing the address of __libc_start_main. That can be done when
producing an executable if the symbol in preemptable (we create a plt and use
that as the symbol address). But in this case the symbol is not preemptable:

  1488: 000000000004d436    57 FUNC    GLOBAL PROTECTED    9 __libc_start_main

We should produce a better error message and we could even add an option for
ignoring pointer equality (after all that is what we do in icf), but the test
as written is working as intended.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20170327/9eec6525/attachment.html>