Hi @ostylk. Do you intend to return to this PR?

Hi, thank you for the reminder, this PR slipped my mind.
I do intend on returning to it, preferably in this month. Am I impeding anything by working on this slowly?

All good, I was just checking in on PRs that hadn't been touched in a while. I don't think there's any particular hurry on this though.

Ready for review but I have some comments/questions:

1. The CI failure is weird, it only happens on opensuse (I could reproduce it locally with a container). Any pointers on how to approach that?
2. There was a comment in output_section_id.rs about how uninitialized TLS data is padded with zeroes in the output file. However, I couldn't find where that happens in the code? Maybe it was forgotten and that caused this bug in the first place?
3. Funnily enough GNU ld seems to have a bug (to see for yourself: delete the SkipLinker instruction in tls-custom.c test) where uninitialized TLS sections overlap initialized TLS sections. LLD seems to handle the case correctly though.

Welcome back!

Regarding the CI failure - RexMovIndirectToAbsolute vs MovIndirectToLea, that sounds like something that could be caused by different defaults as to whether the binary is position independent or not. So possibly the difference might go away if you override those defaults - e.g. by passing a flag to make both linkers produce a position-independent binary.

The PR in which we started putting zero bytes in the file for .tbss was #489. It unfortunately looks like I didn't add a test for it, so it's not impossible that something around there could have regressed.

Okay, so the linker failure seems to be due to lld behaving differently to GNU ld on openSUSE. While I am unsure what the exact reason is, the startup files (crtbegin.o crtend.o) on openSUSE cannot be linked with --pie. This leads to GNU ld and lld applying different relaxations. GNU ld and also wild choose to relax like this:

0000000000401020 <_start>:
  <snip>
  401038:       48 c7 c7 16 11 40 00    mov    $0x401116,%rdi
  40103f:       ff 15 93 2f 00 00       call   *0x2f93(%rip)        # 403fd8 <__libc_start_main@GLIBC_2.34>

while lld relaxes to this (effectively ignoring pie/no-pie stuff)

0000000000201570 <_start>:
  <snip>
  201588:       48 8d 3d d7 00 00 00    lea    0xd7(%rip),%rdi        # 201666 <main>
  20158f:       ff 15 03 1b 00 00       call   *0x1b03(%rip)        # 203098 <__libc_start_main@GLIBC_2.34>

So ultimately I think it is best to ignore that difference. GNU ld and wild switch to absolute addresses with no-pie files while lld ignores that fact completely.

Done and thank you for your patience with this PR.