[Bug symtab/30031] New: gdb ctf tests fail with gold

[Bug symtab/30031] New: gdb ctf tests fail with gold

[tromey at sourceware dot org]

https://sourceware.org/bugzilla/show_bug.cgi?id=30031

            Bug ID: 30031
           Summary: gdb ctf tests fail with gold
           Product: gdb
           Version: HEAD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: symtab
          Assignee: unassigned at sourceware dot org
          Reporter: tromey at sourceware dot org
  Target Milestone: ---

When I use alternatives to set my system linker to gold,
the gdb ctf tests will fail.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug symtab/30031] gdb ctf tests fail with gold

[tromey at sourceware dot org]

https://sourceware.org/bugzilla/show_bug.cgi?id=30031

Tom Tromey <tromey at sourceware dot org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |nick.alcock at oracle dot com

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug symtab/30031] gdb ctf tests fail with gold

[tromey at sourceware dot org]

https://sourceware.org/bugzilla/show_bug.cgi?id=30031

--- Comment #1 from Tom Tromey <tromey at sourceware dot org> ---
Forgot to mention, I'm using:

prentzel. gold --version
bash: gold: command not found
prentzel. /usr/bin/ld.gold --version
GNU gold (version 2.37-37.fc36) 1.16


The corresponding ld.bfd works fine.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug symtab/30031] gdb ctf tests fail with gold

[nick.alcock at oracle dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=30031

--- Comment #2 from Nick Alcock <nick.alcock at oracle dot com> ---
This is expected: gold (like lld and mold) has no support for linking CTF yet
(though it should be fairly easy to add: see what ld does in ldlang.c and
ldelfgen.c: I've been meaning to do it for a while but haven't got around to it
yet).

Until that's done, the GDB CTF tests really need to pre-arrange the compilation
environment so that they always use the built ld (if built) or don't run
(otherwise). Something similar to what is done in
libctf/testsuite/config/default.exp should work fine, but, obviously, done in a
function invoked by ctf*exp and reversed at the end of them.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug symtab/30031] gdb ctf tests fail with gold

[tromey at sourceware dot org]

https://sourceware.org/bugzilla/show_bug.cgi?id=30031

--- Comment #3 from Tom Tromey <tromey at sourceware dot org> ---
(In reply to Nick Alcock from comment #2)
> This is expected: gold (like lld and mold) has no support for linking CTF
> yet

Aha, good to know, thanks.

> Until that's done, the GDB CTF tests really need to pre-arrange the
> compilation environment so that they always use the built ld (if built) or
> don't run (otherwise).

gdb doesn't generally use the in-tree linker for tests, and I
think it's pretty common for gdb devs to --disable-{ld,gold}.
Instead perhaps the test can be changed to either examine
the linker somehow (this is kind of a pain) or maybe to just
use gdb_test_multiple, detect the failure mode, and kfail.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug symtab/30031] gdb ctf tests fail with gold

[nick.alcock at oracle dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=30031

--- Comment #4 from Nick Alcock <nick.alcock at oracle dot com> ---
Examining the linker is fairly easy. We can do a test compile-and-link and
examine the results with objdump, or take the simple route and just look at the
--help output text  (all lds capable of CTF deduplication currently have at
least one CTF-related command-line option!)

-- 
You are receiving this mail because:
You are on the CC list for the bug.