[Bug dynamic-link/27924] ld.so: Support DT_RELR relative relocation format

["carlos at redhat dot com" <sourceware-bugzilla@sourceware.org>]

https://sourceware.org/bugzilla/show_bug.cgi?id=27924

--- Comment #6 from Carlos O'Donell <carlos at redhat dot com> ---
(In reply to Fangrui Song from comment #5)
> (In reply to Carlos O'Donell from comment #3)
> > The proposal to the ABI seems stalled and probably needs to be pushed again
> > to the gABI group with the updated wording for review again? The alternative
> > is not to extend the gABI but consider this a GNU ABI extension. If Google
> > has interest in moving this forward someone needs to be a change champion
> > here and work with upstream.
> > 
> > Objectively I have no problem with SHT_RELR, but it should be documented in
> > our ABI docs *somewhere* so we have a concrete place to discuss future
> > changes and accept those changes.
> > 
> > Today the gABI list is a "consensus" group with Cary Coutant documenting
> > that consensus. The GNU ABI is implemented by the GNU Toolchain, so there is
> > probably some consensus there that needs forming, but we can do that too.
> 
> http://www.sco.com/developers/gabi/latest/contents.html is unmaintained and
> receives no update since circa 2016. So SHT_RELR is wording adopted by many
> groups, including Solaris which has many differences with the GNU ABI.

No update since 2016 does not mean unmaintained. Please speak with Cary Coutant
to determine the current status.

If Solaris has adopted SHT_RELR I assume they put it into their OS-specific
ABI?

> If https://groups.google.com/g/generic-abi/c/9OO5vhxb00Y "Ongoing
> Maintenance of the gABI" will take some time, perhaps the wording can be
> added to System V Application Binary Interface Linux Extensions temporarily?

It could, but why temporary? If there are objections at the ELF gABI level, and
there might be, then we can just look at a Linux extension [1].

We would submit the constants in the scope of the OS-specific extensions, and
then binaries with the old ABI would not match, but I don't know if that is an
issue you want to address?

[1] Just for the sake of others reading, the generic Linux ABI (GNU ABI) is
maintained by HJ with input from various toolchain developers:
https://gitlab.com/x86-psABIs/Linux-ABI

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