[PATCH v3 00/18] PR ld/24600: Add general linker support for mapless archives

[Hans-Peter Nilsson]

> Date: Thu, 20 Nov 2025 10:55:51 +0000
> From: "Maciej W. Rozycki" <macro@orcam.me.uk>

> Hi,
> 
>  This is v3 of a patch series originally posted here 
> <https://inbox.sourceware.org/binutils/alpine.DEB.2.21.2509251408330.15999@angie.orcam.me.uk/>.  
> As from v2 it includes `--start-lib'/`--end-lib' linker command line 
> option support, however I chose to keep the original subject for the cover 
> letter for the purpose of mailbox sorting.
> 
>  I have now investigated and fixed the issue originally triggered by v1 
> for the `cris*-*-*' target related to commit c1a7d039587b ("cris: bfd: 
> Default selected target to the configured target").  The CRIS targets 
> turned out to be the only ones that make use of the multiple emulation 
> support in GAS for dislike target object formats (the only other target 
> that makes use of the feature is MIPS, but nowadays only as an alternative 
> way to choose the endianness between otherwise ELF formats), so I made it 
> the basis for new tests to verify generic BFD support for switching the 
> emulation where archives are involved, as with other targets extra messing 
> with `objcopy' would be required to change the format of
> assembly output.

Oh, the burden... 1/2 :)  Seriously, you really went above
and beyond with this patchset.

My only worry is that, as mentioned before, I'd like to move
cris*-*-* away from using the gas emulation feature, and
also to deprecate cris-aout (for gas).  Not that I've
noticed (other) problems though; curiously, making use of
the otherwise unused machinery hasn't been much of a hassle,
and nowhere near the top of my TODO list.  The only one
recent enough to stick to my memory is having to update gcc
when for gas, "--em" no longer was a valid short-form for
"--emulation".

The patchset is ok for cris*-*, thanks.

brgds, H-P