[PATCH v2 1/3] cris/testsuite: don't use --em=

[Hans-Peter Nilsson]

> Date: Tue, 24 Jun 2025 13:56:09 +0200
> From: Jan Beulich <jbeulich@suse.com>

> Using such abbreviations is fine when written on an interactive command
> line by a human. In scripts and alike, doing so risks colliding with
> later option additions, as is about to occur for gas: Shortly there'll
> be --emit-local-absolute.

(Aw...but emulations is such a long word that I'll
abbreviate it "e8s". 1/2 :-)

Looks like I'll have to go fix gcc/config/cris.h too, then.
Thanks for the heads-up.

> ---
> I was surprised by only cris-elf failing its testsuite run. Is there a
> reason crisv32-* isn't covered by gas/testsuite/gas/cris/cris.exp and
> ld/testsuite/ld-cris/cris.exp?

>From the top of my head, I think I did it that way just to
avoid framework complexity.  Focus is on testing for
cris-*-* (not forgetting to also test cris-linux-gnu when
testing changes affecting CRIS) where the testsuite caters
also to CRIS v32.  CRIS v32 is a ISA variant with some
overlap in coding (and ELF header distinction through
e_flags).  Maybe, or maybe not, it'd work to just
s/cris-*-*/cris*-*-*/ in gas and ld cris.exp.  I don't even
remember if I've tried that at some time but saved it for a
rainy day due to required churn.

> (In various of the hunks below cris-*-*
> can also be seen as a [perhaps overly restrictive] pattern.)

Now that you mention it, the dg "target cris-*-*" specifier
may possibly be omitted since that's the gate condition.

The patch is ok, thanks.

BTW, speaking of gas e8s, wrt. CRIS: I don't mind if e8s
support is dropped completely, as long as --emulation=<arg>
(or --emulation=crisaout for cris-aout and
--emulation=criself for the rest) is kept as a compatibility
option.

brgds, H-P