Hi Martin,
> It's fix for Darwin target and similar others which do
> not support the asembler directive.
>
> Ready for master?
I'm usually all for using effective-target keywords instead of ad-hoc
tests or target lists, but this seems like a single-use case and thus a
bit too much. Maybe restrict to gas instead? Or is it possible to
change the test in a way that doesn't depend on special unportable
target assembler options? I haven't looked admittedly.
If none of this works out:
* Please change the keyword to as_execstack (assuming that an assembler
supports the positive form support the negative one, too). "wa"
doesn't tell me anything...
* This needs documenting in sourcebuild.texi, as always.
So, if there's really no other option: ok with those changes.
> gcc/testsuite/ChangeLog:
>
> 2020-03-16 Martin Liska <mli...@suse.cz>
>
> PR lto/94157
> * gcc.dg/lto/pr94157_0.c: Add wa_noexecstack effective
> target filter.
> * lib/target-supports.exp: Add
> check_effective_target_wa_noexecstack.
One question (because I noticed it in several of your patch submissions
lately): do you have a particularly short line length configure for
ChangeLogs or in general? Many of the lines are way below a 75
character limit, which doesn't improve readablity...
Rainer
--
-----------------------------------------------------------------------------
Rainer Orth, Center for Biotechnology, Bielefeld University
