Re: Merging MPS a.k.a. scratch/igc, yet again

[Pip Cet]

"Gregor Zattler" <telegraph@gmx.net> writes:

> Hi Pip, Emacs developers,

Hello!

> * Pip Cet via "Emacs development discussions." <emacs-devel@gnu.org> 
> [2024-12-10; 15:10 GMT]:
>> ...and more problems. MPS has made the decision not to work with gcc
> -O3, only with -O2 or less, and LTO in particular is something MPS
> cannot reliably support, IIUC.
>
> is this (-O2) true only for building mps or
> also for building Emacs with
> --with-mps=yes?  If the latter it should
> be documented in README-IGC, I think.

I think you're right.  I do think that the -O2 or less rule was meant to
apply to MPS only, not for all applications using it.

Does anyone remember what our conclusion was wrt
-fno-omit-frame-pointer?  I seem to remember there was a patch to MPS to
avoid relying on setjmp() to save all registers, but I'd still be
happier if we enabled that for all MPS builds, since we don't know
whether our MPS has the patch.

> I built mps as instructed in README-IGC,
> but tried to build emacs with -O3 and it
> crashed.  I can reproduce and file a
> bug, if you are interested.

I'm definitely interested.  It might just be the -fomit-frame-pointer
thing, but if people run into that, clearly our configure script needs
to be modified to explicitly request -fno-omit-frame-pointer.

Pip