Fwd: Re: Linux on VAX

[coypu at sdf.org]

> -------- Vidarebefordrat meddelande --------
> Ämne: 	Re: Linux on VAX
> Datum: 	Thu, 24 Oct 2019 13:50:08 +0200
> Från: 	Klaus Kämpf <kkaempf at suse.de>
> Till: 	Maciej W. Rozycki <macro at linux-mips.org>
> Kopia: 	jbglaw at getshlash.de, Jan Prunk <janprunk at gmail.com>,
> vax-linux at lug-owl.de, jbglaw at lug-owl.de
> 
> 
> 
> * Maciej W. Rozycki <macro at linux-mips.org> [Oct 23. 2019 23:04]:
> 
> > The next issue is moving from reload to LRA, but that is not imminent.
> > I'm sure our assistance would be welcome here, but our port has never
> > reached the stability required for the GCC test suite to produce stable
> > results, which in turn is required for port verification.
> 
> I volunteered some years ago and I volunteer again: Having ported the
> GNU compiler toolchain to VMS/AXP, I believe to have some experience
> maintaining this code ;-)
> 
> Can you point me to a (github?) repo where this code is hosted ?
> 
> Klaus
> 
> -- 
> SUSE Software Solutions Germany GmbH; GF: Felix Imendörffer; (HRB 36809, AG Nürnberg)
> _______________________________________________
> Vax-linux mailing list
> Vax-linux at lug-owl.de
> https://lug-owl.de/mailman/listinfo/vax-linux
> 

Hi,

I know how these posts are often demotivating, so below is my own
progress and self-documentation. However, feel free to completely throw
all of it out.

If you convert upstream GCC to CCmode or LRA, I will have no trouble
merging these patches to that version. Don't worry about it.

CCmode is a minimum requirement because cc0 ports are going to become
deprecated in GCC, then removed.

LRA is less urgent, but it might help us avoid bugs. GCC developers
refer to reload as the devil, and we're one of few neglected
architectures using it.


------------------------------------------------------



I don't know if any of my patches are interesting.
They are split into tasks:
- Merge the netbsd/vax code, which isn't necessarily interesting for
  non-netbsd.

I am discussing them in https://gcc.gnu.org/ml/gcc/2019-09/msg00154.html
This has two remaining issues with the patch:
- The atomic patterns crash when building libatomic.
They previously were not in use because builtins.md was not included.
I'm spending so long stuck on this I might just comment them out.

- flag_dwarf2_cfi_asm = 0, this is crashing when building netbsds libc,
in binutils, like https://github.com/riscv/riscv-binutils-gdb/issues/76
it might be possible to overlook this problem because the fix will not
be in GCC anyway.

- CCmode draft, which might be generally interesting. It cannot compile
basic programs.
https://github.com/coypoop/gcc/tree/vax3

The documentation for turning to CCmode is here:
https://gcc.gnu.org/wiki/CC0Transition


- Additionally, I have my list of distinct crashes with reproducers and
some notes:
https://wiki.netbsd.org/users/maya/vax-gcc/
(These crash on a cross compiler)


- I have some local patches I use to do a complete, clean build. Perhaps
I should create a branch with those for easy testing, with build
instructions I use for netbsd/vax.