Port-vax archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: Toolchain and NetBSD 8



2017-06-14 2:57 GMT+02:00 John Klos <john%ziaspace.com@localhost>:
> NetBSD 8 will be coming soon. The last time I checked, gcc on NetBSD 7 and
> current still don't work on VAX. This means that there'll no longer be an
> active version of NetBSD/VAX that can compile code.

Yes, and to make this clear: Cross-compilation works (vax release can
be built from other archs), native compilation on vax is what is
potentially broken (?).
Furthermore, many hacks (reduced optimization like -O0 for gcc or -O1
for boot) are currently applied to make it work more or less.

> Is anyone working on this? What needs to be done?

You probably saw that I fixed a bug in gcc regarding __builtin_ffs
lately which made the broken install kernel (compiled with -Os...)
work again. As discussed here, variable-length bit field instructions
are "special" and it would be good to revise them all (i. e. their
constraints) in our gcc.

I tried to compile the boot loader with standard optimization (-O2)
recently, but it's still broken then, also see PR port-vax/46098 or:
    https://nxr.netbsd.org/xref/src/sys/arch/vax/boot/boot/Makefile#20

For the -O0 stuff:
    src/external/gpl3/gcc/usr.bin/backend/Makefile
It seems the number of these hacks to make vax work is increasing or
at least alarmingly high, see PR port-vax/51967.

I do not actively work on above topics, up for grabs.
It stands or falls with the compiler...

Felix


Home | Main Index | Thread Index | Old Index