re: Issues with native build of -10 on VAX

[matthew green <mrg%eterna23.net@localhost>]

> I *think* I've worked out what the 'reload' pass has done when compiling 
> this test case, though I'm not sure that I understand it well enough to 
> be able to explain it properly.

nice!

> What seems to have happened is that the instructions to push argument 3 
> have inherited the output reload of --%sp into %r1 from the instructions 
> pushing argument 5, but GCC hasn't noted that the pushl $0 for argument 
> 4 has also changed %sp so the output reload can't be reused.
>
> There's a block of code in gcc/reload1.c which does invalidate old 
> output reloads; the attached patch adds to this block to invalidate a 
> reload of an output register if the output register is a reference to 
> memory using an autoincrement or autodecrement addressing mode.
>
> With the patch applied, I've been able to compile the test case using 
> the resulting native compiler, and a native build of NetBSD hasn't 
> failed yet, which is promising.  I have not applied any workarounds to 
> sancov.c or to tree-switch-conversion.c, and I'm also compiling GCC with 
> the default optimisation level (-O2; I haven't applied any of Matthew's 
> changes mentioned earlier).

i think i actually understand what you're fixing and my review of
this change says i don't see anything obvious wrong about it (but
i'm not going to claim i understand it fully. :)

> I've also applied this change to a copy of GCC 10.5.0 from upstream and 
> on amd64 at least there was no change to the regression test results so 
> I don't think this introduces any regressions.

FWIW, it doesn't seem to affect x86 because it doesn't end up 
having AUTO_INC_DEC set, so not compiled there.  looks like for
netbsd targets, that set that will be aarch64, arm, ia64, m68k,
pa, rs6000, sh, and vax.  (i confirmed x86 with a real test,
but this list is a grep-based one.)

this makes me wonder if we can trigger this failure mode on
these other cpus if we try hard, and this fixes for all..

> I'm curious to know if this will work for anyone else.

my netbsd-10 vax simh hung during the install phase of
build.sh release, using the -O1 hacked version, so for
now i've switched to a -current GCC 10 system, with your
this patch applied and no -O1s for cc1plus.  will be a
couple days, but i should have a result soon :)


.mrg.