Re: Issues with native build of -10 on VAX

On 15/12/23 21:11, matthew green wrote:
> 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 think that one of the prerequisites is that when pushing the value 
from a bit field onto the stack, the backend has to emit a 
multi-instruction sequence and that the destination operand can't have 
any side-effects.  (The vax backend will emit a rotate+mask pair instead 
of an extzv if the bitfield start and length are constant, but both the 
rotate instruction and the bit clear instruction will write to the 
destination operand.)  I had a quick look at the backends for the other 
targets which set AUTO_INC_DEC and couldn't immediately spot any 
multi-instruction sequences for bitfield operations...

kalvis