On 12/8/2021 9:08 PM, Alexandre Oliva wrote:
On Dec 8, 2021, Jeff Law <jeffreya...@gmail.com> wrote:
expr.c (emit_move_multi_word): Skip clobber during lra.
OK.
I found a similar pattern of issuing clobbers for multi-word moves, but
not when reload_in_progress, in expr.c:emit_move_complex_parts. I don't
have a testcase, but I'm tempted to propose '!lra_in_progress &&' for it
as well. Can you think of any reason not to?
The only reason I can think of is we're in stage3 :-) It'd be a lot
easier to green light that if we could trigger an issue.
I also see lots of uses of reload_in_progress in machine-dependent code,
and I suspect many cases involving enabling patterns or checking for
legitimate addresses might benefit from the addition of lra_in_progress,
but that's too many occurrences to try to make sense of :-(
Yea, very likely.
jeff
