alexcrichton commented on issue #3561:
Thanks for pointing this out! I didn't realize this was one of the goals but makes sense to me. I can also audit the other existing aarch64 rules soon as well to make sure they follow this principle.
For this particular function though I think it may be best to match
$I64
perhaps instead of anything bigger than 32-bits since I think that's what the basic intention was, one thing for 64-bit and something else for smaller. Otherwise though r=me
cfallin commented on issue #3561:
Thanks! Yeah I think this is just a natural "bump into the corners of the language semantics as we settle into it" thing, to be expected to some degree.
I scanned over the rest of the aarch64 ISLE code and it looks good to me; and in any case the worst that happens if we miss a rule-ordering-dependent thing is that it pops up when we actually do testing that depends on it. Since the ordering is locked down/deterministic right now and we're testing the result of that, I'm pretty satisfied that's enough for now :-)
Updated as suggested,
$I64
-specific now.
Last updated: Dec 23 2024 at 13:07 UTC