cfallin commented on Issue #2490:
Thanks!
Is this problem an isolated case, or could it potentially affect other load-op merges?
I looked at all the other
input_to_reg_mem()
callsites and I don't see any other instances of this problem. That said, I'll consider if there's a way to work this into the types somehow -- perhapsinput_to_reg_mem()
either takes an operand size and asserts it's as expected, or returns a mem op only if the load would be the right size, or something like that... I don't want to block this PR landing on it but will think further!
Last updated: Nov 22 2024 at 16:03 UTC