Emit the enum range assumption if the range only contains one element

[Rustin170506]

close #82871

[rust-highfive]

r? @lcnr

(rust-highfive has picked a reviewer for you, use r? to override)

[the8472]

Naive question, but couldn't it be turned into a constant instead if there can only be a single value?

[nikic]

This needs a codegen test to make sure the desired optimization takes place. The src/test/codegen/enum-bounds-check*.rs tests can serve as inspiration.

[Rustin170506]

Now it's look like:

	.section	__TEXT,__text,regular,pure_instructions
	.macosx_version_min 10, 7
	.globl	_index
	.p2align	4, 0x90
_index:
	.cfi_startproc
	pushq	%rbp
	.cfi_def_cfa_offset 16
	.cfi_offset %rbp, -16
	movq	%rsp, %rbp
	.cfi_def_cfa_register %rbp
	movl	(%rdi), %eax
	popq	%rbp
	retq
	.cfi_endproc

.subsections_via_symbols

[Rustin170506]

I think it's ready for review.

[lcnr]

this does seem fine to me, would it be possible to instead emit a constant if start == end? it feels a bit weird to me to use range assumptions for this.

edit: ah, that was already asked in #83020 (comment)

[nikic]

this does seem fine to me, would it be possible to instead emit a constant if start == end? it feels a bit weird to me to use range assumptions for this.

That would be possible, but maybe not worth the bother to special case it? LLVM will end up replacing the value with a constant based on the assumption, so I don't think it will make much of a difference.

[lcnr]

in this case

@bors r+

[bors]

📌 Commit d180f91 has been approved by lcnr