JIT: Fold unsigned constant range tests

[EgorBo]

Closes #135418

Enable unsigned constant range folding in optOptimizeBools. For uint n:

bool Pattern() => n is >= '0' and <= '9';

-       mov      ecx, dword ptr [rcx+0x08]
-       cmp      ecx, 48
-       jae      SHORT G_M53568_IG05
-       xor      eax, eax
-       ret
-G_M53568_IG05:
-       cmp      ecx, 57
+       mov      eax, dword ptr [rcx+0x08]
+       sub      eax, 48
+       cmp      eax, 9
        setbe    al
        movzx    rax, al
        ret
-; Total bytes of code: 21
+; Total bytes of code: 16

[azure-pipelines]

Azure Pipelines:
Successfully started running 5 pipeline(s).
11 pipeline(s) were filtered out due to trigger conditions.
There may be pipelines that require an authorized user to comment /azp run to run.

[dotnet-policy-service]

Tagging subscribers to this area: @JulieLeeMSFT, @jakobbotsch
See info in area-owners.md if you want to be subscribed.

[Copilot]

🟡 Changes recommended

The newly enabled unsigned transformation needs targeted regression coverage; builds and runtime tests were not run during this review.

1 open finding

• Add regression tests for unsigned range-folding transformation · New

What changed in this PR

Extends the JIT’s existing range folding to unsigned comparisons, addressing the digit-range optimization in #135418.

Changes:

• Allows unsigned constant range tests.
• Requires matching signedness for folding and preserves the signed-only nonnegative fallback.

File	Description
src/​coreclr/​jit/​optimizebools.cpp	Enables unsigned range folding with signedness guards.

🧠 Review effort: Balanced

---

Give feedback about Copilot approvals in this survey to enter a drawing for a $150 gift card.

[EgorBo]

PTAL @adamperlin @dotnet/jit-contrib The diffs are quite small, but there is no reason to give up on Unsigned in that transformation (I remember I just conservatively gave up on them there). Fixes user's repro: #135418

[Corniel]

@EgorBo Note that in my example, I used an uint but also char. In both cases it was not applied. If understand correctly, the char case should already have worked?!

[adamperlin]

This LGTM. Do existing tests for range folding cover the unsigned case?

[adamperlin]

@EgorBo Note that in my example, I used an uint but also char. In both cases it was not applied. If understand correctly, the char case should already have worked?!

The JIT actually represents char as ushort, so I believe it's consistent that the optimization wasn't being applied to the char case before!

[Corniel]

The JIT actually represents char as ushort, so I believe it's consistent that the optimization wasn't being applied to the char case before!

It makes a lot of sense, and explains why it did not work either way.