Fix ILogB for Half and BFloat16 subnormal values

[Copilot]

• Analyze the issue: ILogB for Half and BFloat16 returns incorrect values for subnormal numbers because BitOperations.LeadingZeroCount only has overloads for uint/ulong/nuint, so the ushort from TrailingSignificand is implicitly widened to uint, adding 16 extra leading zeros
• Fix Half.ILogB to use ushort.LeadingZeroCount which handles the bit-width adjustment
• Fix BFloat16.ILogB to use byte.LeadingZeroCount after changing TrailingSignificand to return byte
• Add tests to verify the fix for both Half.ILogB and BFloat16.ILogB with normal and subnormal values
• Build and run tests to verify the fix (69,569 tests passed with 0 failures)

Security Summary

No security vulnerabilities were discovered or introduced by these changes.

Original prompt

---

This section details on the original issue you should resolve

<issue_title>Possibly incorrect implementation of ILogB for BFloat16 and Half</issue_title>
<issue_description>ILogB is implemented on Half as such:

runtime/src/libraries/System.Private.CoreLib/src/System/Half.cs

Line 1535 in 8ff3668

return MinExponent - (BitOperations.LeadingZeroCount(x.TrailingSignificand) - BiasedExponentLength);

and TrailingSignificand is a ushort

runtime/src/libraries/System.Private.CoreLib/src/System/Half.cs

Line 137 in 8ff3668

internal ushort TrailingSignificand

However, BitOperations.LeadingZeroCount has no overload that accepts a ushort. So it is implicitly widened to the uint overload, and calculates the leading zero count as if the number had a 32-bit width instead of 16-bit.

I believe that means that Half.ILogB may be off by 16 for subnormal values (and probably BFloat16).

For example, for Half.Epsilon I would expect the ilogb to be -24, but is reported as -40.</issue_description>

Comments on the Issue (you are @copilot in this section)

• Fixes Possibly incorrect implementation of ILogB for BFloat16 and Half #123293

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.

[Copilot]

Pull request overview

This PR fixes a bug in ILogB for Half and BFloat16 where subnormal values returned incorrect results due to implicit type widening. When BitOperations.LeadingZeroCount was called with a ushort argument (the type of TrailingSignificand), it was implicitly widened to uint, adding 16 extra leading zeros to the count.

Changes:

• Added internal LeadingZeroCount overloads for ushort and byte in BitOperations that handle the bit-width adjustment
• Refactored LeadingZeroCount in UInt16, Int16, Byte, SByte, and Char to use the new internal overloads, removing redundant manual adjustments
• Added comprehensive tests for ILogB covering normal, special, and subnormal values for both Half and BFloat16

Reviewed changes

Copilot reviewed 8 out of 8 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
BitOperations.cs	Adds internal LeadingZeroCount overloads for ushort and byte that correctly adjust for bit width when widening to uint
UInt16.cs	Removes manual -16 adjustment, now uses internal LeadingZeroCount(ushort) overload
Int16.cs	Removes manual -16 adjustment, now uses internal LeadingZeroCount(ushort) overload
Byte.cs	Removes manual -24 adjustment, now uses internal LeadingZeroCount(byte) overload
SByte.cs	Removes manual -24 adjustment, now uses internal LeadingZeroCount(byte) overload
Char.cs	Removes manual -16 adjustment, now uses internal LeadingZeroCount(ushort) overload
HalfTests.cs	Adds tests for ILogB including normal values, special values, and two subnormal test cases
BFloat16Tests.cs	Adds tests for ILogB including normal values, special values, and two subnormal test cases