stage2: lower float negation explicitly + modify hash/eql logic for floats #11967
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Rather than lowering float negation as `0.0 - x`. * Add AIR instruction for float negation. * Add compiler-rt functions for f128, f80 negation closes #11853
Zig guarantees the memory layout of f16, f32, f64, f80, and f128 which means for generic function purposes, values of these types need to be compared on the basis of their bits in memory. This means nan-packing can be used with generic functions, for example. For comptime_float, the sign is observable, whether it is nan is observable, but not any more kinds of bit patterns are observable. This fixes the std.fmt tests that check printing "-nan".
Stage1 (and LLVM) just directly generates an XOR with an appropriate constant, I think stage2 should too. |
Rather than a compiler-rt call in the case that LLVM does not support lowering the fneg instruction.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Rather than lowering float negation as
0.0 - x, this commit makes Zig lower it with a negation AIR instruction. It also adds compiler-rt functions for f128 and f80 negation.This revealed a problem with the way generic functions work for floats:
Zig guarantees the memory layout of f16, f32, f64, f80, and f128 which
means for generic function purposes, values of these types need to be
compared on the basis of their bits in memory. This means nan-packing
can be used with generic functions, for example.
For comptime_float, the sign is observable, whether it is nan is
observable, but not any more kinds of bit patterns are observable.
This fixes the std.fmt tests that check printing "-nan".
closes #11853