Landing: b74153cc071f

Project / Subsystem

gcc / fold-const-call

Date

2026-08-28

Author

Kyrylo Tkachov

Commit

b74153cc071f4bbedb76f2b72800b2c9467ce996

Source

github

Perf win

No

Breaking

No

All attributes

project
gcc
subsystem
fold-const-call
patch_id
—
commit_hash
b74153cc071f4bbedb76f2b72800b2c9467ce996
source_type
github
headline
Clarify handling of singleton sNaN in constant folding
tldr
Constant folding optimization now includes an explanation for its handling of singleton sNaN encodings.
author
Kyrylo Tkachov
outcome
committed
performance_win
false
breaking_change
false
series_id
—
series_parts
[]
tags
  • • optimization
  • • floating-point
  • • fold-const-call
  • • NaN
discussion_id_link
—
bugzilla_pr
—
date
2026-08-28T00:00:00.000Z

The constant folding optimization pass has been updated to clarify why it uses an exact element-count fallback for singleton sNaN (signaling Not-a-Number) encodings. This explanation is added to the fold_const_reduction function in fold-const-call.cc to provide better understanding of this specific floating-point behavior within the optimization context.