AArch64: Don't set flush-subnormals-to-zero flag on startup
ClosedPublic
Actions

Authored by arichardson on Feb 26 2021, 12:43 PM.

Details

Reviewers

mmel
manu

Commits

rGf979a90f4209: AArch64: Don't set flush-subnormals-to-zero flag on startup
rGd37fb0e37814: AArch64: Don't set flush-subnormals-to-zero flag on startup
rG0e4ff0acbe80: AArch64: Don't set flush-subnormals-to-zero flag on startup

Summary

This flag has been set on startup since 65618fdda0f272a823e6701966421bdca0efa301.
However, This causes some of the math-related tests to fail as they report
zero instead of a tiny number. This fixes at least
/usr/tests/lib/msun/ldexp_test and possibly others.
Additionally, setting this flag prevents printf() from printing subnormal
numbers in decimal form.
See also https://www.openwall.com/lists/musl/2021/02/26/1

PR: 253847

Diff Detail

Repository

rG FreeBSD src repository

Lint

Lint Not Applicable

Unit

Tests Not Applicable

Event Timeline

arichardson created this revision.Feb 26 2021, 12:43 PM

Herald added a reviewer: manu. · View Herald TranscriptFeb 26 2021, 12:43 PM

Herald added subscribers: emaste, andrew, imp. · View Herald Transcript

arichardson requested review of this revision.Feb 26 2021, 12:43 PM

Harbormaster completed remote builds in B37405: Diff 84745.Feb 26 2021, 12:43 PM

typo fix

Harbormaster completed remote builds in B37406: Diff 84746.Feb 26 2021, 12:49 PM

arichardson added inline comments.Feb 26 2021, 12:57 PM

sys/arm64/arm64/vm_machdep.c
58	I just re-ran the lib/msun tests with this set to zero and this results in one fewer failure. Why is this non-default flag being set?

My original idea was to do as much as possible for armv7. Primarily because it has been working on this platform for a long time. My bad is that I didn't remember that flush-to-zero mode was chosen because the little version of armv7 VFP may need software emulation for rounding to denormal values, so we chose the IEEE 754 incompatible mode.