]> sourceware.org Git - glibc.git/commit - ChangeLog
Fix ARM NAN fraction bits.
authorJoey Ye <joey.ye@arm.com>
Thu, 27 Feb 2014 17:44:43 +0000 (17:44 +0000)
committerJoseph Myers <joseph@codesourcery.com>
Thu, 27 Feb 2014 17:44:43 +0000 (17:44 +0000)
commit7d92b78723848ae616709eb8f0191ea067025b18
tree9aea24dcf073852560bd0d88dc8befd39c540603
parent2b7f4f2cada566fe5da5b9260b3ca7b21db2a7e6
Fix ARM NAN fraction bits.

Current ARM soft-float implementation is violating the RTABI
(http://infocenter.arm.com/help/topic/com.arm.doc.ihi0043d/IHI0043D_rtabi.pdf)
Section 4.1.1.1:

When not otherwise specified by IEEE 754, the result on an invalid
operation should be the quiet NaN bit pattern with only the most
significant bit of the significand set, and all other significand bits
zero.

This patch fixes it by setting _FP_NANFRAC_* to zero.

Ran make check test with -mfloat-abi=soft. No regression.

* sysdeps/arm/soft-fp/sfp-machine.h (_FP_NANFRAC_S, _FP_NANFRAC_D)
(_FP_NANFRAC_Q): Set to zero.
ChangeLog
sysdeps/arm/soft-fp/sfp-machine.h
This page took 0.040605 seconds and 5 git commands to generate.