Go to file
Rich Felker 8eead3ef18 math: explicitly promote expressions to excess-precision types
a conforming compiler for an arch with excess precision floating point
(FLT_EVAL_METHOD!=0; presently i386 is the only such arch supported)
computes all intermediate results in the types float_t and double_t
rather than the nominal type of the expression. some incorrect
compilers, however, only keep excess precision in registers, and
convert down to the nominal type when spilling intermediate results to
memory, yielding unpredictable results that depend on the compiler's
choices of what/when to spill. in particular, this happens on old gcc
versions with -ffloat-store, which we need in order to work around
bugs where the compiler wrongly keeps explicitly-dropped excess
precision.

by explicitly converting to double_t where expressions are expected be
be evaluated in double_t precision, we can avoid depending on the
compiler to get types correct when spilling; the nominal and
intermediate precision now match. this commit should not change the
code generated by correct compilers, or by old ones on non-i386 archs
where double_t is defined as double.

this fixes a serious bug in argument reduction observed on i386 with
gcc 4.2: for values of x outside the unit circle, sin(x) was producing
results outside the interval [-1,1]. changes made in commit
0ce946cf80 were likely responsible for
breaking compatibility with this and other old gcc versions.

patch by Szabolcs Nagy.
2015-11-21 21:41:42 -05:00
arch fix dynamic loader library mapping for nommu systems 2015-11-11 17:40:27 -05:00
crt explicitly assemble all arm asm sources as UAL 2015-11-10 00:01:55 -05:00
dist add another example option to dist/config.mak 2012-04-24 16:49:11 -04:00
include fix mismatched parens in CMPLX def for annex-g-conforming compilers 2015-11-02 21:44:57 -05:00
lib new solution for empty lib dir (old one had some problems) 2011-02-17 17:12:52 -05:00
src math: explicitly promote expressions to excess-precision types 2015-11-21 21:41:42 -05:00
tools add CFI generation script for x86_64 2015-10-13 18:09:46 -04:00
.gitignore add musl-clang, a wrapper for system clang installs 2015-07-06 23:52:16 +00:00
configure work around toolchains with broken visibility in libgcc/libpcc 2015-11-07 20:23:49 -05:00
COPYRIGHT update authors/contributors list 2015-03-16 18:43:54 -04:00
INSTALL update notice on broken gcc versions in INSTALL file 2014-07-31 19:02:54 -04:00
Makefile fix build regression from removal of #ifdef SHARED 2015-11-18 19:00:44 -05:00
README update version reference in the README file 2014-06-25 14:16:53 -04:00
VERSION release 1.1.12 2015-10-19 19:12:57 -04:00
WHATSNEW release 1.1.12 2015-10-19 19:12:57 -04:00

    musl libc

musl, pronounced like the word "mussel", is an MIT-licensed
implementation of the standard C library targetting the Linux syscall
API, suitable for use in a wide range of deployment environments. musl
offers efficient static and dynamic linking support, lightweight code
and low runtime overhead, strong fail-safe guarantees under correct
usage, and correctness in the sense of standards conformance and
safety. musl is built on the principle that these goals are best
achieved through simple code that is easy to understand and maintain.

The 1.1 release series for musl features coverage for all interfaces
defined in ISO C99 and POSIX 2008 base, along with a number of
non-standardized interfaces for compatibility with Linux, BSD, and
glibc functionality.

For basic installation instructions, see the included INSTALL file.
Information on full musl-targeted compiler toolchains, system
bootstrapping, and Linux distributions built on musl can be found on
the project website:

    http://www.musl-libc.org/