_strtod_l as well as the gethex function both fetch the decimal point
from the current LC_NUMERIC locale info. This pulls in _C_numeric_locale
unconditionally even on targets not supporting locales at all.
Another problem is that strtod.c and gdtoa-gethex.c are ELIX 1, while
locale information in general isn't. This leads to potential build
breakage on bare metal targets.
Fix this by setting the decimal point to "." on all targets not
defining __HAVE_LOCALE_INFO__.
While at it, const'ify the entire local decimal point info in the
affected functions.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
The C standard says that errno may acquire the value ERANGE if the
result from strtod underflows. According to IEEE 754, underflow occurs
whenever the value cannot be represented in normalized form.
Newlib is inconsistent in this, setting errno to ERANGE only if the
value underflows to zero, but not for denorm values, and never for hex
format floats.
This patch attempts to consistently set errno to ERANGE for all
'underflow' conditions, which is to say all values which are not
exactly zero and which cannot be represented in normalized form.
This matches glibc behavior, as well as the Linux, Mac OS X, OpenBSD,
FreeBSD and SunOS strtod man pages.
Signed-off-by: Keith Packard <keithp@keithp.com>
The string/float conversion functions need to get the locale decimal
point. Instead of calling __localeconv_l (which copies locale data
into lconv form from __get_numeric_locale), use __get_numeric_locale
directly.
Signed-off-by: Keith Packard <keithp@keithp.com>
strtof ("-nan") returned positive NaN instead of negative NaN.
strtod ("-nan") and strtold ("-nan") return negative NaN.
Linux glibc has been fixed
that strto{f|d|ld} ("-nan") returns negative NaN.
https://sourceware.org/bugzilla/show_bug.cgi?id=23007
This commit makes strtof preserves the negative sign bit
when parsing "-nan" like glibc.
By previous commit, strto{d|ld} ("nan")
does not use the definition of NaN.
There is no other function that uses the definitions.
This commit remove the definitions.
The definition of qNaN for x86_64 and i386 was wrong.
strto{d|ld} ("nan") returned wrong negative NaN
instead of correct positive NaN
since it used the wrong definition.
On the other hand, strtof ("nan") returns correct positive NaN
since it uses nanf ("") instead of the wrong definition.
This commit makes strto{d|ld} ("nan") uses {nan|nanl} ("")
like strtof ("nan") using.
So strto{d|ld} ("nan") returns positive NaN.
The gdtoa implementation uses the type long, defined as Long, in lots
of code. For historical reason newlib defines Long as int32_t instead.
This works fine, as long as floating point exceptions are not enabled.
The conversion to 32 bit int can lead to a FE_INVALID situation.
Example:
const char *str = "121645100408832000.0";
char *ptr;
feenableexcept (FE_INVALID);
strtod (str, &ptr);
This leads to the following situation in strtod
double aadj;
Long L;
[...]
L = (Long)aadj;
For instance, on x86_64 the code here is
cvttsd2si %xmm0,%eax
At this point, aadj is 2529648000.0 in our example. The conversion to
32 bit %eax results in a negative int value, thus the conversion is
invalid. With feenableexcept (FE_INVALID), a SIGFPE is raised.
Fix this by always using 64 bit ints here if double is not a 32 bit type
to avoid this type of FP exceptions.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
* libc/stdlib/strtod.c (strtof_l): Set errno to ERANGE when double to
float conversion results in infinity.
(strtof): Likewise.
* libc/stdlib/wcstod.c (wcstof_l): Likewise.
(wcstof): Likewise.
radix char instead of assuming length 1.
* libc/stdlib/gdtoa-gethex.c: Remove use of USE_LOCALE.
(gethex): Allow multibyte decimal point.
Fix compiler warnings due to different signedness of pointer types.
* libc/stdlib/strtod.c: Remove use of USE_LOCALE.
(_strtod_r): Allow multibyte decimal point.
* libc/stdlib/wcstod.c (_wcstod_r): Evaluate correct wide char
endptr position if the decimal point is a multibyte char.
_MB_CAPABLE systems.
* libc/ctype/iswblank.c: Ditto.
* libc/ctype/iswcntrl.c: Ditto.
* libc/ctype/iswprint.c: Ditto.
* libc/ctype/iswpunct.c: Ditto.
* libc/ctype/iswspace.c: Ditto.
* libc/ctype/jp2uc.c (__jp2uc): On Cygwin, just return c.
Explain why.
* libc/ctype/towlower.c: Ditto.
* libc/ctype/towupper.c: Ditto.
* libc/include/sys/config.h: Define _MB_EXTENDED_CHARSETS_ISO
and _MB_EXTENDED_CHARSETS_WINDOWS if _MB_EXTENDED_CHARSETS_ALL is
defined. Define _MB_EXTENDED_CHARSETS_ALL on Cygwin only for now.
* libc/include/sys/reent.h (struct _reent): Mark _current_category
and _current_locale as unused.
* libc/locale/locale.c: Add new charset support to documentation.
Include ../stdio/local.h from here.
(lc_ctype_charset): Set to "ASCII" by default.
(lc_message_charset): Ditto.
(_setlocale_r): Don't set _current_category and _current_locale.
(loadlocale): Add Cygwin codepage support. On _MB_CAPABLE
systems, set __mbtowc and __wctomb function pointers to function
corresponding with current charset. Don't allow non-existant
ISO-8859-12 charset. Add support for Windows singlebyte codepages.
On Cygwin, add support for GBK, CP949, and BIG5. On Cygwin,
call __set_ctype() in case the catorgy is LC_CTYPE. Don't set
_current_category and _current_locale.
* libc/stdlib/Makefile.am (GENERAL_SOURCES): Add sb_charsets.c.
* libc/stdlib/Makefile.in: Regenerate.
* libc/stdlib/local.h: Add prototype for __locale_charset.
Add prototypes for __mbtowc and __wctomb pointers.
Add prototypes for charset-specific _wctomb_r and _mbtowc_r
functions.
Declare tables and functions from sb_charsets.c.
* libc/stdlib/mbtowc_r.c (__mbtowc): Define. Set to __ascii_mbtowc
by default.
(_mbtowc_r): Just call __mbtowc from here.
(__ascii_mbtowc): New function.
(__iso_mbtowc): New function.
(__cp_mbtowc): New function.
(__utf8_mbtowc): New function.
(__sjis_mbtowc): New function. Disable on Cygwin.
(__eucjp_mbtowc): New function. Disable on Cygwin.
(__jis_mbtowc): New function. Disable on Cygwin.
* libc/stdlib/sb_charsets.c: New file, adding singlebyte to UTF
conversion tables for all ISO and CP charsets.
(__iso_8859_index): New function.
(__cp_index): New function.
* libc/stdlib/wctomb_r.c (__wctomb): Define. Set to __ascii_wctomb
by default.
(_wctomb_r): Just call __wctomb from here.
(__ascii_wctomb): New function.
(__utf8_wctomb): New function.
(__sjis_wctomb): New function. Disable on Cygwin.
(__eucjp_wctomb): New function. Disable on Cygwin.
(__jis_wctomb): New function. Disable on Cygwin.
(__iso_wctomb): New function.
(__cp_wctomb): New function.
* libc/stdlib/strtod.c (_strtod_r): Change local variables aadj,
rv, rv0 from double to type U. Use accessor macros dval, dword0
and dword1 for all accesses except for the ULtod call, where rv.i
replaces the pointer cast.
* libc/stdlib/mprec.h (U): Rename member L to i for easier re-use
of access macros. Tweak comment.
Remove #ifdef'd YES_ALIAS code.
(dword0, dword1, dval): Define in terms of uncast union member
access. Ditto for _DOUBLE_IS_32BITS variants.
(Storeinc): Replace aliasing-flawed microoptimized definition with
alternative suggested in comment. Remove now stale comment.
* libc/stdlib/mprec.h [_DOUBLE_IS_32BITS]: Turn off C99 hex
floating-point format support. Also redefine
dword0 and dword1 macros.
* libc/stdlib/strtod.c: Add checks for _DOUBLE_IS_32BITS
to prevent setting dword1 which is an rvalue only.
* libc/stdlib/Makefile.am: Add new gdtoa routines.
* libc/stdlib/Makefile.in: Regenerated.
* libc/stdlib/gd_qnan.h: New file.
* libc/stdlib/gdtoa-gethex.c: Ditto.
* libc/stdlib/gdtoa-hexnan.c: Ditto.
* libc/stdlib/gdtoa.h: Ditto.
* libc/stdlib/mprec.c: Add new helper routines needed by
the new gdtoa code.
* libc/stdlib/mprec.h: Integrate some defines and prototypes
used by gdtoa routines here.
* libc/stdlib/strtod.c: Rebased on David M. Gay's gdtoa-strtod.c
which adds C99 support such as nan, inf, and hexadecimal input
format.
* libc/stdlib/strtod.c (_strtod_r): Add NaN support.
* (strtof): Ditto.
* libc/stdio/vfscanf.c (__svfscanf_r): Ditto.
* Makefile.am (MATHOBJS_IN_LIBC): Add s_nan and sf_nan
functions for use by strtod and strtof.
* Makefile.in: Regenerated.
* libc/include/stdlib.h (strtof): New prototype (from C99).
(strtodf): Changed from prototype to macro which redefines
to strtof.
* libc/stdlib/atof.c: Change documentation to refer to strtof
instead of strtodf.
* libc/stdlib/atoff.c (atoff): Change to call strtof instead of
strtodf.
* libc/stdlib/strtod.c (strtodf): Renamed to strtof.
(strtof): New function.
* libm/test/convert.c (test_strtodf): Renamed to test_strtof which
calls strtof.
* libc/include/stdio.h[!_REENT_ONLY]: Moved various functions together
into one list.
[!__STRICT_ANSI__]: Moved non-ANSI I/O functions in this list.
(vfscanf, vscanf, vsscanf, _vfscanf_r, _vscanf_r, _vsscanf_r): New
function prototypes.
(_fscanf_r, _sscanf_r): Ditto.
* libc/include/stdlib.h: Added _strtod_r prototype.
* libc/stdio/Makefile.am: Add new v*scanf functions.
* libc/stdio/Makefile.in: Regenerate.
* libc/stdio/fscanf.c: Reorganized so HAVE_STDC only affects prototype
and code is shared. Added reentrant _fscanf_r which calls __svfscanf_r. * libc/stdio/scanf.c: Changed to call __svfscanf_r.
* libc/stdio/sscanf.c: Changed documentation to add reentrant routines.
(sscanf): Changed to call __svfscanf_r with _REENT argument.
(_sscanf_r): New routine.
* libc/stdio/local.h: Removed __svfscanf prototype and replaced it
with __svfscanf_r prototype.
* libc/stdio/vfscanf.c (vfscanf, _vfscanf_r: New
routines.
(__svfscanf_r): Reentrant version of __svfscanf which takes reetrancy
structure as argument as calls reentrant versions of helper functions
(e.g. _strtol_r, _strtoul_r). Also replaced calls to atol and atof
to _strtol_r and _strtod_r respectively.
* libc/stdio/vfscanf.c: Also changed __svfscanf to call __svfscanf_r.
* libc/stdlib/strtod.c (strtod): Changed to call _strtod_r with
_REENT argument.
* libc/stdio/vscanf.c: New file.
* libc/stdio/vsscanf.c: Ditto.
Corinna Vinschen
Keith Packard
Masamichi Hosoda
Yaakov Selkowitz
Jeff Johnston
Joel Sherrill
Hans-Peter Nilsson
Eric Blake
Christopher Faylor