This repository continues development of the DJGPP run time library.

The original source is available from any mirror of DJGPP as the archives djcrx205.zip, djlsr205.zip and djtzs205.zip.

Goals

Goals of this project are:

• to fix any bugs that might be found
• to support all of C11
• to support as much of Posix as is practical in a DOS environment
• to support the GCC _FORTIFY_SOURCE facility

How to build

I am cross-compiling from Linux Mint using a compiler built from the Github repository https://github.com/andrewwutw/build-djgpp.git . Building natively does not seem to work, at least on FreeDOS; the make utility crashes for no apparent reason. I have not yet tried it on real MS-DOS.

Use "make" to build. Use "make zip" to create djdev206.zip and djtzn206.zip; use these in your DJGPP installation, whether native or a cross-compiler, in place of the corresponding original archives.

Current status of C11 support

The original djlsr205 supports C90 with a few C99 features. Here is the current support for C11:

Headers provided by GCC

• float.h
• iso646.h
• limits.h
• stdalign.h
• stdarg.h
• stdatomic.h
• stddef.h
• stdnoreturn.h
• tgmath.h

Headers completely supported

• assert.h
• complex.h with conformance to Annex G
• ctype.h
• errno.h
• fenv.h
• inttypes.h
• locale.h
• setjmp.h
• signal.h
• stdbool.h
• stdint.h
• stdio.h
• stdlib.h
• string.h
• time.h
• uchar.h
• wctype.h

Headers partially or incorrectly supported

math.h

Many functions in math.h have two versions. The one defined in libc.a is coded specifically for the x87 unit and sets errno on any math errors. The one defined in libm.a is more generically coded, and may set errno or call matherr() on error. See the file info/libc.info for details.

All functions currently implemented in libc.a and libm.a, and defined in C11, pass a number of tests for conformance to Annex F of the C11 standard, provided that _LIB_VERSION is set to _IEEE_. _IEEE_ is now the default; in the original library, the default is _XOPEN_, contrary to what info/libc.info says.

All math functions defined in C11 are now present in libc.a. The ones present in libm.a are listed below.

Function	float	double	long double
acos	libm.a	libm.a	missing
acosh	libm.a	libm.a	missing
asin	libm.a	libm.a	missing
asinh	libm.a	libm.a	missing
atan	libm.a	libm.a	missing
atan2	libm.a	libm.a	missing
atanh	libm.a	libm.a	missing
cbrt	libm.a	libm.a	missing
ceil	libm.a	libm.a	missing
copysign	libm.a	libm.a	missing
cos	libm.a	libm.a	missing
cosh	libm.a	libm.a	missing
erf	libm.a	libm.a	libm.a
erfc	libm.a	libm.a	libm.a
exp	libm.a	libm.a	missing
exp2	missing	missing	missing
expm1	libm.a	libm.a	missing
fabs	libm.a	libm.a	missing
fdim	missing	missing	missing
floor	libm.a	libm.a	missing
fma	missing	missing	missing
fmax	missing	missing	missing
fmin	missing	missing	missing
fmod	libm.a	libm.a	missing
frexp	libm.a	libm.a	missing
hypot	libm.a	libm.a	missing
ilogb	libm.a	libm.a	missing
ldexp	libm.a	libm.a	missing
lgamma	libm.a	libm.a	libm.a
llrint	libm.a	libm.a	libm.a
llround	libm.a	libm.a	libm.a
log	libm.a	libm.a	missing
log10	libm.a	libm.a	missing
log1p	libm.a	libm.a	missing
log2	libm.a	libm.a	missing
logb	libm.a	libm.a	missing
lrint	libm.a	libm.a	libm.a
lround	libm.a	libm.a	libm.a
modf	libm.a	libm.a	missing
nan	libm.a	libm.a	missing
nearbyint	missing	missing	missing
nextafter	libm.a	libm.a	missing
nexttoward	missing	missing	missing
pow	libm.a	libm.a	missing
remainder	libm.a	libm.a	missing
remquo	missing	missing	missing
rint	libm.a	libm.a	libm.a
round	libm.a	libm.a	libm.a
scalbln	missing	missing	missing
scalbn	libm.a	libm.a	missing
sin	libm.a	libm.a	missing
sinh	libm.a	libm.a	missing
sqrt	libm.a	libm.a	missing
tan	libm.a	libm.a	missing
tanh	libm.a	libm.a	missing
tgamma	libm.a	libm.a	libm.a
trunc	libm.a	libm.a	libm.a

libm.a implements additional functions not present in C11:

Function	float	double	long double
cabs	libm.a	libm.a	missing
drem	libm.a	libm.a	missing
finite	libm.a	libm.a	libm.a
gamma	libm.a	libm.a	missing
gamma_r	libm.a	libm.a	missing
infinity	libm.a	libm.a	missing
isinf	libm.a	libm.a	libm.a
isnan	libm.a	libm.a	libm.a
j0	libm.a	libm.a	missing
j1	libm.a	libm.a	missing
jn	libm.a	libm.a	missing
lgamma_r	libm.a	libm.a	missing
scalb	libm.a	libm.a	missing
significand	libm.a	libm.a	missing
y0	libm.a	libm.a	missing
y1	libm.a	libm.a	missing
yn	libm.a	libm.a	missing

wchar.h

• All functions at least minimally implemented
• No locale support in:
  • wcscoll -- same as wcscmp, no locale-dependent behavior
  • wcsxfrm -- identity map, no locale-dependent behavior

Status of _FORTIFY_SOURCE support

The following functions provide additional boundary checking if _FORTIFY_SOURCE is in effect:

• Buffer overruns trigger a message to stderr and an abort.
• if _FORTIFY_SOURCE is at least 2, printf-like functions may not use the %n format specifier when the format string is not constant. This does not apply to vprintf-like functions, because they are typically used with wrappers to make new printf-like functions.

string.h

• memcpy
• memmove
• memset
• strcat
• strcpy
• strncat
• strncpy
• strxfrm
• memccpy
• stpcpy
• stpncpy
• strlcat
• strlcpy

stdio.h

• fgets
• gets
• fread
• snprintf
• sprintf
• vsnprintf
• vsprintf
• fprintf
• printf
• dprintf
• asprintf
• asnprintf
• vasnprintf

gets is furthermore marked as deprecated, and its use will trigger an error if C11 or higher is in effect.

stdlib.h

• mbstowcs
• wcstombs
• wctomb
• realpath

time.h

• strftime

wchar.h

• wprintf
• fwprintf
• swprintf
• vwprintf
• vfwprintf
• vswprintf
• fgetws
• wcscpy
• wcsncpy
• wcscat
• wcsncat
• wcsxfrm
• wmemcpy
• wmemmove
• wmemset
• wcsftime
• wcrtomb
• mbsrtowcs
• wcsrtombs

unistd.h

• confstr
• getcwd
• getgroups
• read
• gethostname
• getwd
• readlink