CVS log for src/lib/libc/string/__strsignal.c

Up to [cvs.NetBSD.org] / src / lib / libc / string

Request diff between arbitrary revisions

Keyword substitution: kv
Default branch: MAIN

Pull up following revision(s) (requested by kre in ticket #1004):

	tools/compat/configure: revision 1.106
	lib/libc/string/__strsignal.c: revision 1.26
	tools/compat/configure.ac: revision 1.108
	tools/compat/nbtool_config.h.in: revision 1.57
	usr.bin/elf2aout/elf2aout.c: revision 1.24
	tools/compat/Makefile: revision 1.93
	sys/arch/macppc/stand/fixcoff/fixcoff.c: revision 1.13
	tools/compat/compat_defs.h: revision 1.123

Include <arpa/inet.h> for htonl()  (From Jan-Benedict Glaw)
Probably this should be using htobe32() instead of htonl()
(and so <endian.h> rather than <arpa/inet.h>) as there is
no sign of any network anywhere near here.   But that's for
another day, and to be done by someone who can test it.

Portability fixes from Jan-Benedict Glaw
The args to help() and usage() need to be const char *
and main() should be declared as returning int.

Add shquote to libnbcompat
Include shquote() in configure tests
Regen for added shquote()

Stop attempting to return a (char*) in place of (const char *)
This removes a cast that was added in 1994, when __strsignal()
returned a char * to convert the const char * element of sys_siglist
into a type that matched.

It should have been removed when __strsignal() was changed to
return const char * in 1998.   But wasn't.

This should fix an inane warning (treated as an error) from a
compile stupid enough to allow a natural char * (as in char buf[N])
to be returned as a result (which is, of course, fine), but complains
about returning (char *) applied to a (const char *).

Delete unnecessary definitions of SIGRTMIN & SIGRTMAX

Long ago, SIGRTMIN & SIGRTMAX were #ifdef _KERNEL in
<sys/signal.h> so duplicate definitions were added
here (#ifndef SIGRTMIN so the duplicates vanished for
all practical purposes in 2016 when the symbols were
exposed to userland).

NFC (unless someone hides SIGRTMIN & SIGRTMAX again).

KNF.    NFC.

Stop attempting to return a (char*) in place of (const char *)

This removes a cast that was added in 1994, when __strsignal()
returned a char * to convert the const char * element of sys_siglist
into a type that matched.

It should have been removed when __strsignal() was changed to
return const char * in 1998.   But wasn't.

This should fix an inane warning (treated as an error) from a
compile stupid enough to allow a natural char * (as in char buf[N])
to be returned as a result (which is, of course, fine), but complains
about returning (char *) applied to a (const char *).

sync with head

Use C89 Prototypes.
Remove use of __P

Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22280, verified by myself.

Sync with HEAD.

use strlcpy.  correct mis(non-)use of buflen argument.

Catch up to -current

Fix signed/unsigned comparison warnings.

Catch up to -current.

add rt signal names.

Catch up to -current.

make strerror and strsignal use NL_CAT_LOCALE.

back out the #ifdef _DIAGNOSTIC argument checks; too many people complained.
_DIAGASSERT() is still retained.

* use _DIAGASSERT() to check pointer arguments against NULL and file
  descriptors against -1 (as appropriate).
* add actual checks which to detect stuff that would trigger_DIAGASSERT(),
  and attempt to return a sane error condition.
* knf some code
* remove some `register' decls.

the first two items result in the addition of code similar to the
following in various functions:

		_DIAGASSERT(path != NULL)
	#ifdef _DIAGNOSTIC
		if (path == NULL) {
			errno = EFAULT;
			return (-1);
		}
	#endif

Use the (new) sys_nsig instead of NSIG.

delint

const poisoning.

remove obsolete register declarations

Fix RCSID's
Use "extern.h" where appropriate.

- convert unsafe strcpy(), strcat() and sprintf() to the `n' versions.
- some KNF.

Pullup request from Mike Long <mike.long@analog.com>

Avoids buffer overruns as described in PR 2780.

stnrcpy() isn't a function.  strncpy() is.  Use it instead.

Since the user provide his own message catalog (via NLSPATH and LANG)
the __strerror() and __strsignal() functions can not assume that the
messages will not overflow a NL_TEXTSIZE-sized buffer.

Noted by Mike Long <mike.long@analog.com> in PR

snapshot namespace cleanup

don't include <sys/cdefs.h>, it's a 'private' header

cast last argument of catgets() to (char *) to quiet compiler's complaints
about loss of 'const'-ness.

Use namespace.h, back out old mechanism used for namespace purity.

Rework indirect reference support as outlined by my recent message to
the tech-userlevel mailing list.

Private versions of cat*() functions now have only one leading underscore.

case to quiet compiler.

Use __sys_siglist[].

Call internal NLS interfaces: __catopen(), __catgets(), and __catclose().

Don't define sys_siglist, <signal.h> does that already

__strerror() and __strsignal() have been broken out into their own files
and use message catalogs if NLS is defined.
Increased the size of char buffers used by strerror() and strsignal();
the error and signal strings may be much longer in some locales.