🚀 go-pugleaf

------- Start of forwarded message -------
Status: RO
X-VM-v5-Data: ([nil nil nil nil nil nil nil nil nil]
	[nil nil nil nil nil nil nil nil nil nil nil nil "^From:" nil nil nil])
Received: from geech.gnu.ai.mit.edu by albert.gnu.ai.mit.edu (5.65/4.0) with SMTP
	id <AA00486@albert.gnu.ai.mit.edu>; Sat, 5 Dec 92 19:37:46 -0500
Received: from nrtc.northrop.com by geech.gnu.ai.mit.edu (5.65/4.0) with SMTP
	id <AA13543@geech.gnu.ai.mit.edu>; Sat, 5 Dec 92 19:37:37 -0500
Received: from charming.nrtc.northrop.com by nrtc.nrtc.northrop.com id aa07049;
          5 Dec 92 16:32 PST
Received: from bat.nrtc.northrop.com by charming.nrtc.northrop.com (4.1/SMI-4.1.1)
	id AA22645; Sat, 5 Dec 92 16:37:23 PST
Message-Id: <2397.723602241@charming.nrtc.northrop.com>
Sender: jpl@charming.nrtc.northrop.com
From: Jeff Lankford <jpl@nrtc.nrtc.northrop.com>
Subject: gcc bug
To: Roland McGrath <roland@geech.gnu.ai.mit.edu>
Cc: jlankford%nrtc.northrop.com@gnu.ai.mit.edu
Date: Sat, 05 Dec 92 16:37:21 PST
Reply-To: jlankford@nrtc.nrtc.northrop.com
Organization: Northrop Research and Technology Center
Location: One Research Park, Palos Verdes, CA 90274
Telephone: 310/544-5394

Roland,
	i'm posting directly to you, since i believe that
	you maintain gcc and glibc.
	i ftp'd latest copies (gcc-2.3.1 and glibc-1.05)
	from prep.ai.mit.edu a week ago.
	i built for sparc-sun-sunos4.1.1.
	ive never built glibc before, but have built previous
	veersions of gcc, gas, gld, ....

	Problem: build of glibc fails with puts.o and putenv.o,
	with something like the following messages:
	puts.c:10: conflicting types for `puts'
	puts.c:5: previous declaration of `puts'

	Demonstration:
	/* gcc bug demonstration program */
	typedef unsigned long int  ANAME;
	extern int FUNC (const char *__s);
	extern ANAME strlen (__const char *__s); /*comment this line or move up */
	int FUNC (const char * s)
	{
        	return 0;
	}

	compile as "gcc -c junk.c" and get
	junk.c:10: conflicting types for `FUNC'
	junk.c:5: previous declaration of `FUNC'

	Solution:
	i don't have any simple solution at this point, because the
	make of glibc is too hard to change and the bug is in gcc anyway.
	note that the problem has to do with builtins (ie, strlen).
	if you comment out the strlen line or move it up one line,
	the compile succeeds; you can play around with changing strlen to strXXX.
	This is too bizarre for me to grok, but maybe you have a
	simple fix?
	the config line to build gcc was
	./configure --target=sparc-sun-sunos4.1.1 --with-gnu-as --with-gnu-ld
	the config line to build gcc was
	./configure sparc-sun-sunos4.1.1 --with-gnu-as --with-gnu-ld
	and i had a configparms file (which i based on INSTALL instructions):
	override stddef.h =
	override stdarg.h =
	defines := -D__SIZE_TYPE__='unsigned long int'

	Why do i think the INSTALL instructions are wrong?
	Any help you can give is appreciated, since i haven't had time to
	monitor the gnu-utils-bugs newsgroups.
thanks,
jpl

------- End of forwarded message -------