Compaq Tru64 bootstrap problems

Compaq Tru64 bootstrap problems

[Roger Sayle]

Richard Henderson wrote:
> Roger Sayle wrote:
> > Unfortunately, CVS isn't bootstrapping for me on alphaev67-dec-osf5.1
> > at the moment.
>
> Really?  It built for me on linux on Friday...

I've taken the time to track down the cause of my bootstrap problems
on mainline CVS.  I believe its caused by the recent reorganization of
CPP predefined macros, specifically those relating to current language
settings.

The bootstrap is failing because libgcc2.c includes <time.h> via
"tsystem.h".  On OSF 5.1, <time.h> includes <sys/timers.h> which
fails to compile complaining of an incomplete type for "it_interval".
The cause is that it_interval is defined as "struct timespec" and
with the recent changes timespec is no longer being defined.

Normally, <sys/timers.h> gets its definition of "struct timespec"
from <sys/timemisc.h> where the definition is guarded by the CPP
directive:

#if defined(__LANGUAGE_C__) || defined(__cplusplus)

My guess is that with the recent predefines reorganization we are
no longer defining "__LANGUAGE_C__" on alpha*-dec-osf5.1.  Hopefully,
the fix is trivial, but if not I can file a GNATS bootstrap PR.

Roger
--
Roger Sayle,                         E-mail: roger@eyesopen.com
OpenEye Scientific Software,         WWW: http://www.eyesopen.com/
Suite 1107, 3600 Cerrillos Road,     Tel: (+1) 505-473-7385
Santa Fe, New Mexico, 87507.         Fax: (+1) 505-473-0833

Re: Compaq Tru64 bootstrap problems

[Richard Henderson]

On Thu, May 30, 2002 at 01:54:58PM -0600, Roger Sayle wrote:
> Normally, <sys/timers.h> gets its definition of "struct timespec"
> from <sys/timemisc.h> where the definition is guarded by the CPP
> directive:
> 
> #if defined(__LANGUAGE_C__) || defined(__cplusplus)
> 
> My guess is that with the recent predefines reorganization we are
> no longer defining "__LANGUAGE_C__" on alpha*-dec-osf5.1.  Hopefully,
> the fix is trivial, but if not I can file a GNATS bootstrap PR.

Indeed that would appear to be the case.

The code in builtin_define_std doesn't look right.
Given LANGUAGE_C it defines 

	_LANGUAGE_C
	_LANGUAGE_C__

not the expected

	__LANGUAGE_C
	__LANGUAGE_C__

Zack or Neil, can you comment?


r~

Re: Compaq Tru64 bootstrap problems

[Neil Booth]

Richard Henderson wrote:-

> On Thu, May 30, 2002 at 01:54:58PM -0600, Roger Sayle wrote:
> > Normally, <sys/timers.h> gets its definition of "struct timespec"
> > from <sys/timemisc.h> where the definition is guarded by the CPP
> > directive:
> > 
> > #if defined(__LANGUAGE_C__) || defined(__cplusplus)
> > 
> > My guess is that with the recent predefines reorganization we are
> > no longer defining "__LANGUAGE_C__" on alpha*-dec-osf5.1.  Hopefully,
> > the fix is trivial, but if not I can file a GNATS bootstrap PR.
> 
> Indeed that would appear to be the case.
> 
> The code in builtin_define_std doesn't look right.
> Given LANGUAGE_C it defines 
> 
> 	_LANGUAGE_C
> 	_LANGUAGE_C__
> 
> not the expected
> 
> 	__LANGUAGE_C
> 	__LANGUAGE_C__
> 
> Zack or Neil, can you comment?

Silly me.  Does this fix it?  I've only checked it compiles;
I don't have time right now to check it DTRT.

Neil.

	* c-common.c (builtin_define_std): Correct logic.

Index: c-common.c
===================================================================
RCS file: /cvs/gcc/gcc/gcc/c-common.c,v
retrieving revision 1.338
diff -u -p -r1.338 c-common.c
--- c-common.c	30 May 2002 21:28:11 -0000	1.338
+++ c-common.c	30 May 2002 22:12:53 -0000
@@ -4440,10 +4440,13 @@ builtin_define_std (macro)
 
   /* prepend __ (or maybe just _) if in user's namespace.  */
   memcpy (p, macro, len + 1);
-  if (*p != '_')
-    *--p = '_';
-  if (p[1] != '_' && !ISUPPER (p[1]))
-    *--p = '_';
+  if (!( *p == '_' && (p[1] == '_' || ISUPPER (p[1]))))
+    {
+      if (*p != '_')
+	*--p = '_';
+      if (p[1] != '_')
+	*--p = '_';
+    }
   cpp_define (parse_in, p);
 
   /* If it was in user's namespace...  */

Re: Compaq Tru64 bootstrap problems

[Roger Sayle]

Hi Neil,
> Silly me.  Does this fix it?  I've only checked it compiles;
> I don't have time right now to check it DTRT.
>
> 	* c-common.c (builtin_define_std): Correct logic.
>

Yep.  That fixes this problem.

Unfortunately, bootstrap on alphaev67-dec-osf5.1 still fails to
build libgcc2.c only 23 "files" later, but your patch is definately
an improvement.

Roger
--

Re: Compaq Tru64 bootstrap problems

[Neil Booth]

Roger Sayle wrote:-

> Yep.  That fixes this problem.
> 
> Unfortunately, bootstrap on alphaev67-dec-osf5.1 still fails to
> build libgcc2.c only 23 "files" later, but your patch is definately
> an improvement.

OK, I've applied it.  Thanks,

Neil.