This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: PATCH [v4]: Support the FreeBSD 5.0 system thread library model


> On Wed, 23 May 2001, Loren James Rittle wrote:
>> I suspect that you provided --enable-libgcj at configure time.

> Yup.

>> I have an approved patch to allow libjava(libgcj) to build, but it is
>> a kludge that I am embarrassed to apply.

> Well, I guess it's better to somehow fix the problem now somehow if it is
> not that nice and later do a nicer fix, for this might affect other users
> as well (though, for me, changing the configury options is not a problem)?

Agreed.  Crashing `make bootstrap' at library construction-time can be
extremely annoying to unsuspecting users since it happens so late in
the cycle.

Here is the obvious configuration patch I applied to mainline to
handle this situation at initial configuration-time:

2001-05-25  Loren J. Rittle  <ljrittle@acm.org>

	* config.gcc (*-*-freebsd*): Report bad configuration.

Index: config.gcc
===================================================================
RCS file: /cvs/gcc/egcs/gcc/config.gcc,v
retrieving revision 1.61
diff -c -r1.61 config.gcc
*** config.gcc	2001/05/22 19:23:33	1.61
--- config.gcc	2001/05/25 21:57:43
***************
*** 339,344 ****
--- 339,353 ----
  	*) echo 'Unknown thread configuration for FreeBSD'; exit 1;;
  	esac
  	fbsd_tm_file="${fbsd_tm_file} freebsd.h"
+ 	if test x${thread_file} = xposix; then
+ 	  if test x${enable_libgcj} = xyes; then
+ 	    echo 'FreeBSD does not currently handle --enable-threads (now'
+ 	    echo 'the default to match system compiler) and --enable-libgcj.'
+ 	    echo 'Please explicitly configure with --disable-threads or'
+ 	    echo '--disable-libgcj'
+ 	    exit 1
+ 	  fi
+ 	fi
  	;;
  esac
  

I hope all will agree that this is better than building a known bad
configuration of libjava (the library is not buggy only its
configuration for FreeBSD with threads and I am sure I will not have
an acceptable patch done by the 3.0 release).  BTW, this is a better
way to handle this problem than my kludge to handle it inside libjava.

As soon as this configuration at least bootstraps on FreeBSD, I will
remove that fatal configuration error from the mainline.

Regards,
Loren


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]