Aldy Hernandez | 1 Jul 01:59 2002
Picon

gcc.dg/20020312-2.c: check for __PPC__

__powerpc__ is only defined for:

CPP_OS_FREEBSD_SPEC
CPP_OS_NETBSD_SPEC

checking for __PPC__ is a bit more generic.

is this ok?

2002-06-30  Aldy Hernandez  <aldyh <at> redhat.com>

	* gcc.dg/20020312-2.c: Check for __PPC__.

Index: testsuite/gcc.dg/20020312-2.c
===================================================================
RCS file: /cvs/uberbaum/gcc/testsuite/gcc.dg/20020312-2.c,v
retrieving revision 1.7
diff -c -p -r1.7 20020312-2.c
*** testsuite/gcc.dg/20020312-2.c	24 Apr 2002 21:47:44 -0000	1.7
--- testsuite/gcc.dg/20020312-2.c	30 Jun 2002 23:56:00 -0000
***************
*** 72,78 ****
  /* No pic register.  */
  #elif defined(__pj__)
  /* No pic register.  */
! #elif defined(__powerpc__)
  # ifdef __darwin__
  #  define PIC_REG  "31"
  # else
--- 72,78 ----
(Continue reading)

Stan Shebs | 1 Jul 03:00 2002
Picon

Re: Patch: libobjc reformat. (long, gzipped)

rbrown64 <at> csc.com.au wrote:
> 
> No objc testsuite regressions after rebuilding (not bootstrap from scratch) on
> i586-pc-linux-gnu. I'll work my way through the rest of the thr-*.c files but will
> probably not have the platforms to test them on.

Thanks, this looks very nice, feel free to commit it.  There are a
couple arithmetic expressions that could be spread out a little more
(look for 8* and %100).  Also, I think the newer style for '!' is
to not have the space following, but it's not important enough to
hold up the patch.

Stan

> 2002-06-28  Rodney Brown  <rbrown64 <at> csc.com.au>
>       * objc/encoding.h: Fix formatting.
>       * objc/hash.h: Likewise.
>       * objc/objc-api.h: Likewise.
>       * objc/runtime.h: Likewise.
>       * objc/thr.h: Likewise.
>       * archive.c: Likewise.
>       * class.c: Likewise.
>       * encoding.c: Likewise.
>       * gc.c: Likewise.
>       * hash.c: Likewise.
>       * init.c: Likewise.
>       * misc.c: Likewise.
>       * nil_method.c: Likewise.
>       * objects.c: Likewise.
>       * sarray.c: Likewise.
(Continue reading)

Eric Christopher | 1 Jul 05:45 2002
Picon

Re: Patch to fix FUNCTION_ARG_REGNO_P for mips


> Ok to install?

Please. Thanks.

-eric

 
> 
> 2002-06-25  Kaveh R. Ghazi  <ghazi <at> caip.rutgers.edu>
> 
> 	* mips.h (FUNCTION_ARG_REGNO_P): Fix parentheses.
> 

--

-- 
I will not carve gods

Bryce McKinlay | 1 Jul 05:47 2002
Picon

Re: TEXI patch: CNI Invocation example

T.J. Mather wrote:

>This patch is needed for the example to compile correctly using the GCC 
>3.1 release.
>
>Thanks,
>TJ
>
>2002-06-29  T.J. Mather  <tjmather <at> maxmind.com>
>
>	* gcj.texi: Fixed gcj invocation example so that it compiles.
>  
>

I checked this in. Thanks!

Bryce.

Bryce McKinlay | 1 Jul 05:49 2002
Picon

Re: WWW patch: links to GCJ Manual

T.J. Mather wrote:

>This patch adds a link to the GCJ Manual from the docs.html, and
>includes an updated link to the section on Invocation from faq.html.
>
>
>2002-06-29  T.J. Mather  <tjmather <at> maxmind.com>
>
>	* docs.html: Added link to GCC GCJ Manual
>	* faq.html: Updated link to Invocation section in GCJ Manual
>

Thanks. I checked this in, with a slight variation to the faq change.

regards

Bryce.

Gerald Pfeifer | 1 Jul 10:53 2002
Picon

Re: GCC 3.1.1 Status

On Fri, 28 Jun 2002, Gerald Pfeifer wrote:
> Thanks!  I have put the PR c/6957 into feedback state, and will close
> this high priority bug once we have feedback from the submitter.

Bernd, I just noticed that PR 7063 seems to be a duplicate of 6957,
which includes a very nice and small test case by Volker Reichelt:

 void foo()
 {
    char                      (*c)[2];
    float                     *fp;
    int                       i, j;
    union { float r; int i; } u;

    for ( i=0; i<j; ++i )
       if (j==1)
       {
          ++fp;
          u.r = *fp; c[i][0] = u.i>0 ? 0 : (u.r+=1, u.i);
          u.r = *fp; c[i][0] = u.i>0 ? 0 : (u.r+=1, u.i);
       }
 }

 compiled with gcc -march=athlon -O2 -fforce-addr -c

Would you mind adding this to the testsuite?

Thanks,
Gerald
--

-- 
(Continue reading)

Jason Merrill | 1 Jul 11:29 2002
Picon

Re: PATCH: optimization/7145: g++ -O with structure initializer & returnvalue optimization generates bad code

>>>>> "H" == H J Lu <hjl <at> lucon.org> writes:

> I don't think we can do named return value optimization if the return
> value is initialized by "{...}". Here is a patch. Tested on gcc 3.1.1
> on Linux/x86 and Linux/alpha.

I think we can.  Please send a testcase.

Jason

Jason Merrill | 1 Jul 11:29 2002
Picon

Re: PATCH: optimization/7145: g++ -O with structure initializer & returnvalue optimization generates bad code

Oops, I see this is in response to a PR.  I'll get the testcase from there.

Jason

Gerald Pfeifer | 1 Jul 11:50 2002
Picon

Re: Patch to include traditional C support removal in beginner project docs

On Sun, 30 Jun 2002, Kaveh R. Ghazi wrote:
> Index: beginner.html
> ===================================================================
> RCS file: /cvs/gcc/wwwdocs/htdocs/projects/beginner.html,v
> retrieving revision 1.20
> diff -u -p -r1.20 beginner.html
> --- beginner.html	24 Jan 2002 17:19:02 -0000	1.20
> +++ beginner.html	30 Jun 2002 13:31:34 -0000
>  <at>  <at>  -147,6 +147,22  <at>  <at>  still does generate the warning where it
>  to take those patches too.</p>
>  </li>
>
> +<li>Eliminate support for compiling with traditional C compilers in
> +the language frontend subdirectories.
> +
> +<p>See: <a href="http://gcc.gnu.org/ml/gcc/2002-06/msg00871.html">this
> +announcement</a> and the discussion following it which clarifies the
> +guidelines.  In addition to the cleanups <a
> +href="http://gcc.gnu.org/ml/gcc/2002-06/msg00959.html">listed
> +here</a>

Would you mind making this <a href="...">cleanups listed here</a>...

> , one can also consider removing unnecessary casts, such as
> +those on the return value of xmalloc, alloca and other memory
> +allocation routines, casts on the arguments passed e.g. to the mem*
> +functions, casts on 0 e.g. used in assignment, initialization or
> +comparison, etc.  Note casts on values passed to stdarg functions or
> +used in ~ mask operations may still be necessary since they ensure
> +type width.</p>
(Continue reading)

Neil Booth | 1 Jul 12:52 2002
Picon

Re: Finish i386/ CPP builtin conversion.

David O'Brien wrote:-

> > > I'd[*] really prefer that you didn't #undef this.  No one before this
> > > file should be defining EXTRA_OS_CPP_BUILTINS.
> > 
> > But someone does.
> 
> :-((  Why in the world is somone else setting OS specific flags.. don't
> answer that, I'll deal with it in my next clean up pass.

Because they're sharing code, with the EXTRA macro holding OS-specific
stuff.  It was part of my patch.  I don't think there's a better way to
do it, unless you want code duplication.

Neil.


Gmane