Re: partial patch 904_hjl_hppa_whitespace.diff

["H. J. Lu" <hjl at lucon dot org>]

Should my PA space patch be applied?

H.J.
--
On Sun, May 25, 2003 at 11:49:33AM -0400, Jack Howarth wrote:
> HJ,
>    This doesn't impact my arch but in using the debian packaging for
> binutils they had a patch called 904_hjl_hppa_whitespace.diff of which
> only the section...
> 
> --- binutils/gas/testsuite/gas/hppa/parse/parse.exp.space     
> 2002-09-08 18:54:43.000000000 -0700
> +++ binutils/gas/testsuite/gas/hppa/parse/parse.exp    2003-05-18
> 22:24:27.000000000 -0700
> @@ -222,5 +222,8 @@ if [istarget hppa*-*-*] then {
>      # Check for bogus registers in single precision fmpyadd/fmpysub
>      # instructions
>      gas_test_error "badfmpyadd.s" "" "Check for error on bad fmpyadd
> insn"
> +
> +    # Make sure we grok spaces in directives.
> +    gas_test "space.s" "" "" "Test acceptance of spaces in directives"
>  }
> 
> 
> is currently applied to binutils cvs. Any idea why the rest of this
> patch, attched to this message, wasn't applied to the cvs as well?
> Should Debian just drop the rest of the patch for hppa?
>                                 Jack

> From: "H. J. Lu" <hjl@lucon.org>
> To: Matthias Klose <doko@cs.tu-berlin.de>
> Cc: bug-binutils@gnu.org, binutils@sources.redhat.com
> Subject: Re: [hppa-linux] gcc-3.3 fails to build using binutils-2.14.90.0.1
> Date: Mon, 19 May 2003 00:08:42 -0700
> 
> 
> --XsQoSWH+UP9D9v3l
> Content-Type: text/plain; charset=us-ascii
> Content-Disposition: inline
> 
> On Sun, May 18, 2003 at 11:47:33PM +0200, Matthias Klose wrote:
> Content-Description: message body text
> > H. J. Lu writes:
> > > On Sun, May 18, 2003 at 07:06:06PM +0200, Matthias Klose wrote:
> > > > binutils-2.13.90.0.18 works ok.
> > > > 
> > > > See http://buildd.debian.org/fetch.php?&pkg=gcc-3.3&ver=1%3A3.3ds9-2&arch=hppa&stamp=1053264270&file=log&as=raw:
> > > > 
> > > > ./xgcc -B./ -B/usr/hppa-linux/bin/ -isystem /usr/hppa-linux/include -isystem /usr/hppa-linux/sys-include -O2  -DIN_GCC    -W -Wall -Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes -isystem ./include  -fPIC -DELF=1 -DLINUX=1 -g -DHAVE_GTHR_DEFAULT -DIN_LIBGCC2 -D__GCC_FLOAT_NOT_NEEDED  -I. -I. -I../../src/gcc -I../../src/gcc/. -I../../src/gcc/config -I../../src/gcc/../include -DL_mulI -xassembler-with-cpp -c ../../src/gcc/config/pa/milli64.S -o libgcc/./_mulI.o
> > > > ../../src/gcc/config/pa/milli64.S: Assembler messages:
> > > > ../../src/gcc/config/pa/milli64.S:1779: Error: Undefined .EXPORT/.IMPORT argument (ignored): 
> > > > ../../src/gcc/config/pa/milli64.S:1779: Error: Undefined .EXPORT/.IMPORT argument (ignored): millicode
> > > > make[5]: *** [libgcc/./_mulI.o] Error 1
> > > 
> > > Please provide a small assembly testcase. It shouldn't be too hard
> > > to reproduce.
> > 
> > attached, although it's only the preprocessed milli64 source.
> > 
> 
> Could PA maintainer please verify this patch?
> 
> BTW, there may be more PA assembler breakages. Please provide a small
> assembly testcase when reporting assembler bugs.
> 
> 
> H.J.
> 
> --XsQoSWH+UP9D9v3l
> Content-Type: text/plain; charset=us-ascii
> Content-Disposition: attachment; filename="gas-pa-space.patch"
> 
> gas/
> 
> 2003-05-18  H.J. Lu <hongjiu.lu@intel.com>
> 
> 	* config/tc-hppa.c (pa_callinfo): Skip whitespace after
> 	incrementing input_line_pointer.
> 	(pa_export): Likewise.
> 	(pa_import): Likewise.
> 	(pa_param): Likewise.
> 
> gas/testsuite/
> 
> 2003-05-18  H.J. Lu <hongjiu.lu@intel.com>
> 
> 	* hppa/parse/parse.exp: Add "space.s".
> 
> 	* gas/hppa/parse/space.s: New file to test spaces.
> 
> --- binutils/gas/config/tc-hppa.c.space	2003-04-28 07:17:58.000000000 -0700
> +++ binutils/gas/config/tc-hppa.c	2003-05-18 22:14:28.000000000 -0700
> @@ -6200,6 +6200,7 @@ pa_callinfo (unused)
>  	  p = input_line_pointer;
>  	  *p = c;
>  	  input_line_pointer++;
> +	  SKIP_WHITESPACE ();
>  	  temp = get_absolute_expression ();
>  	  if ((temp & 0x3) != 0)
>  	    {
> @@ -6217,6 +6218,7 @@ pa_callinfo (unused)
>  	  p = input_line_pointer;
>  	  *p = c;
>  	  input_line_pointer++;
> +	  SKIP_WHITESPACE ();
>  	  temp = get_absolute_expression ();
>  	  /* The HP assembler accepts 19 as the high bound for ENTRY_GR
>  	     even though %r19 is caller saved.  I think this is a bug in
> @@ -6230,6 +6232,7 @@ pa_callinfo (unused)
>  	  p = input_line_pointer;
>  	  *p = c;
>  	  input_line_pointer++;
> +	  SKIP_WHITESPACE ();
>  	  temp = get_absolute_expression ();
>  	  /* Similarly the HP assembler takes 31 as the high bound even
>  	     though %fr21 is the last callee saved floating point register.  */
> @@ -6242,6 +6245,7 @@ pa_callinfo (unused)
>  	  p = input_line_pointer;
>  	  *p = c;
>  	  input_line_pointer++;
> +	  SKIP_WHITESPACE ();
>  	  temp = get_absolute_expression ();
>  	  if (temp != 3)
>  	    as_bad (_("Value for ENTRY_SR must be 3\n"));
> @@ -6304,7 +6308,10 @@ pa_callinfo (unused)
>  	  *input_line_pointer = c;
>  	}
>        if (!is_end_of_statement ())
> -	input_line_pointer++;
> +	{
> +	  input_line_pointer++;
> +	  SKIP_WHITESPACE ();
> +	}
>      }
>  
>    demand_empty_rest_of_line ();
> @@ -6633,6 +6640,7 @@ pa_export (unused)
>        if (!is_end_of_statement ())
>  	{
>  	  input_line_pointer++;
> +	  SKIP_WHITESPACE ();
>  	  pa_type_args (symbol, 1);
>  	}
>      }
> @@ -6825,6 +6833,7 @@ pa_import (unused)
>        if (!is_end_of_statement ())
>  	{
>  	  input_line_pointer++;
> +	  SKIP_WHITESPACE ();
>  	  pa_type_args (symbol, 0);
>  	}
>        else
> @@ -6981,6 +6990,7 @@ pa_param (unused)
>        if (!is_end_of_statement ())
>  	{
>  	  input_line_pointer++;
> +	  SKIP_WHITESPACE ();
>  	  pa_type_args (symbol, 0);
>  	}
>      }
> --- binutils/gas/testsuite/gas/hppa/parse/space.s.space	2003-05-18 22:24:53.000000000 -0700
> +++ binutils/gas/testsuite/gas/hppa/parse/space.s	2003-05-18 22:23:12.000000000 -0700
> @@ -0,0 +1,24 @@
> +	.code
> +	.align 4
> +	.export $$mulI, millicode
> +	.proc
> +	.callinfo millicode
> +$$mulI:
> +	.procend
> +
> +	.code
> +
> +	.align 4
> +	.PARAM foo, RTNVAL=GR
> +foo:
> +	.PROC
> +	.CALLINFO FRAME=128, NO_CALLS, ENTRY_GR=3,  ENTRY_FR=12
> +	.ENTRY
> +	bv,n %r0(%r2)
> +	.EXIT
> +	.PROCEND
> +
> +	.align 4
> +	.import yabba, code
> +
> +  	ble	R%yabba(%sr4, 	%r0)
>