Correct m68k float flag logic

Correct m68k float flag logic

[Nathan Sidwell]

[-- Attachment #1: Type: text/plain, Size: 671 bytes --]

I've installed this patch (as obvious), for a problem with the m68k's float 
format logic.  The -mfloat flag is used for both m68k and cf floats, which to 
target is determined by the target cpu.  That check should be done _before_ 
testing whether we've been given flags for both m68k & cf architectures.

We've been using this on the binutils-csl-2_17-branch for some time (and we're 
just going through making sure things get back to mainline).

Ben, do you want this on the binutils-2_17-branch too?

nathan
-- 
Nathan Sidwell    ::   http://www.codesourcery.com   ::         CodeSourcery
nathan@codesourcery.com    ::     http://www.planetfall.pwp.blueyonder.co.uk


[-- Attachment #2: all.diff --]
[-- Type: text/plain, Size: 1451 bytes --]

2006-05-19  Nathan Sidwell  <nathan@codesourcery.com>

	* gas/config/tc-m68k.c (m68k_init_arch): Move checking of
	cfloat/m68881 to correct architecture before using it.

Index: config/tc-m68k.c
===================================================================
RCS file: /cvs/src/src/gas/config/tc-m68k.c,v
retrieving revision 1.75
diff -c -3 -p -r1.75 tc-m68k.c
*** config/tc-m68k.c	28 Mar 2006 07:21:49 -0000	1.75
--- config/tc-m68k.c	19 May 2006 10:12:58 -0000
*************** m68k_init_arch (void)
*** 7233,7238 ****
--- 7233,7247 ----
    
    current_architecture &= ~not_current_architecture;
  
+   if ((current_architecture & (cfloat | m68881)) == (cfloat | m68881))
+     {
+       /* Determine which float is really meant.  */
+       if (current_architecture & (m68k_mask & ~m68881))
+ 	current_architecture ^= cfloat;
+       else
+ 	current_architecture ^= m68881;
+     }
+ 
    if (selected_cpu)
      {
        control_regs = selected_cpu->control_regs;
*************** m68k_init_arch (void)
*** 7244,7258 ****
  	}
      }
  
-   if ((current_architecture & (cfloat | m68881)) == (cfloat | m68881))
-     {
-       /* Determine which float is really meant.  */
-       if (current_architecture & (m68k_mask & ~m68881))
- 	current_architecture ^= cfloat;
-       else
- 	current_architecture ^= m68881;
-     }
- 
    if ((current_architecture & m68k_mask)
        && (current_architecture & ~m68k_mask))
      {
--- 7253,7258 ----

Re: Correct m68k float flag logic

[Ben Elliston]

[-- Attachment #1: Type: text/plain, Size: 615 bytes --]

> I've installed this patch (as obvious), for a problem with the m68k's float 
> format logic.  The -mfloat flag is used for both m68k and cf floats, which 
> to target is determined by the target cpu.  That check should be done 
> _before_ testing whether we've been given flags for both m68k & cf 
> architectures.
> 
> We've been using this on the binutils-csl-2_17-branch for some time (and 
> we're just going through making sure things get back to mainline).
> 
> Ben, do you want this on the binutils-2_17-branch too?

It's okay with me if it's okay with the release manager.  :-)

Cheers, Ben

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: Correct m68k float flag logic

[Daniel Jacobowitz]

On Thu, May 25, 2006 at 03:55:35PM +1000, Ben Elliston wrote:
> > I've installed this patch (as obvious), for a problem with the m68k's float 
> > format logic.  The -mfloat flag is used for both m68k and cf floats, which 
> > to target is determined by the target cpu.  That check should be done 
> > _before_ testing whether we've been given flags for both m68k & cf 
> > architectures.
> > 
> > We've been using this on the binutils-csl-2_17-branch for some time (and 
> > we're just going through making sure things get back to mainline).
> > 
> > Ben, do you want this on the binutils-2_17-branch too?
> 
> It's okay with me if it's okay with the release manager.  :-)

Go ahead.

-- 
Daniel Jacobowitz
CodeSourcery