RFA: Patch: Change FRV 'mdrotli' instruction to accept a signed immediate

Dave Brolley brolley@redhat.com
Tue Jun 3 16:04:00 GMT 2003


Hi Nick,

You've begun work at an interesting point in the life of this port. 
Ongoing work to get the frv.cpu and frv.opc files approved for the 
simulator has just (and I mean just a minute ago) resulted in the 
creation of src/cpu/frv.cpu and src/cpu/frv.opc (ask Andrew Cagney if 
you really want to know why these files had to be moved). However, the 
tools have not yet been changes to use the new location for these files.

In the meantime, please commit your changes to both copies please. I 
will be changing the tools to use the new files asap and, once we verify 
that the tools still work, we'll switch over.

Of course, if you would like to make the switch, then please be my guest 
 :-)

Thanks,
Dave

Nick Clifton wrote:

>Hi Dave,
>
>  Any objections to my applying the patch below ?  It changes the
>  mdrotli instruction for the FRV port so that it accepts a signed
>  immediate value instead of an unsigned one.  According to Fujitsu
>  this is the correct behaviour.  (See Issue #23199 for more
>  details).  I could not find an FRV ISA anywhere on the web to
>  confirm this, but I assume that they are correct.
>
>Cheers
>        Nick
>
>cgen/ChangeLog        
>2003-06-03  Nick Clifton  <nickc@redhat.com>
>
>	* cpu/frv.cpu (media-dual-word-rotate-r-r): Use a signed 6-bit
>	immediate value not unsigned.
>
>opcodes/ChangeLog        
>2003-06-03  Nick Clifton  <nickc@redhat.com>
>
>	* frv-desc.c: Regenerate.
>	* frv-opc.c: Regenerate.
>
>
>Index: cgen/cpu/frv.cpu
>===================================================================
>RCS file: /cvs/src/src/cgen/cpu/frv.cpu,v
>retrieving revision 1.1
>diff -c -3 -p -r1.1 frv.cpu
>*** cgen/cpu/frv.cpu	18 Jun 2002 21:12:33 -0000	1.1
>--- cgen/cpu/frv.cpu	3 Jun 2003 14:11:16 -0000
>***************
>*** 6643,6658 ****
>    (dni name
>         (comment)
>         ((UNIT FM0) (MACH fr400) (FR400-MAJOR M-2))
>!        (.str name "$pack $FRinti,$u6,$FRintk")
>!        (+ pack FRintk op FRinti ope u6)
>         (if (orif (register-unaligned FRinti 2)
>  		 (register-unaligned FRintk 2))
>  	   (c-call VOID "@cpu@_media_register_not_aligned")
>  	   (sequence ()
>! 		     (set FRintk (operation FRinti (and u6 #x1f)))
>  		     (set (nextreg h-fr_int FRintk 1)
>  			  (operation (nextreg h-fr_int FRinti 1)
>! 				     (and u6 #x1f)))))
>         ((fr400 (unit u-media-3-quad)))
>    )
>  )
>--- 6643,6658 ----
>    (dni name
>         (comment)
>         ((UNIT FM0) (MACH fr400) (FR400-MAJOR M-2))
>!        (.str name "$pack $FRinti,$s6,$FRintk")
>!        (+ pack FRintk op FRinti ope s6)
>         (if (orif (register-unaligned FRinti 2)
>  		 (register-unaligned FRintk 2))
>  	   (c-call VOID "@cpu@_media_register_not_aligned")
>  	   (sequence ()
>! 		     (set FRintk (operation FRinti (and s6 #x1f)))
>  		     (set (nextreg h-fr_int FRintk 1)
>  			  (operation (nextreg h-fr_int FRinti 1)
>! 				     (and s6 #x1f)))))
>         ((fr400 (unit u-media-3-quad)))
>    )
>  )
>        
>




More information about the Binutils mailing list