cgen-cpu/tracedis.c patch

cgen-cpu/tracedis.c patch

[Ben Elliston]

The following patch adds a few more BFD functions to tracedis.c.  They
are needed because a simulator I am targetting already has a
hand-written disassembler which uses these BFD functions.  Okay to
commit?

At what point will be abandon this approach and make SID dependent on
libbfd?

Ben


2002-01-23  Ben Elliston  <bje@redhat.com>

	* tracedis.c (bfd_getb16): New function.
	(bfd_getl16): Likewise.
	(bfd_getb32): Likewise.
	(bfd_getl32): Likewise.

Index: tracedis.c
===================================================================
RCS file: /cvs/cvsfiles/devo/sid/component/cgen-cpu/tracedis.c,v
retrieving revision 1.2
diff -u -c -r1.2 tracedis.c
*** tracedis.c	2001/06/08 07:18:50	1.2
--- tracedis.c	2002/01/22 23:35:06
***************
*** 60,65 ****
--- 60,106 ----
  }
  
  /* Stolen from libbfd.  */
+ 
+ bfd_vma
+ bfd_getb16 (addr)
+      register const bfd_byte *addr;
+ {
+   return (addr[0] << 8) | addr[1];
+ }
+ 
+ bfd_vma
+ bfd_getl16 (addr)
+      register const bfd_byte *addr;
+ {
+   return (addr[1] << 8) | addr[0];
+ }
+ 
+ bfd_vma
+ bfd_getb32 (addr)
+      register const bfd_byte *addr;
+ {
+   unsigned long v;
+ 
+   v = (unsigned long) addr[0] << 24;
+   v |= (unsigned long) addr[1] << 16;
+   v |= (unsigned long) addr[2] << 8;
+   v |= (unsigned long) addr[3];
+   return (bfd_vma) v;
+ }
+ 
+ bfd_vma
+ bfd_getl32 (addr)
+      register const bfd_byte *addr;
+ {
+   unsigned long v;
+ 
+   v = (unsigned long) addr[0];
+   v |= (unsigned long) addr[1] << 8;
+   v |= (unsigned long) addr[2] << 16;
+   v |= (unsigned long) addr[3] << 24;
+   return (bfd_vma) v;
+ }
+ 
  void
  bfd_put_bits (data, addr, bits, big_p)
       bfd_vma data;

re: cgen-cpu/tracedis.c patch

[matthew green]

   The following patch adds a few more BFD functions to tracedis.c.  They
   are needed because a simulator I am targetting already has a
   hand-written disassembler which uses these BFD functions.  Okay to
   commit?

looks fine.
   
   At what point will be abandon this approach and make SID dependent on
   libbfd?


i'd like to.  it would allow --trace-disasm to get at symbols..


.mrg.

Re: cgen-cpu/tracedis.c patch

[Frank Ch. Eigler]

Hi -

On Wed, Jan 23, 2002 at 01:02:41PM +1100, matthew green wrote:
> [...]
>    At what point will be abandon this approach and make SID dependent on
>    libbfd?
> 
> i'd like to.  it would allow --trace-disasm to get at symbols..

Linking with bfd opens up more cans of worms, some of which we have
discovered in earlier experiments.  Besides, there are other ways
to let trace-disasm see symbols: see for example how addr2line is
used by the (currently defunct) visual memory tracer thingie.


- FChE