[RFC] m32r-tdep.c: Fix sign extension problem during prologue analysis

[RFC] m32r-tdep.c: Fix sign extension problem during prologue analysis

[Kevin Buettner]

I recently learned that there were far too many testsuite failures for
the m32r-elf target.  I found that backtraces (among other things)
were badly broken and traced the problem to a lack of sign extension
for a particular instruction pattern during prologue analysis.

Comments?  (Is there a better way to do the sign extension?)

Kevin

	* m32r-tdep.c (decode_prologue): Sign extend offset for
	"addi sp, xx" case.
	(m32r_frame_unwind_cache): Likewise.

Index: m32r-tdep.c
===================================================================
RCS file: /cvs/src/src/gdb/m32r-tdep.c,v
retrieving revision 1.48
diff -u -p -r1.48 m32r-tdep.c
--- m32r-tdep.c	23 Aug 2007 18:08:35 -0000	1.48
+++ m32r-tdep.c	10 Oct 2007 23:35:12 -0000
@@ -349,7 +349,7 @@ decode_prologue (CORE_ADDR start_pc, COR
       if ((insn >> 8) == 0x4f)	/* addi sp, xx */
 	/* add 8 bit sign-extended offset */
 	{
-	  int stack_adjust = (gdb_byte) (insn & 0xff);
+	  int stack_adjust = (signed char) (insn & 0xff);
 
 	  /* there are probably two of these stack adjustments:
 	     1) A negative one in the prologue, and
@@ -578,7 +578,7 @@ m32r_frame_unwind_cache (struct frame_in
       else if ((op & 0xff00) == 0x4f00)
 	{
 	  /* addi sp, xx */
-	  int n = (gdb_byte) (op & 0xff);
+	  int n = (signed char) (op & 0xff);
 	  info->sp_offset += n;
 	}
       else if (op == 0x1d8f)

Re: [RFC] m32r-tdep.c: Fix sign extension problem during prologue analysis

[Daniel Jacobowitz]

On Wed, Oct 10, 2007 at 05:02:52PM -0700, Kevin Buettner wrote:
> I recently learned that there were far too many testsuite failures for
> the m32r-elf target.  I found that backtraces (among other things)
> were badly broken and traced the problem to a lack of sign extension
> for a particular instruction pattern during prologue analysis.
> 
> Comments?  (Is there a better way to do the sign extension?)

That seems right to me.  You could be more explicit about it, but
there's no point if you assume that char is eight bits anyway.

-- 
Daniel Jacobowitz
CodeSourcery

Re: [RFC] m32r-tdep.c: Fix sign extension problem during prologue analysis

[Kevin Buettner]

On Wed, 10 Oct 2007 21:29:50 -0400
Daniel Jacobowitz <drow@false.org> wrote:

> > Comments?  (Is there a better way to do the sign extension?)
> 
> That seems right to me.  You could be more explicit about it, but
> there's no point if you assume that char is eight bits anyway.

Thanks for looking at this patch.

I've committed it now.

Kevin