VAX PC RELATIVE JMP: gas and gdb perform incorrect sign extension

[John David Anglin]

> On Mon, 21 Aug 2000, John David Anglin wrote:
> 
> > I am looking at the binutils code.  Suggestions on where to look are welcome.
> 
> I'm not particularly familiar with the vax port, but my guess would be a
> problem in tc-vax.c:md_convert_frag, or possibly md_estimate_size_before_relax

I think the problem is in tc-vax.c:md_convert_frag.  It looks as if somebody
changed the addressing mode from VAX_ABSOLUTE_MODE to VAX_PC_RELATIVE_MODE
and forgot to change the address calculation (compare to preceding WORD
mode).

I would guess VAX_ABSOLUTE_MODE might be more efficient thans
VAX_PC_RELATIVE_MODE since there is no addition involved in the address
calculation.

Dave
-- 
J. David Anglin                                  dave.anglin@nrc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6605)