This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [rfa] Add iWMMXt DWARF support, and fixed register numbers
- From: Daniel Jacobowitz <drow at false dot org>
- To: gdb-patches at sourceware dot org
- Cc: Richard Earnshaw <Richard dot Earnshaw at arm dot com>
- Date: Mon, 26 Feb 2007 14:19:04 -0500
- Subject: Re: [rfa] Add iWMMXt DWARF support, and fixed register numbers
- References: <20070213162540.GA13938@caradoc.them.org>
On Tue, Feb 13, 2007 at 11:25:40AM -0500, Daniel Jacobowitz wrote:
> This patch includes an assortment of minor improvements for iWMMXt
> support. It relies on a clever trick that Vladimir pointed out, which
> I may have mentioned before: we can assign fixed register numbers for
> all possible target registers, and then return a name of "" and a type
> of builtin_type_int0 for any that are not currently present. This
> makes the register numbering constant between different targets and
> simplifies writing tdep and nat files by quite a lot.
>
> Changes:
> - Recognize the iWMMXt feature by name, and document its contents.
> - Associate the registers with the correct ARM-assigned DWARF
> encodings.
> - Do not record FPA register saves in the prologue if we do not have
> FPA. Similarly, do not try to print out FPA status.
> - Reformat the register names to match those in the Intel manual.
> Gas accepts both sets, but nothing besides GNU tools seems to use
> the lowercase variants.
>
> Tested on arm-linux-gnueabi with iWMMXt. OK?
I've checked this in.
--
Daniel Jacobowitz
CodeSourcery