This is the mail archive of the
gdb@sources.redhat.com
mailing list for the GDB project.
Re: support for different sized FPU registers
- From: Andrew Cagney <ac131313 at cygnus dot com>
- To: Peter Barada <pbarada at mail dot wm dot sps dot mot dot com>
- Cc: gdb at sources dot redhat dot com
- Date: Sat, 01 Jun 2002 18:33:18 -0400
- Subject: Re: support for different sized FPU registers
- References: <200205292217.g4TMH7300610@hyper.wm.sps.mot.com>
> I'm trying to add GDB remote support for the ColdFire v4e, and I'm
> running into problems with the FPU register definitions.
>
> On the ColdFire v4e, the FPU registers are 64 bit, but the gdb is
> expecting 96 bit register values from a 68881. As a result I'm
> getting some 'Remote reply is too short' messages from gdb when the
> stub responds to a 'g' packet and on the host the register contents
> are all wrong(from 'info reg').
>
> What's the best way of telling gdb(for the coldfire portion of a
> m68k-elf configured build) that the FPU registers are only 64 bits
> instead of 96)? I still want this configuration to support 040s and
> 060s which have the 96 bit FPU registers as well as a ColdFire's 64
> bit FPU registers...
The most immediate thing that is needed is for the m68k family to be
up-dated to the new multi-arch framework. Once that is done, handling
both a coldfire and an m68k is fairly mechanical. Check other targets
such as PPC, xstormy16, mn10300 et.al. for examples.
Do you have an FSF assignment in place?
Andrew