This is the mail archive of the
gdb-patches@sources.redhat.com
mailing list for the GDB project.
Re: [PATCH/RFA] arm-netbsdelf cross-debugging fixes
- From: Richard Earnshaw <rearnsha at cambridge dot arm dot com>
- To: Jason R Thorpe <thorpej at wasabisystems dot com>
- Cc: gdb-patches at sources dot redhat dot com
- Date: Tue, 03 Sep 2002 20:23:01 +0100
- Subject: Re: [PATCH/RFA] arm-netbsdelf cross-debugging fixes
- Organization: ARM Ltd.
- Reply-to: Richard dot Earnshaw at arm dot com
> This fixes a bunch of cross-debugging issues for the arm-netbsdelf
> target, and makes it more like the other NetBSD targets I've worked
> on recently.
>
> Unfortunately, the change disables "gdb_multi_arch=yes" for the
> arm-*-netbsd* entry in configure.tgt. The reason for this is that
> the tm file needs to include "solib.h". As far as I can tell, no
> platforms with shared libraries can be fully multi-arch until all
> the solib stuff is multi-arch'd.
>
Hmm, evil thought. If I said 'no' then perhaps you'd have to go
multi-arch the solib code....
No, couldn't be that mean :-)
> I need approval for a change to the shared ARM target code. I needed
> to add a function to deconstruct/reconstruct the R15 register on the
> 26-bit ARM CPUs. The new functions:
>
> * Split a 26-bit R15 into PC and a 32-bit-format PSR.
>
> * Take a PC and a 32-bit-format PSR and combine them back
> into a 26-bit R15.
This is going to conflict horribly with my WIP for using cooked registers.
Is there any chance it can be separated out, or made less invasive?
>
> While there, I also fixed a problem with arm_addr_bits_remove -- No
> 26-bit systems can run in Thumb mode, and so doing an arm_pc_is_thumb
> on them is unnecessary (and could return incorrect results if debugging
> code which runs in FIQ mode, since (pc & 1) == FIQ mode on those CPUs).
>
Yep, that would be a good move, but it would be easier to say yes if it
were a separate patch.
R.