Re: [rfc/rft] [3/4] Untangle register_addr - v2 - mips-linux

[Daniel Jacobowitz <drow at false dot org>]

On Mon, Apr 23, 2007 at 05:45:32PM +0200, Ulrich Weigand wrote:
> Hello,
> 
> this is the updated mips-linux patch.  Main difference is moving
> mips_linux_cannot_fetch_register and mips_linux_cannot_store_register
> to mips-linux-tdep.c and installing them as gdbarch callbacks.
> 
> Tested by building --target=mips-linux and --host=mips-linux.

Well, no, I don't think this will do.  Sorry I didn't get to look at
this before you redid it.  The cannot_store_register callback refused
stores to badvaddr and cause.  If you're using a mips-linux native
debugger, or talking to gdbserver, that's fine.  But if you're using a
mips-linux debugger and talking to kgdb over a JTAG unit, it's less
fine.  You should be able to write to cause if your target_ops permits.

That's probably typical of user-readable supervisor mode registers.
I bet there are plenty more targets with this situation where GDB just
doesn't support the register in question.

-- 
Daniel Jacobowitz
CodeSourcery