Re: [RFC] gdb/riscv: Improved register alias name creation

[Andrew Burgess <andrew.burgess@embecosm.com>]

* Tom Tromey <tom@tromey.com> [2020-06-09 14:54:23 -0600]:

> >>>>> "Andrew" == Andrew Burgess <andrew.burgess@embecosm.com> writes:
> 
> Andrew> The patch below has had only minimal testing so far, but I'm sharing
> Andrew> it here so it can be discussed while I also get it tested.
> 
> Andrew> This changes how the 'csr%d' aliases are created, and adds the
> Andrew> dscratch to dscratch0 alias.
> 
> I backed out the reversion here; then I did a new merge; then I applied
> this patch and re-ran our internal test suite.
> 
> It still fails -- though it does change the failures I see.  Now it
> complains about a different register:
> 
> (gdb) call debug.trace (me, "You")
> Could not fetch register "mucounteren"; remote failure reply 'E14'

That is so weird!  Based on the dump you included in the other email I
would have expected that to work.

Please would it be possible for you to gather two logs using "set
remotefilelog ..." as before.  For each log I'd like to see:

  1. Connect to the target,
  2. Print $dscratch,
  3. print $mucounteren

Could you do this with current whatever GDB you were using above, and
then again with whatever patches you need to back out in order to get
it working again.

Thanks,
Andrew