Re: [RFC][PATCH v2] Consolidate gdbserver global variables

[Pedro Alves <palves@redhat.com>]

On 05/07/2018 08:16 PM, Frank Ch. Eigler wrote:
> 
> palves wrote:
> 
>> [...]
>> Some comments more.  I still have some design reservations.
>> It seems to me that simply making some of the globals be
>> per client along won't work correctly.  Particularly, the
>> cases where the globals are used in the backends.
>> [...]
> 

> I don't want to speak for Stan here, 

;-)

> but I suspect all this is about is
> the breaking up of the multi-client work along the patchset path your
> requested.  Yup, indeed backends have to do the logical 'union' of the
> work that all the clients request, and each client interface will have
> to filter the notifications correspondingly to undo the 'union'.  

Note that in some cases it's not just even filtering.  Globals
can control modes of operation, like the cs->non_stop case I pointed
out in a previous review.

> 
> But all that is in a separate patch; this part seems to simply
> reorganize the globals in preparation for that, as you asked.

Right, but it doesn't hurt to think about the plan for the
issues I pointed out a bit, since it may avoid churn in a
direction that we may not want.  In the original version of the patch,
when macros were used, these references to client_state fields
from the backends were pretty much invisible (because of the macros),
so it would be unsurprising to me if the full multi-client prototype
that is based on the macros version does not consider these issues.

I'll go look at the updated patch.

Thanks,
Pedro Alves