On Wed, Jul 8, 2009 at 12:54 AM, David Brownell<[email protected]> wrote:
> On Tuesday 07 July 2009, Ųyvind Harboe wrote:
>> The problem is that GDB will assume that the target
>> is in the halted state and query for registers.
>>
>> I'd like to see a model where OpenOCD allows a
>> connect and enters a state where it returns dummy
>> values for register queries(this happens today to
>> an extent). A faux halted state.
>
> Wouldn't a better fix be for GDB to actually *CHECK* the
> target state?

No.

Also we can't change GDB. We've tried pushing fixes
upstream, to no avail in the past. Perhaps some day....

The problem is that GDB command line model is kinda busted,
the command line prompt is unavailable *while* the target
is running. This is why it's not that uncommon to have a telnet
prompt running next to GDB...

(I think there are more esoteric communication protocols
that are better able to handle these concepts available in GDB,
but I don't know that they are for human  beings)

Even if we changed GDB, that wouldn't change all the
GUI's out there.

What I'm suggesting is that we need a few choice white
lies and we can get GDB+GUI's out there working just
like we need/want to.

-- 
Øyvind Harboe
Embedded software and hardware consulting services
http://www.zylin.com
_______________________________________________
Openocd-development mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/openocd-development

Reply via email to