On Wed, Aug 10, 2011 at 12:01 AM, Spencer Oliver <s...@spen-soft.co.uk>wrote:
> On 09/08/2011 22:15, Øyvind Harboe wrote:
>
>> Any objections?
>>
>>
> I would like to give this a test-run tomorrow.
>
> One observation - other targets that do not yet support the new functions
> will output a LOG_ERROR to the user. Maybe this should be a LOG_DEBUG as the
> user will have no idea what it means.
>
> Cheers
> Spen
>
+ if (!target->type->start_algorithm) {
+ LOG_ERROR("Target type '%s' does not support %s",
+ target_type_name(target), __func__);
Are you referring to this? That's in line with the corresponding check in
target_run_algorithm(), I think. If some code is executing
target_xxx_algorithm() when it isn't supported by the target in question,
that's clearly an error, right?
Well, I can see a scenario where for example an architecture specific
checksum_memory function had been (re-)written to make use of
target_start/wait_algorithm, and a target x of that architecture points to
that checksum_memory function as well as the architecture's
start/wait_algorithm functions. If another target of the same architecture
uses the same checksum_memory function but hasn't yet populated the
start/wait_algorithm pointers, then the user would get an ERROR when
checksumming on that target. On the other hand, maybe that's a good way to
make someone update the target's pointers. LOG_ERROR or LOG_DEBUG, I don't
care much. It shouldn't happen in the current tree anyway, unless I've
overlooked something.
/Andreas
_______________________________________________
Openocd-development mailing list
Openocd-development@lists.berlios.de
https://lists.berlios.de/mailman/listinfo/openocd-development
