On Wed, Oct 15, 2008 at 1:44 AM, Duane Ellis <[EMAIL PROTECTED]> wrote:
> Øyvind Harboe wrote:
>>
>> I *know* it is a bit late to suggest this now, but does anyone have
>> any argument against making target_write_xxx() and
>> target->type->write_xxx()
>> return void?
>>
>> jtag_execute_queue() will return an error if the above failed....
>>
>> That's part of the whole queue API design...
>>
>>
>
> What you propose seems wrong - in at least 2 ways:
>
> --------------------------------------
>
> 1) a script that will read lots and lots of data bytes from the target.
>
> Example - my "show register" functions I wrote for AT91SAM parts.
>
> The boil down to 'target->type->read_memory()'
>
> (A) that script function should get an error.
> (B) if it returns VOID - how would it get or display that error?

I was only airing the idea to make *write* return void.

>
> ----
>
> 2) Same questions for GDB   If I cannot write/read memory - it needs to
> know.
>
>  It uses "target_write_buffer()'
>
> Granted, GDB does not always do a good job - but - the intent is this: If
> you have a 'wacky pointer' reading from that memory location should cause a
> read fault of sometype and that read fault should be reported back up to
> GDB.
>
> ------
>
> Or maybe you have a way around this?

I wanted to air this thought, but I believe that we should stick to
having write's return error code.


-- 
Øyvind Harboe
http://www.zylin.com/zy1000.html
ARM7 ARM9 XScale Cortex
JTAG debugger and flash programmer

Free eCos workshop in Oslo October 21!
http://www.zylin.com/workshop.html
_______________________________________________
Openocd-development mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/openocd-development

Reply via email to