> Øyvind Harboe <oyvind.har...@zylin.com>
> Sent by: openocd-development-boun...@lists.berlios.de
>
> On Tue, Aug 3, 2010 at 8:00 PM, Peter Stuge <pe...@stuge.se> wrote:
> > Øyvind Harboe wrote:
> >> I'm not crazy about adding another code path for an aging part...
I don't think STR912 is that "aging" though, its from 2007 and got
ARM966E-s core.
The latest 2Mbyte versions of the STR9 chip is just a year old.
I agree that some parts, like ARM7TDMI-cores are "aging", but not in this
case.
I know Cortex is more hip now, but in some applications the ARM9E still is
a valid choice.
We estimate to have several years ahead with high volume production with
this part in our existing products.
> >>
> >> Thoughts?
> >
> > Why not make improvements as long as it's still supported?
>
> We have something that is slow and tested today.
>
> Switching to something potentially faster and untested is not
> an improvement.
>
> I'm not saying that the code is untested, I'm saying that I'd like
> to *know* that it works before I commit it.
I think we will never 100% know that, like for all other patches.
It's a very un-complicated patch though, you just do one chip-erase
instead of iterating sector-erase.
But I rest my case, we can use the patch here internally when we compile
openOCD.
>
> This particular patch has been tricky in the past w.r.t. what the
> correct timeout handling is across adapters...
Yes, there was a simple timeout issue previously, but unfortunately the
code was misinterpreted and the patch was rejected.
I think timeout-handling is a generic potential issue in openOCD, on alot
of places just "1000 ms" is entered as a generic timeout value, without
any real configuration of what adapter nor JTAG speed is used.
Thanks and Best Regards,
Fredrik Hederstierna
Securitas Direct AB
Malmoe Sweden
_______________________________________________
Openocd-development mailing list
Openocd-development@lists.berlios.de
https://lists.berlios.de/mailman/listinfo/openocd-development
