On Fri, Feb 24, 2012 at 3:54 PM, Spencer Oliver <[email protected]> wrote:
> On 14 February 2012 01:07,  <[email protected]> wrote:
>> This is an automated email from Gerrit.
>>
>> Andreas Fritiofson ([email protected]) just uploaded a new patch 
>> set to Gerrit, which you can find at http://openocd.zylin.com/454
>>
>> -- gerrit
>>
>
> Thought i would reply to the list as this patch series is non-trivial,
> so as many eyes as possible is good.
>
> This all seems a much cleaner implementation than we currently have,
> and is working well in my tests.

Great!

> I saw a negligible speed increase (0.5KiB/s), not sure if this is expected.

Depends on what you tested. Flash programming (with async support)
shouldn't benefit much. Verify and dump is *much* faster here, as well
as raw throughput to RAM. At least on CM3. General debugging should be
snappier too, but it's hard to quantify. I think I've posted these
figures before, which is for STM32F1xx at 64MHz with 3MHz TCK (or was
it 6?):

Flash write 128k, 29.932 KiB/s -> 32.835 KiB/s
Verify 128k, 97.452 KiB/s -> 221.608 KiB/s
Memory dump 128k, 20.645 KiB/s -> 73.848 KiB/s
RAM download 20k, 76.731 KiB/s -> 100.391 KiB/s
Total elapsed time incl. startup, 13.588s -> 7.128s

> Andreas, what is your plan with regards to merging/replacing the
> current ftdi implementation?

First I need a way to specify layouts in Tcl, I really don't wan't
that mess in the driver. I'm going to need some help with that to make
it simple and flexible enough. I've been thinking of having commands
to specify set/clear bitmasks for each operation (assert_srst,
deassert_srst, blink and so on). That should be generic enough, but
it's a huge amount of obscure values that needs to be specified and a
lot of ugly command handlers in C.

The trouble is how to represent the necessary information cleanly in
the Tcl config file and how to get that information into C (I have
very little knowledge of the Jim-to-C binding). Ideas, anyone?

After that it's a matter of writing config files for all the
interfaces that are currently supported in the ft2232 driver. I guess
the new driver will live side-by-side with the old for some time to
gain some mileage. Hopefully the ftdi driver will be the default
choice after a while, at which point the ft2232 driver could be
deprecated and finally removed. Some special interfaces may not be
portable to the new driver, i.e. signalyzer, so the ft2232 code may
need to stay for those, or they can be broken out to separate drivers.

> Has anybody else tested this patch series?

Please do! I realize that the limitation of a single working layout
makes it hard for many people, but if you have a jtagkey-compatible
dongle, give it a spin. Most appreciated would be some non-ARM,
non-Linux testing.

Regards,
Andreas

------------------------------------------------------------------------------
Virtualization & Cloud Management Using Capacity Planning
Cloud computing makes use of virtualization - but cloud computing 
also focuses on allowing computing to be delivered as a service.
http://www.accelacomm.com/jaw/sfnl/114/51521223/
_______________________________________________
OpenOCD-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/openocd-devel

Reply via email to