Hi Jon,

On Fri, Jul 5, 2013 at 12:35 AM, [email protected] <[email protected]> wrote:
> On Tue, Apr 16, 2013 at 6:32 PM, Drasko DRASKOVIC
> <[email protected]> wrote:
>> Hi Jon,
>> both RT3050 and RT5350 have MIPS 24Kc core. I have successfully used
>> OpenOCD with similar (not exactly these) chips, having the same MIPS
>> core.
>>
>> You should use MIPS 4K as a target.
>
> I've bricked one of my AsiaRF modules and I'm trying to recover it.
>
> For some reason reads over my JTAG are painfully slow. Writes as quick.
>
> Takes 20 seconds....
> mdw 0x80000198 200
>
> Instant...
> mww 0x80000198 200 0
>
> Because reads are terribly broken it take 15 minutes to load uboot
> into RAM since OpenOCD verifies the write.
>
> Any idea why reads are so slow?

Checkout FASTWRITE property of EJTAG, i.e. look in
mips_m4k_read_memory() and see if mips32_pracc_read_mem() gets called
and debug around that.

Timestamp in wait_for_pracc_rw() to see if you are waiting too long
for READY bit of EJTAG to gets set...

You cantake a look at the doc
http://repo.or.cz/w/openocd.git/blob/00d6925b41690df17f81ab3da2f37829d7095e19:/doc/manual/target/mips.txt
to understand a bit how it works...


> #jtag_speed
> adapter_khz 2000

Try playing with adapter_khz. From my point of view this can be fast
for a start, especially if JTAG is unstable.

Try lowering it to 200 for example, suprisingly you might get better
results... Then try to augment it to see how fast you can go.


>>
>> At the time I used it extensively I tried to demystify a code that was
>> there and the code I contributed, so I crafted a doc that you can
>> refer to if you are stuck : http://openocd.zylin.com/#/c/904/ (I am
>> sending you a link to the patchset, although it should be merged in
>> the main tree. However, I do not see it there :
>> http://openocd.git.sourceforge.net/git/gitweb.cgi?p=openocd/openocd;a=tree;f=doc/manual;h=8a9121ca13b9286d258a420f4e301b84bad287b7;hb=08ddb19fd3a708d21057c88e8b86215e04c781ec)
>>
>> Anyway, I can confirm that I was capable to load and JTAG debug both
>> low-level code and Linux, having FASTWRITE and I even added
>> coprocessor manipulations from command line (I added this for correct
>> cache handling, so now both soft and hard breaks should work fine).
>>
>> I am very interested to see how this Chinese TOPLINK story develops.
>> Please keep us informed.
>
> I have the Toplink boards, but I am not so convinced that Toplink
> wants to be in the module business.
>
> I also have the AsiaRF AWM002. AsiaRF is much more interested is
> getting OpenWRT going on their stuff.
> http://www.asiarf.com/Smallest-Tiny-Ralink-802-11n-Wireless-AP-Router-Module-Board-AWM002-product-view-375.html
>
> With 32MB/8MB the modules are around $8.50 Q1000.

This is very nice price, but I guess that is 8MB, and it will be more for 32MB.

> AsiaRF will sell Q1
> for $15 and a dev carrier board for $30. Dev board exposes JTAG, both
> UARTS, two Ethernet and USB.

Fot this price I would actually rather look at Carambola 2, which can
be bought in 1 unity for 15 eur, around 30 eur for a devboard. Plus
you get Atheros and working SDK.

BR,
Drasko
_______________________________________________
openwrt-devel mailing list
[email protected]
https://lists.openwrt.org/mailman/listinfo/openwrt-devel

Reply via email to