Linux + u-boot port to MT6235

[Marcin Mielczarczyk]

Hi Steve

On Sat, Oct 30, 2010 at 2:53 PM, Steve Markgraf <steve at steve-m.de> wrote:
>
> Alright, this is how it works on the MT622x as well.
>

I thought so, as I already saw that bootloader checks hardware ID and
compares it with values 6228, 6226, 6223, etc.

>
> That's pretty interesting, since it could be used to create authentificated
> code for other phones with the secure romloader, too.
>

That's true, it should work.

>
> Well, there is the CECT C3100 which has JTAG solder pads. OpenOCD
> successfully detected the CPU, but I couldn't halt the ARM core (same
> problem on the Calypso btw). I tried other random ARM7 targets which had the
> same EmbeddedICE revision and there it worked.
>

I see that we were going through the same stuff.
I had the same problem with Sciphone G2. I could attach to core but I
couldn't halt it.
At the beginning I detected just four JTAG lines (TCK, TMS, TDI, TDO)
and could attach to CPU, but couldn't do more.
Then I discovered two more lines: RTCK and nTRST. When I connected
them, I still couldn't halt core, but it was just matter of
configuration.
When PLL is configured (CPU: 208/104MHz, AHBx4: 52MHz and AXBx8:
104MHz) 10MHz clock for JTAG should be used.
When PLL is not configured, then RTCK clock should be used.
I use Lauterbach for JTAG communication, but in openOCD should work as well.

>
> Getting osmocon to work with the secure romloader could be an option as
> well. Currently it only supports the non-secure romloader without that
> *.auth and SLA_Challenge.dll stuff.
> Plus we have a CFI-flash driver for the Compal phones, which might work on
> the MTK platform with a few modifications. If that works, we could flash
> U-Boot without any proprietary Flashtool/DownloadAgent.
>

Sounds good.

BR,
Marcin