On Sunday 28 February 2010, Ian wrote:
> The best reference page for the Bus Pirate is probably the manual:
> http://dangerousprototypes.com/bus-pirate-manual/

If so, the next iteration of this patch should use it.  :)


> Microchip's 16bit compilers are gcc-based, but you're correct that 
> vanilla gcc won't compile for the PIC.

... yes ... Microchip is one of the vendors that's not gotten
their support into mainline.  (Atmel's been fairly good with
AVR8, though they didn't start that work.  And they've not been
good with AVR32.)


> I went with the 24FJ chip because  
> it has a peripheral pin select cross-bar that can assign timers, 
> counter, PWM, SPI, UART, etc to any reprogrammable pin. A lot of the Bus 
> Pirate libraries use true hardware modules instead of software 
> emulation, and PPS makes it possible to swap these modules to the same 
> IO pins.

I think I read that somewhere.  Sort of understandable, if you value that
"swap modules to the same I/O pins" mechanism.  I've never wanted that in
I/O connectors though .... "use these pins for SPI, these for I2C, ..." and
so on is usually fine.  (Not particularly error prone, for one thing, and
I've not suffered pin shortages.)

 
> How's the USB driver situation with the M3?

They exist.  Few are yet GPL-compatible, I understand.  I've not
looked to see whether TI/Luminary, NXP, ST, and the other vendors
share any of the USB modules ... for some others (like SSP/SPI), there
is a lot of commonality, since they licensed the same IP from ARM.

That situation is probably better on ARM9 cores (or even ARM7) due
to platform maturity.  (Worst comes to worst, Linux-USB drivers
may be available to start the work.)


> There are plenty of  
> Microchip USB chips, probably an even cheaper solution than an M3,

I know that Microchip has some PIC-USB chips.  Dunno about cheaper;
I browsed prices at Digi-Key and found a lot of prices near $8 (US)
with at least 64K flash (and enough SRAM).

And in any case I'd gladly pay a couple bucks more to have hackable
tools using ARM instead of needing to invest time (and maybe $$) in
new tools I wouldn't use otherwise.


> but  
> there's currently no open source USB driver. Microchip has a free USB 
> stack, but it's gratis and not libre, so I can't actually distribute it 
> with any projects.

That seems to be a common problem.  For avr8 there is a libre stack.
SInce it's the same full speed USB module, that'd work on AVR32 too
(but not high speed USB, ISTR).

I've not tracked PIC vs AVR8 in any detail, so I'm roughly expecting
them to have the same performance levels.  For JTAG usage a downside
of the AVR8 is that their SPI doesn't clock particularly fast; 1/4 of
the CPU clock is what I recall, with 20 MHz being a quick AVR8.

The XMega chips go faster than that ... but they have no USB (unless
you like bitbanging the low-speed flavor).

And the AVR32 UC3 series combine high speed USB options with fast clocks
for SPI .... but there's that issue getting its GCC support mainlined.
(Someone -- not from Atmel! -- finally submitted a patch late October,
but I don't know if it merged yet.)



Yeah, there are a lot of microcontrollers with USB support, which could
be used for something like this.  Not many with GCC support though, and
Libre software stacks aren't necessarily available ... sometimes the
marketers in that space are offended by the dynamics of open sourcing.

But that would be part of the hackability too.  None of that stuff is
especially hard.  Get a decent start, and you're flying.

- Dave



_______________________________________________
Openocd-development mailing list
Openocd-development@lists.berlios.de
https://lists.berlios.de/mailman/listinfo/openocd-development

Reply via email to