RE: [avr-gcc-list] using the 43USB355 (was - Using IAR compile d lib with GCC?)

[Bryan Jarmain]

Keith,

Atmel probably made the right decision in keeping the Hub/Function together
to begin with, as it covers most options, and as you said, it is easier to
optimize and keep small like this. I also understand your concern in keeping
certain key parts of the device as proprietary as possible to maximize the
barrier to entry from potential competitors. It still seems to be a pity to
make such a nice device less attractive, just in order to confuse potential
competitors. Anyhow, a split library solution (proprietary/open) would be a
great next step.

The USB initialization part is where my attempted development ended. The
reason was that there was just insufficient information on how the device
fitted together, in what sequence the registers should be set, and what
their exact functions were (some items also seemed to be undocumented) - I
generally like to program a device at the hardware level - this helps me get
a better understanding of the technology as a whole, and should allow me
better flexibility in my design.

For these reasons, I would find example code (and hopefully more detailed
documentation) greatly helpful. I don't really intend/need to use the hub
just yet - this is not really required by me. The function in itself already
provides a great solution that has the excellent AVR performance, the single
chip compactness, and all at a very compelling price. 

Once all the T's are crossed and the i's dotted, I can't see any reason why
this family of devices won't become the number one choice as single chip
full speed USB solutions - better than the EZUSB... I also look forward to
seeing the anticipated new Atmel devices planned for next year.

Bryan

-----Original Message-----
From: Keith Gudger [mailto:address@hidden 
Sent: Tuesday, September 02, 2003 6:15 PM
To: Bryan Jarmain
Cc: address@hidden
Subject: RE: [avr-gcc-list] using the 43USB355 (was - Using IAR compile d
lib with GCC?)

Bryan:

I'll get to work on the boss & code (in that order :).  

What (we think) is nice about our current crop of AVR / USB devices is
that a lot of what others do in hardware is left to firmware.  That gives
us a lot more flexability.  In my chats with other USB chip/firmware
developers, I hear some amazing horror stories about what you cannot /
have to do when most of the stuff is in hardware.  I'm sure the
participants on this list can appreciate this...

There are 2 components of the firmware:  hardware configuration and USB
interrupt / response routines.  Configuring the hardware is easy.  All of
the many interrupt routines to satisfy the USB requirements - that's a lot
of effort.  The code must pass chapter 9 (and chapter 11 for the hub).  To
save code space, our library uses the same routines for the hub and the 
function.  That's why there isn't a separate hub and function library.  We
put a lot of effort into using pointers and structures so that we could
reuse the code.

BTW, if you crack open an x-box controller, you'll find the AT43USB355.
Winning that design required a bulletproof library.  That's partly why we
say the USB libraries have been thoroughly checked out.

Keith

On Tue, 2 Sep 2003, Bryan Jarmain wrote:

> The 355 seems to be a really nice device (even without the HUB), but the
USB
> configuration side seems to be very difficult to configure, and example
code
> would go a long way in making the device more usable.
> 
> I never immagined that the hub would have firmware that would need to be
> protected - logically and traditionally these types of propriatry features
> are typically hidden in hardware. I believe that if the hub firmware needs
> to be protected, this could be most elegantly done in a separate "hub
> library" - but this is just my own speculative opinion.
> 
> So, function only example firmware would be really great and helpfull. 
> It would help substantually with using the device more efficiently. An
open
> function only library would also serve a similar purpose, and would be
> usefull to many other potential users.
> 
> I look forward to your success in getting this open function code together
> :-)
> 
> Bryan
> 
> -----Original Message-----
> From: Keith Gudger [mailto:address@hidden 
> Sent: Tuesday, September 02, 2003 5:12 PM
> To: Jason Kyle
> Cc: Bryan Jarmain; address@hidden
> Subject: Re: [avr-gcc-list] using the 43USB355 (was - Using IAR compiled
lib
> with GCC?)
> 
> Here is the story about the 355 libraries.  The reason for not releasing
> the source code is to protect the code for the hub.  If you look around,
> you will find that "other USB devices" are function only.  Sorry, but it
> appears we have something worth protecting in the hub.
> 
> Are you are building a function only device?  If so, maybe I should put
> together code that works only for a function.  Maybe the boss will let me
> release that source code?  If that's what you want, I'll look into
> releasing a GCC version.
> 
> Keith
> 
> On Mon, 1 Sep 2003, Jason Kyle wrote:
> 
> > At 09:24 01/09/2003 +0200, Bryan Jarmain wrote:
> > >I recently tried to use the AT43USB355, but could not get the USB side
to
> > >work, even after quite considerable effort. I found the root cause to
be
> > >inadequately and inaccurately documented hardware (USB) functionality.
> The
> > >rest of the device seemed easy enough to use (although considering
there
> is
> > >no emulator support - I implemented a 115kbaud bit-bang serial port in
> > >software for this purpose).
> > >
> > >Atmel sent me some beta 43USB355 USB libraries with a wizard (I believe
> they
> > >may be properly released now), but I was reluctant to use them because
> their
> > >source code was not open (proprietary) - I don't believe using closed
> > >libraries is an industrial standard way of writing firmware. Especially
> with
> > >the library having such a fundamental influence on the firmware
structure
> > >(e.g. no main function in your C code). They also didn't provide other
> > >source-code examples on how to use their USB hardware features - I
> couldn't
> > >even initialize the USB functionality.
> > >
> > >Anyway, to cut a long story short I decided it would be prudent to find
a
> > >better documented USB device (where I had better knowledge and
> understanding
> > >of what was happening behind the scenes), until Atmel decides to better
> > >document the 43USB355 (and hopefully expose their library source code
> )...
> > >
> > >All this was a pity because the 43USB355 seems to be a very nice
> > >microprocessor with lots of promise. Currently it is probably very
usable
> > >for anyone who is prepared to use Wizards and Proprietary Libraries
(they
> > >claim their libraries are quite stable - I have no reason to doubt
this).
> I
> > >just believe it still needs a bit of work from their side before it
> becomes
> > >a widely acceptable micro.
> > 
> > Next year we should see some decent USB micros from Atmel, the current
> crop 
> > is based around the original AVR core.  Hopefully Atmel (hint hint) will

> > prepare both open (for gcc) and proprietary (for IAR C & others)
libraries
> 
> > for the new batch of USB AVR's.  All this proprietary talk reminds me of

> > the FPSLIC devices, is anybody still using them?
> > 
> > Jason Kyle
> > 
> > 
> > >Has anyone been successful with this device (without using their
> libraries).
> > >If anyone has had success and has source-code to share, I would be very
> > >interested to see some example code on the hardware interface.
> > >
> > >Bryan
> > >
> > >
> > >-----Original Message-----
> > >From: Keith Gudger [mailto:address@hidden
> > >Sent: Friday, August 29, 2003 6:14 PM
> > >To: Tim Lapawa
> > >Cc: address@hidden
> > >Subject: Re: [avr-gcc-list] Using IAR compiled lib with GCC?
> > >
> > >I have USB libraries for the AT43USB35x USB AVRs, compiled for GCC.  I
> > >haven't gotten the 320 libraries compiled for GCC yet, but there is a
> > >io43u32x.h file, and one can set -mmcu=at43usb320. (WinAVR release, at
> > >least).
> > >
> > >I can send you the 35x files anytime, but I will have to compile the
32x
> > >libraries next week (when I'm back from vacation ;) ).
> > >
> > >Keith
> > >
> > >On Fri, 29 Aug 2003, Tim Lapawa wrote:
> > >
> > > > Hello list,
> > > >
> > > > atmel send me a proprietary library for the AT43USB320 SDK kit.
> > > > It's an 4port USB Hub with integrated 8515 integrated.
> > > > But they do not want to send me the firmwares source code.
> > > > They send me libraries compiled with the IAR and the Imagecraft
> compiler.
> > > > Can I use these libs with the GNU toolchain?
> > > > I've read in this list, that the IAR and the GNU compiler use
> > > > different registers for Stack- and Framepointer.
> > > > And what is about parameter passing in function calls?
> > > >
> > > >
> > > > Greetings Tim Lapawa
> > > > --
> > 
> > 
> > _______________________________________________
> > avr-gcc-list mailing list
> > address@hidden
> > http://www.avr1.org/mailman/listinfo/avr-gcc-list
> > 
> > 
> > 
> 
> 
> _______________________________________________
> avr-gcc-list mailing list
> address@hidden
> http://www.avr1.org/mailman/listinfo/avr-gcc-list
> 
> 
>