Re: [Barry-devel] Barry version 0.6 (alpha) released

[Chris F. <cd...@fo...>]

On Sun, Mar 11, 2007 at 06:39:37PM -0700, troy engel wrote:
> On 3/11/07, Miles Sabin <mi...@mi...> wrote:
> >
> > But, OTOH, this was more or less working for you a week or so ago with
> > the earlier firmware, right? So even if the newer firmware solves the
> > problem there ought to be a fix which works with the older?
> 
> Yes, it did! I had it working on 2007-02-19 of CVS with my old
> firmware; try rolling back to whatever code was in that CVS tree at
> that time and see if btool works for you. Something in the code was
> changed after that point which started causing the hang; how long
> after that point, I'm not sure.

The probe code did change on Feb 23/24th.  But what I've been seeing
in the lsusb output from both you and Miles is that it seems to be
something more basic:  the endpoints btool needs to use to talk
the database protocol just don't show up.

The possible causes for this are:
	- a kernel issue... you are both running the same kernel, so there
		is a slight chance this could be a factor... I'm running
		a stock 2.6.18.1 kernel if anyone is curious.

	- a firmware issue... this looks likely since Troy has had success
		with new firmware... but I have a hard time believing RIM
		would release firmware that Windows can't make backups with,
		so I'm curious how Windows does manage to talk to it.

	- a bcharge issue... this is always possible.  I reverse engineered
		the Pearl codes with a quite recent version of the Pearl,
		so maybe I'm using something that is not fully backwards
		compatible.  Unfortunately I don't have an older Pearl
		to snoop with.

Running "bcharge -o" seems to work no matter which version of firmware
you are running, but then you don't have usb_storage support until you
reconnect the device.  It is an available workaround though.

- Chris