Re: [Bluez-devel] [ANN] Zaurus BlueZ kernel packages for ROM v3.10

[Justin Karneges <justin-qt@affinix.com>]

> the two big parts with BlueZ are the kernel modules and the userspace
> stuff. The kernel stuff is the essential and the most problematic part.
> And I think that we should have a package which only contains the BlueZ
> kernel modules. Maybe we need different builds from the same source for
> Sharp Kernel 2.4.6, 2.4.18 and OpenZaurus Kernel 2.4.18, 2.4.19.

This would be nice, if indeed we can support this on 2.4.6.  Although, I 
believe the earlier Sharp ROM has similar suspending issues related to 
cardmgr (this is what Max told me in January).  For bluetooth to work 
reliably here, we may need to offer a cardmgr upgrade.

> I don't have a Zaurus, so I can't help here. Who wants to volunteer for
> the Zaurus kernel package maintainer?

I would like to volunteer for this, when I get a build setup working (I will 
try Stephen's soon).

> The BlueZ CVS module "kernel" contains a build system for building the
> BlueZ kernel modules outside a kernel source tree. I am willing to
> extend this as needed.

What extensions do you suppose are needed?

> > Also, I think I found a bug in dtl1_cs and/or bluez itself:  the card
> > cannot survive a suspend.  Turning off / on the Zaurus, or simply running
> > "cardctl suspend ; cardctl resume" will leave the driver in an unusable
> > state. Running "cardctl eject ; cardctl insert" fixes the problem.  I'm
> > willing to hack on the driver and remedy this, if I could just figure out
> > how to cross-compile the modules...
>
> This is a known problem, but I don't have a Zaurus and so I can't test
> it. On my development machine is this working without any problems.

Do you have any hint about where it may be?  Maybe the Zaurus does not send 
appropriate signals to the driver that it is shutting down?  I'd like to look 
into this, but I need some direction first.  For instance, if it affects all 
drivers, not just dtl1_cs, then either all drivers have the same "bug", or 
the bug is in the bluez core, right?

What is significant about cardctl suspend?  This will break the driver also, 
and perhaps it easier to look into this than the full pda suspend.

-Justin