Inexplicably yesterday's (and the day before) Debian testing snapshot
seems to have the problem fixed!
I tested kernel 3.2.x and 3.1.x. udev still is at version 175.
I'm completely ignorant to what change in Debian testing led to the
fixing of this problem. I guess I should have tested everyday, for
every round of updates to testing, but I lacked the time.
I'm attaching a dmesg, though it shows nothing of interest AFAIK.
I hope this issue doesn't come back to haunt me. I'll upgrade to
wheezy on this machine soon so I can monitor everything more closely
instead of relying on live images.
Cheers,
Andres
> Inexplicably yesterday's (and the day before) Debian testing snapshot
> seems to have the problem fixed!
>
> I tested kernel 3.2.x and 3.1.x. udev still is at version 175.
>
> I'm completely ignorant to what change in Debian testing led to the
> fixing of this problem. I guess I should have tested everyday, for
> every round of updates to testing, but I lacked the time.
>
> I'm attaching a dmesg, though it shows nothing of interest AFAIK.
>
> I hope this issue doesn't come back to haunt me. I'll upgrade to
> wheezy on this machine soon so I can monitor everything more closely
> instead of relying on live images.
I finally installed wheezy by default on the laptop... it ended up
being rather hard...
Anyways the problem is back. I don't know what change in the Usb
images I tested days ago may have resulted in a fix.
I've tested the most recent Debian kernel -> basically 3.2.5 and also 3.3.0-rc3
Both throw the same error. I'm attaching dmesg from 3.3.0-rc3, but
there's absolutely no new information.
Should I enable the debugging symbols to test if there's more output?
my build had them disabled, otherwise the kernel image was huge.
Andres
I believe I have finally narrowed down the problem. It lies in the
'libmtp-runtime' package with the command 'mtp-probe' (only present in
wheezy or newer).
I'm not sure what this is doing to my system, but when the package
mentioned is installed, it "messes" up my atheros bluetooth device so
that the kernel/udev cannot load the firmware into it.
As soon as I uninstalled this package and thus caused the boot process
to throw some errors about missing mtp-probe command, my bluetooth
module was correctly loaded and the led turned on.
>> Inexplicably yesterday's (and the day before) Debian testing snapshot
>> seems to have the problem fixed!
I guess this snapshot that I tested earlier didn't have that package installed
I found out about this completely by chance. For whatever reason, on
my second upgrade to wheezy 'apt-get dist-upgrade' didn't pull
'libmtp-runtime' and my bluetooth module just worked.
I tried to get the bootlog messages so I can send this info, but even
with bootlogd enabled these messages happen too early on the boot
process to be logged in /var/log/boot.
I feel this is not necessarily a kernel bug anymore (all kernels
tested work fine 3.0.x, 3.1.x, 3.2.x and even 3.3.0-rc3). But the
experts will know better.
This is definitely progress!
Cheers
Andres
PS: please forward the info to other bug reports I don't have access
to, like Red Hat / Fedora and Ubuntu. Additional testing is needed.