[karmic] Bluetooth killswitch doesn't reliably work on Dell BT370 or BT410

[Mario Limonciello]

Binary package hint: bluez

In karmic, the whole way that Dell BT adapters work has changed and in the process a race condition has been exposed causing the killswitch to work about 20 percent of the time.

When you press the killswitch, here's what happens on a Dell laptop now:

1) The dell-laptop kernel module attaches the BT adapter to the USB bus using a SMI call to the SMBIOS.
2) When the adapter is attached to the bus, a keyboard and mouse (both HID) device are each visible on the bus. No BT adapter is available.
3) A udev rule runs the hid2hci tool which is now part of the udev package (formerly bluez, then udev-extras, now udev).
4) That udev rule matches on when the mouse device shows up on the bus and tries to do some usbfs operations.

Unfortunately there is a race condition for when those usbfs symlinks are created and when you are actually able to do an operation on the bus.

[Mario Limonciello (superm1)]

The attached patch by Alan Stern fixes the issue. According to him,
"It will appear during the next merge window and afterward will go into 2.6.31-stable (and probably 2.6.27-stable too)."

[Mario Limonciello (superm1)]

This is this commit in the kernel tree:

http://kernel.ubuntu.com/git?p=ubuntu/ubuntu-karmic.git;a=commit;h=6a6fe62e9d1420ae674a52f795a53164d5205962