The slave_busy status doesn't need to persist, so just use a stack variable
instead.

The work queue would have been leaked in the case where the GPIO request failed
because the error handling cases were in the wrong order.

 - Remove trailing whitespace
 - Wrap lines at 80 chars
 - Format else statements according to linux coding style

when there is an error stopping record push timer when a
session stopped event is received.

stopped.

GPIO requests were passing pointers to stack allocated strings under the
assumption that the label string contents would be copied to another
storage location during the request.  This assumption was incorrect.
The string pointer is just copied, so the pointer becomes unsafe to
follow once the requesting function returns or re-uses the string.

The chip isn't capable of reporting this value.

This change resolves a problem reported on the mangOH forum
(http://forum.mangoh.io/t/controlling-wp-gpio-1-gpio22-problem/984)
where GPIO 22 could not be exported to userspace because it was
incorrectly registered as an interrupt in the bq24296 kernel module.

to insert a 0-byte into the Rx stream

 - added UART shutdown req/rsp msg
 - updated location where UART close called so its not called while atomic

was a NULL pointer was getting passed to cfg80211.

Put back commented out CAN drivers