The locking in the driver is far from being obvious, with unlocking
automatically happening at end of operations scheduled by interrupt,
especially for the error paths where one does not necessarily expect
that such an interrupt may be triggered. Let's add a few comments
about what to expect at certain places to avoid misdetecting bugs
which are not.

Link: https://lore.kernel.org/r/20200331094054.24441-24-w@1wt.euSigned-off-by: Willy Tarreau <w@1wt.eu>
Signed-off-by: Denis Efremov <efremov@linux.com>