firewire: ohci: increase AT req. retries, fix ack_busy_X from Panasonic camcorders...
authorStefan Richter <stefanr@s5r6.in-berlin.de>
Tue, 20 Jan 2009 18:10:58 +0000 (19:10 +0100)
committerStefan Richter <stefanr@s5r6.in-berlin.de>
Sat, 24 Jan 2009 10:17:27 +0000 (11:17 +0100)
commit8b7b6afaa84708d08139daa08538ca3e56c351f1
tree91b1d27a08f6e28c13f8163cc2dfbba4022029f0
parentb006854955254a971096c120d4ef115a7c6145fb
firewire: ohci: increase AT req. retries, fix ack_busy_X from Panasonic camcorders and others

Camcorders have a tendency to fail read requests to their config ROM and
write request to their FCP command register with ack_busy_X.  This has
become a problem with newer kernels and especially Panasonic camcorders,
causing AV/C in dvgrab and kino to fail.  Dvgrab for example frequently
logs "send oops"; kino reports loss of AV/C control.  I suspect that
lower CPU scheduling latencies in newer kernels made this issue more
prominent now.

According to
https://sourceforge.net/tracker/?func=detail&atid=114103&aid=2492640&group_id=14103
this can be fixed by configuring the FireWire controller for more
hardware retries for request transmission; these retries are evidently
more successful than libavc1394's own retry loop (typically 3 tries on
top of hardware retries).

Presumably the same issue has been reported at
https://bugzilla.redhat.com/show_bug.cgi?id=449252 and
https://bugzilla.redhat.com/show_bug.cgi?id=477279 .

In a quick test with a JVC camcorder (which didn't malfunction like the
reported camcorders), this change decreased the number of ack_busy_X
from 16 in three runs of dvgrab to 4 in three runs of the same capture
duration.

Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
drivers/firewire/fw-ohci.c