On Wed, 2012-04-11 at 15:20 +0100, Chris Boot wrote:The FireWire SBP-2 Target is a driver for using an IEEE-1394 connectionSome very nice work to make this all go Chris.. 8-)
as a SCSI transport. This module uses the SCSI Target framework to
expose LUNs to other machines attached to a FireWire bus, in effect
acting as a FireWire hard disk similar to FireWire Target Disk mode on
many Apple computers.
Sorry this latest revision has been a long time coming. I was trying to
chase down a crashing bug (but haven't been able to replicate it), changed
my mind about a large portion of the code twice, and life got in the way as
well!
Thanks again btw for your target_core_mod patches along the way to v3!
Changes in v3:So I've merged a squashed version this series from:
* Updates for target framework API changes
* Attempt to make headers self-contained
* Remove bad use of atomics and memory barriers
* Rework use of locking
* Use system workqueues
* Wrap fw_run_transaction() to retry failed transactions
* Coding style fixes
* Fix a few bugs
* Overhaul sbp_rw_data()
* Overhaul target fetch agent
https://github.com/bootc/Linux-SBP-2-Target.git patch-v3
into lio-core/master, and have now pushed to k.o upstream
This looks to me like a reasonable candidate for a v3.5 mainline merge,
so we need to go ahead and get this into target-pending.. Also, thanks
for getting Stefan's ACK's here on the drivers/firmware/ specific
changes..
So for new fabric drivers (as discused with Sebastian and usb-gadget
recently) that the target team has been enforcing that new fabric
modules convert to use a single source / header file for readability and
maintainability purposes.
It would be great if you can publish one more branch (no need to re-post
the full diff) that takes your patch-v3 down to a single .c/.h file.
Otherwise, I'm fine to take the 15 minutes to do this conversion ahead
of dropping into target-pending for the sunday night linux-next build.