On Mon, 2013-06-17 at 22:13 -0700, Kent Overstreet wrote: > On Fri, Jun 14, 2013 at 03:40:13PM -0700, Nicholas A. Bellinger wrote: > > Hey Kent, > > > > So after rebasing target-tag-pool to include your percpu-tags v2 code > > from earlier this week, the BUG_ON(tag >= pool->nr_tags) below is > > immediately being triggered in the tcm_vhost response -> release path > > with a 4x tcm_vhost endpoint / ramdisk LUN + scsi-mq virtio-scsi client > > setup. > > Hey, sorry for not getting back to you sooner - I ended up heavily > reworking it again, and rewriting ida/idr because Andrew Morton was > complaining so much :) I'm mailing the new version out now - and last > week while I was going over akpm's review feedback I noticed an ABA that > might be the bug you ran into - if so, that should be fixed now. Hey Kent, Apologies for the lack of updates this past week, just catching up on the latest percpu-ida bits now that the immediate v3.11 target items are out of the way.. Just a quick update on my end, target-pending/target-percpu-ida has been rebased following your latest linux-bcache/ida branch. Things are running wonderfully with the last rev, and I'm still making a few other pre-allocation optimizations in vhost-scsi code in order to drop the remaining fast-path allocations for per descriptor SGL + user page list array memory. No further progress on the conversion of iscsi-target/iser-target and other target drivers just yet, as there some cleanups in the former that need to happen before a proper percpu-ida conversion can happen. So now that the chances are looking good for percpu-ida + idr conversion being merged into v3.11, these target conversions will end up becoming v3.12 items. For next development cycle, I'd like to see vhost-scsi, ib_srpt, tcm_qla2xxx and iscsi-target/iser-target conversions, along with one other new fabric driver currently in the works. So all that said, I'm very excited to see percpu-ida upstream! :) --nab -- To unsubscribe from this list: send the line "unsubscribe target-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
