[01:32:58] *** Joins: tomzawadzki (uid327004@gateway/web/irccloud.com/x-tboocelkkxwymftv) [02:42:56] *** Quits: peluse (peluse@nat/intel/x-vodyuajkodygizfx) (Quit: ZNC - http://znc.in) [02:44:12] *** Quits: mszwed (~mszwed@134.134.139.76) (Read error: Connection reset by peer) [03:03:41] *** Joins: mszwed (mszwed@nat/intel/x-uqulpmkocywdutvz) [03:40:13] *** Quits: guerby (~guerby@april/board/guerby) (Remote host closed the connection) [03:41:34] *** Joins: guerby (~guerby@april/board/guerby) [06:36:03] is anybode else experiencing a problem to push commits (internal error returned)? [06:44:04] I see that http://status.gerrithub.io/ is down, so it's probably not only my problem [06:52:28] Pawel confirmed that it is not working for him as well... in that case we need to wait until they fix it [06:59:19] *** Joins: peluse (~peluse@134.134.139.75) [06:59:19] *** ChanServ sets mode: +o peluse [07:01:55] ok, it's up now [08:06:18] *** Joins: travis-ci (~travis-ci@ec2-54-90-128-182.compute-1.amazonaws.com) [08:06:19] (spdk/master) pkgdep: install pycodestyle in favor of pep8 (Seth Howell) [08:06:20] Diff URL: https://github.com/spdk/spdk/compare/12759ab569d9...df517550cc9c [08:06:20] *** Parts: travis-ci (~travis-ci@ec2-54-90-128-182.compute-1.amazonaws.com) () [08:07:44] *** Joins: travis-ci (~travis-ci@ec2-54-90-128-182.compute-1.amazonaws.com) [08:07:45] (spdk/master) event: The reactor now contains a list of threads (Ben Walker) [08:07:45] Diff URL: https://github.com/spdk/spdk/compare/14e443bdfbf9...12759ab569d9 [08:07:45] *** Parts: travis-ci (~travis-ci@ec2-54-90-128-182.compute-1.amazonaws.com) () [10:39:32] *** Quits: tomzawadzki (uid327004@gateway/web/irccloud.com/x-tboocelkkxwymftv) (Quit: Connection closed for inactivity) [11:37:08] @jimharris, see https://review.gerrithub.io/c/spdk/spdk/+/442896 when you get a chance, let me know what you think... [12:38:23] jimharris, I was thinking about one of your compress vbdev comments in the channel callback where I use the same mechanism I used in crypto to assign queue pairs. I'm thinking that besides not having to do anything if the reduce channel has already been set, I don't need the locks or the device_qp struct at all anymore, we'll only ever at most use 1 qp. What do you think? [12:41:02] doing so will simplify and remove a lot of code and I don't see any future changes that would allow > 1 channel to the underlying bdev [14:03:57] jimharris, QQ on channel funneling? [14:05:33] sure [14:06:08] on your previous question - yeah, we should only ever have one 'qp' per compress bdev [14:08:22] OK, cool that's whats up there now. Almost done w/the channel stuff but I notice, and it makes sense to me, that the iscsi driver is tracking the submit thread in their IO context and sending the completion to that thread when done. I wasn't doing that but it wasn't in the review so wanted to make sure I need to [14:10:23] well, actually I don't have a completion function yet so it would be a TODO on this patch :) [14:12:48] *** Quits: lhodev (~lhodev@66-90-218-190.dyn.grandenetworks.net) (Ping timeout: 272 seconds) [14:13:19] *** Joins: lhodev (~lhodev@inet-hqmc08-o.oracle.com) [14:17:53] yeah - you'll definitely need to explicitly send the completion back to the master thread [14:18:13] we have to complete the bdev_io back to the bdev layer on the same thread that it sent it to us [14:21:19] *** Quits: michalwy (~michalwy@192.55.54.42) (Ping timeout: 246 seconds) [14:22:52] I believe the bdev_io already holds the original I/O channel [14:22:55] to send it back [14:31:59] oh really? Wonder why iscsi bdev didn't use that, I'll take a look. thanks [14:34:17] bwalker, do you mean io_submit_ch? It in __bdev_io_internal_fields so I don't think I want to use it, or if it makes more sense to we should take it out of private yeah? [14:34:55] I do mean that one - is there an accessor function? Shuhei may have a patch out to add one actually [14:35:52] I'll check it out... [14:38:12] yup, looks like it was merged 2 days ago [14:39:45] *** Joins: pawelkax (~pawelkax@134.134.139.75) [15:37:05] *** Joins: Shuhei_ (caf6fc61@gateway/web/freenode/ip.202.246.252.97) [17:03:07] yeah - i had shuhei add this as part of his dif series [17:04:16] he needed it for the case where an I/O fails a DIF check, so he issues the I/O again without DIF so he can figure out which block was bad [18:46:06] *** Joins: zhouhui (~wzh@114.255.44.139) [18:47:01] Hi, is GerritHub goes wrong? GerritHub report server error when I sign in or push patches. [19:04:03] Nevermind, the GerritHub recovers. [19:16:12] *** Quits: lhodev (~lhodev@inet-hqmc08-o.oracle.com) (Remote host closed the connection) [19:16:48] *** Joins: lhodev (~lhodev@66-90-218-190.dyn.grandenetworks.net) [20:06:49] *** Quits: Shuhei_ (caf6fc61@gateway/web/freenode/ip.202.246.252.97) (Ping timeout: 256 seconds) [20:19:46] *** Joins: travis-ci (~travis-ci@ec2-52-203-111-139.compute-1.amazonaws.com) [20:19:47] (spdk/master) nvme: add the asynchronous controllers probe/poll APIs (Changpeng Liu) [20:19:47] Diff URL: https://github.com/spdk/spdk/compare/df517550cc9c...bad30d5366e4 [20:19:47] *** Parts: travis-ci (~travis-ci@ec2-52-203-111-139.compute-1.amazonaws.com) () [20:31:17] *** Joins: Shuhei_ (caf6fc61@gateway/web/freenode/ip.202.246.252.97) [20:49:57] *** Quits: zhouhui (~wzh@114.255.44.139) (Quit: WeeChat 1.9.1) [21:34:17] *** Joins: travis-ci (~travis-ci@ec2-34-229-48-169.compute-1.amazonaws.com) [21:34:18] (spdk/master) rdma: add return to avoid address points to the zero page (yidong0635) [21:34:18] Diff URL: https://github.com/spdk/spdk/compare/bad30d5366e4...9d838d24ad4d [21:34:18] *** Parts: travis-ci (~travis-ci@ec2-34-229-48-169.compute-1.amazonaws.com) ()