[01:09:37] *** Joins: ziyeyang_ (~ziyeyang@134.134.139.75) [01:27:57] *** Joins: tomzawadzki (~tzawadzk@134.134.139.74) [01:45:43] *** Quits: ziyeyang_ (~ziyeyang@134.134.139.75) (Remote host closed the connection) [02:01:51] *** Joins: ziyeyang_ (~ziyeyang@134.134.139.75) [02:13:23] *** Joins: ziyeyang__ (ziyeyang@nat/intel/x-cootfnmwktvfxlep) [02:13:23] *** Quits: ziyeyang_ (~ziyeyang@134.134.139.75) (Remote host closed the connection) [02:17:33] *** Quits: ziyeyang__ (ziyeyang@nat/intel/x-cootfnmwktvfxlep) (Remote host closed the connection) [02:18:43] *** Joins: ziyeyang_ (~ziyeyang@134.134.139.75) [02:21:31] *** Quits: ziyeyang_ (~ziyeyang@134.134.139.75) (Remote host closed the connection) [02:22:14] *** Joins: ziyeyang_ (~ziyeyang@134.134.139.75) [02:31:43] *** Quits: ziyeyang_ (~ziyeyang@134.134.139.75) (Remote host closed the connection) [02:47:19] *** Joins: ziyeyang_ (ziyeyang@nat/intel/x-ygmaglzpamratwqp) [02:57:13] *** Quits: ziyeyang_ (ziyeyang@nat/intel/x-ygmaglzpamratwqp) (Remote host closed the connection) [08:07:33] *** Quits: tomzawadzki (~tzawadzk@134.134.139.74) (Ping timeout: 260 seconds) [09:09:50] *** Quits: drv (daniel@oak.drv.nu) (Ping timeout: 240 seconds) [09:09:58] *** Joins: drv (daniel@oak.drv.nu) [09:09:58] *** ChanServ sets mode: +o drv [09:59:01] drv: i had to rebase my bdev patch set - no differences since your last review - bwalker has already added his +2 [09:59:43] I will check it out [10:01:37] cool - thanks [11:23:36] I have started tagging reviews that I think are critical to get in for the 17.07 release with topic 'spdk-v17.07' [11:23:39] https://review.gerrithub.io/#/q/status:open+project:spdk/spdk+topic:spdk-v17.07 [11:23:56] please take a look and let me know if there are any you think should be added (or removed) [12:31:22] drv: not sure we will get the vhost hotplug stuff in time - cunyinch, darek and i have an e-mail thread working through this [12:31:41] but darek and cunyinch will be driving this next week while i'm out [12:55:35] ok, we can drop that one from the list [13:01:20] i've provided input on the rest (and pushed one of them which was ready) [13:03:17] bwalker: i added some feedback on roman's vtune patch [13:03:28] i'm trying to picture what this looks like long-term [13:03:48] if i'm reading the algorithm correctly, it only really works for something like bdevperf [13:41:55] yeah - I can't really devise an algorithm that measures what we want to measure [13:43:03] I could put something into the reactor, but that means vtune only works with our event framework [13:43:23] and essentially count the time spent in each iteration of the reactor loop where no real work was accomplished [13:43:48] and then the idle time also can't be traced back to the disks [13:43:52] are we OK with deferring this until after the release, then? (does the VTune integration work without it?) [13:43:53] which maybe it can't [13:44:16] i was thinking along those lines too until my brain broke trying to think how the pollers tell the reactor whether they performance any work or not [13:44:37] it has to be a return value [13:44:40] it seemed like that would a slew of return codes all over the place [13:45:24] maybe we need to try to prototype that [13:45:37] I can take a quick shot at it and see how bad it looks [13:45:39] with something relatively complicated, like a vhost-scsi poller [13:59:50] hmm - the other io stats really do need to be on the bdev [14:00:04] so I can get rid of the SPDK_CONFIG_VTUNE stuff from the nvme bdev module [14:00:09] but not the others in the bdev layer [14:00:24] yeah - the bdev stuff has to conditionally compile in the itt stuff [14:00:24] I'm also not sure what the rules are for creating two itt domains [14:01:16] you in the lab? [14:01:19] yeah [18:36:11] *** Joins: guerby_ (~guerby@ip165.tetaneutral.net) [18:37:25] *** Quits: guerby (~guerby@april/board/guerby) (Remote host closed the connection)