[00:03:13] *** Joins: tkulasek (~tkulasek@134.134.139.83) [01:59:04] *** Joins: tkulasek_ (~tkulasek@134.134.139.75) [02:02:20] *** Quits: tkulasek (~tkulasek@134.134.139.83) (Ping timeout: 245 seconds) [03:41:41] *** Quits: dlw (~Thunderbi@114.255.44.143) (Ping timeout: 265 seconds) [04:35:44] *** Joins: lyan (~lyan@2605:a000:160e:3c8:4a4d:7eff:fef2:eea3) [08:07:41] *** Joins: bwalker_ (~bwalker@ip70-190-226-244.ph.ph.cox.net) [08:07:41] *** ChanServ sets mode: +o bwalker_ [08:24:42] *** Joins: tomzawadzki (~tomzawadz@192.55.54.42) [08:25:27] did someone just pull the power cord? [08:25:43] as far as I can tell, yes [08:25:57] I think we are having some technical difficulties with webex [08:26:11] webex is having some technical difficulties because it is a terrible program [08:26:23] I think peluse was the host and left the meeting, which ended it [08:26:45] I can't restart the meeting either as far as I can figure out [08:27:15] I rejoined as host - maybe it should be back now? [08:27:37] I wanted to describe the issue and propose a solution, but this can be done through email as well [08:27:38] and the last trello item is from @mszwed who's not here yet [08:27:50] we're back in [08:27:51] yes, I think we can talk it over via email [08:28:23] but basically, we do need some allocations to be guaranteed physically contiguous; I'm not sure if DPDK 18.05 provides some new interface for this [08:29:06] not beyond 2mb boundaries though right? [08:29:37] depends on how big your queues are [08:30:21] submission queue entries are 64 bytes, and if you had a 64K entry queue, it would be 4MB, but that's an extreme case [08:35:31] ? [08:35:32] hmm, meeting ended again? [08:35:43] y [08:35:46] we can't continue with webex [08:36:53] drv: we can still use DPDK memzones to request physically contiguous memory [08:37:10] oh, ok [08:38:13] memzones have a less nice API, but I suppose we could try to make that work [08:38:14] it will probably do the same "hacks" you mentioned [08:38:31] *** Quits: bwalker_ (~bwalker@ip70-190-226-244.ph.ph.cox.net) (Quit: Leaving) [08:39:18] I can't join this webex call with audio [08:40:00] there was no host - it might be working now [08:43:48] ahh, it's my cpu not keeping up [08:43:58] I had to left the meeting [08:53:49] leave* [08:59:14] *** Joins: philipp-sk (~Philipp@ktnron0916w-lp140-04-65-92-69-234.dsl.bell.ca) [09:01:19] hi there, a question about CI: i'm hitting on on the freebsd-0 node: [09:01:32] 10: fatal error: 'infiniband/verbs.h' file not found [09:02:08] do i need to ifdef the include for freebsd somehow? [09:02:25] the FreeBSD machine doesn't have the RDMA stack installed, yes [09:03:13] do you mean this patch? https://review.gerrithub.io/#/c/spdk/spdk/+/416748/ [09:03:34] yep [09:03:43] currently, all of the infiniband verbs code is isolated in rdma.c, and its compilation is controlled by CONFIG_RDMA in lib/nvmf/Makefile [09:04:13] you could add #ifdef SPDK_CONFIG_RDMA around all uses of ibverbs code, or try to keep it all in rdma.c if that is possible [09:05:12] well, i thought bringing it up to nvmf to be honest [09:05:23] it's not particular useful right now, since RDMA is the only transport at the moment, but there will be more coming soon [09:05:39] so we want to keep the generic NVMe-oF target code isolated from the transport layer [09:06:48] so more spdk_transport_xxx api to come, i guess.. [09:07:58] hopefully we won't need many new spdk_transport_* APIs, but there will be multiple spdk_nvmf_transport_ops implementations [09:17:30] the ugliest hack that comes to mind is to ifdef all ibv stuff i've added with that SPDK_CONFIG_RDMA [09:25:37] oh, but you mentioned it already, drv :) thanks for the tip though, will reconsider the code [09:37:33] *** Quits: tomzawadzki (~tomzawadz@192.55.54.42) (Ping timeout: 264 seconds) [11:33:57] *** Joins: travis-ci (~travis-ci@ec2-54-204-190-81.compute-1.amazonaws.com) [11:33:58] (spdk/master) spdkcli: remove print_array from bdev create calls (Karol Latecki) [11:33:58] Diff URL: https://github.com/spdk/spdk/compare/8c71c208a2c4...a648660be566 [11:33:58] *** Parts: travis-ci (~travis-ci@ec2-54-204-190-81.compute-1.amazonaws.com) () [11:55:37] *** Joins: alekseymmm (bcf3adf1@gateway/web/freenode/ip.188.243.173.241) [12:00:58] *** Quits: tkulasek_ (~tkulasek@134.134.139.75) (Ping timeout: 264 seconds) [12:25:14] *** Quits: philipp-sk (~Philipp@ktnron0916w-lp140-04-65-92-69-234.dsl.bell.ca) (Read error: Connection reset by peer) [12:25:38] *** Joins: philipp-sk (~Philipp@ktnron0916w-lp140-04-65-92-69-234.dsl.bell.ca) [13:07:24] *** Joins: travis-ci (~travis-ci@ec2-54-158-111-123.compute-1.amazonaws.com) [13:07:37] (spdk/master) bdev: encapsulate members of spdk_bdev_module (Seth Howell) [13:07:37] Diff URL: https://github.com/spdk/spdk/compare/a648660be566...2c52691b48d4 [13:07:37] *** Parts: travis-ci (~travis-ci@ec2-54-158-111-123.compute-1.amazonaws.com) () [13:19:33] Hello everyone . Is it true that now all loggs are -L not -t ? [13:25:17] I think we got them all, yes [13:25:25] -t was conflicting with tools that used it to mean "time" [13:32:07] ok [13:37:02] *** Joins: travis-ci (~travis-ci@ec2-54-144-70-104.compute-1.amazonaws.com) [13:37:03] (spdk/master) scripts/rpc: Add save_subsystem_config RPC (Shuhei Matsumoto) [13:37:03] Diff URL: https://github.com/spdk/spdk/compare/2c52691b48d4...800a14233291 [13:37:03] *** Parts: travis-ci (~travis-ci@ec2-54-144-70-104.compute-1.amazonaws.com) () [13:38:44] it could definitely use a changelog entry [14:02:18] https://review.gerrithub.io/#/c/spdk/spdk/+/416879: CI put -1 but the link to logs seems to be invalid. any thoughts? [14:34:10] sometimes it takes awhile for the logs to copy over to the public web server [14:34:17] I think at least [14:34:28] sethhowe would know [14:35:17] looking at the log on the build server itself, it shows that the patched failed astyle [14:35:24] so code format issue [14:35:29] run ./scripts/check_format.sh [14:36:18] I think the way the current system works is that the publish step that copies to ci.spdk.io happens right away, but if it fails, it is never retried [14:36:35] there is also a failure due to including the ibverbs headers on FreeBSD in that patch [14:38:17] phillip-sk: The copy step might have failed on this one. I'll see if I can't manually push it over. [14:39:29] bwalker, sethhowe - thanks, will looki into styling [14:40:05] phillip-sk: actually, it just looks like the wrong link was pasted into gerrithub. I think this is your review. https://ci.spdk.io/spdk/builds/review/3664e40ea61ee8210ca8dcf28e084341c3ed8ec9.1530046426/ [14:41:09] thanks, sethhowe [14:55:01] *** Joins: philipp-sk_ (~Mutter@ktnron0916w-lp140-04-65-92-69-234.dsl.bell.ca) [14:56:12] *** Quits: philipp-sk_ (~Mutter@ktnron0916w-lp140-04-65-92-69-234.dsl.bell.ca) (Client Quit) [14:59:31] *** Joins: philipp-sk_ (~Mutter@ktnron0916w-lp140-04-65-92-69-234.dsl.bell.ca) [14:59:42] *** Quits: philipp-sk_ (~Mutter@ktnron0916w-lp140-04-65-92-69-234.dsl.bell.ca) (Client Quit) [15:32:57] *** Quits: alekseymmm (bcf3adf1@gateway/web/freenode/ip.188.243.173.241) (Quit: Page closed) [15:54:42] *** Joins: travis-ci (~travis-ci@ec2-54-158-111-123.compute-1.amazonaws.com) [15:54:43] (spdk/master) nvme_spec: add VWC flush broadcast bits (TP 4035) (Daniel Verkamp) [15:54:43] Diff URL: https://github.com/spdk/spdk/compare/800a14233291...d4ef57c0724e [15:54:43] *** Parts: travis-ci (~travis-ci@ec2-54-158-111-123.compute-1.amazonaws.com) () [16:34:44] *** Joins: travis-ci (~travis-ci@ec2-54-205-136-216.compute-1.amazonaws.com) [16:34:45] (spdk/master) scsi: Use not pthread_self but spdk_get_thread for LUN affinity (Shuhei Matsumoto) [16:34:46] Diff URL: https://github.com/spdk/spdk/compare/d4ef57c0724e...322e50468cb7 [16:34:46] *** Parts: travis-ci (~travis-ci@ec2-54-205-136-216.compute-1.amazonaws.com) () [18:19:37] *** Quits: lyan (~lyan@2605:a000:160e:3c8:4a4d:7eff:fef2:eea3) (Quit: Leaving) [18:33:57] *** Joins: dlw (~Thunderbi@114.255.44.143) [19:07:33] *** Joins: darsto_ (~darsto@89-64-41-182.dynamic.chello.pl) [19:08:46] *** Quits: darsto (~darsto@89-68-131-200.dynamic.chello.pl) (Ping timeout: 264 seconds) [19:08:46] *** darsto_ is now known as darsto [19:28:52] *** Quits: philipp-sk (~Philipp@ktnron0916w-lp140-04-65-92-69-234.dsl.bell.ca) (Quit: Leaving) [19:29:22] *** Joins: philipp-sk (~Philipp@ktnron0916w-lp140-04-65-92-69-234.dsl.bell.ca)