[00:45:27] *** Quits: dlw (~Thunderbi@114.255.44.143) (Ping timeout: 240 seconds) [00:47:39] *** Joins: sage_ (~quassel@2607:f298:5:101d:f816:3eff:fe21:1966) [00:53:12] *** Joins: tkulasek (~tkulasek@192.55.54.40) [00:54:57] *** Quits: darsto (~darsto@89-68-112-135.dynamic.chello.pl) (Ping timeout: 240 seconds) [01:00:07] *** Joins: darsto (~darsto@89-68-112-135.dynamic.chello.pl) [05:38:14] *** Joins: tkulasek_ (~tkulasek@134.134.139.76) [05:39:57] *** Quits: tkulasek (~tkulasek@192.55.54.40) (Ping timeout: 240 seconds) [06:30:46] *** Joins: tkulasek__ (~tkulasek@134.134.139.76) [06:31:29] *** Joins: pohly (~pohly@p54849CF1.dip0.t-ipconnect.de) [06:34:09] *** Quits: tkulasek_ (~tkulasek@134.134.139.76) (Ping timeout: 244 seconds) [06:40:39] Anyone around who might help me debug why app/vhost never responds to a JSON request? See https://github.com/spdk/spdk/issues/322 for details. [06:57:39] *** Quits: darsto (~darsto@89-68-112-135.dynamic.chello.pl) (Ping timeout: 268 seconds) [07:02:09] *** Joins: darsto (~darsto@89-68-112-135.dynamic.chello.pl) [07:57:30] *** Joins: johnmeneghini (~johnmeneg@pool-100-0-53-181.bstnma.fios.verizon.net) [07:59:50] @bwalker, are we done with https://review.gerrithub.io/#/c/spdk/spdk/+/410302/ ? [08:00:14] Can this patch get merged? [08:29:03] *** Quits: pohly (~pohly@p54849CF1.dip0.t-ipconnect.de) (Remote host closed the connection) [09:08:48] drv: scan-build didn't like part of the payload type patch: https://ci.spdk.io/spdk/builds/review/a3eabc68a3693416e5512b805af2805586157d76.1527821533/fedora-01/scan-build/report-6e79ad.html#EndPath [09:27:49] hmm, it likes it again in the next patch; I'll see if I can work around it [09:40:47] *** Quits: darsto (~darsto@89-68-112-135.dynamic.chello.pl) (Ping timeout: 256 seconds) [09:41:27] jimharris: did you see Jonas's comment on https://review.gerrithub.io/#/c/spdk/spdk/+/413146/ ? [09:45:36] *** Joins: darsto (~darsto@89-68-112-135.dynamic.chello.pl) [10:07:21] *** Quits: tkulasek__ (~tkulasek@134.134.139.76) (Ping timeout: 240 seconds) [10:12:07] hmm, nightly test on master has failed for the last two days - looks like a real issue introduced by a recent patch [10:12:24] last passing nightly is commit b530d23b8: https://ci.spdk.io/spdk/builds/review/b530d23b8541b4c274f9f546db4bb829e474eca6.1527682236/ [10:12:59] and vhost test start failing somewhere between that and commit 6f95ad0d8: https://ci.spdk.io/spdk/builds/review/6f95ad0d8120dbc2a8624c182451514b99e86065.1527854424/ [10:13:51] hm, sorry, those commits are from the gerrit review, not actually in master [10:14:02] but the dates should be enough to get a range for bisect [10:18:15] actual good-bad range is 47c7470d..40b6f761 [10:19:32] if I had to guess, it is probably 89236d8fc86a ("test/vhost:Remove vhost configuration file from live migration test.") [10:19:39] which removes the NoPci option [10:23:19] jimharris: I fixed the scan-build thing; I'd be interested in seeing if the overhead test results change at all on your system [10:23:29] my box has turbo enabled and doesn't get reproducible enough results to tell [10:25:56] can you send an e-mail to the mailing list about the nightly test failure? and ask Pawel N (pniedzwx) to take a look [10:26:42] yes [10:27:08] i'll pull your latest patch set, take out the PPC/wmb change, test for performance and then push the updated patchset [10:27:27] I just looked at the DPDK barriers, and I think you're actually right about the PPC rmb thing [10:27:39] rmb or wmb? [10:27:47] sorry, wmb [10:28:16] at least in DPDK 18.05, rte_wmb is lwsync on PPC64 [10:28:26] but I'll defer to the POWER experts on this one :) [10:29:34] i suspect that jonas is correct, but that dpdk is using rte_wmb instead of rte_io_wmb in a bunch of the drivers [10:32:21] although i'm looking at dpdk 18.02 - maybe this has all been magically fixed in 18.05 [10:33:42] in 18.05 at least, PPC rte_io_wmb == rte_wmb [10:34:02] I'm not sure which one is supposed to be more restrictive, though [10:58:59] drv: with your patches i see a small improvement (1.0-1.5ns) [11:00:32] nice [11:00:57] I was just talking with bwalker this morning - it might be interesting to try moving nvme_allocate_request to nvme_internal.h as an inline [11:01:20] I think we originally moved it out of line to nvme.c since it was using DPDK mempools at that time, and mempool allocation expands to a huge amount of code [11:01:32] but now that we're just pulling a request off a stailq, the code size increase should be much smaller [11:03:54] in addition to nvme_allocate_request_contig? [11:04:45] we don't actually use the non-SGL variants in the bdev driver, so we wouldn't have to move that function [11:05:16] sure, that's worth a try as well [11:05:52] hmmm - the overhead tool uses the contig variants [11:06:00] we should probably change that to use the SGL variant [11:06:51] _nvme_ns_cmd_rw() uses nvme_allocate_request() directly, which is probably the most important hot path [11:07:07] but the other allocate_request_* helpers are also small enough that they are fine to inline [11:07:28] i think i'll leave the user_copy one in the .c file [11:07:36] yeah, that one is fine to leave in the .c [11:07:40] it's not intended for performance paths [11:09:10] also nvme_complete_request [11:09:16] and nvme_free_request [11:09:29] yeah, definitely [11:11:25] ugh - g_pid [11:14:39] oh, is that currently static? [11:14:48] hmm [11:15:37] *** Joins: Clark__ (0cda5282@gateway/web/freenode/ip.12.218.82.130) [11:20:35] drv: To give you an update, keeping SPDK code base same, with older version i.e. Kernel 3.10.0-693+Mellanox 3.0-1, I do not see the memory allocation error. The error is seen with Kernel 4.16.8-1+Mellanox 5.0-0. Will spend some time to isolate it. Thanks [11:22:11] inlining those functions saves another 4-5ns [11:23:09] *** Quits: darsto (~darsto@89-68-112-135.dynamic.chello.pl) (Ping timeout: 256 seconds) [11:23:41] Clark__: hmm, sounds like it's related to the RNIC driver then - are you using the Mellanox OFED stack? [11:28:08] *** Joins: darsto (~darsto@89-68-112-135.dynamic.chello.pl) [12:08:34] *** Joins: johnmeneghini1 (~johnmeneg@pool-100-0-53-181.bstnma.fios.verizon.net) [12:10:42] *** Quits: johnmeneghini (~johnmeneg@pool-100-0-53-181.bstnma.fios.verizon.net) (Read error: Connection reset by peer) [12:10:44] *** Quits: ziyeyang (~ziyeyang@134.134.139.72) (Ping timeout: 244 seconds) [12:11:12] *** Joins: ziyeyang (ziyeyang@nat/intel/x-ctrqofxmnulahtnr) [12:13:35] *** Quits: lhodev (~lhodev@66-90-218-190.dyn.grandenetworks.net) (Ping timeout: 256 seconds) [12:15:10] drv: Correction - The version which is working is Kernel 3.10.0-693+Mellanox OFED 4.3-1.0.1 (and not Mellanox 3.0-1 as mentioned before - got this version wrongly from the client side). The error which is seen with Kernel 4.16.8-1+Mellanox 5.0-0, this Mellanox driver is built in Kernel 4.16.8-1. [12:50:58] *** Quits: Clark__ (0cda5282@gateway/web/freenode/ip.12.218.82.130) (Quit: Page closed) [13:17:49] *** Joins: travis-ci (~travis-ci@ec2-50-19-20-167.compute-1.amazonaws.com) [13:17:50] (spdk/master) event/app: fix setting "single-file-segments" (Dariusz Stojaczyk) [13:17:50] Diff URL: https://github.com/spdk/spdk/compare/51806fa7b64c...f8387acaa5f6 [13:17:51] *** Parts: travis-ci (~travis-ci@ec2-50-19-20-167.compute-1.amazonaws.com) () [13:18:42] *** Joins: travis-ci (~travis-ci@ec2-50-19-20-167.compute-1.amazonaws.com) [13:18:43] (spdk/master) include/bdev.h: add comments for callback functions (Yanbo Zhou) [13:18:43] Diff URL: https://github.com/spdk/spdk/compare/f8387acaa5f6...cfbae9c23b74 [13:18:43] *** Parts: travis-ci (~travis-ci@ec2-50-19-20-167.compute-1.amazonaws.com) () [13:19:42] *** Joins: travis-ci (~travis-ci@ec2-54-211-92-252.compute-1.amazonaws.com) [13:19:43] (spdk/master) subsystem/iscsi: Add iSCSI options to JSON config file (Shuhei Matsumoto) [13:19:44] Diff URL: https://github.com/spdk/spdk/compare/cfbae9c23b74...fafec18e3c33 [13:19:44] *** Parts: travis-ci (~travis-ci@ec2-54-211-92-252.compute-1.amazonaws.com) () [13:24:30] Thanks for approving patch https://review.gerrithub.io/#/c/spdk/spdk/+/410302/ ! [13:58:31] *** Joins: lhodev (~lhodev@inet-hqmc07-o.oracle.com) [14:00:33] *** Quits: lhodev (~lhodev@inet-hqmc07-o.oracle.com) (Remote host closed the connection) [14:04:22] *** Joins: travis-ci (~travis-ci@ec2-54-211-92-252.compute-1.amazonaws.com) [14:04:23] (spdk/master) nvme: optimize nvme_allocate_request memsets (Jim Harris) [14:04:23] Diff URL: https://github.com/spdk/spdk/compare/fafec18e3c33...d76cd984aafd [14:04:23] *** Parts: travis-ci (~travis-ci@ec2-54-211-92-252.compute-1.amazonaws.com) () [14:04:28] *** Joins: lhodev (~lhodev@inet-hqmc07-o.oracle.com) [14:06:30] *** Quits: lhodev (~lhodev@inet-hqmc07-o.oracle.com) (Remote host closed the connection) [14:07:02] *** Joins: lhodev (~lhodev@inet-hqmc07-o.oracle.com) [14:09:03] *** Quits: lhodev (~lhodev@inet-hqmc07-o.oracle.com) (Remote host closed the connection) [14:09:34] *** Joins: lhodev (~lhodev@inet-hqmc07-o.oracle.com) [14:11:36] *** Quits: lhodev (~lhodev@inet-hqmc07-o.oracle.com) (Remote host closed the connection) [14:12:07] *** Joins: lhodev (~lhodev@inet-hqmc07-o.oracle.com) [14:14:09] *** Quits: lhodev (~lhodev@inet-hqmc07-o.oracle.com) (Remote host closed the connection) [14:14:42] *** Joins: lhodev (~lhodev@inet-hqmc07-o.oracle.com) [14:16:45] *** Quits: lhodev (~lhodev@inet-hqmc07-o.oracle.com) (Remote host closed the connection) [14:17:16] *** Joins: lhodev (~lhodev@inet-hqmc07-o.oracle.com) [14:19:18] *** Quits: lhodev (~lhodev@inet-hqmc07-o.oracle.com) (Remote host closed the connection) [14:19:50] *** Joins: lhodev (~lhodev@inet-hqmc07-o.oracle.com) [14:21:51] *** Quits: lhodev (~lhodev@inet-hqmc07-o.oracle.com) (Remote host closed the connection) [14:25:42] *** Joins: lhodev (~lhodev@inet-hqmc07-o.oracle.com) [14:27:45] *** Quits: lhodev (~lhodev@inet-hqmc07-o.oracle.com) (Remote host closed the connection) [14:28:17] *** Joins: lhodev (~lhodev@inet-hqmc07-o.oracle.com) [14:30:18] *** Quits: lhodev (~lhodev@inet-hqmc07-o.oracle.com) (Remote host closed the connection) [14:37:29] *** Joins: lhodev (~lhodev@inet-hqmc07-o.oracle.com) [14:39:31] *** Quits: lhodev (~lhodev@inet-hqmc07-o.oracle.com) (Remote host closed the connection) [14:40:02] *** Joins: lhodev (~lhodev@inet-hqmc07-o.oracle.com) [14:42:04] *** Quits: lhodev (~lhodev@inet-hqmc07-o.oracle.com) (Remote host closed the connection) [14:42:35] *** Joins: lhodev (~lhodev@inet-hqmc07-o.oracle.com) [14:44:37] *** Quits: lhodev (~lhodev@inet-hqmc07-o.oracle.com) (Remote host closed the connection) [14:45:08] *** Joins: lhodev (~lhodev@inet-hqmc07-o.oracle.com) [14:47:10] *** Quits: lhodev (~lhodev@inet-hqmc07-o.oracle.com) (Remote host closed the connection) [14:47:40] *** Joins: lhodev (~lhodev@inet-hqmc07-o.oracle.com) [14:49:43] *** Quits: lhodev (~lhodev@inet-hqmc07-o.oracle.com) (Remote host closed the connection) [14:50:14] *** Joins: lhodev (~lhodev@inet-hqmc07-o.oracle.com) [14:52:16] *** Quits: lhodev (~lhodev@inet-hqmc07-o.oracle.com) (Remote host closed the connection) [14:52:48] *** Joins: lhodev (~lhodev@inet-hqmc07-o.oracle.com) [14:54:50] *** Quits: lhodev (~lhodev@inet-hqmc07-o.oracle.com) (Remote host closed the connection) [14:55:20] *** Joins: lhodev (~lhodev@inet-hqmc07-o.oracle.com) [14:57:23] *** Quits: lhodev (~lhodev@inet-hqmc07-o.oracle.com) (Remote host closed the connection) [14:57:54] *** Joins: lhodev (~lhodev@inet-hqmc07-o.oracle.com) [14:59:55] *** Quits: lhodev (~lhodev@inet-hqmc07-o.oracle.com) (Remote host closed the connection) [15:00:26] *** Joins: lhodev (~lhodev@inet-hqmc07-o.oracle.com) [15:02:02] *** Quits: darsto (~darsto@89-68-112-135.dynamic.chello.pl) (Ping timeout: 244 seconds) [15:07:07] *** Joins: darsto (~darsto@89-68-112-135.dynamic.chello.pl) [16:07:41] *** Joins: DT1 (c6ba0105@gateway/web/freenode/ip.198.186.1.5) [16:08:57] Have a question on nvmf initiator mode connecting to discovery controller. [16:15:45] anyone available? [16:19:13] *** Quits: DT1 (c6ba0105@gateway/web/freenode/ip.198.186.1.5) (Quit: Page closed) [20:26:14] *** Quits: johnmeneghini1 (~johnmeneg@pool-100-0-53-181.bstnma.fios.verizon.net) (Quit: Leaving.) [21:09:45] *** Quits: darsto (~darsto@89-68-112-135.dynamic.chello.pl) (Ping timeout: 260 seconds) [21:14:35] *** Joins: darsto (~darsto@89-68-112-135.dynamic.chello.pl) [21:46:02] *** Quits: darsto (~darsto@89-68-112-135.dynamic.chello.pl) (Ping timeout: 245 seconds) [21:51:09] *** Joins: darsto (~darsto@89-68-112-135.dynamic.chello.pl)