[00:04:02] *** Quits: darsto (~darsto@89-68-115-185.dynamic.chello.pl) (Ping timeout: 255 seconds) [00:07:01] *** Joins: darsto (~darsto@89-68-115-185.dynamic.chello.pl) [00:49:03] *** Joins: tomzawadzki (~tomzawadz@134.134.139.72) [00:52:49] *** Joins: tkulasek (~tkulasek@134.134.139.72) [01:08:39] pwodkowx, anyone: quick question regarding the JSON API - do both "remove_vhost_scsi_target" and "add_vhost_scsi_lun" remove/add SCSI targets, despite the "lun" in "add_vhost_scsi_lun"? [01:08:59] Looks like "add_vhost_scsi_lun" is misnamed, right? [01:09:14] Internally it calls spdk_vhost_scsi_dev_add_tgt(). [01:19:46] *** Joins: tkulasek_ (~tkulasek@192.55.54.42) [01:22:21] *** Quits: tkulasek (~tkulasek@134.134.139.72) (Ping timeout: 240 seconds) [01:23:30] *** Quits: Shuhei (caf6fc61@gateway/web/freenode/ip.202.246.252.97) (Ping timeout: 260 seconds) [01:26:58] pohly: we support only LUN per target [01:27:32] so add tgt is equivalent of add LUN. [01:28:27] if we ever support more than one LUN per target we will have to split this into spdk_vhost_scsi_dev_add_tgt and spdk_vhost_scsi_dev_add_lun [01:52:00] pwodkowx: thanks for the confirmation. Is there a maximum number of targets per controller? [01:52:52] pohly: yes, current default is 8. [01:53:21] can be changed -> vhost_internal.h -> #define SPDK_VHOST_SCSI_CTRLR_MAX_DEVS 8 [01:57:54] pwodkowx: when a VHost SCSI controller is constructed without CPU mask, what CPU mask should it get? [01:58:24] I started vhost without CPU mask, and a new controller ended up with CPU mask 0x1. Is that alright? [02:15:32] *** Joins: bwalker (~bwalker@134.134.139.72) [02:15:32] *** Server sets mode: +cnt [03:42:45] pohly: it will use the same CPU mask as APP [03:43:37] * as APP cpu mask. [03:46:05] *** Quits: dlw (~Thunderbi@114.255.44.143) (Ping timeout: 240 seconds) [04:28:36] pwodkowx: and if the app has no CPU mask set? Is 0x1 the default for that? [04:28:49] pohly: correction - it will choose one core from APP cpu mask. [04:28:53] yes [04:31:18] there is some logic behind this [04:31:27] see http://www.spdk.io/doc/vhost.html#vhost_start [05:45:25] *** Joins: tim___ (b649cc4a@gateway/web/freenode/ip.182.73.204.74) [06:02:36] *** Joins: dlw (~Thunderbi@123.122.66.130) [06:37:24] I seem to have fixed my IRC disconnect issue [06:37:27] lucky me! hooray! [06:37:51] after hours of debugging, a znc update did the trick [07:27:05] *** Quits: dlw (~Thunderbi@123.122.66.130) (Ping timeout: 260 seconds) [08:12:57] oh geeze [08:22:16] drv: can you take a look at these patches from bwalker again? https://review.gerrithub.io/#/c/spdk/spdk/+/414699/ [08:22:22] ok [08:22:33] i think we should deprecate io_channel.h but we can do that in a later patch [08:22:49] i'd like to get these changes in since my spdk_thread_get_count patch will need to be rebased [08:22:50] oh yeah, I agree we should put something in changelog [08:23:05] but it definitely can be another patch (just have to make sure bwalker remembers) [08:23:45] ok, I re-+2'd those two [08:27:11] *** Joins: travis-ci (~travis-ci@ec2-54-205-60-99.compute-1.amazonaws.com) [08:27:12] (spdk/master) thread: Replace #include of io_channel.h with thread.h (Ben Walker) [08:27:12] Diff URL: https://github.com/spdk/spdk/compare/9f697e0fe24a...a83f91c29a47 [08:27:12] *** Parts: travis-ci (~travis-ci@ec2-54-205-60-99.compute-1.amazonaws.com) () [09:04:38] *** ChanServ sets mode: +o bwalker [09:14:24] I may throw this up onto the web page for use in documentation: https://roughjs.com/ [09:14:26] what do you guys think [09:15:54] we need some way to draw diagrams, basically [09:16:11] there are lots of choices, but I thought making them look hand-drawn was neat [09:16:43] I'm looking at this as a better way to solve https://review.gerrithub.io/#/c/spdk/spdk/+/414816/ [09:17:49] i like the idea - could we ask tkulasek_ to try modifying his diagram to use it so we can see a real-world SPDK example? [09:18:03] bwalker: could you take a look at my bsdump patches again? https://review.gerrithub.io/#/c/spdk/spdk/+/414479/ [09:18:12] I think that sounds great - I'm going to just do a quick overview on the best available options and then I'll ask him to [09:21:39] done [09:30:38] bwalker, looks pretty cool. [09:36:05] *** Quits: tomzawadzki (~tomzawadz@134.134.139.72) (Ping timeout: 256 seconds) [09:40:36] *** Joins: travis-ci (~travis-ci@ec2-54-205-60-99.compute-1.amazonaws.com) [09:40:37] (spdk/master) test/rocksdb: run blobcli -D after test is complete (Jim Harris) [09:40:37] Diff URL: https://github.com/spdk/spdk/compare/a83f91c29a47...5f0c61a589fa [09:40:37] *** Parts: travis-ci (~travis-ci@ec2-54-205-60-99.compute-1.amazonaws.com) () [09:42:06] *** Joins: travis-ci (~travis-ci@ec2-54-90-168-129.compute-1.amazonaws.com) [09:42:07] (spdk/master) bdev/qos: add the bandwidth rate limit control (GangCao) [09:42:07] Diff URL: https://github.com/spdk/spdk/compare/5f0c61a589fa...7191c4bdfbb0 [09:42:07] *** Parts: travis-ci (~travis-ci@ec2-54-90-168-129.compute-1.amazonaws.com) () [09:42:37] bwalker: could you add a CHANGELOG entry for deprecating io_channel.h, and add a #warning to io_channel.h that it will be removed? [09:43:18] sure [09:43:52] ok, so I can definitely do the rough.js thing for most of the web page, but I can't embed javascript into doxygen output [09:43:55] so that idea is dead in the water [09:44:17] because the diagrams are most useful for things in doc/, which are generated by doxygen [09:50:16] so I am absolutely not a Doxygen nor javascript expert - so this is probably a really stupid question... [09:50:35] ...but does the doxygen htmlonly tag help? [09:50:52] that's what I'm looking at, and it does work for "regular" doxygen docs [09:51:04] but I can't find a way to put the @htmlonly into a markdown file [09:51:22] ah [09:52:00] I'm going to do an experiment actually [09:53:14] i assume you're already seen https://www.stack.nl/~dimitri/doxygen/manual/markdown.html#md_html_blocks [09:53:50] yeah - that's describing how html blocks get translated [09:53:56] you can use text in your markdown [09:54:13] but it strips the , changes it to bold, then re-generates HTML or latex as necesasry [09:54:32] I may be able to get this @htmlonly thing to work [09:54:43] the diagrams would only work for the HTML version of the documentation [09:54:46] but that may be ok [10:00:54] I'm not aware of any, surely they're out there though... [10:01:43] I've got it working! [10:01:56] still having issues with QAT AEAD auth even following the advice from DPDK folks, going to try the same setup but with the SW driver so its easier to see what the heck is going on. Would like to get that working before tackling the larger issues [10:02:01] just putting together a patch of the infrastructure, then I'll tell them to try to draw their svg's using this [10:02:16] bwalker, sweet! [10:32:45] drv: can you take a look at this one again? https://review.gerrithub.io/#/c/spdk/spdk/+/411842/ [10:43:38] *** Joins: travis-ci (~travis-ci@ec2-54-205-60-99.compute-1.amazonaws.com) [10:43:39] (spdk/master) nvme: add PI to spdk_nvme_ns_cmd_read/writev() (Xiaodong Liu) [10:43:39] Diff URL: https://github.com/spdk/spdk/compare/7191c4bdfbb0...5f1c1189fade [10:43:39] *** Parts: travis-ci (~travis-ci@ec2-54-205-60-99.compute-1.amazonaws.com) () [10:44:37] *** Joins: travis-ci (~travis-ci@ec2-54-90-168-129.compute-1.amazonaws.com) [10:44:38] (spdk/master) copy/ioat: Change the default of IOAT from Enabled to Disabled (Shuhei Matsumoto) [10:44:38] Diff URL: https://github.com/spdk/spdk/compare/5f1c1189fade...239e63be7d15 [10:44:38] *** Parts: travis-ci (~travis-ci@ec2-54-90-168-129.compute-1.amazonaws.com) () [11:04:37] *** Joins: travis-ci (~travis-ci@ec2-54-90-168-129.compute-1.amazonaws.com) [11:04:38] (spdk/master) nvme: add Get/Set Feature wrappers with NSID (Jakub Radtke) [11:04:38] Diff URL: https://github.com/spdk/spdk/compare/c73908316973...b3497d5da77f [11:04:38] *** Parts: travis-ci (~travis-ci@ec2-54-90-168-129.compute-1.amazonaws.com) () [11:06:17] *** Joins: travis-ci (~travis-ci@ec2-54-205-60-99.compute-1.amazonaws.com) [11:06:17] (spdk/master) bdev/iscsi: add poller when there is no master channel (Pawel Wodkowski) [11:06:18] Diff URL: https://github.com/spdk/spdk/compare/b3497d5da77f...bd52bfecb5ec [11:06:18] *** Parts: travis-ci (~travis-ci@ec2-54-205-60-99.compute-1.amazonaws.com) () [11:53:09] *** Quits: tkulasek_ (~tkulasek@192.55.54.42) (Ping timeout: 264 seconds) [12:25:07] *** Quits: pohly (~pohly@p54BD5098.dip0.t-ipconnect.de) (Quit: Leaving.) [12:40:22] *** Joins: travis-ci (~travis-ci@ec2-54-90-168-129.compute-1.amazonaws.com) [12:40:22] (spdk/master) bdev/nvme: remove redundant check for active NS (Changpeng Liu) [12:40:23] Diff URL: https://github.com/spdk/spdk/compare/bd52bfecb5ec...a52b35069ac1 [12:40:23] *** Parts: travis-ci (~travis-ci@ec2-54-90-168-129.compute-1.amazonaws.com) () [13:06:02] *** Joins: travis-ci (~travis-ci@ec2-54-90-168-129.compute-1.amazonaws.com) [13:06:03] (spdk/master) nvmf: Add trace points for the RDMA state machine (Ben Walker) [13:06:03] Diff URL: https://github.com/spdk/spdk/compare/a52b35069ac1...6a5ae72b4734 [13:06:03] *** Parts: travis-ci (~travis-ci@ec2-54-90-168-129.compute-1.amazonaws.com) () [13:10:24] jimharris: I responded to your comment on https://review.gerrithub.io/#/c/spdk/spdk/+/414687/ [13:27:40] *** Joins: travis-ci (~travis-ci@ec2-54-205-60-99.compute-1.amazonaws.com) [13:27:41] (spdk/master) nvmf: allow setting namespace UUID in config file (Daniel Verkamp) [13:27:41] Diff URL: https://github.com/spdk/spdk/compare/c06345bfafa5...194d6af65db4 [13:27:41] *** Parts: travis-ci (~travis-ci@ec2-54-205-60-99.compute-1.amazonaws.com) () [13:28:22] *** Joins: travis-ci (~travis-ci@ec2-54-205-60-99.compute-1.amazonaws.com) [13:28:23] (spdk/master) doc/vhost: add SVG diagram (Dariusz Stojaczyk) [13:28:23] Diff URL: https://github.com/spdk/spdk/compare/6a5ae72b4734...c06345bfafa5 [13:28:23] *** Parts: travis-ci (~travis-ci@ec2-54-205-60-99.compute-1.amazonaws.com) () [13:30:19] jimharris: what do you think of that rough.js example? [13:30:38] I don't see a good way to put text onto the diagrams, so now I'm concerned it isn't as good as I had hoped [13:55:46] *** Joins: travis-ci (~travis-ci@ec2-54-156-70-11.compute-1.amazonaws.com) [13:55:47] (spdk/master) iscsi: Correct return code values of bdev iscsi pollers (Vishal Verma) [13:55:47] Diff URL: https://github.com/spdk/spdk/compare/194d6af65db4...9380f06bdcc7 [13:55:47] *** Parts: travis-ci (~travis-ci@ec2-54-156-70-11.compute-1.amazonaws.com) () [14:16:28] *** Joins: travis-ci (~travis-ci@ec2-54-90-168-129.compute-1.amazonaws.com) [14:16:29] (spdk/master) nvmf/rpc: ze-out name array in old format (Pawel Wodkowski) [14:16:29] Diff URL: https://github.com/spdk/spdk/compare/9380f06bdcc7...3b0759d955f3 [14:16:29] *** Parts: travis-ci (~travis-ci@ec2-54-90-168-129.compute-1.amazonaws.com) () [15:40:34] adding comments to bdev.h to specify possible errnos that can be returned: [15:40:35] http://spdk.intel.com/public/spdk/builds/review/ac04835ca4ddbaf71111efba00130a581ad2e7bc.1528842589/doc/html/bdev_8h.html#a4b500ce84df1d2551f76e635c9dafbd5 [15:40:46] formatting look OK? [15:46:31] yeah, looks OK to me [15:51:01] *** Joins: Shuhei (caf6fc61@gateway/web/freenode/ip.202.246.252.97) [15:53:43] why does spdk_bdev_write_zeroes_blocks() have this -ERANGE errno? [15:55:20] i don't think it's needed - the spdk_bdev_io_valid_blocks() call right afterwards effectively does the same check [15:55:31] i'd ask seth but he's not here :) [16:06:38] the first one is to make sure the len = spdk_bdev_get_block_size(bdev) * num_blocks multiplication doesn't overflow, I think [16:07:13] but it probably should get caught by spdk_bdev_io_valid_blocks() too [16:07:30] might have become redundant at some point when we did the _blocks refactoring, not sure [16:07:53] yeah - if it's really needed here then wouldn't it also be needed in read and write and unmap? [16:08:28] it should probably be validated at bdev registration time, if anything [16:08:38] I think we can probably remove it [16:10:05] bwalker: can you look at https://review.gerrithub.io/#/c/spdk/spdk/+/414709/ [16:10:16] then i can get the first three patches in this series merged [16:10:25] does SPDK NVMeOF initiator send a host NQN? [16:11:22] write_zeroes was also not checking desc->write [16:12:59] vermavis: yes, there's a default one based on randomly-generated UUID, or you can override it [16:13:28] OK thanks! [16:14:02] we don't have the hostnqn config piped through into the bdev layer or test apps, but the NVMe library does have hostnqn in struct spdk_nvme_ctrlr_opts [16:14:12] so actually overriding it will take some work :) [16:16:35] Ok got it. Joe was trying to run SPDK initiator w/ Attala and couldn't run with SPDK NVMeOF initiator because Attala target expects initiator to send hostNQN. So, I was wondering about that [16:17:53] *** Quits: guerby (~guerby@april/board/guerby) (Write error: Connection reset by peer) [16:18:04] *** Joins: guerby (~guerby@ip165.tetaneutral.net) [16:18:04] *** Quits: guerby (~guerby@ip165.tetaneutral.net) (Changing host) [16:18:04] *** Joins: guerby (~guerby@april/board/guerby) [16:18:09] *** Joins: travis-ci (~travis-ci@ec2-54-90-168-129.compute-1.amazonaws.com) [16:18:10] (spdk/master) Add copy module dependencies to Makefiles (Daniel Verkamp) [16:18:10] Diff URL: https://github.com/spdk/spdk/compare/3b0759d955f3...9c577fe7c022 [16:18:10] *** Parts: travis-ci (~travis-ci@ec2-54-90-168-129.compute-1.amazonaws.com) () [16:18:43] *** Quits: guerby (~guerby@april/board/guerby) (Read error: Connection reset by peer) [16:18:54] *** Joins: guerby (~guerby@april/board/guerby) [16:19:36] vermavis: if you wanted to do a quick hack, you can change the line in nvme_ctrlr.c that sets the default hostnqn to whatever they're looking for [16:19:38] snprintf(opts->hostnqn, sizeof(opts->hostnqn), "2014-08.org.nvmexpress:uuid:%s", host_id_str); [16:19:52] but we should probably add configuration options for that anyway [16:20:35] OK maybe I can create a patch for it later. But, first let me tell them this hack. Thanks! [17:27:25] drv: thank you for reviewing my copy ioat patch. Carelessly I overlooked error of format. Will you review again when you can? [17:41:36] Probably I didn't overlook but didn't run format check, but anyway I fixed it. Thank you for your review. [17:43:29] *** Joins: travis-ci (~travis-ci@ec2-54-90-168-129.compute-1.amazonaws.com) [17:43:30] (spdk/master) test/bdev: use DEFINE_STUB for spdk_conf_* stubs (Jim Harris) [17:43:31] Diff URL: https://github.com/spdk/spdk/compare/9c577fe7c022...34de1542eb41 [17:43:31] *** Parts: travis-ci (~travis-ci@ec2-54-90-168-129.compute-1.amazonaws.com) () [18:58:44] *** Joins: dlw (~Thunderbi@114.255.44.143) [19:15:38] *** Quits: dlw (~Thunderbi@114.255.44.143) (Remote host closed the connection) [19:15:55] *** Joins: dlw (~Thunderbi@114.255.44.143) [20:41:24] *** Quits: Shuhei (caf6fc61@gateway/web/freenode/ip.202.246.252.97) (Ping timeout: 260 seconds) [22:31:16] *** Parts: tim___ (b649cc4a@gateway/web/freenode/ip.182.73.204.74) () [22:32:00] *** Joins: rick_ (b649cc4a@gateway/web/freenode/ip.182.73.204.74) [23:09:54] Hey guys, i have compiled spdk master. I am trying to run the hello_world example in examples/nvme/hello_world directory. I am getting the following error [23:10:08] no NVMe controllers found. [23:10:20] what am i doing wrong ? [23:10:53] here is the link to pastebin https://pastebin.com/hXH1GHaM [23:22:04] *** Joins: pohly (~pohly@p54BD5098.dip0.t-ipconnect.de) [23:24:22] *** Joins: Shuhei (caf6fc61@gateway/web/freenode/ip.202.246.252.97) [23:35:02] rick_: did you unbind your NVMe from the kernel driver? [23:35:40] You can use ./scripts/setup.sh script to do that [23:37:02] It doesn't need any arguments, but you can refer to ./scripts/setup.sh -h for some extra configuration [23:39:56] yeah i ran the script. It said "Active mountpoints on /dev/nvme0n1, so not binding PCI dev 0000:00:04.0" [23:41:23] yeah - that's a security check. we don't want to unbind an NVMe that holds your OS [23:41:31] or other previous data [23:42:01] precious* [23:44:21] unbinding a drive from the kernel driver means that it disappears from lsblk [23:44:53] so your only option is to umount all mountpoints from that drive [23:54:47] *** Quits: darsto (~darsto@89-68-115-185.dynamic.chello.pl) (Quit: /quit) [23:55:05] *** Joins: darsto (~darsto@89-68-115-185.dynamic.chello.pl)