[00:14:08] *** Joins: vmysak (~vmysak@192.55.54.39) [03:54:32] *** Joins: tomzawadzki (~tomzawadz@134.134.139.74) [07:16:07] *** Quits: vmysak (~vmysak@192.55.54.39) (Ping timeout: 252 seconds) [08:25:03] *** Quits: tomzawadzki (~tomzawadz@134.134.139.74) (Ping timeout: 245 seconds) [08:57:39] *** Joins: bluebird (~bluebird@p5DE941C4.dip0.t-ipconnect.de) [09:59:31] *** Joins: vmysak (~vmysak@192.55.54.38) [11:10:21] *** Quits: vmysak (~vmysak@192.55.54.38) (Remote host closed the connection) [11:10:29] *** Joins: vmysak (vmysak@nat/intel/x-hbogtvoqdqjqzjzb) [11:46:06] *** Quits: vmysak (vmysak@nat/intel/x-hbogtvoqdqjqzjzb) (Ping timeout: 272 seconds) [11:48:36] *** Joins: vmysak (~vmysak@192.55.54.38) [11:59:06] *** Quits: vmysak (~vmysak@192.55.54.38) (Ping timeout: 252 seconds) [12:02:26] *** Joins: Tracy35 (~Tracy35@12.218.82.130) [12:04:49] If spdk 18.07 is downloaded and "git submodule update --init" is run, will it pick dpdk 18.05 or the latest 18.08.x version? [12:15:00] *** Joins: vmysak (~vmysak@192.55.54.38) [12:36:46] *** Quits: vmysak (~vmysak@192.55.54.38) (Ping timeout: 272 seconds) [12:37:26] Tracy35: you can always type the command and check, but in my system is downloading 18.05 :) [12:41:32] pwodkowx: Thanks. Will do. [12:50:36] bwalker: I moved out to a new flat just recently and don't have that bad habit of working overtime just yet [12:50:42] but I'm here now if you need something [13:08:34] *** Joins: JC_intel (c0373629@gateway/web/freenode/ip.192.55.54.41) [13:38:01] sethhowe: As reported in https://github.com/spdk/spdk/issues/434#issuecomment-420799947 once I cherry-picked your patches, I was happy to report nvmf_tgt no longer segfaults. Reviewing your patches, I'm questioning: is there a need to retain both a LIBS and a SYS_LIBS? Can we simplify the build a little by consolidating that into one variable? [13:51:24] lhodev: I don't think we can, for the same reason that you don't pass $(filter-out -Wl$(comma)--no-whole-archive,$(LIBS)) as the first argument to the second call to spdk_build_realname_shared_lib [13:52:17] Hi, I'm new to SPDK/DPDK, I'm trying to run a DPDK example built from the SPDK sources but it doesn't find any ethernet device whereas it works when I use the app built in dpdk-stable [13:53:08] lhodev: Namely, LIBS contains a lot of linker information about libraries internal to spdk. I encountered a lot of linker issues when I tried to pass LIBS into that call. [13:54:18] JC_intel: Welcome! We'd be happy to help you spin up. Can you give a little more information about which SPDK example you are trying to run? [13:56:56] Digging deeper, it appears that actually your changes not only of course address the segfault problem, but also brings the use of LIBS vs SYS_LIBS into better distinction. Use LIBS for the naming SPDK libs and SYS_LIBS for those already installed on a system. Perhaps, serendipitously, your patch highlighted this conflation of use and now will make maintenance of the build easier. [13:57:15] I haven't revisited *all* of the instances, but it does seem a little more (if not entirely) more consistent. [13:57:50] *** Joins: vmysak (vmysak@nat/intel/x-okgmjdfiemqvsibr) [13:59:53] sethhowe: Thanks! [13:59:57] Yeah, I haven't dug all the way in either, but that seemed like the original intent of those two variables and increasing that distinction helped out in this use case. [14:00:24] Tracy35: it depends on what version of SPDK you have checked out. Right now on master it's 18.05 [14:01:02] sethhowe: Thanks! The SPDK examples seem to run fine (I wrote an app that is using SPDK nvme) but the DPDK examples from within the SPDK sources do not find any device... [14:02:23] JC_intel: I think if you build DPDK via the SPDK submodule, we don't actually build the NIC drivers [14:02:38] just because the SPDK stuff doesn't need them and it cuts the build time down [14:02:50] Oh, I see! [14:03:07] you can build DPDK on your own however you'd like and then tell SPDK to use your version instead of the submodule [14:03:14] ./configure --with-dpdk=/path/to/dpdk/build [14:03:24] So, I should try to link against my DPDK or rebuild the DPDK from the sources to get it compliant? [14:03:38] Thanks! [14:03:54] the submodule is really just for convenience [14:04:02] for those who want to run SPDK without messing with any DPDK stuff [14:05:39] I imagine that I should rebuild DPDK with gcc instead of ICC, I didn't see a SPDK icc option [14:05:58] we don't have built-in icc support for SPDK right now [14:06:01] we have gcc and clang [14:06:09] (we'd accept a patch to add it!) [14:06:41] No problem, thanks! [14:07:05] I think I'll work on it as soon as I'll have a working reference [14:15:47] *** Quits: vmysak (vmysak@nat/intel/x-okgmjdfiemqvsibr) (Ping timeout: 240 seconds) [15:02:27] *** Quits: Tracy35 (~Tracy35@12.218.82.130) (Ping timeout: 240 seconds) [15:11:48] *** Quits: bluebird (~bluebird@p5DE941C4.dip0.t-ipconnect.de) (Quit: Leaving) [16:19:59] Rebuilt DPDK for gcc and spdk using my dpdk build, all compiles like a charm but I can't get my application working... I rewrote my Makefile to specify the libs/includes/flags... and now it seems to work. [21:00:00] *** Joins: travis-ci (~travis-ci@ec2-54-163-173-183.compute-1.amazonaws.com) [21:00:01] (spdk/master) lvol: _spdk_add_lvs_to_list: break loop immediately if find (wuzhouhui) [21:00:01] Diff URL: https://github.com/spdk/spdk/compare/9637e17d2d43...8220121f4475 [21:00:01] *** Parts: travis-ci (~travis-ci@ec2-54-163-173-183.compute-1.amazonaws.com) () [23:46:05] *** Joins: tomzawadzki (~tomzawadz@134.134.139.75)