[04:50:31] *** Joins: darsto_ (~darsto@89-78-174-111.dynamic.chello.pl) [04:51:36] *** Quits: darsto (~darsto@89-78-174-111.dynamic.chello.pl) (Ping timeout: 252 seconds) [04:51:36] *** darsto_ is now known as darsto [05:58:41] *** Quits: peluse (peluse@nat/intel/x-spaviogdwvvmbqoi) (Quit: ZNC - http://znc.in) [07:13:56] *** Joins: pwodkowx (pwodkowx@nat/intel/x-hxlwzazugwreszoi) [07:39:22] jimharris: about bdev_crypto "it requires a patch that is in our dpdk submodule" [07:41:23] This patch only allow switching on '-fPIC' during build. I'm not sure whay we need this. [07:41:35] *** Joins: sethhowe (sethhowe@nat/intel/x-ckvnvdluetfbmkpr) [07:43:35] Are we are alread building all objects with '-fPIC' [07:51:38] and DPDK shared libraries are also sompiled with fPIC [07:51:56] So, is there any issue with this? [07:55:47] pwodkowx: b724ebb commit in our dpdk submodule [08:07:43] *** Joins: travis-ci (~travis-ci@ec2-23-23-26-127.compute-1.amazonaws.com) [08:07:44] (spdk/master) scripts/spdkcli: override configshell init (Karol Latecki) [08:07:44] Diff URL: https://github.com/spdk/spdk/compare/99ca58e0827b...ad06d03a8666 [08:07:44] *** Parts: travis-ci (~travis-ci@ec2-23-23-26-127.compute-1.amazonaws.com) () [08:14:23] Hi pwodkowx: I believe the decision to specify -fPIC by default was put in place to ease the ability for the generation of shared libs, if desired, before we formally began added shared lib support to the individual libraries. This made it easy for Daniel to build a single shared lib from all the static ones months ago. [08:14:54] Now with the new '--with-shared' option, we could revisit the build code such that it's only activated when that's specified. [08:15:50] jimharris, lhodev_: thank for explanation :) I'm asking in face of SPDK vs DPDK master test as I'm trying to fix this job [08:15:52] Curious: are you concerned that for the case when shared libs aren't build, that we're incurring a significant performance penalty with -fPIC ? [08:16:49] I checked dpdk commit b724ebb and I think I have too small mind to understan this :) [08:17:12] for me most of this patch (in context of bdev_crypto) is a NOP [08:17:51] check with peluse - if we really don't need it we should remove it from our dpdk submodule - but i'm fairly certain it's required [08:18:32] first two ifdefs are NOP for sure [08:18:48] I need to understand last one :) [08:20:10] i don't think the first two are NOP - without these patches, there is no way to specify separate source and destination buffers [08:20:20] so dpdk crypto will always just encrypt/decrypt in place [08:20:41] that doesn't work well for storage - especially vhost :) [08:20:50] *** Quits: lhodev_ (~lhodev@66-90-218-190.dyn.grandenetworks.net) (Quit: Textual IRC Client: www.textualapp.com) [08:21:15] *** Joins: lhodev (~lhodev@inet-hqmc06-o.oracle.com) [08:59:24] *** Joins: travis-ci (~travis-ci@ec2-54-211-21-201.compute-1.amazonaws.com) [08:59:25] (spdk/master) crypto: Use bulk API to return mbufs and crypto ops in poller (paul luse) [08:59:25] Diff URL: https://github.com/spdk/spdk/compare/ad06d03a8666...7bcdc55926a3 [08:59:25] *** Parts: travis-ci (~travis-ci@ec2-54-211-21-201.compute-1.amazonaws.com) () [09:19:28] *** Joins: travis-ci (~travis-ci@ec2-54-221-129-99.compute-1.amazonaws.com) [09:19:29] (spdk/master) scripts/vagrant: Add fc28 image and update docs. (Seth Howell) [09:19:29] Diff URL: https://github.com/spdk/spdk/compare/7bcdc55926a3...b7ceb7508fa4 [09:19:29] *** Parts: travis-ci (~travis-ci@ec2-54-221-129-99.compute-1.amazonaws.com) () [11:11:50] *** Joins: peluse (~peluse@134.134.139.72) [11:11:50] *** ChanServ sets mode: +o peluse [11:34:55] jimharris: I see. But, as I have task to revive SPDK vs DPDK master tests the question is: disable the bdev crypto tests, patch DPDK Paul will prepare copy-mode for bdev crypto? [11:35:55] * patch DPDK or Paul will prepare copy-mode for bdev crypto? [11:38:56] for SPDK vs. DPDK master tests - I think we'll need to disable crypto tests [11:42:55] ok [11:42:59] oh [20:32:56] *** Quits: lhodev (~lhodev@inet-hqmc06-o.oracle.com) (Remote host closed the connection) [20:33:31] *** Joins: lhodev (~lhodev@66-90-218-190.dyn.grandenetworks.net)