[06:54:25] *** Joins: krk_ (~krk@2601:644:8502:4820:a165:9576:ca8:1829) [07:20:12] *** Quits: krk_ (~krk@2601:644:8502:4820:a165:9576:ca8:1829) (Remote host closed the connection) [07:26:52] *** Quits: peluse_ (~peluse@ip72-201-144-44.ph.ph.cox.net) (Remote host closed the connection) [07:27:18] *** Joins: krk_ (~krk@2601:644:8502:4820:a165:9576:ca8:1829) [07:31:40] *** Quits: krk_ (~krk@2601:644:8502:4820:a165:9576:ca8:1829) (Ping timeout: 246 seconds) [08:20:50] *** Joins: krk_ (~krk@2601:644:8502:4820:a165:9576:ca8:1829) [09:33:01] drv, sethhowe: can you check the two pool.py patches I just pushed out for review? [09:34:38] *** Quits: krk_ (~krk@2601:644:8502:4820:a165:9576:ca8:1829) (Remote host closed the connection) [09:35:24] *** Joins: krk_ (~krk@74.217.79.250) [09:41:48] Will do! [10:20:30] *** Joins: jstern_ (~jstern@134.134.139.73) [11:47:30] *** Joins: ed_ (d8f01e19@gateway/web/freenode/ip.216.240.30.25) [11:47:49] *** Quits: ed_ (d8f01e19@gateway/web/freenode/ip.216.240.30.25) (Client Quit) [11:48:27] *** Joins: edwinrodriguez (d8f01e19@gateway/web/freenode/ip.216.240.30.25) [11:56:14] the pool agents are automatically updated, right? [11:56:25] so it's safe for me to push my dpdk submodule patch again? [12:11:33] is it possible to use uio_pci_generic with IOMMU enabled? i'm not following this setup.sh script change from karol [12:59:51] *** Quits: krk_ (~krk@74.217.79.250) (Ping timeout: 258 seconds) [13:17:51] I don't think uio works with iommu enabled [13:18:08] jimharris: can you take a look at the LTO patch set? that adds compiler detection that you can use in your DPDK submodule patch [13:29:01] done - but seems trying to call this shell script from within the makefile isn't much better than what i'm currently doing? [13:33:52] well, it detects the compiler type by looking at the version string, not the name of the binary [13:34:06] so if you are running /usr/bin/cc, that could in theory be clang or gcc [13:37:59] fair enough - so you're suggesting running this script from within the makefile, put the output into some makefile variable and then do regexp magic on it to determine if its gcc or clang? [13:39:04] *** Quits: jstern_ (~jstern@134.134.139.73) (Ping timeout: 255 seconds) [13:39:41] btw - we run autopackage.sh with ubsan enabled which adds about 20 seconds on my system [13:39:43] this script already generates a makefile fragment (mk/cc.mk) that exports the CC and CXX vars, and we can just add CC_TYPE to that [13:40:00] hmm, we should probably turn ubsan off for autopackaget hen [13:40:30] yes :) [13:40:44] I'm surprised it makes the compilation that much slower [13:40:50] it's actually the linker [13:41:05] maybe you should get a faster computer ;) [13:42:00] I am going to go ahead and merge this LTO patch, and then I can put together a review that also exports CC_TYPE so you can use it in the dpdkbuild makefile [13:42:26] or you can roll that into your change if you prefer [13:43:07] i follow you now - i was missing that we already get mk/cc.mk [13:43:15] sure - go ahead and add the CC_TYPE [13:43:19] i'll rebase off of that [13:47:10] *** Joins: jstern_ (jstern@nat/intel/x-kpvrqhycazkmbpdn) [13:50:04] bwalker: i suggested adding that check for the bdev_io_pool but it is showing there are some leaks that need to be resolved [13:50:35] i think you should keep the messages in there but don't fail - then let's have someone dig into those leaks [13:50:40] ok [13:51:25] these vhost failures are killing me [13:51:31] we need to revert [13:52:35] we can put in karol's patch for now (and change it so it does not report failure if we have to revert to sigkill) [13:53:34] i'll send another e-mail out to make sure this gets attention tomorrow - if not we'll revert in the morning [13:58:43] bwalker: posted a comment on your patch - i think you should only remove the bdev_io_pool assert [13:59:02] the other asserts are ok - they've always worked and your patches shouldn't change that [14:00:17] k [14:28:29] i think we can probably cut off a good 30-45 seconds from our iscsi testing by refactoring the test scripts so that each one does not start a new instance of iscsi_tgt [14:28:37] similar for nvme-of probably [14:29:10] that would be awesome [14:29:34] also would be nice if we could cut down on the repetitive test script stuff that starts iscsi_tgt [14:29:42] we have a bunch of tests that do things like copy spdk to an iscsi lun and do a build there - but we don't re-run configure on the copy to disable ubsan and coverage [16:21:43] this new CC_TYPE is broken on my system [16:22:23] empty? [16:22:26] % gcc -v 2>&1 | grep version (git)-[dev/dpdk_submodule]- [16:22:26] Configured with: ../src/configure -v --with-pkgversion='Ubuntu 5.2.1-22ubuntu2' --with-bugurl=file:///usr/share/doc/gcc-5/README.Bugs --enable-languages=c,ada,c++,java,go,d,fortran,objc,obj-c++ --prefix=/usr --program-suffix=-5 --enable-shared --enable-linker-build-id --libexecdir=/usr/lib --without-included-gettext --enable-threads=posix --libdir=/usr/lib --enable-nls --with-sysroot=/ --enable-clocale=gnu --enable-libstdcxx-debug --enable-lib [16:22:26] stdcxx-time=yes --with-default-libstdcxx-abi=new --enable-gnu-unique-object --disable-vtable-verify --enable-libmpx --enable-plugin --with-system-zlib --disable-browser-plugin --enable-java-awt=gtk --enable-gtk-cairo --with-java-home=/usr/lib/jvm/java-1.5.0-gcj-5-amd64/jre --enable-java-home --with-jvm-root-dir=/usr/lib/jvm/java-1.5.0-gcj-5-amd64 --with-jvm-jar-dir=/usr/lib/jvm-exports/java-1.5.0-gcj-5-amd64 --with-arch-directory=amd64 --with- [16:22:28] ecj-jar=/usr/share/java/eclipse-ecj.jar --enable-objc-gc --enable-multiarch --disable-werror --with-arch-32=i686 --with-abi=m64 --with-multilib-list=m32,m64,mx32 --enable-multilib --with-tune=generic --enable-checking=release --build=x86_64-linux-gnu --host=x86_64-linux-gnu --target=x86_64-linux-gnu [16:22:30] gcc version 5.2.1 20151010 (Ubuntu 5.2.1-22ubuntu2) [16:22:32] ~ [16:22:55] see that Configured with: blah blah --with-pkgversion [16:23:00] I think a grep -w version would fix it [16:23:09] i'll add that now [16:23:11] yeah crap [16:26:02] pushed [16:30:26] I don't think this works right on FreeBSD either [16:30:36] (before or after the change) [16:30:58] $ cc -v [16:30:58] FreeBSD clang version 3.8.0 (tags/RELEASE_380/final 262564) (based on LLVM 3.8.0) [16:31:10] so it's going to pick out "FreeBSD" instead of clang [16:31:34] should probably just look for 'gcc version' and 'clang version' explicitly [18:06:08] *** Joins: changpeng (~changpeng@134.134.139.75) [18:18:58] *** Quits: jstern_ (jstern@nat/intel/x-kpvrqhycazkmbpdn) (Ping timeout: 260 seconds) [18:19:00] *** Joins: peluse_ (~peluse@ip72-201-144-44.ph.ph.cox.net) [19:23:33] *** Joins: tsuyoshi (b42b2067@gateway/web/freenode/ip.180.43.32.103) [19:44:57] *** Joins: ziyeyang_ (~ziyeyang@192.55.54.40) [20:34:19] *** Quits: ziyeyang_ (~ziyeyang@192.55.54.40) (Remote host closed the connection) [23:47:01] *** Quits: tsuyoshi (b42b2067@gateway/web/freenode/ip.180.43.32.103) (Quit: Page closed)