--- Log opened Fri Aug 16 00:00:43 2019 00:56 < jonasschnelli> Anyone knows why "make check" of a Win64 builds complains with: /bin/bash: test/test_bitcoin.exe: cannot execute binary file: Exec format error 00:56 < jonasschnelli> ? 00:57 < jonasschnelli> while "wine ./src/test/test_bitcoin.exe" runs 01:00 < aj> jonasschnelli: sounds like the make check doesn't invoke it via wine; you need to tell binfmt_misc to identify .exe's and run them via wine so that you can just invoke a .exe directly 01:00 < jonasschnelli> thanks... aj. Let me have a look 01:04 < jonasschnelli> aj: that should have been done automatically when installing the wine-binfmt package I guess? 01:05 < fanquake> review beg for #16371. It's pretty trivial. 01:05 < aj> jonasschnelli: maybe the binfmt_misc kernel module isn't available? 01:11 < jonasschnelli> aj: Looks like it running... since this is about the CI. I think wine-binfmt should register that automatically 01:11 < jonasschnelli> Maybe order or apt install packages? 01:16 < aj> jonasschnelli: shouldn't matter... is this in docker? you might need some docker magic to let binfmt_misc work in a container? 01:16 < jonasschnelli> aj: It's not running in docker... directly in a qemu vm 01:16 < jonasschnelli> (bitcoinbuilds.org) 01:17 < jonasschnelli> the issue appeared when I create a new base-vm-image (that will be loaded for a blank vm to run a job inside)... I apt-installed all packages to avoid reinstall them when building a job 01:18 < jonasschnelli> And since then the W64 make check fails 01:18 < jonasschnelli> full log: https://bitcoinbuilds.org/index.php?ansilog=b93020b3-f118-486d-938c-145ed262b9a2.log 01:23 < provoostenator> I restarted Travis machines 5 and 8 on a bunch of pull requests that I'm reviewing. 01:23 < fanquake> provoostenator: yea they've all been failing all day 01:24 < provoostenator> I was hoping it was just the weather... :-( 01:24 < fanquake> Probably is. The weather here is garbage. 01:27 < aj> jonasschnelli: do you have /rpoc/sys/fs/binfmt_misc/ ? status, wine files in that dir, maybe? 01:28 < jonasschnelli> I have a directory there, yes: /proc/sys/fs/binfmt_misc/ 01:28 < jonasschnelli> there is a register file,... but can't cat it 01:29 < jonasschnelli> cat /proc/modules | grep binfmt shows me binfmt_misc 20480 1 - Live 0xffffffffc0544000 01:29 < aj> should have status file and wine file? 01:29 < jonasschnelli> and ls gives me: llvm-6.0-runtime.binfmt python2.7 python3.6 register status 01:30 < jonasschnelli> status has content "enabled" 01:30 < jonasschnelli> but there is no wine file 01:30 < aj> makes sense 01:30 < aj> does "sudo update-binfmts" fix it? 01:31 < aj> err, "update-binfmts --enable" maybe 01:31 < jonasschnelli> "sudo update-binfmts --display" shows nothing about wine 01:31 < jonasschnelli> (also after enable) 01:32 < jonasschnelli> oh 01:32 < jonasschnelli> -import 01:32 < jonasschnelli> \o/ 01:32 < jonasschnelli> "sudo update-binfmts -import" is the magic comand! 01:32 < jonasschnelli> thanks aj! Own you a beer 01:33 < aj> if you get CI working, everyone will owe you all the beers :) 01:33 < jonasschnelli> hehe... 01:34 < jonasschnelli> I wonder why it worked before.... 01:34 < jonasschnelli> but figuring that out seems to be a rabbit hole 01:34 < aj> --import should've been run when wine-binfmt got installed i think 01:34 < aj> but 3.0 is a bit old compared to debian i think, so could be anything 01:38 < jonasschnelli> I'll try now to explicitly run --import before the build starts 01:38 < jonasschnelli> (after all the apt commands) 01:46 < jonasschnelli> Now the next things.... 01:46 < jonasschnelli> asan detections in function tests: 01:46 < jonasschnelli> https://bitcoinbuilds.org/index.php?ansilog=bec8b464-8cf4-44bf-8457-f9a328c94917.log#l5938 01:46 < jonasschnelli> I wonder why those not fire in travis 01:46 < jonasschnelli> Looks like we'r not using a suppressions for ASAN 03:32 -!- jonatack [~jon@2a01:e35:8aba:8220:6627:dad:d967:649d] has quit [Quit: WeeChat 2.3] 03:34 -!- jonatack [~jon@2a01:e35:8aba:8220:6627:dad:d967:649d] has joined #bitcoin-builds 04:51 < jonasschnelli> MarcoFalke: "test_bitcoin --run_test=fs_tests" fails on my debian system with sanitizers 04:51 < jonasschnelli> unknown location(0): fatal error: in "fs_tests/fsbridge_fstream": boost::system::system_error: boost::filesystem::path codecvt to string: error 04:52 < fanquake> jonasschnelli: Is that the same error as #14948? 04:52 < jonasschnelli> MarcoFalke: the functional tests do also fail on bitcoinbuilds with ASAN, lsan enabled: https://bitcoinbuilds.org/?job=bec8b464-8cf4-44bf-8457-f9a328c94917 04:52 < jonasschnelli> fanquake: let me check... 04:53 < jonasschnelli> Oh.... yes. 04:53 < jonasschnelli> Thanks fanquake. 05:48 < jonasschnelli> p2p_invalid_messages.py fails on my debian with sanitizers: 05:48 < jonasschnelli> Memory usage increased over threshold of 50.000% from 150172 to 417496 (178.012%) 06:17 -!- elichai2 [uid212594@gateway/web/irccloud.com/x-gwpzuvjzenerifyu] has joined #bitcoin-builds 07:16 < provoostenator> jonasschnelli: p2p_invalid_messages.py almost always fails for me on macOS, I've seen that particular error too. 07:22 < jonasschnelli> it really looks that our tests are optimized for travis 07:25 < provoostenator> I believe it happens mostly when building with --enable-debug 07:25 < provoostenator> Which I like to do by default, otherwise I forget 07:26 -!- grubles [~grubles@unaffiliated/grubles] has joined #bitcoin-builds 07:26 < grubles> hi 07:26 * dongcarl waves 07:26 < dongcarl> grubles: what's the hiccup? 07:26 < provoostenator> jonasschnelli: if you want to look at functional test performance, I listed the worst ones here: https://github.com/bitcoin/bitcoin/issues/16613 07:27 < grubles> i'm getting a hash mismatch for a couple things when doing the initial `guix pull` 07:27 < jonasschnelli> provoostenator: thanks 07:28 < fanquake> provoostenator: the p2p_invalid_messages issues on mac should mostly be fixed now I think. 07:28 < dongcarl> grubles: Hash mismatch for initial `guix pull`...? You mean you've performed the Guix build and have hash mismatches against what people posted on the PR, or `guix` is complaining about something? 07:28 < fanquake> Pretty sure the failing part of the test is just being fixed, as the cause in deep in some macOS specific Python buffer handling. 07:28 < fanquake> *skipped 07:29 < grubles> dongcarl: https://gist.github.com/grubles/1bd923e730fae346d5fb6c77e0648592 07:30 < dongcarl> grubles: Urgh it's cuz `http://www.mr511.de/software/libelf-0.8.13.tar.gz` is down... 07:31 < dongcarl> grubles: I'll send you an archive that you can import tho, this isn't a show-stopper in guix land 07:33 < grubles> ahh i see 07:33 < grubles> ok send it my way :-) 07:36 < provoostenator> fanquake: still breaks as of the latest master, and it's not getting skipped for me. 07:36 < dongcarl> grubles: Kk gimme a bit 07:37 < provoostenator> Also there's at least test, haven't triangulated it, which tries to bind to something other than 127.0.0.1 thus triggering a Firewall popup after every compile 07:37 < provoostenator> *at least one 07:41 < dongcarl> grubles: Single download https://send.firefox.com/download/22ca96682a0c94a6/#yCh9gDP9if-FAStass69Gw 07:42 < dongcarl> grubles: just `guix archive --import` should work 07:42 < dongcarl> oh wait I think it's `guix archive --import < libelf-for-grubles.nar` 07:43 < dongcarl> Lmk if that gets you past it 07:47 < grubles> nope 07:47 < grubles> $ guix archive --import < libelf-for-grubles.nar 07:47 < grubles> guix archive: error: program `/gnu/store/ncknl03pkmamrxg7q9nxi1rn1qhvwbi9-guix-1.0.1/libexec/guix/authenticate' failed with exit code 1 07:48 < dongcarl> grubles: Ah right you gotta authorize my key 07:48 < dongcarl> one moment 07:50 < dongcarl> grubles: https://send.firefox.com/download/8093d588f5ad7c22/#U9428e86CaJPgiAKN3ZEhw 07:50 < dongcarl> `sudo guix archive --authorize < signing-key.pub` 07:57 < grubles> ok that got the import working 08:00 < dongcarl> \o/ 08:13 < grubles> guix pull is now continuing too. thanks carl! 08:14 < dongcarl> grubles: That's really good to hear :-) 08:42 < dongcarl> cfields: I think the boostrap part of `boost` is actually "native", as in, should be moved to a `boost_native` package like the native part of `protobuf` was... Does that ring somewhat true to you? 11:02 < jb55> ooo there's a PR for adding guix to nixos, might be able to try these builds finally 11:02 < dongcarl> jb55: Dope, the store format is identical anyway 11:03 < jb55> dongcarl: yeah I think it just does the guix build and copies into nix store? if it does that would be idea, then could use guix builds for nix-bitcoin 11:03 < jb55> ideal* 11:04 < dongcarl> jb55: Hopefully that's the case! I know that there was talk of rewriting the builder daemon for Guix, which might break compatibility, but it hasn't happened yet... Hopefully the two projects can agree on a store/builder spec so that things fit together neatly 11:05 < jb55> going to try this PR and report back my findings ... 11:05 < dongcarl> jb55: Cool 11:13 -!- cfields [~cfields@unaffiliated/cfields] has quit [Remote host closed the connection] 11:16 -!- cfields [~cfields@unaffiliated/cfields] has joined #bitcoin-builds 12:02 -!- rain2 [~My_user_n@unaffiliated/rain1] has joined #bitcoin-builds 13:11 < provoostenator> I find the combination of Travis and Docker directory structure pretty confusing. 13:12 < provoostenator> It might be useful to explain that a bit better in the README, in particular where the source code lives, where the building happens and where the binaries go. 13:13 < provoostenator> And where the functional test suite lives relative to that. 14:12 < dongcarl> Fix for Guix Inferior corner case has been pushed to master upstream, could fanquake or wumpus create a new repo called “guix-channel” in the bitcoin-core org and add me to it? 14:17 < fanquake> dongcarl can take a look for you tomorrow 😴 14:18 < dongcarl> Np, gnite! 18:07 -!- elichai2 [uid212594@gateway/web/irccloud.com/x-gwpzuvjzenerifyu] has quit [Quit: Connection closed for inactivity] 23:20 -!- rain2 [~My_user_n@unaffiliated/rain1] has quit [Ping timeout: 258 seconds] --- Log closed Sat Aug 17 00:00:48 2019