**** BEGIN LOGGING AT Mon Dec 21 02:59:59 2015 Dec 21 08:47:13 when i run powertop i get only one cpu. i have 8 cpus. is there any utility which displays cpufreq in % Dec 21 12:57:49 gm Dec 21 14:59:33 anyone investigate recent (approx last 3d) vim breakage in do_configure before I investigate myself? Dec 21 14:59:36 | Makefile:4: ../auto/config.mk: No such file or directory Dec 21 14:59:37 | make[1]: *** No rule to make target '../auto/config.mk'. Stop. Dec 21 15:01:39 * paulg_ notes vim dir itself doesn't appear to have changed since ~ October. Dec 21 15:03:49 aha. race. went away after I did a clean. :-/ Dec 21 15:04:28 so probably been there forever ; just odd that it hit both my ppc and x86 builds at the same time. Dec 21 15:53:34 paulg_: have a look at the log, maybe it was trying to do a make clean before re-running configure? if so, the recipe needs CLEANBROKEN=1. Dec 21 15:54:24 do_configure attemps to run make clean before it does stuff to remove stale files. some recipes totally break when you try and do that though. Dec 21 15:57:19 rburton, seems my clean killed the log for the failed instance. If I see it again, I will check before clean Dec 21 15:58:12 but you are correct in assuming I'm reusing the same build dir every time, so what you suggest seems plausible. Dec 21 15:58:40 (if not 95% likely, even) Dec 21 20:19:17 is it possible to PLBLACKLIST a recipe for selected architectures? does PNBLACKLIST play well with COMPATIBLE_MACHINE? Dec 21 20:25:02 it doesn't play well since it's flag based, can't combine flags with overrides. but there was a recent patch submitted so you could just set a variable in the recipe itself instead of a flag, so you could use overrides with that Dec 21 20:35:17 kergoth: my understanding, though, is that if i set PNBLACKLIST in the recipe then the entire recipe will be ignored for all architectures. i only want it ignored for qemuarm but included in Martin's world builds for other architectures (such as qemuarm64, qemuarmx86-64...) Dec 21 20:35:51 maybe that's just not possible at this point (?) Dec 21 20:37:42 read what i just said again Dec 21 20:38:45 as i just said, there's a patch on the list that makes it so you can set a variable in the recipe to blacklist it, and you could use overrides with that to set it conditionally. Dec 21 20:38:50 see "[PATCH 1/1] blacklist: add SELFBLACKLIST support" on the oe-core list Dec 21 20:39:05 i.e. SELFBLACKLIST_qemuarm = "foo" Dec 21 21:52:56 what's wrong with compatible machine for "don't build this"? Dec 21 21:55:20 rburton: if a MACHINE doesn't build a recipe due to COMPATIBLE_MACHINE the build fails (returns non-zero), but if that recipe is blacklisted Martin's world build won't even try to build it (is my assumption) so it succeeds (in a manner) Dec 21 21:56:21 tlwoerner: that's not the case Dec 21 21:56:41 PNBLACKLIST and COMPATIBLE_* use the exact same underlying mechanism, which makes the recipe provider unbuildable Dec 21 21:57:44 kergoth: ok Dec 21 21:58:07 so? Dec 21 21:59:07 otavio: please make someone fix uboot :) Dec 21 21:59:09 (http://errors.yoctoproject.org/Errors/Search/?items=10&query=88d20387c7f991f4cd683911277b4db80b202217) Dec 21 23:10:29 What is the preferred way of testing a change before sending a patch for OE-Core? Is the a document or instructions for it? Dec 21 23:13:09 well make sure it builds Dec 21 23:13:37 test on more than one arch, if you suspect it maight have issues Dec 21 23:13:47 Is there some sort of local.conf that should be used to make certain it builds? Dec 21 23:14:58 What sort of automated test are performed on the repository to check that it builds? Dec 21 23:30:08 runs on the autobuilder Dec 21 23:30:17 big matrix of machines Dec 21 23:37:38 rburton: which is that in? Dec 21 23:37:45 *repo even Dec 22 00:01:35 nerdboy: what repo is what in? Dec 22 00:02:15 CipherWizard: depends on what your patch changes. if its recipes then at least a few different machines (this is where the qemu machines are very useful) Dec 22 00:02:43 CipherWizard: the autobuilder has automated sanity testing (meta/lib/oeqa/runtime) and selftests (meta/lib/oeqa/selftest) Dec 22 00:04:14 The OE-Core only has QEMU machines in it. So just "bitbake world" for a couple different qemus? Dec 22 00:04:57 a representative sample of packages is fine - no point building qt and webkit if your change doesn't impact them Dec 22 00:06:05 if sstate is there it should not be a problem though, Right? Dec 22 00:07:21 sure Dec 22 00:22:16 the u-boot error... Dec 22 00:22:39 and did someone delete libssh2_1.4.3.bb by mistake? Dec 22 00:23:08 * nerdboy updating fsl repo thing and it choked on meta-oe Dec 22 00:23:42 i see a new libssh2-1.4.3/CVE-2015-1782.patch but the recipe is deleted... Dec 22 00:24:50 since libssh2_1.6.0.bb does not use the patch that seems like a "doh!" Dec 22 00:25:21 * nerdboy waits for homer to speak up Dec 22 00:36:24 nerdboy: oe-core and meta-fsl master Dec 22 00:36:52 and the other thing? Dec 22 00:36:52 nerdboy: https://autobuilder.yoctoproject.org/main/builders/nightly-fsl-ppc-lsb/builds/577/steps/BuildImages/logs/stdio is a master run Dec 22 00:37:34 i aborted the rebase locally Dec 22 00:39:24 merge ff doesn't either Dec 22 00:41:25 i think it's time for me to abandon the whole repo manifest thing and make my own layer cake... Dec 22 02:47:51 nm, git and/or me was confused by the odd branch setup **** ENDING LOGGING AT Tue Dec 22 02:59:58 2015