**** BEGIN LOGGING AT Mon Jan 21 02:59:56 2019 Jan 21 08:08:23 good morning Jan 21 10:20:35 hi, some ptests are dlopening libs from .so symlink which is only in -dev packages. Are patches to switch loading .so.6 etc ok? Jan 21 10:21:57 halstead: AUH is fine, however I would like RRS to be either fixed, or switched off, as it keeps sending outdated reports to yocto@ ML Jan 21 10:25:21 RP: " Can't open file /home/pokybuild/yocto-worker/qemux86-64-x32/build/build/tmp/work/x86_64_x32-poky-linux-gnux32/perl/5.28.1-r0/image/usr/libx32/perl5/5.28.1/x86_64-linux/.packlist: No such file or directory at installman line 183." Jan 21 10:25:26 let me check what that means... Jan 21 11:16:27 Hi guys! Is it possible to select the GCC (GCC SDK) version for a single recipe? I just want to compile an older pkg that wouldnt compile otherwise Jan 21 11:18:08 malanecora: you can patch the source in the older package instead, I don't think cherry-picking compilers is possible Jan 21 11:18:15 RP: patch for perl sent Jan 21 11:19:07 kanavin: OK, fair enough...I'll try to figure out how to do it. Ty! Jan 21 11:29:58 kanavin: Could be possible to remove some compilation flags from the list (CFLAGS?)? Jan 21 11:30:54 If it is would be way more straightforward Jan 21 11:51:31 malanecora: absolutely Jan 21 11:51:33 what flag? Jan 21 12:02:45 kanavin: thanks! :) Jan 21 12:11:27 rburton: -lnsl Jan 21 12:11:51 rburton: but sadly, I think it won't do the trick :/ Jan 21 12:12:35 that looks like something the makefile is doing itself Jan 21 12:12:55 if you pastebin the errors we might be able to help Jan 21 13:43:59 rburton: Ty, I managed to solve if pathcing the configure.ac in the sources Jan 21 13:44:53 Now I'm getting "Error: No manifest generated from: wl12xx-fw " and it is weird, because this recipe worked as expected in Rocko Jan 21 13:45:21 Is not the build system suposed to do that job? Jan 21 13:50:52 The build of the standalone package doesn't break Jan 21 13:51:05 But it does when I try to add it to my image Jan 21 13:52:03 Well, I get this warning: wl12xx-fw: /wl12xx-fw/lib/firmware/ti-connectivity/wl128x-fw-4-plt.bin is owned by uid 1000, which is the same as the user running bitbake. This may be due to host contamination [host-user-contaminated] Jan 21 13:52:13 May be it has something to do here... Jan 21 14:18:19 Hi, all is it possible to evaluate a yocto variable (MACHINE) in a recipe, (I have a defconfig which is same for both the machines apart from one config), so I would like to check ${MACHINE} inside do_compile_prepend() and append .config file depending on the MACHINE is that possible ? Jan 21 14:21:09 prabhakarlad: its possible, but probably easier to do without code at all. i guess we are talking about the kernel config? Jan 21 14:26:56 LetoThe2nd: yes the kernel config, (I dont want to have 2 copies of the config), could you please point me to some example. Jan 21 14:27:12 prabhakarlad: in shell just do ${MACHINE}, it gets expanded for yo Jan 21 14:28:10 prabhakarlad: well we use two configs and SRC_URI_append_machine = "machine_defconfig", basically Jan 21 14:39:07 rburton: LetoThe2nd: thank you for the pointers :) Jan 21 15:13:14 jonmason: let me know if you're around Jan 21 15:14:15 RP: I'm here, but just getting to my computer Jan 21 15:14:47 jonmason: ok. I tried the qemuarm patchset but couldn't get it to work, on the autobuilder or locally :( Jan 21 15:15:02 is it the defcnfig? Jan 21 15:15:10 that is a second series I sent out first Jan 21 15:15:21 it need that or it won't work Jan 21 15:15:22 jonmason: I have no idea. Doesn't show any signs of life under qemu Jan 21 15:16:12 ok, i'll take a look Jan 21 15:16:39 jonmason: Which second series? was this kernel changes Bruce merged? Jan 21 15:17:31 "[kernel-cache][PATCH 0/2] qemuarm: config changes needed for rename" Jan 21 15:17:40 let me update and see if he pulled it in over the weekend Jan 21 15:18:15 doesn't appear to be in Jan 21 15:18:23 jonmason: ok, that explains it Jan 21 15:18:36 * RP will wait for that to merge and update then we can retest Jan 21 15:18:36 but in his defence, I sent out late on friday and this is a holiday weekend Jan 21 15:18:48 jonmason: right, I just hadn't realised the dependency Jan 21 15:19:03 jonmason: We're due M2 today but I'd really like to have this in Jan 21 15:19:29 I debated putting that in the 0/2 patch, but since it was OE decided that it was unrelated. poor decision? Jan 21 15:19:43 jonmason: a mention would have been helpful Jan 21 15:20:16 * zeddii is here Jan 21 15:20:23 will know for next time. I'm still a bit confused about the relations between the projects Jan 21 15:20:25 RP: I did test the series locally. Jan 21 15:20:35 I can have the patch out to the list before the top of the hour Jan 21 15:20:57 zeddii: thanks, I can then test the next stage and see if this is good for M2 :) Jan 21 15:21:03 I always want to run some of the beasties myself, everything looked sane here. Jan 21 15:21:20 ok. I'll get to finishing it up. Jan 21 15:23:36 worth noting, there might be an issue with the 4.19-tiny. It looks to be referencing a 4.15 brance for qemuarm(v5) Jan 21 15:23:43 err.branch Jan 21 15:24:04 Afternoon! Is there an easy way to get build details / status / summary after running bitbake from the command line, much in the way that toaster can show status? Wondering if its trivial to do this so I can hook into a notification API ... Jan 21 15:24:25 no_such_user: there is a log file Jan 21 15:24:28 oh yah. armin pointed that one out to me. I can tweak it here. Jan 21 15:24:39 :q Jan 21 15:27:43 jonmason: Sure, but its not really in an easily parsable format - Im wondering how toaster gets its status summaries Jan 21 15:31:52 jonmason: I'd wondered about that 4.15 reference Jan 21 15:36:29 Have the WL12xx and WL18xx FW of meta-ti been replaced by linux-firmware of meta-oe? Jan 21 15:36:55 RP: I'm seeing compile warnings on it too. So, I think it needs to TLC. I was going to open bugs to track all the qemuarm* kernel compiler warnings Jan 21 15:37:40 jonmason: We can't get the bitbake warnings 100% cleaned up yet but it is nice to fix these things... Jan 21 15:37:47 no_such_user: toaster is a bitbake UI, so interacts with the server programmatically with python. it's not emitted in json or something Jan 21 15:40:27 no_such_user: have a look at bitbake events, you can drive things from them Jan 21 15:52:23 RP: Ah, that looks handy, thanks! Jan 21 16:19:22 Piraty, thanks, just saw your message...that page is an archive, not the upstream. Up to v0.3.0, opkg code home was on google code, but then it moved to the yocto git servers (http://git.yoctoproject.org/clean/cgit.cgi/opkg/) Jan 21 16:20:04 i thought you still maintained it, because the mailing list is still active too (accoring to you) Jan 21 16:21:20 RP: http://git.yoctoproject.org/cgit.cgi/poky-contrib/log/?h=akanavin/virgl-gtk is once more ready for AB Jan 21 16:26:08 hope it's more or less green this time Jan 21 16:27:06 Piraty: maybe we should put directions on that page to be explicit about were upstream is Jan 21 16:27:29 I'll look into it Jan 21 16:33:18 if I define a SRCREV in a recipe and in local.conf, is local.conf supposed to take precedence? Jan 21 16:45:54 I see that my override is parsed as expected, but it doesn't actually override the value, it's in the form of SRCREV_foo_pn-bar = "" Jan 21 16:50:41 dkc: no, recipes override config metadata unless overrides are used. if overrides are used they take precedence. if an override is used by both local.conf and ther ecipe then it depends on overrides priority (order of items in the OVERRIDES variable) Jan 21 16:50:57 you sohuld use bitbake -e to examine both the variable and its value and where it was defined Jan 21 16:51:03 bitbake -e yourrecipe, tha tis Jan 21 16:51:15 yep, that's what I'm doing, let me paste you what I get Jan 21 16:53:46 kergoth: https://pastebin.com/kKvtpya5 sorry it's slightly obfuscated but everything is there Jan 21 16:55:32 so I have SRCREV_gin = "${AUTOREV}" in the recipe, and SRCREV_gin_pn-foobar = "", so I would expect the override to... well override ^^ Jan 21 16:56:15 that's.. odd. the listed operations from bitbake -e seem to indicate it should be the overridden value Jan 21 16:56:24 kanavin: https://autobuilder.yoctoproject.org/typhoon/#/builders/85/builds/139 Jan 21 16:56:40 RP: thanks :) fingers crossed... Jan 21 16:56:52 kergoth: f* me, there was typo in my "foobar" Jan 21 16:57:07 haha, that'll do it Jan 21 16:57:41 as it was listed in my SRCREV_gin overrides, I just read it without paying attention Jan 21 16:57:58 hmm, gin Jan 21 16:57:59 sorry for the noise Jan 21 16:59:10 rburton: have you tonic though? Jan 21 17:18:09 RP: neat it is then Jan 21 17:18:14 RP: i do have frozen lime slices Jan 21 17:18:20 #preparation Jan 21 17:20:06 rburton: thats ok then :) Jan 21 17:22:48 RP: got a bottle of https://www.facebook.com/TarquinsGin/photos/a.551503644888767/2040974872608296/, practically a crime to mix that. released on my birthday and only 500 bottles ever, couldn't resist :) Jan 21 17:24:29 rburton: I like like description too :) Jan 21 17:58:11 fray: the submodule regression is not fixed with your patch Jan 21 17:58:26 fray I see it in three recipes Jan 21 17:59:08 http://cgit.openembedded.org/meta-openembedded/tree/meta-oe/recipes-support/cli11/cli11_1.6.2.bb?h=master Jan 21 17:59:39 heh, must be getting sick of mucking with gitsm by now eh Jan 21 17:59:43 http://cgit.openembedded.org/meta-openembedded/tree/meta-oe/recipes-core/dbus/dbus-broker_git.bb?h=master Jan 21 18:00:12 OT, but http://app.programmingfonts.org Jan 21 18:00:52 https://github.com/YoeDistro/meta-ti/blob/yoe/mut/recipes-bsp/rwmem/rwmem_1.1.bb Jan 21 18:01:29 kergoth: he's braver than me, I hate fetcher changes as they have so much potential to go wrong :( Jan 21 18:01:41 me too, makes me nervous every damn time Jan 21 18:02:11 we've hit gitsm issues too :( Jan 21 18:02:22 a colleague is looking into it, or will look tomorrow, as it's late Jan 21 18:02:30 * RP has tried to ensure we have a good test suite Jan 21 18:02:47 especially gitsm Jan 21 18:03:13 which is beast in itself Jan 21 18:04:03 kergoth: handy link for fonts i use fira code Jan 21 18:04:06 it's tough since the behavior changes with the state of DL_DIR, and trying to reproduce every possible state of that is non-trivial Jan 21 18:04:09 nice Jan 21 18:04:13 and noto mono Jan 21 18:04:33 i'm using fira code right now too, experimenting, until then i was really happy with Input Mono Narrow ExLight Jan 21 18:04:58 still rather nostalgic for Terminus and Inconsolata, i loved those while i used them years ago Jan 21 18:05:27 Input's flexibility is nice, yo ucan tweak almost everything when yu download it Jan 21 18:08:01 other than some particular cases (i.e. distinguishing 1 and l and I) font seems to be mostly preference and just getting used to whatever it is. they're all pretty readable. personally not a big fan of ones that are too compressed width wise, like Iosevka Jan 21 18:11:51 i've been experimenting with fonts with ligature support, still not sure if i like themo r not Jan 21 18:14:49 there's even stranger setups too like https://github.com/kencrocken/FiraCodeiScript Jan 21 18:21:44 Inconsolata is good Jan 21 18:22:19 Fira Code Retina is best for me Jan 21 18:54:45 RP I think we have an issue with the default testsuite. they assume sshd on the target, core-image-minimal does not provide that from what I can tell Jan 21 18:55:42 the old devtest suite bypasses the sshd pkg check Jan 21 19:02:14 armpit: doesn't that mean the tests just get skipped for minimal though? Jan 21 19:02:49 armpit: this is why we need a final step of builds where we compare the test results against a master list Jan 21 19:03:27 armpit: for minimal we'd only ever expect the ping test to work (and serial login and any serial tests we add) Jan 21 19:03:57 no, the build* ones try to talk to the target still in the setup and tear down Jan 21 19:04:19 outside the skip ckeck for the actual test Jan 21 19:04:25 armpit: so why aren't we seeing problems on the autobuilder? Jan 21 19:04:41 I am.. via Thud Jan 21 19:04:58 armpit: ah, didn't I make some change in master that we need there then> Jan 21 19:05:31 I have been looking. The one thing I need to do is double check master Jan 21 19:05:52 armpit: I'm sure I changed some of the setup/teardown due to this Jan 21 19:06:40 on with the hunt ! then :) Jan 21 19:19:19 we can use dropbear Jan 21 19:19:27 actually just virtual ssh Jan 21 19:19:32 would be nicer Jan 21 19:20:56 khem, sshd was used in the generic sence (openssh & dropbear) Jan 21 19:27:28 I See Jan 21 19:27:29 * armpit found it... Jan 21 19:29:06 khem since you are here.. the "Upgrade to latest on 2.31 release branch", any reason why Thud should not get it ? Jan 21 19:33:05 RP, I see 3 quick tests running. should I wait to launch one for thud? Jan 21 19:34:17 armpit: yes thud could get it Jan 21 19:34:38 khem, cool.. thanks Jan 21 19:36:11 halstead, can we archive build logs for 2018 since we are in a new year? Jan 21 20:14:43 Hm, building an older bsp - getting an error fetching uninative / nativesdk. Googling I can see that it seems to have been a known bug but I cant easily find what the fix / workaround is. Any ideas where to look? Jan 21 20:29:19 * armpit launches thud AB build... Jan 21 21:02:18 no_such_user: you need to provide more info Jan 21 21:14:24 armpit: should be fine... Jan 21 21:14:36 armpit: sorry, was eating :) Jan 21 21:19:07 RP, no worries Jan 21 22:20:11 RP: is there a way to force fetch of SRC_URI without wget fetcher trying to understand it ? Jan 21 22:20:38 I have a url which results in a tgz file but doesn't end with tgz Jan 21 22:21:25 ;downloadfilename= Jan 21 22:28:48 khem: Is it unpack causing the problem? Jan 21 22:29:59 khem: ;unpack=0 might do it Jan 21 22:30:09 khem: or as kergoth says, rename it Jan 21 22:34:37 jonmason: backtrace from testimage now: https://autobuilder.yoctoproject.org/typhoon/#/builders/53/builds/200 Jan 21 22:37:16 jonmason: serial console timeout and X didn;t start when I tried it here Jan 21 22:37:35 jonmason: serial console seems very slow too :/ Jan 21 22:37:49 at least its showing signs of life I guess :) Jan 21 22:39:18 armpit: your build looks happier :) Jan 21 22:43:22 RP: I've been testing with `runqemu qemuarm nographic` Jan 21 22:43:40 I just did a fetch and build, and it is working the same for me Jan 21 22:43:51 I'll look more at your use case to see what I missed Jan 21 22:43:58 jonmason: what about with graphics enabled? Jan 21 22:44:07 RP: yeah seems that is going to work Jan 21 22:44:48 jonmason: we do use graphics support Jan 21 22:45:01 RP: since that was failing with the legacy stuff, I was going to get around to it after getting this stuff in Jan 21 22:45:20 it is on my radar Jan 21 22:45:36 jonmason: graphics works today but doesn't work after the patches Jan 21 22:45:51 oh? damn... Jan 21 22:45:51 jonmason: it doesn't work on 64 bit but 32 is ok Jan 21 22:46:17 jonmason: the other thing to get working is "bitbake core-image-sato -c testimage" Jan 21 22:46:42 to clarify, it doesn't work any more on qemuarmv5 or the new qemuarm doesn't work ? Jan 21 22:46:51 jonmason: sorry, its going to need a little more work Jan 21 22:47:01 jonmason: qemuarmv5 today boots graphics into X Jan 21 22:47:16 jonmason: new qemuarm does not Jan 21 22:47:23 ok, so its the limitation of not having the virtio graphics then Jan 21 22:47:32 I think so Jan 21 22:47:37 rp, happy good Jan 21 22:48:12 that is my next task, as I think the fix will match what is needed for qemuarm64 Jan 21 22:48:21 is this a show stopper for these patches? Jan 21 22:48:33 jonmason: autobuilder going red is show stopper Jan 21 22:48:41 (red == regressions) Jan 21 22:48:58 * armpit or too much beer Jan 21 22:49:49 * armpit kernel merging hurts Jan 21 22:50:15 jonmason: I think the duplicate console declaration is going to be problematic Jan 21 22:50:19 RP: I'll make it my highest priority to get that working ASAP. Do I need to push a revert? Jan 21 22:50:33 jonmason: I also had to tweak your patches to apply on top of zeddii's Jan 21 22:50:44 jonmason: no, its not merged, just in testing Jan 21 22:50:46 the double consoles is a bit ugly, but does allow for the virtio consoles to be used Jan 21 22:51:04 jonmason: its showing timeout errors on the console though Jan 21 22:51:34 it doesn't show it if hvc0 is being used iirc Jan 21 22:51:53 jonmason: well, runqemu wasn't using it :/ Jan 21 22:52:12 i was just about to say that Jan 21 22:52:31 I looked at adding it and it would be not too bad, but would take a little bit of work Jan 21 22:52:57 if you want to open a bug to do this and assign it to me, then I can get to it eventually Jan 21 22:53:32 jonmason: can you sort the bug? I think this goes on the list of things that need sorting to get this merged Jan 21 22:53:47 sure, I'll be happy to do that Jan 21 22:54:23 jonmason: this change was always going to be 'fun' so these gremlins aren't entirely surprising btw Jan 21 22:55:01 I appreciate your patience. Lots of learning happening in these patches Jan 21 22:57:07 I'll do my best to get graphics working in the next couple of days (of course, that probably means end of next week based on past promises) Jan 21 22:57:53 hahah Jan 21 23:02:23 jonmason: I think I'll roll M2 without this but we at least have a V1 and an idea of where it stands so its demonstrable progress Jan 21 23:02:59 that's fair Jan 21 23:07:48 New news from stackoverflow: Yocto rebuild is not overwriting **** ENDING LOGGING AT Tue Jan 22 02:59:56 2019