**** BEGIN LOGGING AT Sun Feb 28 02:59:57 2021 Feb 28 03:00:00 hey there, having a really strange behavior with do_fetch where I'm getting a segfault.  I'm doing this through an automated process so the container goes bye bye - but what's strange is I can build core-image-minimal with no package adds just fine, but when I start adding additional packages, I get this: Feb 28 03:00:01 https://gist.github.com/rpcme/c9ed1dfda5fee1c9384b55e5e778d22c Feb 28 03:01:51 Hoping this is a known pattern and something I can just tweak.  I can spin up the container separately and run the command fine. Feb 28 03:05:06 could be psuedo faulting Feb 28 03:05:14 i ran into an issue with psuedo recently Feb 28 03:06:07 https://wiki.yoctoproject.org/wiki/Pseudo_Abort Feb 28 03:06:12 i never figured out why it happened Feb 28 03:10:35 it's bizarre how so many packages build fine. Hm Feb 28 03:37:54 in my case i think it only happened in glibc Feb 28 03:38:03 i forget which step, probably do_package Feb 28 03:41:42 this is specifically a git issue in do_fetch Feb 28 03:41:59 outside bitbake, it's working fine in the container Feb 28 03:42:09 I think it's the built git that's having a problem Feb 28 03:49:34 ok I reproduced it in the container but outside bitbake. methodically working it, but it's a process for sure... Feb 28 04:06:56 Unbelievable, it's these flags:  --bare --mirror Feb 28 04:20:35 Ok ... setting bareclone=0 in the SRC_URI "fixed" it but this is happening to a number of github repos... wondering if there's something about the repos that could be causing this? **** BEGIN LOGGING AT Sun Feb 28 05:02:04 2021 Feb 28 09:05:40 RP: right, maybe it's best that I ask that the other dosfstools patch is rebased on top of mine? I'm not super comfy adding changes to my patches that I don't understand and can't test. Feb 28 09:09:28 RP: ah wait, I looked at it again Feb 28 09:09:34 it's removing a dependency Feb 28 09:09:40 which is fine, I'll add that Feb 28 09:10:58 RP: I also have a cleaner fix for gettext/iconv issues (inherit gettext) Feb 28 09:58:25 Hi, I'm trying to add iproute2 to an image, and I added IMAGE+INSTALL += iproute2; baked the image, I see the files in my rootfs folder, but if I flash the iproute2 artifacts are not there. Where am I goofing ? Feb 28 10:12:51 nevermind, problem was between chair and desk :-) Feb 28 10:42:35 kanavin: right, its just a dependency removal which I think should be fine. I noted it in particular as its hard to catch those kind of changes Feb 28 10:42:52 kanavin: cool on gettext/iconv Feb 28 10:43:11 kanavin: I'm slowly getting there with reproducibility Feb 28 10:43:30 RP: yep, go is the difficult item Feb 28 10:43:41 RP: I think ideally we need to involve the upstream Feb 28 10:44:00 and those who know the toolchain well, e.g. khem Feb 28 10:44:15 I have some tweaks, but would rather have the specialists take over Feb 28 10:44:35 kanavin: right, I'm ignoring go atm, it will need upstream help Feb 28 10:45:29 RP: I wouldn't be surprised if upstream's response will be 'take our prebuilt tarballs' :( Feb 28 10:47:29 kanavin: I really don't know what they'll say... Feb 28 10:49:32 JPEW: "136893 out of 137376 (99.65%) packages tested were reproducible" - I don't think the tests understand a-full test runs :) Feb 28 13:44:19 I'm having some issues with packagegroups. I want to specify the exact version of each package in my RDEPENDS_packagegroup-whatever = "package-a (=1.0.0) package-b (=1.0.0)" but when I do this I get the error "Nothing provides package-a = 1.0.0 needed by packagegroup-whatever-1.0.0-r0.all" Feb 28 13:44:38 If I remove the versions from the RDEPENDS string then it works fine. Any ideas? Feb 28 13:46:43 Hmm, it seems to work if I say "(=1.0.0-r0)" instead of "(=1.0.0)". Is there a way to ignore then revision and just match the major/minor/patch version? Feb 28 15:27:27 TheComet: Is the error from bitbake or the package manager? Feb 28 15:27:42 TheComet: this could be an issue we've not considered before :/ Feb 28 15:38:49 RP: I believe it's from bitbake. Here's a paste as I'm not 100% sure https://pastebin.com/0jS0A6BX Feb 28 15:39:10 RP: patches sent, they're a-full tested, so I'm fairly confident in them :) Feb 28 15:39:24 there's also a meta-mingw patch that needs to be picked up Feb 28 16:19:32 TheComet: its from opkg Feb 28 16:20:00 TheComet: you could try a different package backend, otherwise that looks like an opkg level issue Feb 28 16:20:03 kanavin: thanks! Feb 28 21:14:45 looks like i just stumbled across the "error 134" issue, but with a completely up-to-date build on master this morning Feb 28 21:15:36 https://bugzilla.yoctoproject.org/show_bug.cgi?id=14175 Feb 28 21:16:01 the suggested fix had no effect (PSEUDO_IGNORE_PATHS_append = ":/run/") Feb 28 21:16:29 but doing a "bitbake -c clean" then re-running the build fixed it Feb 28 21:55:28 Hello I am not an python expert ... do you know what is the @ i ${@time.strftime... found in the doc ? Feb 28 21:58:11 s/i/in/ Feb 28 22:02:12 hello, I am trying to do a simple `find ... -exec rename.ul foo bar {} \;` inside a bitbake task but I keep getting this error: find: ‘rename.ul’: No such file or directory Feb 28 22:02:50 rename.ul is available in my devshell and I can do a simple find -print on the line before and clearly see that the files are there, so I don't know what gives Feb 28 22:06:20 I also tried a verbatim rename command and it complained that the command is not found, so I guess I'm wondering how I can get this utility to be available inside a bitbake task Feb 28 22:06:51 I have included util-linux-native in my DEPENDS Feb 28 22:10:09 what provides rename.ul? Feb 28 22:10:52 util-linux(-native), or so I thought. I could be wrong but I'm not quite sure how I can know Feb 28 22:11:33 try uh Feb 28 22:12:02 maybe: oe-pkgdata-util list-pkg-files util-linux-native Feb 28 22:12:33 woops, I just tried oe-pkgdata-util find-path /usr/bin/rename and it gives util-linux-rename Feb 28 22:12:38 didn't realize it was a separate package Feb 28 22:12:56 ah, there ya go :) Feb 28 22:13:04 :) Feb 28 22:13:16 thanks a lot! Feb 28 22:13:24 though IIRC packages are distinct from recipes Feb 28 22:13:28 let's see if that works if I put it in my DEPENDS Feb 28 22:13:51 yes the thing is that I did a grep of the util-linux recipe for rename and nothing turned up Feb 28 22:14:00 i dont think `util-linux-rename` is valid in DEPENDS if it was made by util-linux packaging it Feb 28 22:15:09 its because the recipe is doing python magic Feb 28 22:15:13 https://git.yoctoproject.org/cgit.cgi/poky/plain/meta/recipes-core/util-linux/util-linux_2.36.2.bb Feb 28 22:15:19 its not an explicit list of packages to split Feb 28 22:15:24 ERROR: Nothing PROVIDES 'util-linux-rename-native' (but foo.bb DEPENDS on or otherwise requires it). Close matches: Feb 28 22:15:24 util-linux-native Feb 28 22:15:25 util-linux RPROVIDES util-linux-rename-native Feb 28 22:16:10 hmm, how am I to specify this? I tired already util-linux-native in DEPENDS and it didn't work Feb 28 22:18:35 i think its just a path issue Feb 28 22:19:48 the path in devshell is /usr/bin/rename Feb 28 22:19:48 bps: can you try to find rename in your recipe's recipe-sysroot-native/ dir? Feb 28 22:19:58 sure let me check Feb 28 22:20:07 bps: yea thats the host's rename. not available in a recipe Feb 28 22:20:23 devshell lets you access any host stuff, inside recipe does not Feb 28 22:21:34 /usr/share/quilt/rename Feb 28 22:21:45 what is quilt? Feb 28 22:21:56 wrong one, probably not the rename from util-linux Feb 28 22:22:05 quilt is like a patch file manager Feb 28 22:22:17 I did a find in the recipe-sysroot-native directory so I guess it's not there Feb 28 22:22:47 can u set DEPENDS = "util-linux-native" again and rerun bitbake Feb 28 22:22:50 then go look Feb 28 22:23:16 yup doing it now - think I removed it in the last bitbake run Feb 28 22:23:30 nope, that's still the only one Feb 28 22:25:34 one remark - I currently have the recipe checked out with devtool modify. could it be that the DEPENDS are not getting parsed when I do this? Feb 28 22:26:08 it is difficult to verify what I am trying to achieve unless I have the sources checked out Feb 28 22:26:17 not familiar with devtool very much Feb 28 22:26:32 maybe the recipe is in the devtool workspace and not in your layer? Feb 28 22:26:50 it threw an error when I modified the DEPENDS in the original recipe so I think it's ok Feb 28 22:29:07 aha! now I ran bitbake -c do_populate_sysroot and now I can see a rename in my sysroot Feb 28 22:29:24 I guess it's because I was executing a specific task which was not triggering do_populate_sysroot Feb 28 22:29:27 let me try from the top Feb 28 22:32:42 alright, it works! thanks a lot mischief, you saved a couple of hours of my life :) Feb 28 22:45:54 grats Feb 28 22:46:03 now if you could make my router work that'd be great Feb 28 22:47:59 what's wrong with it? Feb 28 22:49:30 extremely OT here, but i just bought a https://www.hardkernel.com/shop/odroid-h2plus/ with the extra 4-port NIC card, and i'm setting up openbsd on it to be my new router. Feb 28 22:50:12 after like 2 days of fucking around i finally got a set of ansible scripts working that configure it but i think it's not 100%. sometimes forwarding just stops Feb 28 22:50:28 nice machine though. probably make a good yocto target Feb 28 22:53:10 nice, I use a https://www.pcengines.ch/apu2.htm as my home router but I am running linux on it... never used openbsd :) Feb 28 22:59:46 well, i got this for 2.5GBASE-T. Feb 28 23:00:00 also, i am not a fan of iptables. **** ENDING LOGGING AT Mon Mar 01 02:59:57 2021