**** BEGIN LOGGING AT Wed Feb 17 02:59:59 2016 Feb 17 09:06:20 hi. i repeatedly have this problem but i'm not sure how to debug it, except trial and error: say there is a recipe xyz.bb that adds a file ("example.txt") using SRC_URI. i want to replace that file, so i create a bbappend (xyz.bbappend) and a replacement "example.txt". i prepend the subfolder where my example.txt is placed to FILESEXTRAPATHS. but when building the original example.txt is used instead of my replacement. Feb 17 09:14:36 Hi. Is there an easy way to add all the files in a folder to SRC_URI Feb 17 09:15:01 Instead of doing this individually --> SRC_URI += "file://${PN}.service" Feb 17 09:36:35 good morning Feb 17 10:03:32 hi.. i see .so files in package-split-dev but not in package-split .. why? i need the *.so also in my image not only in sdk Feb 17 10:04:13 ping rburton Feb 17 10:04:14 rtr_ not sure if it helps, but you can add a directory to SUB_URI, and it adds all files and subdirectories recursively Feb 17 10:09:51 ok, could someone take a look at this recipe and bbappend? http://pastebin.com/EhQE2dQM Feb 17 10:10:31 ericbutters: is the .so file a symlink or not? Feb 17 10:11:04 i'm trying to replace a file "boardenv/config-expansions" that is included in the original recipe. i have a subdirectory "env" which contains a custom "boardenv/config-expansions" Feb 17 10:11:15 but the original is used, no matter what i do... Feb 17 10:11:34 ionte: you don't need to add file://boardenv \ again in the bbappend Feb 17 10:12:00 the variable already contains that name because it was added in the original bbfile Feb 17 10:12:26 bluelightning: it is not a symlink Feb 17 10:12:26 aratiu: i thought so. i added it when i couldn't get it to work. Feb 17 10:12:39 aratiu: same problem if i remove it Feb 17 10:12:50 ericbutters: ok, in which case you will need to set FILES_${PN}-dev such that the .so file isn't matched by it Feb 17 10:15:38 ionte: boardenv/ is a directory containing the config-expansions file? Feb 17 10:15:46 aratiu: yes Feb 17 10:16:09 you need to put in SRC_URI files, not directories Feb 17 10:17:14 SRC_URI can take directories, but their behaviour is not well defined Feb 17 10:17:46 personally I would recommend referring to files instead Feb 17 10:18:04 ionte: either try putting the boardenv directory in the FILESEXTRAPATHS_prepend path and list only the files in SRC_URI or try putting the entire path to the file boardenv/config-expansions in SRC_URI Feb 17 10:20:46 bluelightning: is FILES_${PN} and FILES_${PN}-dev installed into sdk? or only -dev? Feb 17 10:21:12 ericbutters: ${PN}-dev depends on ${PN} so you'll get both Feb 17 10:21:31 unless the recipe disables that dependency of course Feb 17 10:21:42 (which would be unusual) Feb 17 10:22:33 No fixes for the getaddrinfo vuln in git? Feb 17 10:23:34 bluelightning: okay, so i could do FILES_${PN}-dev = "" and do FILES_${PN} = "/usr/lib" -- that will bring the *.so to SDK and IMAGE right? Feb 17 10:25:25 ericbutters: should do yes Feb 17 10:25:50 thanks Feb 17 10:31:02 alexlarsson: I don't believe so Feb 17 10:31:32 looks like we should get on it immediately though Feb 17 10:37:34 aratiu: directories in SRC_URI does seem to work - the contents are included recursively. this is how it is done in the other layer. but i tried to change SRC_URI to include "boardenv/config-expansions", but the default file is *still* included, instead of my custom file... Feb 17 10:37:59 alexlarsson: I've sent an email to our security list FWIW Feb 17 10:38:16 bluelightning: cool Feb 17 10:38:30 thanks for bringing it to our attention Feb 17 10:38:40 aratiu: if i change the filename to a file that does not exist (i tried with boardenv/config-expansionsXYZ) i get a file-not-found error Feb 17 10:38:56 aratiu: but there must be a better way to debug this than trial and error? Feb 17 11:15:29 ionte: if you get a file not exists error and the file indeed does not exist, what is trial and error about that? Feb 17 11:16:47 aratiu: i would rather see a path list and debug the operations that causes the wrong file to be included, rather than "well, let's see if i rename it. if it complains it is not found, then the operation at least runs. if it doesn't, the problem must be elsewhere - the bbappend not being used or something like that..." Feb 17 11:17:38 you can see the lists of paths very simply, it is an environment variable Feb 17 11:17:40 just print it Feb 17 11:18:05 I usually just print it in a recipe function append and do an exit(1) to see the value Feb 17 11:37:16 Hello, Feb 17 11:37:22 who is Fido's maintainer? Feb 17 11:41:08 otavio: https://wiki.yoctoproject.org/wiki/Stable_branch_maintenance Feb 17 11:41:18 I am, what's up? Feb 17 11:41:55 joseppc: thx Feb 17 11:41:59 joshuagl: https://sourceware.org/bugzilla/show_bug.cgi?id=18665 Feb 17 11:42:00 Bug 18665: was not found. Feb 17 11:42:29 joshuagl: this is a glibc CVE issue, not yet fixed Feb 17 11:42:52 otavio: OK, I'll work on integrating that in a bit. Thanks Feb 17 11:43:27 joshuagl: I think master and jethro also need fixing Feb 17 11:43:33 joshuagl: but did not check Feb 17 11:43:55 otavio: sure, I'll check **** BEGIN LOGGING AT Wed Feb 17 12:00:09 2016 Feb 17 12:56:58 Hi, how compatible is prserv with different bitbake and yocto versions? Feb 17 13:46:50 why isn't the ipk created and the log says that it is empty when installing only hidden folders from a recipe? Feb 17 15:12:18 Hi Feb 17 15:13:59 I'm writing a bbappend file for ppp --> http://pastebin.com/gqYZrvM2 Feb 17 15:14:36 I'm facing this error --> ERROR: QA Issue: ppp: Files/directories were installed but not shipped /lib/systemd/system/ppp@provider.service [installed-vs-shipped] Feb 17 15:14:50 Can someone please help me with this Feb 17 15:14:52 ? Feb 17 15:54:37 Spent some time looking for a patched version of glibc Feb 17 15:54:50 But there seems to be no movement upstream, at least at https://github.com/openembedded/openembedded-core/tree/master/meta/recipes-core/glibc Feb 17 15:55:28 The relevant patch on glibc's repo is at https://sourceware.org/git/?p=glibc.git;a=commit;h=e9db92d3acfe1822d56d11abcea5bfc4c41cf6ca Feb 17 15:55:41 I'm not even sure if my distros are vulnerable Feb 17 15:56:07 i'd guess that needs to be applied to every oe-core branch, but *shrug*, not sure either Feb 17 15:56:11 been trying to figure it out, but I seen to have a problem with resolv.conf being ignored... (I never use dns on this particular system) Feb 17 15:57:15 *been trying to figure out if the distro I maintain is vulnerable, but resolv.conf is being ignored, so dns does not work Feb 17 16:03:48 Ok... looking at the list of branches on openembedded-core I see what you mean (daisy danny denzil dizzy dora dylan fido jethro master) Feb 17 16:04:52 blue123, we need that patch. I was waiting for the fix to show up in the repo Feb 17 16:05:20 * armpit glibc repo Feb 17 16:05:40 master, jethro and fido for sure Feb 17 16:06:10 anyway, did anyone manage to run the proof-of-concept? https://github.com/fjserna/CVE-2015-7547 I tried and failed, I think my distro is vulnerable, it just that I've got some problem with resolv.conf for some reason (so dns does not work) Feb 17 16:07:26 Anyway the patch is now on libc's repo, as mentioned above it's at https://sourceware.org/git/?p=glibc.git;a=commit;h=e9db92d3acfe1822d56d11abcea5bfc4c41cf6ca Feb 17 16:08:10 not sure if we could cherry-pick that one... Feb 17 16:10:07 anyway, I'not in a hurry here (the distro I maintain runs in somewhat isolated environments). But I imagine there might be others open to the world. Feb 17 16:10:44 blue123: scary Feb 17 16:13:09 I particularly worry about those which don't have PAE (the NX-bit), or ASLR enable. Feb 17 16:14:23 But maybe there are not many systems like that anymore (without PAE and ASLR). Feb 17 16:15:46 But to be honest I was not running PAE until farily recently Feb 17 16:17:45 and without a PAE enabled kernel, it is semi-trivial to pull off a remote code execution exploit. Feb 17 16:18:46 (at least that is my impression, I might be wrong... ) Feb 17 16:22:41 Hi. What is the reason for "file installed but not shipped" error? Feb 17 16:23:26 you installed the file in do_install but it's not being packaged Feb 17 16:23:35 see the yocto project documentation about teh FILES and PACKAGES variables Feb 17 16:29:31 @kergoth ln -s ${D}${systemd_unitdir}/system/ppp@.service ${D}${systemd_unitdir}/system/ppp@provider.service Feb 17 16:30:10 read what i just said again Feb 17 16:30:23 or just go read the ppp recipe, where the ppp@.service is explicitly added to the list of files to package Feb 17 16:42:23 @kergoth Sure. Thanks Feb 17 19:39:31 ping bluelightning Feb 17 19:40:17 morning all Feb 17 19:40:22 ericbutters: pong Feb 17 19:42:33 bluelightning: i got: ERROR: The recipe my-connectivity is trying to install files into a shared area when those files already exist. And i want that.. how to use SSTATE_DUPWHITELIST for this? Feb 17 19:42:46 it complains about tmp-glibc/sysroots/tegra-t18x/usr/lib/perl/5.20.0/strict.pm Feb 17 19:43:26 ericbutters: you'd have to look for examples... git grep SSTATE_DUPWHITELIST Feb 17 19:53:28 bluelightning: even when i pass full path it does not work Feb 17 20:29:25 make[2]: *** No rule to make target `../readline/libreadline.a', needed by `gdb'. Stop. Feb 17 20:29:28 * kergoth shakes head Feb 17 20:33:02 wtf.... **** ENDING LOGGING AT Thu Feb 18 02:59:59 2016