**** BEGIN LOGGING AT Mon Apr 21 02:59:58 2014 Apr 21 05:23:55 okay, looks like xkeyboard-config switched default rules and broke my cursor keys Apr 21 05:24:13 was evdev rules, now base rules Apr 21 06:37:46 Hi yocto. Is there anyway to ignore a executable when creating the package? Apr 21 12:14:07 vicky: How do you mean ignore? If you don Apr 21 12:14:21 don't put it in your install rule it won't be included. Apr 21 15:48:13 RP: I'm guessing that it isn't possible for sstate to fetch the intermediate sigdata/siginfo necessary to get useful output from -S printdiff, since at the point where we run sstate_fetch, we don't have information about the dependent tasks within the recipe. Is that correct? Apr 21 16:32:38 as a distro maintainer, https://gist.github.com/kergoth/11147594 is slightly useful. wonder if we should add that to insane.bbclass as something the user can inspect Apr 21 17:07:25 Hi all Apr 21 17:07:43 RROR: Function failed: do_install (see /nobackup/dvasadia/ea3/iosxe/build/tmp/work/mips64_nf_n32-iosxe-linux-gnun32/aer-inject/0.1-r0/temp/log.do_install.19043 for further information) ERROR: Logfile of failure stored in: /nobackup/dvasadia/ea3/iosxe/build/tmp/work/mips64_nf_n32-iosxe-linux-gnun32/aer-inject/0.1-r0/temp/log.do_install.19043 Log data follows: | DEBUG: SITE files ['endian-big', 'bit-64', 'mips64-common', 'common-linux', 'c Apr 21 17:08:11 xe/build/tmp/work/mips64_nf_n32-iosxe-linux-gnun32/aer-inject/0.1-r0/image SBINDIR=/usr/sbin INCLUDEDIR=/usr/include | make: *** No rule to make target `install'. Stop. | ERROR: oe_runmake failed | ERROR: Function failed: do_install (see /nobackup/dvasadia/ea3/iosxe/build/tmp/work/mips64_nf_n32-iosxe-linux-gnun32/aer-inject/0.1-r0/temp/log.do_install.19043 for further information) Apr 21 17:08:18 any pointer on this error Apr 21 17:08:34 i am writing a recipe for aer-inject package Apr 21 19:36:53 Hi everyone, I have a skeleton for rootfs. It has 15 folders and many files and sub-folder. What is the best way to populate the rootfs? Apr 21 20:39:24 So I found out what I was confused about with regards to chmod/setxattr()/etc. Apr 21 20:39:46 On some systems (CentOS 5.x/6.x among them), cp -a ends up using setxattr(..., "system.posix_acl_access", ...) for the base permissions. Apr 21 20:40:04 Tricky part being that the extended attribute for this is, it turns out, basically a binary blob in a custom format. Whee. Apr 21 20:47:44 * kergoth peruses http://users.suse.com/~agruen/acl/linux-acls/online/ Apr 21 21:19:34 So, I have a bit of time to work on adding *xattr() support to pseudo, which means it may actually happen. And I am just curious: Anyone got any funny war stories about extended attribute support? Things I should be watching out for? Apr 21 21:20:10 The initial support is mostly in, except that the result is that on CentOS 6.x, "cp -a" no longer preserves modes because it helpfully tries to copy them more-precisely as a posix ACL extended attribute rather than using chmod. Apr 21 21:22:10 Hmm. Or something, anyway. I'm not actually sure what's happening here. Apr 21 21:30:30 seebs: I owe you a reply to an email don't I? :/ Apr 21 21:30:59 Probably. Not in a huge hurry, I have at least one pure bug to resolve before I'm quite sure I need to fix stuff. Apr 21 21:31:41 *thinking* Oh, and there's the "sqlite version" question: Do we expect to ever need pseudo to work with sqlite older than 3.6.19? I am pretty sure our build system currently assumes we'll be building a native sqlite to avoid host problems. Apr 21 21:32:53 seebs: From a persona/yocto perspective, I care about the version of sqlite in the metadata Apr 21 21:33:27 seebs: currently 3.8.3.1 Apr 21 21:33:32 That appears to be 3.8.x in my local tree, so I'm good. Apr 21 21:33:38 seebs: yes Apr 21 21:33:45 The issue: 3.6.19 is when foreign key support got added. Apr 21 21:33:57 And I much prefer ON DELETE CASCADE to trying to store a list of rows that got deleted and then delete rows in other tables. Apr 21 21:33:58 seebs: quite happy to depend on that from my perspective Apr 21 21:35:49 seebs: FWIW with oe-core/yocto we always build sqlite-native. I'm not aware of "host" issues there Apr 21 21:36:32 I couldn't remember the order. I know pseudo gets built VERY early on. Apr 21 21:36:53 So it seemed plausible that pseudo might be getting built before sqlite-native was, except I think that just means sqlite-native also gets built really early. Apr 21 21:38:17 I've also fixed a couple of minor nuisances. I've made the pseudo build smart enough to rebuild if you touch the things included in pseudo_wrappers.c, and I've made the pseudo-to-get-a-shell behavior shut down the server when it's done. Apr 21 21:39:26 X keyboard still dorked up Apr 21 21:39:47 rebuilding after pulling agagin Apr 21 21:39:56 *again even Apr 21 21:40:23 but riddle me this, yocto-bats Apr 21 21:41:15 when i have two linguas installed by default, why does setxkbmap think the locale is "C" instead ? Apr 21 21:42:44 if IMAGE_LINGUAS has more than one value, how does the default get set? Apr 21 22:10:38 With DEPENDS I can say that package A depends on package B. Does this also mean that every time package A change, package B will recompile? Apr 21 22:34:02 I'm trying to enable the uinput kernel module for linux-yocto-3.10. I put the kernel config line in a *.cfg file and added it to the SRC_URI_append_qemuall define in a linux-yocto_3.10.bbappend file, but the module wasn't enabled. Is there a good way to debug that? Apr 21 22:36:17 How is the recipe-skeleton/useradd/useradd-example.bb used? Where / how do I need to reference it to be picked up and added to the image? Apr 21 22:38:38 daren: Don't know exactly in your case, the following works for me: http://pastebin.com/0KaR59An Apr 21 22:39:59 volker: Yeah, that's what I did with the exception of SRC_URI was qualified with _append_qemuall Apr 21 22:40:29 daren: bitbake -c clean linux-yocto and -c configure and -c compile for debugging. Then find the .config file in build/tmp/work/qemux86_64-poky-linux/linux-yocto/ Apr 21 22:41:36 daren: I would try it first configured as in my example. if this works with -c clean & -c configure then I would modify it and see if it still works (trial and error wise ;-) Apr 21 22:42:48 volker: I will try your example. I should also have the .config file still in the tmp directory.... let me check Apr 21 22:43:13 daren: in my case it only overwrote the one line I configured, not the entire .config Apr 21 22:43:20 'find . -name .config' Apr 21 22:44:42 daren: what I learned with the kernel, it might be compiled as module but not automatically added as package. Check build/tmp/deploy///kernel-module* Apr 21 22:45:20 daren: in my case I didn't need to modify it as in the example I showed you. I just had to add the kernel-module package to my image. Apr 21 22:46:35 volker: I've tried searching for the module, but didn't find it. Doesn't look like my config line made it into the *.config file Apr 21 22:48:11 daren: k, can't help you future, not too familiar with it. I would also check the default setting in .config for that kernel driver Apr 21 22:49:43 volker: Thanks for your help. By default the setting is disabled (according to menuconfig). Once I get a change, I'll clean the kernel build and force a new one with your example, then mine. Apr 21 23:13:21 this is intereesting. I'm playing in dylan right now (a painful forward migration). Anyways, bitbake linux-yocto has a patch failure, says you can resolve patch failures using bitbake -c devshell linux-yocto... Then the devshell calls dopatch and fails with the same message. :( Apr 21 23:14:07 Prowell, which MACHINE? Apr 21 23:15:22 one of my own, that's what I'm trying to fix/bring upto date. What caught me by surprise is that devshell appears to require dopatch to be successful, but dopatch failures recommend running devshell to fix them. Apr 21 23:15:58 one out of two machines is moved forward to work in dylan, then it's straight on to dora. Apr 21 23:16:01 Prowell, that is odd. But you are using linux-yocto? the do_patch() there is a bit special Apr 21 23:16:12 Okay, there's probably bugs galore to be encountered, but pseudo can now have xattr support *and* still handle "cp -a" on CentOS machines that try to use system.posix_acl_access instead of chmod when xattr support is available. Apr 21 23:16:15 Yes, I am using linux-yocto. Apr 21 23:16:48 Prowell, are you making use of something like the emgd-1.18 feature branch? there was a known regression there Apr 21 23:16:53 which is being worked/fixed Apr 21 23:18:02 both machines are based off of /standard/common-pc/atom-pc. Apr 21 23:18:16 which kernel version? Apr 21 23:18:20 3.4 Apr 21 23:18:26 hrm Apr 21 23:19:03 should devshell be calling dopatch? Apr 21 23:19:06 if you can provide more detail on how do_patch() is failing, that would help. Should be some detail in the log. Apr 21 23:19:21 I honestly do not know how devshell interacts with linux-yocto - I don't use it. Apr 21 23:19:56 I use it sometimes, never had patch fail before so this poor interaction hadn't been noted before. Apr 21 23:49:15 right now, I'm trying to tease it out. I was more curious to know if it was a known issue that the hint for what to do is broken. Apr 21 23:52:05 and as far as the error: A meta series could not be created for blah blah blah/linux/(blah blah blah)/omc1734-standard.scc Apr 21 23:52:34 volker: yes, if you change package A and B depends on A, B will rebuild too Apr 22 01:54:22 well, found my issue with the upgrade. seems I managed to upcase a word in the scc file which ended up making a file not found. Still not sure if I should report a bug against bitbake because of the bad error message it gave. **** ENDING LOGGING AT Tue Apr 22 02:59:58 2014