[00:02:05] *** perezd has quit IRC
[00:02:06] *** saintlou has joined #chromium-os
[00:02:06] *** ChanServ sets mode: +v saintlou
[00:02:36] *** oc80 has joined #chromium-os
[00:04:34] *** behdad has joined #chromium-os
[00:04:34] *** piman__ has joined #chromium-os
[00:04:34] *** deshantm_ has joined #chromium-os
[00:04:56] *** wfrichar has quit IRC
[00:07:33] *** quannnum_ has joined #chromium-os
[00:09:32] *** oc80z has quit IRC
[00:09:32] *** crosbot has quit IRC
[00:09:33] *** behdad1 has quit IRC
[00:09:33] *** deshantm has quit IRC
[00:09:33] *** piman_ has quit IRC
[00:09:33] *** achuith has joined #chromium-os
[00:09:33] *** ChanServ sets mode: +v achuith
[00:09:34] *** achuith has quit IRC
[00:09:34] *** achuith has joined #chromium-os
[00:09:34] *** adams.freenode.net sets mode: +v achuith
[00:09:34] *** quannnum has quit IRC
[00:10:47] *** BladeFreak has joined #chromium-os
[00:11:56] *** sbyer has quit IRC
[00:20:10] *** Styx has quit IRC
[00:23:28] *** ser has joined #chromium-os
[00:24:14] <ser> hey, which --board= should i choose to build for samsung chromebook?
[00:24:47] <adlr> x86-alex
[00:30:39] *** sbyer has joined #chromium-os
[00:30:39] *** ChanServ sets mode: +v sbyer
[00:30:53] <ser> adlr: thanks :-)
[00:33:21] *** sadrul has quit IRC
[00:44:52] *** nick761 has joined #chromium-os
[00:48:55] *** reveman_ has joined #chromium-os
[00:50:14] *** achuith has quit IRC
[00:50:14] *** jshin has quit IRC
[00:50:14] *** nick761_ has quit IRC
[00:50:14] *** pastarmovj has quit IRC
[00:50:14] *** kerz has quit IRC
[00:50:14] *** Justasic has quit IRC
[00:50:14] *** reveman has quit IRC
[00:50:19] *** reveman_ is now known as reveman
[00:51:37] *** Justasic has joined #chromium-os
[00:55:10] *** achuith has joined #chromium-os
[00:55:10] *** jshin has joined #chromium-os
[00:55:10] *** kerz has joined #chromium-os
[00:55:10] *** zelazny.freenode.net sets mode: +vvv achuith jshin kerz
[00:58:53] *** unreal has joined #chromium-os
[01:01:21] *** unreal_ has quit IRC
[01:04:21] *** unreal has quit IRC
[01:14:36] *** unreal has joined #chromium-os
[01:29:33] <gsam_> what's the ebuild magic to fetch a package from an external site?
[01:34:50] *** Adys has quit IRC
[01:41:42] *** deshantm_ has quit IRC
[01:41:58] *** deshantm has joined #chromium-os
[01:54:37] <steev_> gsam_: ebuild ebuidlname fetch, alternatively, emerge -f blah
[02:04:10] *** JasonO has joined #chromium-os
[02:04:10] *** JasonO has joined #chromium-os
[02:05:02] <gsam_> RESTRICT="mirror" was what i was looking for
[02:08:32] <steev_> ah, i misunderstood what you meant by fetch from an external site :)
[02:17:35] *** sadrul has joined #chromium-os
[02:17:35] *** ChanServ sets mode: +v sadrul
[02:43:44] *** arkaitzr has quit IRC
[02:44:18] *** arkaitzr has joined #chromium-os
[02:44:18] *** ChanServ sets mode: +v arkaitzr
[03:13:44] *** saintlou has quit IRC
[03:26:32] *** lipsinV2 has joined #chromium-os
[03:29:13] *** magn3ts has quit IRC
[03:29:38] <yusukes> sheriffs: sorry for the break. i'll revert my recent change against app-i18n/ibus-mozc-chewing shortly.
[03:32:24] <yusukes> done
[04:10:49] <ser>  ./image_to_usb.sh seeks for chromiumos_image.bin but ./build_image creates chromiumos_base_image.bin
[04:11:27] <ser> is it the same file? or something went wrong?
[04:12:56] <sosa> ./build_image defaults create both chromiumos_base_image.bin and chromiumos_image.bin
[04:13:40] <ser> ok so there is no chromiumos_image.bin :(
[04:14:36] <sosa> there should be one if you used the default flags
[04:15:21] <ser> i am issuing ./build_image --board=x86-alex --withdev --noenable_rootfs_verification --statefulfs_size=11046
[04:15:35] <sosa> should work
[04:15:55] <sosa> maybe build_image is erroring out in creating the devimage?
[04:15:58] <ser> but it does not :(
[04:16:06] <ser> checking the log, one sec
[04:16:25] *** cooled has quit IRC
[04:18:49] <ser> no, there is nothing suspicious on the screen during the ./build_image process
[04:18:54] <ser> hmm
[04:19:29] <ser> it is unfortunately a fedora machine not ubuntu i am using
[04:20:13] <sosa> well
[04:20:18] <sosa> at least you have the base image :P
[04:20:31] <sosa> you can use image_to_usb --image-name=chromiumos_base_image.bin
[04:20:39] <sosa> the base image is missing extra dev packages
[04:20:42] <ser> yes but it does not boot
[04:20:47] <sosa> oh?
[04:20:58] <ser> samsung refuses to see it as chromeos
[04:21:04] <sosa> oh of course
[04:21:13] <sosa> cause it's not :)
[04:21:16] <sosa> it's chromium os
[04:21:50] <sosa> you have to follow these directions to install a Chromium OS-based image on a Chrome OS device
[04:21:51] <sosa> http://dev.chromium.org/chromium-os/developer-information-for-chrome-os-devices/samsung-series-5-chromebook
[04:22:13] <ser> i have followed them precisely
[04:22:25] <sosa> so you get the Blue dev screen when you boot?
[04:22:30] <ser> yes
[04:22:35] <ser> and later it tries to boot
[04:22:44] <ser> but prints "no chrome disk"
[04:22:58] <ser> i am using an sd card
[04:23:01] <ser> not usb
[04:23:07] <ser> 32G sdhc
[04:23:12] <ser> maybe it is a reason?
[04:23:16] <sosa> ah, i'm not familar with the sd card process
[04:23:58] <ser> in theory it should work, my friend run it from sd
[04:24:11] <sosa> yeah
[04:24:11] <ser> but he had chromiumos_image.bin as well
[04:24:20] <sosa> it seems like your image is just bad beacuse build_image didn't complete
[04:24:27] <sosa> can you use pastebin to show me your log?
[04:24:44] <ser> yes i will try, wait a sec please
[04:34:19] <ser> http://lists.tldp.org/log.txt
[04:35:41] *** cooled has joined #chromium-os
[04:38:51] *** D|sT has quit IRC
[04:47:51] <sosa> i have to run soon, but build_image is definitely failing
[04:48:05] <sosa> you're getting:  mount: /home/ser/trunk/src/build/images/x86-alex/0.14.784.2011_07_21_0330-a1/rootfs not mounted already, or bad option
[04:48:12] <sosa> and it's cleaning up afterwards
[04:48:31] <ser> oops i missed it.
[04:48:32] <sosa> probably want to send an email to chromiumos-dev for help
[04:49:49] <ser> ok maybe i have screwed it because i did crtl-c during the first attempt
[04:50:04] <ser> checking mounted filesystems one sec
[04:51:23] <ser> yep. it looks it the reason, thanks for a hint - i will try to umount it
[05:04:10] *** D|sT has joined #chromium-os
[05:26:35] *** cooled has quit IRC
[05:27:45] *** deshantm is now known as deshantm_away
[05:40:14] *** JasonO has quit IRC
[05:44:12] *** cooled has joined #chromium-os
[06:04:39] *** behdad has quit IRC
[06:48:26] *** magn3ts has joined #chromium-os
[07:02:27] *** sergiu_ has joined #chromium-os
[07:02:27] *** sergiu_ has joined #chromium-os
[07:05:53] *** OtherJakeSays has joined #chromium-os
[07:08:27] *** JakeSays has quit IRC
[07:13:47] *** behdad has joined #chromium-os
[07:20:03] *** BladeFreak has quit IRC
[07:26:59] *** Styx has joined #chromium-os
[07:27:51] *** behdad has quit IRC
[07:59:00] *** sergiu has quit IRC
[08:22:07] *** stalled has joined #chromium-os
[08:32:17] *** Styx has quit IRC
[08:32:23] *** Styx__ has joined #chromium-os
[08:32:59] *** Styx__ is now known as Styx
[08:41:13] *** stalled has quit IRC
[08:52:43] *** stalled has joined #chromium-os
[09:07:37] *** vmil86 has joined #chromium-os
[09:12:57] *** Ethan has quit IRC
[09:13:18] *** jochen__ has quit IRC
[09:13:24] *** jochen__ has joined #chromium-os
[09:13:24] *** ChanServ sets mode: +v jochen__
[09:30:46] *** magn3ts has quit IRC
[09:30:53] *** jujugre has joined #chromium-os
[09:45:45] *** magn3ts has joined #chromium-os
[09:47:47] *** yoshiki has joined #chromium-os
[09:50:32] *** SoulRaven has joined #chromium-os
[10:32:24] *** superm1 has quit IRC
[10:34:25] *** superm1 has joined #chromium-os
[10:34:25] *** superm1 has joined #chromium-os
[10:54:29] *** vandebo has quit IRC
[10:54:37] *** vandebo has joined #chromium-os
[10:54:37] *** ChanServ sets mode: +v vandebo
[11:13:10] *** zork has quit IRC
[11:32:47] *** patcito has quit IRC
[11:37:10] *** eggy has quit IRC
[11:37:42] *** eggy has joined #chromium-os
[11:42:26] *** mascondante has joined #chromium-os
[11:46:25] *** mascondante has quit IRC
[11:48:35] *** mascondante has joined #chromium-os
[12:07:06] *** hashimoto has left #chromium-os
[14:49:55] *** sadrul has quit IRC
[15:05:10] *** rbyers has joined #chromium-os
[15:14:07] *** SoulRaven has quit IRC
[15:37:14] *** rharrison_chrome has joined #chromium-os
[15:37:14] *** ChanServ sets mode: +v rharrison_chrome
[15:44:28] *** sadrul has joined #chromium-os
[15:44:28] *** ChanServ sets mode: +v sadrul
[15:58:44] *** Adys has joined #chromium-os
[16:09:24] *** behdad has joined #chromium-os
[16:09:24] *** ChanServ sets mode: +v behdad
[16:17:26] *** Adys has quit IRC
[16:30:47] *** Adys has joined #chromium-os
[16:43:09] *** lipsinV2 has quit IRC
[16:43:12] *** lipsinV21 has joined #chromium-os
[16:44:47] *** lipsinV21 has quit IRC
[16:44:51] *** lipsinV2 has joined #chromium-os
[17:08:09] *** Styx has quit IRC
[17:08:58] *** cjb has quit IRC
[17:09:06] *** cjb has joined #chromium-os
[17:40:19] *** cros_ has joined #chromium-os
[17:40:59] <cros_> Hi folks, having a heck of a time with the portage build system.
[17:41:07] <ellyjones> hi
[17:41:09] <ellyjones> howso?
[17:41:46] <cros_> ellyjones: I've been working with it for a few weeks but still too much of a mystery. First, when I run build_packages how does it determine what packages(ebuilds) to build?
[17:42:12] <ellyjones> I think it just builds the 'chromeos' package and the transitive closure of its dependencies :)
[17:43:50] <cros_> ellyjones: Thanks, that's the best I could gather. Now, if I modify one particular package how do I make sure it builds with my changes? I used the cros_workon / cros_workon_make approach to do this.
[17:43:54] *** saintlou has joined #chromium-os
[17:43:54] *** ChanServ sets mode: +v saintlou
[17:45:16] <cros_> ellyjones: When it comes time to include the package in my rootfs image it seems there are several ways to do this. I tried a build_image and an emerge -k --root but the resulting image seems to hang.
[17:46:21] <ellyjones> is your package depended upon (transitively) by chromeos?
[17:47:18] <cros_> Yes, the chromeos-wm package.
[17:47:29] <ellyjones> hm, okay
[17:47:37] <ellyjones> just build_packages then build_image should work then
[17:47:54] <ellyjones> can you ssh to it? (if you modded it for test)
[17:47:58] <cros_> ellyjones: All I did was add 1 log statement. Even after I backed it out init just seems to hang with no console messages.
[17:48:32] <ellyjones> odd.
[17:49:16] <cros_> ellyjones: Let me backup a little bit. I notice there is some way the build system knows what snapshot of the source tree I have and based on that it download the correct version of prebuilt binaries at build time.
[17:49:50] <ellyjones> it knows because the ebuilds it have revisions, I think
[17:51:24] <cros_> ellyjones: I've seen a URL stored in prebuilt.conf in one of the overlays. It contains some version made up of the date and some unique number and during the build it spits that number out.
[17:52:03] * ellyjones shrugs
[17:52:19] <ellyjones> we are kind of out of my depth here; try asking powderluv or sosa perhaps :)
[17:54:43] <cros_> ellyjones: Heh, yeah a bit in the weeds. I'm just trying to demystify things to try and understand what's happening with my builds. Are you familiar with the window manager?
[17:54:56] <ellyjones> not even slightly
[17:55:00] * ellyjones is a kernel hacker mostly :)
[17:55:10] <cros_> ellyjones: Ha, got you. Thanks for the help!
[18:00:18] *** ScriptRipper has quit IRC
[18:00:34] *** ScriptRipper has joined #chromium-os
[18:02:22] *** jujugre has quit IRC
[18:06:29] <cros_> Another question for anyone out there. So even with the full layout I don't get the source for every package that goes into chromiumOS. It seems some of them get downloaded at some point during the build. Is there any way to see the source? x11 is such a package.
[18:06:37] <zbehan> cros_: if build_packages hangs, it could be a bug in your ebuild that you're trying to add in?
[18:07:14] <cros_> zbehan: build_packages does not hang when I try to run the resulting /sbin/init it hangs without any messages from init.
[18:07:45] <cros_> zbehan: I'm also not adding in an ebuild, it is an existing package that I just modified to add some debug statements to.
[18:08:10] <zbehan> i see
[18:09:16] <zbehan> cros_: full layout consists of chromium-os specific projects, we still use lots of third-party packages, for which the source is freely available through the sites but not part of the checkout, google simply provide a distfiles mirror of snapshots of the source tarballs, which are located in /var/lib/portage/distfiles-target/, and you can unpack them anywhere and have a look
[18:15:05] *** wfrichar has joined #chromium-os
[18:15:05] *** ChanServ sets mode: +v wfrichar
[18:28:38] *** cjb has quit IRC
[18:28:45] *** cjb has joined #chromium-os
[18:29:27] *** jrbarnette has joined #chromium-os
[18:38:09] *** lipsinV2 has quit IRC
[18:39:52] *** mtennant has joined #chromium-os
[18:39:52] *** ChanServ sets mode: +v mtennant
[18:40:41] <mtennant> All: I am one of the build sheriffs on duty now.
[18:43:49] <darinski> mtennant: a good sheriff project -- figure out how to enforce tree closures on gerrit and prevent change submission :-)
[18:45:13] <mtennant> That's more than a "sheriff" project.  More like an FBI agent project.
[18:45:33] <mtennant> But not a bad idea.
[18:45:55] <darinski> I'm not saying it needs to be complete in 2 days :-)
[18:54:28] *** sergiu has joined #chromium-os
[19:06:06] *** jrbarnette_ has joined #chromium-os
[19:07:55] <kliegs> cros_: If you're still trying to find the source, you can do a FEATURES=noclean on your build. The source files used to build each package will leave in a portage temp directory then and you can view what's being built there
[19:08:46] <kliegs> Unfortunately if you want to change the source for packages that have no forked version in Chromium OS you'd need to download the source from the appropriate upstream location and (I think) tweak the ebuild to build from the location you downloaded the source to
[19:09:47] *** jrbarnette has quit IRC
[19:09:50] *** vmil86 has quit IRC
[19:10:05] <kliegs> mtennant, jglasgow, sheriffs: I have a change to submit that removes a repo from the manifest.  It will cause a brief tree failure on the first run and then work when reapplied
[19:10:14] <kliegs> Is now a good time to work with you to commit it?
[19:11:05] <ellyjones> jglasgow is doing an interview
[19:11:08] *** jrbarnette_ has quit IRC
[19:11:18] <ellyjones> I dunno, mark the tree as throttled and go for it?
[19:11:34] <darinski> kliegs: is there no way to avoid failures?
[19:11:35] <kliegs> I'd rather wait on a sheriff to be present for it
[19:11:55] <kliegs> darinski: there is no way to do so - its a known issue in our buildbots that removing a repo from a manifest causes a single red build
[19:12:07] <darinski> single or all of them?
[19:12:46] <kliegs> I don't think it impacts the full builders, just the preflights
[19:12:54] <kliegs> and possibly only the master pfq, not each slave
[19:13:37] <kliegs> well, the internal preflight bot in this case.
[19:13:51] <darinski> that's pretty bad...
[19:14:44] <darinski> i have trouble saying "yeah, go ahead and make the waterfall red for the day" :-)
[19:14:46] *** radhermit has quit IRC
[19:15:31] <kliegs> it should be <30 minutes
[19:15:33] *** radhermit has joined #chromium-os
[19:15:59] <kliegs> and hadn't realized you were petkov - I'm not good at translating ID's into irc
[19:19:50] <mtennant> Right, petkov is sheriff, too
[19:20:10] <mtennant> I suppose if such a change has to be done sometime we can do it now.
[19:20:14] <darinski> asking around...
[19:20:31] <mtennant> But maybe we can stop other commits in the meantime... mark tree as "closed" for others
[19:20:35] *** powderluv has joined #chromium-os
[19:20:36] *** ChanServ sets mode: +v powderluv
[19:20:46] <kliegs> there's no need to stop other commits really
[19:21:14] <mtennant> we'd have no way to know if bad commits are going in
[19:21:39] <cmasone> close the tree, land the change, and wait for it to clear...
[19:21:57] <kliegs> well, I should say stop other commits until its time to go in
[19:22:27] <kliegs> any bad changes will trigger on the next build. but yes - this commit does need to get in at some point
[19:22:29] <darinski> yes, we should definitely close the tree before commiting the CL and wait for it to clear
[19:22:41] <cmasone> darinski: just do it.  Now's as good a time as any
[19:23:17] <darinski> cmasone: maybe... friday night also works :-)
[19:23:26] <ellyjones> that's my line
[19:23:37] <ellyjones> kliegs: just land it, there's nothing magical about sheriffs :P
[19:23:41] <cmasone> sure, then when something goes wrong, we can keep the tree nice a closed till 10AM PST monday :-)
[19:24:09] <kliegs> ellyjones: I should still respect the sheriffs.  its also good to coordinate through them in case someone else is doing something similar
[19:24:40] <kliegs> darinski: To confirm - its fine with me doing this now?  its really minor and when 95% of the people won't notice
[19:24:51] <darinski> ok, whatever... close the tree and push... this is messed up and wrong though
[19:25:23] <kliegs> darinski: Its a minor closure that happens rarely when we clean up the manifests
[19:25:26] <cros_> zbehan: Thanks, for now I am trying to modify a package for which there is source but I was wondering about third party packages as well and you answered that.
[19:25:38] <cros_> kliegs: How do I pass FEATURE=noclean when using build_packages?
[19:25:41] <cmasone> darinski: at this point, we've probably wasted more time talking about it than will be wasted by just doing it :-)
[19:25:45] <kliegs> darinski: but thank you - starting now
[19:25:54] <kliegs> cros_: FEATURES=noclean ./build_packages
[19:26:00] <kliegs> or you can export FEATURES=noclean and then run it
[19:26:04] <cros_> kliegs; Much appreciated!
[19:26:06] <darinski> maybe... or maybe we should fix the issue by fixing the bots
[19:26:10] <kliegs> its just an environment variable to set - be warned - it will leave a lot of junk around
[19:26:15] <darinski> davidjames says its not too hard to fix...
[19:26:25] <cmasone> darinski: sounds great :-)
[19:26:59] <kliegs> darinski: Not arguing it shouldn't be fixed, just for a relatively minor occurance probably not high on the priority list
[19:27:00] <darinski> cmasone: but that we've let it go, we won't fix it :-)
[19:27:27] <powderluv> darinski: I missed the context of the requirement to restart the bots. please file a tracker item
[19:27:31] <powderluv> and we can fix it
[19:27:33] <cmasone> darinski: you could fix it, right?
[19:28:01] <darinski> cmasone: yeah, right...
[19:28:35] <ellyjones> does anyone know if you can create custom per-project presubmits now that we're using gerrit? I'd like to force a unit test run before uploading or something
[19:28:35] <mtennant> kliegs is going to go ahead with this commit, and tree will be closed for a brief period.  Should only be minutes.
[19:28:43] <darinski> powderluv: i've asked davidjames to file an issue and assign to you. i don't understand the issue enough to describe the problem
[19:28:58] <ellyjones> we have now spent nearly 15 minutes talking about a 5-minute tree closure :P
[19:29:20] <darinski> ellyjones: unless it ends up not being a 5 minute tree closure...
[19:29:29] <ellyjones> yeah, it's true
[19:29:43] <darinski> ... and it's not a change that you can just revert if things go wrong...
[19:29:59] <darinski> anyway... we're going ahead... everybody, keep coding, stop slacking
[19:30:21] <kliegs> one slight delay - had to rebase the CL
[19:31:27] *** jrbarnette has joined #chromium-os
[19:31:27] *** ChanServ sets mode: +v jrbarnette
[19:33:46] <kliegs> unfortunately in the time it took to debate this the internal pfq started running on its own for what looks like a non-cros commit
[19:34:06] <kliegs> davidjames: ping - should I let it finish then go through the restart process?  and is it safe to open the tree in the meantime?
[19:34:29] <davidjames> kliegs: You need to make the bot fail somehow, just stop the builder, that'll fail it :)
[19:34:48] <davidjames> kliegs: Once the bot is stopped, then reopen the tree, and just watch for it to go green :)
[19:35:14] <kliegs> it doesn't need to fail on the specific CL I put in?  Any failure causes it to work correctly?
[19:35:17] <kliegs> or should I double-fail it?
[19:35:48] <davidjames> kliegs: The reason it needs to fail is to stop the tree from closing automatically when it fails later :)
[19:36:49] <kliegs> Oh. it just failed itself on its own
[19:36:51] <davidjames> kliegs: Great now that it's failed you can open the tree. It'll fail 1 more time when it pulls in your change, but no problem, we know about that already
[19:36:53] <kliegs> i'll reopen and restart it
[19:37:06] <davidjames> kliegs: Ah, good, it failed on your change :)
[19:37:12] <kliegs> the one that failed already seemed aware of my change. that kinda worries me
[19:37:23] <davidjames> kliegs: Yeah that's cause it ran repo sync :)
[19:37:34] <kliegs> oh. my change got in before the repo sync phase
[19:37:40] <kliegs> davidjames: so its good to open now and we should stay green?
[19:37:46] <davidjames> kliegs: Yup
[19:37:59] <kliegs> ok done.
[19:38:06] <kliegs> davidjames: thanks for the assistance/sanity checking
[19:38:28] <kliegs> mtennant, darinski: thanks for the assistance.  tree should be back to normal now
[19:39:18] <mtennant> thanks
[19:40:03] <cjb> Hi all.  Could someone tell me how to build an individual package inside the chroot capturing all the gcc invocations and output, please/
[19:40:04] <cjb> ?
[19:40:15] <ellyjones> cjb: emerge-$board package ?
[19:42:53] <cjb> ellyjones: thanks, perfect -- am not familiar with Gentoo, obviously :)
[19:47:40] <darinski> x86-generic pfq is about to fail, i guess
[19:48:18] <darinski> checking with zel to see who can investigate the chrome crash...
[19:50:42] *** powderluv has quit IRC
[19:50:53] *** powderluv has joined #chromium-os
[19:50:53] *** ChanServ sets mode: +v powderluv
[19:54:48] <powderluv> cjb: you may have to go deeper if you want the gcc invocations :)
[19:54:56] <ellyjones> read build.log!
[19:54:58] <powderluv> maybe replace the gcc binary with your wrapper
[19:55:30] <powderluv> so you will only get gcc/g++ invocations and not deal with all the other build logs
[19:55:59] <cjb> powderluv: nah, they showed up okay, thanks
[19:56:16] <powderluv> cool
[19:56:28] <powderluv> curious what you are debugging with it
[19:57:55] <cjb> I'm trying to bring up chromiumos on a new device, but I'm getting SIGILL on every compiled binary on my platform when built against arm-generic
[19:58:21] <cjb> I've been investigating the make flags, and I think I just found out that -march=armv7-a (used by arm-generic) causes the SIGILL, and it goes away if you cahnge to -march=armv7
[19:58:39] <cjb> this doesn't really tell me what the problem is, but it's enough to make some progress
[19:59:33] <kliegs> cjb: you can try checking the floating point settings set in the make.conf files
[19:59:51] <kliegs> look at ARM_FPU and MARCH_TUNE values
[20:00:09] <cjb> kliegs: yes, did those too.  They didn't affect the SIGILL, but -march did.
[20:01:10] *** cros_ has quit IRC
[20:09:36] *** Fractal has joined #chromium-os
[20:15:19] *** mtennant has quit IRC
[20:17:30] *** Styx has joined #chromium-os
[20:38:34] <zbehan> cjb: what about binaries you compile by hand? a simple hello world program?
[20:39:16] <ellyjones> or run one under gdb and see which instruction is giving you SIGILL
[20:41:28] *** leavittx has joined #chromium-os
[20:41:57] <Fractal> Hi, is it possible to build a private server and make Chromium OS connect and synchronize everything with this server?? (instead of a Google account on Google servers)
[20:44:51] *** leavittx has quit IRC
[20:47:53] *** ScriptRipper has quit IRC
[20:47:53] *** ScriptRipper has joined #chromium-os
[20:54:11] *** Inumedia has quit IRC
[20:56:16] <kliegs> Fractal: you mean for user authentication or for OS updates?
[20:59:18] <ellyjones> it sounds like they mean for chrome sync and stuff
[20:59:41] *** leavittx has joined #chromium-os
[21:05:49] <cjb> rats.  my SIGILL goes away if I switch from armv7-a to armv7, but then media-o3d and chromeos-chrome fail to build.
[21:07:07] <Fractal> kliegs: user authentication and synchronization of chrome stuff like settings, bookmark, passwords, apps, etc.
[21:09:28] <Fractal> (by apps I mean the list of installed apps, not the apps? data, that would be a different question)
[21:10:39] *** Inumedia has joined #chromium-os
[21:10:39] *** Inumedia has quit IRC
[21:10:39] *** Inumedia has joined #chromium-os
[21:11:02] <kliegs> Fractal: that I'm not sure about - I haven't played with any of the sync/auth stuff.  I suspect its technically possible but not sure how much work would be involved
[21:15:50] *** Inumedia has quit IRC
[21:17:45] <Fractal> kliegs: and do you know if the protocol and server software used for synchronization are open source?
[21:18:29] <kliegs> Fractal: I don't, sorry.  The sync code is part of chromium I believe so you might want to check in #chromium
[21:18:51] <kliegs> I knew a bit about how autoupdate for builds works so if you were looking at that I could have helped
[21:18:53] <redpig> Fractal: the client code will be in the source for chromium and chrome os. At present you can't supply alternate endpoints without changing the code
[21:25:41] <Fractal> redpig: kliegs: ok, thanks, I?ll try to see if I find something interesting in the chromium source code :)
[21:25:47] <Fractal> I have another question
[21:27:08] <Fractal> If I build a custom Chromium OS image, run in on my Samsung Chromebook and run the "chromeos-install" command, will it wipe the whole SSD or only the relevant partitions? (I have Ubuntu installed in KERN_C and ROOT_C and I don?t want to wipe it)
[21:29:19] *** Inumedia has joined #chromium-os
[21:32:35] <Fractal> Or can I just dd the partitions from the USB key onto KERN_{A,B} and ROOT_{A,B} instead of "chromeos-install"? Will it work?
[21:34:26] <cjb> Does anyone know how to disable NEON support for the arm-generic builds?
[21:35:44] <cjb> I tried changing ARM_FPU from neon to vfpv3-d16, and removing "-opengl opengles neon" from the use flags
[21:35:55] <cjb> but chrome still tries to build arm-neon.h, which fails
[21:45:03] *** patcito has joined #chromium-os
[21:45:27] <powderluv> cjb: can you check the chrome ebuild
[21:45:44] <powderluv> i think there is a gyp define for it and can be toggled
[21:45:54] <redpig> Fractal: it'll repartition
[21:46:02] <redpig> you can just dd the partitions you want
[21:46:09] <redpig> KERN-A and ROOT-A into place manually
[21:48:03] <cjb> powderluv: ah, I don't know how to do that, sorry
[21:48:15] <cjb> powderluv: any chance of a pointer?
[21:49:18] <Fractal> redpig: Thanks :)
[21:50:50] <cjb> powderluv: I see, the ebuild is in ~/trunk/src/third_party/chromiumos-overlay/chromeos-base/chromeos-chrome/
[21:50:55] <cjb> but no hits for neon in that file
[21:50:55] <powderluv> yes
[21:51:04] <powderluv> look for armv7=1
[21:51:18] <cjb> found that
[21:51:25] <cjb> but I have an armv7 cpu :-)
[21:51:27] <cjb> just no neon
[21:52:14] <powderluv> say arm_neon=0
[21:52:22] <cjb> ooh, thank you
[21:52:26] <powderluv> let me me know how it goes
[21:52:35] <Hexxeh> cjb: What board/device are you using?
[21:52:48] <cjb> I think I need to fix up o3d first, perhaps, since it's broken and depended on by chrome
[21:52:55] <cjb> Hexxeh: the new OLPC laptop, XO-1.75
[21:53:09] <cjb> (Marvell MMP2 SoC, PJ4 CPU)
[21:53:15] *** saintlou has quit IRC
[21:53:27] *** saintlou has joined #chromium-os
[21:53:27] *** ChanServ sets mode: +v saintlou
[21:53:29] <Hexxeh> Ah, nice
[21:56:12] *** mtennant has joined #chromium-os
[21:56:12] *** ChanServ sets mode: +v mtennant
[21:56:52] <kliegs> cjb: in the make.conf file look at EXTRA_BUILD_ARGS
[21:57:11] <kliegs> those are options passed to the Chrome build (when it uses gyp to generate makefiles)
[21:57:23] <kliegs> Some arm builds have arm_neon=1 set, so you'd want to unset it
[21:57:27] <kliegs> I'm not sure of the default unfortunately
[21:59:57] <cjb> kliegs: the overlay-arm-generic/make.conf doesn't have an EXTRA_BUILD_ARGS
[22:00:31] <kliegs> cjb: it might be worth adding one then as:  EXTRA_BUILD_ARGS="arm_neon=0"
[22:00:37] <kliegs> that would shut off neon for chrome
[22:01:02] <cjb> thanks!
[22:01:04] * cjb tries.
[22:01:46] <kliegs> cbj: your welcome - good luck!  You can try looking at some of the other arm packages for common flags (overlay-tegra2 for example)
[22:03:54] <cjb> oh!  yes, tegra2 is much closer to my hardware than arm-generic!
[22:04:42] <cjb> hm, chrome build failed differently:
[22:04:50] <cjb> {standard input}:332: Error: selected processor does not support `pkhbt r5,r4,r5,lsl#16'
[22:05:05] <cjb> building third_party/skia/src/opts/SkBitmapProcState_opts_arm.cpp
[22:05:13] <cmasone> cjb: maybe try building the tegra2overlay :-)
[22:05:20] <cjb> yeah, was just thinking that :-)
[22:05:28] <cjb> although it still uses -march=armv7-a
[22:05:47] <cjb> which I think I've determined gets me SIGILL.  (though using armv7 instead gets me these assembler errors..)
[22:06:19] <cmasone> cjb: well, maybe make your  own board using the tegra2 overlay as a jumping off point, if it's closer to your needs than arm-generic
[22:06:36] *** sbyer is now known as sbyer-at-lunch
[22:07:00] <cjb> yup, sounds right
[22:07:02] <kliegs> cjb: You might want to make sure you do a clean build just in case some flags didn't clear object staleness
[22:07:14] <cjb> kliegs: good idea, ok
[22:07:15] <kliegs> from inside the chroot, rm -rf /var/lib/portage/distfiles-target/chrome*
[22:07:23] <cjb> I'm doing a tegra2 overlay build now, so that'll be clean
[22:07:26] <kliegs> that will clear out any chrome source that's been downloaded
[22:07:34] <cjb> ok
[22:07:51] <kliegs> cjb: it will be mostly clean.  the actual directory of the chrome code that gets checked out is shared between different boards
[22:08:04] <kliegs> but it gets build in a different out directory (out_tegra2 vs out_arm-generic)
[22:09:07] <cjb> I see
[22:21:43] *** saintlou has quit IRC
[22:24:10] *** saintlou has joined #chromium-os
[22:24:10] *** ChanServ sets mode: +v saintlou
[22:27:20] *** JasonO has joined #chromium-os
[22:27:20] *** JasonO has joined #chromium-os
[22:27:49] *** unreal has quit IRC
[22:29:50] *** unreal has joined #chromium-os
[22:31:48] *** behdad has quit IRC
[22:31:50] <Hexxeh> Where is the mapping from keymap files to friendly names shown on the Welcome screen stored?
[22:36:19] *** kochi has quit IRC
[22:42:36] *** JasonO has quit IRC
[22:43:23] *** JasonO has joined #chromium-os
[22:46:52] *** behdad has joined #chromium-os
[22:56:35] *** rharrison_chrome has quit IRC
[22:58:26] *** jrbarnette_ has joined #chromium-os
[22:58:26] *** ChanServ sets mode: +v jrbarnette_
[22:58:39] *** jrbarnette_ has quit IRC
[22:59:44] *** jrbarnette has quit IRC
[23:01:07] *** saintlou has quit IRC
[23:04:10] *** leavittx has quit IRC
[23:05:57] <Fractal> Is the Chromium OS kernel using an initramfs?
[23:06:19] <adlr> for normal boots, no
[23:09:33] <Fractal> adlr: But there are boots using an initramfs? What is the chromeos-base/chromeos-initramfs package for?
[23:09:40] *** saintlou has joined #chromium-os
[23:09:40] *** ChanServ sets mode: +v saintlou
[23:09:59] <adlr> in some cases (i think relating to installation, but perhaps there are others?) we do use an initramfs
[23:09:59] *** OtherJakeSays is now known as JakeSays
[23:10:12] <gauravsh> it's used by the recovery and factory installer
[23:10:23] <gauravsh> I don't believe it is used anywhere else
[23:12:41] *** Sith has joined #chromium-os
[23:13:07] *** Sith is now known as Guest85737
[23:13:56] *** JasonO has quit IRC
[23:15:49] *** Guest85737 has quit IRC
[23:16:17] *** JasonO has joined #chromium-os
[23:16:17] *** JasonO has joined #chromium-os
[23:16:59] *** jrbarnette has joined #chromium-os
[23:16:59] *** ChanServ sets mode: +v jrbarnette
[23:17:07] *** jrbarnette has quit IRC
[23:17:40] *** jrbarnette has joined #chromium-os
[23:17:40] *** ChanServ sets mode: +v jrbarnette
[23:19:14] *** rbyers has quit IRC
[23:20:55] *** leavittx has joined #chromium-os
[23:22:46] *** Styx has quit IRC
[23:23:11] *** kochi has joined #chromium-os
[23:24:23] <darinski> why didn't the tree close automatically on arm-generic?
[23:27:36] *** kochi has quit IRC
[23:35:23] *** superm1 has quit IRC
[23:35:25] <mtennant> Don't know.  Just saw that you did that.
[23:36:11] *** behdad has quit IRC
[23:38:35] <ojn> arm-generic is a secondary citizen
[23:38:41] <ojn> I think... at least some builders are.
[23:38:49] <ojn> so they just fail and don't close the tree
[23:45:54] *** sbyer-at-lunch is now known as sbyer
[23:45:58] <davidjames> All external binary builders *should* close the tree if they fail
[23:46:29] <davidjames> They are 1st class citizens (and very important) because preflight queue won't push any changes until they pass
[23:47:20] <davidjames> So yeah, that includes x86-pineview-binary, x86 generic pre flight queue, arm generic binary, and arm tegra2 binary
[23:49:58] *** behdad has joined #chromium-os
[23:49:58] *** ChanServ sets mode: +v behdad
[23:50:17] *** falken has quit IRC
[23:52:12] <kliegs> davidjames: I seem to recall seeing a CL to remove the arm builders from the list a bit ago. but don't remember if it went through or not (within the last month)
[23:52:25] <davidjames> kliegs: Yeah they were removed temporarily and then added back
[23:52:41] <kliegs> ahh. ok.  at least I'm not imagining things i've seen





top