=== mwhudson is now known as zz_mwhudson [04:27] Kernel Experts: [04:27] Could i get the exact kernel source git tree that was used to generate: trusty-preinstalled-boot-armhf+maguro.img [04:29] Maguro(samsung galaxy nexus) Kernel source tree is what i am looking for Ubuntu Touch. === zz_mwhudson is now known as mwhudson [07:41] moin === mwhudson is now known as zz_mwhudson === lag` is now known as lag [09:27] ppisati, moin [09:36] apw, If you ain't hearing anything your mumble is broken [09:36] or my ear is currently remote [09:36] * apw bets 10p on the latter [09:36] or that [09:37] apw needs ears on strings [09:41] hi, I run precise on armhf (beagleboard-xm) and need to add xenomai. I tried codesourcery but that doesn't do it. Linaro people directed me here. Can someone please help me out. Thank you. [09:44] cking, hey - which of your ppa's has the version of fwts with the stdout-summary fix? [09:47] https://launchpad.net/~colin-king/+archive/ppa, but only a trusty build [09:48] brendand, we hope to spin a 14.01.01 release to address this before the week is out [09:49] cking, ok [09:55] hvn3, not sure what you are asking really? [09:58] cking, when will 14.01.01 be out? [10:00] brendand, hopefully end of this week if it builds OK today [10:00] apw: I'm asking if someone here can direct me to info on cross-compiling a xenomai-patched precise-kernel to armhf [10:01] apw: it needs to run on omap3 [10:02] apw: the tools I have found so far only cross-compile to either armel or armv6, not to armhf and armv7 [10:02] hvn3: i'm reading the xenomai doc right now [10:02] hvn3: hold on [10:03] ppisati: ty [10:03] hvn3: you already did the "scripts/prepare-kernel.sh --linux=..." step, right? [10:03] hvn3: you just need to recompile a patched kernel, don't you? [10:04] ppisati: at got stuck at that point, because it tells me the kernel is unsupported. [10:05] ppisati: I used a vanilla-kernel btw, not a ubuntu kernel-source [10:05] hvn3: https://wiki.ubuntu.com/KernelTeam/ARMKernelCrossCompile [10:06] hvn3: FAQ section, point 6 [10:06] hvn3: make ARCH=arm omap2plus_defconfig [10:06] hvn3: edit .config to suit your need [10:07] hvn3: (you better try it with make ARCH=arm menuconfig since - i guess - xenomai adds some menu entries and you want to select that) [10:07] hvn3: and then [10:07] hvn3: make ARCH=arm CROSS_COMPILE=/usr/bin/arm-linux-gnueabihf- uImage [10:07] ppisate: I read that now....I stopped reading that because it mentioned omap4 [10:08] hvn3: when you did "scripts/prepare-kernel.sh --linux=..." did you use the --arch= switch? [10:08] hvn3: "3) Ok, but what about my Beagle/XM? How do i compile a kernel for an omap3 board?" [10:08] ppisati: does arch=arm differ from ARCH=arm ? [10:09] ppisati: I used arch=arm [10:09] ppisati: not in capitals [10:09] hvn3: ARCH=arm is used during the kernel compilation, --arch=arm is when you patch the kernel src tree with xenomai [10:09] hvn3: ok, there's some confusion let me recap... [10:10] hvn3: these are the steps that you should take to make it work: [10:11] hvn3: (since you are using a vanilla kernel) checkout a vanilla tree somewhere [10:11] hvn3: make ARCH=arm omap2plus_defconfig [10:12] hvn3: edit .config as stated in https://wiki.ubuntu.com/KernelTeam/ARMKernelCrossCompile, FAQ point 6, subsection b [10:12] hvn3: make ARCH=arm CROSS_COMPILE=/usr/bin/arm-linux-gnueabihf- uImage [10:12] hvn3: and follow the remaining step [10:13] hvn3: if it boots, then you have kernel tree and .config that you know it works with your hw and your userspace [10:13] hvn3, arch=arm and ARCH=arm are completly different to make [10:13] hvn3: after that (make a backup copy of .config) [10:14] hvn3: and follow the xenomai instructions [10:14] hvn3: scripts/prepare-kernel.sh --linux=$YOURKERNELTREE [10:14] hvn3: sorry, add the "--arch=arm" switch at the command above [10:15] hvn3: make ARCH=arm menuconfig and tweaks the config [10:15] ppisati: yes, I tried that and got error, but will try again [10:15] hvn3: and finally "make ARCH=arm CROSS_COMPILE=/usr/bin/arm-linux-gnueabihf- uImage" and follow the rest [10:16] ppisati: thank you for helping out so far [10:16] hvn3: first make sure you have a vanilla kernel AND a vanilla config working with your hw, then take the xenomai patches [10:16] hvn3: because if you start patching a kernel that you never tested, you'll never know if those patches are bad, the config was already broken, the kernel didn't support your board, etcetc [10:17] hvn3: baby steps [10:17] ppisati: I have the 3.2.0-55-omap kernel running right now....would it be safer to get a ubuntu-kernel source ? [10:23] hvn3: the kernel that you are going to use depends on xenomai (since you want to patch it with it later on), so which kernel does xenomai support? [10:24] cking, someones hooking me up with a test system - i'll let you know in a few hours if it works [10:25] brendand, ok, thanks [10:26] ppisati: as far as I see, the xenomai FAQ only mentions the hardware it supports, not kernels. So I assume it supports all recent kernels. [10:27] hvn3: i wouldn't be so sure, you better find out first [10:28] hvn3: and when you found out which kernel it supports, you make that version work (config and compilation wise) with your board [10:29] hvn3: and only after that, you apply the xenomai patches, reconfig and recompile [10:29] ppisati: ok, ty for the advise [10:32] ppisati: ty for your help and time [10:32] ppisati, do we still support linux-fsl-imx51 in lucid (i presume not) [10:42] apw: nope [10:42] apw: we dropped support for all the arm topic branches in L AFAIK [10:43] great [10:43] apw: cve? [10:45] yeah, just noting we are running processing on that branch for nothing, so getting rid of it === ricmm_ is now known as ricmm [11:27] cking, the fix looks good. cheers! [11:27] brendand, cool, thanks for letting me know :-) [11:54] infinity: There's a problem with the saucy -lowlatency. From what I can tell, it includes the right code, but the debian/changelog file only shows 3 master commits since 3.11.15 [11:59] zequence: Oops. Want to fix that up? [12:00] zequence: I'm in no rush to copy it over ASAP. [12:00] zequence, you hsould be able to reset to the previous lowlatency tag and remake it [12:11] apw: Ok. I'll try that [12:29] infinity: apw: I'm applying for upload rights to ubuntustudio-* packages. Also, linux-lowlatency, but I suppose that might not be necessary anymore. If anyone of you could endorse me that would be really helpful. micahg is the person who has been most active in sponsoring our packages, so he'll probably add a few words here too https://wiki.ubuntu.com/zequence/DeveloperApplication [12:30] I realize you don't know that much about my packaging skills, and tbh, I need a lot more practice, but the ubuntustudio-* packages are pretty straightforward, and I do maintenance on most of them [12:32] Things went better with saucy this time. I'm uploading soon [12:39] infinity: did you get the reject message too? [12:41] zequence: Why would I get reject messages on your uploads? [12:41] zequence: You didn't reuse the same version number did you? [12:43] zequence: You can't reuse versions, even if you delete the old one. Just bump the build number, so it's -7.10 [12:43] (Don't bother with a new changelog entry, just edit the top line and re-tag and rebuild the source) [12:44] 17.10, even. [13:03] infinity: Just guessing. LP might be able to send pigeon post until I know it doesn't ;) [13:04] infinity: It's up now. [13:07] zequence: Great, thanks. [13:08] does anyone know what is happening with the -proposed kernels? [13:08] are they waiting for 12.04.4 release? [13:35] brendand, that is my understanding yes, that they are all holding for the point release which in and of itself was delayed [13:36] * apw muses that if they hold much longer we might consider respinning them for stables throughout [13:40] apw, here is a 3.13 bug we ought to keep an eye on (from LKML): '[BISECTED] Linux 3.12.7 introduces page map handling regression' [13:40] smb, ^^ [13:40] rtg, looking [13:41] rtg, hm looking [13:41] dunno if it requires 3.13 host as well [13:45] Hm, yeah that sounds like something we had reports too and was somewhat related to anonymous pages (at least that flag seemed present most of the times) [13:47] apw, smb: here is another good one: 'fanotify use after free.'. Seems like 3.13 might have some gotchas for awhile. [13:48] rtg lets be glad we ahve a few stables before release [13:48] oh yeah [13:53] rtg, Uhm, well for those bad page thing... not sure how much of numa code was around then but we got a bug report unresolved for something looking reaaalyly similar for Lucid [13:53] Need to try that testcase there [14:03] apw, you (or your minion) are still planning to produce a low-latency kernel for Trusty, right ? [14:37] rtg, yeah, the question which has been raised is as to whether to pull it into master-next [14:37] apw, are there no extra patches now ? [14:38] apw, perhaps you should add this to the sprint agenda [14:38] there is one patch, but that changes the default for a configuration option [14:38] so we could carry that and it be only config them [14:38] rtg, already on there [14:38] apw, ack [14:39] i was presuming i would sort out the fat kernel and meta in a ppa, we could decide that on monday and upload them together or not [16:16] apw: rtg: did a rebase for manta (aosp 4.4.2) based, tested and working fine with the 4.4.2 port that I'm working on: http://kernel.ubuntu.com/git?p=rsalveti/ubuntu-trusty.git;a=shortlog;h=refs/heads/manta-kitkat [16:16] can any of you get this merged into the 'manta' branch and upload the package to the ppa? don't yet upload to the archive as we need to first switch the android code officially (like done with mako and flo) [16:17] rsalveti, acl [16:17] ack* [16:18] rtg: thanks [16:19] rtg: would also be nice to have the meta package in the ppa as well [16:20] rsalveti, I didn't think y'all were using a meta package [16:20] rtg: it's used by our android build system to find the package name (because of the abi) [16:20] rsalveti, ok, can do [16:20] awesome, thanks! [17:07] * apw reboots to test a grub2 beta [17:11] * apw is somewhat supprised to be back .. [17:12] cjwatson, just installed and booted my efi system (secure boot _off_) with your new grub2 from -proposed [17:12] cjwatson, all seems to be ok, only noticed a '*' marker on the menus, other than that it seemed the same [17:14] I think that was deliberate [17:15] it didn't seem like a bad thing, just noticed it [17:16] apw: yep, http://git.savannah.gnu.org/gitweb/?p=grub.git;a=commitdiff;h=a59a9826fd56efe350933ef73cba3a42b97a52b9 [17:21] cjwatson, it seems to let me select other kernels, and edit things, i have an issue with ctrl- not working in the exec screen but i seem to remember that was there before for me [17:22] cjwatson, but worth checking next reboot if ctrl-x works ok for you [17:28] apw: ok. does the help screen still advise ctrl-x? [17:28] yeah, the help is the same, and F12 or whatever it is works ok [17:28] i think it is a bios issue, as hitting C-x gives me an X [17:28] even outside grub? [17:29] only in grub [17:29] bit sceptical of that being a bios issue then [17:29] I vaguely recall some attempt to improve EFI keyboard handling but I can't find it now [17:29] i'll downgrade and confirm its there before either way === jono is now known as Guest52764 [17:30] ta [17:34] cjwatson, ok confirmed ctrl- issue is there in the 2.0.22 version as well [17:34] ah ok [17:34] so I guess that could just be crazy bios [17:34] thanks for the testing [17:34] yeah i suspect so, and it is a pre-release job, so i am not so concerned about it [17:34] rsalveti, manta -5.20 uploaded. I'll get the meta package in a bit [17:35] and i have "or Fn" keys which work just fine [17:35] rtg: great, mind also doing the meta package for flo? [17:35] rsalveti, np [17:35] awesome [18:57] I'm curious if anyone here has experience using amtterm to get console output on Ubuntu? [18:58] I have a couple of systems that have Intel AMT enabled; and I can use it successfully to remotely power on/off systems [18:58] but I haven't succeeded yet with getting console output [18:58] I've tried following http://manpages.ubuntu.com/manpages/trusty/en/man7/amt-howto.7.html [18:58] but that manpage talks about inittab, which, of course, doesn't exist in Ubuntu [19:02] cking: you seem like my most likely candidate to have tried ^ :-) [19:02] kirkland, i've not used that [19:03] cking: bummer [19:03] I have also not tried that [19:04] mjg59: wow, you either, huh? [19:05] I've played with most of the other features. And cried. [19:05] mjg59: it's a bit wonky, isn't it [19:05] where would I put "T2:2345:respawn:/sbin/getty ttyS2 115200 vt100-nav", if not inittab? I'm guessing somewhere upstarty... [19:06] kirkland: /etc/init/ttyS2.conf [19:06] kirkland: Compare to tty1.conf, etc [19:07] mjg59: excellent, thanks [19:08] * cking likes the man page section "More fun with AMT", yeah, right [19:09] cking: :-P [19:09] cking: self-extracting archive [19:09] fun as in "poking oneself in the eye with a red hot poker" [19:09] heh [19:10] cking: oh, btw, I encountered something very odd this week, something that gave me many hours of headache [19:10] cking: realtek USB ethernet adapters [19:10] cking: we had a bunch of them, all identical, some worked, some didn't [19:10] cking: ie, dmesg could see some, but not others [19:11] kirkland, duds or something more bizarre? [19:11] cking: perhaps more bizarre; not dudes [19:11] cking: after poking at it for several days, we noticed that the ones that didn't show up as NICs [19:11] cking: actually showed up as cdrom/optical devices [19:11] cking: when running windows, you get a popup, that a cdrom exists with some drivers you need to install [19:11] kirkland, there is some special magic that makes them switch mode, I can't recall what it is [19:12] cking: you do that, and then it clears some flag [19:12] cking: after which it looks like a nic again [19:12] cking: we couldn't figure out how to clear that flag in Linux (though we didn't try too hard) [19:12] cking: cool, so you've encountered these before [19:13] kirkland, something like usb-modeswitch, I've seen this on a USB 3G modem dongle [19:13] isn't it something like "ZeroCD" mode? [19:14] http://en.wikipedia.org/wiki/Virtual_CD-ROM_switching_utility perchance? [19:25] cking: interesting, I'll definitely try it if I get access to another one of these dongles [19:26] kirkland, try ejecting the CD rom device, that is commonly what occurs on the 3G modems [19:26] kirkland, you ask it to eject and that makes it go awayand the modem or whatever appear [19:28] that's a nice trick [19:33] apw: heh, cool :-) [19:48] cking, it makes some sense for sure === zz_mwhudson is now known as mwhudson === mwhudson is now known as zz_mwhudson === zz_mwhudson is now known as mwhudson