**** BEGIN LOGGING AT Mon Jan 23 02:59:57 2012
Jan 23 09:07:59 <Jay7> Shane Wang proposed kexec-tools 2.0.3 into oe-core
Jan 23 09:08:07 <Jay7> but seems unrefactored
Jan 23 09:35:11 <Jay7> ant_work: hi
Jan 23 09:35:14 <Jay7> [13:04] <Jay7> Shane Wang proposed kexec-tools 2.0.3 into oe-core
Jan 23 09:35:16 <Jay7> [13:05] <Jay7> but seems unrefactored
Jan 23 09:35:22 <ant_work> hi Jay7
Jan 23 09:35:34 <Jay7> ant_work: try to propose latest version you made
Jan 23 09:35:56 <ant_work> isn't the fix already in?
Jan 23 09:39:14 <Jay7> http://patches.openembedded.org/patch/19945/
Jan 23 09:39:18 <Jay7> here is proposed patch
Jan 23 09:40:16 <Jay7>  PACKAGES =+ "kexec kdump"
Jan 23 09:40:47 <Jay7> iirc, you are removed kexec-tools in favor of separate kexec and kdump
Jan 23 09:40:53 <Jay7> am I wrong?
Jan 23 09:44:12 <ant_work> it's ok
Jan 23 09:44:27 <ant_work> one recipe creates 2 packages
Jan 23 09:44:41 <ant_work> with ^^^ line
Jan 23 09:45:27 <Jay7> ah.. I've thought you removed kexec-tools
Jan 23 09:46:08 <ant_work> in fact we should rename the dir  /kexec  -> /kexec-tools
Jan 23 09:46:11 <ant_work> :)
Jan 23 09:46:39 <ant_work> project name changed long ago
Jan 23 09:50:24 <ant_work> Jay7: btw
Jan 23 09:50:25 <ant_work> After afternoon tea, I watched two talks on bootloaders. The first, on using Linux as a bootloader (given by Peter Chubb), was interesting, though was really only useful as a method in a small number of cases. Most systems are better off sticking to something like GRUB
Jan 23 09:50:39 <ant_work> (linux.conf.au 2012)
Jan 23 09:51:46 <Jay7> look at grub as very small linux kernel ;)
Jan 23 09:54:24 <ant_work> http://www.youtube.com/watch?v=pteHg54WBbQ
Jan 23 09:54:26 <ant_work> :)
Jan 23 09:54:46 <ant_work> add it to the site!
Jan 23 09:57:07 <ant_work> hi bluelightning
Jan 23 09:57:13 <bluelightning> hi ant_work
Jan 23 09:57:26 <ant_work> http://www.youtube.com/watch?v=pteHg54WBbQ
Jan 23 09:57:34 <ant_work> sorry
Jan 23 09:57:57 <bluelightning> ant_work: did you see this? http://lwn.net/Articles/475590/
Jan 23 09:58:07 <bluelightning> ant_work: some hope for offline charging :)
Jan 23 09:58:48 <ant_work> interesting
Jan 23 09:59:13 <ant_work> btw I'm pretty sure with 3.x corgi and spitz did recharge..at least the led was steady orange
Jan 23 10:00:15 <ant_work> we should send that link to our kernel hackers
Jan 23 10:04:14 <ant_work> he he
Jan 23 10:04:30 <ant_work> eter Chubb used collie to how kexecboot
Jan 23 10:06:51 <ant_work> Jay7: undoubtely the biggest propaganda we ever had!
Jan 23 10:07:02 <Jay7> ant_work: :)
Jan 23 10:07:04 <Jay7> hi bluelightning
Jan 23 10:07:15 <bluelightning> ant_work: :)
Jan 23 10:07:16 <bluelightning> hi Jay7
Jan 23 10:08:37 <ant_work> heh, the video is an important how-to for beagle users...
Jan 23 10:08:48 <Jay7> ant_work: is kexecboot noted there? :)
Jan 23 10:24:46 <ant_work> yes, around 11:00
Jan 23 10:25:38 <ant_work> schematics  at 13:00
Jan 23 10:27:41 <ant_work> thenbad notes about broken kexec on opamp's
Jan 23 10:37:30 <Jay7> ant_work: imho, we should release 0.6
Jan 23 10:37:39 <Jay7> have you any pending ideas for it?
Jan 23 10:37:47 <Jay7> how fast is interface/
Jan 23 10:37:53 <ant_work> yes...still waiting klibc with ubi
Jan 23 10:38:22 <ant_work> we could try to copy from util-linux-ng
Jan 23 10:38:27 <ant_work> as Maks started to do
Jan 23 10:38:37 <Jay7> I just want to implement loop+switch_root method
Jan 23 10:38:45 <ant_work> but can be added with 0.7, no problems
Jan 23 10:38:56 <ant_work> ah, that will be for 1.0 ;)
Jan 23 10:39:18 <Jay7> yeah.. and may under different name :)
Jan 23 10:39:23 <Jay7> execboot :)
Jan 23 10:39:53 <ant_work> can you execve another binary afte kexecboot?
Jan 23 10:40:00 <Jay7> yes
Jan 23 10:40:04 <ant_work> hm..
Jan 23 10:40:11 <ant_work> good
Jan 23 10:40:23 <Jay7> good idea, btw
Jan 23 10:40:28 <Jay7> another method - exec :)
Jan 23 10:40:35 <ant_work> exactly
Jan 23 10:41:15 <ant_work> anyway, adding an initrd containing another /init should also worl
Jan 23 10:41:21 <ant_work> *work
Jan 23 10:43:42 <ant_work> bluelightning: about having kerenl in boot, maybe we should force that for zaurus too
Jan 23 10:43:55 <ant_work> i.e. core-image-minimal lacks kernel atm
Jan 23 10:44:11 <bluelightning> depends, if it's needed for booting or not
Jan 23 10:44:19 <ant_work> once you'll apply my latest patchset, we'll have
Jan 23 10:44:25 <ant_work> MACHINE_ESSENTIAL_EXTRA_RRECOMMENDS_append = " kexecboot-cfg "
Jan 23 10:44:30 <ant_work> and maybe  +linux
Jan 23 10:44:43 <ant_work> as you did  MACHINE_ESSENTIAL_EXTRA_RDEPENDS = "kernel ipaq-boot-params"
Jan 23 10:45:06 <ant_work> zaurus.conf seems overcomplicated
Jan 23 10:46:04 <ant_work> surely  RDEPENDS_kernel-base = "kernel-image"  is bad ;)
Jan 23 10:46:13 <ant_work> if put in machine.conf
Jan 23 10:47:06 <bluelightning> what does that actually do?
Jan 23 10:47:24 * bluelightning does not know what kernel-base is
Jan 23 10:47:28 <ant_work> in oe-clasic that made sure kernel is installed in /boot
Jan 23 10:47:41 <ant_work> kernel-base is apackege (see kernel.bbclass)
Jan 23 10:47:59 <ant_work> even, the postinst was broken, I fixed that in oe-core
Jan 23 10:48:54 <ant_work> so, unless an image wipes it out (oe-core-minimal i.e.) , we should have kernel in boot
Jan 23 10:49:46 <ant_work> we should possibly share one single knob between machines.conf
Jan 23 11:10:47 <ant_work> bluelightning: wrt label of kexecboot.cfg, maybe MACHINE + DISTRO_VERSION
Jan 23 11:11:09 <bluelightning> machine will always be the same though surely?
Jan 23 11:11:11 <ant_work> i.e. "h3600_oe-core.0"
Jan 23 11:11:29 <ant_work> well, kexecboot has concetp of machine-kernel
Jan 23 11:12:11 <ant_work> mage one would deploy one distro and 3 kernels (zubuntu ?)
Jan 23 16:04:45 <ant_work> Jay7: what is your suggestion for precompilating LABEL= in kexecboot.cfg ?
Jan 23 16:05:01 <ant_work> I'm puzzled by the narrow qvga
Jan 23 16:05:25 <Jay7> have no idea..
Jan 23 16:05:38 <Jay7> iirc, it was $distro name
Jan 23 16:05:47 <Jay7> but you are building distroless
Jan 23 16:05:53 <ant_work> can happen
Jan 23 16:06:23 <ant_work> on poodle it's very limited indeed
Jan 23 16:14:53 <Jay7> "${DISTRO_NAME} ${IMAGE_NAME}"
Jan 23 16:19:58 <ant_work> eh, no..image would take the first
Jan 23 16:20:42 <ant_work> maybe some date tag?
Jan 23 16:30:55 <Jay7> why not
Jan 23 16:31:16 <Jay7> 'custom build $date' :)
Jan 23 17:21:40 <ant_work> bluelightning suggested to create a $KEXECBOOT_LABEL ?= var
Jan 23 17:22:01 <ant_work> maybe defaulting to $MACHINE - $Short-date
Jan 23 17:22:52 <Jay7> better than nothing :)
Jan 23 17:23:08 <ant_work> in fact DISTRO_VERSION says 'oe-core.0'
Jan 23 17:24:05 <bluelightning> what would be wrong with including ${DISTRO} btw? surely for most distros it's pretty short
Jan 23 17:24:25 <ant_work> well, if you add a check for it (can be unset) nothing against
Jan 23 17:24:25 <bluelightning> obviously for distroless it's blank
Jan 23 17:24:38 <ant_work> worst, is unexpanded :/
Jan 23 17:25:37 <Jay7> name=${DISTRO:-"$MACHINE - $short_date"}
Jan 23 17:25:43 <Jay7> something like this :)
Jan 23 17:30:29 <bluelightning> you can always do something like ${@d.getVar('DISTRO', True) or ''}
Jan 23 17:30:43 <bluelightning> somewhat messy perhaps
Jan 23 17:30:47 <bluelightning> but works
Jan 23 17:31:02 <ant_work> pls feel fre to do
Jan 23 17:31:04 <ant_work> it
Jan 23 17:31:27 <ant_work> maybe use DISTRO_VERSION instead ?
Jan 23 17:31:43 <ant_work>  ... or  DISTRO_VERSION
Jan 23 17:32:12 <ant_work> I can't remember why Khem added it
Jan 23 17:32:49 <ant_work> bluelightning: see, go for it for the moment  ${@d.getVar('DISTRO', True) or ''}
Jan 23 17:33:05 <ant_work> we'll change if/when needed with a custom var
Jan 23 17:33:20 <bluelightning> I can do the custom var, that's no problem
Jan 23 17:33:39 <ant_work> would be top
Jan 23 17:34:30 <ant_work> I suppose majority (mee too asap) will use a Distro
Jan 23 17:34:59 <ant_work> asap = once things work
Jan 23 17:36:02 * ant_work has survived a rogue Gentoo update...18 init files changed
Jan 23 17:48:20 <bluelightning> hmm but how to get the dash in there only if DISTRO is set..
Jan 23 18:05:13 <ant_work> bluelightning: what about adding DISTRO ?= "" in kexecboot.cfg ?
Jan 23 18:05:34 <bluelightning> ant_work: that should be ok... but it doesn't solve that issue
Jan 23 18:05:41 <ant_work> and let LABEL = ${DISTRO}
Jan 23 18:05:53 <bluelightning> unless you are happy with "-oe-core.0" as a label?
Jan 23 18:06:07 <bluelightning> (is leading - an issue?)
Jan 23 18:06:24 <ant_work> I'm just against unexpanded ${DISTRO}  :)
Jan 23 18:06:59 <ant_work> I'm open to *all* solutions :p
Jan 23 18:07:24 <bluelightning> ok, if leading dash isn't an issue then I can fix it :)
Jan 23 18:08:02 <bluelightning> KEXECBOOT_LABEL ?= "${@d.getVar('DISTRO', True) or ''}-${DISTRO_VERSION}-${MACHINE}"
Jan 23 18:08:12 <bluelightning> or should we drop the machine?
Jan 23 18:08:26 <ant_work> I'll test on poodle.. I fear we have max 12-15 chars
Jan 23 18:09:33 <ant_work> wait, maybe there is now a second line..can't remember offhand
Jan 23 18:09:39 <ant_work> http://kexecboot.org/screenshots
Jan 23 18:10:10 <ant_work> I'll be bacl later this evening, I'll tell you how many chars are displayed
Jan 23 18:10:13 <Jay7> ant_work: second line is filled automatically
Jan 23 18:10:27 <ant_work> iirc fstype is right-aligned
Jan 23 18:10:32 <Jay7> yes
Jan 23 18:10:50 <Jay7> screenshot should be updated
Jan 23 18:11:02 <ant_work> I lack qemu :)
Jan 23 18:11:08 <Jay7> same here..
Jan 23 18:11:21 <ant_work> lumag was working on a collie emulation...
Jan 23 18:11:34 <Jay7> need to spend some time to make new environement
Jan 23 18:11:51 <Jay7> it's easier to take screensots from qemu-x86
Jan 23 18:12:02 <ant_work> otherwise I'll take some picture and let my wife do the editings
Jan 23 18:13:06 <ant_work> but I'd prefer a pixel-like
Jan 23 18:13:27 <ant_work> Jay7: hm.. and adding an option to dump framebuffer on file?
Jan 23 18:13:40 <ant_work> just for dev
Jan 23 18:13:45 <Jay7> no, just screenshot qemu window :)
Jan 23 18:14:03 <Jay7> you should convert fb dump to some image then
Jan 23 18:14:13 <Jay7> should be easy imho but need work
Jan 23 18:14:38 <bluelightning> ant_work: ok, will be around
Jan 23 18:14:49 <ant_work> going now, bye bye
Jan 23 23:46:14 <ant__> bluelightning: http://postimage.org/image/g281nc56x/
Jan 23 23:47:42 <ant__> oe-core.0 has same length
Jan 24 00:26:00 <ant__> bluelightning: ERROR: No recipes available for:
Jan 24 00:26:00 <ant__>   /oe/meta-handheld/recipes-core/udev/udev_173.bbappend
Jan 24 00:26:36 <ant__> aren't the ts rules already installed?
Jan 24 00:27:24 <ant__> (I've removed udev from meta-openembeded atm)
Jan 24 00:35:39 <bluelightning> ant__: not for h1940 they aren't
Jan 24 00:36:21 <ant__> I see. I stay away from meta-oe's udev
Jan 24 00:36:44 <ant__> it still borks boot
Jan 24 00:37:06 <ant__> well, tbh tested one month ago...
Jan 24 00:38:00 <ant__> I think it does not cope very well with initscripts ordering
Jan 24 00:38:12 <ant__> probably is only tested with systemd
Jan 24 00:38:33 <ant__> and Koen evilish disabled udev-cache for systemd only ;)
Jan 24 00:40:10 <ant__> have you seen how narrow is the LABEL on qvga?
Jan 24 00:41:03 <ant__> we even discussed/tested variable spaced chars
Jan 24 00:41:18 <ant__> finally Yuri chosed this one
Jan 24 00:41:46 <ant__> which looks pretty rounded
Jan 24 00:41:53 <bluelightning> ant__: it works OK on h1940 at least
Jan 24 00:42:08 <ant__> you mean you have touchscreen
Jan 24 00:42:11 <ant__> ?
Jan 24 00:42:15 <bluelightning> yes
Jan 24 00:42:31 <ant__> ok, we have it on Z too, with both udev
Jan 24 00:42:57 <ant__> I guess the ads7846 rule is common
Jan 24 00:43:07 <bluelightning> I did not test recently with an older version, but either way that additional rule is necessary
Jan 24 00:43:19 <ant__> how did we load it then?
Jan 24 00:43:43 <ant__> ok, let me see
Jan 24 00:44:24 <ant__> udev 164
Jan 24 00:45:04 <ant__> http://paste.debian.net/153410/
Jan 24 00:47:47 <bluelightning> ok...
Jan 24 00:48:07 <bluelightning> so what can I do here?
Jan 24 00:48:29 <ant__> hm..
Jan 24 00:50:27 <ant__> http://paste.debian.net/153412/
Jan 24 00:51:08 <ant__> is in local.rules
Jan 24 00:52:29 <bluelightning> I gotta go to bed, I'm really tired :/
Jan 24 00:52:42 <ant__> http://paste.debian.net/153414/
Jan 24 00:52:46 <ant__> me too
Jan 24 00:52:54 <ant__> seems local.rules takes care of the ts
Jan 24 00:53:22 <bluelightning> that rule looks like the first one from touchscreen.rules for 173 which for whatever reason does not match for h1940
Jan 24 00:53:54 <ant__> I'll retest with actual ude from meta-oe then
Jan 24 00:54:06 <ant__> launching a build ang going to zZZ ...
Jan 24 00:54:39 <bluelightning> I'm not suggesting you have to use udev 173... it's just what meta-oe provides and I have to have meta-oe enabled
Jan 24 00:54:56 <bluelightning> we should discuss this further tomorrow
Jan 24 00:55:01 * bluelightning -> bed
Jan 24 00:55:07 <ant__> it's only about commenting DEVICE_CACHE is /etc/defaults/udev   not a big thing
Jan 24 00:55:14 <ant__> gnb
Jan 24 00:55:14 <bluelightning> hmm
Jan 24 00:55:16 <bluelightning> cya
**** ENDING LOGGING AT Tue Jan 24 02:59:57 2012