[05:35] <sid__> hi anybody there
[05:36] <sid__> i want to know the exact flow of rootstock script
[05:36] <sid__> and its functioning
[08:05] <zyga> good morning
[08:07] <lool> morning
[08:35] <neure> hi
[08:35] <neure> how good is 10.04 on beagleboard?
[08:37] <amitk> neure: it works fairly well on the C4 board
[08:37] <DanaG> though, rcn-ee's kernels are better than the official Ubuntu one.
[08:38] <persia> DanaG: Do we know precisely how?  Is this fixable?
[08:39] <DanaG> https://bugs.launchpad.net/ubuntu/lucid/+source/linux-ti-omap
[08:41] <persia> amitk: Do we have an SRU in progress?
[08:41] <amitk> persia: some of those bugs will be fixed as SRUs, yes
[08:41] <DanaG> I'm also interested in this: https://blueprints.launchpad.net/ubuntu/+spec/foundations-m-uefi-support
[08:42] <DanaG> though, it'd be nice to get HP to fix the first-gen elitebooks' broken UEFI.
[08:42]  * persia looks again, having thought all of them were listed as SRU candidates
[08:42] <persia> DanaG: That ought drop in fairly smoothly.  EFI already (mostly) works.  The main issue with UEFI is the CDs.
[08:44] <DanaG> My issue, specifically, is that my firmware seems to claim GraphicsOutputProtocol support... but the framebuffer is at 0x0.
[08:44] <DanaG> also: http://h30499.www3.hp.com/hpeb/?category.id=business-support
[08:44] <persia> Ah, that's broken firmware then *)
[08:44] <DanaG> er
[08:45] <DanaG> fail.... urls are now dead.
[08:45] <hrw> morning
[08:47] <DanaG> Next-gen models are supposedly fixed... but that does me no good. =(
[08:47] <DanaG> It'll be interesting to see uefi + ARM.
[08:47] <neure> well i'd like to know what does not work
[08:48] <neure> "fairly well" is rather meaningless
[08:48] <amitk> OTG is broken ATM, and DanaG has pointed you to the buglist
[08:48] <neure> bugs at launchpad? hmm let me see
[08:48] <neure> my computer is really slow right now though :/
[09:03] <neure> which package has 7za?
[09:06] <tmzt> p7zip at one point
[09:06] <ogra> neure, apt-cache search 7za ;)
[09:07] <neure> thx
[09:12] <neure> "Everything is Ok"
[09:12] <neure> lol
[09:12] <neure> so comforting
[09:15] <neure> im on my x86 linux now, how do i figure out what device is my memory card reader?
[09:15] <hrw> dmesg?
[09:16] <ogra> or if its formatted it should just automount :)
[09:16] <ogra> (once you plug an SD in)
[09:16] <hrw> ogra: depends on system config - my ubuntus do not automount
[09:16] <ogra> mine does
[09:17] <ogra> definately in karmic and lucid
[09:17] <ogra> with the builtin mmc reader as well as with an external USB one
[09:17] <hrw> [161018.255969] sd 8:0:0:0: [sdi] Assuming drive cache: write through
[09:17] <hrw> [161018.255973]  sdi: sdi1
[09:17] <hrw> 10:17 hrw@home:~$ mount|grep sdi
[09:17] <hrw> 10:17 hrw@home:~$
[09:18] <ogra> wow, file a bug
[09:18] <ogra> devkit-disks should definately pick it up
[09:18] <hrw> ogra: for me it is not a bug but proper behaviour
[09:18] <ogra> it should only not automount if you added a specific udev rule to omit it
[09:18] <neure> hmm mine is sdb
[09:19] <ogra> by default all block devices are supposed to be automounted
[09:19]  * neure ./setup_sdcard.sh --mmc /dev/sdb --uboot beagle --swap_file 100
[09:20] <hrw> neure: I have hdds as sda/b/c, one card reader as sdd/e/f/g, pendrive as sdh and second card reader as sdi
[09:20] <neure> i hope i get this Lucid 10.04 Xfce4 to boot on my C2 beagleboard
[09:21] <ogra> for a C2 make sure to have a proper power supply attached
[09:21] <ogra> and a powered hub as well
[09:21] <neure> check, check
[09:21] <ogra> it has HW issues wrt USB/power
[09:21] <neure> keyboard and mouse is in the powered hub
[09:21] <neure> the mini usb thing, also ethernet is there
[09:22] <neure> i wonder if i need any specific boot arguments
[09:22] <hrw> no, just boot
[09:22] <ogra> beyond that i'd recommend the netinstall image and then install xubuntu-desktop (or xfce )
[09:22] <ogra> https://wiki.ubuntu.com/ARM/Beagle
[09:22] <neure> would that be somehow better?
[09:22] <ogra> its tested and supported :)
[09:23] <neure> ok i'll go that instead then
[09:23]  * ogra just closed a bunch of bugs as invalid from people that use their own kernels
[09:23] <hrw> 10 years ago my desktop had 128MB ram and it was enough for gnome1. today with beagleboard/256MB ram it is hard to fit something
[09:24] <ogra> 640k is enough for everyone !!
[09:24] <neure> why is the netinstall image half a gig?)
[09:24] <amitk> hrw: that's called progress
[09:24] <ogra> neure, the netinstall image isnt half a gig
[09:24] <hrw> neure: its SD card image
[09:24] <hrw> neure: lot of empty space inside
[09:24] <ogra> neure, dont mix up netinstall with netbook :)
[09:25] <ogra> netinstall is about 10MB
[09:25] <neure> "Download installation software" Download the ubuntu-10.04-netbook-armel+omap.img image file from ...
[09:25] <hrw> amitk: add >logger "autogetty: $GETTY_ARGS"< to /etc/init/autogetty.conf and restart. then tell what was in syslog
[09:25] <neure> that is the first thing on https://wiki.ubuntu.com/ARM/BeagleNetbookInstall
[09:25] <neure> and that image is half a gig
[09:25] <ogra> https://wiki.ubuntu.com/ARM/BeagleNetInstall
[09:25] <ogra> you are looking at the wrong one
[09:25] <neure> oh
[09:26] <ogra> the netbook image contains the whole netbook release :)
[09:26] <ogra> which gets you what the screenshot at https://wiki.ubuntu.com/ARM/Beagle shows
[09:27] <ogra> (only tested on beagle C4 though, not sure it works on a C2)
[09:27] <hrw> I got netbook installed on C3
[09:27] <hrw> not that I would suggest using it
[09:27] <ogra> does C2 have 256M ?
[09:27]  * ogra thought that was still 128
[09:27] <ogra> netbook definately needs 256
[09:28] <ogra> (and a lot of swap)
[09:29] <ogra> hrw, yeah, netbook was mainly thought as a base for the upcoming panda/blaze images in maverick ... probably also good to demo whats possible but surely not for daily use
[09:29] <hrw> Cx have 256MB
[09:29] <hrw> Ax/Bx were 128MB
[09:29] <ogra> right, i wasnt sure if there wasnt some overlap
[09:29] <hrw> there was cpu overlap
[09:30] <hrw> B7 got newer silicon
[09:30] <hrw> C4 got even newer one
[09:30] <ogra> ah
[09:32] <ogra> cooloney, so i was trying to build touchbook kernels on my blaze over the weekend ... seems there are still some issues, usually make uImage works fine but in make modules i then get bus errors (or the other way round if i build modules first on the second run when using make uImage i get the same)
[09:32] <amitk> ogra: I thought the only problem with touchbook was the display resolution?
[09:32] <ogra> i was using make -j2 so it might be a heat problem or it might trigger the same issue we see with git clone, not sure
[09:33] <ogra> amitk, nope, there are DSS2 patches (that break all other boards) that arent in our tree
[09:33] <ogra> i finally found a 2.6.32 tree to build that actually has all patches i need, but now my touchbook seems ot have power issues
[09:34] <ogra> so i didnt get very far yet but seeing a tty on console
[09:34] <ogra> somehow the battery doesnt charge ... i wonder if the power supply has issues with 220v (even though the printed specs disagree)
[09:35] <ogra> it shuts down after a minute or two
[09:35] <hrw> ogra: OE 2.6.32 works on TB iirc
[09:35] <ogra> hrw, i dont use OE, i want a plain kernel tree
[09:35] <ogra> so i can diff patches against the ubuntu tree
[09:35] <XorA> ogra: my 4430 SDP hard locks on occasion, but this is the sort of thing we expect with ES1.0 hardware :-D
[09:35] <ogra> but anyway the TB patches are to intrusive
[09:36] <ogra> we would need a standalone kernel for it i guess
[09:36] <ogra> (which i'm fine to provide from people.u.c once i have one that works though)
[09:36] <ogra> (as unsupported build indeed :) )
[09:36] <cooloney> ogra: what's touchbook kernel? i have no idea about that. -:/
[09:37] <ogra> cooloney, its just a kernel
[09:37] <ogra> cooloney, the point is thet the blaze kernel misbehaves when i try to build a kernel with it
[09:37] <ogra> it either overheats or exposes the same issue we see with git
[09:37] <ogra> its fine after i reboot the blaze
[09:37] <cooloney> ogra: so what's the difference between blaze kernel and touchbook kernel?
[09:38] <ogra> cooloney, forget about the touchbook kernel
[09:38] <ogra> my point is that building soemthing heavyweight on the blaze with -j2 makes it die at some point
[09:38] <ogra> gcc spills out bus errors after 1-2h of building
[09:38] <cooloney> ogra: ok, terrible.
[09:39] <ogra> which are gone after a reboot
[09:39] <cooloney> bus errors
[09:39] <ogra> we need to talk to nicolas
[09:39] <ogra> but its a bank holiday across most of europe today
[09:39] <cooloney> ogra: right, maybe they did similar test on blaze
[09:39] <ogra> so it has to wait until tomorrow
[09:39] <cooloney> ogra: ok, understand
[09:40] <ogra> just wanted to notify you that there is some issue
[09:40] <ogra> and a kernel is the heaviest i can build atm ... thats why i took a touchbook kernel
[09:40] <ogra> (i didnt like to build OO.o :P )
[09:40] <cooloney> ogra: ok, so if ~test6 works on your hardware,
[09:41] <cooloney> i am going to ask merge the kernel as a branch in lucid
[09:41] <ogra> ok
[09:41] <cooloney> and upload it to PPA
[09:41] <ogra> i'm still running test5
[09:41] <ogra> what changed ?
[09:41] <cooloney> oh, i went throught the email from Sebastein
[09:41] <ogra> ah
[09:41] <cooloney> and built-in SysLink driver which will cause linking error before
[09:42] <cooloney> and rebased to omap4 latest kernel from omapzoom
[09:42] <cooloney> they updated some wifi MMC things in their kernel
[09:42] <ogra> ok
[09:42] <cooloney> so mostly are not big change since ~test5
[09:43] <cooloney> and ogra, before I upload it to PPA, need i solve the ABI dump and modules list issue in kernel?
[09:43] <ogra> well, you discuss some SMP related option there
[09:43] <cooloney> oh, that KGDB,
[09:43] <ogra> i guess for the first upload you need to disable the abi and module checks
[09:43] <ogra> amitk will know though
[09:44] <cooloney> they did not implement a function for KGDB in SMP
[09:44] <ogra> cooloney, since my building fails with -j2 it might be related
[09:44] <cooloney> so we have to disable KGDB in our kernel
[09:44] <ogra> i havent tried without -j2 though
[09:44] <cooloney> ogra: so -j1 works fine in your blaze?
[09:44] <cooloney> ok
[09:45] <amitk> cooloney: for first upload, just ignore the abi and modules list (see wiki)
[09:45] <ogra> havent tried yet, building a kernel with -j1 will take ages :)
[09:45] <cooloney> ogra: ok, IC, heh
[09:45] <cooloney> amitk: thx, i failed to find the wiki about that.
[09:45] <cooloney> amitk: where is it?
[09:46] <ogra_> whoops
[09:46] <amitk> cooloney: see the first few omap uploads in lucid, there is a ignore and modules.ignore file in the  abi dir
[09:46] <ogra_> -j2 finishes building uImage in 45min ... i think -j1 will take 3h or more :)
[09:47] <cooloney> amitk: so i just touch those 2 files and upload the kernel source.
[09:48] <ogra> hmm ?? where does that underscore guy come from
[09:48] <ogra> weird, i have two xchats running but see only one on the desktop
[09:49] <amitk> cooloney: each of those files should contain a single line with '1' in it.
[09:50] <cooloney> amitk: thanks, i understand. so for uploading, the page in our ppa is quite simple.
[09:50] <cooloney> amitk: 'dput <source.changes>' what does that mean?
[09:51] <ogra> use the .changes file for uploading with dput
[09:51] <amitk> cooloney: https://wiki.ubuntu.com/KernelTeam/KernelMaintenance (seearch for ignore)
[09:52] <amitk> cooloney: you have to build the kernel (debuild) and upload the resulting .changes files to PPA
[09:52] <ogra> cooloney, you need to generate a debian source package (which generates <packagename-version>_source.changes) and then upload the .changes file (it pulls the other bits of the source package with it)
[09:53] <cooloney> ah, got it. so how to generate a debian source package for kernel?
[09:54] <ogra> either using "dpkg-buildpackage -S -sa" or debuaild -S in the source tree
[09:54] <ogra> *debuild
[09:54] <ogra> make sure to have your gpg key set up right before doing that, you need to sign the package
[09:56] <cooloney> ogra: thx, will try soon
[09:56] <ogra> (debuild/dpkg-buildpackage will automatically ask for your passphrase and do everything for you)
[09:57] <cooloney> ogra: it is so smart now, much easier in lucid than before
[09:58] <ogra> building packages ?
[09:58] <ogra> that has always been like that :)
[09:58] <ogra> (cant speak for the kernel though, not sure that was different before lucid, but i dont think so)
[09:59] <cooloney> ogra: need i use root or fackroot?
[09:59] <ogra> fakeroot is ok, but you shouldnt even need that for building a source package
[09:59] <ogra> err ... hmm, you might
[09:59] <ogra> use fakeroot :)
[10:00] <ogra> never use root for building packages
[10:00] <hrw> ogra: you should like this: http://hrw.pastebin.com/aR3VSxJS
[10:00] <ogra> hrw, great !
[10:00] <ogra> hrw, how about dropping all the if/then ? :)
[10:01] <ogra> [ -z $BITS ] && BITS=8
[10:01] <ogra> works the same way and is more readable
[10:01] <hrw> I like if/then/fi more
[10:01] <ogra> (makes no difference execution wise)
[10:01] <hrw> ogra: I prefer programing languages then shell ones so if/then/fi
[10:02] <ogra> hrw, will you put that into initramfs-tools by default ?
[10:02]  * ogra votes for it 
[10:02] <hrw> ogra: asac did put it in some package already
[10:02] <hrw> ogra: so ask him
[10:02] <ogra> oh, /bin/autologinroot wont work in initramfs-tools
[10:02] <ogra> hrm
[10:03] <ogra> hrw, i want that function in ubuntu proper too
[10:03] <ogra> but /bin/autologinroot will make that impossible
[10:04] <hrw> ogra: I think that /etc/default/autogetty can solve that
[10:04] <ogra> you mean by just checking for a variable ? yeah, that would help
[10:05] <ogra> asac, where do you plan to put that script ?
[10:06] <ogra> (it wasnt uploadedd to maverick apparently)
[10:06] <hrw> ogra: http://hrw.pastebin.com/LtbQBMa0
[10:06] <hrw> ogra: I think that so far it landed in newcore package
[10:07] <ogra> right, i'd like to have that function in maverick (and i'd like to use it for 10.07)
[10:07] <ogra> why do you set line 17 ?
[10:07] <ogra> if AUTOGETTY_ARGS="-n -l /bin/autologinroot" is in /etc/default/autogetty that should be fine
[10:07] <hrw> cause I am testing it on system where I do not want to create such one
[10:08] <ogra> ah
[10:09] <ogra> well, for having it in maverick it would have to not use autogetty by default
[10:09] <ogra> so you need the eaxct opposite here :/
[10:10] <hrw> ogra: or add "START=yes/no" in /etc/default/autogetty
[10:10] <ogra> right, something like that
[10:10] <ogra> though that would even require to have the autogetty binary in the main distro
[10:11] <ogra> i would propose to just source the default file if it exists, put your stuff into the distro and for your project just have a package that creates the file
[10:11] <ogra> i.e. an autogetty package that ships binary and /etc/default/autogetty
[10:12] <ogra> that way the distro can have serail gettys automatically without using root access on them by default
[10:13] <hrw> ok, will discuss with asac
[10:13] <ogra> cool, thanks :)
[10:13] <ogra> i guess he's off today as well
[10:14] <ogra> and i'll go enjoying a bit of my free day too now ... back later
[10:15] <hrw> national holiday?
[10:18] <ogra_cmpc> yes whit monday
[10:21] <hrw> I need to readd US/UK/FR national holidays to calendar
[10:23] <amitk> ogra_cmpc: curious, .fi has a WhitSunday
[10:24] <ogra_cmpc> amitk, we have that too
[10:24] <ogra_cmpc> its both days in .de
[10:27] <kblin> amitk: what's the point on having sunday off? ;)
[10:28] <hrw> kblin: if you work in supermarket then it matters
[10:28] <kblin> hrw: not in .de
[10:28] <hrw> kblin: but in .pl yes
[10:28] <kblin> at least in my state supermarkets aren't allowed to open on sundays
[10:29] <hrw> kblin: basically if you work in businnes which is open 7 days a week then sundayOff matters
[10:29] <ogra_cmpc> same here
[10:29] <kblin> which is funny, given how much more religious the polish people are in general
[10:29] <hrw> in Poland supermarkets can be open on sundays and I even like it.
[10:29] <hrw> during week I do not have a time to go and make big shopping
[10:30] <kblin> well, people tried to do that in .de as well, but unions and churches lobbied against it
[10:31] <kblin> so if you have a gas station with an attached supermarket you can open on sunday, but if you're a normal supermarket you can't
[10:31] <kblin> obviously god likes gas stations ;)
[10:55] <ogra_cmpc> heh
[10:55]  * ogra_cmpc just found out why his touchbook batteries didnt charge
[10:55] <ogra_cmpc> apparetnly you can only charge it while its running
[10:58] <hrw> common problem with cheap hardware
[10:59] <amitk> ogra_cmpc: I'm guessing they have a driver to do charging then?
[11:00] <ogra_cmpc> apparently
[11:00] <ogra_cmpc> i wish the NIC would work :(
[11:03] <XorA> quit complaining, at least you have one to play with :-D
[11:03] <hrw> ok, lets plug some dongles and report bugs against linux-omap-ti
[11:03] <lool> linux-ti-omap
[11:03] <hrw> right - have it open in browser
[11:03] <hrw> test1: rt2570
[11:03] <lool> ogra_cmpc: Yes, on N900 the phone actually boots in a special mode when charging the batteires
[11:04] <ogra_cmpc> well, its supposed to be a rt2879
[11:04] <ogra_cmpc> err
[11:04] <ogra_cmpc> 2870
[11:04] <hrw> ogra: no, I have rt2570 and rt3072 here
[11:04] <ogra_cmpc> loading the shipped module in OE doesnt get me a device
[11:04] <hrw> ah
[11:05] <ogra_cmpc> and plugging the device into an x86 system also only gets me errors in dmesg
[11:05] <ogra_cmpc> i suspect the HW is dead
[11:05] <hrw> let me check
[11:06] <ogra_cmpc> plugging in my zd1211rw NIC makes the system shut down ... i guess i need to charge some more hours
[11:06] <hrw> ok, got same in desktop. shit... it was working
[11:06] <hrw> 6€ turned to dust
[11:07] <ogra_cmpc> sadly OE only ships that single one USB wlan NIC driver :(
[11:07] <hrw> will check under maverick kernel later (on desktop)
[11:07] <ogra_cmpc> so even if it wouldnt shut down, my zd1211rw wouldnt work :(
[11:08] <ogra_cmpc> without network its a pretty useless thing ...
[11:09]  * amitk boots a 2.6.34 maverick kernel on the beagle
[11:09]  * ogra_cmpc is envious
[11:10] <hrw> amitk: share
[11:10] <ogra_cmpc> will surely be in the archive soon
[11:11] <hrw> cool
[11:11] <hrw> rt3072 under lucid/desktop suxx
[11:13] <ogra_cmpc> geez
[11:13] <neure> erm
[11:13] <neure> i cant get ARM/BeagleNetInstall to work
[11:13]  * ogra_cmpc reads the AI FAQ
[11:13] <neure> fatload mmc 0 0x82000000 boot.scr says 227 bytes read
[11:13] <neure> source 0x82000000 says Unknown command 'source'
[11:14] <ogra_cmpc> "My batteries does not seem to charge ..."
[11:14] <ogra_cmpc> "With a voltmeter, check the voltage of pin 7 of the battery connector"
[11:14] <ogra_cmpc> oh my
[11:14] <ogra_cmpc> neure, ouch, your u-boot is to old i guess
[11:14] <ogra_cmpc> neure, open the boot.scr file in vi on your desktop and type in the lines manually
[11:16] <ogra_cmpc> ugh
[11:16] <ogra_cmpc> "Your wifi dongle may not work. Try to plug it into another USB port an wait a minute: the LED should blink. If not, please reinstall your OS"
[11:16] <amitk> hrw: display is broken, dss2 patch was merged only after 2.6.34, cherry-picking now...
[11:16] <ogra_cmpc> reinstall your OS, yah
[11:16] <persia> Oh my.  That:s not a good recommedation.
[11:16] <ogra_cmpc> persia, the AI FAQ is full of that stuff
[11:17] <ogra_cmpc> grab your voltmeter ... reinstall your OS
[11:17] <persia> Right.  Need to get those folk off email and in here and then everything should work
[11:17] <ogra_cmpc> hahah
[11:17] <ogra_cmpc> There seems to be something loose in my bottom part
[11:17] <ogra_cmpc> You might have a loose counterweight: you can easily re-glue it by un-mounting your bottom part
[11:18] <neure> ogra, could you pastebin that http://ports.ubuntu.com/ubuntu-ports/dists/lucid/main/installer-armel/current/images/omap/netboot/omap/boot.scr somewhere for me?
[11:18] <neure> i cant seem to see it properly in windows :/
[11:19] <neure> hmm i think i can read this somehow..
[11:20] <ogra_cmpc> fatload mmc 0:1 0x80000000 uImage
[11:20] <ogra_cmpc> fatload mmc 0:1 0x82000000 uInitrd
[11:20] <ogra_cmpc> setenv bootargs vram=12M omapfb.mode=dvi:1280x720MR-16@60
[11:20] <ogra_cmpc> bootm 0x80000000 0x82000000
[11:21] <neurre> :/
[11:22] <amitk> ogra_cmpc: what do you think about a feature in flash-kernel where it can flash to a different area in flash, so that if the new/test kernel is broken, I can stop uboot autobooting and point it to the alternate (old) address for a working kernel
[11:22] <ogra_cmpc> amitk, planned for maverick already
[11:23]  * amitk ^5s ogra_cmpc 
[11:23] <ogra_cmpc> amitk, since we will provide preinstalled SD images its no prob to keep a uImage.bak in place in the vfat partition
[11:23] <persia> amitk: Mind you, if you implement it first/faster, that's lovely :)
[11:23] <neurre> i still cant get keyboard work :/
[11:23] <ogra_cmpc> we wont use NAND in maverick (since most OMAP4 devicdes we'll suppport dont have NAND)
[11:24] <persia> Rather, we won't use NAND for any images, or by default.  No reason we can't have support there.
[11:24]  * amitk better figure out why mmc isn't working in the 2.6.34 kernel then :)
[11:24] <persia> In fact, that makes it significantly easier to do other stuff later.
[11:24] <ogra_cmpc> we wont use NAND at all in flash-kernel with the new images
[11:26] <amitk> neurre: are you connected via your usb-hub to the ehci port of beagle?
[11:30] <neure> if netinstall fails, what is the next best option
[11:30] <persia> SD install
[11:30] <neure> url?
[11:32] <amitk> neure: what reason for the failure?
[11:32] <neure> amitk, when i boot the netinstall image on my C2 beagle, usb does not seem to work
[11:32] <neure> usb works when i boot to ångström
[11:33] <neure> this is powered usb hub on miniusb
[11:33] <neure> so im stuck at the language selection screen in the beginnin
[11:35] <neure> amitk, any suggestions?
[11:35] <persia> neure: Do you want GTK, Qt, or server?
[11:35] <amitk> neurre: is your keyboard connected via your usb-hub to the ehci port of beagle?
[11:35] <neure> persia, i just want X
[11:35] <neure> amitk, usb-hub
[11:35] <neure> powered usb hub
[11:35] <amitk> neure: where is the usb hub connected?
[11:35] <persia> neure: OK.  For SD install, we have two images that have X, one with GTK and one with Qt.
[11:35] <amitk> to the ehci port or the smaller otg port?
[11:36] <neure> amitk, the smaller usb connector on the board
[11:36] <neure> smaller one
[11:36] <neure> the normal looking usb port has nothing in it
[11:36] <amitk> neure: that is broken ATM, see bug list
[11:36] <neure> amitk, so, what should i do?
[11:37] <amitk> neure: if you can fix it, I'll take patches. Otherwise, you'll have to wait while we fix that bug.
[11:37] <neure> persia, which one has less space wasted?)
[11:37] <neure> can i use preinstalled images?
[11:37] <persia> neure: They're about the same size: you can look at manifests at http://releases.ubuntu.com/lucid/ or http://cdimage.ubuntu.com/kubuntu-netbook/ports/releases/lucid/release/
[11:37] <neure> i hate those huge borders in ångström enlightenment
[11:38] <ogra_cmpc> neure, you should use the netinstall image but using the OTG port wont work, use a powered hub on the big USB port
[11:38] <neure> persia, i was talking about wasted screen pixels :)
[11:38] <persia> There are no preinstalled images available.
[11:38] <neure> ah
[11:38] <amitk> ogra_cmpc: the B2 boards don't have the ehci port populated, I believe. So he is out of luck
[11:38] <neure> so if i'd use usb hub on normal usb, it'd work?
[11:38] <persia> I think the Qt interface scales better, but it requires 2D accelleration to be acceptable.  The GTK interface breaks down at less than 1024x600, but works with no accel
[11:38] <neure> i have C2
[11:38] <neure> not B2
[11:38] <ogra_cmpc> neure, and a proper power supply on the board itself since the C2/3 have HW issuess otherwise
[11:39] <neure> i have proper power supply
[11:39] <ogra_cmpc> ok
[11:39] <ogra_cmpc> then just plug your powered hub into the big USB port and it should just work
[11:39] <ogra_cmpc> the OTG port is not supported in the lucid kernel as amitk said above
[11:40] <persia> The port is entirely unsupported, or it fails to work in one or the other mode?
[11:41] <ogra_cmpc> entirely unsupported with that kernel version
[11:41] <persia> Ugh.  That's awkward.
[11:45] <neure> i had to take one usb hub from my computer
[11:45] <neure> i had two in a chain
[11:45] <hrw> ok, I will not fight with ogra or amitk but OTG works in lucid 10.04 kernel on C3 BB
[11:45] <neure> this usb hub is bigger than the beagleboard itself :D
[11:45] <neure> it's HUGE
[11:45] <persia> hrw: Maybe different chip revs?
[11:46] <neure> great, now that i moved usb serial thing to new usb port.. it's now COM6 :D
[11:46] <hrw> neure: my desktop has 7 serials out-of-box
[11:46] <amitk> hrw: Are you sure you run the ubuntu kernel? CONFIG_USB_MUSB_OTG is disabled on it...
[11:47] <neure> hrw, my "desktop" is macbook running windows xp with.. 2 usb ports :D
[11:48] <neure> now, i did not get that error
[11:48] <neure> lets see if keyboard will work ..
[11:48] <neure> yes
[11:48] <neure> \o/
[11:49] <hrw> amitk: fun... I could swear that I installed on otg
[11:49] <XorA> hrw: switch in Angstrom kernel, do install :-D
[11:49] <neure> netinstall proceeds :)
[11:50] <amitk> lunch-time
[11:50] <hrw> XorA: I have ubuntu installed
[11:50] <hrw> XorA: 0.22uF capacitor hack helps
[11:50] <hrw> amitk: smacznego
[11:51] <XorA> hrw: my usb died completely, but I have Lucid on my omap4430
[11:51] <amitk> hrw: thanks!
[11:51] <XorA> which is the worlds hugest board
[11:51] <amitk> XorA: is that the OMAP4 SDP?
[11:52] <XorA> amitk: yes
[11:52] <XorA> amitk: took my rootstocked image and used custom kernel
[11:52] <neure> erm
[11:53] <neure> it doesnt detect my network :/
[11:53] <neure> i get a redscreen during netinstall coz my network is not detected :/
[11:55] <neure> :(
[11:56] <ogra_cmpc> might be that we are missing some drivers here
[11:56] <ogra_cmpc> a) file a bug with the exact description of your NIC so we can fix it in maverick, b) use the server image
[11:57] <neure> its some logilink usb 2.0 to fast ethernet thing i have
[11:57] <ogra_cmpc> (or the netbook one)
[11:57] <persia> Will the server image help?  I thought it was the same initramfs as the netinstall image.
[11:57] <ogra_cmpc> it doesnt neet network :)
[11:57] <neure> it was in the beagleboard box
[11:57] <persia> Or do you mean: perform the install without network, get network working, install X.
[11:57] <neure> ok where do i get server image?
[11:57] <ogra_cmpc> network works fine once the kernel is installed
[11:58] <ogra_cmpc> there are issues with the udeb
[11:58] <ogra_cmpc> neure, same wikipage
[11:58] <neure> i rebooted..
[11:58] <ogra_cmpc> neure, note though that you need a usb disk/key to install to
[11:59] <neure> got one
[11:59] <persia> Um, the lucid server image seems to have gone missing, unfortunately :(
[11:59] <ogra_cmpc> netinst is the only one that can diretly install to the SD you booted from
[11:59] <neure> so what, there is no server image?
[11:59] <persia> No.  Just odd URL: http://cdimage.ubuntu.com/ports/releases/lucid/release/
[11:59] <neure> so now what?-|
[11:59] <neure> thx!
[12:00] <neure> will try that now
[12:00]  * persia was expecting http://cdimage.ubuntu.com/ubuntu-server/ports/releases/lucid/release/
[12:00] <neure> erm
[12:00] <neure> which image?
[12:00] <persia> Texas Instruments OMAP server install image
[12:00] <ogra_cmpc> http://cdimage.ubuntu.com/ports/releases/lucid/release/
[12:00] <neure> armel+omap
[12:00] <ogra_cmpc> right
[12:01] <ogra_cmpc> nothing odd about that url
[12:01] <neure> so once i have that .img.. what do i do with it?
[12:01]  * hrw -> lucnh
[12:01] <ogra_cmpc> dd it to the SD card
[12:02] <neure> device?
[12:02] <ogra_cmpc> ?
[12:02] <neure> dd to /dev/sdb ?
[12:02] <ogra_cmpc> if sdb is your sd card, yes
[12:02] <persia> Depends on how your SD is mapped, really.
[12:03] <neure> and why would i need usb stick?
[12:03] <ogra_cmpc> to install to
[12:03] <ogra_cmpc> the img is an installer image
[12:03] <neure> and thats the destination where it will be running?)
[12:03] <ogra_cmpc> right
[12:03] <neure> so i wont need sdcard after?
[12:04] <neure> it well keep booting from the usb key?
[12:04] <ogra_cmpc> it will start from NAND and boot the OS from USB, right
[12:04] <neure> not exactly ideal
[12:04] <neure> is there any other way?
[12:05] <ogra_cmpc> only wiht the netinstall, that will boot from NAND and then be able to run the OS from SD
[12:05] <ogra_cmpc> but you will need a different NIC for it apparetnly
[12:05] <neure> :/
[12:06] <neure> would be great to have install from usbkey
[12:06] <ogra_cmpc> all will get better in maverick with the preinstalled images :)
[12:06] <ogra_cmpc> they will just run from SD directly
[12:07] <ogra_cmpc> so your thrid option would be to wait 6 months *g*
[12:07] <neure> this is AU0025C
[12:07] <persia> neure: The main blocker to install-from-USB is the lack of support for it on the hardware.
[12:08] <neure> really?
[12:08] <persia> neure: if you want to play, and you have NAND, you can try to put the bootloader in NAND, and then have that boot from USB, but it might be tricky.
[12:08] <ogra_cmpc> well, thats what we currently do post install
[12:08] <persia> Yeah, the ROM in the processor on that chip can only load the bootloader from MMC or NAND.
[12:08] <ogra_cmpc> but pre-install you have to rely on the possibilities the HW offers
[12:08] <persia> ogra_cmpc: I thought it was still bootloader-on-MMC postinstall: glad to hear it's not.
[12:09] <ogra_cmpc> nah, we have kernel and initrd in NAND
[12:09] <ogra_cmpc> so we can boot off whatever device we want
[12:09] <persia> Well, no.
[12:09] <ogra_cmpc> we could even do bluetooth NFS
[12:09] <ogra_cmpc> :)
[12:09] <persia> When you put kernel and initrd in NAND, you mostly boot off NAND.
[12:10] <ogra_cmpc> but your rootfs can live wherever you want
[12:10] <persia> The location of root is completely arbitrary, but that's mounted post-boot
[12:10] <persia> Right.
[12:10] <ogra_cmpc> right
[12:10]  * persia is quibbling about nomenclature and semantics, but likes that NAND is used
[12:10] <ogra_cmpc> well, not anymore after lucid
[12:10] <neure> where should i file the bug report about the missing driver?
[12:11] <neure> http://www.itbank.us/Logilink+UA0025C/
[12:11] <neure> thats the device i have
[12:11] <ogra_cmpc> neure, launchpad against linux-ti-omap
[12:11] <ogra_cmpc> mention that its missing in the udeb package
[12:11] <persia> neure: Best if you can run `ubuntu-bug linux-ti-omap` from the board, with the nonfunctional device attached.
[12:11] <ogra_cmpc> haha
[12:11] <ogra_cmpc> without network :P
[12:11] <persia> Sure.
[12:12] <persia> the way it works is that it prepares a report, and then, if you don't have X, gives instructions for getting that offline and filing the bug.
[12:12] <ogra_cmpc> sure
[12:12] <neure> well
[12:12] <ogra_cmpc> its easier to do it via the web interface though
[12:12] <neure> first i would need to INSTALL :)
[12:12] <persia> So you copy the files containing the report details, and then do something like apport-submit (I forget: instructions for the process are part of the process), which submits the right stuff.
[12:12] <ogra_cmpc> for a simple omission in the udeb at least
[12:13] <persia> But less complete :)
[12:13] <persia> Anyway.
[12:13] <ogra_cmpc> we only need info about the NIC
[12:13] <ogra_cmpc> and we most likely have the module in the std kernel package
[12:13] <ogra_cmpc> its just about making sure it ends up in the udeb
[12:14] <neure> lsusb in vmware says Bus 001 Device 002: ID 9710:7830 MosChip Semiconductor MCS7830 Ethernet
[12:14] <neure> is that sufficient?
[12:14] <ogra_cmpc> yeah
[12:14] <ogra_cmpc> moschip only has one driver
[12:14] <neure> url for filing a bug?
[12:14] <ogra_cmpc> launchpad
[12:14] <neure> and its not included?
[12:15] <ogra_cmpc> it isnt in the udeb package which the installer uses
[12:15] <ogra_cmpc> it is the the actual kernel after installation
[12:16] <neure> i should first search if it is already registered as a bug?
[12:22] <neure> https://bugs.launchpad.net/ubuntu/+source/linux-ti-omap/+bug/584920
[12:22] <ubot2> Launchpad bug 584920 in linux-ti-omap (Ubuntu) "netinstall fails, it has no network driver for moschip (affects: 1)" [Undecided,New]
[12:22] <neure> there
[12:25] <ogra_cmpc> ah, great
[12:26] <neure> since this was in the beagleboard box that we bought, i would expect there are other ppl with the same adapter
[12:27] <neure> unfortunately it also comes with the
[12:27] <neure> otg usb hub :/
[12:27] <ogra_cmpc> in maverick everything will be better
[12:27] <ogra_cmpc> we only had three weeks to build the lucid image
[12:27] <neure> well i need this working today :)
[12:28] <ogra_cmpc> it will work after install
[12:28] <neure> well i will endup running it from the usb stick :D
[12:28] <ogra_cmpc> just use the server image and it will be fine after installation
[12:28] <ogra_cmpc> yeas
[12:28] <neure> i just had sdcard bought for me
[12:29] <neure> now i need to tell 'oops i need usb stick too' :D
[12:29] <neure> but hmm
[12:29] <neure> i could get even bigger usb stick?)
[12:29] <ogra_cmpc> you can even use a real USB HDD
[12:39] <neure> cool
[12:40] <neure> looks like i can actually get a real usb hdd
[12:40] <neure> is that possibly more performant than the sdcard?
[12:41] <rcn-ee> depending on the usb hdd.. Yes. ;)
[12:50] <kblin> it also won't wear out as fast
[12:51] <neure> well
[12:51] <neure> turns out i only got usb stick :/
[12:51] <neure> well, more portable
[12:51] <xorAxAx> are there NEON mplayer/ffmpeg packages for ubuntu lucid?
[12:52] <kblin> so what I use is an sdcard for the system an put all the bigger data on /data
[12:52] <kblin> which is, depending on the system either a usb stick, a usb hdd or a cifs network mount
[12:55] <neure> ah, ok
[12:55] <neure> so i can add that later, yeah
[12:55] <neure> half way downloading network image
[12:55] <neure> i mean server image
[13:03] <neure> how does that dd work anyway?
[13:03] <neure> my sdcard is 16gig
[13:04] <neure> i can safely dd anything smaller that that to it?
[13:04] <ogra_cmpc> yeas
[13:11] <neure> just dd img -of=/dev/sdb or do i need something more?
[13:11] <ogra_cmpc> dd if=/path/to/img of=/dev/sdb
[13:11] <ogra_cmpc> (with sudo)
[13:12] <neure> this is going to take a while ...
[13:22] <neure> only 10 minutes :)
[13:23] <neure> so..
[13:23] <neure> how do i boot from this image?
[13:24] <neure> mmcinit - init mmc card
[13:24] <neure> ** Unable to use mmc 0:1 for fatload **
[13:28] <neure> got ut
[13:30] <neure> erm
[13:30] <neure> the installer seems to hang
[13:30] <neure> [!] Partition dsiks
[13:30] <neure> Installation medium on /dev/mmcblk0p1
[13:31] <neure> i dont seem to be able to get past that dialog
[13:31] <ogra_cmpc> no, you want to use the usb disk
[13:31] <neure> as installation medium?
[13:31] <ogra_cmpc> asw target device
[13:31] <neure> yeah but i cant get rid of this dialog
[13:32] <neure> installer gets stuck there
[13:32] <ogra_cmpc> it doesnt here
[13:32] <ogra_cmpc> you should just be able to hit enter there
[13:32] <neure> well i do, nothing happens
[13:32] <ogra_cmpc> and then get to the partitioner
[13:33] <ogra_cmpc> thats before the paritioner ?
[13:33] <ogra_cmpc> (before it asks you to select a device for paritioning?)
[13:33] <neure> yes
[13:33] <neure> before
[13:33] <ogra_cmpc> strange
[13:34] <ogra_cmpc> and you are sure you have the USB key attached already ?
[13:34] <neure> yes
[13:34] <ogra_cmpc> it should really just tell you that mmc isnt a valid media to install to
[13:34] <ogra_cmpc> thats what happens on mz B3 and C4 at least
[13:34] <Martyn> morning
[13:35] <neure> detecting disks...
[13:35] <Martyn> ogra : do you think the 2.6.35 merge window will last as long as Jun 1?
[13:35] <neure> starting up the partitioner
[13:35] <neure> then i get this dialog
[13:35] <neure> oh
[13:35] <neure> now it worked
[13:35] <neure> on second boot
[13:36] <neure> what should i select?
[13:36] <neure> guided - use entire disc?
[13:36] <ogra_cmpc> yeah
[13:36] <neure> its 16GB kingston usb stick
[13:37] <neure> Failed to create a file system
[13:37] <ogra_cmpc> ugh
[13:37] <ogra_cmpc> Martyn, no idea, ask the kernel team
[13:37] <Martyn> which image is neure using, on which hardware, out of curiosity?
[13:37] <neure> The ext4 file system creation in partition #1 of /dev/sda failed.
[13:38] <amitk> Martyn: doubtful, Linus should release the -rc1 this week IMO
[13:38] <neure> beagleboard C2
[13:38] <xorAxAx> ogra_cmpc: can i use the versatile kernel to run the netinstall of the beagle initrd in qemu and then copy the resulting image file to an sd card?
[13:38] <neure> image is ubuntu-10.04-server-armel+omap.img
[13:38] <xorAxAx> (because i dont have an usb-ethernet card)
[13:38] <ogra_cmpc> xorAxAx, well, you wont have a kernel in the target and your NAND wont be set up right
[13:38] <neure> and when i get that red screen, keyboard no longer works
[13:39] <Martyn> amit : damnit... i need one more week to get the mach-sstone patch ready
[13:39] <ogra_cmpc> for creating a rootfs you can for sure use qemu netinstall
[13:39] <xorAxAx> ogra_cmpc: can i somehow manually execute those steps?
[13:39] <ogra_cmpc> neure, but your hub is powered, right ?
[13:39] <amitk> Martyn: you're assuming that the patches will be accepted in the first submission...
[13:39] <neure> yes
[13:39] <neure> it is powered
[13:40] <Martyn> amitk : yes, i know
[13:40] <ogra_cmpc> xorAxAx, for setting up the nand and kernel ? thats a bit tricky ... but you can look into the source of flash-kernel what it does for the beagle
[13:41] <ogra_cmpc> xorAxAx, you should be able to use the netbook or the server image (at least thats properly tested on C3 and C4 beagles, looking at neure the C2 seems to have issues)
[13:41] <ogra_cmpc> both dont need networking
[13:42] <xorAxAx> ogra_cmpc: but i want to install to the sd card
[13:42] <xorAxAx> ogra_cmpc: why isnt that possible using the server image?
[13:42] <ogra_cmpc> because the installer runs from the SD
[13:42] <ogra_cmpc> its not capable to install to the media it runs from
[13:43] <neure> silly limitation
[13:43] <ogra_cmpc> the netinst image copies the 10MB installer to ram after booting and doesnt need the SD afterwards ... the server install has all the packages on the SD
[13:43] <ogra_cmpc> neure, well, add a gig of ram to your beagle and i can overcome that :P
[13:44] <ogra_cmpc> if you have a whole package archive on the SD that needs to live somewhere
[13:44] <neure> couldnt you use usb stick as a temp file system during installation or something like that?
[13:44] <xorAxAx> ogra_cmpc: can i run the server installer from an usb stick?
[13:44] <ogra_cmpc> we could copy it to ram as the netinst installer does but that would require at least the amount of ram the SD carries as packages
[13:45] <neure> and why not usb stick?
[13:45] <neure> you could install the installer to usb first :)
[13:45] <ogra_cmpc> xorAxAx, with some tricks you should be able to have the archive on usb, but dont ask me how
[13:45] <ogra_cmpc> should be described in the debian installer docs somewhere
[13:46] <ogra_cmpc> neure, because beagle cant boot from USb
[13:46] <ogra_cmpc> we had that discussion before
[13:46] <neure> ogra, it doesnt need to boot from there
[13:46] <ogra_cmpc> for the installer it does
[13:46] <neure> really?
[13:46] <ogra_cmpc> the installer is the initrd
[13:47] <ogra_cmpc> in case of debian-installer images (server and netinst)
[13:47] <ogra_cmpc> so uImage and initrd need to come from SD
[13:47] <ogra_cmpc> at least with this HW
[13:48] <neure> but once loaded they dont need the sdcard
[13:48] <lool> hrw: I've renamed the wiki page with the arm-m-cross-compilers spec to CrossCompilers instead of CrossCompilation
[13:48] <lool> hrw: https://wiki.ubuntu.com/Specs/M/ARMCrossCompilers
[13:48] <lool> hrw: Also, did a large pass on it
[13:48] <lool> hrw: Could you have a look?
[13:48] <ogra_cmpc> there are tricks that you can tell d-i to use the archive from a different media
[13:48] <ogra_cmpc> but as i said above you have to look them up in the d-i docs
[13:49] <ogra_cmpc> and have to modify the image accordingly
[13:49] <lool> hrw: If you're happy with the overall text, I'll add work items
[13:50] <hrw> lool: in few minutes I will look
[13:50] <lool> hrw: Once thing which we failed discussing is how the runtime libs should be named and packaged, either they could use the dpkg-cross namespace (package name and file locations), or the cross-compiler one; the latter seems safer, but it could confuse dpkg-cross later and means we have two runtime libs (one dpkg-crossed, the other cross-built) which is also confusing
[13:55] <hrw> or we move to sysroot and keep all crossarch stuff in sysroot only. maybe will require adaptations in dpkg-cross to follow. and I would follow dpkg-cross naming of packages
[13:55] <hrw> but would be nice to have 'cross/' section maybe for them
[13:56] <neure> cross?
[13:56] <neure> "no, we do not cross compile"
[13:57] <xorAxAx> ogra_cmpc: what if i install to a usb storage device which is a sd-card and then later plug the sd card into the beagle, will it still find the root fs?
[13:57] <hrw> xorAxAx: UUID is same so it should
[13:57] <xorAxAx> ok
[13:57] <lool> hrw: What does moving to a third location gain us?
[13:58] <lool> Besides the location is only half of the problem, the other namespace issue is with package names, dpkg-cross also converts dependencies
[14:01] <hrw> lool: location is small problem which can be solved in any way. naming/deps are more important. I think that it depends on how much multiarch ubuntu will be. so far on x86-64 we have ia32-* packages for 32bit compatibility and they are stored in defined places and have own deps. but they are also ugly.
[14:02] <hrw> lool: dpkg-cross way gives simple visibility which pacakges are cross ones but require all that converting (which can be also done at build time of packages if we will go that way)
[14:02] <hrw> lool: I mean: how to show that libpng12 is x86-64 (host) or armel (cross)...
[14:06] <lool> hrw: Multiarch isn't in the picture yet; we know that's what we want in the end, but it's not available right now
[14:06] <lool> hrw: ia32-libs is horrible
[14:13]  * ogra_cmpc dances ...
[14:13] <ogra_cmpc> finally got the touchbook booting
[14:13] <hrw> and I got auto-serial-console working in a way wanted
[14:31] <lool> hrw: Are you happy with the wiki page?
[14:32] <hrw> yes, I am
[14:47] <neure> now im trying again with a different usb stick
[14:49] <neure> also fails for another partition :D
[14:50] <neure> i mean stick
[14:52] <rsavoye> lool: personally, I prefer sysroot too. nice wiki page
[14:54] <ogra_cmpc> neure, but it is listed in the partitioner ?
[14:56] <neure> fuck
[14:56] <neure> i switched to another usb hub
[14:57] <neure> now the usb hub i tried on beagle .. is giving me issues on the pc :D
[14:57] <neure> like, mouse is not working through it :D
[14:57] <ogra_cmpc> well, the C2 definately has USB power issues
[14:57] <neure> now it is
[14:58] <ogra_cmpc> see http://elinux.org/BeagleBoard#EHCI
[14:58] <lool> hrw: BTW did you manage to build an emdebian style cross-compiler using the Ubuntu sources?
[14:58] <hrw> yes, I did
[14:59] <rsavoye> hrw: did you hit the configure problem with libstdc++ and libiberty ?
[14:59] <hrw> rsavoye: during build of gcc? nope
[14:59] <rsavoye> no c++ ?
[15:00] <hrw> g++-4.4-arm-linux-gnueabi_4.4.3-4ubuntu5_amd64.deb
[15:01] <hrw> I built gcc, g++, gfortran, gobcj, gobcj++
[15:01] <rsavoye> using the code sourcery patches ?
[15:02] <hrw> no, ubuntu sources
[15:03] <hrw> I plan to recreate it from scratch tomorrow anyway
[15:03] <rsavoye> I'd have to check, mostly I'm just curious as I use FSF sources
[15:04] <neure> oh dear
[15:19] <lool> hrw: Do you have some tasks you feel like you can attack already with the cross-toolchains?
[15:21] <hrw> lool: was thinking about dropping dh_movefiles for start and then looking how binary-cross is made
[15:22] <hrw> lool: sysroot changes from 4.5 would be nice to have in 4.4 but did not looked at 4.5 yet
[15:29] <lool> Hmm ok, movefiles was actually assigned to me
[15:30] <hrw> ok, can skip it
[15:33] <ogra_cmpc> mumble
[15:33] <ogra_cmpc> so i got a working kernel but the screen shuts off after some minutes
[15:34] <armin76> energy saving feature :D
[15:34] <ogra_cmpc> i wish it would be that easy
[15:40] <amitk> ogra_cmpc: oops?
[15:40] <ogra_cmpc> amitk, yes, but unrelated to the framebuffer
[15:41] <ogra_cmpc> arch/arm/mach-omap2/dpll.c:439 omap3_noncore_dpll_set_rate+0x220/0x264()
[15:41] <ogra_cmpc> seems it doesnt like that i picked performance as governor at compile time
[15:42] <ogra_cmpc> but its unrelated to the screen
[15:42] <ogra_cmpc> (note that is the 2.6.34 touchbook tree i'm trying here)
[15:43] <amitk> should be unrelated to the governor, but a real bug in the dpll code (that could take down the dss on its way down)
[15:44] <ogra_cmpc> if i start/restart gdm i see some message about GFX underrun
[15:44] <ogra_cmpc> saldy that doesnt show up in any log
[15:46] <ogra_cmpc> aha !
[15:46] <ogra_cmpc> this time it wrote to dmesg
[15:46] <ogra_cmpc> [ 3859.026519] omapdss DISPC error: GFX_FIFO_UNDERFLOW, disabling GFX
[15:47] <ogra_cmpc> hmm, and the screen didnt shut down yet i still see my xterm
[15:47] <ogra_cmpc> wow and the touchscreen works out of the box
[15:48] <ogra_cmpc> so apparently issuing xset dpms force on and restarting gdm made it work
[15:49]  * ogra_cmpc is brave and installs ubuntu-netbook
[15:49] <ogra_cmpc> i hope the 4G SD card is sufficiant
[15:51]  * robclark|pb finds 4gb a bit small once you install build-essentials and a handful of builddeps..
[15:52] <ogra_cmpc> yeah, its only a test setup to get a touchbook image together
[15:52] <ogra_cmpc> i wont really do work on it
[15:52] <ogra_cmpc> basically i just want a std ubuntu install to work on the touchbook
[15:52] <ogra_cmpc> that SD is supposed to be my reference
[15:53] <persia> ogra_cmpc: How much variance do you see?  Do you expect that we could have a single image that worked on that and on the beagles?
[15:54] <ogra_cmpc> persia, not until someone uinfies u-boot
[15:54] <persia> Oh, heh.  same problem as always
[15:55] <ogra_cmpc> persia, but the pkan is to provide a script or at least a wikipage to replace MLO and u-boot on the maverick preinstall images we,ll provide
[15:55] <ogra_cmpc> since amitk fights for a unified kernel at least :)
[15:55] <amitk> ogra_cmpc: having a brain freeze here, what is the program that writes the kernel/initrd to nand on beagle? flash-kernel?
[15:55]  * persia omits a repitition of the the extended whining about the parlourous state of bootloaders
[15:55] <persia> amitk: Yes
[15:55] <ogra_cmpc> amitk, in lucid and later its update-initramfs calling flash-kernel
[15:56] <ogra_cmpc> before it was the kernel posinst via kernel-img.conf
[15:56] <amitk> what package?
[15:56] <ogra_cmpc> kernel-img.conf doesnt come from any package afaik
[15:56] <ogra_cmpc> its cretaed by the installer
[15:56] <persia> Indeed, it is.
[15:57] <ogra_cmpc> the ac tual work is done by flash-kernel though
[15:57] <amitk> no, I meant flash-kernel. I want to find out the commands and addresses you write the kernel to
[15:57] <ogra_cmpc> and the flash-kernel-installer.postinst which is used by d-i/ubiquity
[15:57] <ogra_cmpc> ah
[15:58] <ogra_cmpc> look at debian/flash-kernel-installer.postinst in the flash-kernel source then
[15:58] <ogra_cmpc> it sets up the adresses etc
[15:58]  * amitk is trying to rescue a broken (test) kernel upgrade
[15:58] <ogra_cmpc> ow
[15:58]  * ogra_cmpc checks the source
[15:59] <persia> Isn't flash-kernel-installer.postinst basically just a reimplmentation of flash-kernel for use in a udeb?
[16:00] <ogra_cmpc> persia, nope
[16:00] <ogra_cmpc> its doing all the setup for flash-kernel
[16:00] <ogra_cmpc> amitk, so basically its just writing to the mtdblock device that holds the kernel
[16:01] <ogra_cmpc> dd if=/dev/zero of=$kmtd bs=$kmtdsize count=1
[16:01] <ogra_cmpc> cat $tmp.uboot > $kmtd
[16:01] <ogra_cmpc> $kmtd is the mtdblock device that is labeled as Kernel in /proc/mtd
[16:01] <ogra_cmpc> and $tmp.uboot is the uImage
[16:02] <ogra_cmpc> and $mtdsize is the size of the mdtblock device in bytes
[16:03] <ogra_cmpc> *$kmtdsize
[16:03] <ogra_cmpc> and that part is definately in flash-kernel not the installer
[16:04] <amitk> ogra_cmpc: and initramfs?
[16:04] <ogra_cmpc> same thing
[16:04] <ogra_cmpc> but its using the "File System" space for the uInitrd
[16:05] <amitk> ohh, they probably use offset
[16:05] <ogra_cmpc> using a fixed size of 16777216 bytes
[16:05] <ogra_cmpc> (for zeroing it)
[16:05] <amitk> ogra_cmpc: What is "File system" space?
[16:06] <ogra_cmpc> cat /proc/mtd
[16:06] <ogra_cmpc> try that
[16:07] <ogra_cmpc> you should get u-boot, u-boot config, kernel and file system
[16:07] <ogra_cmpc> the device labeled filesystem is the one we use for initrd
[16:08] <amitk> ogra_cmpc: I'm at uboot prompt, transferred over a kernel using ymodem. Now hoping to flash that to nand
[16:08] <ogra_cmpc> oh
[16:08] <ogra_cmpc> no idea
[16:08] <ogra_cmpc> i always do it from a running system
[16:08] <ogra_cmpc> i guess i have to refer you to the beagle wiki
[16:08] <amitk> could I do it from initramfs? (I don't think so)
[16:09] <ogra_cmpc> you should be able to
[16:09] <ogra_cmpc> as soon as the /dev/mtdblock* nodes exost
[16:09] <ogra_cmpc> exists
[16:09] <armin76> ssvb: cosmicparrot :D
[16:11] <hrw> did someone built maverick's gcc-4.4 for armel? ports.ubuntu.com has it but no libstdc++ package
[16:11] <ogra_cmpc> i think doko uploaded it, yes
[16:12] <ogra_cmpc> should be named libstdc++6 though
[16:12] <hrw> yes, but there is -dbg, -pic, -dev but no lib
[16:13] <hrw> anyway I will rebuild it here
[16:13] <amitk> ogra_cmpc: the problem is that the test kernel broke mmc :-/
[16:13] <persia> hrw: Check the build log: maybe something fixable went wrong.
[16:13] <hrw> k
[16:14] <ogra_cmpc> amitk, ouch, you could x/yload kernel and initramfs and boot from ram though
[16:15] <ogra_cmpc> hrw, http://ports.ubuntu.com/pool/main/g/gcc-4.4/libstdc++6-4.4-dev_4.4.4-2ubuntu2_armel.deb
[16:15] <ogra_cmpc> built may 19
[16:15] <hrw> facepalm...
[16:16] <ogra_cmpc> hmm, only -dev though
[16:16] <ogra_cmpc> and -pic
[16:17] <hrw> ~curse apt-cross
[16:17] <ogra_cmpc> there is no non-pic version
[16:21] <hrw> exactly
[16:21] <ogra_cmpc> well, as persia said and also ask doko
[16:21] <hrw> yep
[16:22] <xorAxAx> anybody knows what i need to pass on the kernel param line to get omapfb run on s-video out?
[16:23] <ojn> xorAxAx: Check on #beagle or #linux-omap?
[16:24] <lool> hrw: the runtime libs are built from gcc-4.5 in maverick
[16:25] <hrw> thx
[16:26] <lool> https://launchpad.net/ubuntu/+source/gcc-4.5/4.5.0-3ubuntu1 has links to the .debs
[16:26] <ogra_cmpc> now /me facepalms :P
[16:34] <lool> hrw: I added work items to https://blueprints.edge.launchpad.net/ubuntu-arm/+spec/arm-m-cross-compilers the dh_movefiles action was originally assigned to me, but I don't mind if you pick it up
[16:35] <amitk> hrw: ogra: 2.6.34-based omap3 kernel - display and usb works, nand driver broken, but xubuntu works fine. Find it here -> http://people.canonical.com/~amitk/ti/
[16:36] <hrw> lool: tomorrow I will rebuild toolchain with maverick sources and then start from movefiles
[16:36] <amitk> it looks "good enough" to start the maverick cycle, IMHO
[16:37] <hrw> amitk: fetching
[16:50] <ogra_cmpc> amitk, thats beagle only ? or did you add any other nofty HW support
[16:50] <ogra_cmpc> *nifty
[16:51] <amitk> ogra_cmpc: nothing notably new, except for OTG support (I have no HW to test though). Once this is verified on beagles, we can start playing with more features.
[16:51] <hrw> amitk: during UDS persia was giving otg host cables
[16:52] <lool> JamieBennett: It seems one can retarget blueprints to a different project
[16:52] <lool> JamieBennett: I wonder whether it would make sense to move the bps back to /ubuntu
[16:52] <lool> It would give us all the regular milestones
[16:52]  * persia suspects this would make it easier to collaborate effectively
[16:58]  * hrw -> off
[16:58] <hrw> have a nice rest of day
[17:10] <JamieBennett> lool: That is possible, I'm not completely convinced its the right option though
[17:22] <DanaG1> hmm, is a headless box supposed to automount usb drives?
[17:23] <DanaG1> ah, usbmount just didn't use ntfs.
[17:30] <persia> JamieBennett: What would it take to convince you?
[17:32] <DanaG1> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=321694#17
[17:32] <ubot2> Debian bug 321694 in usbmount "usbmount: use vendor/model in mountpoints" [Wishlist,Open]
[17:43] <persia> DanaG1: I'm really tempted to mark that Won'tFix, for several reasons.  Firstly, I have several devices with the *same* vendor/modle that I use simultaneously.  Secondly, I have other devices with nice names I want to keep.  Thirdly, I think most users would be confused.
[17:43] <persia> Please rebut, and I'll delay doing that :)
[17:47]  * zyga enjoys evening working spree :-)
[17:48] <lool> JamieBennett: Do you have some time for a call?
[18:41] <NCommander> ndec: you around?
[19:27] <lool> NCommander: national holiday today
[19:28] <NCommander> lool: d'oh :-/
[19:43] <ogra_cmpc> NCommander, so i dont undrestand what your issues with genext2fs are, it works fine for me on the babbage and rolls a netbook image in ~30min
[19:51] <NCommander> ogra_cmpc: when I tried it, it ate more and more RAM until it eventually OOMed. If it works for you, then I'll leave the code be :-)
[19:52] <ogra_cmpc> well, you have a weird way of computing the size, it might get confused because you allocate more than actually needed
[19:53] <ogra_cmpc> the -b parameter should take exactly the size of the chroot contents
[19:53] <ogra_cmpc> the tool will allocate enough to add a journal later
[19:54] <ogra_cmpc> also there is no need to add 10MB spare space, we wont need them
[19:55] <ogra_cmpc> the partition will be resized before the first boot anyway
[19:56] <ogra_cmpc> i agreed that genext2fs isnt actually a beauty with its habit to allocate the full image size in ram before writing to it, but we have 30G swap on the builder so no worries
[19:56] <ogra_cmpc> *agree
[19:58]  * ogra_cmpc goes back to slurp cool Cidre and watch TV
[20:06] <persia> So, the reason the magic 10M exists is to contain pool/
[20:07] <persia> If debian-cd isn't going to insert that later, it's not needed, but...
[20:07] <NCommander> ogra_cmpc: tune2fs fails if there's no free space
[20:07] <NCommander> ogra_cmpc: feel free to improve the code, its free software afterall :-)
[20:07]  * NCommander runs
[20:08]  * persia lassos the running target
[20:08] <persia> NCommander: So, where does pool/ end up living with this new order?
[20:09] <NCommander> persia: what pool?
[20:09] <persia> The set of stuff that some folk need to be able to connect to the network to get other stuff.
[20:09] <NCommander> If we need a pool, then we need to expend toe spec
[20:09] <persia> We need a pool.
[20:09] <persia> I thought that it was going to prepopulate /var/cache/apt/... somehow, but there are other ways to make it work.
[20:10] <NCommander> persia: this is a corner case IMHO.
[20:10] <persia> It's really not.
[20:10] <NCommander> persia: this image is for demostration purposes only
[20:10] <NCommander> do we *really* need a pool?
[20:10] <persia> Three of the last four machines I've installed needed pool/
[20:10] <NCommander> what did you need from the pool?
[20:10] <persia> Drivers for network
[20:10] <persia> Drivers for video
[20:10] <NCommander> persia: drivers for the network are included on OMAP4 in kernel
[20:11] <persia> You sure?
[20:11] <NCommander> video can be downloaded from a PPA
[20:11] <NCommander> persia: its supposed to work
[20:11] <persia> So, what kind of network drivers are in the kernel?
[20:11] <persia> Do they work for my odd network environment?
[20:11] <persia> What about with my 3G dongle?
[20:11] <NCommander> persia: we don't have a USB port on blaze
[20:12] <NCommander> That kinda limits what drivers are needed :-)
[20:12] <persia> And there's onboard network of some sort?
[20:12] <persia> And this will remain true for other devices that would use this class of image?
[20:12] <NCommander> persia: can't speak for hardware I don't have
[20:14] <ogra_cmpc> persia, why would you want /pool on an installed system ?
[20:14] <ogra_cmpc> NCommander, tune2fs works for me without leaving a gap in the image
[20:14] <persia> ogra_cmpc: I don't.  I want to be able to get to the network to install other stuff.
[20:15] <ogra_cmpc> persia, right, and even if we would ship a pool dir it wouldnt be inside the rootfs
[20:15] <NCommander> persia: we'll provide drivers for the machines the image is supposed that will make the onboard stuff work
[20:15]  * persia has had all sorts of devices, the most obscure network requirement was PPP-over-ISDN-on-CF card, which caused no end of annoyance to StevenK when trying not to have pool/ for MID
[20:15] <NCommander> persia: ...... my eyes just bled from that sentence
[20:16] <ogra_cmpc> pool is pointless for that kind of image
[20:16] <persia> I'm against the very concept of designing software to build images that can only be used for a single device.
[20:16] <NCommander> persia: why can't we pre-install this stuff? (or is it in restricted)
[20:16] <persia> I've no issues with designing software that ends up being used to generate an image that can only be used for a single device.
[20:16] <persia> Notice the different.
[20:16] <ogra_cmpc> but thats what we'll do, feel free to come up with a better concept
[20:16] <persia> Err, Difference.
[20:17] <persia> ogra_cmpc: As I suggested last week, we could prepopulate /var/cache/apt/ archives with some stuff.
[20:17] <ogra_cmpc> indeed
[20:17] <ogra_cmpc> but with what ? all thats intresting cant be shipped anyway
[20:17] <NCommander> persia: why don't we just preinstall it now?
[20:17] <ogra_cmpc> i.e. the 3D drivers
[20:17] <persia> But we really should include the debs from the ship seed *somewhere*.  This may happen to be the empty set for some specific image, but that's coincidence.
[20:18] <persia> PPPoE is a good example of software most folks don't need and some cannot live without.
[20:18] <NCommander> persia: at risk of stupid question, why can't we pre-install it then?
[20:18] <ogra_cmpc> if you can tell a good reason to have the ship seed for anything i'm happy to prepopulate /var/cache/apt
[20:18] <persia> NCommander: Because it's useless for most folks.
[20:19] <NCommander> persia: we ship IPv6 stuff out of the box, that's useless for more folks than PPPoE :-P
[20:19] <ogra_cmpc> the only reason i see for ship is to have the closed stuff available
[20:19]  * NCommander agrees with ogra_cmpc 
[20:19] <persia> ogra_cmpc: Because some use cases for some devices require it.  Like I said, I don't care if you make an image where it's empty, but please make sure the facility is there to use the ship seed to populate *somewhere* on the result image for folks/devices/environments that do require it.
[20:19] <ogra_cmpc> in which case neither preinstallinf nor perpopulating the cache will work
[20:20] <persia> It's not about open/closed.  it's about stuff that most folk don't want and others need.
[20:20] <persia> Is there an open SD slot?
[20:20] <ogra_cmpc> no
[20:20] <ogra_cmpc> else we could use an installer
[20:20] <persia> No USB, no SD.  Anything?
[20:20] <persia> Any expansion of any sort?
[20:21] <ogra_cmpc> we will have devices that only come with a single SD slot and nothing else
[20:21] <persia> No USB?
[20:21] <persia> No expansion port?
[20:21] <NCommander> persia: nothing usable out of the box without a custom kernel
[20:21] <persia> Nothing?
[20:21] <persia> NCommander: Why not?
[20:21] <NCommander> persia: you'd have to remove the UART gadget and replace it with something else as far as I can tell
[20:21] <NCommander> and I'm not even sure if that's possible
[20:21] <persia> Oh, it has a serial port?
[20:21] <NCommander> persia: it has a mini-USB port, and a micro-USB port
[20:22] <NCommander> my Blaze won't boot if I have anything in the micro port
[20:22] <persia> That's USB.  I can attach all sorts of things to that.
[20:22] <ogra_cmpc> well, the blaze has eMMC
[20:22] <persia> And I might need to have my 3G dongle attached to be able to download updated software.
[20:22] <ogra_cmpc> and could boot from it
[20:22] <NCommander> persia: mini-USB is OTG, and stuck in Host mode out of the box
[20:22] <persia> Host mode means I can attach a dongle.
[20:22] <ogra_cmpc> thats a matter of dip switches
[20:22] <NCommander> ogra_cmpc: I thought it was panada that had eMMC, not Blaze, or do i have it backwards.
[20:22]  * NCommander probably has it backwards
[20:22] <ogra_cmpc> both do
[20:23] <NCommander> oh
[20:23] <ogra_cmpc> just have a look :)
[20:23] <NCommander> ogra_cmpc: ... then why can't we have an installer to the eMMC?
[20:23]  * persia didn't think panda specs were available yet, and wants a URL
[20:23] <ogra_cmpc> you have mmcblk0 and 1
[20:23] <ogra_cmpc> NCommander, because panda and blaze arent the only devices
[20:24] <persia> So, anyway, regardless of what's in the device, at least someone is going to have an awkward use case.
[20:24] <persia> And that person is going to need the stuff that usually goes in pool/
[20:24] <persia> So that stuff belongs somewhere in the image.
[20:24] <ogra_cmpc> erm, i'm wrong
[20:24] <ogra_cmpc> panda wont have eMMC
[20:26] <ogra_cmpc> persia, make a list of poackages that need to be in pool and dont need a pre-download EULA agreement and i'll happily add them to the cache
[20:26] <persia> Use the content of the ship seed.
[20:26] <persia> Same as for any other sort of image.
[20:27] <DanaG1> hmm, how do I get usb hard drives to automount, by-label, on a headless box?
[20:28] <prpplague> ogra_cmpc: the eMMC is available for expansion, so you could have a secondary mmc card, or use eMMC
[20:28] <DanaG1> usbmount doesn't work for me... I need persistent labels.
[20:28] <persia> DanaG1: Watch for udev events, and do something with them :)
[20:29] <ogra_cmpc> persia, so looking at the lucid netbook ship seed i dont see a single package i would ship
[20:30] <DanaG1> Unfortunately, udev alone doesn't seem to expose the label.
[20:31] <DanaG1>  udevadm info --query=all -a -p $(udevadm info -q path -n /dev/sda2)
[20:31] <persia> DanaG1: Your program would respond to udev, then perform operations on the device to do things like get the label, then mount it.
[20:31]  * persia looks at the maverick netbook seed
[20:31] <DanaG1> If I can run gnome-volume-manager (or an equivalent) in xpra, that'd work, too.
[20:32] <ogra_cmpc> DanaG1, just create a custom rule
[20:32] <ogra_cmpc> /etc/udev/rules.d/README has info
[20:33] <DanaG1> I found one here that tries to use an ENV that doesn't exist as an ATTR.
[20:33] <persia> ogra_cmpc: So, based on http://bazaar.launchpad.net/~ubuntu-core-dev/ubuntu-seeds/netbook.maverick/annotate/head:/ship-live I'd include pptp-linux, sl-modem-daemon, bpalogin, libatm1, setserial, b43-fwcutter, bcmwl-kernel-source,
[20:33]  * DanaG1 digs up link...
[20:33] <persia> Extra langpacks are nice, but it's the network stuff that7s critical.
[20:33] <DanaG1> http://superuser.com/questions/53978/ubuntu-automatically-mount-external-drives-to-media-label-on-boot-without-a-us
[20:34] <ogra_cmpc> we wont ship any langpacks
[20:34] <DanaG1> there's a udev rule... but it doesn't work.
[20:34] <ogra_cmpc> oem-config will install them if network is available
[20:34] <ogra_cmpc> if not, the image will stay english
[20:35] <persia> ogra_cmpc: Fine, but the network stuff is critical.
[20:35] <persia> Lots of folks have broadcomm.  Lots of folks still use modems.  Lots of folks need pptp auth to get out of the local net.
[20:36] <ogra_cmpc> what does b43-fwcutter gain you without the firmware ?
[20:37] <persia> I don't know.  I just know that if I do a fresh install on a machine, and install b43-fwcutter and bcmwl-kernel-source I get a working network,.
[20:37] <ogra_cmpc> you dont
[20:37] <persia> I do.  I did it on the 11th of May.
[20:37] <ogra_cmpc> you need trird party frimware files
[20:38] <persia> I really didn't.
[20:38] <ogra_cmpc> thats what b43-fwcutter does
[20:38] <ogra_cmpc> then you only need the source file
[20:38] <ogra_cmpc> b43-fwcutter extracts windows firmware files
[20:39] <persia> OK.  Maybe I have a windows driver CD that came with my dongle.
[20:39] <ogra_cmpc> fine then put the CD in your pandaboard :P
[20:40] <persia> No problem.  I have USB DVD drives around.
[20:40]  * persia should really get a new one, as the one that uses PS/2 power is mostly obsolete, and the other one requires *two* USB ports
[20:41] <ogra_cmpc> my point is that you have used another computer to download your image and ot write the SD, why cant you just copy debs with that computer on the SD ?
[20:41] <persia> Because there's no space in the filesystem
[20:41] <ogra_cmpc> ??
[20:41] <ogra_cmpc> the filesystem is as big as the card
[20:41] <DanaG1> weird.... ata_id and scsi_id are both returning empty strings!
[20:41] <persia> I'd have to do the install, and then post-install monkey with the SD card, and then install more stuff, etc.
[20:42] <ogra_cmpc> right
[20:42] <persia> ogra_cmpc: Not at the time I dd it.  Only post-install.
[20:42] <persia> So, that's an unpleasant experience.
[20:42] <ogra_cmpc> alternatively you have to bastardize livecd-rootfs to install ship inside the image
[20:42] <persia> Plus it fails the case where I get passed an already burned SD card somewhere from some guy in an overcoat
[20:42] <ogra_cmpc> pathces accepted if the cdimage team accepts them
[20:42] <persia> (yes, this happens: I have been on both sides of such an SD exchange)
[20:43] <persia> OK,  As a rep of the CD image team, do you expect them to be accepted?
[20:43] <ogra_cmpc> no idea, depends on the code :)
[20:43] <persia> NCommander: Please present ogra with code to review :)
[20:44] <ogra_cmpc> pfft
[20:44] <ogra_cmpc> thats lame
[20:44] <ogra_cmpc> NCommander has enough work with debian-cd atm
[20:44] <persia> Well, I guess.  I'm just not expecting to get to it for 3-4 weeks, and that's probably late to jam it in.
[20:44] <persia> Very true.
[20:45] <ogra_cmpc> just do it if you have time
[20:45] <ogra_cmpc> i'm happy to review it
[20:45] <ogra_cmpc> and to merge it too
[20:45] <persia> OK, consider my protest registerd.  I'll send a patch when I can.  Otherwise, I'll complain about it for next time.
[20:45] <ogra_cmpc> next time livecd-rootfs wont exist anymore
[20:46] <persia> What?  Why not?
[20:46] <persia> Or is it being reimplemented differently?
[20:47] <ogra_cmpc> d-cd and l-r are being dropped
[20:47] <persia> In favour of?
[20:47] <ogra_cmpc> apparently someone decided to switch everything to luvehelper
[20:47] <ogra_cmpc> *live
[20:47] <persia> Everything?
[20:47]  * persia clearly needs to read more of the specs
[20:47] <ogra_cmpc> i havent seen a spec
[20:48] <ogra_cmpc> it was a m+1 decision apparently
[20:48] <ogra_cmpc> cody told me at the corridor
[20:48] <ogra_cmpc> at UDS
[20:48] <persia> I'm reluctant to believe that level of infrastructure migration would happen wholesale without a spec.
[20:48] <ogra_cmpc> m+1 qill likely have a spec then
[20:48] <ogra_cmpc> will
[20:49] <persia> But anyway, I'll send you a patch in July, *OR* I'll make sure there is a pool/ equivalent on produced images in m+1 (regardless of the underlying technologies.)
[20:51] <ogra_cmpc> pool/ wont happen, cache prepopulation can though
[20:51] <persia> Oh, of course.
[20:51] <ogra_cmpc> hmm
[20:52] <persia> I'm not fussed about the name, so long as users can end up with a working system post-install without needing to mount the storage on *another* device.
[20:52] <ogra_cmpc> pool/ might if we pick to ship with a huge vfat
[20:52] <ogra_cmpc> though my plan waws to make that hidden actually
[20:52] <persia> huge vfat?  Why?
[20:53] <persia> But yeah, pool/ on vfat is fine, and can be wiped if the user doesn't need it.
[20:53] <ogra_cmpc> because we boot from that vfat
[20:53] <persia> If it's just a boot vfat, it ought be small, I think.
[20:53] <ogra_cmpc> so the boot partition would be bigger but carry pool/
[20:53] <ogra_cmpc> indeed that makes the plan to make it a hidden partition tricky
[20:53] <persia> I don't like that.  I'd rather prepopulate the cache.
[20:53] <ogra_cmpc> ok, fine as well
[20:53] <persia> I don't think you want to hide the partition.
[20:54] <persia> For the EFI/GPT case, the FAT partition is unhidden, which people seem to like.
[20:54] <ogra_cmpc> i want to hide the partition
[20:54] <ogra_cmpc> else it gets mounted
[20:54] <DanaG1> persia: you can hack up a ps/2 to usb-power-only cable.
[20:54] <DanaG1> or rather, the other way around.
[20:55] <ogra_cmpc> that fat partition is our replacement for the NAND and will be handled similar
[20:56] <persia> DanaG1: both ways, actually, but I've seen new devices that have an onboard capacitor to handle spinup with only a 5V/500mA draw.
[20:56] <DanaG1> ah yeah, usb->ps/2->usb.
[20:56] <persia> ogra_cmpc: Doesn't have to mount.  The FAT partition for EFI/GPT doesn't get mounted.
[20:56] <DanaG1> Suck power from it, while passing it right back to USB protocol.
[20:56] <ogra_cmpc> persia, it shows up in nautilus
[20:56] <persia> ogra_cmpc: Not on a new macbook.
[20:56] <ogra_cmpc> i dont want people to be able to muck about with it
[20:57]  * ogra_cmpc didnt know there will be panda based macbooks :P
[20:57] <DanaG1> http://www.cablesdirect.com/prodimages/USB-PS2F_LR.jpg
[20:57] <ogra_cmpc> TI will be happy to hear that :D
[20:57] <DanaG1> I wish my UEFI weren't broken. =(
[20:57] <persia> ogra_cmpc: The point isn't panda-macbooks.  The point is that we have technology that causes special bootloader partitions not to be mounted by nautilus.
[20:58] <ogra_cmpc> right
[20:58] <persia> So we ought use that to do this, rather than fiddling with hidden partitions.
[20:58] <ogra_cmpc> we also have udev rules that hide such partitions completely
[20:58] <persia> Right.
[20:58] <DanaG1> HP_TOOLS is another example.
[20:58] <persia> Which is the right way to do it, not with the hidden flag in the parition table.
[20:58] <DanaG1> HP boots in CSM mode, but has their EFI partition for stuff like QuickLook.
[20:58] <ogra_cmpc> i didnt talk about flags
[20:59] <persia> Oh, then I've misunderstood you.  Please forget I said anything :)
[20:59] <ogra_cmpc> but i want it completely hidden so users dont fiddle manually with the contents and instead use the provided tools
[20:59] <persia> You can't do completely hidden.
[20:59] <ogra_cmpc> you can through udev
[20:59] <persia> Closest is with partition table flags (which I'll argue against)
[20:59] <persia> No, you can't.
[21:00] <persia> All udev does is not provide a /dev/ link, at best.
[21:00] <ogra_cmpc> the system wont see it
[21:00] <DanaG1> Isn't there something like UML that'll let different apps see different file systems?
[21:01] <persia> DanaG1: Yes.
[21:01] <DanaG1> sudo scsiadd -p
[21:01] <DanaG1> scsiadd:scsidump():could not open /proc/scsi/scsi (r): No such file or directory
[21:01] <DanaG1> argh... for some reason, that udev rule doesn't work.
[21:01] <persia> ogra_cmpc: You really don't need to go to that extent.  Look up how the EFI/GPT case is handled for Intel Macs.  That7s well-tested code.
[21:01] <DanaG1> It acts as if ENV{ID_FS_LABEL_ENC} does not exist.
[21:01] <DanaG1> Intel Macs are weird... they don't use protective MBR.
[21:02] <persia> Well, kinda.
[21:02] <DanaG1> I wish Apple would recommend using EFI-installed Windows instead of CSM-installed Windows.
[21:02] <persia> they leave space for it, but don't pay any attention to it.
[21:03] <ogra_cmpc> i just plan to add it to /lib/udev/rules.d/80-udisks.rules at the bottom
[21:03]  * ogra_cmpc needs to go now
[21:30] <DanaG1> ah, fixed the udev rule.
[21:30] <DanaG1> had to change SUBSYSTEM=usb to SUBSYSTEM=block
[21:30] <DanaG1> and rename a file I had mis-named.
[22:25]  * NCommander is dealing with some really sucky internet in his test builds :-/