[04:55] <eggonlea> ogra: new rootstock works here! Thanks!
[04:56] <eggonlea> after install qemu-arm-static from Karmic (i'm using Jaunty)
[06:50] <siji> need help
[06:50] <siji> to compile clutter with SGX support on OMAP
[09:35]  * NCommander installs bootchart
[09:35] <ogra> eggonlea, good to hear !
[09:35]  * ogra did tests until 2am
[09:36] <ogra> eggonlea, if you find any issues, please file bugs, i'm scared that something regressed
[10:17] <lool> NCommander: So usplash just works on dove right?
[10:17] <NCommander> lool, works now. No idea what changed.
[10:17] <NCommander> lool, probably new kernel or driver updates
[10:17]  * NCommander is doing the usplash boot charts now
[10:18] <NCommander> it doesn't look like theres much difference if its enabled or not
[10:24] <lool> I think there was an issue in initramfs-tools
[10:24] <NCommander> lool, possibly. no point really dwelling on it unless it re-emerges
[10:27] <NCommander> lool, ogra http://people.canonical.com/~mcasadevall/bootchart/ - initial bootcharts (for some reason, a bunch of the usplash ones, and one of the no-usplash ones didn't fully write out, and corrupted the tgz). No real difference between having usplash and not having it on dove
[10:27] <ogra> give it time :)
[10:27] <ogra> the java conversion tool is very slow
[10:28] <ogra> erm
[10:28] <ogra> why doesnt your bootchart stop ?
[10:28] <ogra> how did you install it ?
[10:28] <NCommander> ogra, apt-get install bootchart?
[10:29] <NCommander> ogra, it stops ...
[10:29] <ogra> way after mine
[10:29] <ogra> http://people.canonical.com/~ogra/babbage2-karmic-20090916-2.png
[10:29] <ogra> http://people.canonical.com/~ogra/babbage2-karmic-20090916-3.png
[10:29] <NCommander> ogra, the conversion tool is very slow on ARM, its fine on my desktop
[10:29] <NCommander> ogra, I don't get it ....
[10:30] <ogra> i dont see the S99bootchart-stop initscript in yours at all
[10:30] <ogra> its java
[10:30] <NCommander> ogra, I'm not even logging in ...
[10:31] <ogra> so your xorg comes up after 14sec
[10:31] <NCommander> ogra, I don't get it, but I usually let the system sit at gdm for awhile, more thatn a minute, so it is stopping
[10:31] <NCommander> ogra, the framebuffer comes up at the 6-7 second mark
[10:32] <ogra> it should be up after the first modprobe ran
[10:32] <NCommander> ogra, oh, it comes up, but it take a bit of time to settle
[10:33] <ogra> woah, 49MB/s
[10:33] <ogra> is that SATA ?
[10:33] <NCommander> ogra, :-)
[10:33] <ogra> or USB ?
[10:33] <NCommander> SATA
[10:33] <ogra> ah
[10:33] <NCommander> but USB pretty fast on this too
[10:33] <ogra> well, i get 33 on my slow USB key
[10:33] <NCommander> lool, ogra, https://bugs.edge.launchpad.net/ubuntu/+source/linux-mvl-dove/+bug/435046 - can you hit the nominate button
[10:34] <ogra> no, but the approve one :P
[10:34] <NCommander> ogra, er :-)
[10:34] <ubot4> Launchpad bug 435046 in linux-mvl-dove "Ethernet port on the dove sometimes changes MAC addresses" [Undecided,New]
[10:35] <lool> NCommander: done
[10:35] <NCommander> thanks
[10:35] <lool> NCommander: Give a heads up to bjf on it when you catch him
[10:35]  * NCommander is on eth30 right now 
[10:35] <NCommander> Yeah
[10:35] <NCommander> I think its going up every boot now
[10:35] <ogra> why do you still use mem=512M ?
[10:35] <ogra> and console=tty0 ?
[10:35] <NCommander> ogra, because it doesn't bring up the full memory if you don't :-/
[10:36] <ogra> is that filed ?
[10:37] <ogra> i also see no "ro" in your cmdline
[10:37] <ogra> that should be added
[10:37] <NCommander> ogra, it was a known issue from Marvell I believe, but I don't think we have a specific bug on it
[10:37] <lool> NCommander: You might want to attach dmesg to the bug
[10:37] <ogra> and get it in the right order so quiet and splash sit at the end
[10:37] <lool> NCommander: Did you try lamont's uinitrd?
[10:37] <ogra> NCommander, we need to track it
[10:37] <NCommander> lool, not yet, my TFTP server is broken at the moment, and I just got console access to the dove back about an hour ago
[10:38] <ogra> tftp ?
[10:38] <ogra> i thought you made it run from disks
[10:38] <lool> ogra, NCommander: Please fix the default kernel cmdline now or file a bug about it
[10:38] <lool> The mem= should be a separate bugs
[10:38] <ogra> yeah
[10:39] <NCommander> lool, fixing it now in my branch, once I go over it and make sure its ok, I'll post a comment asking for review
[10:39] <ogra> NCommander, root=UUID=<$uuid> ro quiet splash
[10:39] <ogra> thats all you should have ... in this order
[10:40] <ogra> the ro is important for fsck
[10:40] <NCommander> ogra, at the risk of a stupid question, does that order actually matter?
[10:40]  * NCommander is geniunely curious ...
[10:40] <ogra> its mentioned in docs all across the wiki
[10:40] <ogra> so it makes sense to keep the same order the rest of ubuntu uses
[10:40] <lool> NCommander: it does not but let's keep it consistent
[10:41] <NCommander> right, no problem
[10:41] <ogra> doesnt matter technically
[10:41] <NCommander> ogra, good catch on that, changed on my flash-kernel branch. Once I make sure the installer still works with my changes, I'll kick a branch up
[10:41] <ogra> good
[10:42] <NCommander> yeah, mem=512M is still needed
[10:42]  * NCommander files a bug
[10:43] <ogra> hmm, if i only could find my voltmeter
[10:44] <ogra> lool, battery bug still presists even after 2 days of charging
[10:45] <lool> ogra: Reopen it and note that you didn't test the actual voltage
[10:47]  * NCommander is at eth31
[10:50]  * ogra wonders when NCommander runs out of numbers :)
[10:53] <NCommander> ogra, probably eth255
[10:53] <NCommander> ogra, I already ran out of DHCP leases -_-;
[10:53] <NCommander> ogra, (that's how I found the magic MAC numbers)
[10:54] <ogra> set a shorter lease time :)
[10:54] <NCommander> ogra, I did
[10:54] <NCommander> udev assigns a new ethXXX number because the MAC address differs I think
[10:54] <ogra> indeed
[10:55]  * NCommander filed a bug on this but still this is redicious
[10:55] <NCommander> (32)
[10:57] <NCommander> ogra, lool: https://bugs.edge.launchpad.net/ubuntu/+source/linux-mvl-dove/+bug/435151 - someone accept the nomination please :-)
[10:57] <ubot4> Launchpad bug 435151 in linux-mvl-dove "dove kernel requires mem=512M to use all available memory" [High,New]
[11:01] <lool> NCommander: Do you also see oo.o not starting on dove?
[11:01] <lool> NCommander: I dont think that needs to be tracked for release
[11:01] <lool> Ethernet issue is important, mem=512m cleanup is nice to have
[11:02] <lool> "High" uh
[11:02] <NCommander> *coughs*
[11:02] <NCommander> lool, the problem is if you have a board with more that 512MiB (like at Marvell) it won't be able to see it.
[11:02] <lool> NCommander: now that you learnt how to properly get a bug on the radar you need to learn how not to  :-)
[11:03] <lool> NCommander: I can see it'sa real life issue which wont hit any of us and will only limit amount of RAM
[11:03] <lool> Is this blocking the release of karmic?  no
[11:03] <lool> Is this a nice to have for karmic?  certainly
[11:03] <lool> Could we decently release an image without that fix?  yes
[11:04]  * NCommander doesn't remember the last time he didn't file a bug that was High or Critical ...
[11:04] <NCommander> -_-;
[11:04] <lool> Critical should really be exceptional
[11:04] <lool> I might have filed a couple, or perhaps three total
[11:05] <NCommander> Two or three, dunno the specifics
[11:05] <NCommander> Its usually one per cycle
[11:05] <NCommander> OOo starts ...
[11:06] <NCommander> and then crashs with the InteractiveAugmentedIOException
[11:06] <NCommander> But I do get the splash screen
[11:06] <NCommander> ^- lool ogra
[11:08] <lool> NCommander: that's with latest karmic?
[11:08] <NCommander> lool, do you want me to start investing time in the OOo bug after the last few dove issues are firmly squashed?
[11:08] <NCommander> uh
[11:08]  * NCommander didn't apt-get update today actually
[11:08] <ogra> well, thats the normal oo.o behavior
[11:08] <lool> NCommander: You get the same output as in https://bugs.launchpad.net/ubuntu/+source/openoffice.org/+bug/417009 ?
[11:08] <ubot4> Launchpad bug 417009 in openoffice.org "all openoffice apps die in 'com::sun::star::ucb::InteractiveAugmentedIOException' on armel in karmic" [High,Confirmed]
[11:08] <NCommander> no
[11:08] <NCommander> I don't get the first error
[11:09] <lool> that one javaldx: Could not find a Java Runtime Environment! ?
[11:09] <NCommander> I only get "terminate called after throwing an instance of 'com::sun::star::ucb::InteractiveAugmentedIOException'"
[11:09] <lool> NCommander: Ok and it fails to start?
[11:09] <NCommander> yeah, fails to start
[11:09] <lool> Then it's the same on non-cortex-a8
[11:09] <ogra> because you installed bootchart
[11:09]  * NCommander notes he DOES have java packages installed though
[11:09] <ogra> it pulls in java
[11:10] <ogra> by default the oo.o java extensions shouldnt be installed due to CD size constraints
[11:10] <NCommander> ogra, right. Anyway, I think it might be worth building a rawhide/Fedora chroot, and see if their armel port works w/ OOo
[11:10] <ogra> i doubt they use our toolchain or compiler version
[11:10] <ogra> and it works fine with libgcc3_uno.so from jaunty copied in place
[11:11] <NCommander> right
[11:11] <NCommander> I knew that
[11:11]  * NCommander needs coffee
[11:12] <NCommander> ogra, it seems like the same issue we had w/ thunderbird
[11:12] <ogra> why do you say that with every second bug ?
[11:13] <NCommander> ogra, say what? need coffee or?
[11:13]  * ogra doesnt see any relation between TB and OO.o 
[11:13] <NCommander> ogra, er, more specifcally, the cause, which was an issue with preprocessor macros, but I'm no expert on this bug
[11:14] <ogra> TB had build failures
[11:14] <ogra> OO.o fails to run
[11:15] <NCommander> ogra, no, it was fail to run mostly
[11:15] <NCommander> lool, I just had a nasty thought that our dove images might break on whatever hardware Marvell might roll out in the future
[11:16] <NCommander> even if its Dove compatible
[11:16] <ogra> suppress such thoughts ?
[11:16] <NCommander> ogra, :-p
[11:25] <lool> NCommander: We need to wait until we see more of the future
[11:25] <lool> NCommander: or perhaps you can check your crystal ball and tell me how Hardware will look like on these devices and how compatible they will be
[11:26] <lool> NCommander: Remember that the kernel needs to be ported too, adding machine-ids etc.
[11:26] <NCommander> My magic 8 ball says "Try Again Later"
[11:26] <NCommander> :-/
[11:27] <NCommander> lool, hrm ... I completely forgot that the kernel also uses the Hardware field
[11:27] <NCommander> Ok, feel free to ignore me
[11:27] <siji> hello all
[11:28] <siji> i successfully compiled clutter with GLX support
[11:28] <siji> While trying to compile clutter aplication
[11:28] <siji> it's giving a error like
[11:29] <siji> Glib ERROR /build/buildd/glib2.0-2.20.1/glib/gmem.c:136:failed to allocate 1989214296 bytes
[11:33] <siji> any idea
[12:17] <lool> NCommander: so you booted lamont's kernel + initrd successfully?
[12:17] <NCommander> lool, yeah, I got to the (initramfs) prompt
[13:39] <lool> ogra: is MX51 TO 3.0 a separate file in redboot-imx or is it the same file for to2 and to3 boards?
[13:39] <lool>    + Add display support on MX51 & MX37 3-stack platforms
[13:39] <lool> cool
[13:39] <lool> ah wait that's LCD
[13:39] <ogra> there is no romram_TO3
[13:39] <ogra> so i dont see where it would have been added
[13:39] <ogra> the changelog mentiones it though
[13:40] <ogra> and yes, no change for our setup wrt displays
[13:40] <lool> ogra: Hmm I only see -rw-r--r-- root/root    171352 2009-09-23 12:53 ./usr/lib/redboot/imx51-babbage-TO2_redboot.bin
[13:40] <lool> ogra: Did you confirm that works on babbage 2.0/2.5?
[13:40] <ogra> thats how we named it, yes
[13:40] <ogra> with 2.5
[13:40] <ogra> i dont have a 2.0
[13:41] <lool> Yeah good enough
[13:41] <lool> ogra: Perhaps we should add a TO3 symlink
[13:41] <ogra> i test the imx51-babbage_redboot.bin on 1.0 and imx51-babbage-TO2_redboot.bin on 2.5 usually
[13:41] <ogra> well, how would we know it actually supports TO3
[13:41] <ogra> there is no indication in the patch
[13:41] <lool> ogra: You quoted the release notes
[13:41] <ogra> i copy pasted the release notes, yes
[13:42] <lool> Well you say in the ubuntu changelog that it now supports TO3
[13:42] <ogra> but couldnt find a specific code reference to TO3
[13:42] <ogra> upstream says that
[13:42] <lool> hmpf
[13:42] <ogra> there is nothing specific in the imx51 patch that has TO3 attached to it
[13:43] <ogra> ogra@osiris:/var/build/redboot_200933/package$ grep TO3 patch-redboot-200933-mx51
[13:43] <ogra> ogra@osiris:/var/build/redboot_200933/package$
[13:43] <ogra> and thats the original upstream patch, not the modified dpatch ... (which has CVS dirs stripped)
[13:44] <ogra> ogra@osiris:/var/build/redboot_200933/package$ grep TO3 patch-redboot-200933-base
[13:44] <ogra> ogra@osiris:/var/build/redboot_200933/package$
[13:44] <ogra> same here
[13:45] <ogra> if TO3 hw works with it, fine ... but nobody added a comment or anything when the code was added
[13:45] <ogra> i simply trust what they write in their changelog, what else can i do
[13:47] <lool> ogra: Well either you believe there's TO3 support, then you quote the release notes and assume that what you're building TO3 support, or you dont in which case you dont quote the release notes
[13:47] <ogra> well, i belive upstream
[13:47] <ogra> and there is no specific additional TO3 patch or something
[13:48] <ogra> patch-redboot-200933-mx51.bz2 is the only thing we have for babbage and 3stack imx51
[13:48] <ogra> and from that i only build the babbage targets
[13:49] <ogra> i guess the TO2 binary currently has basic support for TO3 and a proper target will be added in a future release, thats how they did it for TO2
[13:52] <lool> So if the TO2 binary has support for TO3 my proposal stands
[13:52] <lool> 14:41 < lool> ogra: Perhaps we should add a TO3 symlink
[13:52] <ogra> ah well, but how would we test TO3 support ?
[13:53] <ogra> sure i can easily add a link
[13:53] <lool> ogra: It's in there since you trust upstream  :-)
[13:53] <ogra> :P
[13:53] <ogra> ok, i'll add a link :)
[13:56]  * ogra wonders about 435221
[13:56] <lool> bug #435221
[13:56] <ubot4> Launchpad bug 435221 in project-rootstock "rootstock-0.1.2 crashes on missing /dev/pts" [Undecided,New] https://launchpad.net/bugs/435221
[13:56] <ogra> how can a system be in such a state if deboostrap finished properly
[13:56] <ogra> (no update-rc.d no udevd)
[13:59] <lool> Well do you have sysv-rc?
[14:00] <lool> hmm probably you do
[14:00] <ogra> its a finished debootstrap (at least it should be)
[14:00] <ogra> there should be a /dev/pts as well as a udevd and update-rc.d
[14:01] <ogra> i wonder if the new debootstrap from today causes that
[14:01] <ogra> i did my tests before it was uploaded
[14:01] <ogra> but we'll see, i need more info first
[15:11] <lool>   * ARM: IMX51: Make fec ethernet driver compile
[15:11] <lool>   * [Config] Disable FEC2 for imx51
[15:12] <ogra> yeah, just saw that
[15:13] <ogra> * ARM: IMX51: Fix USB errors ... should make GrueMaster happy
[15:13] <lool> GrueMaster: ^ check it out
[15:13] <ogra> once it built
[15:13] <GrueMaster> yea, I talked to amitk about that yesterday during the sprint.
[15:13] <ogra> i guess he did already though, since he spnt a day on amits lap to get it fixed
[15:14] <GrueMaster> Well, we weren't that close.  :P
[15:14] <GrueMaster> Has anyone else tried SATA?
[15:15] <ogra> no cable
[15:15] <GrueMaster> (not sata through usb).
[15:15] <ogra> well
[15:15] <ogra> SATA is SATA through USB
[15:15] <ogra> no matter how you attach it :P
[15:15] <GrueMaster> on-board sata.
[15:15] <ogra> is a USB bridge
[15:16] <GrueMaster> understood, but electrically it is very different than an external HD USB adapter.
[15:40] <rabeeh> hi
[15:40] <rabeeh> after 'apt-get upgrade' on karmic i get the following -
[15:40] <rabeeh> root@dove:~# gconftool
[15:40] <rabeeh> gconftool: relocation error: /usr/lib/libglib-2.0.so.0: symbol __abort_msg, version GLIBC_PRIVATE not defined in file libc.so.6 with link time reference
[15:40] <rabeeh> anyone? ogra? NCommander? lool?
[15:41] <NCommander> rabeeh, that error sounds vaguely familiar
[15:42] <rabeeh> looking at the logs; i see the 'apt-get upgrade' failed too with the same reason.
[15:42] <ogra> weird, i dont get it on imx51
[15:43] <NCommander> ogra, I know there was something dealing with __abort_msg's as a build failure a month or so back
[15:44] <rabeeh> i think that the reason might be libglib is being upgraded before libc
[15:45]  * NCommander notes he usually uses dist-upgrade verus upgrade ...
[15:46]  * ogra notes that he doiesnt use apt at all since we need to test update-manager :P
[15:50] <lool> rabeeh: You need to update glib and glibc
[15:50] <lool> rabeeh: Can you file this against glibc please?
[15:50] <lool> and sub us
[15:51] <NCommander> rabeeh, stupid question, are you using one of the official dove images?
[15:51] <ogra> us being ubuntu-armel :)
[15:51]  * NCommander sighs
[15:51] <rabeeh> NCommander: i'm just using karmic from ports.ubuntu.com --> not a live image
[15:52] <rabeeh> i'm building my own kernel
[15:52] <NCommander> rabeeh, how'd you install it?
[15:52] <ogra> well, the live image is an install media ;)
[15:52]  * NCommander notes we do have official images now, with our u-boot modifications public
[15:52] <rabeeh> NCommander: can you please send a link?
[15:52] <NCommander> rabeeh, https://wiki.ubuntu.com/ARM/MarvellDoveKarmicInstall
[15:53] <NCommander> http://cdimage.ubuntu.com/ports/releases/9.10/alpha-6/
[15:53] <rabeeh> NCommander: For installing my Karmic, I used debootstrap under a Debian/Lenny chroot, then used that as a base
[15:53] <rabeeh> NCommander: I haven't used rootstock or other tools
[15:54] <NCommander> rabeeh, that works, but you don't end up with an Ubuntu ramdisk, or our fancy autostart solution :-)
[15:54] <rabeeh> :)
[15:54] <rabeeh> Well, in my mind ramdisk, initrd and other stuff just makes my boot slower :)
[15:55] <ogra> well, you dont have an ubuntu system then
[15:55] <ogra> lots of stuff in userspace relies on stuff that was set up in initramfs
[15:55] <rabeeh> i have a karmic filesystem
[15:55] <rabeeh> like what?
[15:56] <ogra> right, you have a filesystem
[15:56] <NCommander> rabeeh, actually, after leaving u-boot, we hit desktop pretty quickly
[15:56] <ogra> dm-raid, encrypted dirs, lvm, splashscreen are some i can name off the top of my head
[15:57] <NCommander> rabeeh, the slowest bit ATM is the physical read from disk (and I'm starting to think its just the ext2 code)
[15:57] <ogra> beyond that i heard that the new upstart has issues with missing initramfs
[15:57] <ogra> but thats only rumors :)
[15:57] <NCommander> ogra, might explain why my SPARC currently panics on startup
[15:57] <rabeeh> ogra: i noticed that after apt-get update, udev doesn't show up (so i don't have /dev/ttyS0 to open a console on)
[15:58] <ogra> right
[15:58] <ogra> make sure to use an initramfs, its not a slowdown in karmic anymore
[15:58] <NCommander> rabeeh, in general, debootstrap based installs come with a big unsupported label :-/
[15:58] <ogra> yeah
[15:58] <NCommander> actually
[15:58] <ogra> thats why i'm so cautious to promote rootstock to loudly
[15:58] <NCommander> Pretty much anything short of the alternative or live installer has that label
[15:59] <ogra> while its noce for unsupported HW its really nothing you should use on imx51 or dove anymore
[15:59] <ogra> *nice even
[15:59] <rabeeh> i'll try the dove images.
[16:00] <rabeeh> NCommander: mechanical drives suffer a lot from seek time on boot. I previously compared boot with SSD vs. 3.5" drive and that was ~50% faster
[16:00] <rabeeh> (with xfce4)
[16:00] <NCommander> rabeeh, actually, I think its the filesystem code just being ....
[16:01] <NCommander> rabeeh, it shouldn't take 10 seconds to load a 3MiB file from a rotary drive, especially if that drive is spinned up
[16:01] <rabeeh> NCommander: then something is wrong with your drive.
[16:01] <Martyn> NCommander: Can, if the drive is ATA and neither PIO nor DMA is engaged
[16:01] <rabeeh> I know for a fact that dove can easily drive HDD full wire speed without issues (~80MB/sec on read with latest 7200rpm drives)
[16:02] <NCommander> rabeeh, in u-boot?
[16:02] <NCommander> rabeeh, once the system boots, its fine
[16:02] <rabeeh> oh
[16:02] <NCommander> But its the IPL of the kernel and ramdisk that just D-R-A-G-S on
[16:02] <ogra> SSD is definately a lot faster than rotary
[16:02] <rabeeh> NCommander: so probably you need to mkfs.ext2 with -b 4096 (4kb block size)
[16:03] <NCommander> rabeeh, I'll make sure I teach the installer about that when I teach it about u-boot partition types :-/
[16:03] <ogra> all the 10 second boot targets for karmic+1 rely on SSD
[16:03] <rabeeh> i recall i had an issue like this; and that seriously boosted the performance
[16:03] <ogra> you will boot 5 sec slower with rotary
[16:04] <NCommander> rabeeh, that's good to know. I wouldn't be suprised if u-boot makes some assumptions about partition information that's disappeared on a booted system
[16:05] <NCommander> rabeeh, anyway, the recommended way to install Ubuntu on dove is to grab the Canonical modified u-boot, grab an alpha 6 image, and play flip the DIPs
[16:05] <rabeeh> NCommander: it's definetely ext2 code in u-boot since it's running with I-cache enabled, but with D-cache disabled
[16:05] <NCommander> rabeeh, ow
[16:05] <NCommander> rabeeh, the only other choice is vfat ...
[16:05] <NCommander> ow
[16:05] <NCommander> wb Martyn
[16:05] <Martyn> thanks
[16:05] <rabeeh> NCommander: with '-b 4096' block sizes will be 4 times larger, so much less code to run
[16:05] <Martyn> That's what I get for trying to type on two consoles at once...
[16:05] <Martyn> I rebooted the wrong machine
[16:06] <NCommander> rabeeh, *wince*, we have an ext2 /boot partition, thats going to make files larger though
[16:06] <NCommander> Martyn, have you seen any issues with ext2 speed in u-boot?
[16:06] <Martyn> No
[16:06] <NCommander> Martyn, what version of u-boot?
[16:06] <Martyn> fat16/fat32 and ext2 work fine in my u-boot
[16:06] <Martyn> tips
[16:06] <rabeeh> Martyn: how big is your drive?
[16:07] <NCommander> Martyn, work, and work well are two different things ;-)
[16:07] <Martyn> I can go back from HEAD and see if there may have been an issue in the past, but at the moment, at least of three days ago, no issues
[16:07] <NCommander> Martyn, oh, *ahem*
[16:07] <NCommander> Martyn, our u-boot is 1.3.4 based :-)
[16:07] <rabeeh> NCommander: all u-boots are the same.
[16:07] <Martyn> rabeeh : Depends on which test bench machine .. the PBX A9's all have relatively small drives (80Gb)  and the other one has a 1Tb drive, but boots the same
[16:07] <Martyn> (other one = platform I can't talk about)
[16:08] <NCommander> rabeeh, I honestly won't be suprised if the ext2 code has been heavily modified in four major versions since 1.3.4
[16:08] <Martyn> rabeeh : Indeed NCommander is right there .. take a look at the checkins surrounding ext2
[16:08] <Martyn> NCommander: Are you stuck with the old version for a reason?
[16:08] <NCommander> rabeeh, at the moment, the lag is tolerable though
[16:09] <NCommander> Martyn, vendor provided u-boot
[16:11]  * NCommander hasn't tested load sleep from SSD/ext2 or SATA/vfat to rule out the drive
[16:12] <rabeeh> Who is Michael Casadevall
[16:12] <ogra> rabeeh, that guy you talked to the last minutes ;)
[16:12] <rabeeh> NCommander?
[16:13] <rabeeh> :)
[16:13] <ogra> yeah
[16:13]  * NCommander has been identified
[16:13] <rabeeh> NCommander: you have lots of warnings on upgrading u-boot on dove board. you seem hurt :)
[16:13] <Martyn> Beep beep :)
[16:13]  * Martyn laughs
[16:14] <NCommander> rabeeh, none of us have JTAG, or a known-working recovery method
[16:14] <Martyn> warnings?  Something new on the list I should have read?
[16:14] <NCommander> Martyn, https://wiki.ubuntu.com/ARM/MarvellDoveKarmicInstall
[16:14] <rabeeh> NCommander: come on; I wrote for you how to configure the bootrom to boot from the rs-232
[16:14] <Martyn> NCommander: <--- waiting for his Dove board
[16:14] <Martyn> then again, I'm also waiting on a new A9 based board too .. lots of frigging _waiting_
[16:14] <NCommander> rabeeh, and we never managed to get it work :-/. That's why we install u-boot to NAND now and have SPI as a backup
[16:14] <NCommander> (or at least I never did, I dunno if anyone else tried)
[16:15] <NCommander> er
[16:15] <NCommander> whoelse
[16:15] <ogra> NCommander, ? we all have jtags ?
[16:15] <NCommander> bah
[16:15] <NCommander> ogra, and no support scripts to debrick with them :-/
[16:15] <rabeeh> NCommander: i'll try and provide you exact instructions (again) :)
[16:15] <ogra> no idea
[16:15] <ogra> but i definately got that little board thats wrapped in a condom and fits into the jtag socket on the dove
[16:15] <NCommander> rabeeh, well, the problem might be that we don't use windows here
[16:16] <rabeeh> NCommander: as a cookie, we are considering adding FTDI chip on next Dove board; that will give you USB to serial and USB to JTAG for openocd
[16:16]  * Martyn goes back to fighting to build more of the userspace full v7
[16:16] <NCommander> ogra, oh, sure, I got that too, but openbcd doesn't support Dove
[16:16] <rabeeh> NCommander: minicom works too. it supports kermit
[16:16] <Martyn> rabeeh : PLEASE do so, if at all possible.
[16:16] <NCommander> rabeeh, no love here, when encoded to 0x0E, I believe I just get BootROM
[16:16] <NCommander> *BootROM 1.07>
[16:17] <rabeeh> NCommander: it will. like openocd didn't support sheevaplug; but now it does
[16:17] <rabeeh> Martyn: This is on the top of the list
[16:17] <NCommander> rabeeh, then I can be a little less scary with the dire warnings ;-)
[16:17] <rabeeh> NCommander: congratualations. you'v got bootrom now :)
[16:18] <Martyn> rabeeh : Marvel is now diverging from the ARM roadmap though, right?   No dual core A9 based chip, but something different instead?
[16:18] <NCommander> rabeeh, ENEEDINSTRUCTIONS ;-)
[16:18] <rabeeh> bootrom is the rom inside the Dove chip.
[16:18] <NCommander> rabeeh, I gathered that much. There's a distinct lack of documentation on what you can do at that prompt
[16:18] <ojn> Martyn: Why should Marvel follow ARMs roadmap?
[16:18] <ojn> they make their own cores
[16:18] <ojn> +l
[16:18] <Martyn> I know
[16:19] <ojn> (since we're not talking comics here :)
[16:19] <Martyn> Yeah yeah ...
[16:19] <NCommander> ojn, I'm just waiting for some superhero called THE ARM to come into existence and then make the resulting bad puns
[16:19] <Martyn> However, I'm curious as to whether a multicore chip is in the works...
[16:19] <Martyn> (that's v7, rather than v5 based)
[16:20] <ojn> Something to talk to them under NDA about, most likely.
[16:21] <NCommander> rabeeh, stupid question, why were you looking for me? (I have my nick on the Dove installation page)
[16:22] <Martyn> ojn : Well,a s long as rabeeh is on channel .. can't hurt to ask :)
[16:22] <rabeeh> NCommander: I missed your nick name on the wiki; just saw your name at the bottom of the page
[16:23] <rabeeh> NCommander: do you use minicom?
[16:23] <NCommander> rabeeh, I tried it
[16:23]  * NCommander prefers screen
[16:23] <ojn> Martyn: not everyone are loose lipped what concerns confidential information. :)
[16:23] <rabeeh> NCommander: change that to 0x10, instead of 0xe
[16:23] <Martyn> ojn : Depends on what's confidental ...
[16:23] <NCommander> rabeeh, I'll check it when I next powercycle the board
[16:24] <NCommander> rabeeh, I found a pretty decent work around which let me make our modifications to u-boot
[16:24] <rabeeh> NCommander: ok. when  I used minicom you would see sent 'spaces' from the board. those are xmodem send me some juice commands
[16:24] <NCommander> rabeeh, actually, you just got an email from me on that subject dated ~2-3 hours ago
[16:24] <Martyn> ojn : ARM fully published and released the roadmaps, documentation, etc for just about everything on the A9, PBX a9, and partners that are producing chips now .. so I figured Marvell (or it's reps) might now be more open as well.   In fact, there is plenty of information now on the ubuntu wiki regarding the Y0, etc...
[16:25] <Martyn> ojn : Point being that you never know, until you ask :)
[16:25] <NCommander> rabeeh, right, I tried to point sx at it. I'll look at that though, since I can remove some of the dire warnings
[16:25] <rabeeh> I haven't got any email
[16:25] <NCommander> rabeeh, your rabeeh@marvell.com, right?
[16:25] <rabeeh> yes
[16:25] <rabeeh> that's my (spam-me) email
[16:25] <NCommander> rabeeh, interesting, you should have been Cc'ed on it
[16:26] <NCommander> check your spam folder from anything from michael.casadevall@canonical.com
[16:26] <ojn> Martyn: Right, hurting doesn't ask I suppose. :)
[16:26] <Martyn> ojn : Polish notation revering you are
[16:27] <rabeeh> NCommander: my junk mail is full on ads on viagras, my friends at Africa that has 100 million $ and doesn't know what to do with; but nothing from Michael
[16:27] <ojn> Martyn: whatever. :)
[16:28] <NCommander> rabeeh, woo, disappearing email
[16:28] <rabeeh> invisible ink email :)
[16:28] <NCommander> rabeeh, I'll take your invisible ink email, and raise you a 500 mile email
[16:28] <NCommander> http://www.ibiblio.org/harris/500milemail.html
[16:29] <Martyn> slowly compiling ubuntu's packages on a PBX-a9 is s-l-o----oooooo---w
[16:29] <NCommander> Martyn, pedal faster ;-)
[16:30] <Martyn> NCommander: I need faster A9 hardware, damnit.
[16:30] <NCommander> Martyn, bah, no you don't
[16:30] <NCommander> Martyn, I built *KDE* on an NSLU2
[16:30] <ojn> Just do it on qemu. :)
[16:30] <Martyn> ojn : I get build fail issues under qemu
[16:30] <NCommander> rabeeh, see PM
[16:31] <NCommander> Martyn, system or user emulation?
[16:31] <Martyn> system
[16:31] <NCommander> Martyn, odd
[16:31] <NCommander> Martyn, its stable for ARMv5 for me
[16:31] <NCommander> Never tried to abuse ARMv7 on it though
[16:31] <Martyn> NCommander: Which is kind of the point.
[16:32] <NCommander> Martyn, it could be worse
[16:32] <Martyn> NCommander: Real hardware is better anyway.  Build -might- finish by friday
[16:32] <NCommander> Martyn, WTF are you building?
[16:32] <Martyn> NCommander: Karmic
[16:32] <NCommander> Martyn, o_o;
[16:33] <Martyn> a.k.a "everything needed for Karmic server"
[16:33] <NCommander> Martyn, karmic .... server?
[16:33] <Martyn> and there are a lot of packages that are completely fubaring because I don't fully grok the launchpad system
[16:33] <NCommander> Martyn, oh god, don't use Launchpad for archive rebuilds
[16:33] <NCommander> PLEASE
[16:34] <NCommander> That's like using a steamroller to open a bag of peanuts
[16:34] <Martyn> NCommander: You have a better way to rebuild the entire package archive?  If so, I'm all ears
[16:34] <Martyn> because from what I gleaned from Duncan and company .. I needed the whole launchpad rigamarole
[16:35] <NCommander> Martyn, rebuildd is one choice for one shot rebuilds. wanna-build/buildd/sbuild from Debian probably work better but require manual intervention on uploads to get them published (unless you tweak buildd)
[16:35] <Martyn> Well, this isn't just going to be one-shot, hopefully.
[16:35] <NCommander> lool, ^
[16:36] <NCommander> lool was working on a spec to do exactly this actually.
[16:36] <Martyn> Yes, don't know where that got to though
[16:36] <NCommander> Martyn, do you actually expect to see significant performance increases from ARMv6->armv7?
[16:36]  * NCommander didn't think ARMv7 added too much to the base instruction set that would improve general application time.
[16:37] <rabeeh> NCommander: nice :) i'll trade
[16:37] <Martyn> I'm not doing it for performance.
[16:37] <Martyn> I'm doing it for power
[16:37] <Martyn> and to compile with full NEON support for those applications that can use it
[16:37] <NCommander> rabeeh, I think your spamfilter ate that email
[16:37] <NCommander> rabeeh, (did you see the private message i sent you?)
[16:37] <NCommander> as in server side spam filter
[16:41] <rabeeh> Martyn: is this autovectorization?
[16:42] <Martyn> rabeeh : Can't comment
[16:45] <NCommander> Martyn, w.r.t. to public dove documentation, its mostly just a by-product of the Ubuntu development cycle
[16:45] <NCommander> (i.e., the installation wiki page and such)
[16:45] <Martyn> *nod*
[16:46] <Martyn> frankly, I wish a lot of the ARM crap would stop being so cloak-and-dagger
[16:46] <NCommander> Martyn, the entire process is driven to be open and visible, so the documentation follows the same route.
[16:46] <Martyn> ARM architecture that is
[16:46] <armin76> rabeeh: i'm happy with ssh access to dove, gimme :)
[16:46] <Martyn> so many manufacturers wanting to keep things under wraps, that it slows down innovation and cross-pollination at a time when people's interest in ARM architecture is really starting to explode
[16:46] <armin76> Martyn: tell me that to me :)
[16:47] <NCommander> Martyn, I just want OpenFirmware for ARM or equivelent
[16:47] <Martyn> armin76: Syntax error?
[16:47] <Martyn> NCommander: UEFI
[16:47] <Martyn> that's the thing that will eventually be that
[16:48] <armin76> Martyn: the thing about manufacturers keeping it secret
[16:48] <armin76> that affects me, because i could do stuff if i had access to the hardware as well
[16:49] <rabeeh> armin76: you have 78100 and sheevaplug
[16:49] <rabeeh> marvell is making those announced products public !!!
[16:49] <armin76> rabeeh: but thats not v7
[16:50]  * NCommander got to play with a Sheevaplug
[16:51] <Martyn> armin76 : BTW, I reinstalled the babbage, and am waiting for permission from the CEO to put it in the DMZ so you can ssh into it
[16:52] <ojn> armin76: there are other armv7 platforms out there. Or do you need to run on a specific implementation to performance tune something? Either way, "do stuff" is barely much of a selling point for hardware access by vendors. :)
[16:53] <Martyn> ojn : perhaps not, but locking away the hardware from view certainly doesn't help get a robust and vibrant community of developers going either :)
[16:53]  * NCommander would love to see a vendor successfully see a product that didn't do stuff
[16:53] <armin76> ojn: i'm eating an icecream, i'll write better answers now
[16:53] <armin76> after i'm done with it, that is
[16:53] <Martyn> Best thing ATmel ever did for the mega series of AVR processors, was come out with the inexpensive butterfly dev kit
[16:53] <NCommander> armin76, what flavor ;-)
[16:54] <NCommander> Martyn, ooh, I have an old AVR8 board in the closet
[16:54] <ojn> Martyn: Sure. ARM has some of it going already with the beagle and other cheap platforms
[16:54] <Martyn> and the best thing TI/ARM ever did, was back the creation of the OMAP3 based Beagleboard
[16:54] <NCommander> That thing taught me ASM could be fun
[16:54]  * NCommander was runover by x86 asm as a small child ...
[16:54] <armin76> NCommander: almond :)
[16:54] <ojn> armin76: heh
[16:54] <armin76> ok, i'm done
[16:54] <armin76> Martyn: thanks :) how much space can i have on it?
[16:54]  * NCommander goes to hit the installer with a brick
[16:55] <Martyn> I'm working on an A9 based beagleboard-like implementation, based on the SoC my company is working on
[16:55] <Martyn> that way there will be the expensive server platform, but also an uber-cheap little board for people to get addicted to
[16:55] <armin76> ojn: yes, there are other platforms but unlike you canonical guys(are you one of them as well?) i don't have the contacts
[16:55] <NCommander> Martyn, sounds like the NSLU2
[16:55] <ogra> armin76, the babbage is really not as exciting as it seems ... the dove with its cool SATA controller is way beyond
[16:55] <ojn> armin76: no, i dont' work for canonical
[16:55] <Martyn> armin76: I'll put in one of the 16Gb SDHC cards, should be enough to keep you hapyp
[16:56] <armin76> ojn: also i don't get paid for doing development on it
[16:56] <NCommander> Martyn, 16GiB? Bah, I can only find 2-4 GiB around here
[16:56] <Martyn> ogra : There's a little sata controller on the babbage too :)  It's just not -as- exciting
[16:56] <armin76> Martyn: it should! thanks :)
[16:56] <ogra> Martyn, its a fake :P
[16:56] <NCommander> Martyn, its USB based
[16:56] <ogra> Martyn, the SATA on the babbage is a SATA to USB bridge
[16:56] <Martyn> Yeah, I know it's usb based ... but it works anyway
[16:56] <ogra> it only operates at USB speed
[16:56] <armin76> ojn: the only manufacturer contact i have is rabeeh, thats why i'm bothering him :)
[16:56] <Martyn> ogra : True .. that's a bummer
[16:57] <ojn> armin76: heh
[16:57] <ogra> i get 60MB/s out of my dove with a 7200rpm disk
[16:57] <Martyn> ogra : Stop making me jealous
[16:57] <armin76> ogra: but its still armv7, i'm happy with being able to compile and having nfs
[16:57] <rabeeh> ogra: is it an old drive?
[16:57] <ogra> only 16-20 out of the babbage
[16:57] <ogra> rabeeh, well, what i had lying around
[16:57]  * NCommander should plug in his UltraSCSI board into his dove then connect a 10,000 RPM drive
[16:57] <Martyn> ogra : Yeah yeah, but are YOU going to put a dove outside a firewall for armin to play with?
[16:57] <NCommander> :-)
[16:57] <ogra> Martyn, no :)
[16:57] <armin76> ogra: maybe it sucks for you, but if its better than the beagle then its enough for me :)
[16:58]  * NCommander would if armin76 is willing to access it over IPv6
[16:58] <NCommander> bahahahah
[16:58] <Martyn> Well then, the whole idea of which sata implementation is moot then :)
[16:58] <ogra> its better than the beagle B1 for sure
[16:58] <armin76> NCommander: i do have ipv6
[16:58] <ogra> cant compare to the C rev. beagles
[16:58] <NCommander> armin76, >.<;
[16:58] <armin76> NCommander: gimme :)
[16:58] <NCommander> armin76, sadly, my dove's ipv6 address keeps changing
[16:58] <Martyn> ogra : C rev is the same, but the USB host works, and there is twice the memory
[16:58] <NCommander> armin76, https://bugs.edge.launchpad.net/ubuntu/+source/linux-mvl-dove/+bug/435046 - due to this
[16:58] <ubot4> Launchpad bug 435046 in linux-mvl-dove "Ethernet port on the dove sometimes changes MAC addresses" [Undecided,New]
[16:58]  * Martyn again, has to patiently wait for smooth-stone to get a dove to touch one
[16:59] <NCommander> Martyn, BTW, do you know how many eth devices the kernel can have?
[16:59] <ogra> right, twice the ram will make some difference
[16:59] <ogra> 65535 ?
[16:59] <NCommander> Martyn, I'm trying to figure out if the dove will panic if it hits eth255
[16:59] <armin76> ojn: also believe me that i've tried to contact freescale and TI about boards, to no avail, so i only have rabeeh :)
[16:59] <rabeeh> NCommander: just set the 'ethaddr' in u-boot and it won't random your mac address
[16:59] <ojn> armin76: see /msg
[16:59] <NCommander> rabeeh, I did
[16:59] <NCommander> rabeeh, it doesn't work.
[16:59] <rabeeh> NCommander: saveenv?
[16:59] <rabeeh> :)
[16:59] <NCommander> rabeeh, yup
[17:00] <NCommander> rabeeh, see the bug :-/
[17:00] <NCommander> rabeeh, its like my MAC is connected to a random number generator
[17:00]  * NCommander ran out of DHCP leases last night ...
[17:00] <ogra> get more dhcp servers
[17:00] <NCommander> ogra, I just moved to IPv6
[17:00] <ogra> you never see the obvious fix
[17:00] <NCommander> ;-)
[17:01] <rabeeh> NCommander: touch /ubuntu/etc/udev/rules.d/75-persistent-net-generator.rules
[17:01] <lool> ojn: Hey what are you working on?  (in terms of ARM stuff)
[17:01]  * NCommander blinks
[17:01] <lool> Martyn: I'm afraid I didn't manage to reserve time to work on my spec so I didn't complete anything remotely useful
[17:01] <rabeeh> that will keep ethX to eth0; but regardless the MAC address should change after you do saveenv
[17:01] <ojn> lool: Working at a startup that is currently considering using an arm soc in a product. just doing random stuff
[17:01] <lool> I only read API docs and played with some scripts
[17:01] <NCommander> rabeeh, it keeps changing regardless
[17:02] <ogra> lool, dont say that, rootstock is happily using your kernel hack ;)
[17:02] <lool> ojn: netbook class, server class, beagleboard like?
[17:02] <rabeeh> NCommander: can you use the Marvell u-boot
[17:02] <lool> ogra: True, I did put a shitload of time in qemu in the beginning of the cycle
[17:02] <NCommander> rabeeh, I can, but I haven't made any code changes that would affect this
[17:02] <ojn> lool: Sorry, not public information. :)
[17:03] <NCommander> armin76, 2001:470:1d:8d:250:43ff:fe68:1438
[17:03] <rabeeh> NCommander: after two boots - http://dpaste.com/97255/
[17:03] <lool> ojn: armv7?
[17:04] <NCommander> rabeeh, thats a manually set one, I was just using the default in u-boot
[17:04] <lool> ojn: it's ok if it's private too  :-)
[17:04] <ojn> lool: well, most interesting ARM SoC's out there today (and/or soon) are ARMv7.
[17:04] <lool> Yeah you know I was hoping to bring you on the road to revealing something about your work   :-)
[17:05] <lool> ojn: You coming to UDS?
[17:05] <ogra> and if not, whats your excuse :)
[17:05] <armin76> NCommander: which password? :)
[17:05] <ojn> lool: hah.   where, when?
[17:05] <armin76> and user
[17:05] <NCommander> armin76, wow, you weren't joking when you said you have IPv6
[17:05] <lool> ojn: mid Nov somewhere in the US
[17:05] <lool> Final place not decided yet   :-(
[17:05] <rabeeh> NCommander: that's what i'm saying; you need to set it manually to different MAC
[17:05] <armin76> NCommander: owned? :)
[17:06] <ojn> lool: depends on if I can find business justification, I suppose.
[17:06] <NCommander> armin76, well, I learned my firewalls rules are a bit more permissive than I realized ;-)
[17:06] <armin76> NCommander: had to set it up to access the mv78100 rabeeh talked about :)
[17:06] <lool> ojn: It's where we plan for next 6 months of Ubuntu dev
[17:06] <ojn> lool: right, I know what the event is. :)
[17:06] <ogra> ojn, you can take influence on behalf of your company ...
[17:06] <NCommander> armin76, there's something sexy about not popping a hole in the NAT for you
[17:06] <lool> ojn: I was suggesting that might be good justification   :-)
[17:07] <ogra> ojn, that should be enough justification :)
[17:07] <ojn> heh
[17:07] <ojn> well, locate it in austin and it's an easy decision.
[17:07] <lool> ojn: But then you cant share much so I cant really give better advice; I cant say we could have a session on netbooks or a session on servers for instance   :-)
[17:07] <lool> ojn: It might be in dallas
[17:07] <lool> at which point you can basically walk there
[17:07] <ogra> ojn, one of the places on the plate is very near
[17:07] <ojn> lool: that'd be ok
[17:12] <NCommander> lool, can I pick your mind on a technical question on a how best to implement?
[17:12] <NCommander> lool, I rewrote flash-kernel so it gets the UUID dynamically by reading /, but I'm not sure how to make this work from flash-kernel-installer, since it will end up calling both f-k-i, and flash-kernel
[17:12]  * NCommander is kinda thinking of forging a mtab entry in the chroot ...
[17:13] <NCommander> (https://code.edge.launchpad.net/~mcasadevall/flash-kernel/standalone_boot.scr)
[17:21] <armin76> Martyn: is the cortex-a9 armv7a or not?
[17:21] <NCommander> bbiab
[17:22] <Martyn> armin76 : Yes
[17:23] <Martyn> lool : Can't wait to have the Ubuntu 6mo here in Austin :)
[17:23] <Martyn> lool : I finally get to meet all of you in person :)
[17:24] <Martyn> lool : 3 hour drive to dallas, still would do it :)
[17:41] <lool> rabeeh: Can you explain how you got the glib warning exactly?
[17:41] <lool> rabeeh: Cause that should work in karmic, I see a > 2.10 dep on libc on amd64
[17:41] <rabeeh> apt-get update
[17:41] <rabeeh> apt-get upgrade
[17:42] <lool> NCommander: Cant f-k-i just call flash-kernel??
[17:42] <lool> NCommander: there are examples of UUID computations in flash-kernel already I think
[17:42] <lool> rabeeh: That's a pure karmic system?
[17:42] <rabeeh> yes
[17:42] <lool> libglib2.0-0 Depends: libc6 (>> 2.10), libc6 (<< 2.11), libgcc1 (>= 1:4.4.0), libpcre3 (>= 7.7), libselinux1 (>= 2.0.85)
[17:43] <rabeeh> root@ubuntu:/tmp# dpkg-query -l | grep libc6
[17:43] <rabeeh> ii  libc6                             2.10.1-0ubuntu7                       GNU C Library: Shared libraries
[17:43] <rabeeh> ii  libc6-dev                         2.10.1-0ubuntu7                       GNU C Library: Development Libraries and Hea
[17:43] <rabeeh> ii  libc6-vfp                         2.10.1-0ubuntu7                       GNU C Library: Shared libraries [VFP version
[17:45] <lool> rabeeh: odd, that version should be fine
[17:46] <rabeeh> hmm... shouldn't that build depends on ... ?
[17:51] <rabeeh> NCommander: where can i download dove live images? and u-boot?
[17:55] <Martyn> ARGH
[17:55] <Martyn> build crashed
[17:55] <Martyn> rabeeh : www.ubuntu.com/testing
[17:59] <rabeeh> 17-sep is latest ?
[17:59] <rabeeh> (alpha-6)
[18:11] <Martyn> yes, but there are some packages that need upgrading after install
[18:11] <Martyn> there were a lot of changes applied in the last few days
[18:25] <lool> rabeeh: it does too