loolDanaG: lol00:01
loolDanaG: Yeah, it will definitely we without NAND00:01
rcn-eeDanaG, they ubuntu is currently only targeting C4... ;)00:02
DanaGNow, if u-boot had ext4 support, that would be pure win.00:02
rcn-eeif u-boot had any extX support.. ;) (that worked)00:02
loolu-boot needs to be read from somewhere too00:02
loolrcn-ee: ISTR we're using ext2 support in the dove images, not quite sure anymore00:03
loolproblem is that we still need to load u-boot from MMC I guess00:03
loolso with vfat00:03
loolbecause the ROM will load it00:03
rcn-eei know what you mean, i a couple other boards (avr32, etc) that has no problems with extX partitions running mainline u-boot.. Only the beagle has issues...00:03
ogra_cmpclool, http://paste.ubuntu.com/414595/ ... i need uboot-envtools in main and integrate it into flash-kernel-installer (for cmdline/root UUID etc), but it apparently works well already00:04
Martynrcn-ee : Right now the bulk of the work I'm doing is getting the smooth-stone SoC supported, and trying to figure out ways of recycling as much of the ARM tree in an abstracted way to do it00:05
ogra_cmpclool, thanks so much for poking me towards NAND00:05
DanaGoh... right, you do still need fat32.00:05
DanaGIt's a chicken-and-egg problem.00:05
rcn-eeOf course Martyn, cant' wait to see what that board can actually do. ;)00:06
DanaGCan't read u-boot itself from ext4, anyway.00:06
Martynrcn-ee : One of the big problems, at least for me, is that Grant Likely device tree patches aren't going to make it in until mid 2.6.35, even .3600:06
Martynrcn-ee : Coming to Brussels?00:06
DanaGI wonder if you could fit u-boot in an MBR.00:06
MartynDanaG : No.  i've tried00:06
MartynDanaG ; The problem is keeping it under 63 sectors (512bytes*63)00:06
rcn-eeNope, not across the pond anytime soon, but i will be in esc san jose..00:07
loologra_cmpc: eh00:07
ogra_cmpcrcn-ee, thats a shame !00:07
rcn-eeDanaG, actually someone did intergrate X-loader into u-boot as a side project... can't remember how many months back on the beagleboard group...00:08
ogra_cmpclool, i got a freeze exception until fri. btw00:08
Martynrcn-ee : Pity .. I'm bringing a prototype of sorts00:08
ogra_cmpcfor that bit at least00:08
Martynrcn-ee : nVidia is helping :)00:08
rcn-eewhat? no way! ;)00:09
rcn-eei know they are kinda betting on the tegra platform, so it makes some sense...00:09
DanaGnow, a Tegra netbook could be cool.00:09
Martynrcn-ee : Heh .. true.  Although I feel kind of guilty being one of the reasons the tegra2 200 platform is hard to get right now00:09
DanaGI may not much like nvidia... but at least I'd probably be able to use compiz on the thing.00:10
DanaG... as opposed to PowerVR.00:10
MartynDanaG : I already saw four prototype 'smartbooks' based on the tegra2 .. but the 200 has some problems with Mini PCIe00:10
Martyn(weak ordering is supported, strong ordering is not, which breaks the PCI driver in linux_00:10
MartynDanaG : Try the more exciting idea -- the potential for using CUDA00:11
rcn-eeYour just helping to get things supported before everyone starts yelling about the issues...00:11
DanaGhmm, weak ordering?00:11
MartynDanaG : Yeah.00:11
rcn-eei just want a PCI-E, then stick ati/nouveu based card into it.. ;)  the open-rd was interesting, but backordered..00:11
* ogra_cmpc falls dead00:12
Martynogra :Sleep time?00:13
MartynDanaG : Bingo00:13
DanaGSucks that the Base and Client have different sets of I/O.00:14
Martynand because the assumption is strong ordering (where the tegra2 demands weak ordering) ... we get brokenness all over the place00:14
DanaGI want a superset of both... "client" without the onboard IGP, with PCIe slot.00:14
DanaGhmm, still doesn't quite say what strong / weak ordering is.  Is it presence or lack of specific instructions?00:14
rcn-eeOh ogra, found another device with no battery backed rtc...  The Always-innovating touchbook doesn't have a backup battery...  There's a mod on their wiki to add one...00:17
Martynthink of it as the difference between TCP and UDP00:17
MartynTCP ensures that the order of packets is perserved and checked00:18
MartynUDP .. ordering is -not- guaranteed or checked00:18
DanaGah, but instead of packets, it's "things you tried to fetch from memory"?  (would have to include a sequence marker, of course.)00:19
Martynsomething like that00:20
Martynthere are entire chapters in my PCI book dedicated to the subject00:20
DanaGhmm, that reminds me of something I had thought would be interesting: use a PCIe FPGA to do all sorts of stuff.00:21
DanaGSomething I wanna' see: http://www.sondigo.com/sirocco/overview  -- something like that, but with 2.6 kernel.00:21
DanaGIt's a c-media PCI audio chip on a Realtek Lexra MIPS CPU.00:22
DanaGMost ARM SOCs don't have surround sound.00:22
Martyn.. thinks00:24
MartynNO ARM SoC's have surround sound yet, that I can think of00:24
MartynI don't even think the Tegra2 is more than 2.100:24
DanaGyeah, I realized that after I spoke / typed / pressed enter.00:25
philoi need a simple way to setup a virtual machine with ubuntu-arm running it... can anyone point me to some doc ?06:04
=== JaMa|Zzzz is now known as JaMa
ographilo, see topic :)07:15
aaron_liujmy board have two output hdmi and lcd ,but it's default out lcd ,how to change it to hdmi display07:31
aaron_liujneed i modify the frame buffer ?07:32
amitkaaron_liuj: what board is it?07:33
aaron_liujone type arm board07:37
aaron_liuj2 output device , but default is lcd ,i want change it to hdmi tv07:38
aaron_liujhow to do that ?07:38
=== ogra_ is now known as ogra
amitkaaron_liuj: I'm afraid we need more info than that. What processor does it use? Is it a custom board, etc..07:47
aaron_liujbut the i have he right worked hdmi driver07:52
aaron_liujbut the i have the right worked hdmi driver07:52
aaron_liujbut i don't know how to change the default output device07:52
amitkaaron_liuj: that is not an SoC that is supported by Ubuntu, so I'm not familiar with its drivers07:53
aaron_liuji think it's vary samilar with pc have vga and hdmi port07:55
aaron_liuji think it's very samilar with pc have vga and hdmi port07:55
aaron_liuji think it's very samilar with pc have vga and dvi, hdmi port07:55
amitkit is probably in the framebuffer driver options somewhere07:56
=== DanaG1 is now known as DanaG
DanaGhmm, my Cowon S9 has some telechips thingy of some sort.08:21
amitknever heard of them before08:24
aaron_liujyeah from telechips08:25
NCommanderaaron_liuj: ususally its a kernel command line option, or if you just care about X, you might be able to get by with just xrandr, but I can't tell you the specifics08:27
NCommanderMartyn: dove has surrond sound08:28
ograhrm, so broken_system_clock in e2fsck.conf doesnt help :(08:28
saeedhi there08:33
saeedogra: hi08:40
ograhey saeed08:40
saeedI would like to boot lucid on dove with console on serial08:41
saeedit would be very usefull if we can append args to bootargs of the boot.scr08:41
ograsaeed, that will breake the splash screen on all desktop installs, do you  want that ?08:44
ograplymouth sadly breaks once a console= option is on the cmdline and wont display any splash anymore08:45
saeedogra: what is splash sceen?08:46
saeedok, I googled it08:47
saeedanyway, the console option will be added only for debug08:47
ograheh, its the progressbar screen you see on boot of ubuntu desktop systems08:47
saeedalso, we may add other options08:48
ograit's a bit late to change these options, we need to talk to the release team given that final freeze was yesterday08:49
ograhow confident are you that they are 100% not causing any problems (the additional options i mean)08:49
saeedI see, please consider it for next release08:49
ograwe'll do, we can get a freeze exception but it must be a 100% bulletproof change08:50
ograif the images get unbootable due to some cmdline option you request thats a disaster08:50
saeedogra: I'm not looking to add permanent options to the cmdline08:51
saeedonly temporary options for debugging specific issue08:51
ograyou can easily change boot.scr yourself08:52
ograif its only for you for debugging08:52
ograjust grab it, remove the binary header, save it as boot.script and edit bootargs to your liking08:53
ograthen run: mkimage -A arm -T script -C none -n "Ubuntu boot script" -d <path to your boot.script file> /boot/boot.scr08:54
ograits trivial08:54
=== nosse1_away is now known as nosse1
nosse1Morning everyone09:04
aaron_liujgood night all09:09
loologra: I think saeed was just asking for a local workaround09:15
ogralool, yeah, i didnt get that09:16
loolsaeed: What we kind of miss is some UI to offer multiple kernels and versions with options09:16
loolsaeed: Could probably be done with the soft bootloader PoC, or some other bootloader like petitboot and such09:16
loolsaeed: But locally you could regenerate boot.scr or break into u-boot I guess?09:17
saeedlool: I used to create my own bootcmd based on the boot.scr one, I also tried to modify the boot.scr as ogea suggested. both methods work09:18
saeedhowever, if the next release is comming with softbootloaded, then that will be great09:19
saeedbtw, what is the status of the softloader?09:20
saeedNcommander worked on that, right?09:20
amitklool: atleast openmoko uses a u-boot that supports menu entries IIRC09:21
loolsaeed: No it's not coming with softbootloader09:21
loolsaeed: I think ogra and NCommander had a working proof of concept, but I'm not sure what prevented its wider use; for a long time kexec was badly broken09:21
loolsaeed: Perhaps the kexec breakage prevented its use?  Not sure09:22
loolsaeed: project name is now https://launchpad.net/mukluk09:22
saeedlook: issues with kexec were fixed09:22
loolSo I cant comment in place of NCommander and ogra09:23
ogralool, kexec broke suspend/resume on most arches we added it ...09:23
ograi think it was withdrawn from most of our kernels again until thats better understood by the kernel guys09:24
ograamitk, since you are affected by bug 563618, would you mind to confirm it ?09:25
ubot4`Launchpad bug 563618 in util-linux (Ubuntu Lucid) (and 3 other projects) "there is no way to tell fsck to ignore broken clocks on embedded systems (affects: 1) (heat: 8)" [High,New] https://launchpad.net/bugs/56361809:25
ograelse Keybuk will close it immediately again and tell me its fixed :)09:26
amitkogra: done09:27
ograthanks :)09:27
* ogra waits for dpms to kick in so he can reproduce the DSS2 bug finally09:31
nosse1ogra, does the new ti-omap -5 cover the changes you made to the kernel yesterday?09:35
nosse1comparing to the -4 config, there doesn't seem to be anything changed related to console09:36
amitknosse1: that isn't built yet... long backlog of arm packages just before freeze09:37
nosse1amtik, thanks. I can extract the config from those sources09:38
amitknosse1: use tab for my nick :)09:38
nosse1amitk, sorry. Didn't mean to offend. I've tried using the tab, but maybe this isn't a function in chatzilla09:40
amitknosse1: none taken, just trying to point you to a (timesaving) feature. :)09:40
nosse1trust me, I have been looking... ;)09:41
saeedogra: I'm trying to debug the hibernate issue https://bugs.launchpad.net/bugs/50900609:44
ubot4`Launchpad bug 509006 in linux-mvl-dove (Ubuntu Lucid) (and 1 other project) "[dove] hibernation failed to resume (affects: 2) (heat: 12)" [High,Confirmed]09:44
saeedI suspect that the issue is that the initramdisk mounts the root partition before doing the resume09:45
loolsaeed: You can break into the initrd by passing extra args09:48
loolsaeed: break=top break=bottom etc.09:48
loolsaeed: if you unpack an Ubuntu initrd and grep for maybe_break, you should see all the possible break points09:48
saeedlool: thank, I'll try it09:49
ograor look under /usr/share/initramfs-tools09:49
ogra(easier than explicitly unpacking)09:49
ogralool, http://paste.ubuntu.com/414821/09:51
ogralool, i have one prob left ... if you look below "Erasing Initramfs NAND space" you will notice that i use $kmtdsize (4MB) to erase the initramfs space09:52
ogralool, the default size of that space is just to big to erase it all ... without erasing we cant write (generates IO errors) what would you recommend to use for a erase size ? i was pondering to hardcode a value somewhere09:53
loologra: You mean it's too slow to erase it all?09:54
loologra: how large is it?09:54
ograyeah, its 2Gig09:54
loolFor Kernel?09:54
ograits the filesystem space for the rootfs originally09:54
ograno, kernel is 4MB09:54
loolWho creates the Kernel partition?09:54
ograthe kernel09:54
ografor a later release we need to sort that stuff in kernel09:55
ograbut for 10.04 i doubt we can09:55
loolOk, there's a space for a rootfs and you use it for the initramfs09:55
ograand its to big to erase it all09:55
loologra: Can't you change the partitions to be sensible?09:55
ogratakes 10-15min09:55
ograno, they are set in kernel09:55
loologra: pointer?09:56
ograin the omap mtd driver actually09:56
ogralool, http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=blob;f=arch/arm/mach-omap2/board-omap3beagle.c;h=962d377970e9bb198d410abf1ac1b418d0658341;hb=HEAD09:56
nosse1amitk: Are there any other changes to the kernel except for config changes?09:57
loologra: Could you possibly use the kernel part for both, or is it two small?09:57
ografor both ?09:57
ograwhats both ?09:57
ogra4MB is definately to small09:57
loologra: So I have some ideas, but first could you explain where these IO errors come fronm?09:58
loolI thought you could write whatever you like to mtdblock parts, the kernel will do the NAND flashing to 0xff for you if needed09:59
ogralool, no idea, i get them if i write to unerased space09:59
amitknosse1: no other changes, they'll happen after release as SRUs09:59
loologra: I suspect it's a bug in the offset / size you're writing09:59
ograas soon as i zero it everything works flawless09:59
ograi got the offset sizes from the beagle docs09:59
ograthey seem to be identical across all beagles10:00
loologra: I would say you should have at least 16 MiB for the initrd data10:01
loolfor safety, you could go to something higher10:01
loollike 6410:01
ograwhere would you put it ?10:01
loolin the rootfs part10:01
ograin the script (i dont use flash-kernel.conf yet and would happily not)10:01
loologra: how do you load the initrd ATM?10:02
ogranand load10:02
loolcertainly *some* components needs to be made aware of the actual size we care about10:02
loologra: do you have control of that script?10:02
nosse1amitk: How do you generate the .config? Cat everyting together from debian.ti-omap/config ?10:02
ograsee the top part of my paste10:02
ograline 10-2610:02
ograit uses uboot-envtools to set the u-boot env10:03
loologra: yeah, so you're essentially already listing the initnrd size10:05
ograi have to change it in both places10:05
ograor would you suggest to parse bootcmd in flash-kernel ?10:05
loologra: I guess that's ok; if would be nice if you followed the usual u-boot variables10:05
loolbeagleboard comes with a nice set of u-boot envs vars10:05
ogramine doesnt anymore :P10:06
ograafter i played with the flash :)10:06
loologra: Well if you were to use proper vars, it would be nice to parse in flash-kernel10:06
loole.g. (I'm not sure what exists already) initrd_nand_addr, initrd_nand_size etc.10:06
loologra: you can reset the values to their defaults10:06
loolresetenv or something10:06
ograah, right10:07
loologra: I suspect they don't use the rootfs as an initrd though10:07
loolsorry initramfs10:08
loologra: Perhaps it would be more elegant to just write what you need to write and use it as an initrd instead of initramfs?10:08
ograthere are no vars at all for initramfs usage10:08
ogranor for initrd10:08
loologra: Hmm really10:08
ograinitrd isnt a concept on beagle10:09
looljffs2args=setenv bootargs console=${console} ${optargs} root=/dev/mtdblock4 rw rootfstype=jffs210:09
ograeven finding proper docs with google is painful10:09
ograright, thats for a rootfs10:09
loolOh it's certainly supported for sd10:09
nosse1I see you changed the HZ. Why is that?10:09
=== XorA|gone is now known as XorA
ogranosse1, ??10:10
loologra: So I'd introduce a nandinitrdboot or similar which would set root=UUID=something and load some configurable megabytes of the rootfs as an initrd; I'd read that value from flash-kernel to zero it10:10
loolYou can re-use rdaddr10:11
ograamitk, bug 563650 with kisses for you :)10:12
ubot4`Launchpad bug 563650 in linux-ti-omap (Ubuntu Lucid) (and 1 other project) "DSS2 oops when shutting down while DPMS is active (affects: 1)" [Medium,New] https://launchpad.net/bugs/56365010:12
loolI wonder whether it would be possible to point the kernel at the nand for the initramfs and unpack it in RAM10:12
ogralool, what scares me a bit is to rely to much on uboot-envtools10:12
ogralool, thats for 10.10 :)10:13
* ogra will already be happy if he ends up with a bootable *something* on lucid beagle ... given bug 563618 and Keybuks attitude towards the prob thats not looking good atm10:15
ubot4`Launchpad bug 563618 in util-linux (Ubuntu Lucid) (and 3 other projects) "there is no way to tell fsck to ignore broken clocks on embedded systems (affects: 1) (heat: 8)" [High,Confirmed] https://launchpad.net/bugs/56361810:15
amitknosse1: right, cat together all of it10:17
amitkogra: thanks10:17
NCommandersaeed: softbootloader was blocked on lack of kexec()10:32
NCommandersaeed: kernel support landed too late for lucid unfortunately10:32
loologra: i'm glad I told you 64 MiB because the usual ubuntu config is actually 4 MiB10:33
ograi would have gone with 16 :)10:33
ograi dont think i have ever seen a bigger initramfs in my life10:33
loologra: given that the size in the kernel config is 64 MiB, that would seem a prudent value to use10:34
loologra: are you checking the file size before writing it?10:34
ogranot explicitly, no but thats just a copy paste job10:35
loologra: Make sure there's some check in place?10:35
loolquoting is a bit inconsistent in your diff too10:35
ogracheck_size "Kernel" $(($kfilesize + 8 + 64)) $kmtdsize10:36
ogralike that ?10:36
loolapw: Would you be open to bumping the default ramdisk size in the dove config to 65536 instad of 4096?  :_)10:36
ogra(stolen from QNAP)10:36
loologra: something like that, I'm afraid I lack the specifics10:36
loologra: note that you're actually writing an u-boot image which is slightly bigger than the initrd10:37
apwlool, i think that is becoming the default on the other arm branches10:37
ogralool, indeed10:37
apwi presume you mean for SRU10:37
loolapw: Yes, it's the case in pretty much all other branches, not sure why not on dove10:37
loolapw: I guess it's for $next-upload, SRU or late upload; happy either way10:38
loolapw: I would like if it's in git and I can forget about this discrepancy  :-)10:38
apwlool, as dove is in the can is there a bug as we'll need one to SRU the change10:39
loolapw: no open bug, can file one now10:45
apwlool thanks10:45
apwassign it to me10:45
loolapw: https://bugs.launchpad.net/ubuntu/+source/linux-mvl-dove/+bug/56367910:49
ubot4`Launchpad bug 563679 in linux-mvl-dove (Ubuntu) "CONFIG_BLK_DEV_RAM_SIZE should be 65536 as in other kernels (affects: 1)" [Undecided,New]10:49
apwlool ta10:49
saeedNCommander: what is the current status of mukluk10:53
NCommandersaeed: on hold. Will look at implementing for 10.1010:54
saeedNCommander: does the current version partialy work?10:54
ograNCommander, what was the prob again with missing uboot-mkimage on certain dove installs ?11:09
ograNCommander, were these d-i based or live ones ?11:10
NCommanderogra: live images, sometimes blew up due to network not being present11:10
NCommandersaeed: yes, but the script ot make the config file automatically is buggy11:11
ograNCommander, weird, i'm just looking at the redommends11:11
ograin ubiquity11:11
ograand uboot-mkimage is definately in the list so it should be inside the livefs11:11
ograit was added on Fri Mar 2611:13
=== rsalveti_ is now known as rsalveti
nosse1is there a link to the armel build queue somewhere?11:39
ogralool, erasing 64M is still taking more than 2min, i'll go with 1611:44
ograasac, so looking at uboot-envtools ...12:47
asacogra: thanks. and sorry i didnt fix it yet for you ;)12:47
ograasac, using make clean breaks because the quilt patches that clean up Makefile are not applied12:47
ogradonty worry, i'm nearly done with all the omap stuff ... just needs that package12:48
ograasac, so, i cant easily fix up the build system without adding my completely own Makefile debian/patches has Makefile patches that need to be applied to exec anything at all but rules calls unpatch in clean12:51
ograi guess thats the reason for the rm commands in rules12:51
ograasac, laso it doesnt remove source files it seems but it removes the link12:52
ograasac, the missing config is caused because its actually a snipped from uboot, it would be configured during uboot build but the package here tries to be generic and thus ships only example configs12:57
=== ogra_ is now known as ogra
=== ocs_ is now known as faLUCE
=== JaMa is now known as JaMa|Away
ocshi, which is the cheapest arm-board where ubuntu can run ?15:16
rsalvetiocs: beagleboard?15:17
rsalvetiocs: you can get it for $15015:17
ocsrsalveti: are you sure there are not cheaper ones?15:18
rsalvetiocs: if you want to run lucid, then I guess it's the cheaper one15:18
ocsrsalveti: lucid <--- ?15:19
rsalvetiotherwise if you are happy with older releases, you can get an armv5 or armv6 board15:19
rsalvetiocs: ubuntu lucid, the next release15:19
ocsrsalveti: thanks15:21
rsalvetiocs: sheevaplug is cheaper ($99), but it's armv515:23
rsalvetiocs: you can run ubuntu 9.04 or debian15:23
rsalvetior oe based distros, like angstrom15:23
ocsrsalveti: but sheevaplug doesn't have a video connector15:25
rsalvetiocs: yep, you're right15:26
ograrcn-ee, your /etc/e2fsck.conf works with mountall ?15:36
ogra(note that mountall doesnt use e2fsck but fsck from util-linux which doesnt respect the config file)15:36
=== alecrim_ is now known as alecrim
=== jamie is now known as Guest80122
ograGrueMaster, did you test the d-i fix for the netinst images already ?17:08
ogra(for dove)17:08
GrueMasterYes.  Netboot install now works.17:09
ograwith the default initrd without having to fiddle with mkimage etc ?17:09
GrueMasterI even wrote a little netboot.scr script for it that gets pulled via bootp.17:09
GrueMasterI used the latest images from http://ports.ubuntu.com/dists/lucid/main/installer-armel/20081029ubuntu98/images/dove/netboot/dove/ with out any modification.17:10
ograperfect, thanks17:10
GrueMasterI'm still trying to figure out how to use the bootp protocol to send uboot commands.17:11
GrueMasterBut I don't think uboot supports it.17:11
ograprobably not uboot, but kernel/initramfs should be able to read dhcp strings you send17:13
=== Meizirkki_ is now known as Meizirkki
=== Guest80122 is now known as JamieBennett
MartynIs there a list of packages that are -not- attempted to be built on ARM?21:12
armin76libx86 :D21:13
armin76s390-tools :D21:13
asacMartyn: we have a architecture specific blacklist file yes. but dont ask me where ;)21:22
asacogra: ^^ where is pas?21:23
persiaasac: P-a-s isn't the entirety of it, really.  There's also package arch: fields.21:38
asacpersia: sure21:38
asacbut thats not in list form ;)21:38
persiaNo, my point is that such a list doesn't exist.21:39
persiaAnd I know we attempt builds of silly stuff (e.g. libx86, wine1.2).21:39
persiaAnyway, https://code.launchpad.net/~ubuntu-core-dev/packages-arch-specific/ubuntu21:40
Martynthanks :)21:52
Martynpersia : That will do as a starting point21:52
persiaMartyn: patches accepted for that, if you think there's something listed as not-for-ARM that should be for ARM.  I had to unburden a number of packages when working on the lpia port.21:53
Martynright now, I'll take stability over coverage21:55
Martynsmooth-stone is looking over all the packages, and building a blueprint for UDS-M21:55
Martynwe're currently looking for a way to use eucalyptus either with LXC, or completely non-virtualized21:56
persiaThat shouldn't be hard at all.21:58
persiaJust replace eucalyptus-nc with another component.21:58
persiaI strongly recommend chatting with kirkland or smoser in #ubuntu-server about it: they are likely to be able to share good strategies.21:59
persiaAnd I know at least one of them shares your timezone, which may make that easy :)21:59
persiaActually, they're both there now :)22:00
=== jldugger is now known as pwnguin

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!