=== Jack87|Away is now known as Jack87 === Jack87 is now known as Jack87|Away === Jack87|Away is now known as Jack87 [02:56] Is there an easy way to determine which /usr/bin/qemu-X-static corresponds to a "debian --print-architecture" architecture string? e.g. armhf -> arm, amd64 -> x86_64 [02:57] Plan B is just to copy in qemu-*-static and hope for the best. [03:08] twb`: wouldn't it be -cpu as a command line option === Jack87 is now known as Jack87|Away === Jack87|Away is now known as Jack87 [03:25] Well, I'm not actually on armhf at the time, so I don't think I can ask dpkg === Jack87 is now known as Jack87|Away === Jack87|Away is now known as Jack87 [03:32] twb`: Perhaps dpkg-architecture -qDEB_BUILD_GNU_CPU? [03:33] twb`: Although, if you want to know without being able to run the native arm dpkg, that could be more entertaining. [03:39] Let me show you the actual code... [03:40] http://paste.debian.net/129496/ (the QemuUserEmulation bit) [03:40] IOW I want to know before I can execute arm code [03:52] twb`: dpkg-architecture -aarmhf -qDEB_HOST_GNU_CPU [03:52] twb`: And might want to 2>/dev/null to ignore the whining about gcc mismatches. [03:53] twb`: Assuming that qemu-$arch-static matches the GNU CPU types, that should do it for you. [03:55] Ah, thanks === Jack87 is now known as Jack87|Away [05:45] so running ubuntu 11.04, my beagle board doesn't turn on the usb bus. but this works fine with angstrom. thoughs? [05:47] travalas: Try the latest oneiric kernel. I vaguely recall some known issues surrounding USB on beagle. [05:48] where would I look for that kernel? [05:53] travalas: http://ports.ubuntu.com/ubuntu-ports/pool/main/l/linux/linux-image-3.0.0-11-omap_3.0.0-11.17_armel.deb [05:58] thanks [06:15] lilstevie: what happens if you do "make localyesconfig" and "make localmodconfig" from inside a running TF101? [06:16] twb`: hmm [06:16] twb`: I have only done make uImage from inside a running TF101 [06:17] You don't have to actually build the image there, that just generates the .config [06:17] hmm [06:17] well IDK [06:17] I do it on x86_64 and it's handy for getting mostly-right localized/hw-specific .config [06:17] I'd check myself but I left the TF101 at home [06:22] Why is both MACH_PICASSO and MACH_TF101 set? [06:37] in the 2.6.38.3 kernel it is set because the guy who did it wove tf101 stuff into picasso [06:38] not only that is it is common place on these tegra devices [06:38] the asus defconfig has both MACH_HARMONY and MACH_VENTANA set [06:39] Hum. [06:39] Kernel config is pretty overwhelming these days [06:39] I end up doing things like grep -r --include Kconfig [06:39] yeah [06:39] heh === doko_ is now known as doko [07:42] lilstevie: remind me; what doesn't work when booting via u-boot? [07:44] in the boot process or after boot [07:45] Whatever you were saying was needed to boot your CrOS kernel [07:46] no I mean, what is broken of u-boot or once you are in userland :p [07:47] either [07:47] both [07:47] ok, most of the u-boot env is broken [07:47] "What's the downside of using CrOS + u-boot" [07:47] without attaching to the serial port you have 0 interactivity with u-boot apart from boot.scr [07:47] you must use an MBR layout on the emmc [07:48] and the MBR part must start at a set address [07:48] That's not so bad considering I have no interaction with the default bootloader except "use LNX" vs "use SOS" [07:48] yeah well it is less [07:48] :p [07:48] cause you don't even have that [07:48] well, I mean you can run a cascase boot [07:48] so like test dock usb, dock SD emmc [07:49] µSD is broken too [07:49] u-boot does not see the card [07:49] so that is u-boots limitations set out [07:49] as for the kernel, it is mostly fine [07:49] wifi works with network manager [07:49] "connect serial port" as in open the case and connect to to a jumper header, or like the USB to 40-pin cable? [07:50] well I haven't traced out the serial port yet [07:50] I am certain it plumbs out on the 40pin connector somewhere [07:50] just no idea where [07:51] So it would mean getting a 40-pin plug and wiring that up to a 9-ping serial port? [07:51] *9-pin [07:51] well just 3 pins are needed [07:51] tx, rx and common ground [07:51] when I say serial, I mean UART :) [07:51] k [07:52] I could live with that if I could tell uboot to try booting off the dock's SD card first, and then fall back to the eMMC [07:52] That way, I could put a complete kernel and rootfs rescue image on an SD card [07:52] yeah, just build it in to your scripting [07:52] I have a small patch which uses some space on emmc for environment variables too :) [07:52] I've used systems that have u-boot preinstalled, I haven't installed it before tho [07:53] heh [07:53] Do you build a uboot image and then nvflash --download it to the EBT partition? [07:53] no [07:53] it can only be placed in with a --create [07:53] nvflash will not write EBT at any other time [07:53] Because nvflash is retarded? [07:53] probably [07:54] Don't know if mini/updater prevent it [07:54] or if it is just some retardation from nvflash [07:55] anyway, back to kernel [07:55] with the u-boot kernel most things work [07:55] in fact nearly everything works [07:55] except the TSP is inverted [07:55] and clocking is a little off [07:56] so from time to time the framebuffer corrupts when using acceleration [07:56] and unexplained slowdowns occur [07:57] What determines if a kernel is u-boot compatible vs. ASUS(?) compatible? [07:58] it seems to be a memory map issue [07:58] I have been far too busy with uni to properly track it down [08:06] What are you studying? [08:06] education [08:07] Man, that's a gruelling career [08:07] heh [09:21] janimo, back from vacation ? [09:27] ogra_, hello, back [09:27] looking into the mx53 scripts now [09:28] janimo, do we have anything for mx5 already ? [09:28] bah i type to slow :) [09:28] I'll send you a post-boot script for debian-cd later today [09:28] cool [09:28] that should be all we need [09:29] later some changes to jasper as resize needs to deal with the partitioning that is not like on omap [09:29] k [09:29] but I still need to check as linaro media create and the factory image that came with the board have slightly different partition layouts [09:30] k, we need to get it in today (and working) [09:30] freeze starts later today === zyga is now known as zyga-afk [10:12] ogra_: do you know where persia id hiding? [10:13] is even [10:27] lilstevie, at home i guess, not sure [10:27] heh [10:27] haven't seen him lately === zyga-afk is now known as zyga [11:31] diwic, do you know any good spectrum analyzer SW in the archive (i got that new set of speakers and want to get all the room related issues fixed) [12:13] ogra_, hmm, no I'm afraid not - I mean, audacity sure has one if you just want to see the frequency spectrum of a recorded wave file [12:13] ogra_, but maybe you're after something more sofisticated? [12:13] nah, i need something that takes a mic and outputs a sweep through my speakers [12:14] and generates the frequency spectrum from that ... so i can see wheer the speakers are lacking [12:14] and adjust accordingly [12:14] i can do that with my ears, but only to a certain level :) [12:15] ogra_, hmm, haven't heard of such an app. My receiver has one built in, so I've just used that. [12:16] yeah, mine too ... but that doesnt work well for the subwoofer and base frequencies [12:17] finding the right position in the room for it to eleminate resonances etc [12:17] if you happen to find such a software, i'm interested in hearing how it works out [12:17] i have a very bad room here and i think there are some bad peaks around 30Hz [12:18] i also have some ypao-crap built in the amplifier [12:18] if i try to fix that "by ear" i will never stop to experiment :) [12:18] ah, you got a yamaha then [12:18] yeah [12:19] * ogra_ is planning to upgrade the amp too ... are you satisfied with yamaha ? [12:19] yes [12:19] i've had it for the past 6 years now [12:19] still working perfectly [12:19] i'm considering the a810 ... but i'm undecided, there are also some good marantz ones in that area [12:19] and they look a lot better :) [12:19] the sound is balanced and works nicely with my b&w speakers [12:20] the onkyo 508b i have atm is really disappointing since i have speakers that are bigger than plastic cups [12:21] it doesnt really manage to replicate the details properly unless i turn very loud [12:23] you should also check out the denon models in the pricerange too [12:23] ah, and one important point is phono-in [12:23] none of them gets above 100W [12:23] as in if you need one, most amps don't have one anymore [12:23] and i dont need phono in, my LPs are all turned into flac already [12:24] and i also have a pretty good hifi system (i rarely use) i could use to lead my luxman pre-amp through to the surround amp if needed [12:25] ogra_: with ac100 you are using nvflash to flash it? [12:25] hrw, for the first flashing operation, yeah, later (with teh flashed kernel booted) you can flash locally [12:25] ogra_: so same as me for the tf [12:26] would be cool if someone would write an android installer :) [12:26] would be [12:26] but meh [12:26] not going to happen on my device [12:26] something that just grabs the boot.img and goes into flash mode [12:26] same here [12:26] I need to modify the flash layout [12:27] why is that ? [12:27] because the original layout is a pita [12:27] should be possible though, there are howtos for the ac100 [12:27] boot.img and recovery.img are not within the fs per se [12:28] ogra_: on device flashing is via mtd subsystem? [12:28] like the partition table covers a certain portion and boot and recovery are not within it [12:28] hrw, eMMC [12:28] its juts a partition [12:29] ah [12:30] and you can just dd an image you created with abootimg to it [12:30] yeah I cant [12:30] I need to package into an nvidia blob [12:30] dd to a staging partition [12:31] then run through a reboot which flashes it into the correct place [12:32] phew, why ? [12:32] there is code that enables you to see the LNX and SOS partitions from the kernel [12:33] becuase of how asus implemented the flash layout [12:33] and our device does not use tegraparts [12:34] not at all ? [12:34] i thought all tegras use the partition table offset in the tegra sd code [12:34] tegrapart is just to make them visible afaik, theoretically you should have all bits in the kernel already [12:36] they use tegra parts for the very minimum [12:37] but there is certainly nothing in the kernel to make any of it visable [12:37] or to use it [12:38] because at a kernel level everything the kernel needs is within the GPT [12:47] ogra_: anyway it is a pain of an implementation [12:48] I move the kernels, and tack on my partition for ubuntu [12:49] which gnu parted cant do (work on the partitions that is) [14:14] I want to mess with the initrd on the pandaboard, eg. move the root filesystem around. On a normal system I'd run update-initramfs and maybe update-grub as necessary. What do I need to do here to cope with u-boot, to update the u-boot partition with the new initrd? [14:35] rbasak, the same minus update-grub [14:35] rbasak, though, do you want to actually change the initrd content ? [14:35] or just the cmdline [14:36] ogra_: I want to change the fstab entry for the root partition, so I think that needs to get duplicated into the initrd [14:36] no [14:36] initrds dont use fstab internally [14:37] you want your root= setting on the cmdline to point to the right uuid [14:37] oh, just the kernel cmdline? [14:37] thats all you need to change [14:37] there should be a /boot/boot.script file, edit it and just run sudo flash-kernel [14:38] (you could run update-initrsamfs toon it calls flash-kernel in the end, if you only want to memorize one command ;) ) [14:47] aha [14:47] ok, I know why I want to update the initrd [14:48] It's for /conf/conf.d/cryptroot [14:48] (I always assumed that fstab and crypttab were copied to the initrd, but it looks like update-initramfs generates /conf/conf.d/cryptroot instead) [14:49] so flash-kernel will update the FAT initrd file with the one in /boot? [15:05] right [15:05] as well as vmlinuz and boot.script [15:10] rbasak: Morning. Anything I can do to help? [15:11] hi GrueMaster [15:12] I'm poking with LUKS right now [15:12] It seems to work for non-system partition stuff just fine [15:13] afaik, the easiest way to configure it is at install time. [15:13] Which means using the netinstall image. [15:13] Only note is that if using /etc/crypttab and /etc/fstab to mount something on boot, then the prompt for the passphrase isn't very clean over serial (it echoes the passphrase in plaintext and kernel messages appear async over the top) [15:13] And installing to USB. [15:14] Ah OK, I'll switch to the netinst image then [15:14] oops. That's not good. [15:14] Although at least it is harder to hack. Hard to get a man in the middle with a serial console. [15:14] heh [15:14] :-) [15:15] What will be the recommended/supported boot method? Serial only? Or via USB keyboard + monitor or something? [15:15] (or both?) [15:15] we default to serial atm iirc [15:16] For server, serial console. For desktop, KVM. [15:16] Also I think GrueMaster just answered this - I was going to ask whether we would support encrypted root via the installer or not, since the pre-installed image didn't give me an option. But I think the answer is to use netboot :) [15:16] you would need to modify the cmdline before first boot to change that to monitor [15:18] It doesn't need a man in the middle at the moment, just a man over the shoulder :-P [15:20] I've been using http://cdimage.ubuntu.com/ubuntu-server/daily-preinstalled/current/ so far - where do I get the netinst image from? [15:20] * rbasak has just hunted through ubuntu-server/ [15:20] well, a good over-the-sholder-man doesnt need to see the cleartext, he can guess from the keys you press ;) [15:21] For server I made a keyscript that required a USB key to be inserted (back in Hardy) [15:22] Never got round to contributing it though :-/ [15:22] The netinstall images are here: http://ports.ubuntu.com/dists/oneiric/main/installer-armel/current/images/omap4/netboot/ [15:23] However, if you use them I recommend a different approach than using the img file for now. [15:23] DYO image :) [15:23] bug 806751 will prevent you from finishing. [15:23] Launchpad bug 806751 in debian-installer "Boot partition on SD is too small on omap/omap4" [Medium,New] https://launchpad.net/bugs/806751 [15:24] luckily NCommander is back today to work on that :) [15:25] So you need to download these files (in order) to an SD with a fat partition >30M: MLO, u-boot.bin, uImage, uInitrd, boot.scr-serial (rename to boot.scr). [15:25] Yes, he should have it fixed this week. [15:26] In order? [15:26] Well, MLO needs to be first. [15:26] Specifically. [15:27] First on a fresh FAT partition so it ends up at the beginning? [15:27] Bug in the boot rom and how it looks at the fat table. [15:27] yes. [15:27] OK, understood [15:27] * rbasak hopes that the vfat driver is deterministic in this regard :-P [15:28] So if you have an SD with server preinstalled already on it, just run "sudo mkfs.vfat -F 32 " [15:28] Then mount and copy the files, MLO being first. [15:32] OK I've done that. Before I try booting this, I'd like to remove the second partition on SD now and install to a USB disk instead - will that work? [15:35] * rbasak tries it [15:37] Don't worry about the second partition. It will be ignored. [15:42] It booted, installing now [15:45] Does anyone else's pandaboard whistle? [15:45] Actually it appears to be the PSU [15:45] interesting. [15:46] I once had a cheap replacement laptop PSU that did that and it made the touchpad not work properly when using it [15:46] * rbasak wonders if he has a dodgy PSU [15:50] rbasak: many cheap "wall-warts" are nothing more than a rectifier, cheap transformer, and a 7805 [15:52] rbasak: without proper impedence matching and filtering you can get a audible frequency Oscillation [15:52] * GrueMaster is using an old AT power supply with something like 30A on the 5V rail. No problems so far. [15:52] stop welding your pandas dude ! [15:54] Heh. I've even thought of installing a solar panel outside to run my panda stack from. [15:54] (living in Oregon has pretty much dashed that thought, though. [15:55] well, if you accidentially shorten the right pins with a screwdriver, you might need a new screwdriver in such a setup :) [15:55] GrueMaster: I recently burnt old ATX psu [15:56] ATX power supplies are easy to fry. They usually melt down in order to protect the $1 fuse they have built in. [15:57] Plus a lot of them require a fair load on the 12v rail for the 5v rail to be stable. [16:07] The install worked, but on reboot nothing after "Uncompressing Linux... done, booting the kernel." - at a guess it's not configured for console output? [16:07] also AT-powers have a real power-switch [16:09] GrueMaster: yeah the original boot.scr-serial had a console=ttyO2... line in it, the installer has replaced it and put nothing. I could fix this manually by replacing the file I suppose? [16:10] That is only usefull if you want to see the kernel boot messages. It should still boot with a login prompt on serial. [16:13] I see nothing after two minutes. I'd like to see the kernel boot message to see what's up. [16:14] The USB disk keeps powering down. It does that when on my laptop as well - evidently by default the power save is quite aggressive. I hope that's not affecting it. [16:14] Try hitting enter on the serial console. Sometimes I have to do that to get a login. [16:15] No joy [16:15] It's not really been hitting the USB disk enough to convince me that it's even booting [16:15] If that fails, you will need to manually edit the boot.scr on the SD from your desktop. I'll walk you through it. [16:15] (and bear with me. I have an IRC lag from my overloaded server). [16:16] Is your usb drive powered separately, or from the panda? [16:16] Powered separately [16:16] I've got the SD mounted on my desktop. boot.scr is binary, though I can edit the ascii part with vim [16:17] Hmmm. Ok, lets edit the boot.scr. First, do "dd bs=1 skip=72 if=boot.scr of=boot.script. [16:17] OK, boot.script now looks like http://paste.ubuntu.com/687735/ [16:18] Then you can edit boot.script to add "console=ttyO2,115200" and "earlyprintk=ttyO2,115200". [16:18] Delete "quiet splash". [16:19] OK done [16:19] Now stick those 72 bytes at the front again I suppose? [16:19] Save and exit. [16:20] Yes. Use "mkimage -A arm -O linux -T script -C none -d boot.script boot.scr" [16:20] If you don't have mkimage installed, install u-boot-mkimage. [16:21] That's it. [16:22] It wanted u-boot-tools rather than u-boot-mkimage, but working now [16:22] and...aha! [16:22] I got the kernel output [16:23] Of course it was asking for the passphrase :-) [16:23] ok [16:23] heh [16:25] No plaintext echo problem this time. I need to see if that's a initrd vs. normal init prompt thing or a netinst-fixes-it thing [16:28] Cool. Also, you might try removing early printk, and adding quiet & splash to see how much you need for it to still come up. If all we need is a console= line, then we can add that. [16:31] It seems to die when my usb drive powers down. Looks like a separate issue I better tackle first, as I only get about a minute at a login otherwise [16:32] I'll work out what the minimum required is for the passphrase prompt after that [16:58] ogra_: indeed, I'm working on it (and did some work on it on Sunday). I already can feel the madness setting in [17:48] GrueMaster: OK, it works with quiet spash and no early_printk - ie. only console= is required. For a full disk encrypted from the installer, apart from that it works including using *s for echoing the passphrase as it is entered. If I manually add a non-root partition to crypttab and fstab, then it works but echoes the plaintext bad and generally pollutes the console during passphrase input. [17:48] can anyone check what's going on with actinidiaceae? [17:49] buildlog has not been updated for a while [17:49] GrueMaster: so that's basic functionality. Any need for stress/benchmarking? I presume I need to write this up and automate the testing? [17:51] rbasak: Thanks, good work. For now, if you can just write up the steps to get it enabled, we will be working on automation between Beta 2 and UDS-P (possibly unitl P-A1). [17:51] Goal for now is to do a once through and flush out any issues (like the console=). [17:52] File a bug against debian-installer that states that console= line needs to be added when doing a serial console install. [17:53] OK, writeup at https://wiki.ubuntu.com/ARM/QA/Server I assume? [17:54] I wonder if it needs to be there regardless. Did yhe passphrase prompt come up on the monitor at all? (If you didn't test it, no worries). [17:54] I didn't test it. I don't have the the right lead to hook up a monitor [17:54] yes, that is the correct link, thanks. [17:55] I suspect it asks on whatever the console is [17:56] I'll test it further after Beta 2 with a KVM enabled system to verify. Might require a different cmdline to enable monitor out. [18:07] rbasak: I marked the Test LUKS under server-o-arm-image-qa as done per your results. Thanks again for the help. [18:28] GrueMaster: No problem, pleased to get into something arm server related. I'm off now, I'll write up the results and file the bug against debian-installer in the morning. [20:17] Is there an ubuntu-server build I can get hold of? (Pandaboard) [21:04] SysTom: The official latest release is Oneiric Beta 1, found here:http://cdimage.ubuntu.com/releases/oneiric/beta-1/ Daily builds are at http://cdimage.ubuntu.com/ubuntu-server/daily-preinstalled/. You can also do a netinstall from http://ports.ubuntu.com/dists/oneiric/main/installer-armel/current/images/omap4/netboot/, but be aware of bug 806751. netinstall is for installing to either USB or iSCSI (both have been tested, iSCSI still has a [21:04] post-install bug). [21:04] Launchpad bug 806751 in debian-installer "Boot partition on SD is too small on omap/omap4" [Medium,New] https://launchpad.net/bugs/806751 [21:05] iSCSI bug is 838809. [21:06] sigh. Bug 838809 [21:06] Launchpad bug 838809 in Ubuntu Oneiric "authenticated and unauthenicated iscsi clients fails to complete boot" [High,Confirmed] https://launchpad.net/bugs/838809 [21:16] GrueMaster: thanks :) [21:17] GrueMaster: How's it looking, I assume it's a lot light on resources? [21:17] Well, imagine running a server instance on your cell phone. [21:18] Other than the lack of SATA, it really isn't bad. [21:18] What type of server are you planning? [21:20] Just having a play at this time tbh [21:24] If you just want to see what it feels like, download the preinstalled image and boot it from SD. If you are doing anything that involves more disk i/o, you need a usb drive (much faster). [21:40] Interesting [21:40] I have a spare USB drive here [21:42] What's the method for preparing a USB stick for the netboot edition (http://ports.ubuntu.com/dists/oneiric/main/installer-armel/current/images/omap4/netboot/) [21:43] The netboot edition still boots from SD, but the rootfs can be installed on USB. You need an SD with a primary fat 32 (LBA) partition marked bootable. The first file should be MLO from the netboot directory. [21:44] The next files should be: u-boot.bin, uImage, uInitrd, and boot.scr-serial (for serial console) or boot.scr-fb (for keyboard/monitor installation) renamed to boot.scr. [21:45] Unfortunately, there is a bug in the installer (see above) that prevents you from using the boot.img files. [21:45] Ah ok, thanks for the info- certainly something to look into [21:46] Also, be aware that it pulls a lot of files across the internet. Depending on your network connection, the ports.ubuntu.com mirror may be a big bottleneck, and there are no mirrors (unless you have your own). [21:49] Let me know if you run into issues. I have been doing netinstalls since Alpha 2. [21:49] Thanks GrueMaster [22:01] GrueMaster: ogra_ infinity need a sanity check before I upload http://paste.ubuntu.com/687934/ [22:03] From what I can tell, this looks sane. [22:48] GrueMaster: uploading then [22:48] Will test first thing in the morning.