**** BEGIN LOGGING AT Sun Sep 11 02:59:56 2011 Sep 11 03:13:02 re defconfig -- that's what I guessed. Would be nice to distill it down sometime. Sep 11 03:23:45 lilstevie: what is "the blobs-be-gone enhancement"? Sep 11 03:28:21 Hi. So I installed an ubuntu chroot on my andoid tablet using debootstrap. Debootstrap didn't give me a sources.list. (Actually it gave me an empty one). So I made myself a sources.list and ran apt-get update. It 404ed on all the repos. What is up? Here's my output from apt-get update: http://pastebin.ca/2080718 Sep 11 03:28:40 Also, why is http://lilstevie.geek.nz/ports/nvflash-ubuntu.tar.gz a tarball when it contains only one file. Sep 11 03:28:50 Galaxor2: resolv.conf Sep 11 03:35:34 Hm. Resolv.conf exists and has the contents I expect. debootstrap doesn't seem to have installed any other programs capable of attempting to resolve domain names, so I can't check it with another program. I don't have host or ping or wget pr nslookup... Sep 11 03:36:53 libc performs hostname resolution Sep 11 03:36:59 Try "getent hosts" Sep 11 03:37:05 Or just put IPs in sources.list for now Sep 11 03:37:32 I mean "getent hosts example.net". Also check nsswitch.conf, I guess. Sep 11 03:41:29 getent hosts ubuntu.com resolved it. Sep 11 03:43:25 Also I notice that the error I was seeing was not a hostname lookup failure but a 404. The webserver was in fact contacted. Sep 11 03:49:20 Then I dunno Sep 11 03:49:34 Sorry, I should've read your error pastebin, I just jumped to the DNS conclusion Sep 11 03:50:42 Your problem is that armel isn't carried on that mirror Sep 11 03:51:15 It looks like armel isn't hosted in the normal archive at all. Sep 11 03:52:21 Mrr...? Where should I look for the armel stuff? Debootstrap seems to have found it somewhere... Sep 11 03:52:57 Dunno, I use Debian. Sep 11 03:57:50 Ha. I have sometimes gone to #debian for help if I have a general enough problem but they always get mad at me when they find out I'm actually using ubuntu. I guess the rivalry isn't so intense in the other direction. Sep 11 03:58:42 It's more that I'm asking lilstevie for help and he is here, not in #debian-arm Sep 11 03:59:02 I do use ubuntu x86-64, but not arm. Sep 11 04:02:42 Aha. Sep 11 04:05:08 Ok, I found the answer. In my sources.list, I s/us.archive.ubuntu.com\/ubuntu/ports.ubuntu.com/. That did the trick. Sep 11 04:09:18 twb: it is a tarball to make it smaller Sep 11 04:09:34 it is a 4GB image or 950MB compressed Sep 11 04:11:16 Why not gzip it Sep 11 04:11:23 An archive of one file is silly Sep 11 04:15:25 I mean: ubuntu.img.gz rather than ubuntu.img.tar.gz Sep 11 04:17:00 eh Sep 11 04:17:05 doesn't really matter tbh Sep 11 04:20:40 I guess not, it's just a bit wasteful Sep 11 04:27:55 lilstevie: what's the GPT partition table for? Sep 11 04:28:55 When you said yesterday that I can blow away everything after EBT, I think that GPT needs to stay Sep 11 04:29:42 when I say you can blow it away I mean that they are mobile Sep 11 04:29:54 GP1 and GPT are bookends Sep 11 04:30:37 GP1 at the start and GPT at the end, everything between them will be able to be addressed from /dev/mmcblk0px Sep 11 04:35:49 OK, let me put it this way. default.cfg ships with: BCT, PT, EBT, SOS, LNX, BAK, GP1, APP, CAC, MSC, USP, PER, YTU, UDA and GPT. Of these, Sep 11 04:35:50 BCT, PT, EBT, GP1, GPT -- leave these completetly as is. Sep 11 04:35:50 BAK, APP, CAC, MSC, USP, PER YTU, UDA -- can all be removed. Sep 11 04:35:50 LNX and SOS should exist, but everything except the label can change Sep 11 04:36:15 Is that right? Sep 11 04:40:08 everything else is irrelevant if you aren't using android Sep 11 04:40:47 my current layout is MBR for u-boot Sep 11 04:41:07 but I have BCT PT EBT as standard Sep 11 04:42:23 Oh, OK Sep 11 04:42:45 I thought you needed e.g. EBT still if you wanted to use nv3p/nvflash Sep 11 04:43:17 then MBR UES (u-boot environment store) UIM (u-boot information and media) UBT (ubuntu) MPT(MBR Partition Table endpoint) Sep 11 04:43:29 nv3p resides in the bootloader you upload Sep 11 04:44:04 And the APX stuff is baked into a separate ROM, and I *can't* accidentally host it? Sep 11 04:44:05 EBT is so it is picked up as the bootloader during boot Sep 11 04:44:09 s/host/hose/ Sep 11 04:44:10 correct Sep 11 04:44:22 APX is baked into the SoC Sep 11 04:57:26 in allocation_attribute, which 8 in 0x808 is "allocate to the end of the disk"? And what does the other 8 mean? Sep 11 04:59:13 the allocation attribute as a whole means that Sep 11 04:59:27 0x808 is the allocation attribute to allocate to end of disk Sep 11 04:59:29 OK Sep 11 04:59:54 Oh, I misread. YTU has 8, UDA has 808 Sep 11 04:59:55 in which case size= is an exclusion Sep 11 05:00:18 YTU has 8 UDA has 0x808 :p Sep 11 05:00:22 there is a difference Sep 11 05:00:26 Yeah sorry Sep 11 05:00:40 What do you mean by exclusion? As in, leave that much off the end unallocated? Sep 11 05:00:47 yep Sep 11 05:04:48 You omitted "ro" from the recoveryimg.cfg -- any reason for that? Sep 11 05:05:02 accident Sep 11 05:05:07 k Sep 11 05:05:24 nobodys perfect :p Sep 11 05:06:59 OK, do you expect this to result in something that'll boot? Sep 11 05:06:59 http://paste.debian.net/129314/ Sep 11 05:07:30 Oops, the id=3 in flash.cfg doesn't match your bootimg.cfg abootimg args Sep 11 05:08:11 ? Sep 11 05:09:02 I'm doing a first test using your kernel, ramdisk and rootfs, so that I know that much of the process works, and because building a rootfs is a bit more tedious than I hoped Sep 11 05:09:32 You have root=/dev/mmcblk0p8, but I has set the partition number to 3 in flash.cfg Sep 11 05:10:56 Although when I look, your purelinux.cfg sets UBT to id=16, so clearly those don't need to match up Sep 11 05:11:59 it is a bit different to that Sep 11 05:12:03 your thing is wrong Sep 11 05:12:05 btw Sep 11 05:12:18 BCT must be ID2 PT needs to be 3 Sep 11 05:12:45 and the config starts from 2 Sep 11 05:12:51 Thought you said I didn't need that stuff since I wasn't using android Sep 11 05:12:59 nonono Sep 11 05:13:15 if you read I said that is the bare minimum you must keep always Sep 11 05:13:21 OK, sorry Sep 11 05:13:27 that is a requirement from the tegra2 Sep 11 05:14:27 http://paste.debian.net/129315/ how about that Sep 11 05:15:18 ok better Sep 11 05:15:35 but Sep 11 05:15:40 still one more thing Sep 11 05:16:55 Yeah? Sep 11 05:17:02 you need to wrap in a partition format Sep 11 05:17:53 for the partitions you want to have a /dev/mmcblk0px Sep 11 05:18:24 Aaaah, that's how it gets to p8 being UBT in your one Sep 11 05:18:57 well it is p8 because it is the 8th partition after the GP1 Sep 11 05:19:04 count them :p Sep 11 05:19:05 Yeah Sep 11 05:19:12 I get that now Sep 11 05:19:36 Do you also put LNX and SOS "inside" GP1/GPT, so that you can reflash them from within the TF101? Sep 11 05:19:46 Or can flash-kernel(?) do that without a mmcblk0px ? Sep 11 05:21:23 yes that is the "blobs-be-gone" enhancement Sep 11 05:21:25 anything outside is a pita Sep 11 05:21:40 OK, I think I get that Sep 11 05:22:18 And if I calculate correctly, your purelinux.cfg leaves space at the end for 1MB more than the size of LNX+SOS+UBT -- is that the space required for the GPT partition? Sep 11 05:22:43 thats the placeholder space yes Sep 11 05:23:38 OK. And you have GP1 and GPT because it's basically GPT-style partitioning, and that's why you're also passing "gpt" to the kernel in bootimg.cfg? Sep 11 05:23:57 yes Sep 11 05:24:07 the GPT on the command line is to force though Sep 11 05:24:13 OK Sep 11 05:24:24 because the GPT at the end of the disk ALA traditional GPT doesn't actually work Sep 11 05:24:36 it needs to use the backup GPT which is located at GP1 Sep 11 05:25:05 Is that because nvflash is fucked up, or what? Sep 11 05:25:15 I mean: why is the end-of-disk GPT buggered Sep 11 05:25:17 yeah Sep 11 05:25:27 nvflash is badly put together Sep 11 05:25:38 OK, I'm trying it now Sep 11 05:28:40 Heh, no asus splash, it was going straight into APX mode and I didn't realize at first. Sep 11 05:29:06 (My first upload failed because I told it to use EBT.bin but the file was called EBT.img.) Sep 11 05:29:37 heh Sep 11 05:30:16 And again with recovery.img vs. SOS.img Sep 11 05:30:30 It seems that after it bork borks, I need to reboot back into APX mode Sep 11 05:32:45 lilstevie: could you use e.g. parted from inside the TF to fix the GPT partition? Sep 11 05:32:57 Goddammit, OS.img is too large for partition Sep 11 05:33:02 oh yeah Sep 11 05:33:18 this is another issue due to how poorly nvflash is coded Sep 11 05:33:35 remove filename= from the config Sep 11 05:33:44 after the create go back into APX Sep 11 05:34:09 and use --download? Sep 11 05:34:13 then --download Sep 11 05:34:14 yeah Sep 11 05:34:25 also no parted does not work Sep 11 05:34:34 Is that because the validation thing doesn't check if size is 0x808'd ? Sep 11 05:34:37 due to the lack of pmbr Sep 11 05:34:42 correct Sep 11 05:34:45 parted speaks GPT Sep 11 05:34:54 Not just msdos I mean Sep 11 05:35:03 yeah parted speaks mbr but it likes a protective mbr Sep 11 05:35:08 Oh right Sep 11 05:35:09 to let it know that it is GPT Sep 11 05:35:53 Ooh Sep 11 05:36:08 It rebooted and I got penguins and into initramfs prompt Sep 11 05:36:17 It was even fast Sep 11 05:36:30 :) Sep 11 05:36:40 ok so now drop her back into APX Sep 11 05:36:48 and upload that nice osimg Sep 11 05:37:01 now, do you have a dock? Sep 11 05:37:04 yeah Sep 11 05:37:10 ok Sep 11 05:37:19 Can't hit volup fast enough anymore, it boots too quick :-/ Sep 11 05:37:36 hold volup and power Sep 11 05:37:44 not power then volup Sep 11 05:37:45 :) Sep 11 05:37:46 oh, that kicks in *after* the penguins Sep 11 05:38:02 I thought I had to hit power, then hold both while I had the asus splash logo Sep 11 05:38:07 nah Sep 11 05:38:11 bootrom man :p Sep 11 05:38:31 needs to detect that button within a few hundred miliseconds of the reset vector Sep 11 05:38:31 I'm used to PCs where if you get penguins you have missed the boat Sep 11 05:39:03 so as long as both are triggered at the same time if not before power you are good Sep 11 05:41:56 Oh fuck, I really shouldn't have done --download from within Emacs' terminal Sep 11 05:42:07 I can see it writing each char of b y t e ... s Sep 11 05:42:35 lol Sep 11 05:42:48 Hmm, it seems to be doing the writes themselves OK, though Sep 11 05:42:56 It's done 10% of the 2GB already Sep 11 05:43:11 Maybe nvflash is smart enough to STFU and write progress slowly if you have TERM=dumb or something Sep 11 05:43:40 Am I write in thinking the eMMC is going to have a "modern" number of writes? Sep 11 05:43:47 yeah Sep 11 05:43:51 it is a toshiba nand Sep 11 05:43:57 I mean, it's not like those really early MTD devices where you get like 100 writes and then you're fucked Sep 11 05:43:59 well at least on the 16GB one Sep 11 05:44:24 and a hardware drivven FTL Sep 11 05:44:54 Yeah Sep 11 05:45:19 Although it would be nice if I had an open implementation of the FTL, to go with Coreboot and OpenFirmware ;-) Sep 11 05:45:43 50% done Sep 11 05:46:09 heh Sep 11 05:48:22 well when we are looking at the FTL is so hw driven that the raw device is post processing Sep 11 05:49:45 100% done, now to reboot Sep 11 05:50:15 ./nvflash -r --go will work too Sep 11 05:50:21 Can I issue a reboot via nvflash, or do I need to let the kernel- Sep 11 05:50:25 Ahahahaha Sep 11 05:50:39 It's refusing to work because the ext last-mount timestamp is in the future. Sep 11 05:50:49 lol Sep 11 05:51:04 you never booted android once did you? Sep 11 05:51:10 Right Sep 11 05:51:22 RTC wasn't set :p Sep 11 05:51:44 you could always drop into single and run fs recovery though right? Sep 11 05:51:47 Actually I did but then I reset it to factory defaults, and I couldn't get it on the network because I use WPA2 Enterprise, and I couldn't get it to recognize an ext2 SD card with the SSL cert on it Sep 11 05:51:57 single ≠ break Sep 11 05:52:15 idk Sep 11 05:52:15 I will try tho, because otherwise I need to reupload the whole rootfs again Sep 11 05:52:33 I have not had a problem with the RTC losing time though Sep 11 05:52:35 oh nm I can just hit "stfu fsck, continue anyway" Sep 11 05:52:46 hah Sep 11 05:52:47 I am so used to mountall just completely failing Sep 11 05:53:00 esp. with diskless clients Sep 11 05:53:05 ah Sep 11 05:53:31 Screen is nice and high-res compared to my 1005PE Sep 11 05:54:36 The map part of ubuiquity still needs work tho, hard to pick melbourne instead of hobart or sydney with my chubby fingers Sep 11 05:54:59 so tsp is working? Sep 11 05:55:01 awesome Sep 11 05:55:23 yeah single touch anyway Sep 11 05:55:33 heh I have skinny fingers and it is still hard to pick melbourne Sep 11 05:55:38 The trackpad thing beneath the keyboard isn't Sep 11 05:55:47 yeah expected Sep 11 05:55:57 there is a patch for that somewhere in the thread on xda-devs Sep 11 05:56:02 keyboard is, which is the importantest thing Sep 11 05:56:13 yeah Sep 11 05:56:18 keyboard works great Sep 11 05:56:27 keymap could do with a tweak Sep 11 05:56:34 which it does have in teh uboot kernel Sep 11 05:56:53 if you compiled from my 2.6.36.4 kernel though you have the important 2 Sep 11 05:57:19 esc, and ctrl Sep 11 05:57:28 I'm currently just testing your pre-compiled kernel Sep 11 05:57:37 The APX-flavoured one, not the u-boot flavoured one Sep 11 05:58:03 oh so you didn't compile your own? Sep 11 05:58:04 :p Sep 11 05:58:20 Not yet Sep 11 05:58:32 I think I want the CrOS one but then I need to learn about u-boot Sep 11 05:59:08 And having my own rootfs is more important than my own kernel, at least until Debian starts pre-rolling them for me with everything =m Sep 11 05:59:10 :-) Sep 11 05:59:28 heh yeah Sep 11 05:59:30 Stupid password checker thinks passwords are weak unless they contain at least one uppercase character Sep 11 05:59:47 I already have tat Sep 11 06:00:05 but meh, bruteforcable passwords are not the issue that todays society faces Sep 11 06:00:10 it is password entropy :p Sep 11 06:00:15 My old laptop has no password at all Sep 11 06:00:24 If you want to log in remotely you have to use multifactor auth Sep 11 06:00:44 I use cert based authentication myself Sep 11 06:01:20 One of the things I'm thinking about doing is LUKSing the user partition Sep 11 06:01:34 And keeping the something-you-have on an SD card or something Sep 11 06:02:04 heh Sep 11 06:04:15 Hmm, it's looking for /lib/modules/2.6.38-8-omap/modules.dep Sep 11 06:05:00 When oem-config finishes I have a white bar down the left, presumably there are supposed to be apps there Sep 11 06:05:20 hmm you are using the ubuntu fs? Sep 11 06:05:30 the bar is white due to a bug in unity Sep 11 06:05:35 ok Sep 11 06:06:54 the apps take quite a while to come up Sep 11 06:07:14 Even for gnome Sep 11 06:09:22 Is it supposed to wake up from suspend? Sep 11 06:10:15 button Sep 11 06:10:22 the power one, push it Sep 11 06:10:28 it didn't like it Sep 11 06:10:35 sometimes it glitches waking up though Sep 11 06:10:35 Also the backlight was still on Sep 11 06:10:46 oh, that isn't suspend Sep 11 06:10:51 I managed to power it off by holding the power button down Sep 11 06:10:53 that is just screensvaer Sep 11 06:11:06 I put it into suspend by hitting power button and picking suspend frmo teh gnome-power-manager popup Sep 11 06:11:07 power management is up the creek though Sep 11 06:11:21 Nod Sep 11 06:11:25 due to how android handles it vs. a real os Sep 11 06:11:41 I remember all that wakup crap that android wanted to get into mainline Sep 11 06:11:56 wakelocks Sep 11 06:12:00 ya Sep 11 06:12:21 the most god aweful system ever :p Sep 11 06:12:59 yeah, well I'm not a fan of upstart or polkit/consolekit or udisks either :P Sep 11 06:13:27 heh Sep 11 06:20:16 I guess i can expect at least 30% faster once I switch to armhf Sep 11 06:22:46 I was talking to somebody yesterday and they said "so when you have the tablet part disconnected, how does the kb work... bluetooth?" Sep 11 06:22:54 And damn that would be awesome if it did Sep 11 06:26:03 In unity, where's the equivalent of the gnome preferences menu? Sep 11 06:26:13 well Sep 11 06:26:31 there isn't really one Sep 11 06:27:21 the scrollbar in the gnome appearances pane is really fucking with my head, I can't work out how to make it scroll Sep 11 06:27:45 it is a bit odd to get used to Sep 11 06:27:47 It's got a little thin bar instead of a scrollbar, and when I try to scroll on it, an up/down arrow thing appears, but doesn't seem to do anything Sep 11 06:28:10 hold the up/down arrow thing and drag it Sep 11 06:28:31 OK I got it Sep 11 06:28:53 That would give Fitts a fit. Sep 11 06:29:12 Is that part of that libhildon stuff that nobody uses? Sep 11 06:29:29 no idea where it comes from but it is in unity Sep 11 06:31:04 It's not smart enough to know it's upside-down, either Sep 11 06:32:19 Unity does not make use of hildon Sep 11 06:32:32 k Sep 11 06:35:59 There should probably be an easy way to start onboard from within unity or gnome Sep 11 06:36:54 there is :p Sep 11 06:37:11 Boy, I hope I can remap the magnifying glass to Alt_L Sep 11 06:37:25 twb: yeah look at my 2.6.36.4 kernel Sep 11 06:37:27 already done :p Sep 11 06:37:32 That's what I'm running isn't it Sep 11 06:37:46 nah, the prebuilt one is the asus drop Sep 11 06:37:52 But surely you can do it in /etc/default/kbd or so Sep 11 06:37:56 yeah Sep 11 06:38:05 you can edit the keymap too Sep 11 06:39:50 Looks like the asus one has RT stuff in it too Sep 11 06:39:54 That's what "PREEMPT" means, right? Sep 11 06:40:48 What's /bin/adbd? Sep 11 06:41:19 adbd :p Sep 11 06:41:33 Yeah but what is it. There's no manpage Sep 11 06:41:48 thats because it is from android Sep 11 06:41:58 android developer bridge daemon Sep 11 06:41:59 And it's not provided from deb Sep 11 06:42:01 Ew Sep 11 06:42:17 that is so people without a dock can get through ubiquity Sep 11 06:42:57 Presumably it does something useful even when there's no android? Sep 11 06:43:18 yeah it allows usb connection using adb Sep 11 06:43:21 Oh, it's an RPC listener to allow you to connect from other other- right Sep 11 06:43:34 FYI, it's still running even after ubiquitity is done and a restart Sep 11 06:43:42 yeah, Sep 11 06:44:02 it is started every boot Sep 11 06:44:12 it does have more uses than just ubiquity Sep 11 06:44:28 k Sep 11 06:44:45 It just sounds to me like an attack vector Sep 11 06:45:04 heh Sep 11 06:45:34 Gah, I'm so used to Caps being a control key Sep 11 06:45:54 This is why I never learnt dvorak – too disonnant on random other devices Sep 11 06:45:59 *dissonant Sep 11 06:48:59 twb: adbd may be an attack vector, but that is moot as you would already have the device in your hands :p Sep 11 06:51:41 Yeah, I *mostly* agree with that Sep 11 06:51:57 adbd ONLY works over usb Sep 11 06:51:59 :p Sep 11 06:52:23 as it requires the kernel interface Sep 11 06:53:00 resize2fs worked on the device, as expected Sep 11 06:53:19 IOW I made a 2G image from your 4G one, uploaded that, then grew it to the full ~30GB Sep 11 06:54:34 yeah Sep 11 06:54:39 my new image is 2GB Sep 11 06:54:47 just need to get around to uploading it Sep 11 07:02:51 Hmm, bluetooth doesn't work? Sep 11 07:06:18 Hmm, in /sys/class/power I have ac, battery and usb Sep 11 07:06:23 Where's the second battery? Sep 11 07:08:53 Hmm, acpid isn't installed Sep 11 07:37:08 lilstevie: when you build your kernels, do you use "make deb-pkg zImage" or what? Sep 11 07:54:25 twb: the second battery stats arent in all kernels Sep 11 07:54:38 bluetooth requires you to compile brcm_patchram_plus Sep 11 07:54:47 Hum Sep 11 07:55:09 and no I just make CROSS_COMPILE= LOCALVERSION= BLAH= Sep 11 07:55:14 On all the other Eee's I just leave this stuff to SynrG to get folded into the debian kernel :P Sep 11 07:55:26 and been using linaro packaging to get debs build Sep 11 07:55:39 deb-pkg is handy because it gives you a deb with the =m .kos Sep 11 07:55:58 linaro packaging does too Sep 11 08:02:56 Well, I'm done for today. Sep 11 08:03:09 See you next weekend or so Sep 11 08:04:01 haha kk Sep 11 12:56:01 Hello, I'm trying to get networking up on a beagleboard-xM Sep 11 12:56:17 what is the recommended module Sep 11 12:56:18 ? Sep 11 15:58:18 travalas: smsc95xx Sep 11 16:56:01 is anybody here using natty on a beagle board-xM rev C? I installed it and ethernet and usb doesn't work **** ENDING LOGGING AT Mon Sep 12 02:59:57 2011