**** BEGIN LOGGING AT Mon May 23 02:59:58 2011 May 23 08:36:34 persia, did you find a working nvflash yet ? May 23 08:40:12 No, but I'm not sure that's the issue. May 23 08:40:21 nvflash claims it can't see the USB device. May 23 08:40:35 Looking at syslog, it seems that when I attach, two USB devices show, and then both disconnect. May 23 08:40:43 * persia reproduces to get a viewable snippet May 23 08:42:22 hrm. Reproduction failed. I wonder what changed. May 23 08:43:12 well, are you sure the device was actually in flash mode ? May 23 08:43:30 the usb device will only show up if its in that mode May 23 08:44:50 Right, but it was the immediate disconnect that was the issue. As long as it stays connected, it ought be OK. May 23 08:45:12 yep May 23 08:49:17 Ugh. I tried to backup partition 0, and it failed, and now I get protocol errors. I guess this will be a bit of fun trial and error. May 23 08:50:10 there is no 0 partition ;) May 23 08:50:16 it starts at 2 May 23 08:50:28 hence needing to power cycle :) May 23 08:50:36 nvfalsh has a get partition table command May 23 08:50:37 There's no 1 either? May 23 08:50:44 no 1 either May 23 08:50:47 What's that command? May 23 08:50:56 * persia doesn't have a manpage :( May 23 08:50:58 dont ask me what they are smoking at toshiba :) May 23 08:51:08 it should have a help command May 23 08:51:26 nvflash help ? May 23 08:51:33 --help i think May 23 08:51:42 with all the LD_PRELOAD mess May 23 08:51:54 Indeed it does. May 23 08:53:52 --getpartitiontable May 23 08:54:00 with a textfile as output May 23 08:54:21 Yep. So, I don't really like this tool. May 23 08:54:35 nobody does May 23 08:55:17 1) It's non-free. 2) It doesn't qualify for multiverse. 3) It's not packaged. 4) It has no manpage, and poorly formatted --help. 5) it has some of the worst subcommand grammar I've ever encountered. May 23 08:55:35 heh, yeah May 23 08:55:43 sadly we need it May 23 08:55:50 i dont know of any alternative yet May 23 08:56:23 i wonder how trimsclice does it May 23 08:56:38 they have largely the same board in their device May 23 08:57:04 and u-boot is missing input support yet May 23 08:57:26 and even then you would have to flash u-boot first May 23 08:57:36 Lovely. I seem to get to run nvflash once per powercycle. May 23 08:57:53 you can use the --resume command May 23 08:58:11 replace the --bl flastboot.bin with it May 23 08:58:25 it can only upload the bootloader once May 23 08:58:53 Aha! And --resume lets me run another command without getting "usb read error (71): Protocol error" ? May 23 08:59:14 i think so May 23 08:59:17 try it May 23 09:00:10 Hmm. Less than ideal. This power cycle the bootloader failed to install :/ May 23 09:00:49 well, luckily you only have to do it once May 23 09:02:22 --resume seems to work. May 23 09:02:46 No, I'm super-extra-paranoid: I have to do it ~15 times, plus any that are unsuccessful. May 23 09:03:49 i mean once you flashed successfull you dot have to do it anymore May 23 09:04:12 though i would recommend also flashing a fallback kernel to part 5 May 23 09:04:31 in case you mess up anything you can still boot from it May 23 09:04:33 Why? Is there some way to select the partition from which one boots? May 23 09:04:38 yes May 23 09:05:02 if you hold down the home key during power up you get into fastboots recovery menu May 23 09:05:09 If one bricks the device, can one restore with nvflash? May 23 09:05:15 if you press 1 there it boots from part 5 May 23 09:05:20 yes May 23 09:05:26 the device is unbrickable May 23 09:05:51 as long as you can restore the partitioning you will always get it back to life May 23 09:06:36 That's nice. May 23 09:06:48 part 5 is sadly only 5M big May 23 09:07:01 so there is no space for nifty initrds May 23 09:07:06 On most of my devices there is some flash that, if overwritten, will make the device completely useless. May 23 09:07:30 tegra seems to be similar to omap here ... the rom code will always work May 23 09:07:33 I suspect that is true for this, except that there is no way to write that flash (who said updating device firmware was a good idea) May 23 09:07:40 and it seems the flash mode lives there May 23 09:08:32 Oh, nifty. So it's not like BIOS flash, or the boot NOR on mx5 devices (yes, you can work around the boot NOR with the DIP switches, but they aren't exposed in real devices) May 23 09:08:50 right May 23 09:09:12 Seems that --resume only works once, and after two operations, I really have to power cycle. Oh well. May 23 09:09:34 do you do a full backup ? May 23 09:10:23 That's my plan. My hardware is different from yours, and my software is very different. May 23 09:10:32 yeah, makes sense May 23 09:10:38 (unless you have the REGZA control suite by default, etc.) May 23 09:10:45 lol, no May 23 09:10:56 what do you do with that, remote control your TV ? May 23 09:11:02 That's what I thought :) May 23 09:11:19 Rather, participate in your entertainment network. May 23 09:11:36 ah, REGZA is more than a TV brand in .jp ? May 23 09:12:20 At one level you can consider this a remote control for your TV, but you can also do things like switch the audio track from your TV to your stereo to your laptop headphones, control video streaming from your PVR to your TV or your laptop, and switch seamlessly, etc. May 23 09:12:34 ah May 23 09:13:07 REGZA is Toshiba's home entertainment brand. They do TVs, PVRs, Speaker systems, Keitais, Tablets, etc. May 23 09:13:17 ah, i didnt know that May 23 09:13:28 i only know trhe term from my TV :) May 23 09:13:41 I haven't seen a REGZA Stereo, but the protocol should work with Onkyo or Denon hardware. May 23 09:14:07 Yeah. The nifty stuff takes a while to export: someone has to figure out how to explain it to foreigners. May 23 09:14:15 heh May 23 09:15:18 No, seriously. Here they just stick a couple stickers on it, and the guy in the shop says "It connects to your network", and nobody asks questions. May 23 09:15:30 Other places people want to know why they should pay more for feature X. May 23 09:15:49 * persia suspects this has something to do with not having a law against selling electronics more than three years old May 23 09:17:05 hey persia May 23 09:17:22 XorA, Hey. May 23 09:17:46 Im amused as I just bought an item that is now 30 years old May 23 09:20:03 now I just need to build a circuit to adapt its video output to actually work with something modern so I can see it working :-D May 23 09:21:06 * persia laughs May 23 09:24:47 Grr... Partition 8 consistently fails. May 23 09:25:07 is that the biggest one ? May 23 09:25:49 i think nvflash has a prob with partitions bigger than 4G ... usually thats only the user data partition anyway and empty ... May 23 09:26:23 Not by a long way. Biggest is 6774016 sectors. 8 is only 153600 May 23 09:27:48 weird May 23 09:29:22 Yeah, well, it's the oddities that make me do a full backup before I even boot the device. May 23 09:33:21 * ppisati tried the mmu patch for kexec but still fails.. May 23 09:47:52 ogra_, The partitions I can't read are labeled APP, UDA, and UDB. Do you know what these codes mean? May 23 09:48:26 APP is android apps i think, did you look whats indside ? May 23 09:48:34 ah, no. you said you didnt boot yet May 23 09:49:59 (though the part table is completely shifted anyway in userspace ... you will never see part 2-6 in android May 23 09:50:05 ) May 23 09:50:37 UDA and UDB would be User Data A and User Data B? May 23 09:50:45 so you have to guess which ones are APP, UDA, UDb May 23 09:50:53 might be May 23 09:51:35 Ugh. 8 gives a consistent format error, maybe some sort of copy protection. 12 and 14 just time out attempting to copy. May 23 09:52:18 14 should be mmcblk0p7 in userspace May 23 09:52:32 that falls under the "biggest partition" stuff i mentioned above May 23 09:52:37 no need to back it up May 23 09:52:43 should be empty May 23 09:53:06 I suspect so: it's 13,230M May 23 09:53:15 yeah May 23 09:54:18 But the 300M APP partition seems like something I want backed up, and I got just over a GB from the 1235 UDA parittion on one try, although most tries have been lower. May 23 09:55:51 oh, if your device should offer you to upgrade to android 2.2, dont do it if you want to keep the dual boot option May 23 09:56:17 if you should boot it into android before installing May 23 09:56:49 I don't intend to ever boot Android on it, unless I have to verify a hardware issue for the warranty. May 23 09:59:41 Connecting to a full speed hub rather than a high speed hub seems to help, although it takes *MUCH* longer. May 23 10:00:50 HUB ?!? May 23 10:01:06 * ogra_ just connects directly May 23 10:01:38 You connect directly to the controller? What sort of device exposes that? I've never seen one. May 23 10:02:18 i connect a mini usb cable between my x86 laptop and the ac100 May 23 10:02:30 without extra hubs attached May 23 10:03:54 Right. How many USB ports does your laptop have? May 23 10:03:59 3 May 23 10:04:21 and i usually connect to a full speed port May 23 10:04:37 Then you probably have 2 hubs in your laptop. The first providing those three ports and uplink to the second, and the second conneting to internal devices. May 23 10:04:51 right, i thought you referred to an external hub May 23 10:05:04 I was on high speed, which is something like 30x as fast, but it kept crashing. Now I'm on full speed, which works better. May 23 10:05:24 k May 23 11:04:59 ogra_, Aha! I encountered the 4G issue. You don't happen to know if there is a 64-bit nvflash floating around somewhere, do you? May 23 11:05:53 hmm... sheet.zoho.com is default viewer of xls files under ubuntu? May 23 11:06:30 hrw, In some environments (like those where openoffice/libreoffice was considered too heavy/slow/broken) May 23 11:06:45 arm for example May 23 11:07:08 From what I understand, natty libreoffice works, but it's kinda slow. natty koffice is a bit faster, but predates the revival/rebranding. May 23 11:07:53 my daughter just sent panda x11 session to logout... power of power key ;:D May 23 11:08:18 heh May 23 12:08:05 Hey all, Could someone suggest anything? I'm running ubuntu on my arm tablet, I've built and installed touchscreen drivers and the cursor now follows my touch. The only problem is, no click event is launched on touch release/tap. Is there any way I could configure this? Thanks May 23 12:08:32 Which tablet? May 23 12:08:49 Vega May 23 12:09:12 Try `xinput list` to find the id of your touchscreen May 23 12:09:40 Yup, I have that May 23 12:09:54 `xinput list-props ` will tell you what it belives it can do. May 23 12:10:27 `xinput watch-props ` will let you see what it is sending. May 23 12:10:51 ok, thanks May 23 12:11:40 Depending on how the driver is implemented, you may need to have it send tap events, you may need to trap and process tap events to turn them into button activations, you may need to fiddle with input-utils to make sure the kernel has the right bits for the event driver... May 23 12:12:12 hmm ok May 23 12:13:27 It seems that despite moving my pointer, watch-props isn't providing any feedback. (I've tried all of the pointer devices and the touchscreen identifier specified in xorg.conf isn't in the list) May 23 12:15:53 Right. I presume you're well past the xev stage, so diving in: does your driver provide a kernel event interface? May 23 12:20:35 Umm, not too sure. This is the first time I've had to compile and install drivers manually so I'm kinda feeling my way around still. Through inspection of the source it seems it provides the click event via a call to xf86EventButtonPressP (or something similar to that) May 23 12:21:11 Then it's just an X driver, making xinput the best tool for analysis. May 23 12:21:17 Check your X log to make sure it's being loaded. May 23 12:26:27 It seems to attempt to load it, but fails to initialise a touchscreen device and unloads due to EV_SYN missing. I can only assume it loads in some description of state as prior to installation, the cursor would just jump to 0,0 on touch. Post installation it follows touch May 23 12:28:18 Hrm. I'm not sure. You might ask the team in #ubuntu-x, although they may send you back here if there are architecture-specific confusions. May 23 12:28:27 (or someone else might know, if you wait a while) May 23 12:28:57 isnt there also #ubuntu-touch ? May 23 12:29:20 Is there? I hadn't encountered it before, but yeah, if it exists, it's a better place :) May 23 12:29:46 Indeed there is! May 23 12:30:01 Great, I'll ask there - thanks for the help ^^ May 23 12:30:07 they focus more on enabling multitouch in existing drivers though May 23 12:30:26 Well going from 0 to N is multi, right? May 23 12:30:26 so they might not know much about new ones May 23 12:31:06 the prob smells a bit like evdev would just select a mouse driver for the device May 23 12:31:41 Or any other class of incorrect autodetect. May 23 12:31:50 yep May 23 12:32:35 ogra_, My notes have a confusion: I'm writing boot-2.6.37-1-ac100-SD.img to partition 6, right? May 23 12:32:44 yes May 23 12:33:08 OK. The command was listed once with a 6 and once with a 5, so I just wanted to double check. May 23 12:34:14 Oh my. Partition 14 reduced to 0.1% of raw size with gzip :) May 23 12:35:10 heh May 23 12:35:53 OK. partition flashed. Device OFF. USB disconnect. SD insert. Close eyes and cross fingers... May 23 12:36:00 :) May 23 12:36:09 * ogra_ crosses May 23 12:36:48 Yeah. ALSA errors then bootsplash. Thanks a lot! May 23 12:36:56 awesome ! May 23 12:37:03 the rest should be trivial May 23 12:37:42 Yep. May 23 12:38:51 Heh. Jasper. May 23 12:38:57 jasper ? May 23 12:39:20 there shouldnt be any jasper in these images May 23 12:39:43 This is casper? What's giving me initial configuration? May 23 12:39:52 oem-config ? May 23 12:40:08 there should be neither casper nor jasper May 23 12:40:50 Oh. It's an OEM image. I suppose that makes sense. May 23 12:40:55 right May 23 12:41:06 its directly derived from the omap3 one May 23 12:41:06 (although it's a bit annoying now knowing that I have to go through this config process twice) May 23 12:41:32 thats why i said you should set a rootpw or edit /etc/shadow May 23 12:41:39 for the SD boot May 23 12:41:56 this will be automated by an init script for the actual image May 23 12:42:10 so you only need to cp the tarball onto a formatted SD May 23 12:42:25 Really, we should use the installer. May 23 12:42:38 But as we discussed before, this requires tracking down why cp is slow. May 23 12:42:39 no May 23 12:42:57 i think the oem team has a udeb that can install tarballs May 23 12:43:00 Yes. Really yes. I'm inflexible on this. Maintenance overhead. May 23 12:43:03 that might be an option May 23 12:43:26 but i'm really not going through setting d-i to work with universe kernels etc May 23 12:43:27 It's all part of the same suite that includes live-helper. May 23 12:43:41 No, but NCommander has a WI to do that for a different spec. May 23 12:43:45 So you don't have to worry about it. May 23 12:43:48 so it will stay live/oem May 23 12:44:12 NCommander, has no WI to make d-i work with universe bits May 23 12:44:51 there is only one d-i related workitem ... and thats just a partitioning reciepe for partman May 23 12:45:48 It's on a foundations spec. May 23 12:45:56 universe ? May 23 12:45:59 Yes. May 23 12:46:05 that would need a complete redesign May 23 12:46:25 d-i needs the kernel in main May 23 12:46:37 and all other bits it uses to create images May 23 12:47:15 You were *at* the UDS discussion about it. Now isn't the time to complain. May 23 12:47:35 well, *shrug* May 23 12:48:07 So, anyway, assuming the component issue is resolved, and the speed issue is resolved, I'll insist on proper installer images. May 23 12:48:17 i wont change the ac100 image, since i find wasting hours of user time for copying or formatting tasks completely pointless May 23 12:48:22 But that's not soon: make them however you want until then. May 23 12:48:44 and i wont change them in the future either ... May 23 12:48:46 Please read what I wrote above again? Your objection is already included. May 23 12:49:21 it adds massive complexity for maintenance as well as for the enduser May 23 12:49:52 Really it doesn't. It *streamlines* maintenance and provides a consistent experience for end users. May 23 12:49:56 we have oem-config exactly for this task ... i dont plan on moving away from it May 23 12:50:01 Mind you, currently it's painfully slow, which needs sorting. May 23 12:50:10 It's also impossible, which needs sorting. May 23 12:50:14 there is more than slowness May 23 12:50:32 you have to do a d-i rebuild for every kernel version change May 23 12:50:41 that doesnt scale to universe kernels May 23 12:51:24 (it is already annoying enough for the non mainline kernels we have in main ... which was one of the most pressing reasons for preinstalled oem images) May 23 12:51:36 So, this will be sorted. May 23 12:51:44 not by me, mind you May 23 12:51:51 No, not by you. May 23 12:52:07 and as mentioned above, you will need a lot to convince me to move away from oem images May 23 12:52:11 But when everyone else has done their bit, these images will switch to using an installer. May 23 12:52:21 not mine May 23 12:52:24 I have a Lanya smart book X6-7A with an SSD and I was wondering if putting ubuntu on it would be any different. It doesn't have a normal load up like another computer would, it just pops up the SmartBook screen and does the 25 second loading process. Would I put a live load in a USB and go from there? May 23 12:52:27 Whether you do this or not at that point can be discussed later. May 23 12:52:40 D34X, What's the processor? May 23 12:52:50 i find it a massive waste of manopower to even put time into that May 23 12:52:59 persia - ARM 926 May 23 12:53:33 its fine for server but i dont see even the slightest reason to push us back into stonegae with the desktop images May 23 12:53:51 D34X, Ubuntu doesn't support that processor. You might try Debian. May 23 12:54:14 Poo.... May 23 12:55:27 ogra_, I feel the opposite. I have no interest in remaining in an age of hardware-dependent images. I want generic solutions, that work for arbitrary hardware. We are getting much closer with the tools to be able to do this, and when we can, I'll have no interest in supporting less generic solutions. May 23 12:55:47 where would a generic image help on the ac100 ? May 23 12:56:09 well, let's assume we are able to construct a generic tree that supports all Tegra2 devices. May 23 12:56:13 and whom would a generic image help on a specific board that can only run this specific image ? May 23 12:56:20 LOL May 23 12:56:24 Then we should only maintain that image. May 23 12:56:25 that will never happen May 23 12:56:26 Why LOL? May 23 12:56:34 It will. This is the *point*. May 23 12:56:49 because nvidia doesnt care at all and the board specific implementations collide heavily May 23 12:56:58 If this was not a goal (and a believeably achievable goal), we wouldn't do doing any of this. May 23 12:57:16 well, you wont see such a kernel for a long long time May 23 12:57:44 it might happen at some point indeed ... like it did for the beagle after what ... 4 years or was it five May 23 12:58:11 I'll continue to expect it tomorrow, and be disappointed with anyone that feels otherwise, simply because if we don't believe it will happen, we won't help make it happen, and we'll be discouraging the folk that are trying. May 23 12:58:20 the only organization working on that (linaro) has no support at all for tegra May 23 12:58:40 and there is no movement from either side May 23 12:58:58 That's true today. I'm not convinced it has always been true nor that it will always be true. May 23 12:59:36 *if* you get such support (which is already unlikely because every odm seems to implement its own bootloader solution) it will happen long after your ac100 collects dust in your basement May 23 13:00:05 All I ask of you is that if the valid bugs you raise can be sorted, and we can make generic solutions, you help with those, rather than refusing. If you're right, and it's a long time before we get there, the cost to you of agreeing is nothing. May 23 13:00:27 oh, i surely wont block any progress either way May 23 13:00:39 i just wont change my images :) May 23 13:00:56 even in a generic setup i would go for oem images for desktop May 23 13:01:12 For all architectures? May 23 13:01:19 at least on devices wheer you have things like hardcoded partition tables etc May 23 13:01:51 its just pointless to count on d-i for two features we dont use/need at all May 23 13:01:52 I'd rather have a generic installer that could recognise those devices, and skip bothering the user about partitioning, but rather just do the right thing. May 23 13:02:08 I think you aren't understanding what I'm saying. May 23 13:02:09 thats what an initrd script can do as well May 23 13:02:21 in an easier way for the user May 23 13:02:26 I don't intend to show the user anything pointless: that's poor interaction. May 23 13:02:32 and with zero maintenance for the developer May 23 13:02:56 With two pieces of code to maintain, to have bugs, to have security issues, and two places to fix them all, and inconsistency to confuse the support and advocacy teams,etc. May 23 13:03:31 i only care for one place really ... May 23 13:04:04 So someone else has to care for another? No opportunity for collaboration? May 23 13:04:36 i wont pull d-i into the installation May 23 13:05:05 d-i has its areas where it makes sense ... in this particular install to totally doesnt and only gets in the way May 23 13:05:11 You do know that live-helper is a d-i component, so you'll be doing that anyway for all the preinstall images that use livecd-rootfs today, right? May 23 13:05:26 not during install time May 23 13:05:32 only during build time May 23 13:07:10 what i would prefer to see would be a generic oem images initramfs hook May 23 13:08:03 so that you hand over filesystem and target device on the cmdline or some such and a tarball gets unpacked to the target after verifying teh taball md5 sum ... May 23 13:08:19 without all the crap i need to do for d-i May 23 13:08:24 Have you looked at live-helper? It does something very similar to that. May 23 13:08:31 And it does it from within d-i May 23 13:08:40 no, i havent yet, i will have to before A1 though May 23 13:08:43 And it can do it without bothering the user with lots of d-i prompting. May 23 13:09:05 You may want to: I believe it both represents that which you protest against and that which you wish existed, simultaneously. May 23 13:09:06 it bothers me as maintainer with the complexity i add through d-i May 23 13:09:35 i dont want to a) maintain d-i b) have d-i booted at all for these kind of images May 23 13:10:32 having to maintain a ton of d-i components vs a 10 line initramfs shell script is really not wnat i'm after May 23 13:12:22 Who said anything about "a ton of d-i components"? May 23 13:12:40 And d-i is just some initramfs scripts. May 23 13:13:48 So, for instance, if someone wrote partman-ac100, it could simply return, never prompting the user, and adding hints to everything else about how/where to install everything. May 23 13:13:56 So, for instance, if someone wrote partman-ac100, it could simply return, never prompting the user, and adding hints to everything else about how/where to install everything. May 23 13:14:05 d-i is a huge complex pile of stuff May 23 13:14:12 of which i dont need 99% May 23 13:15:24 Then don't put those bits in your image. May 23 13:15:34 It's *designed* to be flexible, modular, etc. May 23 13:16:34 i know May 23 13:16:39 you wont convince me May 23 13:17:38 I'm not trying to convince you. I'm trying to inform you that as a result of the change to live-helper, *every* image will become a d-i image. May 23 13:17:57 now that would be really bad May 23 13:18:01 And I'm trying to tell you that you needn't fear this change, and that it's not that different from what you do now. May 23 13:18:08 and i doubt thats true May 23 13:18:09 That's what the shift to live-helper means. May 23 13:18:23 you think we will drop casper ? May 23 13:18:27 No. May 23 13:18:31 see May 23 13:18:42 I think we'll wrap casper in a filesystem that we deliver with d-i through the live-helper support infrastructure. May 23 13:18:56 heh May 23 13:19:12 after we ported l-h to upstart indeed :) May 23 13:19:27 given that upstart will now run the whole initrd May 23 13:19:39 oh, and indeed all of d-i too May 23 13:20:17 sorry, but i dont see us changing the whole structure on one release May 23 13:20:30 Writing a single compatibility upstart job is trivial, so there's no blocking porting effort. May 23 13:20:40 we'll see May 23 13:20:49 Slowly moving more of the early boot stuff to upstart can happen over time. May 23 13:21:05 i expect that in function and form of casper or the live initrd there wont be many changes May 23 13:21:31 Really, go read about live-helper. Learn how it works. May 23 13:21:53 I think you'll be pleased as much as you're annoyed. May 23 13:22:17 * persia decides thrudhr has a nice ring to it May 23 13:22:21 i wont be pleased for sure :) May 23 13:27:03 how to change boot.scr without regenerating it by hand? panda May 23 13:29:06 .. use my awesome flash-kernel boot.script->boot.scr demangler :D May 23 13:29:23 hrw: or like this May 23 13:29:27 sudo mkimage -A arm -T script -C none -n "Ubuntu boot script" -d boot.script boot.scr May 23 13:29:30 guys does anyone have knowledge of a quick way to disable these horrible new scrollbars? May 23 13:30:11 Don't install overlay-scrollbar? May 23 13:30:18 both kinds.. gedit seems to have a kind of floating on the side scrollbar and other apps just have a kind of little orange line May 23 13:30:25 I want traditional ones back May 23 13:30:29 Neko: for efikas it uses /boot/boot.script May 23 13:30:36 Neko: as base of boot.scr May 23 13:30:38 I don't think I have a choice not to install overlay-scrollbar May 23 13:30:52 hrw, run flash-kernel again it should do it on all boards? May 23 13:30:57 Why not? The rdepends list is short. May 23 13:32:05 Neko: right.. May 23 13:32:32 but you want to do it "not by hand", you can't... you have to invoke it somehow with manual labor May 23 13:33:25 persia, I just removed the package and I still get orange-line scrollbars May 23 13:35:09 Hrm. That's unexpected. That package is supposed to be where the orange widget lives. May 23 13:35:35 If you want different defaults, use a different seed. May 23 13:35:57 the whole point is I am trying to replace the metapackages so we can basically define a behavior for our not-quite-ubuntu natty build May 23 13:36:08 except, if I take these out, the scrollbar thing still happens May 23 13:36:50 ahhh here we are May 23 13:36:59 you need to remove liboverlay-scrollbar-0.1-0 too May 23 13:37:07 Then I'm wrong. I thought it was that. Maybe grep for "scrollbar" in the natty-changes mbox to see what else got patched. May 23 13:37:16 Ah, you found it. Excellent. May 23 13:37:28 it's just reaaaaally difficult to hit those little orange lines with a netbook trackpad May 23 13:37:57 also since firefox and thunderbird are blacklisted we get different behavior per app which is just not desirable May 23 13:38:03 I wonder how this affects unity though May 23 13:38:09 Try it with one of the tiny little "optical pointers". It becomes essentially impossible. May 23 13:38:59 weurghhh May 23 13:39:12 saving a file in gedit defaults to a directory called "chrome" May 23 13:40:35 maybe because I am root.. but that's strange May 23 13:46:07 Is it not possible to remove X / completely from the ARM version ? Everything I try just prompts it to want to download another desktop setup ... May 23 13:46:51 ZeZu, It certainly is possible. I don't have X installed on my beagle. May 23 13:47:35 You will have to either 1) start from a minimal image (you might be interested in the natty headless developer images), or 2) spend a lot of time making sure you have every reverse dependency removed. May 23 13:52:01 I figured it would be something along those lines ... dependency mess , do you have the sgx drivers installed on your beagle ? May 23 13:53:39 I didn't have nearly as many issues w/ beagle as i do w/ panda ... the pvr kernel drivers refuse to cross compile w/ 2.6.39-rc7 May 23 13:55:05 if your .39 kernel is properly packaged they wont ;) May 23 13:56:12 ZeZu, I don't have the sgx drivers on my beagle (no monitor, no need) May 23 13:57:35 oh, wait you said cross ... May 23 13:57:41 indeed May 23 13:57:47 i fear there even a properly packaged kernel wont help May 23 13:58:11 i dont think dkms gets along well with cross building May 23 13:58:28 sounds like a good future project for hrw *g* May 23 13:59:00 yes well i found out the hard way ... actually you have to pass ARCH=arm or else it tries to build bits and pieces for host system ... May 23 13:59:46 define properly packaged btw, might help ... i do recall having to package the headers differently when building for beagle, but i don't think it used dkms at that time May 23 14:02:32 ogra_: D: May 23 16:18:31 hi , does any one have a u-boot for omap4 panda board, where we 'saveenv' works ? .. any help pls May 23 16:18:36 j /#panda May 23 16:19:30 avinashhm, Does it not work for the u-boot in the Ubuntu images? May 23 16:20:12 persia, sorry .. wasn't aware of the ubuntu images .. can u point me to .. i ll try May 23 16:21:52 Most recent images at http://cdimage.ubuntu.com/releases/11.04/release/ May 23 16:22:32 You can choose between a full environment (optimised for netbooks, but suitable for desktops), or a headless image as a base for arbitrary development. May 23 16:23:09 ogra_, Do you know why there aren't any .torrent files for the omap images? May 23 16:23:50 persia, thanks man .. i ll pick up these net book image and check out ... May 23 16:30:21 avinashhm: When you say "saveenv", are you trying to save the u-boot environment between reboots? The panda doesn't have any emmc or nand for storing u-boot settings. May 23 16:30:54 GrueMaster, can't we store to MMC .. just asking , i am not sure ? May 23 16:31:43 Not from uboot. But you can modify the /boot/boot.script in our images and reflash it to the boot partition with flash-kernel. May 23 16:32:14 (or modify it manually and reapply the u-boot crc with mkimage). May 23 16:32:20 heh. kernel segfault during debootstrap. Extra points! May 23 16:34:34 janimo: poke re chromium> any luck? May 23 16:48:06 GrueMaster, ok .. i ll try to modify boot.script then .. thanks man May 23 17:06:55 rsalveti: have the PXE boot patches landed in the Linaro u-boot tree? May 23 17:13:46 NCommander: not that I know, jcrigby should know better May 23 17:14:15 NCommander, no not yet May 23 17:14:38 jcrigby: any idea when it will land? May 23 17:15:57 NCommander, I could push them to the linaro-next if that is useful to you May 23 17:16:27 jcrigby: indeed it would be May 23 17:16:28 or are we talking pushed to archive? May 23 17:16:37 ok, I'll do that May 23 17:17:17 jcrigby: when it hits the linaro u-boot, it will get piced by the archive next time we update it May 23 17:17:27 jcrigby, For pushed-to-archive, if you can get a candidate package ready, it's easy to get someone else to help drop it in. May 23 17:20:34 NCommander, persia ok I'll prepare a 2011.05 release based on current upstream rc and the PXE patches May 23 17:20:59 jcrigby: thanks, that's a huge help May 23 17:23:57 jcrigby: it'd be good to include the tftp patches, but probably need more work at the ones that enables ehci for omap4 May 23 17:25:02 rsalveti, the patches don't break anything so we can include them for those that want to experiment? May 23 17:25:19 I know we have ehci support for omap3 already, so I believe we can also make it work with beagle xM May 23 17:25:35 rsalveti, yes that would be great May 23 17:25:37 jcrigby: sure, it's not touching anything outside usb, ehci and smsc driver May 23 17:25:50 guess we can also include them May 23 17:29:40 Thanks! That helps us play. May 23 17:53:36 micahg, none so far, but I did not spend time on it since the UDS May 23 17:57:12 janimo: k, thanks May 23 19:50:51 did anybody try webm in firefox in natty arm ? **** ENDING LOGGING AT Tue May 24 01:08:22 2011 **** BEGIN LOGGING AT Tue May 24 05:32:15 2011 May 24 08:39:50 morning May 24 10:51:16 (gst-plugin-scanner:1303): GStreamer-WARNING **: Failed to load plugin '/usr/lib/gstreamer-0.10/libgstffmpeg.so': /usr/lib/neon/vfp/libavformat.so.52: symbol __aeabi_d2lz, version LIBAVCODEC_52 not defined in file libavcodec.so.52 with link time reference May 24 10:51:27 someone got such? May 24 10:51:31 natty on panda May 24 10:51:33 \o/ How did you generate that? May 24 10:51:46 gst123 movie.divx.avi May 24 10:58:33 bug 787486 May 24 10:58:35 Launchpad bug 787486 in libav "/usr/lib/neon/vfp/libavformat.so.52: symbol __aeabi_d2lz, version LIBAVCODEC_52 not defined in file libavcodec.so.52 with link time reference" [Undecided,New] https://launchpad.net/bugs/787486 May 24 11:46:12 hey guys does anyone know how to get opengl running on a toshiba tegra 2 with ubuntu / xorg? May 24 12:59:51 persia, how is your ac100 adventure going btw ? May 24 14:14:00 I'm having trouble booting an OMAP4 panda board with 11.04 May 24 14:14:10 can anyone help me? May 24 14:15:52 nanomad: sure, post the question and hopefully someone will be able to answer it :-) May 24 14:16:18 we zcat the headleass image to the sd May 24 14:16:28 and we get no output from the serial console May 24 14:16:31 the validation FS works fine though May 24 14:19:15 nanomad: ubuntu does not use serial console too much May 24 14:19:37 well, you should see initial bootloader messages May 24 14:19:49 exactly May 24 14:19:54 but we get nothing May 24 14:20:01 once the kernel is up its all graphical May 24 14:20:05 and no leds blinking May 24 14:20:19 sounds like you messed up the SD somehow May 24 14:20:22 we attach the serial cable first and then then we give power to the board May 24 14:20:38 that's the third time i'm zeroing the SD May 24 14:20:47 any suggestions? May 24 14:20:49 how exactly did you write the image to it ? do you still have the cmdline in your history ? May 24 14:20:55 yes May 24 14:21:02 can you paste it here May 24 14:21:09 zcat ubuntu-11.04-preinstalled-headless-armel+omap.img.gz > /dev/sdd May 24 14:21:24 as root i assume May 24 14:21:31 yes May 24 14:21:39 and sdd is actually the right device ? May 24 14:21:51 we tried also the dd way May 24 14:21:54 of course, the validation FS from Texas boots fine May 24 14:22:21 are you 100% sure sdd is your SD card ? May 24 14:22:26 yes May 24 14:22:42 (ouch) May 24 14:22:49 if you re-plug the card after writing the image, does it get automounted ? May 24 14:22:55 yes May 24 14:23:04 weird May 24 14:23:16 that indicates the filesystems are ok May 24 14:23:28 i wouldnt know why it doesnt find the bootloader May 24 14:29:46 i'm trying with dd again May 24 14:31:40 nanomad: Reading the backscroll, it looks like you are using the omap image on a panda? You need the omap4 image. May 24 14:32:15 GrueMaster, cool ! nice catch ! May 24 14:32:16 hello May 24 14:32:26 zcat ubuntu-11.04-preinstalled-headless-armel+omap.img.gz > /dev/sdd Should be ubuntu-11.04-preinstalled-headless-armel+omap4.img.gz May 24 14:32:31 GrueMaster, and that *before* coffee ???!?!??!?!!! May 24 14:32:39 someone could tell me if is possible to install adobeflash on ubuntu arm? May 24 14:32:52 Well, I had my first sip seconds before responding. May 24 14:33:04 sha__, if you find a binary for your architecture May 24 14:33:16 usually the flash binaries are highly optimized for the HW May 24 14:33:34 so an omap flash would have probs to run on tegra for example May 24 14:33:42 toshiba ac100-10d :) May 24 14:33:52 yeah, indeed, i thought so :) May 24 14:34:09 where i could find that binary? May 24 14:34:15 i think phh had a solution, someone talked about a libflashplayer.so recently May 24 14:35:14 ogra_: i have the library for 32bit May 24 14:35:30 but i think it doesn work May 24 14:37:28 for 32bit ? May 24 14:39:33 what do you mean by that May 24 14:43:50 ogra_: not for ARM but for x86 (downloaded from adobe website) May 24 14:44:27 :( May 24 14:44:27 heh May 24 14:44:30 GrueMaster: thanks, i definitely need more coffee May 24 14:44:35 what do you plan to do with it on arm May 24 14:44:55 nothing good :) im looking for an arm version of that lib May 24 14:45:19 sha__: Let us know when you find one. :P May 24 14:45:25 sha__: which one? May 24 14:45:44 * nanomad is interested in flash for omap4 May 24 14:45:48 gildean: libflashplayer.so May 24 14:45:55 http://www.enst.fr/~husson/libflashplayer.so May 24 14:46:32 that's the arm version i included in the modified rootfs for the ac100 May 24 14:47:36 does it work on our boards? May 24 14:48:18 gildean: but i can't see video right? May 24 14:48:27 where is the right installation path? May 24 14:49:12 sha__: /usr/lib/mozilla/plugins May 24 14:54:01 nanomad: it's originally from the atrix ROM May 24 14:54:10 nanomad: firefox doenst recognize the plugin May 24 14:54:12 so no idea on which machines it works May 24 14:54:34 sha__: i think the path is /usr/lib/firefox/addons/plugins May 24 14:55:57 no, it's /usr/lib/firefox-addons/plugins/ May 24 14:56:06 you will likely need the full nvidia xserver and GL setup May 24 14:56:20 yeah, the overlay is needed May 24 14:56:26 * ogra_ bets it is linked against them) May 24 14:56:36 it surely is May 24 14:57:23 so that one only works with the .32 and the l4t overlay May 24 15:04:02 the roght folder is /usr/lib/mozilla/plugins/ May 24 15:04:09 but it doesn't work for me May 24 15:04:18 it's doesn't recognized May 24 15:06:24 sha__, do yuo run a .32 kernel and the L4T overlay FS so you have EGL support ? May 24 15:06:24 well, afaik atm the only way to have flash at least with the ac100 is to do as i did and have the l4t overlay and all that shit May 24 15:07:45 i run .29 May 24 15:08:43 Ok, we booted ubuntu correctly but the installer crashes near the end with DSPC timeout waiting for EVSYNC May 24 15:09:10 It shouldn't crash. Try hitting enter. May 24 15:09:21 That is a console message only. May 24 15:09:40 yea, we discovered that 3 seconds after i posted May 24 15:10:03 Ok. Happy hacking. :) May 24 15:10:20 We should have that error fixed in an upcoming kernel update. May 24 15:10:30 good May 24 15:11:07 does the GPU work with 11.04 or we are missing the drivers? May 24 15:11:43 I thought you wanted to run headless? May 24 15:12:00 yup, while we wait for our brand new hdtv May 24 15:12:26 If you want the GPU drivers, you should install the netbook image. It has an icon in unity to install them from ppa. May 24 15:12:37 thanks May 24 16:54:19 anyone using "claws mail"? May 24 16:54:32 it seems to corrupt attachments on save May 24 20:23:55 ogra_, It mostly works. There's still issues with the keymap, some complaints about where I stuck root (not mmkblk0p7), and it crashes at least once in a while. May 24 20:37:29 crashes how? May 24 20:43:36 marvin24, There seem to be main classes of kernel panic: when I use the wireless too much (e.g. download 1GB of data as several parallel streams), 2) Something random when I leave it alone too long, and 3) Something involving loss of access to MMC, which seems to happen randomly. May 24 20:44:59 I'll turn these into bugs at some point, but I'm never comfortable reporting bugs on something I can't reproduce, and system setup is a time when I have little patience for reproduction (excepting when I'm trying to sort some specific feature for installers) May 24 20:49:17 (note that the above is the result of playing with linux-ac100 on a Dynabook AZ, and does not reflect the results anyone might see in environments where the kernel and installation device are more closely aligned) May 24 20:49:41 thanks persia May 24 20:49:51 I cannot say much about 1) May 24 20:50:11 but 2 und 3 looks like some random decision of the device to go to sleep May 24 20:50:25 black screen, no chance to wake up May 24 20:50:37 or mmc sleeping, and not recovering May 24 20:50:59 Those aren't my symptoms though: I can always restore the screen by pressing the spacebar: I just am greeted with a kernel panic dump screen. May 24 20:51:17 urh May 24 20:51:28 I havn't seen any kernel panic for a long time May 24 20:51:44 maybe you can make a photo for me if it's fatal May 24 20:53:42 Now that the system is actually installed, it's been running more smoothly, but I'll admit that after spending time getting it installed, I've not been using it as much (although that's in part because I haven't been doing much computing away from my desk). May 24 20:54:41 But sure, if I get another one, I'll capture the output before restarting the device. Is there enough information from the screen dump to understand, or does one also need to know how one got to that state? May 24 20:55:37 problem is often that the useful info gets scrolled away May 24 20:55:42 too many tasks May 24 20:55:59 maybe you'll find something in the syslog May 24 21:00:53 [16:38:18] i think phh had a solution, someone talked about a libflashplayer.so recently <--- motorola atrix May 24 21:01:33 I don't see anything obvious (although wireless is fairly chatty). May 24 21:02:07 phh, Does that contain a redistributable blob, or is it just an example of something that works? May 24 21:02:29 persia: well it's the "full" libflashplayer.so May 24 21:02:35 I don't know how much it is linked to nvidia stuff May 24 21:02:42 i'd say it should be mostly independant May 24 21:03:47 I was asking more about licensing than compatibility. I would be surprised to see Adobe prepare device-specific flash solutions. May 24 21:08:43 To put that another way, if it's licensed for redistribution, I'll toss it in multiverse. If there is a known reliable download site, it should be easy to extend flashplugin-installer to pull from there. If it's not-for-distribution, that makes it trickier. May 24 21:11:10 ah. May 24 21:11:24 i'm quite sure i'm not even allowed to use it. May 24 21:11:58 Ah. That was my fear. Oh well. May 24 21:12:01 if they would like to distibute it, they would have put it on they main site (beta version) May 24 21:13:09 marvin24, From what I understand, Adobe doesn't especially want to distribute it. I've seen device vendors stick it behind registration sites before, for download and device recovery. I keep hoping someone who does that fails to have the registration step. May 24 21:14:14 I wonder if they ever make an arm plugin free for download May 24 21:14:34 maybe platforms are too different (even more than on x86) May 24 21:17:24 There are a few places that have free arm plugins working for download, and several people have reported that something downloaded for one device works on another (with the same instruction set: e.g. ARMv7a). The issue is more that the current download locations of which I am aware all require registration, which includes an assertion that one has the device in question, and I'm not going to put anything in flashplugin-installer that encourages May 24 21:17:25 users to lie. May 24 21:18:51 that's understandable May 24 21:20:02 the whole proprietary stuff is just ... crap May 24 21:24:36 I believe that free software is a better model, but complaining about stuff that doesn't work doesn't help: the better solution is to make something else work better (who uses Adobe's PDF readers for Linux these days?). May 24 21:27:46 with a little luck, html5 will eat flash May 24 21:28:10 I think even Adobe seems to accept that May 24 21:29:46 the main problem is not proprietary flash, but proprietary video codecs May 24 21:30:20 marvin24_DT: +1 May 24 21:30:36 BTW, is there a good set of instructoins ATM on how to best setup natty on an ac100? May 24 21:32:10 NCommander, I've just completed, and was planning to draft those. May 24 21:32:13 Here's the precis: May 24 21:32:25 NCommander: you can use my guide and just substitute the files with the .37 ones http://ac100.tunk.org/wiki/phh May 24 21:32:39 gildean: how well does it work? May 24 21:33:03 what, natty or my guide? May 24 21:33:29 natty May 24 21:33:29 Ugh. Now I can't find ogra's prebuilt files :( May 24 21:33:37 i haven't tried the latest .37 or the rootfs with it May 24 21:33:41 also, I don't have Android on my AC100 anymore, so no way to update to 2.2 May 24 21:34:14 NCommander: no need, you just need to flash parts 2 and 4 May 24 21:34:22 You don't need to run 2.2: I never launched Android on my Dynabook May 24 21:34:37 https://launchpad.net/~ac100/+archive/ppa is a PPA with kernel/settings/etc. May 24 21:34:45 and btw. my portal contains links for about everything you need here : http://ac100.tunk.org/ May 24 21:34:49 http://people.canonical.com/~ogra/tegra/2.6.37/ has some binary files. May 24 21:34:59 You need to get the nvflash utility from somewhere May 24 21:35:02 I take it works reasonable well? May 24 21:35:12 (nVidia isn't distributing it anymore). May 24 21:35:15 th .37 prebuilt is here : http://people.canonical.com/~ogra/tegra/2.6.37/ May 24 21:35:29 persia: that is also on my portal May 24 21:35:38 gildean, Cool! May 24 21:37:23 So what I did was 1) get the rootfs (non-SD), untar that to a freshly-formatted SD, 2) get the boot*SD.img, and nvflash it to partition 6, 3) insert the SD and boot, 4) use that target environment to hack around and build a system based on natty and the PPA. May 24 21:37:55 persia: how well does it work? My ac100 has performacne issues ... May 24 21:38:11 For the lazy, just untarring ubuntu-natty-netbook-2.6.37-1-ac100-rootfs.tgz onto a formatted partition on the MMC, adjusting /boot/bootimg.cfg, and running flash-kernel ought be the totality of the required hacking around. May 24 21:39:18 NCommander, it works. Attempting to type '|' results in '9', and one of my '\' keys doesn't work, but that won't affect you (you are unlikely to have as many keys on your keyboard). May 24 21:40:23 jpkeybd May 24 21:40:28 ftw May 24 21:40:59 jhobbs, This is a software package that intends to work around the kernel bug, or just agreement that more buttons makes a better keyboard? May 24 21:41:21 agreement =) May 24 21:54:57 gildean, Looking at http://salaliitto.com/~gildean/ac100/wiki/phh/ , I can't find a download link for nvflash. Am I blind, or is there another page I need? May 24 21:55:55 persia: check the front page of http://ac100.tunk.org it's under downloads May 24 21:56:42 from scoopr.fi? Excellent. Thanks! May 24 21:56:54 but the link is also on the guide May 24 21:57:23 so yeah, you were a bit blind ;D May 24 21:58:41 Ah, "... and tools from nvidia ...". Yes, I'm blind. **** ENDING LOGGING AT Wed May 25 02:59:57 2011