**** BEGIN LOGGING AT Fri Jul 02 02:59:58 2010 Jul 02 04:30:56 * mozzwald is away: sleeping Jul 02 10:23:48 Desperate times call for desperate measures! Jul 02 11:31:34 berco: xem3 copies sur .../xem3. Y'en a un pour chaque core Ducati Jul 02 11:31:56 oops Jul 02 11:41:58 ogra: cool ! I just noticed you pushed the 10.10 images for OMAP! Jul 02 12:19:56 ndec, yeah, working on the omap4 ones now, hoping to have them ready by mid next week (there are some issues with the kernel package naming i need to have sorted first) Jul 02 12:20:35 ogra_cmpc: ok. do you have instructions to use these new type of pre installed images? Jul 02 12:21:20 ndec, bunzip and dd, boot, be patient :) Jul 02 12:22:06 ogra_cmpc: this looks quite complex ;-) Jul 02 12:22:16 hehe Jul 02 12:22:29 ogra_cmpc: do you have the offsets if I want to mount loop to instpect the content? Jul 02 12:22:32 We should develop some GUI for ndec! Jul 02 12:22:40 lool: yes, please Jul 02 12:23:29 lool: that said I would love to be able to create my own images as well... Jul 02 12:23:31 ndec, not yet, noted on my todo now though, the rootfs starts at 72M Jul 02 12:23:51 ogra_cmpc: thx Jul 02 12:24:32 livecd-rootfs is the tool we use for creating the rootfs, for the rest i can put a script online soon Jul 02 12:26:36 ogra_cmpc: it creates preinstalled images? is that easy to use? how long does it take to create a UNE image? Jul 02 12:27:51 livecd-rootf creates ext2/3 rootfs images and spits out a kernel and initrd not the partitioned image Jul 02 12:28:12 for that we have additional wrapper scripts (the one i mentioned above) Jul 02 12:29:48 currently these scripts live in debian-cd but thats very complex to set up, i'll extract them and merge them into a single script yuo can feed the files from livecd-rootfs to Jul 02 12:30:38 creating the rootfs on a imx51 babbage board takes about 2-2.5h for netbook atm Jul 02 12:31:10 i suspect you will get that down to 1.5h on a panda :) Jul 02 12:34:35 * lag_ has his internet-net in tow and it going to bag some cloud (going to buy an internet dongle) Jul 02 12:42:33 ogra_cmpc: thanks for the details. that would be really nice to get this. Jul 02 14:23:14 https://bugs.edge.launchpad.net/jockey/+bug/271288 Jul 02 14:23:15 Launchpad bug 271288 in jockey (Ubuntu Intrepid) (and 2 other projects) "Require the user to confirm the license before downloading a driver if it is non-free or if it has patent issues (heat: 4)" [Wishlist,Won't fix] Jul 02 15:04:36 mpoirier: Oh hey there Jul 02 15:04:57 lool; hello ! Jul 02 15:05:08 mpoirier: Would you mind helping me out on the SDHC timing issue; I wanted to confirm whether there is a fix for it pending, candidate patch or anything Jul 02 15:05:29 lool: completely in the dark still... Jul 02 15:05:32 lool, turn off preemtion Jul 02 15:05:33 mpoirier: If not, I'd love confirming which exact CONFIGs I should turn off from the linux source package to workaround Jul 02 15:06:17 lool: CONFIG_CPU_VOLUNTARY needs to be turned off. Jul 02 15:06:29 mpoirier: Ok; thanks Jul 02 15:06:32 lool: CONFIG_CPU_PREEMPT turned on. Jul 02 15:06:45 I think I'll send my buggy card to india. Jul 02 15:06:55 TI folks that is. Jul 02 15:07:08 they said they'd be willing to look at it. Jul 02 15:07:44 lool: at this point an interrupt doesn't seem to be coming up after a transfer. Jul 02 15:08:11 hence the call back function not called, leading to a time out on the transfer. Jul 02 15:08:37 I have 2 exact same cards, one works, the other one doesn't. Jul 02 15:13:15 mpoirier: Thanks a lot Jul 02 15:13:31 lool: by the way, Jul 02 15:13:53 if you find instances where this work around doesn't work, please tell me. Jul 02 15:14:00 mpoirier: I have a relatively good card, SanDisk Extreme III; I dont know whether speed makes that more likley Jul 02 15:14:18 lool: it's a wierd one... Jul 02 15:14:28 mpoirier: Ok; I'll try that out; right now my xcompiler is broken and I cant boot the board anymore, I'll rebuild a kernel as soon as I have a cross-compiler working Jul 02 15:14:39 we haven't entirely narrowed down type of card and speed, Jul 02 15:14:49 if that makes a difference at all... Jul 02 15:14:53 Ok Jul 02 15:15:13 I've seen a lot of failures on a variety of cards... Jul 02 15:15:35 again, if you still get a failure with PREEMPT_NONE, I need to know. Jul 02 15:16:50 mpoirier, btw, another prob arose yesterday, seems the NAND driver got lost between lucid and maverick, i guess there is a config option missing Jul 02 15:17:26 lool: by the way, I just looked in the config file. It is CONFIG_PREEMPT_VOLUNTARY and CONFIG_PREEMPT_NONE Jul 02 15:17:40 Yes, I noticed that. Jul 02 15:17:54 the NAND flash I mean... Jul 02 15:18:03 can we get that back ? :) Jul 02 15:18:11 we will need to yes. Jul 02 15:18:19 I'll investigate and open a bug if need be. Jul 02 15:18:21 else upgrades from lucid installs will break for C4 users Jul 02 15:18:35 in my testing, class 4 cards seem to work best, although I only have a handful of cards. Class 2 fail consistently (2 SD 4G cards tested), class 6 fails 70% of the time (1 4G). Both of my Kingston Class 4 cards work fine (8G &16G). Jul 02 15:18:53 fresh maverick installs wont use NAND but for lucid users NAND is essential Jul 02 15:20:33 ogra: nec was talking about generic udev rules Jul 02 15:20:42 Are they in the filesystem? Jul 02 15:21:40 lag_, yes, /lib/udev/rules.d/ ... they need somthign like /lib/udev/rules.d/61-gnome-bluetooth-rfkill.rules for their device that allows access for all locally logged in users Jul 02 15:21:59 ogra: I just found them *embarrassed smiley* Jul 02 15:22:00 lag_, its all about access to the device Jul 02 15:23:13 Do you know where this 'alias' is specified? Jul 02 15:30:33 lag_, alias ? Jul 02 15:32:16 ogra: nec mentioned that if I used MODULE_ALIAS macro there was a rule in udev that would pick up syslink's uevent Jul 02 15:32:42 lag_, well, if you use the KERNEL parameter it will just apply ot the module Jul 02 15:33:10 lag_, man udev btw Jul 02 15:33:54 lag_: Yes, I discussed this with mpoirier already Jul 02 15:34:02 lag_: It seems your re-doing this research Jul 02 15:34:17 lag_: /lib/udev/rules.d/80-drivers.rules, DRIVER!="?*", ENV{MODALIAS}=="?*", RUN+="/sbin/modprobe -b $env{MODALIAS}" Jul 02 15:34:32 KERNEL=="syslink*", ENV{ACL_MANAGE}="1" Jul 02 15:34:40 Which means that if there's a DRIVER and a MODALIAS, that triggers a modprobe on the MODALIAS Jul 02 15:34:43 lool, urgh Jul 02 15:34:52 ogra: "urgh"? Jul 02 15:34:54 lool, you defiantely dont need the modprobe stuff Jul 02 15:35:17 the driver gets already loaded by the platform uevent trigger Jul 02 15:35:25 ogra: That's what I'm speaking of Jul 02 15:35:39 loading the driver by the platform uevent trigger Jul 02 15:35:59 That's the only reason I can think of someone adding a MODULE_ALIAS Jul 02 15:36:08 lool, i thought that doesnt need a rule since its done inside udevd if there is a paltform event Jul 02 15:37:00 i.e. doesnt need to run scripts to speed up Jul 02 15:37:40 ogra: Really, which rule is that? Jul 02 15:38:00 dunno, thats how Keybuk explained it to me once Jul 02 15:38:33 ogra: I'm pretty sure this is the rule used Jul 02 15:39:12 well, i was convinced platfrom drivers dont need a rule at all and udevd has that bit builtin, but i might have understood that wrongly Jul 02 15:39:38 ogra: Are you sure of this, or could you look it up? Jul 02 15:40:05 i'm not sure of it, thus my comment above :) Jul 02 15:40:25 indeed the driver rule will match anyway even if i'm wrong Jul 02 15:41:28 ogra: grep-ing the udev source for modprobe and insmod, I see no direct loading; it's only done via rules, and the only place is from rules/rules.d/80-drivers.rules, so I'm pretty sure I pointed at the right spot Jul 02 15:41:45 lool, ok Jul 02 15:41:47 Anyway, so much for trying to help out :-) Jul 02 15:41:50 * lool leaves for WE Jul 02 15:42:03 lool, might be that i mixed it up with devtmpfs stuff Jul 02 15:42:08 enjoy your WE :) Jul 02 15:54:45 lool: Well the drivers aren't loading Jul 02 15:56:04 lool: http://paste.ubuntu.com/458378/ Jul 02 15:56:47 ogra: See above paste - I am using KERNEL Jul 02 15:57:23 you shouldnt add a rule for loading, as lool said, the 80-drivers.rules should just pick it up on booting Jul 02 15:58:00 ogra: I had no intention Jul 02 15:58:16 Did you see the above paste? Jul 02 15:58:27 yes Jul 02 15:58:38 It doesn't use DRIVER Jul 02 15:58:44 It uses KERNEL and UDEV Jul 02 16:05:33 it uses MODALIAS Jul 02 16:07:51 They all use that though don't they? Jul 02 16:08:30 The udev rule that I was pointed to uses DRIVER doesn't it? Jul 02 16:11:22 Help me udev geeks :) Jul 02 16:11:50 I've raised 2 uevents for you Jul 02 16:11:55 you have modalias so udev should load it Jul 02 16:12:05 But it doesn't Jul 02 16:12:17 BB or panda? Jul 02 16:12:17 So where do I start looking for a solution? Jul 02 16:12:27 Panda Jul 02 16:14:26 But the system is common throughout isn't it Jul 02 16:14:44 I raised the two uevents seen in the paste above Jul 02 16:14:58 But udev doesn't load the module(s) Jul 02 16:15:01 What gives? Jul 02 18:28:28 lag_: hey Jul 02 18:28:50 lool: Good evening Jul 02 18:29:04 lag_: So the rule only works when DRIVER and MODALIAS arent empty Jul 02 18:29:26 I don't have a DRIVER variable Jul 02 18:29:37 It has KERNEL instead Jul 02 18:30:15 lag_: So perhaps you need to extend the event to carry the relevant information Jul 02 18:30:38 Yes, I think I know what I have to do Jul 02 18:30:54 lag_: If you prefer keeping the current event, you'll have to submit new rules to udev Jul 02 18:30:59 lag_: Ok cool Jul 02 18:31:19 lool: Who controls the rules? Jul 02 18:31:24 Where do they come from? Jul 02 18:31:24 lag_: Is there anything left open for discussion or where you were looking for input? Sorry, I'm not sure I followed the status Jul 02 18:31:31 lag_: The rules are from udev upstream Jul 02 18:31:41 Okay Jul 02 18:31:53 So, I'm guessing I must conform to them Jul 02 18:32:14 lag_: It's a quite specific case that you have here, so it might make sense to handle them specially, but it might be more efficient or better to try to fit into the expected scheme Jul 02 18:36:22 lool: Do we (Ubuntu) have any entries in udev? Or are they all standard? Jul 02 18:38:13 lag_: We add some, but ideally only upstreamable stuff, so you should check wiht upstream if in doubt Jul 02 18:40:07 I don't think this project is upstreamable Jul 02 18:43:02 Okay, I'm going to try and do it the hard way first Jul 02 18:43:06 Thanks lool Jul 02 18:43:26 Right, it's nearly 8 o'clock here - I'm offski **** ENDING LOGGING AT Sat Jul 03 02:59:56 2010