**** BEGIN LOGGING AT Sat Mar 25 10:59:57 2006 Mar 25 11:02:15 hi there, anyone knows when next release is out ? Mar 25 11:03:41 someone told me to wait to switch to openslug, because next release was right at the corner Mar 25 11:03:55 but this was a month ago...and nothing out yet.... Mar 25 11:08:34 yeah, yeah, I'm working on it ;) Mar 25 11:09:03 It was delayed to include the 2.6.16-kernel Mar 25 11:09:11 and there's a couple of issues that have to be sorted Mar 25 11:11:48 so, if I'm lucky, the testing will start later today Mar 25 11:12:12 and hopefully a release tomorrow (or the day after, since I'm proably gonna have a hangover tomorrow) Mar 25 11:25:14 <|kami|> hi can i set up a mailserver with something else then postfix? Mar 25 11:37:55 NAiL: thanks for the presicion Mar 25 11:38:03 i will wait until then :) Mar 25 11:49:35 rwhitby: I seem to recall you having opinions on how the leds should work on boot... care to remind me a bit? :) Mar 25 11:50:33 <|kami|> hi can someone please tell me why i can send emails with postfix but cant login to postfix with telnet? Mar 25 11:51:59 no idea, I never used postfix Mar 25 12:02:45 koen: could you look at this: http://pastebin.com/621497 Mar 25 12:02:52 I have no idea what goes wrong there... Mar 25 12:03:45 ehm Mar 25 12:03:48 I've seen that before Mar 25 12:03:52 *thinks* Mar 25 12:03:57 wrong path to pythin? Mar 25 12:04:01 python* Mar 25 12:04:08 what should the python path be? Mar 25 12:04:48 `which python` should tell you Mar 25 12:04:57 eddyp: `which python` ;) Mar 25 12:05:19 it could be something more subtle Mar 25 12:05:32 *light goes on* Mar 25 12:05:34 /usr/bin/python Mar 25 12:05:41 can you exec stuff from the disk? Mar 25 12:05:58 mounted with noexec? Mar 25 12:06:03 it could be that it's mounted with the 'user' flag, with implies noexec Mar 25 12:06:07 let me try Mar 25 12:06:10 ah, oh Mar 25 12:06:32 anyway. Anyone that wanna test the latest leds script should check it out of svn Mar 25 12:06:46 * koen is searching for a network cable Mar 25 12:07:05 oops Mar 25 12:07:08 noexec Mar 25 12:07:18 hehe Mar 25 12:07:49 fuck, I assumed that +x would mean is executable Mar 25 12:08:17 bah, leds script available at http://osx.kynisk.com/leds Mar 25 12:08:43 I've done some untested changes. If anyone has a slug and can test, I'd be grateful :) Mar 25 12:21:37 NAiL: so, now it makes the kernel for open slug, but after it compiles it, I would like to make my modified kernel Mar 25 12:21:49 ? Mar 25 12:22:00 how do Ido that? Mar 25 12:41:18 then you go into tmp/work/ixp4xx-kernel/linux-2.6.16/, "ARCH=arm make menuconfig", then copy .config to openembedded/packages/linux/ixp4xx-kernel/2.6.16/defconfig, and re-make the kernel. Mar 25 12:48:17 this info should be on the wiki Mar 25 12:48:47 Good idea. Put it there :D Mar 25 15:31:32 eFfeM-log: did you already see http://www.gnu.org/software/gmediaserver/ ? Mar 25 15:58:59 where should I put this info on the wiki? a new page? some existing page? Mar 25 15:59:06 NAiL: Mar 25 16:58:11 eddyp: put it on the master makefile page Mar 25 16:58:23 just make a note on how to recompile just the kernel Mar 25 16:58:42 NAiL: ok Mar 25 17:01:34 not http://www.nslu2-linux.org/wiki/OpenSlug/CustomisingTheBuild ? Mar 25 17:01:57 hmm Mar 25 17:02:01 possibly Mar 25 17:02:06 the wiki is too large :-P Mar 25 17:04:07 The wiki is often an example of how too much information can become more confusing than too little information.... Mar 25 17:17:46 I can't find the file IPL_ixp400AccessLibrary-2_1.zip, just BSD_ixp400AccessLibrary-2_1.zip. is that the correct one and the wiki is wrong, or should I look for it more? Mar 25 17:18:39 look harder ;) Mar 25 17:19:10 pull harder! Mar 25 17:20:11 err, pulling my finger, aren't you? Mar 25 17:20:24 not really Mar 25 17:20:30 it's not the correct library Mar 25 17:20:48 the zipfile is somewhere on the intel site though Mar 25 17:20:58 intel's site sucks Mar 25 17:23:15 true Mar 25 17:23:23 as most large corporate sites Mar 25 17:25:44 the .bb should have a link to that Mar 25 17:34:15 man, 2.0 version is there, but for 2.1 the only file I can find is BSD.... Mar 25 17:34:33 can somebody check this? Mar 25 17:34:49 what's the link? Mar 25 17:36:27 http://www.intel.com/design/network/products/npfamily/ixp425swr1.htm Mar 25 17:36:33 login Mar 25 17:36:53 I meant, or do you want the one after login? Mar 25 17:42:02 heh Mar 25 17:42:07 looks like they've changed the name Mar 25 17:42:13 (and released v2.1.1) Mar 25 17:42:34 I see that... Mar 25 17:42:41 but will it work? Mar 25 17:43:10 might work if you rename it, not sure Mar 25 17:44:30 damn Mar 25 17:45:11 if not we'll have to work around it Mar 25 17:45:43 NAiL: the module built for a kernel without this module, will it be compatible with the old one? Mar 25 17:46:10 on second tought, I think the kernel has changed, hasn't it? Mar 25 17:47:14 damn, I am tired of trying to set the NSLU as a router on my net and this is keeping me from doing real work Mar 25 17:57:57 * eddyp made a note on the wiki about the filename change Mar 25 18:07:25 NAiL: the pppoe option is checked as module in the kernel configuration for 2.6.16, but I don't see any pppoe module package for pppoe for the openslug. Was this added recently? Mar 25 18:07:43 yes, and IIRC I haven't rebuilt the feed Mar 25 18:07:55 I'm a bit busy ;) Mar 25 18:08:08 Things are moving fast these days Mar 25 18:09:50 NAiL: sorry for keeping you; one more question, 'though: should I have seen an error for ixp4xx driver if the IPL* files were not downloaded during the kernel build? Mar 25 18:10:49 afterwards, when it builds the ixp stuff Mar 25 18:12:24 that means during the openslug-image build? Mar 25 18:12:36 I did a "make kernel" before Mar 25 18:13:27 yeah Mar 25 18:14:43 cool Mar 25 18:15:11 I have seen the pppoe ipk, but that is for 2.6.16 :( Mar 25 18:15:40 and I have 2.6.12.2 on my slug Mar 25 18:18:45 is /tmp ever automatically erased? Mar 25 18:19:07 no Mar 25 18:20:35 good Mar 25 18:21:19 so is there a way to compile the pppoe module ipk for the older 2.6.12.2 kernel? Mar 25 18:53:13 can anyone explain to me what the "devio" statements are for in the kernel Makefile?? Mar 25 18:54:25 in cvs? Mar 25 18:55:00 it prepends a few opcodes to set the kernel in LE mode (if a LE kernel), and pass the correct mach_id Mar 25 18:55:25 basically a hack to avoid replacing the bootloader, while making a unified kernel possible Mar 25 18:56:03 OK, so if I make a new platform, if my bootloaders sets this OK, I do not need to add these hacks. Mar 25 18:56:33 yeah Mar 25 18:56:45 and if the kernel is the same endianness as the CPU when it executes Mar 25 18:57:05 NAiL: Understood. Thanks for the explanation! Mar 25 18:57:07 I assume you've done all the machine_is_() stuff Mar 25 18:57:10 np Mar 25 18:57:44 I have an existing set of 2.4.x patches for my platform and am bringing them upwards to 2.6.x Mar 25 18:58:11 ah Mar 25 18:58:23 So, yeah, I have done this earlier under 2.4.x, but some stuff has moved around and been modularized (Deepak's work I guess). Mar 25 18:59:56 NAiL: you Norwegian? Mar 25 19:00:03 yup Mar 25 19:00:57 why? Mar 25 19:01:07 I *thought* I recognized your name from the guy doing zylin.com (CDT, eCos, Eclipse stuff) but it's only you guys have the same first name. Mar 25 19:01:33 ah Mar 25 19:02:33 NAiL: Does the NSLU use its PCI bus? Mar 25 19:03:12 hmmm, I *think* so Mar 25 19:03:22 for the usb controller, I think Mar 25 19:03:45 Not entirely sure about it though Mar 25 19:03:58 yeah, it does Mar 25 19:04:55 Our platform has to work around the second PCI bug in the IXP4xx chips, but it seems no existing platforms exhibit this bug. Mar 25 19:05:07 second PCI bug? Mar 25 19:06:24 IXP4xx in PCI Master mode does not respect less then 4 byte lane read OR writes (cannot remember read or write) Mar 25 19:07:42 This bug: http://www.intel.com/design/network/applnots/300375.htm Mar 25 19:08:17 It's a PCI non-compliance bug for less then 32-bit reads (writes?) Mar 25 19:08:29 hmm Mar 25 19:08:50 So it affects only target devices needing these accesses. (And ours does...) Mar 25 19:08:55 Can't say I've heard about any such problems on the slug or any of the other machs we do work on. Mar 25 19:09:09 The workaround costs as much PCB space as the IXP4xx itself... :-/ Mar 25 19:09:34 Nah, most devices are 32-bit only. Ours has some 16-bit registers. Mar 25 19:10:49 OK, thanks for the info. I guess I first checked if the workaround was available in the 2.6.x series, but it is not. Besides, it requires additional logic hardware. Mar 25 19:12:40 NAiL: you work on the NSLU kernel as well? Mar 25 19:13:22 not much Mar 25 19:13:37 I mainly work on the DS101 kernel, which is quite similar to the NSLU2 Mar 25 19:14:01 same CPU, but it has an IDE-interface, DoC flash and 64MB ram Mar 25 19:15:16 OK, I noticed two things: the IPL_ixp400AccessLibrary-2_1.zip is now dsitributed as BSD_ixp400AccessLibrary-2_1.zip by Intel, thus seemingly a BSD license. Mar 25 19:15:21 Also, 2.1.1 is available. Mar 25 19:15:33 http://www.intel.com/design/network/swsup/ixp400-2_1_1-relnotes.pdf Mar 25 19:15:39 "the ixp42x A0 stepping is no longer supported... this software release does not support little-endian" Mar 25 19:15:53 2.1 did? Mar 25 19:15:56 2.1.1 Mar 25 19:16:12 found out a couple of hours ago Mar 25 19:16:14 Yes, but 2.1 *did* support A0 and LE ? Mar 25 19:16:19 yeah Mar 25 19:16:36 OK, good to know. I have to refresh my memories about this stuff :-) Mar 26 09:37:53 hi, can someone with write access on nudi remove the cache dir, so the build goes through (as the cache fmt has changed we get errors and removing the cache fixes this) Mar 26 09:51:08 looks like the people responsible for the buildfarm don't read oe@ :( Mar 26 09:51:33 yup **** ENDING LOGGING AT Sun Mar 26 10:59:56 2006