**** BEGIN LOGGING AT Mon Apr 17 09:59:56 2006 Apr 17 11:18:06 Hi, I'm trying to build OpenSlug, but it fails with 'gzip: stdin: decompression OK, trailing garbage ignored' on this file: http://sources.nslu2-linux.org/sources/ports_sources.redhat.com__20050627.tar.gz Apr 17 11:18:22 It seems the file really is broken (tried using tar tvzf on the file after retrieving it manually). Any ideas to what I could try? Apr 17 12:39:57 morning everyone Apr 17 12:41:02 hey eFfeM Apr 17 12:44:03 hi koen, how are things today in openslug? Apr 17 12:44:20 eFfeM: quiet Apr 17 12:44:28 my slug is disconnected at the moment Apr 17 12:44:42 have the usb module problems already been solved? Apr 17 12:45:47 haven't had time to work on the slug Apr 17 12:46:03 have been locked up in a hotel in blr for a day due to the riots Apr 17 12:46:52 ouch Apr 17 12:47:12 actually it was on the dutch news Apr 17 12:47:18 (not me) Apr 17 12:50:19 actually i heard some UK companies ran into trouble as their BLR-based call centers closed Apr 17 13:19:50 hmmph. Finally got kernel compiled. Apr 17 13:20:05 Unfortunately kernel which I cross-compiled on a PC, won't boot. Apr 17 13:20:33 and kernel that I compiled natively on slug, is too big according to upslug2 Apr 17 13:31:29 hmm.. Apr 17 13:31:47 propably because scsi can scsi disk are enabled only as modules :D Apr 17 13:32:38 and usb mass storage.. Apr 17 13:33:48 Hi, I'm trying to build OpenSlug, but it fails with 'gzip: stdin: decompression OK, trailing garbage ignored' on this file: http://sources.nslu2-linux.org/sources/ports_sources.redhat.com__20050627.tar.gz Apr 17 13:34:13 It seems the file really is broken (tried using tar tvzf on the file after retrieving it manually). Any ideas to what I could try? Apr 17 13:35:10 and scsi generic is also module.. Apr 17 13:35:49 I got no ideas, but I never tried to compile openslug.. Apr 17 13:38:29 jake, thanks Apr 17 13:38:43 pfavr: for what? Apr 17 13:43:28 jake: for the reply :-) Apr 17 13:44:14 :) Apr 17 13:46:01 re-cross compiling kernel now with scsi+scsi disk+scsi generic+usb storage in kernel (not as modules, wish me luck) Apr 17 14:15:04 just my luck. Kernel I compiled was too big.. :/ Apr 17 14:48:22 Turned some features to modules now and turned some stuff off that I think isn't that necessary.. Apr 17 14:48:36 so trying now if kernel is small enough for nslu2.. Apr 17 14:50:17 jake: are you compiling the kernel inside openembedded tree, or on the nslu2? Apr 17 14:51:51 pfavr: none Apr 17 14:52:35 I don't use openslug.. Apr 17 14:52:38 argh!!! Apr 17 14:52:42 I got troubles! Apr 17 14:52:52 I can't get my slug to go to upgrade mode!!! Apr 17 14:54:36 yes Apr 17 14:54:38 managed Apr 17 14:54:48 hold reset - turn on power - wait 10 seconds - release reset Apr 17 14:54:54 good :-) Apr 17 14:55:18 yeah, did that and it failed about 5 times before succes.. Apr 17 14:55:40 ready/status started blinking yellow or device shut itself down.. Apr 17 14:55:52 but still no luck, kernel too small or big.. Apr 17 15:08:07 jake, why are you hanging out in #openslug if you're not using openslug? Apr 17 15:10:46 eh, there kinda ain't gentooSLUG channel, or if it is it's not that active :) Apr 17 15:11:07 §second, I fetched sources for openslug's kernel and patches.. Apr 17 15:11:17 and changed that to suite gentooSLUG.. Apr 17 15:13:42 :-) Apr 17 15:17:09 and I'd propably use openslug instead of gentooslug, but I'm keen to using systems that work the same way, currently I already need use windows(xp),linux used by dreambox&dbox2, openwrt, so I run gentoo on my X clients and at my server, I'd really like the idea of using same system with slug too. Jumping to another device, but everything's same :) Apr 17 15:17:51 FWIW, the dreambox is also built by OE Apr 17 15:18:19 koen, yeah, but rootfs on dreambox+dbox2 running enigma is read-only.. Apr 17 15:18:32 so I only can manipulate under /var :) Apr 17 15:26:08 openwrt is OE? Apr 17 15:27:28 koen: you experience with building openslug? Apr 17 15:35:23 gotta go Apr 17 15:45:44 damn. Still too big kernel.. :( Apr 17 16:06:55 just booting 2.6.16 kernel on a gentoo slug :) Apr 17 16:10:33 my normal luck :) Apr 17 16:10:37 it didn't get ip.. Apr 17 19:37:55 anyone awake? Apr 17 19:38:19 * ByronT rubs his eyes.... Apr 17 19:38:45 I've spend most of today trying to build openslug from source Apr 17 19:38:49 without success Apr 17 19:39:42 I'm not doing anything advanced - just getting OpenSlug-2.7-beta-source.tar.gz and typing make image Apr 17 19:40:14 then I tried pulling the trunk from subversion and I get the same error Apr 17 19:40:21 (almost) Apr 17 19:41:12 after fetching ports_sources.redhat.com__20050627.tar.gz it fails while untar'ing it: gzip: stdin: decompression OK, trailing garbage ignored Apr 17 19:42:18 that's about it Apr 17 19:43:24 try grabbing it from the berlios repo - linked via slug-firmware.net Apr 17 19:45:28 thats what I just did - or do you mean grabbing only ports_sources.redhat.com__20050627.tar.gz? Apr 17 19:51:25 it is one of the files referred to by openembedded which is bad... Apr 17 19:52:40 pfavr: we replaced the file upstream, and wget tried to resume it when downloading Apr 17 19:52:49 so it corrupted your file on disk Apr 17 19:52:58 remove the file and download again Apr 17 19:53:45 koen: thanks! Apr 17 19:54:23 (just removed the whole dir and untar'ed OpenSlug-2.7-beta-source.tar.gz again Apr 17 20:03:29 koen: I still get the error when trying to build from subversion trunk Apr 17 20:11:13 and I still get the error when building OpenSlug-2.7-beta-source.tar.gz Apr 17 20:13:36 I get the NOTE: Fetched libc_sources.redhat.com__20050627.tar.gz from tarball stash, skipping checkout Apr 17 20:13:55 followed by NOTE: fetch http://sources.nslu2-linux.org/sources/ports_sources.redhat.com__20050627.tar.gz Apr 17 20:16:49 pfavr: is this an error when untaring the file? Apr 17 20:16:57 yes! Apr 17 20:17:13 and if I get it using wget I still get the error Apr 17 20:17:27 (using e.g. tar tvzf) Apr 17 20:17:37 so the file is damaged somehow (on the server) Apr 17 20:17:51 pfavr: you can untar the file and retar then generate a new md5sum, then put those in your downlaod directory. Apr 17 20:18:09 pfavr: that will fix the problem. Apr 17 20:18:32 but if gzip reports that the file is corrupt... won't I miss some files then? Apr 17 20:19:20 I have done this it worked for me, and others have suggested that here. Apr 17 20:19:30 mproctor: ok, thanks Apr 17 20:19:51 * pfavr wonder why it isn't fixed on the server instead Apr 17 20:20:33 I think it is a problem at the origninal source the nslu2 is just a mirror. Apr 17 20:20:49 * mwester wonders if he should tel pfavr that we don't have access to all the servers... :) Apr 17 20:21:04 :-) Apr 17 20:21:37 where do I put the md5sum btw? Apr 17 20:22:07 is there another way - e.g. making it pull the ports files from cvs? Apr 17 20:22:23 (if anyone happens to know :-) Apr 17 20:22:46 * mwester shudders at the thought of all that instability... Apr 17 20:23:31 Yes, it is possible. Each package definition in the build describes the source, and it *can* call for fetching from cvs or svn. Apr 17 20:23:42 maybe with a -r20050627 Apr 17 20:24:28 guess it'll be fixed soon snyway and that is why nobody bothers fixing it in the archive Apr 17 20:24:36 ;-) Apr 17 20:24:40 In fact such an event delayed the Unslung release - the problem being that if something is built from the head of an svn repo, it's not necessarily very reproducable. So it's more stable to "lock" the build to a tarball. Apr 17 20:25:31 It might be that a new tarball has been put in place at all the archives, but if it has a new name, then someone has to update the package info to pull the new name. Apr 17 20:25:33 but then the tarball should work I think Apr 17 20:25:44 and directory listing is denied Apr 17 20:25:45 One would think. Apr 17 20:26:07 who is in control of that server? Apr 17 20:26:23 (anyone tried to alert them to the problem?) Apr 17 20:26:36 I would think one would download and un-tar it to make sure it was good, so perhaps there is a Linux distro out there that actually thinks that file is good... Apr 17 20:28:00 Don't know if anyone has been alerted. It's possible that this is just what the 2.7 snapshot builds, and the problem is already fixed in the monotone repo (which version are you building?) Apr 17 20:28:35 I've tried the 2.7 beta and the "trunk" from subversion Apr 17 20:28:40 same result Apr 17 20:28:45 the problem is tar -cj files > file.tar.bz2 instead of tar -c files | bzip2 - > file.tar.bz2 Apr 17 20:28:48 (different glibcs though) Apr 17 20:28:52 file contents are fine. Apr 17 20:29:18 it is gzip Apr 17 20:29:42 gzip instead of bzip2 then but same problem. Apr 17 20:29:42 gzip: stdin: decompression OK, trailing garbage ignored Apr 17 20:29:43 mproctor's solution (untar, then tar back up and generate a new md5sum) works; others have used that just fine. Apr 17 20:30:00 mwester: thanks, I'm trying to do it :-) Apr 17 20:30:20 (and really happy about this tip - guess I should have read the archives) Apr 17 20:31:06 I just have run into the problem with files I created. Apr 17 20:31:35 but I've never had the problem - I believed tar czf and tar cf; gzip would do the same thing Apr 17 20:32:04 stupid question: where do I put the checksum? Apr 17 20:32:36 ah Apr 17 20:32:38 sorry Apr 17 20:32:42 downloads dir with the tars Apr 17 20:33:05 yup Apr 17 20:35:19 while I'm building I dare asking another question: what is the easiest way to build (crosscompile) a simple c-program for the openslug? Apr 17 20:35:57 (I guess I should add a new package to openembedded - but is there a skeleton for doing that?) Apr 17 20:36:07 3500 skeletons ;) Apr 17 20:36:16 you could do 'bitbake devshell' Apr 17 20:36:36 that should give you a file which you can source to setup all the env vars Apr 17 20:36:49 and do arm-linux-gcc foo.c -o foo or something like that Apr 17 20:37:16 thanks! Apr 17 20:38:03 uhm! now it builds, really nice. Thanks for the help! Apr 17 20:38:26 what is the most recent kernel tried on the nslu2? Apr 17 20:39:41 2.6.16 and 2.6.17-git iirc Apr 17 20:39:50 nice Apr 17 20:40:09 I need to talk to a usb-serial thingy which is only in 2.6.13 and later Apr 17 20:42:26 koen: bitbake devshell yields: ERROR: Nothing provides devshell Apr 17 20:42:41 path Apr 17 20:42:52 ups, sorry, wrong window Apr 17 20:43:29 D'oh Apr 17 20:43:36 I guess it isn't in openslug-packages Apr 17 20:43:44 or whatever it's called nowadays Apr 17 20:47:41 not sure if this question is appropriate: how is openslug related to the other nslu2 projects, and which is the most recent/bigger/supported/developed? Apr 17 20:48:02 It looks like most projects use openembedded? Apr 17 20:48:26 (dunno where I got that idea from) Apr 17 20:50:59 pfavr: have you read the wiki? Apr 17 20:51:09 some of it yes Apr 17 20:51:26 unslung, openslug, debianslug, ucslugc, optware are all orthogonal projects Apr 17 20:51:27 the difference between the versions, like unslung having linksys functionality etc. Apr 17 20:51:50 you make a choice based on the functionality you need. Apr 17 20:51:51 debianslug not having support for eth0... although it has Apr 17 20:52:09 pfavr: please edit that wiki page then Apr 17 20:52:09 but are all projects based on openembedded/the same code base? Apr 17 20:52:15 no Apr 17 20:52:35 thansk Apr 17 20:52:39 thanks Apr 17 20:52:47 * pfavr goes back to study the wiki Apr 17 20:54:08 debianslug and unslung has recent releases but openslug is old and beta Apr 17 20:54:21 all nslu2-linux projects are beta Apr 17 20:54:49 openslug is as recent as the others - it just hasn't had a binary release in a while. Apr 17 20:55:05 ok Apr 17 20:55:33 it is just a bit strange to a newcomer that's all Apr 17 20:55:57 pfavr: well, now you have enough info to write up a wiki page explaining it - are you going to do so? Apr 17 20:56:10 checking out the source - trying the build and it fails. And it is not news... Apr 17 20:56:26 (if not, then you now know the reason why such a page does not exist ...) Apr 17 20:56:30 it ought to be on this page: http://www.slug-firmware.net/ Apr 17 20:56:47 (or a link to the wiki right next to the "source code" link) Apr 17 20:57:01 Um, there is a link to the wiki on the first line of that page. Apr 17 20:57:49 yup- to the main page for all the projects. Not in the "I just grabbed the source and I'm stupid and now I'm building it"-path Apr 17 20:58:05 (You know about community rule #4, right?) Apr 17 20:58:13 any suggestions to where I should put the info on the wiki? Apr 17 20:58:31 pfavr, where would you expect to find it .... Apr 17 20:58:41 I know that I should contribute and not complain - and that is what I would like to do. But I need to be a little educated before I do it Apr 17 20:58:41 rwhitby, reel him in! (hehe) Apr 17 20:58:43 that seems a good place to put it Apr 17 20:58:53 I'm the wrong person to ask, cause I know the information already. Apr 17 20:59:21 pfavr: Seriously - put it exactly where you would have expected to see it. Apr 17 20:59:27 eFfeM: you're absolutely right. In the README or INSTALL file in the tar.gz Apr 17 21:00:03 pfavr: the information is too dynamic for the README. The README points to the wiki - the wiki should be updated. Apr 17 21:00:21 but then it would be just as easy to write a fix in python (I guess) to untar, retar and md5 the bugging file Apr 17 21:01:18 I really think the problem itself should be fixed. It is a nicer and smotther experience to be able to grab the source and type make... then flash it and see it works... then modify a file and break everything :-) Apr 17 21:01:38 pfavr: fully agree. Apr 17 21:01:40 rwhitby: I'm a great fan of wikis. I host a farm of the Apr 17 21:01:41 m Apr 17 21:02:27 any hints on how I fix this - who to contact to put up a new ports.tar? Apr 17 21:02:49 or could it be fixed in the python.bb files? Apr 17 21:02:49 pfavr: what's the URL of the errant file? Apr 17 21:02:57 just a sec Apr 17 21:03:11 http://sources.nslu2-linux.org:/sources/ports_sources.redhat.com__20050627.tar.gz Apr 17 21:04:47 http://how.dk/~pfavr/ports_sources.redhat.com__20050627.tar.gz Apr 17 21:04:49 which package downloads that? Apr 17 21:04:51 is a file that works Apr 17 21:05:15 uh, how do I find out? Apr 17 21:05:33 ah OTE: package glibc-2.3.5+cvs20050627 Apr 17 21:05:34 which package was building when it failed? Apr 17 21:06:08 ok, I've fixed it on the build machine - that should sync to sources Apr 17 21:07:09 strange thing: original tar was 611K and my "fixed" version is only 270K Apr 17 21:08:07 rwhitby: how did you fix it? by putting up a re-tar'ed file? Apr 17 21:08:14 yep Apr 17 21:08:20 yours Apr 17 21:08:37 I really hope it works then... I'm still building here Apr 17 21:08:47 (after I checked you hadn't inserted a trojan) Apr 17 21:09:19 ;-) Apr 17 21:09:34 but what about the missing ~400K? Apr 17 21:09:44 that was the garbage at the end Apr 17 21:10:12 but where did that come from? (was that a trojan?) Apr 17 21:11:07 that wasn't a trojan either. Apr 17 21:11:32 I guess a script pulled the source from CVS then tar'ed it and gzipped it... and failed in doing so. Then nobody noticed for so long... then everybody learned the workaround. And then I came about and complained so loudly that rwhitby fixed it :-) Apr 17 21:12:44 can I remove the file from my webserver or will it be pulled from there from now on? Apr 17 21:18:13 you can remove it Apr 17 21:26:09 gotta go to work tomorrow so I'm off to bed. Thanks for all your kind help. CU Apr 17 21:29:31 pfavr: bye Apr 18 04:27:36 goodmorning, build of OpenSlug-2.7-beta-source.tar.gz completed :-) **** ENDING LOGGING AT Tue Apr 18 09:59:57 2006