**** BEGIN LOGGING AT Wed Feb 16 23:59:56 2005 Feb 17 00:05:36 morning Feb 17 00:05:47 hi giel! Feb 17 00:06:45 I finally got that damn console-message of the Oops, that kills my slug from time to time. Feb 17 00:06:52 nice Feb 17 00:07:13 but I have no clue what it wants to tell me... ;) Feb 17 00:07:24 hey, a little busy now Feb 17 00:07:31 i'll be back in two hours or so Feb 17 00:49:31 morning Feb 17 00:52:30 morn Feb 17 02:18:59 there Feb 17 02:19:00 i rule! Feb 17 02:19:08 ~hail me Feb 17 02:19:10 * jbot bows down to giel and chants, "I'M NOT WORTHY!!" Feb 17 02:19:19 jbot: seen christopher ? Feb 17 02:19:20 xav: i haven't seen 'christopher ' Feb 17 02:19:20 heh Feb 17 02:19:42 jbot: seen christopher Feb 17 02:19:43 christopher is currently on #nslu2-linux Feb 17 02:19:58 jbot: you're weird Feb 17 02:22:11 ~jbot Feb 17 02:22:12 methinks jbot is the shipboard computer, but you may call me eddie if it helps you relax Feb 17 02:22:26 eddie: seen xav Feb 17 02:22:34 stupid Feb 17 02:31:39 ~seen xav Feb 17 02:31:40 xav is currently on #nslu2-linux. Has said a total of 3 messages. Is idling for 11m 43s Feb 17 02:49:54 ~seen jacques Feb 17 02:49:56 jacques is currently on #intimate #nslu2-linux #oe #elinux #handhelds.org Feb 17 02:53:47 ~seen me Feb 17 02:53:48 me <~me@ip503db214.speed.planet.nl> was last seen on IRC in channel #debian, 11d 16h 44m 38s ago, saying: 'Hey I'm trying to set up my scanner for a 2.6.9 kernel, but I can't seem to find the scanner module, did they change anything since 2.4?'. Feb 17 02:54:04 I never said that Feb 17 02:57:03 rwhitby around? Feb 17 03:42:25 pushed. Feb 17 03:45:40 Can someone test unslung-4.x building please? Feb 17 03:46:27 my sysconf is hosed beyond recognition. Feb 17 03:46:48 switchbox 4.7 is pushed and works fine Feb 17 03:47:19 I can reboot with impunity. ;-) Feb 17 03:50:56 yay! Feb 17 04:12:09 dyoung, look at this Feb 17 04:12:24 Reading specs from /export/home/tjyang/slug/unslung/toolchain/armv5b-softfloat-linux/gcc-3.3.5-glibc-2.2.5/lib/gcc-lib/armv5b-softfloat-linux/3.3.5/specs Feb 17 04:12:24 Configured with: /export/home/tjyang/slug/unslung/toolchain/crosstool/build/armv5b-softfloat-linux/gcc-3.3.5-glibc-2.2.5/gcc-3.3.5/configure --target=armv5b-softfloat-linux --host=i686-host_pc-linux-gnu --prefix=/export/home/tjyang/slug/unslung/toolchain/armv5b-softfloat-linux/gcc-3.3.5-glibc-2.2.5 --with-float=soft --with-cpu=xscale --enable-cxx-flags=-mcpu=xscale --with-headers=/export/home/tjyang/slug/unslung/toolchain/armv5b-soft Feb 17 04:12:24 cc-3.3.5-glibc-2.2.5/armv5b-softfloat-linux/include --with-local-prefix=/export/home/tjyang/slug/unslung/toolchain/armv5b-softfloat-linux/gcc-3.3.5-glibc-2.2.5/armv5b-softfloat-linux --disable-nls --enable-threads=posix --enable-symvers=gnu --enable-__cxa_atexit --enable-languages=c,c++,objc --enable-shared --enable-c99 --enable-long-long Feb 17 04:12:26 Thread model: posix Feb 17 04:12:28 gcc version 3.3.5 Feb 17 04:13:09 Objective-C host compiler is enabled on linux. Feb 17 04:13:15 ... Feb 17 04:13:20 But does it work? ;-) Feb 17 04:13:52 Congratulations BTW. youre probably the first to attempt that. Feb 17 04:14:08 I will use it to comiple a hello world objc file and use this to create crossstool-native. Feb 17 04:14:56 There must be a testing suite for it. Feb 17 04:15:10 to verify its doing The Right Thing (tm). Feb 17 04:15:22 should be, I will dig more. Feb 17 04:17:05 at first I didn't compile host compiler with objc enabled, I got error message complaining host compiler is not objecitve-C enabledcective-c enab Feb 17 04:17:20 when I try to comple crosstool-native. Feb 17 04:18:20 BTW, make crosstool should really clean up the .build and .configured, can you fix the Makefile ? Feb 17 04:18:57 that was intentional. Feb 17 04:19:10 because in general you dont want to rebuild the toolchain everytime. Feb 17 04:19:58 even not the developer type "make crosstool-clean" ? Feb 17 04:20:01 I throughly pissed off at unslung-4 now so I'm gonna go to sleep. Feb 17 04:20:33 yes, sleep it off, it helps at least to me ;) Feb 17 04:21:07 night Feb 17 04:24:25 If someone can look into it... It would help determine what is busted in my system. The diagnostic is something is blowing away the sysconf; and the jffs2 appears to be in the wrong place. I have a feeling this is specific to me though because I twidded one too many things and probably screwed up some header or trailer someplace. Feb 17 05:03:31 * giel sings, "You must be the bass!" Feb 17 05:42:56 is anybody using the watchdog in openslug? Feb 17 05:57:36 cool, the watchdog seems to work... the slug is rebooting on it's own... eerie! Feb 17 05:57:46 hehe, there it does! Feb 17 06:01:08 <[g2]> dog bites man! -- News at 11:00 Feb 17 06:01:53 <[g2]> siddy, you seem to be having a bunch of fun Feb 17 06:02:03 Yeah, I was wondering if that worked or not. Feb 17 06:02:07 morning. Feb 17 06:02:12 <[g2]> morning Feb 17 06:02:17 How did you get it to kick? Feb 17 06:02:51 the watchdog? oh, quite easy, it's already in the kernel, just mknod its device and run some daemon to regularly write to the device. Feb 17 06:03:27 <[g2]> siddy, kernel / device hacker :) Feb 17 06:04:17 oh! I see. I noticed it's init banner on bootup but always assumed it was running already. Feb 17 06:04:21 [g2], it's just an act of desperation, because I can't find out where my Oops comes from. Feb 17 06:05:18 What's the watchdog's minor/major? Feb 17 06:05:42 mknod /dev/watchdog c 10 130 Feb 17 06:05:49 Thanks. Feb 17 06:06:21 I'm running wd_keepalive out of http://archive.ubuntu.com/ubuntu/pool/universe/w/watchdog/watchdog_5.2.4.orig.tar.gz to write to the device. Feb 17 06:08:03 Cool. I'll add this to my notes. Feb 17 06:08:37 Does it monitor something in particular? Feb 17 06:09:39 ? no, it's a hardware-watchdog, so it just monitors if you write to the device each 60 secs. If you do not => hard reset! Feb 17 06:10:07 so it's very handy if you have regular kernel-Oopses... like I do. :-\ Feb 17 06:12:27 I meant, where is the reset code residing? In the kernel? Or user-space? Feb 17 06:12:54 in hardware usually Feb 17 06:12:55 no, in hardware. That's the "hard" in "hard reset" ;) Feb 17 06:13:32 sorry, I'm not expressing myself correctly again (waiting for coffee to kick in!): Feb 17 06:13:56 I mean, where is the code that kicks the dog? kernel? Feb 17 06:14:27 no, that's in userspace. Feb 17 06:14:27 a watchdog is just a timer connected to the reset line Feb 17 06:14:57 I'm wondering because if it's in user space than it's possible for that process to die but not the kernel. right. I know it's not full proof. Feb 17 06:15:30 (I know the hardware part) Just wondering about the actual code that keeps waking up the dog. Feb 17 06:15:44 that's exactely the point Feb 17 06:15:48 yes, if wd_keepalive dies unexpectedly, the watchdog will trigger a reset. Feb 17 06:16:00 I guess that does the job. Feb 17 06:16:18 but, it's a _very_ simple app, so it won't die too often without reason! ;) Feb 17 06:16:25 i see. Feb 17 06:19:49 <[g2]> Be nice... please pet the dog Feb 17 06:22:43 * siddy pets his dog to make it start barking when the next Oops arrives. Feb 17 06:26:21 hehehe Feb 17 06:28:09 [g2], I got the kernel-messages from a crash last night via netconsole. Feb 17 06:28:18 but it doesn't tell me too much... :-\ Feb 17 06:29:51 I noticed the posting about the hotplug-ng thing yesterday. Looks interesting. Feb 17 06:30:14 juj for s5! Feb 17 06:41:46 kergoth: You're planning on adding hotplug-ng as a package you said? That would be really nice. Thanks in advance. Feb 17 06:42:00 OE is very cutting edge! Feb 17 06:48:59 kergoth: I just compiled the hotplug-ng and it compiled without problem with my cross compiler. I can't access my slug from work today though so we'll see later. Feb 17 06:51:25 good! Feb 17 07:35:27 hello all Feb 17 07:35:34 hi Feb 17 07:36:15 what's the best ftp daemon to run on the NSLU2? Feb 17 07:37:31 openssh - run sftp for secure ftp. if you want insecure ftp, then vsftpd or proftpd Feb 17 07:38:43 * giel suggests vsftpd Feb 17 07:39:05 thanks-I wanted to set up an insecure FTP to see how G4U works with it Feb 17 07:45:07 03rwhitby * 10unslung/Makefile: Promoted gdb Feb 17 07:45:43 What am I doing wrong - I tried ipkg install vsftpd and I get: No space left on device Feb 17 07:48:38 03rwhitby * 10unslung/Makefile: Promoted gconv-modules. Feb 17 07:48:46 did you unsling ? Feb 17 07:49:06 if you have filled your flash, then you're in trouble .. Feb 17 07:49:11 if you are asking me - yes Feb 17 07:49:31 is your conf partition full? Feb 17 07:49:53 conf partition has 4% used Feb 17 07:50:17 root directory is full Feb 17 07:51:03 03rwhitby * 10unslung/make/gift-opennap.mk: Fixed the version. Feb 17 07:52:16 is /opt a real directory, or a symbolic link Feb 17 07:53:41 (I'm assuming you're running 3.18, otherwise you would have told me what version of Unslung you are running ...) Feb 17 07:53:48 sec - rebooting Feb 17 07:54:15 hmm - not a wise thing to do if your root directory is full .... Feb 17 07:54:15 yes - just downloaded the latest this morning from sourceforge Feb 17 07:54:33 d'oh Feb 17 07:54:42 how did your root directory get full? Feb 17 07:55:16 beats the heck out of me - it's my first day playing with this Feb 17 07:55:40 did you run /sbin/unsling after you flashed the firmware? Feb 17 07:55:52 yes I did Feb 17 07:56:23 ok, so ipkg should be installing stuff in /opt which should be a symlink onto your attached storage device Feb 17 07:56:35 then updated busybox Feb 17 07:57:17 it looks like it's having trouble rebooting Feb 17 07:57:37 hi Feb 17 07:58:24 Kirk_22: I expect that somehow you have filled the flash (perhaps the disk did not mount properly?) and that will probably mean you will need to reinstall the firmware Feb 17 07:58:32 I ask myself how we want to provide perl modules. as ipk or only by cpan shell? Feb 17 07:59:08 pTweety: I would suggest cpan shell, cause then we have no maintenance overhead. Feb 17 07:59:20 I think you're right, I suppose I should check the wiki to recover from a bad flash? Feb 17 07:59:28 Kirk_22: tep Feb 17 07:59:30 yep Feb 17 08:00:25 rwhitby-away, if user want to install e.g. spamassin with a lot of prerequisits than they can't have an ipk Feb 17 08:01:23 well, they can, but it won't work without the right perl modules installed :-) Feb 17 08:01:27 I see your point Feb 17 08:01:57 I dislike the idea to package spamassassin and all deps in one package Feb 17 08:02:03 therefore I asked Feb 17 08:02:11 packaging perl modules into ipks will be an interesting challenge for you .... Feb 17 08:02:33 agreed that the deps should be separate Feb 17 08:02:34 either prepare a lot of ipk or leave the users alone with some small comments ;-) Feb 17 08:03:38 once you start packaging some perl modules, then users will expect *all* CPAN modules to be packaged for them ... where do you draw the line? Feb 17 08:04:06 IMO, packaging can only be done native. Does it become an issue for the native feed when we have to much modules for native only? Feb 17 08:04:06 night all Feb 17 08:04:43 the line is drawn by my wife :-) Feb 17 08:05:23 my first goal will be to have a mail server with spamassassin and that's it Feb 17 08:06:14 well, but if we prepare the most important modules, then we are fine in most cases Feb 17 08:06:32 pTweety: i want that too :) Feb 17 08:07:18 <[g2]> VoodooZ_work, is it hotplug-ng ? Feb 17 08:08:14 [g2]: what? Feb 17 08:08:41 <[g2]> you said you compile hotplug-ng did kergoth add it to OE ? Feb 17 08:08:45 giel, I'll start with some modules, check them into cvs and then the community may decide Feb 17 08:09:04 no, I just downloaded the tar file for fun. Feb 17 08:09:10 pTweety: well, as i'm using openslug i guess i can only derive from your efforts Feb 17 08:09:12 I doubt he added it yet. Feb 17 08:09:32 The bk version would be more complete. Feb 17 08:10:21 It's not totally finished yet but from the sound of it most of the stuff missing is useless on the slug so it's looking good. Feb 17 08:10:57 <[g2]> great. Thats the way it sounded yesterday from the post kergoth referenced Feb 17 08:11:27 <[g2]> I'd love to be able to include that in the beta release Feb 17 08:11:36 <[g2]> with a much quicker boot Feb 17 08:11:37 I can't wait to see the speed diff. Feb 17 08:11:59 yep. If kergoth adds it soon we should be able to test it. Feb 17 08:12:16 <[g2]> VoodooZ_work, do you run jffs2 ? Feb 17 08:13:09 yep. Whatever is included with Openslug. Feb 17 08:13:20 <[g2]> do you use switchbox at all ? Feb 17 08:13:27 Not really no. Feb 17 08:13:37 I know it boots the slug but that's it. Feb 17 08:13:55 I might end up removing it if I see it's slowing down my boot. Feb 17 08:14:05 Unless it's required? Feb 17 08:14:29 <[g2]> I think you can change you boot command line and boot straight to the jffs2 partition Feb 17 08:14:43 <[g2]> it'll save you some time Feb 17 08:14:59 * [g2] guesses 4-8 seconds Feb 17 08:15:08 <[g2]> maybe even 10 Feb 17 08:15:53 really? Which file would that be in? Feb 17 08:16:31 I'm a bit confused when it comes to the whole initrd/jffs2/switchbox stuff. Feb 17 08:16:48 <[g2]> just change your kernel command line to root=/dev/mtdblock4 rootfstype=jffs2 Feb 17 08:16:58 <[g2]> along with the console line Feb 17 08:17:36 <[g2]> You can test loading the kernel from the Redboot Prompt and you'll just need to load only the kernel Feb 17 08:17:45 yeah, but I was told that that line in defconfig gets overidden by a bb file so which file is the one I should mess with? Feb 17 08:18:01 <[g2]> the .bb file Feb 17 08:18:13 let me check... Feb 17 08:19:15 openslug-kernel_2.6.9.bb right? Feb 17 08:19:29 currently: CMDLINE_ROOT = "root=/dev/slug rootfstype=ext2,jffs2 initrd=0x01000000,10M mem=32M@0x00000000" Feb 17 08:20:03 so I just replace the root and rootfstype options? Feb 17 08:20:51 <[g2]> #CMDLINE_ROOT = "root=/dev/slug rootfstype=ext2,jffs2 initrd=0x01000000,10M mem=32M@0x00000000" Feb 17 08:20:51 <[g2]> CMDLINE_ROOT = "root=/dev/mtdblock4 rootfstype=jffs2 mem=32M@0x00000000" Feb 17 08:20:56 <[g2]> That's my attempt Feb 17 08:23:46 Can anybody send me the modules md.o and raid1.o? Feb 17 08:23:47 Cool. I'll have to try that tonight when I get home. Feb 17 08:24:47 * giel sings, "You must be the bass!" Feb 17 08:26:18 [g2]: so that line totally bypasses the initrd/switchbox thing and boots the kernel directly? I'll still have my jffs2 root too? Feb 17 08:27:47 <[g2]> I think we may need init=/sbin/init on there too! Feb 17 08:27:56 Ok. I'll add it to my notes. Feb 17 08:28:04 <[g2]> yeah, it'll bypass it Feb 17 08:28:30 cool. thanks. I can't wait to see how much time I'll save Feb 17 08:29:13 Is the command line order important? Feb 17 08:29:34 <[g2]> don't think so Feb 17 08:29:57 ok. Feb 17 08:30:11 <[g2]> hmmm, I wonder where OE adds the console part ... Feb 17 08:30:19 same file. Feb 17 08:30:22 also in the .bb Feb 17 08:30:33 why? Feb 17 08:31:13 [g2]: init=/sbin/init is probably not needed, as it is in the standard probing list for init prgs in the kernel sourdce Feb 17 08:31:16 source* Feb 17 08:31:21 but it won't hurt Feb 17 08:31:28 ok Feb 17 08:31:33 [g2]: Did you heard about my stupid mistake? Feb 17 08:31:33 okay, i'm off to home Feb 17 08:31:37 later Feb 17 08:31:43 <[g2]> no Feb 17 08:32:15 I had my ccache path pointing to my home folder (i didn't know) so whenever I was building openslug it was accessing my NFS mounted home dir!!! Feb 17 08:32:29 No wonder my machine was so laggy and my builds were so long! Dope! Feb 17 08:33:08 <[g2]> Doh! Feb 17 08:33:12 I've since configured CCACHE_DIR to point to a local drive and I'm now getting ~72 minutes builds with bluetooth and camsource stuff. Feb 17 08:33:16 :( Feb 17 08:33:31 <[g2]> Glad you found it! Feb 17 08:33:40 yep. Building is fun again! Feb 17 08:33:49 <[g2]> Did you hear about my fubar with memory ? Feb 17 08:33:58 I'll have to try it on the sun V40Z (quad opteron) soon too. Feb 17 08:34:07 I don't think so no. What happened? Feb 17 08:34:39 <[g2]> Well when I upgraded the last .5G I noticed the original .5G was only running at 333 Feb 17 08:35:03 .5G? Feb 17 08:35:14 on your build workstation you mean? Feb 17 08:35:17 <[g2]> I had upgraded before and the bios recognized it, but defaulted it to 333Mhz Feb 17 08:35:21 <[g2]> .5G memory Feb 17 08:35:25 ic Feb 17 08:35:34 What speed is it supposed to be at? Feb 17 08:35:48 DDR? Feb 17 08:35:56 <[g2]> I knew it was capable of 400Mhz but said, "Oh, I'll just use the 333 setting" Feb 17 08:35:59 <[g2]> nod Feb 17 08:36:26 I see. Gives you a bit of speed boost. Feb 17 08:36:30 <[g2]> I'd been running for almost a year that way leaving all that speed on the table :( Feb 17 08:36:37 that sucks! Feb 17 08:36:51 Is that on an AMD board? dual channel? Feb 17 08:36:57 <[g2]> It's actually quite funny (Intel) Feb 17 08:37:01 <[g2]> 865 Feb 17 08:37:07 * VoodooZ_work nods Feb 17 08:37:19 <[g2]> My systems be blazingly fast Feb 17 08:37:25 <[g2]> been Feb 17 08:37:33 <[g2]> but it got even faster Feb 17 08:37:35 <[g2]> :) Feb 17 08:37:44 you appreciate it even more now! Feb 17 08:37:50 a free upgrade! Feb 17 08:40:01 <[g2]> I'm booting directly to the jffs2 now Feb 17 08:40:34 using your recommended command line? Feb 17 08:40:51 <[g2]> nod. Feb 17 08:40:58 cool! how does it look? Feb 17 08:40:59 <[g2]> I don't know if that "recommended" Feb 17 08:41:02 <[g2]> ok Feb 17 08:41:23 no apparent problems? Feb 17 08:41:37 <[g2]> none yet Feb 17 08:41:44 still booting? Feb 17 08:41:54 <[g2]> it's booted twice already Feb 17 08:42:02 I see. how long approx? Feb 17 08:42:14 <[g2]> I don't have my watch on :( Feb 17 08:42:32 how about a nearby PC. :) Feb 17 08:42:32 <[g2]> heading out for lunch Feb 17 08:42:38 well, does it "feel" faster? Feb 17 08:42:45 <[g2]> dunno Feb 17 08:42:50 VoodooZ_work: yes, i have hotplug-ng in my local tree here. had to rename the original hotplug package and stuff to make sure the two packages providing /sbin/hotplug played nice with one another Feb 17 08:42:50 np. thanks. Feb 17 08:42:52 <[g2]> I think so on the boot Feb 17 08:43:05 ~praise kergoth Feb 17 08:43:08 All hail kergoth! Feb 17 08:43:23 wow! all this at the same time. I love it. I can't wait to try all this tonight. Feb 17 08:43:26 <[g2]> kergoth, COOL! Are yo planning on pushing that anytime soon Feb 17 08:43:53 * VoodooZ_work kneels and scream "pretty please!" Feb 17 08:44:03 yeah, just checking to make sure nothing breaks (namely, image creation), since the two packages both RPROVIDES hotplug, need to ensure that one or the other is in IPKG_INSTALL Feb 17 08:44:09 to break the tie for ipkg Feb 17 08:44:19 good. Feb 17 08:44:50 <[g2]> awesome, That'll be really cool to try out. Feb 17 08:44:56 so we'd only need to comment out the hotplug option of the appropriate bb file and replace it with hotplug-ng right? Feb 17 08:45:05 <[g2]-away> lunch Feb 17 08:45:21 VoodooZ_work: the image .bb? not exactly. which hotplug you use is a distro decision. Feb 17 08:45:25 remember, I'm the king of breaking stuff so expect to hear me whine in the next few days! :) Feb 17 08:45:31 VoodooZ_work: so i need to sprinkle the appropriate bits into the distro .conf's Feb 17 08:45:37 i see. Feb 17 08:48:25 Rod, are you around ? Feb 17 08:49:47 hehe, my openslug just crashed again! but... my fellow watchdog rebooted it and it came back up! better than nothing... ;) Feb 17 08:50:22 * siddy pets and feeds his cute watchdog Feb 17 08:51:52 How do I update a pw protected Wiki page ? Feb 17 08:52:52 siddy: Thanks again for the watchdog info. It's kind of important for a speeding robot! Feb 17 08:53:21 yeah, right, np! Feb 17 08:53:43 but in return... you have to find that kernel-bug for me!! ;) Feb 17 08:53:57 anyways, lunch-time (meat-lovers pizza!). see you later. Feb 17 08:54:18 hehehe. Yeah, don't worry, lucky as I am I'll probably discover it too tonight. Feb 17 08:54:22 bbl Feb 17 10:34:17 siddy: What kind of watchdog system do you have? Feb 17 10:49:09 VoodooZ_pizza, [g2]-away: Feb 17 10:49:10 12:44 < CIA-3> openembedded: kergoth * r1.2993.1.1 openembedded/ (37 files in 10 dirs): Add Feb 17 10:49:13 hotplug-ng, rename hotplug to linux-hotplug, and move the decision about which Feb 17 10:49:16 /sbin/hotplug binary you use into the hands of the distributions. Feb 17 10:49:23 feel free to modify openslug.conf to use hotplug-ng and try that Feb 17 10:50:26 <[g2]> kergoth, I'll pull and try it out in the next hour or so Feb 17 10:50:34 cool, thanks Feb 17 10:50:38 i ahvent had time to test it Feb 17 10:50:39 <[g2]> I'm just trying out the latest switchbox 4.7 Feb 17 10:51:04 <[g2]> kergoth, the apache package is marked as broken Feb 17 10:51:12 the apache package probably is broken. Feb 17 10:51:19 i'll look at it Feb 17 10:51:24 <[g2]> thx. Feb 17 10:51:56 <[g2]> it looks like it's pretty close, the big problem is a alocal.m4 template after the init and httpd.conf files Feb 17 10:52:27 the big problem is the vast apache buildsystem. getting autoreconf happy on it isnt an easy task Feb 17 10:53:04 <[g2]> there's some dude that built it for unslung, maybe we can get some help from him Feb 17 10:53:18 wont help Feb 17 10:53:31 highly unlikely that he bothered running aclocal/autoconf/automake/libtoolize/etc on it. Feb 17 10:53:34 and thats the source of the issues Feb 17 10:53:53 it isnt _difficult_, its just _time consuming_ Feb 17 10:54:15 it used to build, just a pain to keep working Feb 17 10:56:58 bob_tm, I'm using nslu2's buildin hardware-watchdog. Feb 17 10:57:11 And it works? Feb 17 10:57:21 yes, flawless. Feb 17 10:57:51 Awesome - it is not controlled by the watchdog daemon, is it? Feb 17 10:58:30 So can it be used with unslung? Feb 17 10:58:39 you mean unslung's watchdog-daemon? I don't know, I'm on openslug. You could try by killing the daemon and waiting for the slug to reboot. Feb 17 10:58:59 The "WatchDog" there is just a shell script run from cron Feb 17 10:59:21 I don't use it anymore but would love to have a proper (ie. hw) watchdog. Feb 17 10:59:40 hm... that script-thing is indeed unlikely to be a "real" watchdog. Feb 17 10:59:42 It is in the openslug kernel then? Feb 17 10:59:46 yes Feb 17 10:59:48 <[g2]> siddy, did you you by chance write up the howto for the watchdog ? Feb 17 11:00:04 I will take a look. That was indeed a pleasant surprise! thnx Feb 17 11:00:21 poking the watchdog from userspace isnt necessarily a bad idea, if you want the device to reboot even if userspace gets hung up, but the kernel has not Feb 17 11:00:35 howto? no. It's not too hard, so no need for a howto! ;) Feb 17 11:00:36 for a headless embedded device, i could see that being a valid thing to do Feb 17 11:01:12 <[g2]> siddy, The shorter the sweeter :) Feb 17 11:03:19 <[g2]> VoodooZ_pizza, ping Feb 17 11:03:42 Somebody said pizza? Feb 17 11:03:45 heheh Feb 17 11:03:53 <[g2]> VoodooZ_Work, I'm trying out dyoung's 4.7 switchbox Feb 17 11:03:55 I think I'll puke if I see more food. Feb 17 11:04:01 What about it? Feb 17 11:04:06 <[g2]> it boots straight to the prompt in switchbox Feb 17 11:04:16 <[g2]> from the + to # it's 14 secnods Feb 17 11:04:37 so he sped up just the switchbox part? Feb 17 11:04:50 <[g2]> well this is *just* switchbox Feb 17 11:04:57 <[g2]> and the linuxrc Feb 17 11:05:08 I mean, there's still going to be the extra junk after taht of course. Feb 17 11:05:18 Can't be bad. Feb 17 11:05:26 What's the change? Feb 17 11:05:27 <[g2]> unless you ran from switchbox Feb 17 11:05:40 <[g2]> and just mounted the jffs2 part Feb 17 11:05:50 But then again, If I don't need switchbox I can save that 14 seconds right? Feb 17 11:06:16 http://www.nslu2-linux.org/wiki/OpenSlug/Watchdog Feb 17 11:06:17 <[g2]> switchbox is probably the last 3 seconds of that Feb 17 11:06:30 thanks siddy Feb 17 11:06:39 hmm... seems I *was* wrong - "map system=yes" is default in Linksys firmware Feb 17 11:06:47 my shortest wiki-page ever! ;) Feb 17 11:06:48 <[g2]> ~praise siddy Feb 17 11:06:50 All hail siddy! Feb 17 11:06:52 cool. I guess I'll try to make sense of it all once I get home. Feb 17 11:07:03 <[g2]> siddy, That's 3 for 3 Feb 17 11:07:07 I'll take whatever is fastest. Feb 17 11:07:10 <[g2]> one each day Feb 17 11:07:32 hehe, I'll have to think about something cool for tomorrow! Feb 17 11:07:33 Is dyoung's switchbox in the repo? Feb 17 11:07:44 siddy: pressure is on! :) Feb 17 11:08:23 what's wrong with me? i don't feel like openslugging Feb 17 11:08:40 i'm gonna watch tv and drink beer and be generally dumb for a while Feb 17 11:08:56 hard work today, and presentation went remarkably well Feb 17 11:09:05 pbly thanks to the goodluck wishes of y'all :) Feb 17 11:09:33 <[g2]> giel, CONGRATS! Relax kick back, we know you'll be back in a while :) Feb 17 11:10:15 <[g2]> VoodooZ_Work, nod. nslu2-switchbox Feb 17 11:10:29 <[g2]> or something like that, and the command line needs to be changed Feb 17 11:11:03 i got compliments from my manager, which is good as i have a review meeting tuesday Feb 17 11:11:14 or how do you call something like that in english Feb 17 11:12:14 so it's not the default right now then? Feb 17 11:12:42 what's the change btw so I can add it to my notes while it's fresh. Feb 17 11:13:58 <[g2]> VoodooZ_Work, the default right now is 3.9 Feb 17 11:14:24 ic. Is it simple to change? Do you plan on making it the default in the near future? Feb 17 11:14:27 <[g2]> check with dyoung or the readme I think it also support nfs booting Feb 17 11:15:11 someone mention readme? Feb 17 11:15:12 If the switchbox bypass trick works fine for me I won't need to do it but I'm curious in case. Feb 17 11:15:13 <[g2]> I'd like to be able to have users boot from flash stick/harddrive and I'm verifying that this supports it Feb 17 11:15:23 good. Feb 17 11:15:48 my sister bought a slug, but she has a full FAT32 drive which she wants to hook up to it... so that'll be fun for me to fix that soon :) Feb 17 11:15:57 <[g2]> I think we can default to that, and it's a simple command line change to boot straight to the jffs2 parition Feb 17 11:16:03 btw, Do you still get that annoying modutils script error on bootup? Feb 17 11:16:22 <[g2]> VoodooZ_Work, nod Feb 17 11:16:25 <[g2]> 127 thingy Feb 17 11:16:29 crap. Feb 17 11:16:48 What's that? Isn't it anybody else using a 2.6 kernel with module-init-tools? Feb 17 11:17:10 <[g2]> I think it's from a bogus dependancy on the modutils-sh which shouldn't be there, so it's not causing real problems from my understanding Feb 17 11:17:20 same for the modules.conf, it should be modprobe.conf now. Feb 17 11:17:41 yeah, I know but I wish I could fix it to clean up the boot. Feb 17 11:18:24 <[g2]> nothing stopping you, I'm sure if you provide the proper patch kergoth would put it in OE Feb 17 11:18:26 It's that ipk post-install configuring thing. Unless I can pretend it ran successfully so It won't try it again. Feb 17 11:18:52 yeah I wish It was that simple. I'll try. Feb 17 11:19:42 <[g2]> I'd think that some ppl would help you fix it if you asked nicely in OE Feb 17 11:19:44 I still don't quite understand the oe structure. :( Feb 17 11:19:56 <[g2]> me either :) Feb 17 11:20:07 the build piece? or the distro? Feb 17 11:20:10 and I hate to beg. heard on my pride. Feb 17 11:20:18 <[g2]> but I don't understand my wife either, but that doesn't stop me from being married Feb 17 11:20:25 hehehehe Feb 17 11:20:35 That's probably I'm still dating at 30! :) Feb 17 11:20:48 <[g2]> I was! Feb 17 11:21:15 oops. I forgot a "why" in there. Feb 17 11:21:17 <[g2]> ah... but that was a long long time ago Feb 17 11:21:47 kergoth: Do you know what causes the modutils error on startup. Feb 17 11:21:53 <[g2]> it's a little bit like the slug Feb 17 11:22:00 I did remove all modutils packages. Feb 17 11:22:09 "the modutils error" Feb 17 11:22:12 my slug is collecting dust. Feb 17 11:22:18 want to try giving me some real actual information? Feb 17 11:22:20 like the error? Feb 17 11:22:41 sorry. My memory is crap right now as I've been working on the robot's power system so I haven't use the slug in a while Feb 17 11:23:25 it happens during the "configuring...." part. (package post-installation). It tries to run some modutils stuff that is no longer present. Feb 17 11:24:02 oh that stuff Feb 17 11:24:19 yeah, what's the error again? Feb 17 11:24:22 blegh, that was just some editing of an ipkg script iirc Feb 17 11:24:22 i'd need more information. Feb 17 11:24:23 details Feb 17 11:24:47 sorry, I don't have access to my slug atm. worst-case I'll ask later. Feb 17 11:24:50 k Feb 17 11:24:52 It's mostly esthetic. Feb 17 11:25:08 (but it slows it down a bit too) Feb 17 11:25:09 iirc one of the scripts calls (imo totally useless) ipkg-update Feb 17 11:25:13 and that gives an error Feb 17 11:25:25 something like that. Feb 17 11:25:26 the last error like that i saw was depmod failing to run depmod.old, but that only happens with a module-init-tools rootfs on a 2.4 kernel Feb 17 11:25:28 and thats fixed Feb 17 11:25:30 ah, okay Feb 17 11:25:49 VoodooZ_Work: you really should dig through your bootscripts and kick out everything you don't need ;) Feb 17 11:25:56 It's like the package is unaware of the 2.6 stuff. Feb 17 11:26:09 I'll do that real soon. Feb 17 11:27:02 I don't know why It's in there in the first place so it's hard to remove it. Feb 17 11:27:25 I could change the scripts but I'd rather make the changes permanent in bk. Feb 17 11:33:29 <[g2]> VoodooZ_Work, you'll probably want to wait on the switchbox 4.7 Feb 17 11:33:54 <[g2]> there are some unslung specific things in there we'll need to fix those up Feb 17 11:34:02 VoodooZ_Work: if you have questions about what various scripts do, just ask. Feb 17 11:34:06 Yeah. I'll try the jffs2 workaround for now and I'll play with the new hotplug. Feb 17 11:34:18 Thanks guys. Feb 17 11:34:20 i maintain the initscripts package Feb 17 11:34:33 cool. I'm pulling in your hotplug stuff now. Feb 17 11:34:41 its never been tested :) Feb 17 11:34:47 I can't run it from here. Feb 17 11:34:55 oh okay Feb 17 11:34:58 I know. I'll break it for you! :) Feb 17 11:35:03 good plan Feb 17 11:35:11 wow Feb 17 11:35:13 I just want to have it built by the time I get home. Feb 17 11:35:13 i'd rather see it break now that when a ton of users or distros are usin it Feb 17 11:35:15 i completely misread that Feb 17 11:35:24 i read: I'll break you for it! Feb 17 11:35:27 hah Feb 17 11:35:29 heheh Feb 17 11:35:47 I'm a good dog, I don't bite the hand that feeds me. Feb 17 11:36:30 I just have to pull in the proper hotplug package right? Feb 17 11:37:00 adjust openslug.conf to have hotplug-ng instead of linux-hotplug Feb 17 11:37:05 thatll do it Feb 17 11:37:09 cool Feb 17 11:37:49 * [g2] will race VoodooZ_Work :) Feb 17 11:37:56 I hope so! Feb 17 11:38:29 I can't run it until after my hockey game tonight so you still have 6 hours! :) Feb 17 11:39:00 i should try it on one of my zauruses Feb 17 11:39:12 the 6k probably, since it has usb host Feb 17 11:39:22 The slug is a good candidate as It has very minimal h/w support. Feb 17 11:42:45 Here goes nothing...build starting... see you at the finish line [g2]! Feb 17 11:45:16 <[g2]> VoodooZ_Work, I've been building for 5+ minutes :) Feb 17 11:45:22 damn! Feb 17 11:45:23 :) Feb 17 11:45:36 mine should be done in ~80 minutes. Feb 17 11:48:15 <[g2]> hotplug-ng is build and building image Feb 17 11:48:46 good. Feb 17 11:48:58 ~praise kergoth Feb 17 11:49:00 All hail kergoth! Feb 17 11:49:08 * [g2] flashing Feb 17 11:49:17 <[g2]> hold your breath Feb 17 11:49:42 * VoodooZ_Work is turning blue. Feb 17 11:49:56 * [g2] ready with the stopwatch Feb 17 11:50:24 good! Feb 17 11:52:49 <[g2]> WHOA! Feb 17 11:53:53 WhOa what? Feb 17 11:54:05 flew by so fast you didn't see it? Feb 17 11:54:28 heh Feb 17 11:54:39 i bet that hotplug-ng package didnt include a startup script, so no coldplugging happened Feb 17 11:55:12 <[g2]> probably, installing drivers seems a little funny Feb 17 11:55:28 didn't work? Feb 17 11:55:45 <[g2]> driver didn't come up Feb 17 11:55:55 try modprobing the usb host controller driver manually Feb 17 11:55:56 damn. Feb 17 11:55:58 then see if hotplug works Feb 17 11:56:04 <[g2]> maybe it did hold on Feb 17 11:56:11 if its just coldplug thats missing, thats no big deal Feb 17 11:56:17 * kergoth pokes into the hotplug-ng ipk Feb 17 11:56:46 <[g2]> ok ignore the funny thing Feb 17 11:56:54 <[g2]> I'm rebooting with the disk attached Feb 17 11:57:04 hehe Feb 17 11:57:21 <[g2]> it was 1 Minute 11 sec for the full first boot Feb 17 11:57:35 It's like the teachers here at work, they call me panicking that their HDs died when it's simply a floppy left in the drive! :) Feb 17 11:57:43 that much? Feb 17 11:58:09 IS that with switchbox 4.7? Feb 17 11:58:23 <[g2]> no Feb 17 11:58:33 <[g2]> 40 seconds from + to # on second boot Feb 17 11:59:20 <[g2]> that includes ixp driver install, but it doesn't seem to recognize the hd Feb 17 12:00:12 okay, so hotplug-ng installs sane default hotplug.d scripts for usb, scsi, and pci Feb 17 12:00:23 but has no startup script, so no coldplug occurs Feb 17 12:00:30 yeah and that ixp stuff is kind of long so once removed it should be good. Feb 17 12:00:47 <[g2]> kergoth, the hotplug for the usb/scsi doesn't trigger Feb 17 12:01:07 I wonder if the new csr fixes the auto-neg stuff. Feb 17 12:01:10 you dont know that. Feb 17 12:01:14 what you know is that your modules arent being loaded Feb 17 12:01:22 <[g2]> nod. Feb 17 12:01:25 either the hotplug.d scripts arent being executed, or they are, and they arent behaving correctly Feb 17 12:01:31 you havent yet determined which is the case. Feb 17 12:01:41 <[g2]> a set -x would tell me right Feb 17 12:01:42 drop a custom script in hotplug.d to confirm that the /sbin/hotplug multiplexor works Feb 17 12:01:47 set -x on what? Feb 17 12:01:51 hotplug-ng is all C. Feb 17 12:01:53 <[g2]> hotplug.d Feb 17 12:01:58 including the default usb, scsi, and pci agents Feb 17 12:02:51 <[g2]> well I don't get any dmesg from the hotplug event Feb 17 12:14:27 do we have any MTA in openslug? Feb 17 12:14:49 postfix Feb 17 12:15:03 dont think any others have been added yet Feb 17 12:15:08 oh, missed that one, thx! Feb 17 12:15:13 np Feb 17 12:15:18 <[g2]> bb postfix :) Feb 17 12:15:21 course i've never tested it :) **** BEGIN LOGGING AT Thu Feb 17 12:17:11 2005 Feb 17 12:22:57 <[g2]> siddy, that's for tomorrow Feb 17 12:23:19 <[g2]> besides, that last one was.... just a tease right :) Feb 17 12:23:50 :-D Feb 17 12:26:17 I don't know how he handles a real job and this one Feb 17 12:27:37 <[g2]> who's that ? Feb 17 12:27:47 siddy, of course Feb 17 12:28:02 me? oh... real job... hm... student? Feb 17 12:28:16 <[g2]> college ? Feb 17 12:28:43 err... university, dunno what you call it over there. Feb 17 12:29:02 <[g2]> nod. some ppl are in high school Feb 17 12:29:13 ah... so that's the secret Feb 17 12:30:26 well "way back when" I was in college, free time was for riding mountain bikes, chasing women, and other non-productive activities Feb 17 12:30:52 but don't tell anyone, they might come take my diploma away Feb 17 12:34:59 03ingeba * 10unslung/sources/parted/control: Added dependencies Feb 17 12:36:38 ByronT: you mistake "non-productive" with "not college related, but still necessary for development and life experience in general, so in that way productive" Feb 17 12:37:13 I bet you're one heck of a rationalist... ;) Feb 17 12:37:19 heh Feb 17 12:37:26 if i wasn't i'd have to kill myself :) Feb 17 12:37:37 i have to rationalise all my stupid actions afterwards Feb 17 12:37:54 maybe pragmatist is a better word Feb 17 12:38:36 that's where I begin to get in trouble... thoughts like "now why *did* I do that?" Feb 17 12:38:47 nuhuh, don't even go there Feb 17 12:39:02 as much as ppl like to believe they thought about their actions, noone really does Feb 17 12:39:15 all decisions are made on gut feeling and are instantly rationalised Feb 17 12:39:39 your mind is constantly making up stories to make you think you're in control ;) Feb 17 12:39:55 so if you skip that part and just act on feeling most of the time, it'll make your life easier Feb 17 12:40:01 it gets more fun when you have kids.... Feb 17 12:40:48 because then you occasionally try to be the example, but then think "wouldn't it be cute if he did that though..." Feb 17 12:41:08 well, i wanna have kids but i don't know if it's a wise decision Feb 17 12:41:13 first get mahself a ladeh :P Feb 17 12:41:18 rationalise it later Feb 17 12:41:18 * kergoth is going to avoid it as long as possible Feb 17 12:41:23 yup Feb 17 12:41:39 I wouldn't trade my little stinker for the world though Feb 17 12:41:41 so, i take it you've got kids? Feb 17 12:41:52 www.toddcomp.com/Christian.html Feb 17 12:42:03 03ingeba * 10unslung/make/parted.mk: Added libiconv to dependencies Feb 17 12:42:18 heh, cute lil' laddy Feb 17 12:42:32 but somehow i wanna stabilize somewhat first Feb 17 12:42:39 eh, who just did that checkin? Feb 17 12:42:43 you dont need libiconv. Feb 17 12:43:15 he's a mess now... he applauds himself when he burps Feb 17 12:44:01 heh Feb 17 12:44:10 just like his father :P Feb 17 12:44:29 I set a bad example the wife says Feb 17 12:45:07 real life intrudes (work)... Feb 17 12:45:25 * ByronT is away -( work )- at 02:45p -( P:On / L:On )- Feb 17 12:45:45 have fun Feb 17 12:45:54 as always Feb 17 12:46:37 giel, watching soccer tonight? Feb 17 12:47:06 not really... my favorite match is only aired on pay tv Feb 17 12:47:16 which i don't have, 'cause i don't like tv at all :) Feb 17 12:47:41 i keep track of the standings tho Feb 17 12:47:44 ah, I'm watching the good old krauts vs dutchmen match... (we'll win!) ;) Feb 17 12:47:54 oh that one Feb 17 12:48:03 well, national soccer doesn't really do it for me Feb 17 12:48:22 i used to like the netherlands in general, but it's getting less and less Feb 17 12:48:32 i like the local team of the place i grew up tho Feb 17 12:48:34 actually I'm only watching the game, because it's "my" team. living in aken. Feb 17 12:48:38 they're playing UEFA cup tonight Feb 17 12:48:51 heh Feb 17 12:48:53 oh wait Feb 17 12:48:55 that's you Feb 17 12:48:59 freakin hell no you won't win! Feb 17 12:49:15 I will. ;) Feb 17 12:49:43 hey, you live in aken? Feb 17 12:49:46 that's scary nearby Feb 17 12:49:52 i live in eindhoven these days Feb 17 12:50:09 really? oh, 30 muinutes car-drive I'd say. Feb 17 12:50:31 knowing dutch traffic i'd say 45, but yeah Feb 17 12:50:38 I bet you know Budel near Weert, right? Feb 17 12:50:46 well, i know of it Feb 17 12:50:54 i don't know the surroundings here, only live here for 3 months Feb 17 12:51:00 hehe, been there for 10 months of my life. Feb 17 12:51:21 i'm going to belgium tomorrow, so i'll be even closer for a moment Feb 17 12:51:28 and saturday i'm going to germany Feb 17 12:51:56 ah, driving your manta on our highways? ;-) Feb 17 12:52:11 no, manta is waiting in a garage for time and money Feb 17 12:52:17 i'm going to look at a smart roadster Feb 17 12:52:26 maybe i'm gonna buy me one of those Feb 17 12:52:48 ah, swett little cars. a bit too expensive for poor students... Feb 17 12:53:35 s/swett/sweet/ Feb 17 12:54:02 yep, i've driven one in the ardennes Feb 17 12:54:04 that was nice Feb 17 12:54:15 no oversteer, no understeer, just rock solid Feb 17 12:54:35 if i'm buying one, i'm going to put on a different turbo and chiptune it Feb 17 12:54:41 that gets you like 125bhp Feb 17 12:54:45 in a 750kg car Feb 17 12:55:22 wow... quite a good hpp/kg ratio! Feb 17 12:56:00 yep Feb 17 12:56:13 and if i don't like the smart roadster after all, i'm thinking about an audi tt Feb 17 12:56:17 german cars for me :) Feb 17 12:57:00 good decision! :) Feb 17 12:57:02 you know rommerskirchen? Feb 17 12:57:07 no Feb 17 12:57:19 Germany, NRW (= Nordrhein-Westfalen), 41569 Rommerskirchen (between Cologne, Feb 17 12:57:22 Neuss, Aachen, Venlo) Feb 17 12:57:27 that's where i'm going saturday Feb 17 12:58:09 oh, I found it... a bit in the middle of nowhere! ;) Feb 17 12:58:32 there lives a guy who's selling a smart Feb 17 12:58:47 it looks good, but there's a little shady side to the story Feb 17 12:59:05 he says he bought it new, but i'm sure i've seen the car on ebay before Feb 17 12:59:19 so, he's gotta clear that up with a good coherent story Feb 17 13:00:45 [g2]: I'm getting this: ERROR: Nothing provides hotplug-ng Feb 17 13:01:08 I did create my symlinks. I must be confused as usual. Feb 17 13:01:13 sounds like you did a pull and forgot to bk -r co -q or something Feb 17 13:01:21 does openembedded/packages/hotplug-ng/hotplug-ng*.bb exist? Feb 17 13:01:42 wow! It's been a while! I totally forgot about that! Feb 17 13:02:03 giel, sounds a bit strange... and be careful: the guys in that area (near Bergheim) are the craziest car-drivers ever... never come to close to someone with a BM license plate! ;) Feb 17 13:02:18 ok, i'll remember Feb 17 13:02:44 i'm taking someone with me who speaks german and that guy speaks english Feb 17 13:02:49 kergoth, I think I haven't done that in like a month! dope! Feb 17 13:02:53 and if i don't like it i'll just get myself another car ;) Feb 17 13:03:34 sounds reasonable! Feb 17 13:04:37 i was looking at this beefed up audi tt quattro Feb 17 13:04:44 four wheel drive, 270bhp Feb 17 13:05:59 wow... and, was it a good one? Feb 17 13:06:20 well, it was a bit too expensive Feb 17 13:06:41 not the purchase, but insurance was pretty heavy Feb 17 13:06:51 Sorry guys: I need advice about unslung. GNU Parted needs a lib from e2fsprogs, but the package is broken. I mailed him (Christopher Blunck) a week ago, but got no answer. Can I just make alterations now, you think? Feb 17 13:07:18 bob_tm: are you the one who did that commit that added libiconv to the parted deps? Feb 17 13:07:27 Yes Feb 17 13:07:31 you dont want that. Feb 17 13:07:35 glibc provides iconv functionality. Feb 17 13:07:45 libiconv is only ever needed if you're using a C library that doesnt provide it Feb 17 13:07:51 like uclibc with NLS support disabled Feb 17 13:07:58 Ok. Feb 17 13:08:21 giel, yeah, TT's tended to crash very often, so the insurance will be a bit higher I guess. Feb 17 13:08:23 I will remove it again. It doesn't change the e2fsprogs issue, though Feb 17 13:08:28 * kergoth nods Feb 17 13:08:36 just wanted to make sure you were aware Feb 17 13:08:43 I wasn't - thanks Feb 17 13:09:02 siddy: well, i'll let you know if i bought something... gotta sleep now, tired already Feb 17 13:09:38 oh, you'll miss the victory of aken! ;) Feb 17 13:09:41 kergoth: I will make the changes to e2fsprogs and he can always revert if and when he comes back Feb 17 13:10:10 that's probably best Feb 17 13:10:31 siddy: it probably stays 0-0, there are only a couple of minutes left, aren't there? Feb 17 13:10:50 kergoth: Thanks Feb 17 13:11:16 man, you're taking my last hopes... but yeah, just 10 minutes left. :-\ Feb 17 13:11:50 siddy: i'm gonna prepare for bed and check in in 10 mins then :) Feb 17 13:12:11 hehe, okay! Feb 17 13:13:27 http://openembedded.org/cgi-bin/moin.cgi/Buildsystems_2fRulesOfThumb .. if anyone has anything to add, do so Feb 17 13:14:30 <[g2]> kergoth, I think hotplug is a little happier Feb 17 13:15:14 <[g2]> when I modprobe'd ehci_hcd, the other modules were automatically loaded and the disk was recognized Feb 17 13:15:27 ah Feb 17 13:15:29 excellent Feb 17 13:15:34 so i just need to get us a startup script Feb 17 13:15:54 good show! Feb 17 13:16:32 <[g2]> I'm adding the modprobe for a couple drivers in the init script and I'll see if it can find the drive Feb 17 13:16:33 for now, obviously you could just echo ehci_hcd >> /etc/modules :) but yeah, actual hotplug event synthesis for coldplug needs to happen Feb 17 13:19:26 <[g2]> kergoth, last time I did that, and this time it's just ignored Feb 17 13:20:20 /etc/modules is a list of modules that /etc/init.d/modutils loads Feb 17 13:20:26 probably just need that startup script Feb 17 13:21:30 keep in mind that although it should be modprobe.conf it's still called modules.conf Feb 17 13:22:31 <[g2]> well all the drivers aren't getting loaded, and the three manually loaded via modprobe are Feb 17 13:22:41 i'm not talkinga bout that. Feb 17 13:22:48 /etc/modules != /etc/modutils.conf or /etc/modules.conf Feb 17 13:22:51 they're completely independent. Feb 17 13:23:22 <[g2]> ixp425_eth, usbcore, ehci_hcd and it boots in 43 seconds with the ethernet up and the hd available Feb 17 13:23:27 ok. I was wondering why there was so many. Feb 17 13:23:42 [g2]: nice Feb 17 13:23:56 I'm stuck loading ohci too as my bluetooth dongle is usb 1.1 Feb 17 13:24:27 <[g2]> kergoth I was talking about the boot speed improvement not the root issue with the modules -- I really do appreciate your help and know how you hate hacks Feb 17 13:24:38 nod Feb 17 13:24:50 how long was the boot time before? Feb 17 13:25:01 must have shaved a good 20 seconds no? Feb 17 13:25:07 and how much longer is it, if at all, if you load ohci as well? Feb 17 13:25:27 giel, okay you were right. but we'll beat you next week! ;) Feb 17 13:25:28 I'll report on my bloated robot testing later tonight. Feb 17 13:25:34 cool Feb 17 13:26:05 if I can get a 20-30 seconds boot I'd be more than happy. Feb 17 13:26:06 and i wonder how much longer itll get if we add in any necessary hotplug event synthesis (coldplug) Feb 17 13:26:06 siddy: heh, no way, when we're playing home we're unbeatable :P Feb 17 13:26:15 though that could be optional Feb 17 13:26:36 I'll be removing the csr junk in my case so that's like 5-8 seconds. Feb 17 13:26:59 giel, just wait and see! Feb 17 13:27:22 we still need to do some real profiling on the boot time. Feb 17 13:28:00 yeah, the instrumented printk thing didn't help much. Need to do the same thing for user scrips. Feb 17 13:28:06 right Feb 17 13:28:09 sysvinit instrumentation Feb 17 13:28:19 some echos in /etc/init.d/rc and /etc/init.d/rcS may be sufficient. Feb 17 13:28:26 okay, I'm off, good night everyone! Feb 17 13:28:50 yep. Feb 17 13:29:38 Anal as I am I'll probably end up trying that even once my boot is 20-30 seconds. More speed! Feb 17 13:30:13 <[g2]> VoodooZ_Work, you'll probably want APEX Feb 17 13:30:24 what's that again? the bootloader? Feb 17 13:30:29 <[g2]> nod. Feb 17 13:30:53 gah Feb 17 13:30:56 yeah, but that's a bit scarier without jtag. Feb 17 13:30:57 i need to check my voice mail more often Feb 17 13:31:03 "you have .. twenty three .. new messages" Feb 17 13:31:11 <[g2]> :) Feb 17 13:31:26 * [g2] misses the big corp. environ Feb 17 13:31:41 <[g2]> not! :) Feb 17 13:32:15 * VoodooZ_Work doesn't Feb 17 13:32:32 (aside for the money and benefits!) Feb 17 13:32:38 heh Feb 17 13:32:39 but now that's all gone so... Feb 17 13:33:07 my bonus from ti this year will be 10.5k Feb 17 13:33:14 nice Feb 17 13:33:14 it was like 15% of the '04 annual income Feb 17 13:33:15 03ingeba * 10unslung/sources/parted/control: Removed iconv as dependency - taken care of by the C library Feb 17 13:33:30 supposed to be anyway, its less for some reason i dont understand Feb 17 13:33:32 ah well Feb 17 13:33:52 I'm looking into getting in a small aerospatial company here called Neptec and they are working with NASA to decommision the Hubble. Feb 17 13:33:56 damn those dutch taxes! Feb 17 13:34:25 03ingeba * 10unslung/make/parted.mk: Removed iconv from staging Feb 17 13:34:29 when i get a bonus, taxes take 48% Feb 17 13:34:35 Should be way more interesting that fixing computers for a bunch of whining teachers Feb 17 13:35:49 <[g2]> morning dyoung Feb 17 13:36:19 I'm sorry you think theres too many unslung specific stuff in switchbox-4.7 Feb 17 13:36:27 I didnt think it was that bad... Feb 17 13:36:51 the borrowing of lib from the jffs2 I felt was the biggest offender. Feb 17 13:37:09 <[g2]> looking for the .unslung Feb 17 13:37:20 <[g2]> loading the ixp drivers Feb 17 13:37:27 getting the ixp drivers from payload Feb 17 13:38:17 <[g2]> dyoung, kergoth has a new hotplug based on C programs Feb 17 13:38:35 ... Feb 17 13:38:44 <[g2]> it boots openslug in 42 seconds from + to # after the first time which only takes 1 minute 11 seconds Feb 17 13:38:46 Right, but thats in 3.9 also. Feb 17 13:39:15 and the ixp drivers are ONLY loaded if youre in a specific unslung mode that openslug can ignore. Feb 17 13:39:49 <[g2]> well we either need to change the default kernel config or load some other drivers Feb 17 13:39:54 Right. Feb 17 13:39:59 I'm not disagreeing. Feb 17 13:40:27 And if we're going to talk about .unslung, why dont we talk about /bin/init too Feb 17 13:40:34 shrug Feb 17 13:41:15 <[g2]> once we get the module init straightened out, then it's probably a lot easier and cleaner just to use the rootfs modules Feb 17 13:41:34 <[g2]> but I digress Feb 17 13:42:21 anyways, i have more immediate issues. Feb 17 13:42:36 <[g2]> slug related or other ? Feb 17 13:42:39 like what the hell changed in the firmware image creation in the past 5 days that make stuff not work. Feb 17 13:42:55 my JFFS2 header is a redbootheader for instance. Feb 17 13:43:00 <[g2]> hmmmm, worked for me Feb 17 13:43:01 [g2], both. Feb 17 13:43:04 <[g2]> on OpenSlug Feb 17 13:43:55 and I notn convinched that I didnt blow something important away... Feb 17 13:44:24 <[g2]> GL Feb 17 13:44:33 <[g2]> I wish I could help more thre Feb 17 13:44:35 <[g2]> there Feb 17 13:50:03 giel: getting ixp drivers from payload is not unslung-specific - it's just the openslug doesn't load the payload with it's drivers ... Feb 17 13:50:32 and the loading of the drivers is then not unslung-specific either (once you are putting the right drivers in the payload) Feb 17 13:51:08 the .unslung is definitely unslung-specific, and the intention would be to add to that to detect a valid openslug drive too, or just remove the check. Feb 17 13:51:38 <[g2]> as is the whole .xxxx convention Feb 17 13:51:46 yep Feb 17 13:51:51 suggestions? Feb 17 13:52:28 Remember too that those bits were pilfered out of a *really* old linuxrc. Feb 17 13:52:32 switchbox is meant to be distro independent (or at least to support the multiple nslu2-linux distros) Feb 17 13:52:43 <[g2]> well for APEX beewoolie has the suggestiong of holding reset down and using the LEDs as indicators to select different loads from by pushing the power button Feb 17 13:52:50 they wre used because they were there. I'm sure there are better ways to detect a disk rootfs Feb 17 13:52:59 so anything we need to do to make that so we will do ... Feb 17 13:53:01 <[g2]> We could possibly put something similar in Redboot Feb 17 13:53:27 not for camps 1-3 we cant ... Feb 17 13:53:41 <[g2]> nod. Feb 17 13:54:02 bbl Feb 17 13:54:20 <[g2]> I'd like to be able to boot directly to sda/sdb Feb 17 13:54:26 anyway, as soon as you replace redboot (for camp 4), then you probably shouldn't use switchbox anymore. Feb 17 13:55:01 <[g2]> Right now I'm playing with booting directly to the jffs2 partition Feb 17 13:55:11 if you made APEX go around switchbox (directly to whatever is chosen via power button and leds), and allow stock redboot to go into a switchbox which works for both unslung and openslug, then we are good I think Feb 17 13:56:46 <[g2]> I'm not sure where the OpenSlug ppl want to go. Feb 17 13:56:57 unless we make a conscious decision to restrict openslug to people who are willing to replace their bootloader, then we need to make sure that openslug still works properly with stock redboot (which doesn't necessarily mean that openslug has to use switchbox). Feb 17 13:57:09 <[g2]> 3 of the 4 (other than me are running totaly non-standard configs) Feb 17 13:57:25 [g2]: we still need to get a beta out, so that we can gauge the community direction for openslug Feb 17 13:57:28 a three characters to the crosstool and crosstool-native package source can enable Objetive-C language. can someone verify ? Feb 17 13:57:41 a three characters change I mean. Feb 17 13:57:45 you can't base it on the alpha testers, because by definition they usually have custom configs Feb 17 13:57:56 <[g2]> rwhitby, we are very close to getting out of beta with ka6sox the list is at 16 Feb 17 13:58:13 nslu2-cross335.sh:GCC_LANGUAGES="c,c++,objc" Feb 17 13:58:35 <[g2]> I think booting to the hd/flash stick and having the jffs2 as a fall back makes a lot of sense Feb 17 13:58:37 tjyang: if jacques gives the nod, then we will include it Feb 17 13:58:54 just mount the hd over the jffs2 / using unionfs :) Feb 17 13:58:54 <[g2]> recovery is simple as pulling the flash stick / hd Feb 17 13:58:55 just add objc in GCC_LANGUAGES variable in nslu22-cross335.sh. Feb 17 13:59:09 <[g2]> also, it could boot jffs2 by holding down the reset Feb 17 13:59:09 kergoth: yeah Feb 17 13:59:26 03ingeba * 10unslung/sources/e2fsprogs/control: Upped version and took over as maintainer at least for a while Feb 17 13:59:33 here is my quick notes http://mediawiki.gnustep.org/index.php/Platform_compatibility#Linksys_NSLU2 Feb 17 13:59:41 should test the unionroot scripts on openslug. Feb 17 13:59:46 should work fine, they work on the Z Feb 17 13:59:52 <[g2]> kergoth, I don't know if we have unionfs support that that makes some sense if we can mount it over / Feb 17 14:00:43 unionfs works fine in every environment i've tested Feb 17 14:00:48 omap boards, desktop, zaurus Feb 17 14:01:11 tjyang: what are the ipk errors in the second step of your gnustep procedure? Feb 17 14:01:28 rod, I am stilling having problem to have ipkg on linux host machine to package all the complete files into 7 split ipk packages. Feb 17 14:02:07 [g2]: is openslug at a stability and feature level where you think it is ready for a public beta (it's a genuine question, not a dig - I haven't been keeping up with it in the last two weeks) ? Feb 17 14:02:38 I ended up use the tar/copy over method to have libobjc.so ... etc installed on my nslu2. Feb 17 14:02:39 <[g2]> rwhitby, It is very close, the punchlist is small Feb 17 14:03:19 tjyang: we'll need to fix those errors before accepting the patch - those symlinks are important, and we can't accept a patch until the ipks install cleanly Feb 17 14:03:29 s/we'll/you'll/ Feb 17 14:03:36 <[g2]> the two big issues are the hotplug, which kerogth has gone a long way to fixing today Feb 17 14:03:37 ;-) Feb 17 14:04:03 <[g2]> and the ability to boot to the hd/flash or overlay the rootfs Feb 17 14:04:24 [g2]: I am looking forward to using OpenSlug to get my new bluetooth dongle running so I can route through the slug for Treo 650 internet connectivity at home ... Feb 17 14:04:25 unionroot will work, its just a question of how the user configures it Feb 17 14:04:40 03ingeba * 10unslung/make/e2fsprogs.mk: Many changes: Rewritten using new template, added missing includes in staging etc. Hardly tested - mainly done for parted, since it provides libuuid Feb 17 14:04:46 <[g2]> a third issue is switchbox compatibility, 3.9 boots OpenSlug, 4.7 doesn't and needs fixes Feb 17 14:04:58 [g2]: do you really want to put unionroot in now (when you are close to a beta release), or hold that over for OpenSlug 2.x ? Feb 17 14:05:26 <[g2]> I don't see that it has to go in now. Feb 17 14:05:33 [g2] right now openslug is supposed to be using switchbox-3.9. Feb 17 14:05:40 [g2]: now that giel has solved the switchbox roadblock, we can get back into active development mode on it and make it work for both openslug and unslung Feb 17 14:05:45 <[g2]> I do see the ability to boot to flash stick /hd as important though Feb 17 14:05:46 the 4 series is supposed to still be under development. Feb 17 14:06:05 [g2]: agreed that we should get switchbox 4 into openslug before the beta release Feb 17 14:06:08 <[g2]> giel is running uClibc and a very hacked environent Feb 17 14:06:22 <[g2]> he's totaly non-standard Feb 17 14:06:32 any method of running the hd as / will need integration work Feb 17 14:06:34 yes, but he gave dyoung and I the key info to fix switchbox reboot and led problems Feb 17 14:06:35 whether it be union or not Feb 17 14:07:04 <[g2]> the /dev/ram0 and init=/linuxrc ? Feb 17 14:07:07 kergoth: we've got it working for unslung in switchbox 4, so it should be trivial to make it work for openslug as well. Feb 17 14:07:11 [g2]: yep Feb 17 14:07:23 unslung will move to that command line style too Feb 17 14:07:32 <[g2]> That's what I working on for OpenSlug Feb 17 14:07:40 the point is, the same testing will have to be done either way. scripts exist for both Feb 17 14:07:52 agreed. Feb 17 14:07:53 I already p ushed that change for unslung-4.x Feb 17 14:07:57 <[g2]> it's as easy as changing rootfs to ext2 and loading it as a ramdisk with /linuxrc Feb 17 14:08:13 <[g2]> 4.7 didn't boot OpenSlug today Feb 17 14:08:26 <[g2]> I wind up in the /bin/sh prompt Feb 17 14:08:37 <[g2]> that was my first test for it Feb 17 14:08:38 ok, let's debug and fix that then ... Feb 17 14:08:39 I said unslung. Feb 17 14:09:17 here's my proposed plan of action (comments welcome): Feb 17 14:09:38 1/ dyoung and I get switchbox 4 stable and integrated into unslung Feb 17 14:10:02 2/ [g2], dyoung, and I (and anyone else) mod switchbox 4 to make it work for openslug too Feb 17 14:10:02 03ingeba * 10unslung/Makefile: Added proftpd, e2fsprogs and parted to cross testing. May compile native - can't test Feb 17 14:11:01 3/ we focus on an Unslung 4.x-alpha release (with just switchbox 4.x and no other new features) and an OpenSlug 1.x-beta release (with both switchbox 4.x and apex 1.0 support and no other new features) Feb 17 14:11:23 (with openslug 1.x-beta having the higher priority) Feb 17 14:11:33 4/ there is no step 4 Feb 17 14:11:40 comments? Feb 17 14:12:15 <[g2]> I'm concerned that 2/ is a waste of time and resources Feb 17 14:12:45 [g2]: why (assuming that openslug is still for camp 3 as well as 4) ? Feb 17 14:12:48 <[g2]> the issue is all the switchbox baggage gets pushed into OpenSlug Feb 17 14:13:01 <[g2]> here's why Feb 17 14:13:27 (and also assuming that we don't impose any of the unslung-specific stuff on openslug) Feb 17 14:13:34 <[g2]> And BTW i'm just being honest here, not trying to point finger, I love and respect you guys! Feb 17 14:13:58 yep, this is a robust technical discussion, no finger pointing or angst :-) Feb 17 14:14:39 <[g2]> my reasoning is that the standard Redboot Feb 17 14:15:15 <[g2]> Rather than switchbox, we can load more full fledged ramdisk Feb 17 14:15:44 <[g2]> that has drivers either directly or in the jffs2 partition in it Feb 17 14:16:21 <[g2]> it's way cleaner, and the real benefit comes from a proper init= (/linuxrc) Feb 17 14:16:49 <[g2]> the jffs2 partition can either be booted to or used for user resident data Feb 17 14:17:31 did you mean ramdisk when you said ramdisk five lines up? Feb 17 14:18:09 <[g2]> initrd Feb 17 14:18:24 how does an initrd have a jffs2 partition it i? Feb 17 14:18:30 ... in it? Feb 17 14:19:06 <[g2]> the initrd can boot to the jffs2 partition or the hd/flash stick Feb 17 14:19:08 rod, either the ipkg or my ipkg on my RedHatAS has problem. it is failing to compress all the files under build/crosstool-native-0.28-rc73-ipk* directory. Feb 17 14:19:10 you've confused me - can we use more precise terminology? Feb 17 14:19:19 <[g2]> it can also just load the jffs2 as a mount point Feb 17 14:19:24 (that was to [g2] Feb 17 14:19:36 tjyang: I can't help you with that problem ... Feb 17 14:19:50 [g2]: how big with the initrd be? Feb 17 14:20:50 (i.e. is a fully bootable and useable rootfs, or just a switchbox type thing which is expanded to include drivers ?) Feb 17 14:20:54 <[g2]> the other possiblity that is possible is append it on the end of the kernel and let the size of the kernel partition float Feb 17 14:21:43 <[g2]> mostly switchbox like, drivers and their location is the issue Feb 17 14:21:52 that is fine. but I think it is ipkg thas issue. because I didn't modifiy anything except putting objc into GCC_LANGUAGE variable. Feb 17 14:21:55 [g2]: what's your proposal again, can you spell it out precisely? Feb 17 14:22:28 tjyang: still your problem to solve, cause it works at the moment for everyone else Feb 17 14:22:38 if you guys don't have this problem that is ok. Feb 17 14:23:06 tjyang: were you running ipk from the root dir? Feb 17 14:23:25 I am not going to solve this problem I am going to work around it by using tar. Feb 17 14:23:38 no, I am using regular user tjyang. Feb 17 14:23:57 tjyang: then the change cannot be accepted if it breaks ipk for everyone ... Feb 17 14:24:07 tjyang: root dir, not root user Feb 17 14:24:11 does ipkg need to be run from root directory? Feb 17 14:24:33 <[g2]> rwhitby, I see three proposals on the table right now: Feb 17 14:24:54 putting objc into GCC_LANGUAGE will break ipkg ? Feb 17 14:24:55 <[g2]> 1/ make switchbox 4.x work with openslug which you suggested Feb 17 14:25:03 Looking at the errors on the page that tjyang referred to, it might for the crosstool-native package. that would be a bug in the crosstool-native package if it does Feb 17 14:25:16 tjyang: I don't know - that's what I want you to determine Feb 17 14:25:33 it is ridiculous. Feb 17 14:25:57 <[g2]> 2/ stub the rootfs partition where switchbox currently is and put a jffs2 parition right after it that allows the user some tayloring of the boot Feb 17 14:26:01 tjyang: you have a broken ipk install of crosstool-native for some reason. you have to prove that it's not the change causing it. Feb 17 14:26:05 I have this problem before and I remember I reported the problem on nslu2 yahoo group. Feb 17 14:26:56 that is why I am asking if someone can veriy/compile and generate a gcc with objc supoprt. Feb 17 14:27:11 <[g2]> 3/ expand the current switchbox to allow booting to either the jffs2 partition, the hd/flashdisk, or using the jffs2 as a mount point / partial overlay Feb 17 14:27:51 <[g2]> Another consideration is the actual rootfs layout and migration of all the packages for the Users Feb 17 14:28:03 on my system(RHAS), ipkg has problem to compress/package big directories/fiiles. Feb 17 14:28:17 [g2]: migration? Feb 17 14:28:38 <[g2]> kergoth, migration of sorts Feb 17 14:29:04 migration of what to what? Feb 17 14:29:07 * kergoth outta the loop Feb 17 14:29:12 <[g2]> I'd guess 85% of the stuff being used by Unslung builds and runs straight in OE Feb 17 14:29:29 <[g2]> Perl doesn't and a couple other things Feb 17 14:29:31 those packages should be added to oe Feb 17 14:29:48 i dont expect there to be real package compatibility ever, personally Feb 17 14:29:51 <[g2]> I understand and agree Feb 17 14:30:03 <[g2]> hence the "migration" Feb 17 14:30:07 ah, gotcha. Feb 17 14:30:16 <[g2]> things get much simpler with OE Feb 17 14:30:18 perl out of oe should work fine today Feb 17 14:30:27 afaik anyway Feb 17 14:30:29 <[g2]> when was it fixd ? Feb 17 14:30:40 <[g2]> it's been broken for weeks Feb 17 14:31:33 built fine last time i tried to build it Feb 17 14:31:42 havent tried armeb in a while Feb 17 14:32:12 <[g2]> fixing Perl would be awesome because isn't AppWeb based on it ? Feb 17 14:32:29 no Feb 17 14:32:39 appweb is a C application Feb 17 14:33:12 <[g2]> well having Perl would still be awesome anyway, it should be able to be built cross Feb 17 14:33:20 it _can_ be built cross Feb 17 14:33:32 all the cross stuff in perl upstream came from a guy who got it building in the old openzaurus buildsystem Feb 17 14:33:36 heh Feb 17 14:34:00 jacques: you've built perl more recently than i. whats the situation? Feb 17 14:34:09 Did someone fix up a arm5 config that works on the nslu2? Feb 17 14:34:13 it built OK but dynamically loaded modules did not work Feb 17 14:34:26 ooh... Feb 17 14:34:30 I tried every trick I knew Feb 17 14:34:30 ah. Feb 17 14:34:43 odd. Feb 17 14:34:54 <[g2]> dyoung, arm5 config ? Feb 17 14:35:08 Read the perl cross build instructions. Feb 17 14:35:23 you can build it cross if you have a config file. Feb 17 14:35:30 and I was using slightly modiified config.sh from native builds Feb 17 14:35:30 <[g2]> thx. Feb 17 14:35:33 but to get the config file you need to run the confurator native. Feb 17 14:35:48 heh Feb 17 14:35:56 ~/code/user/oe/openembedded$ ls packages/perl/perl-5.8.3/ Feb 17 14:35:56 ./ config.sh-arm-linux.patch libperl-5.8.3-create-libperl-soname.patch Makefile.SH.patch Feb 17 14:36:27 [g2]: there's too much other stuff going on in this channel to have a sensible strategy discussion Feb 17 14:36:34 <[g2]> you know I'd like to try running some emulators like qemu locally to get some of that info Feb 17 14:38:15 [g2], perl is so integrated with the OS that you really need a full install to configure it properly Feb 17 14:38:58 03ptweety * 10unslung/ (17 files in 10 dirs): initial checkin of some perl modules. Feb 17 14:39:46 <[g2]> jacques, maybe. But perl is a USER app Feb 17 14:40:28 can anyone here try this in GCC_LANGUAGES="c,c++,objc" in your nslu2-cross335.sh to enable objc ? this will enable the entrance of gnustep apps. Feb 17 14:41:03 my ipkg is broken. Feb 17 14:41:22 and it is not becuase of objc modification. Feb 17 14:41:26 03ptweety * 10unslung/make/perl-digest-hmac.mk: initial checkin of some perl modules. Feb 17 14:41:29 rwhitby, as we discussed some hours ago I've prepared some perl modules. What do you think? Feb 17 14:48:13 tjyang: will try out that change tonight (12 hours time) unless someone else gets to it before then Feb 17 14:48:22 question to anyone: is there any adverse effect of having two files with the same number in /opt/etc/init/d/ (say S52foo and S52bar) ? Feb 17 14:48:26 pTweety: put them in TESTING and we'll try them out Feb 17 14:48:40 *init/d = init.d Feb 17 14:49:09 rwhitby: would you mind adding some more detail about how startup scripts / booting works in unslung to the oe packages on unslung page? Feb 17 14:49:21 rwhitby: i've started fixing all the packages to obey all the target paths. Feb 17 14:49:27 figure thats up next on the todo Feb 17 14:49:40 yep - that will get non-daemon packages going ... Feb 17 14:49:49 * kergoth nods Feb 17 14:50:13 we also need a means of overriding that on a package-by-package basis, for the couple of packages that we do want to install in the base firmware under / instead of under /opt Feb 17 14:50:22 (e.g. ipkg and wget ...) Feb 17 14:50:31 i'll write a quick .inc/.bbclass for that. wont take more than 5 minutes Feb 17 14:50:38 the startup script one is the more annoying Feb 17 14:51:00 thanks rod. are your using debian to build crosstool ? Feb 17 14:51:47 yes Feb 17 14:51:58 debian 3.0 or 3.1 ? Feb 17 14:52:26 3.1 Feb 17 14:52:45 03ptweety * 10unslung/Makefile: several perl modules ready for native testing Feb 17 14:53:09 good, I am going to do the builds using same os as yours instead RHAS. Feb 17 14:54:16 we will be publishing a specific OS and packages for the official nslu2-linux build environment in the next week Feb 17 14:54:25 (as we build the new cross compilation machine) Feb 17 14:54:36 thanks. rod. Feb 17 14:54:48 night all. I'll be off for some days but will read the logs to check comments concering the perl modules Feb 17 14:55:14 I am just exhausted between works,family and hobby(nslu2/gnustep). Feb 17 14:55:19 * ByronT is back -( work )- gone 2 hr 9 min 53 s Feb 17 15:01:20 rwhitby-away: let me know when you've got that startup script info somewhere. i'd really like to get the oe packages working on unslung, reduce the amount of duplicated effort between the teams. Feb 17 15:19:57 I am building crosstool/crosstool-native with objc again on debian 3.1 box. hopefully ipkg on debian is not brokern. Feb 17 15:20:11 with objc -> with objc enabled. Feb 17 15:20:45 here is my note. http://mediawiki.gnustep.org/index.php/Platform_compatibility#Linksys_NSLU2. Feb 17 15:21:05 You are welcome to work on this document. Feb 17 15:21:15 <[g2]> ~nslu2-camps Feb 17 15:21:17 it has been said that nslu2-camps is 1: want full linksys compatibility (including future releases), can load image through web and telnet in to run ipkg, have no clue about redboot, 2: can telnet into redboot and use linux, 3: power user, but no jtag and no custom redboots, 4: bleeding edge - custom redboots, no need or desire for linksys compatibility Feb 17 15:50:11 <[g2]> kergoth ping Feb 17 15:50:18 [g2]: pong Feb 17 15:51:02 <[g2]> on switching the rootfs, can I just change the real-root-dev ? Feb 17 16:34:01 03jp30 * 10unslung/Makefile: core xlibs are ready for testing Feb 17 16:51:09 03jp30 * 10unslung/make/gtk+.mk: gtk+: fix Depends: syntax requiring libjpeg Feb 17 18:33:35 is there a new unslung on the way (meaning should I get the README updates done ASAP)? Feb 17 19:17:02 [g2]-away: ping Feb 17 19:17:32 rwhitby-web, README and NOTES updated and ready for next release Feb 17 19:18:42 in bk? Feb 17 19:18:46 yes Feb 17 19:18:57 just checked to make sure Feb 17 19:19:01 did you send the changeset to nslu-changesets@lists.sf.net ? Feb 17 19:19:28 (using "bk send -r nslu-changesets@lists.sf.net" after you have subscribed to that list on sf.net nslu project) ? Feb 17 19:19:45 ah... that was going to be next question... Feb 17 19:21:41 that's how we notify each other (in addition to the CIA messages in channel) about updates to the bk repo Feb 17 19:22:13 I'm getting an error after a bk push re CIA Feb 17 19:25:32 you can still bk send the changeset even after you have pushed (use bk history I think to make sure the number didn't change on the push) Feb 17 19:26:13 dyoung-web: did you catch the [g2] three options earlier re switchbox? Feb 17 19:26:43 Feb 17 12:24:32 <[g2]> rwhitby, I see three proposals on the table right now: Feb 17 19:26:52 Feb 17 12:24:54 <[g2]> 1/ make switchbox 4.x work with openslug which you suggested Feb 17 19:27:02 Feb 17 12:25:56 <[g2]> 2/ stub the rootfs partition where switchbox currently is and put a jffs2 parition right after it that allows the user some tayloring of the boot Feb 17 19:27:13 Feb 17 12:27:10 <[g2]> 3/ expand the current switchbox to allow booting to either the jffs2 partition, the hd/flashdisk, or using the jffs2 as a mount point / partial overlay Feb 17 19:27:30 isnt 3 == 1 ? Feb 17 19:27:49 (assuming we make it work the way its supoposed to) Feb 17 19:28:04 That's what I thought (apart from the partial overlay bit at the end, which would be an OpenSlug 2.x thing anyway I would think) Feb 17 19:28:10 In my case I'd be happy just booting off the jffs2. Feb 17 19:28:17 I don't understand #2 ... Feb 17 19:28:48 [g2]-away said it worked today but strangely mine crashes with "Kernel panic - not syncing: No init found. Try passing init= option to kernel." Feb 17 19:28:57 VoodooZ_home, you would need to change your bootloader to do that; but I can veify that you CAN boot direct to the jffs2 with Apex as your bootloader Feb 17 19:29:22 what? [g2] did it today? Feb 17 19:29:34 No, *I* did it last weekend. Feb 17 19:29:36 He's the one who said I could bypass it. Feb 17 19:29:43 dyoung-web: I thought that the options were hard-coded in our kernel, and had no connection to bootloader? Feb 17 19:29:46 But I flashed it in there with JTAG. ;-) Feb 17 19:29:59 (the default command line) Feb 17 19:30:02 anyways, I just want to speed up my bootup by removing the fluff. Feb 17 19:30:27 Hmm, good point, ATAGis ignored. Feb 17 19:30:32 Youre probably right then, Feb 17 19:30:33 well, he was doing something else that didn't require a reflash of the bootloader. Feb 17 19:30:52 Voodooz, what cmdline did you use? Feb 17 19:30:59 let me check... Feb 17 19:31:03 bleh Feb 17 19:31:04 VoodooZ_home: you should be in camp #5 (using APEX) ..... (or is APEX just the implementation for camp #4?) Feb 17 19:31:24 I don't mind but I don't have jtag so... Feb 17 19:31:32 kergoth: bleh indeed - if only they hadn't crippled the command line changing support in redboot .... Feb 17 19:32:15 ok, I changed it from: CMDLINE_ROOT = "root=/dev/slug rootfstype=ext2,jffs2 initrd=0x01000000,10M mem=32M@0x00000000" to CMDLINE_ROOT = "root=/dev/mtdblock4 rootfstype=jffs2 mem=32M@0x00000000" in openslug-kernel_2.6.9.bb Feb 17 19:32:23 rwhitby-web: I'm still thinking about what option 2 means... Feb 17 19:33:12 Actually, I had the init=/sbin/init option in there too but I just removed it to try again Feb 17 19:33:28 Any idea? Feb 17 19:33:37 what does it say one line above that? Feb 17 19:33:55 (or do you have a bootlog you can mail me....) Feb 17 19:35:24 dyoung: talking to me? Feb 17 19:36:09 sorry, I'm multitasking (badly!) Feb 17 19:37:48 is [g2] talking about a configuration block for (2) ? Feb 17 19:38:19 That's the line above: KERNEL_IMAGETYPE = "zImage" CMDLINE_CONSOLE ?= "ttyS0,115200n8" Feb 17 19:38:57 VoodooZ_home: Actually, I mean in the boot log... Feb 17 19:39:21 not much to see. but I'll paste it in #flood for you Feb 17 19:39:51 can you mail it to me? I cant get to #flood right now Feb 17 19:40:08 ok. What's your email. Feb 17 19:40:39 dyoung: those three numbered lines were sequential in the logs Feb 17 19:41:08 oh, you're talking about VoodooZ_home's boot log, not [g2]'s messages in the irc log ... Feb 17 19:41:40 dyoung: s/stub/stuff/ and the line makes a bit more sense Feb 17 19:43:24 stuff makes sense. Feb 17 19:43:37 I tried snub; but that didnt quite work.,.. Feb 17 19:43:45 but I don't see the advantages of #2 Feb 17 19:44:36 if we didn't want to support recovery mode and maintenance mode, then I would say we could get rid of switchbox altogether and just boot straight to jffs2 and let it pivot_root anywhere else .... Feb 17 19:45:14 but to support Unslung web upgrade, we need a means of *not* mounting the jffs2, but running what is in the jffs2. Feb 17 19:45:52 (like we do today by creating a big ramdisk, copying the jffs2 contents into it, and real-root-dev'ing to it Feb 17 19:47:02 Right, Feb 17 19:47:39 so for recovery mode and maintenance mode, we absolutely require an initrd, right? Feb 17 19:47:43 heh, you can still do that, you'd just have to set up the jffs2 to know how to pivot onto a temporary ramdisk during the upgrade. Feb 17 19:47:57 if you pivot off the jffs2, it can be unmounted. Feb 17 19:48:09 true Feb 17 19:48:31 it seemed easier to do it in switchbox, cause then we didn't have to mess with the Unslung rootfs Feb 17 19:49:34 having a ramdisk makes it a little more idiot-resistant too; since you have a smaller chance of blowing it away. Feb 17 19:49:57 right, the user can't stuff up our maintenance mode by changing things in the jffs2 rootfs Feb 17 19:50:00 you could accidentally blow away your jffs2 /linuxrc and be unbootable. Feb 17 19:50:06 idiot-resistant - did someone call for me? Feb 17 19:51:21 which reminds me - we need a flag file which is written to the jffs2 at the end of switchbox, and then removed at the end of a good boot, and which is also checked for existence at the start of switchbox and causes a drop into recovery mode if it is found. Feb 17 19:52:04 then a bad change in the jffs2 rc scripts that stop a good boot will cause a drop into recovery mode after power cycling Feb 17 19:53:46 so anyway, I think we push ahead on unslung 4.x, and modify switchbox 4.x to boot openslug 1.x too (all with the current partition table layouts and root filesytem layouts), and defer any discussion of changing partition tables and stuff for Unslung 5.x/OpenSlug 2.x Feb 17 19:54:31 Ok Feb 17 19:54:42 and allow APEX to bypass switchbox if it so desires by booting straight from the jffs2 Feb 17 19:55:02 rwhitby-web: when will you be in a position to test build unslung-4 ? Feb 17 19:55:15 I built it, but haven't flashed it Feb 17 19:55:58 I'm currerntly stuck, and I before going on I want to know if its something about my hardware/flash or if something changed in the build, or ...? Because when I take it apart with slugimage, everything is in the right place. Feb 17 19:56:18 byteswapped, but I think thats od's fault. Feb 17 19:57:11 Once gastro is doing the builds properly, I can boot up devslug and give it a flash Feb 17 19:59:41 btw, I did discover that Linksys samba functionality is NOT default samba functionality with regard to the "map system = yes" samba config Feb 17 20:00:12 right, and unslung functionality matches stock linksys functionality? Feb 17 20:01:45 it does... but that default is not a samba defualt.. almost as if Linksys is trying to *catch* users telnet/ssh in and chmod 777 Feb 17 20:03:49 so I could make a correction, and use resling to save that change? Feb 17 20:08:16 dyoung, I've loop mounted the ramdisk produced by my build and It appears that /sbin/ doesn't contain an init file. Feb 17 20:09:02 there's a few other symlinks pointing to busybox though. Feb 17 20:09:19 s/busybox/switchbox Feb 17 20:10:04 thats correct. Feb 17 20:10:33 ok. So where would that init be? Feb 17 20:10:34 switchbox is to.. well, switch. Feb 17 20:10:43 the init youre looking for is in the jffs2 Feb 17 20:10:59 ok. so the linuxrc file would mount it... Feb 17 20:11:26 ByronT: yes, you could change it and use resling to save and restore it Feb 17 20:11:28 I wonder if [g2] was takling about something else then. Feb 17 20:11:56 right. Feb 17 20:12:29 Can I set init to a shell to experiment? Feb 17 20:12:59 I thought it dropped out to a shell as the last thing it looked for when it couldn't find an init Feb 17 20:13:08 doesn't seem to. Feb 17 20:13:15 kernel panic. Feb 17 20:13:43 rwhitby-web, voodooz is doing something differnt. Feb 17 20:14:00 VFS: Mounted root (jffs2 filesystem) readonly. so it did mount jffs2 Feb 17 20:14:13 dyoung-web: I know he's not using switchbox - I thought that the kernel looked for /bin/sh as a last resort if init is not found Feb 17 20:14:23 VoodooZ_home: thats my standard way of chekcing stuff. init=/bin/sh Feb 17 20:14:34 oh yeah, maybe there isnt a bin/sh Feb 17 20:14:35 hmm Feb 17 20:14:53 Should I have an initrd line? Feb 17 20:15:07 i mean command line option Feb 17 20:15:10 not if you're intending to bypass switchbox Feb 17 20:15:18 thought so.. Feb 17 20:16:16 do you have a valid /bin/sh and /bin/init in your jffs2 filesystem? Feb 17 20:17:11 well, I'm using stock openslug almost so I can only assume there is. Feb 17 20:17:49 The ramdisk mount shows one in /sbin but then again, that's not the jffs2. Feb 17 20:18:01 right - look in the jffs2 Feb 17 20:18:13 Can I simply mount it? Feb 17 20:18:34 don't think so. but you can look in tmp/rootfs in your openslug build Feb 17 20:18:56 yeah, plus my work box doesn't have jffs2 support built-in. crap Feb 17 20:19:36 under tmp/rootfs I only see dev, etc and usr Feb 17 20:19:48 on the openslug rootfs I built 2 weeks ago, there is a bin/sh -> busybox and a sbin/init -> init.sysvinit Feb 17 20:20:25 VoodooZ_home: then your jffs2 build is hosed Feb 17 20:20:53 That's just great. I didn't change anything and it built fine. Feb 17 20:21:02 that's always a nice thing to read Feb 17 20:21:04 let me compare with an archived tmp Feb 17 20:21:20 such is life with alpha software ... Feb 17 20:21:28 yep. you're right. Feb 17 20:21:40 My last backup as a full rootfs. Feb 17 20:21:44 strange. Feb 17 20:21:50 VoodooZ_home: you have a chance to test out the boot time performance of hotplug-ng, no startup script, manually modprobing the ehci/ohci on boot? Feb 17 20:21:53 I wonder what changed. Feb 17 20:22:07 nope. My jffs2 rootfs is hosed. Feb 17 20:22:19 Know of any changes that would cause that? Feb 17 20:22:23 taking it well though Feb 17 20:22:40 ah Feb 17 20:22:56 WHich package can I clean to try it again? Feb 17 20:24:20 Or do I have to do a complete rebuild Feb 17 20:25:51 ? Feb 17 20:25:57 to try what? Feb 17 20:26:31 To rebuild my fcked up jffs2 rootfs. Feb 17 20:26:53 you dont have to clean anything. the -image .bb's dont obey stamps, and wipe tmp/rootfs before they run. Feb 17 20:27:04 just bitbake openslug-image and wait for it to populate the rootfs and produce the .jffs2 Feb 17 20:27:10 or whatever Feb 17 20:27:25 ok thanks Feb 17 20:28:25 damn. My build/tmp/rootfs is practically empty. only etc, usr and dev. Feb 17 20:29:33 dyoung-web: the image looks fine to me. will have to wait for 6 or 7 hours before I can test it Feb 17 20:29:51 so I might do a complete rebuild before then just to be sure Feb 17 20:30:06 anyways, I'll check with [g2] tommorow. THanks for your help. Feb 17 20:31:06 there was something funky for me during the cpio build part. It wanted to mkdir /usr/share and a bunch o fother stuff in /usr. I didnt bother investigating though, and just moved cpio stuff from a old build there. Feb 17 20:34:37 good night Feb 17 20:35:47 dyoung-web: are you saying that the OE cpio package is trying to do stuff to /usr/share on the build machine? That's nasty if it's true. Feb 17 20:39:58 Yes, tahts what I'm saying. Feb 17 20:40:04 its like the patch didnt get installed. Feb 17 20:40:18 but I didnt want to screw with it so I applied Ugly Hack (tm). Feb 17 20:42:17 ok, I'll see if it blows up on my build Feb 17 20:55:07 03kin-kwok * 10unslung/sources/squid/ (rc.squid squid.delay-start.sh): add more comment Feb 17 22:09:50 dyoung: cpio fails for me too Feb 17 22:18:52 back later Feb 17 22:20:59 hi all Feb 17 22:40:56 hi samms Feb 17 22:41:10 hey..sorry installing gentoo Feb 17 22:41:31 hay...watching TV...:-) Feb 17 22:41:52 you know anything about ipkg? Feb 17 22:42:21 little, I am new too... Feb 17 22:42:39 what is your question? Feb 17 22:43:29 ah...well..i'm trying to make a ipkg for apache w/ php but can't figure out the instructions for making ipkg on the nslu2 web site. it seems geared towards making ipkg w/ one package not when you have to compile one/install then compile/install to make one package... Feb 17 22:44:34 It is your choice. Feb 17 22:44:55 you can make one depends on the other. Feb 17 22:45:18 so do you build two ipkgs? Feb 17 22:45:20 So you end up with two ipk; but user have install both Feb 17 22:45:41 You can build 2. Feb 17 22:45:43 with apache w/ php you have to compile php make install which copies its files into the apache source tree then you build apache... Feb 17 22:45:54 static php into apache...not shared. Feb 17 22:46:20 So, in this case, you have to build one ipk. Feb 17 22:46:43 You need to stage the php and build the final. Feb 17 22:47:08 so build php into apache then just do the ipkg for apache w/ the php libs copied into it..? Feb 17 22:48:38 I think so. You can automate this via the .mk file. Feb 17 22:49:41 but when you setup teh .ipkg you have to give it a http/ftp site to download teh package from...is this web site the apache.org web site or .... Feb 17 22:51:19 Yes. Feb 17 22:52:13 so this is where i get confused. if ipkg downloads the source from apache.org. it won't have the php libs already compiled. Feb 17 22:53:38 On the .mk files, you will tell the system to download both apache and php src. Feb 17 22:54:26 And you have tell the build system the order to make/build the apps. Feb 17 22:55:02 yeah, i'm not a software guy, i'm an analog ee so i know enough to be dangerous.. Feb 17 22:55:02 samms71: hi Feb 17 22:55:07 hey Feb 17 22:55:51 It is not that hard ;-) Feb 17 22:55:59 samms71: have an apache.mk and a php.mk Feb 17 22:56:13 have a php-stage target in php.mk which copies files into the staging/... area Feb 17 22:56:27 then grab those files from there after calling make php-stage in apache.mk Feb 17 22:56:35 and copy them to where they are needed in the apache build Feb 17 22:56:39 then package up apache Feb 17 22:57:12 sorry..i'm dense wrt to software and i think i've build one make file in my life... Feb 17 22:57:34 i'll try it....though Feb 17 22:58:06 samms71, you've picked a tough way to learn make :) Feb 17 22:58:25 hehe..feels like it..gotta go do some web searches on make files..;) Feb 17 22:58:40 can you not build apache first, then build php using apxs Feb 17 22:58:58 i tried to build shared libs but ldconfig is missing.at least on my slug Feb 17 22:59:06 so it couldn't finish the .so's Feb 17 22:59:28 yes, as on all, but you don't need ldconfig to make a .so Feb 17 22:59:54 you do if you want to add the .so to ld.so unless you want to have a env variable. right? Feb 17 22:59:55 ldconfig just managed /etc/ld.so.cache on those systems that have it Feb 17 23:00:26 on unslung, ld.so will look in /opt/lib without extra intervention Feb 17 23:00:41 samms71: write up the manual steps on a wiki page, and then someone else will be able to create the .mk files. Feb 17 23:00:44 gotta go Feb 17 23:00:51 ah....ok...that might make things easier... Feb 17 23:00:54 thx Feb 17 23:00:58 no probs Feb 17 23:01:05 cya rwhitby Feb 17 23:01:21 ok..i gotta run..thx all.. Feb 17 23:01:39 bye.. **** ENDING LOGGING AT Thu Feb 17 23:59:56 2005