**** BEGIN LOGGING AT Wed Jun 08 23:59:56 2005 Jun 09 11:18:36 <[cc]smart> booting from emstock /dev/sda1 via /boot/disk seems to always get me a .recovery file on that usbsticks root folder, thus rebooting flicks back to mtdblock4 Jun 09 11:18:40 <[cc]smart> why that ? Jun 09 11:32:32 <[g2]-away> [cc]smart, are you running openslug ? Jun 09 11:38:23 <[cc]smart> yes Jun 09 11:38:33 <[cc]smart> and only openslug :) Jun 09 11:38:42 <[g2]-away> the beta version or a more recent version ? Jun 09 11:38:54 <[cc]smart> bitbak three days ago Jun 09 11:39:00 <[g2]-away> did you turnup ? Jun 09 11:39:23 <[g2]-away> you were having issues with reiserfs right ? Jun 09 11:39:27 <[g2]-away> or was that someone else Jun 09 11:39:29 <[cc]smart> no, did manually, cause i've got not much clue about it yet and am experiementing Jun 09 11:39:39 <[cc]smart> so i will hopefully be skilled enough to write an intro Jun 09 11:40:21 <[g2]-away> the latest turnup is IMHO the way to go Jun 09 11:40:22 <[cc]smart> reiserfs, yes, that's me. but the issue was less reiser than me injecting a typo Jun 09 11:41:03 <[cc]smart> what does trunup do to ... or better what does the system do, to decide i should need a .recovery file ? Jun 09 11:41:04 <[g2]-away> I'd backup the memstick or just flash over it if nothing important is on it and re-turnup Jun 09 11:41:33 <[g2]-away> turnup prepares for the pivot root to an external device Jun 09 11:41:40 <[g2]-away> could be nfs, disk or memstick Jun 09 11:41:46 <[cc]smart> the issue is a bit the other way round, the system is trying to outsmart me Jun 09 11:42:11 <[cc]smart> which is it odes more than i asked for Jun 09 11:42:12 <[g2]-away> well I'm sure you won't let it do that ! :) Jun 09 11:42:46 <[cc]smart> well. i didn't create the .recovery file.... :) Jun 09 11:43:02 <[g2]-away> there's a recovery file on the flashstick ? Jun 09 11:43:10 <[g2]-away> or in the jffs2 ? Jun 09 11:44:00 <[cc]smart> it's being created during each boot from memstick Jun 09 11:44:25 <[cc]smart> so i've set /linuxrc in jffs2 to invoke /boot/disk Jun 09 11:44:31 <[cc]smart> this works fine one time Jun 09 11:44:38 <[cc]smart> it boots the memstick Jun 09 11:44:57 <[cc]smart> during this boot it creates a new .recovery on the memstick Jun 09 11:45:11 <[cc]smart> thus, if i don't remove the file after boot, next boot is from jffs2 again Jun 09 11:45:29 <[cc]smart> for as long as i remove the file each time, all is fine Jun 09 11:45:41 <[g2]-away> Ok... you are short-circuiting what turnup sets up Jun 09 11:46:16 <[cc]smart> ? didn't use turnup yet, cause i didn't know of it till yesterday :) Jun 09 11:46:25 <[g2]-away> nod Jun 09 11:46:34 <[g2]-away> sorry for the confusion Jun 09 11:46:47 <[cc]smart> so what is needed to avoid .recovery to get created ? Jun 09 11:46:58 <[g2]-away> I'd do the followign Jun 09 11:46:59 <[cc]smart> or the other way round, why is it created ? Jun 09 11:47:06 <[g2]-away> remove the memstick and reboot Jun 09 11:47:19 <[g2]-away> you may want to reflash and reboot Jun 09 11:47:53 <[g2]-away> then I'd backup up the memstick if needed and Jun 09 11:48:01 <[g2]-away> format the memstick Jun 09 11:48:20 <[g2]-away> and turnup memstick -i -t ext3|reiserfs Jun 09 11:48:25 <[g2]-away> and reboot Jun 09 11:49:22 <[cc]smart> hmm, that woul prolly be a solution but no explanation... think i'll have to dig a bit. or how should i understand ? Jun 09 11:49:37 <[cc]smart> this stuff shouldn't try to outsmart me. Jun 09 11:49:48 <[g2]-away> nod. Jun 09 11:49:56 <[g2]-away> jbowler did all the updates Jun 09 11:49:56 <[cc]smart> i'm root, go worship :) Jun 09 11:50:24 <[g2]-away> I understand it a little, but not enough to fully explain what's happening Jun 09 11:50:30 <[cc]smart> i'll try to find out and maybe make a suggestion Jun 09 11:51:16 <[g2]-away> turnup overwrites the /linuxrc to just be a couple line file Jun 09 11:51:48 <[g2]-away> my after the turnup looks like this for my memstick Jun 09 11:51:53 <[g2]-away> #!/bin/sh Jun 09 11:51:53 <[g2]-away> exec '/boot/disk' '/dev/sda1' '-t' 'ext3' '-o' 'noatime' Jun 09 11:51:53 <[g2]-away> exec /boot/flash Jun 09 11:52:50 <[g2]-away> the scripts are in /boot on the jffs2 or /initrd/boot if you're on an external device Jun 09 11:53:21 <[cc]smart> i did similar manually: exec /boot/disk /dev/sda1 -t reiserfs -o noatime Jun 09 11:53:23 <[g2]-away> they are pretty small and well commented Jun 09 11:53:54 <[g2]-away> do you have the ' marks ? Jun 09 11:54:08 <[g2]-away> are you running exec /boot/flash afterwards ? Jun 09 11:54:15 <[cc]smart> no, but i figured they were only needed if you've got strange stuff in the options Jun 09 11:54:24 <[cc]smart> to avoid shell interpretation and such Jun 09 11:54:44 <[cc]smart> exec /boot/flash i didn't, too. Jun 09 11:55:01 <[cc]smart> but i don't really se what i could do either Jun 09 11:55:13 <[cc]smart> if i remember rioght exec hands over with now way bacl Jun 09 11:55:37 <[cc]smart> so the first exec will count and the second never be reached Jun 09 11:55:38 <[g2]-away> I'd guess the ' cause each to be it's own parameter Jun 09 11:55:53 <[cc]smart> the psace splits the parameters,t oo Jun 09 11:56:08 <[cc]smart> but the single high quotes inhibit any kind of shell expansion Jun 09 11:56:21 <[g2]-away> probably the - Jun 09 11:56:24 <[cc]smart> so you are on the safe side Jun 09 11:56:25 <[g2]-away> as options to exec Jun 09 11:57:23 <[cc]smart> i wouldn't expect a difference with the - and the result seems to be correct to me, just the .recovery file appearing Jun 09 11:59:09 <[cc]smart> i'm a bit irritated about this redirection in /boot disk Jun 09 11:59:24 <[cc]smart> if checkmount /mnt && :>/mnt/.recovery Jun 09 11:59:30 <[cc]smart> never seens a ":>" Jun 09 12:00:37 <[cc]smart> actually i think it's the culprit Jun 09 12:01:07 <[cc]smart> wondering why anybody should be able to avoid what is happening to me Jun 09 12:01:20 <[cc]smart> seems like it's y design Jun 09 12:04:30 <[cc]smart> and another one in the swivel function: sleep 10 >/.recovery Jun 09 12:25:12 rmrecovery needs to be run at the end of the startup scripts for each run level - i.e. the boot has to run to completion and rmrecovery has to be added to the scripts. Jun 09 18:50:49 So my devslug is now running the latest OpenSlug from yesterday's build. Jun 09 18:53:18 with all the sweet jbowlerisms in it :-) Jun 09 18:54:33 <[g2]> :) Jun 09 18:55:02 <[g2]> and with the magic chant you can reflash from openslug :) Jun 09 18:55:32 <[g2]> hey I coded up my first arm assembler today Jun 09 18:57:06 you got me thinking about learning it Jun 09 18:57:55 <[g2]> what arm asm ? Jun 09 18:58:00 yeah Jun 09 18:58:15 <[g2]> have you done much assembler programming ? Jun 09 18:58:22 much no Jun 09 18:58:49 some z80 and 6502 in high school, MC68000 in college Jun 09 18:59:09 <[g2]> IMHO (as an old-man) It helps to know how the machine works Jun 09 19:00:07 <[g2]> some of the gcc syntax looks pretty freaky though :) Jun 09 19:00:59 <[g2]> stuff like: Jun 09 19:01:03 <[g2]> /* Disable interrupts and set supervisor mode */ Jun 09 19:01:03 <[g2]> __asm volatile ("msr cpsr, %0" : : "r" ((1<<7)|(1<<6)|(0x13<<0))); Jun 09 19:01:48 so I think we should add the update-alternatives stuff to dropbear and openssh in openslug so that you can ipkg install openssh over the top of dropbear, and have ipkg sort out all the symlinks so that your current dropbear session remains active, but your boot starts the openssh server instead. Jun 09 19:01:49 <[g2]> I'd guess that's a move special register (the cp status register) Jun 09 19:02:00 "0x13<<0" ? Jun 09 19:02:29 that's that special rotate 0 command :-) Jun 09 19:02:40 ultra-rotate Jun 09 19:03:11 <[g2]> actually, I think that's good coding practice in this case :) Jun 09 19:04:07 rwhitby-web, yeah being able to install openssh over dropbear would be really useful Jun 09 19:04:18 without --force-overwrite like I did :-) Jun 09 19:04:26 <[g2]> can you do that from jffs2 only ? Jun 09 19:04:31 has anyone else seen those dropbear disconnects? Jun 09 19:04:46 <[g2]> only when I hosed up the shell Jun 09 19:04:48 [g2], I don't think there's space, unles you rm the kernel image Jun 09 19:04:48 nope Jun 09 19:04:58 [g2]: I'm talking on external rootfs Jun 09 19:05:26 <[g2]> if I had a 7.75MB jffs2 would that be enough ? Jun 09 19:06:03 <[g2]> or 6.75 cause a kernel's gotta be *somewhere* Jun 09 19:06:37 heh Jun 09 19:07:14 <[g2]> speaking Openslug development.... Jun 09 19:07:46 <[g2]> any thoughts on bk replacement and snapshoting the meta data for releases ? Jun 09 19:08:04 I was kinda hopinh oe would have a bk replacement by now Jun 09 19:08:18 <[g2]> me too Jun 09 19:08:32 <[g2]> sometimes I wanna be a sheep Jun 09 19:08:36 <[g2]> baaa baaa Jun 09 19:09:36 so does bkbits stop working on 1 July, or do they just stop fixing bugs? Jun 09 19:10:22 my impression is it stops working Jun 09 19:10:30 hmm - I'm remote logged into openslug on external rootfs, and just did ipkg remove dropbear and didn't loose the session Jun 09 19:10:43 doing ipkg install openssh now Jun 09 19:11:26 yeah great - I forgot that trick - it doesn't killall the sshd, so your process keeps running Jun 09 19:11:47 heck it might not even /etc/init.d/sshd stop Jun 09 19:11:54 now openssh is installed, I'm killing the parent dropbear (not the current session dropbear) Jun 09 19:12:21 that should free up the bind for openssh to start Jun 09 19:13:18 * [g2] hugs his geek buddies Jun 09 19:13:24 I think I've had success doing something like "killall sshd && sleep 5 && /etc/init.d/sshd start" Jun 09 19:13:27 generating keys - will need to overwrite those Jun 09 19:13:45 why? Jun 09 19:14:05 jacques: but dropbear doesn't run an sshd - I think you're thinking of an openssh upgrade Jun 09 19:14:06 because of your known_hosts ? Jun 09 19:14:12 yeah Jun 09 19:14:15 convenience Jun 09 19:14:28 rwhitby-web, yeah that's what I was thinking , but I figured dropbear should be similar Jun 09 19:15:11 since I do a two-step to get to another machine on the LAN, I need to make sure that my known_hosts on the gateway machine doesn't have to change Jun 09 19:15:29 otherwise I gotta go in there too and edit that Jun 09 19:16:27 hmm I wonder if the libusb in oe is any good for us Jun 09 19:16:37 <[g2]> rwhitby-web, "The most connected guy down-under" Jun 09 19:17:13 jacques: I'm using the libusb in OE for ftpd-topfield in OE to communicate with my toppy Jun 09 19:17:19 <[g2]> well Jacmet was running LE :) Jun 09 19:18:11 rwhitby-web, wow! how long has if been fixed? Jun 09 19:18:30 rwhitby-web, which version? oe has 0.1.8 and 0.1.10a Jun 09 19:18:39 only 0.1.8 has patches Jun 09 19:18:59 dunno. it wasn't working a long time ago, and when I went to to port ftpd-topfield to Unslung it just worked. Then I did ftpd-topfield in Openslug and it just worked there too. Jun 09 19:19:24 I'm using 0.1.10a in OE Jun 09 19:19:35 ok thanks I will give it a try :-) Jun 09 19:19:47 <[g2]> the "community" is a tremendously powerful force Jun 09 19:20:11 that was the last thing making me use Unslung in production - now Unslung is legacy for me with no driver to develop it further Jun 09 19:20:34 maybe we need a new unslung project leader Jun 09 19:20:41 <[g2]> so rwhitby-web just waits for the 770 and pre-builds in OE Jun 09 19:20:55 <[g2]> :) Jun 09 19:20:58 indeed :-) Jun 09 19:21:31 Unslung may still be useful for the zd1211 wlan dongle support, so I can have a remote slug Jun 09 19:22:31 that won't work on openslug? Jun 09 19:22:37 unless dyoung gets it working in OpenSlug first ... Jun 09 19:22:48 jacques: dunno - had no time to look at it Jun 09 19:23:46 what's the chipset? Jun 09 19:23:55 zd1211 Jun 09 19:24:15 .sf.net Jun 09 19:24:16 oh :-\ Jun 09 19:24:26 802.11b? g ? Jun 09 19:24:37 both Jun 09 19:25:01 aah, usb2.0 ? now it starts to make sense Jun 09 19:25:22 I got this one: http://www.3com.com/products/en_US/detail.jsp?tab=features&pathtype=purchase&sku=3CRUSB10075 Jun 09 19:25:32 you can get a cheaper one for USD$26 at Frys Jun 09 19:26:05 the usb cable on that one that I referenced is stiff - so you can bend it to make the dongle stick up straight into the air behind the slug Jun 09 19:26:12 <[g2]> rwhitby-web, is there a BT how-to for OpenSlug anywhere ? Jun 09 19:26:26 yeah, wiki BlueSlug Jun 09 19:26:42 http://www.nslu2-linux.org/wiki/HowTo/BuildABlueSlug Jun 09 19:26:43 neat! I gotta check that out too! I found my BT dongle Jun 09 19:27:03 and my ethernet dongle... but I'm still not sure the chipset is supported under linux Jun 09 19:27:18 you should be able to just ipkg install everything listed Jun 09 19:27:48 it's dead simple now Jun 09 19:28:26 rwhitby-web, that dongle - it looks like it disconnects from the cable ? Jun 09 19:28:30 <[g2]> rwhitby-web, thx for the link Jun 09 19:29:21 jacques: yes Jun 09 19:29:45 the cable is just a stiff extender - bendable and retains it's stiffnes Jun 09 19:30:02 <[g2]> jacques, which adapter ? Jun 09 19:30:10 sounds well designed then because that's a useful feature (stiff cable) in this case Jun 09 19:30:27 [g2], the 3com one linked above Jun 09 19:30:30 so you're paying an extra USD$25 for style and stiffness :-) Jun 09 19:30:41 well it is kinda cute Jun 09 19:31:07 I had to pay AUD$75 for mine ... Jun 09 19:31:44 it would be great if someone else buys one and makes it work in openslug ;-) ;-) Jun 09 19:32:35 lunch time Jun 09 19:33:19 the ZD1211 company seems to be sending source to the open source team - that's why I bought that one Jun 09 19:33:30 a refreshing change Jun 09 19:34:01 <[g2]> I think the 770's gonna kick the door open with embedded linux Jun 09 19:34:01 back later Jun 09 19:34:07 <[g2]> have a good lunch Jun 09 19:36:06 yeah 770 looks cool - I want one, or two Jun 09 19:56:34 <[g2]> later guys Jun 09 19:56:39 <[g2]> nite Jun 09 23:19:20 jbowler-away: I think the turnup script should copy the new linuxrc across to the target location, or alternatively remove the linuxrc from the target location. Othewise it gets confusing having two different linuxrc's (only one of which is actually used). **** ENDING LOGGING AT Thu Jun 09 23:59:57 2005