**** BEGIN LOGGING AT Sat Mar 12 23:59:56 2005 Mar 13 01:01:32 he's baaaack Mar 13 01:16:35 hi rwhitby-away : I got the memo. Mar 13 01:48:41 <[cc]smart> is there a command or a package with commands that allow changing linksys/nvram variables from openslug ? Mar 13 01:50:16 not easily. Mar 13 01:50:32 <[cc]smart> dd ;/ Mar 13 01:50:42 and dont forget the size header. Mar 13 01:51:03 <[cc]smart> i'm a sinner, ord Mar 13 01:55:29 <[cc]smart> somebody possibly tried an usb ISDN interface ? Mar 13 02:28:03 Its pretty difficult for us to get ISDN here. so no. ;-) Mar 13 02:28:36 Why would you want to diddle the linksys sysconf settings anyways? Mar 13 02:28:41 if youre already openslugged? Mar 13 06:23:43 <[cc]smart> dyvoung: because i frequently reflash. the sysconfig script takes what is i linksys sysconf as defaults, which is quite nifty/nice since it can be used to let your system configuration survive the flash Mar 13 06:50:44 <[cc]smart> heh, and now i know it would be nice to have syconfig read/write tools in openslug, so i can set nfsroot options :) Mar 13 06:51:32 <[g2]> there you go! Mar 13 06:51:48 <[cc]smart> yes, but then, who's writing the linuxrc ? Mar 13 06:52:09 <[g2]> I think a simple program to edit the file Mar 13 06:52:31 <[cc]smart> er, i mean who's coding it ? Mar 13 06:52:44 <[g2]> ~nslu-rules Mar 13 06:52:53 <[g2]> doh! Mar 13 06:53:02 <[cc]smart> other channel :)I Mar 13 06:53:23 <[g2]> :) Mar 13 06:57:06 <[g2]> [cc]smart, what are you using your OpenSlug for ? Mar 13 06:57:17 <[cc]smart> several things Mar 13 06:58:04 <[g2]> are they a secret ? :) Mar 13 06:58:04 <[cc]smart> one slug will be used for fetching some data off a weather station + be cardclient for a dvb receiver Mar 13 06:58:27 <[cc]smart> no, but i takes me a while to find out how to describe the task best :) Mar 13 06:58:49 <[cc]smart> 2) (easy to describe) will become external mail server Mar 13 06:59:04 <[cc]smart> 3) trying new openslug versions Mar 13 06:59:27 <[cc]smart> and possibly at some time i'll buy #4) openslug native compiling service Mar 13 06:59:55 <[cc]smart> 1) more or less is set, but an issue with the cardserver software needs to get resoled Mar 13 07:00:05 <[cc]smart> 2) is in configuration phase now Mar 13 07:00:15 <[cc]smart> 3) is in long time use Mar 13 07:00:40 <[cc]smart> the last one is the only one equipped with serial access Mar 13 07:01:05 <[g2]> excellent! Mar 13 07:01:09 <[g2]> that's impressive Mar 13 07:01:56 <[cc]smart> i'm wondering why linuxrc is probing the existence of all those .files instead of reading the content of one single file and using that as root device Mar 13 07:02:34 <[g2]> [cc]smart, So you are running OpenSlug for all of those ? Mar 13 07:02:38 <[cc]smart> yes Mar 13 07:02:45 * [g2] hugs [cc]smart Mar 13 07:02:48 <[cc]smart> :) Mar 13 07:03:16 <[g2]> For backup recovery do you use switchbox or Redboot Mar 13 07:03:34 <[cc]smart> backup/recovery of what kind Mar 13 07:03:38 <[cc]smart> you mean system backup ? Mar 13 07:04:01 <[g2]> No I mean lets say you hosed up a kernel build and flashed it Mar 13 07:04:23 <[cc]smart> i'm always testing new packages/images on #3 first Mar 13 07:04:42 <[cc]smart> if STILL something goes down the drain i'll restart it from redboot Mar 13 07:04:59 <[cc]smart> all i need to get them going i have avail. via NFS or scp Mar 13 07:05:33 <[g2]> Is #1 running from jffs2 ? Mar 13 07:05:58 <[cc]smart> with respect to system, so i far i always keep the internal image intact Mar 13 07:06:23 <[cc]smart> #1 has an additional 64M USB stick Mar 13 07:06:29 <[g2]> nod. Mar 13 07:06:33 <[cc]smart> or alternatively NFS for /usr Mar 13 07:06:57 <[cc]smart> actually right now it's un NFS Mar 13 07:07:18 <[cc]smart> and i want to get another 256M stick going now for #2 Mar 13 07:07:29 <[g2]> nod. Mar 13 07:07:46 <[cc]smart> right now i'm wondering why an entry in /etc/fstab doesn't make it get mounted during boot Mar 13 07:07:49 <[cc]smart> will find it Mar 13 07:08:09 <[cc]smart> doing so i wondered why /linuxrc check all the files existence instead of reading on single ones content Mar 13 07:08:41 <[cc]smart> which would normally make fiddling with future changes obsolete, too Mar 13 07:09:09 <[g2]> it's all for recovering in a pre-described order Mar 13 07:09:46 <[cc]smart> for this one i can understand it : /mnt/jffs2/.recovery Mar 13 07:09:51 <[g2]> I was going to say that you can tweak your kernel options to skip switchbox and boot straight to the jffs2 partition Mar 13 07:10:03 <[cc]smart> but for all of these, i dont' : .sda1root Mar 13 07:10:08 <[g2]> nod. Mar 13 07:10:40 <[g2]> We need to make the kernel/jffs2 smarter to be able to got straight there Mar 13 07:10:53 <[g2]> I think it might actually work now, but I haven't tested Mar 13 07:10:59 <[cc]smart> ? Mar 13 07:11:07 <[cc]smart> i think switchbox is nifty a thing Mar 13 07:11:17 <[g2]> Right now you can root=/dev/mtdblock4 Mar 13 07:11:49 <[g2]> I would be great if we could root=/dev/sda1 Mar 13 07:11:56 <[cc]smart> just thought of sth. like prefroot=`cat /mnt/jffs2/.newroot` Mar 13 07:12:58 <[g2]> the issue with that is what if jffs2 is hosed ? Mar 13 07:12:58 or . /mnt/jffs2/.newroot Mar 13 07:13:15 fall back to recovery. Mar 13 07:13:25 <[g2]> switchbox is currently written to take that into account Mar 13 07:14:29 <[cc]smart> if /mnt/jffs2/ doesn't carry sth. sensible, looking for .xyz files will be no differnt Mar 13 07:15:03 <[cc]smart> 30 secs i'll return with what i mean.... Mar 13 07:16:53 <[cc]smart> is it ok to paste like... 15 lines here ? Mar 13 07:17:27 <[g2]> kergoth isn't around I won' tell Mar 13 07:17:40 <[g2]> maybe to pastes of 8 ? Mar 13 07:17:46 <[cc]smart> if [ -z "$mounted" ] && /bin/mount -rt jffs2 /dev/mtdblock4 /mnt/jffs2 ; then Mar 13 07:17:46 <[cc]smart> if [ -f /mnt/jffs2/.recovery ] ; then Mar 13 07:17:46 <[cc]smart> /bin/echo "Preferred root device is /dev/ram0 (recovery)" Mar 13 07:17:46 <[cc]smart> mounted= Mar 13 07:17:46 <[cc]smart> prefroot= Mar 13 07:17:46 <[cc]smart> elif [ -f /mnt/jffs2/.newroot ] ; then Mar 13 07:17:48 <[cc]smart> prefroot=`cat /mnt/jffs2/.newroot` Mar 13 07:17:50 <[cc]smart> /bin/echo "Preferred root device is $prefroot" Mar 13 07:17:52 <[cc]smart> mounted=/mnt/newroot Mar 13 07:17:54 <[cc]smart> else Mar 13 07:17:56 <[cc]smart> /bin/echo "Preferred root device is jffs2" Mar 13 07:17:59 <[cc]smart> mounted=/mnt/jffs2 Mar 13 07:18:00 <[cc]smart> prefroot=jffs2 Mar 13 07:18:02 <[cc]smart> fi Mar 13 07:18:04 <[cc]smart> fi Mar 13 07:18:53 <[cc]smart> same as now, just the ot in the middle bitten out Mar 13 07:21:39 <[g2]> I'll have to go back and check the /linuxrc Mar 13 07:22:05 <[g2]> off hand, I think that's a great suggestiong Mar 13 07:22:39 <[g2]> you're just peeking ahead in the jffs2 and adding some extra flexibility Mar 13 07:23:00 <[g2]> jffs2 look-ahead :) Mar 13 07:23:50 <[cc]smart> mostly, you get rid off the responsibility to have to follow to possible changes in the code :) Mar 13 07:24:22 <[g2]> well that is always nice for maintenance :) Mar 13 07:24:51 <[g2]> thx for the suggestion Mar 13 07:26:21 <[cc]smart> well, i'm happy if i can give sth. back Mar 13 07:26:26 <[g2]> On the native compile front kergoth made lots great changes that will really help Mar 13 07:26:41 <[g2]> that's what makes a great community everyone working together Mar 13 07:27:19 <[g2]> Maybe in another week or two I'll have stuff setup for that compile farm Mar 13 07:28:07 <[cc]smart> ok Mar 13 08:22:07 blegh Mar 13 08:22:12 i don't understand bitkeeper Mar 13 08:27:28 <[cc]smart> just don't feel alone because of this Mar 13 08:28:16 heh Mar 13 08:28:31 i don't seem to be able to update my tree Mar 13 08:28:42 i do a pull and the other command, but no openslug-2.6.11 Mar 13 08:29:11 <[cc]smart> bk pull; bk -r co -q Mar 13 08:29:14 <[cc]smart> that's what you do Mar 13 08:29:28 <[cc]smart> and none of them gives errors Mar 13 08:29:45 uhu, i do that Mar 13 08:30:17 and in the stuff that scrolls past, i see it updates openslug-kernel-2.6.11 Mar 13 08:30:24 but it isn't in my tree afterwards Mar 13 08:31:44 <[cc]smart> and no messages in the second part that would suggest that there are changed files or sth. Mar 13 08:32:07 second command doesn't give any output at all Mar 13 08:32:19 <[cc]smart> thats ok Mar 13 08:32:21 there is a dir PENDING with big files in it Mar 13 08:32:24 i don't trust that Mar 13 08:33:45 <[cc]smart> yes, sounds fishy Mar 13 08:33:57 <[cc]smart> but normally, at least bk spits sth. out if it has problems Mar 13 08:34:44 but normally, when you update, files get updated ;) Mar 13 08:34:47 and they don't Mar 13 08:35:08 although they have new timestamps Mar 13 08:41:30 hm, and a bk changes show the last change is from february 2nd Mar 13 08:41:40 something is definitely wrong here Mar 13 08:49:40 takepatch: patch left in PENDING/2005-03-13.01 Mar 13 08:49:44 what does that mean? Mar 13 08:53:59 <[cc]smart> http://www.bitkeeper.com/Documentation.FAQS.Failure.html Mar 13 09:00:57 didn't really help me Mar 13 09:01:07 strange thing is, i get no errors and output seems normal Mar 13 09:01:16 well, fuck it, i'll just do a new clone Mar 13 09:03:25 <[cc]smart> i can hear myself :) Mar 13 09:05:47 :) Mar 13 09:06:04 <[cc]smart> g2: i think chroot command should be moved from /usr/sbin to /sbin. does this make sense/is it possible ? Mar 13 10:12:17 <[g2]> [cc]smart, I'll have to review it sometime soon, I'm running on a different board lately Mar 13 10:43:42 <[cc]smart> /etc/init.d/mountall is not linked, would think it should be from /etc/rc2.d/ Mar 13 10:53:50 <[cc]smart> i added "options ehci-hcd log2_irq_thresh=2" to /etc/modprobe.conf , it seems it makes the USB bus more stable Mar 13 10:55:19 <[cc]smart> maybe the value deserves better tuning Mar 13 10:57:01 <[g2]> [cc]smart, good info. Are you aware that there is a known outstanding sw issue using the PCI bus on the ixp4xx in the 2.6.something kernels ? Mar 13 10:57:45 <[cc]smart> ? Mar 13 10:57:50 <[cc]smart> no clue Mar 13 10:58:14 <[cc]smart> what's it about, or where can the info be taken ? Mar 13 10:59:10 <[g2]> the arm linux ml had a message a couple days ago from dsaxena Mar 13 11:07:03 <[g2]> http://lists.arm.linux.org.uk/pipermail/linux-arm-kernel/2005-March/027503.html Mar 13 11:10:13 <[cc]smart> aha Mar 13 11:13:32 <[cc]smart> maybe another change for /linuxrc Mar 13 11:13:34 <[cc]smart> if [ $prefroot = "sda1" -o $prefroot = "sda2" -o $prefroot = "sdb1" -o $prefroot = "sdb2" ] ; then Mar 13 11:14:45 <[cc]smart> in of [ foo$prefroot != "foo" -a foo$prefroot != "foojffs2" ] ; then Mar 13 11:15:45 <[cc]smart> so that, if the other change is going to be made, it would take on other values we don't know Mar 13 12:45:45 I cant boot to a external disk rootfs. Mar 13 12:46:54 <[cc]smart> prolly because interrupts let it die and/or things like "[" arent visible anymore Mar 13 12:47:43 <[cc]smart> whatever way /dev/sda1 becomes inaccessible during the boot Mar 13 12:47:57 Yeah, its that problem Mar 13 12:48:34 It doesnt even work when I do the pivot by h and. Mar 13 12:49:08 <[cc]smart> hm, that's a bit weird tho Mar 13 12:49:35 <[cc]smart> possibly the PCI issue g2 mentionned is involved Mar 13 12:50:09 It brought me to a grinding halt. Mar 13 12:50:25 no package testing, no switchbox development, no nothing. Mar 13 12:51:34 <[cc]smart> so, you are doing the switchbox ? Mar 13 12:52:39 Yes, not exclusively, but I'm involved in it. Mar 13 12:54:23 <[cc]smart> ah well, you probably read the last couple lines on the channel anyways .) Mar 13 12:54:45 <[cc]smart> there were two suggestion for modifications to /linuxrc Mar 13 12:55:27 <[cc]smart> i'm unaware if g2 actually used them, or wanted to use them, or forward them or... :=) Mar 13 12:56:02 <[cc]smart> so g2 is working on switchbox, too ? Mar 13 12:58:06 I like the first suggestion witha small modification Mar 13 12:58:39 instead of catting the file there we dot it later as we do for nfs. Mar 13 12:58:49 Then you can use it to load modules and other stuff. Mar 13 13:01:01 On the other hand the simplicity of "touch .something" might be nice. I dunno. Mar 13 13:03:19 <[cc]smart> "touch .something" does require you to know the "something" literally precise. the other one, especially the variant that sources a file, allows you to just peek into it, see an example and just do it. Mar 13 13:03:28 <[cc]smart> better for the typical user i think Mar 13 13:03:47 I would almost add it on intead of replacing. Mar 13 13:04:07 . /mnt/jffs2/.altroot or something Mar 13 13:04:23 then in .altroot I can insmod foo.ko Mar 13 13:04:47 Or we could just put it all into the jffs2's linuxrc. Mar 13 13:04:48 <[cc]smart> yep, sourcing makes sense Mar 13 13:04:55 which was the original plan actually. Mar 13 13:05:14 but I never tried pivoting twice. Mar 13 13:05:17 <[cc]smart> hm, i think that would also make things more confusing tho Mar 13 13:05:20 <[cc]smart> even for the devs Mar 13 13:05:32 <[cc]smart> "i changed linuxrc to ..." ... which ? Mar 13 13:05:51 "Ah Crap, which linuxrc was that now.... Mar 13 13:06:34 <[cc]smart> perfectly "embedded" would probably be to read /etc/fstab for a modified / entry Mar 13 13:08:24 <[g2]> dyoung, why can't you boot to an external disk ? Mar 13 13:08:30 <[g2]> rootfs ? Mar 13 13:12:27 <[cc]smart> which should it be ? Mar 13 13:18:10 It just doesnt work. Mar 13 13:18:31 lemme find it again. Mar 13 13:19:16 Root filesystem will be mounted from /dev/sda1 ... Mar 13 13:19:20 [...] Mar 13 13:19:21 then.... Mar 13 13:19:33 ehci_hcd 0000:00:01.2: USB 2.0 initialized, EHCI 1.00, driver 10 Dec 2004 Mar 13 13:19:33 hub 3-0:1.0: USB hub found Mar 13 13:19:33 hub 3-0:1.0: 5 ports detected Mar 13 13:19:33 SCSI error : <0 0 0 0> return code = 0x70000 Mar 13 13:19:35 end_request: I/O error, dev sda, sector 14419 Mar 13 13:19:41 EXT3-fs error (device sda1): ext3_find_entry: reading directory #27 offset 0 Mar 13 13:19:50 Aborting journal on device sda1. Mar 13 13:19:52 ext3_abort called. Mar 13 13:19:56 EXT3-fs error (device sda1): ext3_journal_start_sb: Detected aborted journal Mar 13 13:19:58 Remounting filesystem read-only Mar 13 13:20:02 SCSI error : <0 0 0 0> return code = 0x70000 Mar 13 13:20:04 end_request: I/O error, dev sda, sector 547 Mar 13 13:20:08 and that goes on for a while before it croaks. Mar 13 13:20:47 <[cc]smart> ya, and the funky part is, that it then even reconnects the stick as sdb :) Mar 13 13:21:22 <[cc]smart> but what were you up to suggest when you asked to specify the filesystem for /. sth. specific ? Mar 13 13:21:30 Yeah, and this happens for a USB Stick, a USB->MMC/SD Adapter, or hard drive. Mar 13 13:21:58 I did this both automagically and by hand with the same result. Mar 13 13:24:25 <[cc]smart> i see the posibility that it is related to the PCI problem g2 mentionned and reference by the ink above Mar 13 13:34:01 <[cc]smart> hmm... adding /etc/init.d/mountall.sh into startup, i did as /etc/rc2.d/S01mountall , breaks the tmpfs for /var Mar 13 13:42:55 <[cc]smart> commenting this line in mountall.sh out fixed the issue with /var not being populated: # kill -USR1 1 Mar 13 13:47:21 <[cc]smart> ah no... Mar 13 13:47:26 <[cc]smart> forget my last info Mar 13 13:47:32 <[cc]smart> commenting did not fix it Mar 13 13:47:40 <[cc]smart> i made a mistake with the test Mar 13 13:57:29 <[cc]smart> ah, /etc/init.d/populate-var.sh is meant for that, added it as S02 to rc2 does it. it's static tho. Mar 13 13:57:54 <[cc]smart> this might be another possibility for populating /var: Mar 13 13:58:00 <[cc]smart> mount -o bind / /mnt Mar 13 13:58:12 <[cc]smart> vp -ar /mnt/var/* /var Mar 13 14:00:01 <[cc]smart> umount /var Mar 13 14:00:04 <[cc]smart> ee Mar 13 14:00:07 <[cc]smart> umount /mnt Mar 13 14:07:24 <[cc]smart> but /var gets mounted twice... Mar 13 14:10:04 <[cc]smart> somehow i'm completely off sense Mar 13 14:10:11 <[cc]smart> dunno, will tray again tomorrow Mar 13 18:48:14 [g2]: has putting ehci as a module stuffed something up? Mar 13 18:48:45 <[g2]> rwhitby-web, dunno, checking Mar 13 18:49:26 as soon as the ehci issue is resolved, we have room to put it back in the kernel, right? Mar 13 18:50:03 * rwhitby-web wonders whether he took it out before or after putting ext3 and jbd back in ..... Mar 13 18:50:07 I dunno, I tried with 2.6.11.2 and couldnt make it fit. Inoly spent 10min on it though. Mar 13 18:50:24 how much too big? Mar 13 19:10:21 <[g2]> rwhitby-web, which echi issue are you talking about ? you loading stuff via hubs ? Mar 13 19:10:51 Theres that issue and the dog biting issue. Mar 13 19:11:16 <[g2]> they may be two different issue Mar 13 19:11:38 <[g2]> the may be the same issue Mar 13 19:12:15 <[g2]> I've wiped build/tmp and I've got a switchbox that loads ehci before the sleep Mar 13 19:12:49 <[g2]> but it'll be 1+ hours before the build completes Mar 13 19:19:24 rwhitby, 1050272, so like 2k or something close. Mar 13 19:31:10 ok, so it will be easy to shoe-horn, cause we haven't run out of fat yet Mar 13 19:32:43 Okay, I'll poke at it later pending the outcome of [g2]'s test. Mar 13 19:34:50 ctrlproxy should Just Work (tm) as a unslung package btw. All the support libraries are already there. Mar 13 19:39:19 cool - it's in the CVS ? Mar 13 19:41:34 back later **** ENDING LOGGING AT Sun Mar 13 23:59:56 2005