**** BEGIN LOGGING AT Tue Nov 15 02:59:56 2005 Nov 15 03:31:49 dyoung: Nov 15 03:31:56 i ve done make update Nov 15 03:32:10 and then retried make openslug-build image Nov 15 03:32:16 but i get the same error... Nov 15 08:00:58 Might be I broken the dependency in openslug on the linksys firmware - it should be in slugos-image.bb now. Nov 15 08:02:05 ... um, yes, my openslug builds no longer build the .img Nov 15 08:11:18 Fixed in 62bfda180323845f05492d6c191460c2281c2901 Nov 15 10:30:37 hi Nov 15 10:32:26 is there some way to midificate the bootprocess? I want everytime boot from/with flash system on the nslu2, then a automatic reboot but booting from harddisk Nov 15 10:34:24 ..only after poweron Nov 15 10:35:50 I have to do it so, because my harddisc is not properly mounted on first poweron and many buffer I/O errors are in dmesg Nov 15 10:37:20 but when I boot without harddisk, then turnon the disc, and then do a nslu-reboot, everything works fine Nov 15 11:06:24 Er, Cheef-daniel -- if you're going to modify the boot process, why not figure out what the prob is with mounting the drive and fixing it that way? Probably easier and more reliable than an automatic reboot. BTW, AFAIK OpenSlug boots from flash all the time; it switches to the hard-drive after the kernel is running; which is why I think it might be easier to fix the drive mount prob than... Nov 15 11:06:26 ...multi-boot. BTW, do you have a serial port? That would make figuring out what the problem is and debugging the fix much easier! Nov 15 11:24:52 I use only a USB2Serial converter, but not the internal serial-port Nov 15 11:25:45 Ok - I think OpenSlug is like Unslung in that without the serial port all you have available to debug boot or shutdown messages is the syslog :-( Nov 15 11:25:45 I don't know why on first boot I got this error with the harddisk and I don't know how to find out how to fix it Nov 15 11:26:12 Is this first-boot after a power-on, or also on a reboot from a running openslug? Nov 15 11:26:39 the error is only there when I first-boot Nov 15 11:26:50 after a reboot not Nov 15 11:27:24 Ok - maybe some problem with timing, maybe the drive needs longer to spinup? Nov 15 11:28:05 I tryed to startup the disk, wait longer time and then turn on the nslu, but the same error Nov 15 11:29:18 it' s a 400Gb-Seagte-drive Nov 15 11:30:26 the orange led is blinking on that error Nov 15 11:30:54 but the disk-led is green Nov 15 11:33:00 ls Nov 15 11:34:28 Is the disk spinning, though? Or is it stopped - perhaps you can tell from the noise (mine is pretty noisy when spinning). Nov 15 11:34:51 it's spinning all the time Nov 15 11:35:25 the LED onnthe hardddisk is blinking, so nslu access the harddsik Nov 15 11:36:22 Try turnup disk -s2 , for a 2 second delay, or try longer. If you really want to boot to flash just boot with the disk turned off. Nov 15 11:37:34 does turnup delete/overwrite data on my disk? Nov 15 11:48:35 I think when I do a turnup the root-system will be rewritten and all my config getting lost Nov 15 11:55:01 If you can find the turnup script (I think it is a script), you might be able to figure out what the "-s2" changes, and change it manually? Nov 15 11:57:42 cheef-daniel: no, you are wrong Nov 15 11:58:04 dosn't overwrite my date? Nov 15 11:58:06 data Nov 15 11:58:20 No, turnup (without -i) just modifies /linuxrc Nov 15 11:58:52 E.g. "turnup disk -s2 /dev/sda2" (or whatever the disk is). Nov 15 11:59:45 Also you can hand-hack the sleep delay in /linuxrc, it's pretty obvious. Nov 15 12:00:12 ok I tryed the turnup thing Nov 15 12:01:19 ? does it now boot to the disk? Nov 15 12:04:40 I tryed it with a 10 second delay, but the error is not gone Nov 15 12:07:05 Sounds like it really does need to be turned on after the slug, that's bad. Nov 15 12:07:24 but where is the difference? Nov 15 12:07:52 after reboot everyting works fine Nov 15 12:08:00 its only after a poweron Nov 15 12:08:23 There's no way of delaying the kernel probe for disks, so if the disk isn't ready and the kernel gives up on it that would explain the problem. Nov 15 12:09:07 You know it is fine if it is started after boot and it is fine if it is started well before boot (the reboot case), so it must be starting it close to the slug boot. Nov 15 12:09:14 The LinuxBIOS people have the same problem and their solution is basically to just put in small delay Nov 15 12:11:18 In the kernel? Nov 15 12:11:28 (I.e. to put the delay in the kernel?) Nov 15 12:14:07 Yep as the kernel is the BIOS Nov 15 12:18:00 ok that was it: I turned the disk on after the red-led is turned on, on first powerup the nslu Nov 15 12:24:44 is that the kernel-delay-problem, jbowler mentioned? Nov 15 12:27:02 cheef-daniel: yes, you delayed the disk-on until after the driver is loaded, I guess the driver doesn't look again until the attempt to mount the disk. Nov 15 12:34:05 where can I define scripts that are executed when I plugin my scanner? (there is no /et/hotplug/) Nov 15 12:35:06 found it on the wiki, thx Nov 15 12:52:07 * mwester thinking... Nov 15 12:54:18 Might it be possible, then, to build a boot procedure that could use the flashed openslug firmware image as if it were a BIOS and bootloader, and load another image of Linux from the local hard drive or even the network? Nov 15 12:56:08 might be, but my knowledege about to that is bad Nov 15 13:00:05 I discovered that apache 1.3 doesn't seem to want to compile on openslug Nov 15 13:00:09 (compile properly) Nov 15 16:17:14 anyone know anything about library search paths? Nov 15 16:26:55 Anyone here? Nov 15 16:27:02 I need someone who built the entire package library Nov 15 18:16:04 ldconfig ? Nov 15 18:23:06 nope Nov 15 20:29:59 mwester: that is what OpenSlug does, the jffs2 rootfs is, in effect, an initrd which boots something else (typically OpenSlug or Debian) Nov 15 20:31:48 Ok - I guess I'm looking for a "lazy" way to do development with kernels and modules and such that wouldn't require a reflash. Nov 15 20:33:14 Use a slugos configuration, that allows you to have multiple /lib/modules/* directories, so you can reflash the kernel without reflashing the initrd. Nov 15 20:34:15 The show stopper is that, for any non-flash boot, at least one module is required to mount the rootfs. Nov 15 20:35:07 So if you reflash the kernel then you end up having to have modules for both the new and old kernel. Nov 15 20:35:49 I'd like to be able to associate the modules with the kernel, but I don't currently have a good way to do that - it is major work. Nov 15 20:38:20 Interesting problem. Seems to fall victim to the problem of "do what I say" versus "do what I mean" when it comes to matching modules with kernel. So in reality one must reflash something no matter what if one replaces a kernel? Nov 15 20:40:07 No, so long as the kernel abi does not change then it just works. If you change kernel versions though modules for both versions are required. Nov 15 20:41:28 That's for glibc builds, uclibc builds will probably require a new libc build as well, and that isn't versioned so it won't work. Nov 15 20:43:35 I haven't gotten into the ucslug stuff yet. I'm just reaquainting myself with the kernel stuff as best as I can. And the worry has been just that I've been accustomed to having an alternate boot mechanism available for when I totally botch the kernel, but that seems to be harder to do with the slug. Still lotsa fun. Nov 15 20:44:46 It's very easy to get a non-booting system, and the only recovery is a complete reflash because the NSLU2 upgrade mechanism starts off by erasing the whole flash. Nov 15 20:46:01 I can't see any reasonable way of recovering a system which won't complete a boot to a dropbear prompt. Still, it just slows the development cycle ;-) Nov 15 20:49:13 Ok. I guess I'll just have to hack away carefully! :D Nov 15 20:50:11 Er - if you have a moment, another question? How does openslug determine sda versus sdb currently? In other words, if I have multiple drives, what does it use to locate the *real* root filesystem? Nov 15 20:54:58 sda is the first drive it finds, sdb is the second Nov 15 20:55:06 it == kernel Nov 15 20:58:37 That's all, huh - I kinda thought it might be that simple - that could lead to some interesting problems if one isnt' careful. Thanks ! Nov 15 21:16:33 anyone gotten nfs server setup on openslug? Nov 15 21:17:33 I have it installed and started, but when my othere box tries to mount it says permission denied Nov 15 21:20:48 hmmm, showmount -e claims the export list is empty Nov 15 21:26:14 You did add something to be exported to the exports file, right? Nov 15 22:11:05 mwester: yeah Nov 15 22:11:20 I fixed the exports thing, now showmount -e shows everything Nov 15 22:11:25 but still denied Nov 15 22:12:53 exports looks like: /public 192.168.1.0/255.255.255.0(rw,sync,no_root_squash) Nov 15 22:14:25 lemme try that, I had just rw Nov 15 22:15:58 substitute your own ip network number and netmask, of course! Nov 15 22:16:19 mine says /tmp 192.168.1.1/24(rw,sync,no_root_squash) Nov 15 22:16:28 and now the showmount -e is empty again Nov 15 22:17:24 I tried w/ 255.255.255.0 instead of 24, same deal Nov 15 22:17:53 if this is openslug, then /tmp is a sym link to /var/tmp -- cannot export a sym link. Nov 15 22:22:49 ah Nov 15 22:24:27 well that fixed the export list, still got this: nfs warning: mount version older than kernel Nov 15 22:24:28 NFS: mount program didn't pass remote address! Nov 15 22:24:28 root@jabba:/media/ram/var/tmp# Nov 15 22:24:34 but I'm gonna get some sleep, thanks dude Nov 15 22:25:38 Ok - good night. **** ENDING LOGGING AT Wed Nov 16 03:00:00 2005