**** BEGIN LOGGING AT Mon Jun 06 23:59:56 2005 Jun 07 11:57:42 <[g2]> jbowler-zzz, hey! Jun 07 12:03:32 <[g2]> jbowler-zzz, I've been thinking that a jffs2 update from something other than jffs2 is simply overwriting the jffs2 partition Jun 07 12:06:29 <[g2]> nice timing :) Jun 07 12:06:47 just fixing the merge conflict on nslu2-linux.bkbits.net ;-) Jun 07 12:06:47 <[g2]> I was just about ready to type my dd command :) Jun 07 12:07:13 <[g2]> I haven't pulled since yesterday and all seemed well then Jun 07 12:07:20 Right - I had a discussion with Rod (on #openslug?) which reached the same conclusion. Jun 07 12:07:46 I.e. just as 'update-kernel' drops the kernel image into /dev/mtdblock2, 'update-jffs2' does the same thing to mtdblock4 Jun 07 12:07:50 <[g2]> same conclusion about bb yesterday ? Jun 07 12:08:14 The same conclusion wrt to ffs2 updates/maintenance mode/etc Jun 07 12:08:20 <[g2]> ah. Jun 07 12:08:32 <[g2]> is there an update-jffs2 ? Jun 07 12:08:48 99% of the time users do not need to update jffs2 while running from jffs2 - in fact it is very unusual (or should be) Jun 07 12:09:10 <[g2]> and they should be smart enought to reboot to a ramdisk or external device Jun 07 12:09:26 Well, update-jffs2 is just a matter of writing the jffs2 image (deploy/images) into mtdblock4 with the header. Jun 07 12:09:30 <[g2]> plus we should probably have a check in the update routine :) Jun 07 12:09:51 Yes - users running out of flash must be smart, or they will fill it up and have a 'brike' Jun 07 12:10:14 Checks: yes, that's why 'devio' has an operator to return the dev number of the output device. Jun 07 12:11:02 <[g2]> there isn't a header on mtdblock4 is there ? Jun 07 12:11:19 <[g2]> jffs2 wouldn't work if there was Jun 07 12:11:36 <[g2]> the headers are on 2 and 3 right ? Jun 07 12:11:54 I don't think there is a header - I just need to check slugimage to be sure. Jun 07 12:12:18 <[g2]> I don't think there is... I was about to just reflash mtdblock 4 Jun 07 12:13:04 <[g2]> maybe I was going to try it with out reflashing redboot, I'm running APEX Jun 07 12:13:08 It should just work by dd (though it's sub-optimal because it won't fill the last block with 255). Jun 07 12:13:29 <[g2]> yeah... jffs2 will moan on the first mount Jun 07 12:13:35 <[g2]> I expected that Jun 07 12:13:52 Why? Jun 07 12:14:05 <[g2]> because the jffs2 isn't clean Jun 07 12:14:43 <[g2]> it looks like we dont have the mtd tools in there any more Jun 07 12:15:16 mtd-util isn't in the depend list - I think it isn't even built at present. Jun 07 12:16:02 Why isn't the jffs2 clean? If it was just copied in the flash should end up identical to whatever slugimage produces. Jun 07 12:16:10 <[g2]> was there a reason python and samba were pulled from the openslug-packages ? Jun 07 12:16:35 <[g2]> I think slugimage pads with FF Jun 07 12:17:21 <[g2]> yeah it does Jun 07 12:17:23 Ah. That's important because the journal presumably requires erased space at the end. devio ;-) Jun 07 12:17:56 <[g2]> I think jffs2 is smart enough to know the block are not right and just erases them Jun 07 12:18:14 <[g2]> for the rootfs everything is block aligned so it doesn't matter Jun 07 12:18:48 I don't think samba or python were ever added to openslug-packages, at least I can't see a log message for them. Jun 07 12:20:07 * [g2] dd'ing Jun 07 12:20:52 <[g2]> hmmm... this takes a little while Jun 07 12:21:36 <[g2]> ok... Jun 07 12:21:48 <[g2]> dd if=rootfs.jffs2 of=/dev/mtdblock4 Jun 07 12:21:49 <[g2]> 9472+0 records in Jun 07 12:21:49 <[g2]> 9472+0 records out Jun 07 12:21:55 <[g2]> now for the reboot Jun 07 12:23:44 <[g2]> hmmmm. Jun 07 12:24:57 <[g2]> looks like the dd didn't dd so good Jun 07 12:25:13 That block size is small. I thought /dev/mtdblock4 had a block size of 64k. Jun 07 12:25:43 I suspect you may have to pad with FF (and you must umount /initrd first!) Jun 07 12:26:06 <[g2]> I unmounted /initrd first :) Jun 07 12:26:33 <[g2]> the pad would only make a difference with the error messages Jun 07 12:26:42 <[g2]> the block size would make a difference Jun 07 12:27:09 <[g2]> but 9472 * 512 sounds like 4.7MB ish to me Jun 07 12:27:31 <[g2]> nod. That's the right size Jun 07 12:27:45 <[g2]> 4849664 Jun 07 12:28:03 Bogus busybox dd. The should use st_blksize. 'cat' is probably better. Jun 07 12:28:45 <[g2]> I'd think mtd-utils would be best Jun 07 12:30:13 Well, devio will ensure everything gets flushed correctly Jun 07 12:30:45 <[g2]> did you have any luck with the write-up on that ? :) Jun 07 12:31:15 In progress, trying to find a decent HTML editor (tried quanta, instant SIGSEGV)... Jun 07 12:32:00 It's amazing that there isn't something like Word... Jun 07 12:32:31 <[g2]> I'm really surprised that the dd failed me :( Jun 07 12:36:26 <[g2]> jbowler, so for the next OpenSlug release I'd like to see the following: Jun 07 12:36:36 <[g2]> 1) turnup preserve/restore Jun 07 12:36:54 <[g2]> 2) the update-jffs2 working Jun 07 12:37:04 <[g2]> 3) the update kernel working Jun 07 12:37:13 <[g2]> 4) reboot/shutdown problems fixed Jun 07 12:37:42 <[g2]> I'll take #4 if you'll take #1 and we can meet in the middle on the others Jun 07 12:37:45 <[g2]> Ok ? Jun 07 12:42:56 <[g2]> bbl Jun 07 12:55:34 [g2]: sounds good (I've got to deal with a stray dog now...) Jun 07 14:38:01 dyoung: ping Jun 07 14:50:39 [g2]: don't know if you will see this, but I've just done an 'update-jffs2' with the command devio '<>/dev/mtdblock4;cp$;fb#$-255' (all seems to work fine - jffs2 is the name of the deploy/image file) Jun 07 16:18:45 <[g2]> dyoung, ping Jun 07 17:50:48 What's the latest "official" release of openslug? Jun 07 17:51:03 <[g2]> the last beta Jun 07 17:51:23 <[g2]> but there'll probably be a new beta soon Jun 07 18:05:41 Still running 1.12. I know there were some fixes that I wanted added after that, but I haven't had time to look at it for a while Jun 07 18:35:28 NAiL: ntp is fixed in the latest bk builds (by fixes to problems in the real time clock code in the kernel) Jun 07 18:35:54 jbowler-away: Ah, yes. I remember that as one of the fixes I wanted :) Jun 07 18:36:42 Now, if I could get the irq issue fixed, I'd be a happy camper :P Jun 07 18:37:28 So far as I know no one has looked at it recently. Jun 07 18:38:29 Ah well... It's but a minor annoyance. Jun 07 18:38:42 Hmm... I wonder... Jun 07 18:39:02 Think I'm gonna try building bnetd on my slug **** ENDING LOGGING AT Tue Jun 07 23:59:56 2005