**** BEGIN LOGGING AT Sun Jul 10 23:59:57 2005 Jul 11 00:05:09 NAiL: adduser takes a number of interesting, undocumented, options... --system is probably very useful, along with --no-create-home Jul 11 00:49:16 jbot, test Jul 11 00:49:17 Test Passed! Jul 11 01:34:29 03dyoung 07org.nslu2-linux.dev * r1972dc78... 10/Makefile: Fix the bitbake fixing line Jul 11 03:28:41 03nail 07org.openembedded.nslu2-linux * r49e8d85b... 10/ (5 files in 3 dirs): Fix samba LINUX_LFS problem in a proper way. FIXES #58 (OE) Jul 11 04:05:14 03nail 07org.openembedded.nslu2-linux * r36acbbe1... 10/ (77 files in 57 dirs): Jul 11 04:05:14 propagate from branch 'org.openembedded.dev' (head ef5052f6f05fb50a73e1535b3bfe8a3458070397) Jul 11 04:05:14 to branch 'org.openembedded.nslu2-linux' (head 49e8d85b0430ec83a9e882230e3044c74359e2e4) Jul 11 04:28:44 03nail 07org.openembedded.nslu2-linux * ree06322d... 10/packages/linux/openslug-kernel-2.6.12/defconfig: Added CONFIG_VLAN_8021Q, required for vlan package Jul 11 04:32:30 FWIXP420BL <- "L" for Linksys ? Jul 11 04:33:38 I was wondering about that Jul 11 04:35:01 Obviously Intel has found out about out de-underclocking, and has quickly pushed through a chip revision to counter our nefarious activities. Jul 11 04:35:14 hahah Jul 11 04:39:08 L for Low speed :-) Jul 11 04:41:14 geez would suck if newer ones weren't 266MHz capable Jul 11 04:43:50 Where did the BB number come from? I've got a BL Jul 11 04:44:24 ah cool, and it's running at 266MHz? Jul 11 04:44:27 yes Jul 11 04:44:37 It's the first slug I deunderclocked Jul 11 04:44:57 Does anyone have a BB? Jul 11 04:45:59 http://www.nslu2-linux.org/wiki/uploads/NSLU2-board-front-hires.jpg <- this one does Jul 11 04:46:11 oh Jul 11 04:47:03 Anyone else have a BL? The BL one I have was the one I removed R64 from, not R83 Jul 11 04:47:17 back in a bit - time to upgrade home.rwhitby.net to openslug 2.1 latest ... Jul 11 04:56:12 NAiL, does your BL id as B0 stepping in cpuinfo? Jul 11 04:56:43 oh hmm that's not in there Jul 11 04:57:15 what about these: Jul 11 04:57:20 CPU variant : 0x0 Jul 11 04:57:20 CPU part : 0x41f Jul 11 04:57:20 CPU revision : 1 Jul 11 04:57:24 Both my slugs are BL actually Jul 11 04:57:51 CPU variant : 0x0 Jul 11 04:57:52 CPU part : 0x41f Jul 11 04:57:52 CPU revision : 1 Jul 11 04:58:06 tho I dunno what mine is Jul 11 04:58:14 Probably BL ;) Jul 11 05:21:42 rwhitby: whose slug is that in the picture? Jul 11 05:21:54 NAiL: allan's Jul 11 05:22:10 actually, it could be [g2]'s slug which got salted Jul 11 05:22:24 oh Jul 11 05:40:56 morning Jul 11 05:41:20 VoodooZ_work: morning Jul 11 05:52:33 all mine are BL Jul 11 05:59:55 They're all BL. L is the speed grade Jul 11 06:00:06 It identifies itself as a BB via JTAG Jul 11 06:02:27 so.. L = 133? Jul 11 06:02:45 C = 400, D = 533.. Jul 11 06:02:49 No. We don't know what it stands for Jul 11 06:03:02 A should be 133.. Jul 11 06:03:26 No point speculating as we don't know what it stands for and they've always been like that Jul 11 06:03:42 and, things work, so why complain :) Jul 11 06:03:47 It doesn't appear to be a custom Linksys part as a couple random electronics supply houses in China/Hong Kong have supplies of them Jul 11 06:04:21 yeah, saw that Jul 11 06:15:25 Strange. I'm trying to update my monotone repo and it's stuff at: monotone: [bytes in: 485.6k] [bytes out: 46.1k] [certs in: 427] [revs in: 133] Jul 11 06:15:34 s/stuff/stuck/ Jul 11 06:15:48 not strange Jul 11 06:15:58 It'll take *ages* to check all 133 revs Jul 11 06:15:59 well, it's been a quote a few minutes./ Jul 11 06:16:06 Give it 10-15 Jul 11 06:16:09 damn. Jul 11 06:16:13 yes Jul 11 06:16:24 How hard is it to put a stupid ticker or some progress indicator! Jul 11 06:16:36 I really hope there's some work to speed up that a bit Jul 11 06:16:42 yeah. Jul 11 06:17:03 I wonder if it takes advantage of SMP capbable system? Jul 11 06:17:05 I don't think the monotone guys would mind if you hacked up a patch and sent them ;) Jul 11 06:17:21 I don't think so... afaik it doesn't run any extra threads Jul 11 06:17:23 hehehehe :) I doubt I'd even be able to find my way around! Jul 11 06:20:19 ok. It finally finished. Jul 11 07:03:39 03rwhitby 07org.nslu2-linux.dev * r16b09eea... 10/Makefile: Fixed the setup-developer targets Jul 11 07:16:51 hey! Jul 11 07:29:05 03rwhitby 07org.openembedded.nslu2-linux * rfa9875cd... 10/packages/linux/openslug-kernel_2.6.12.2.bb: Reinstated openslug-kernel 2.6.12.2 Jul 11 07:29:08 03rwhitby 07org.openembedded.nslu2-linux * rb44bd1f0... 10/packages/linux/openslug-kernel-2.6.12.2/ (README defconfig): Reinstated openslug-kernel 2.6.12.2 defconfig and README Jul 11 07:30:10 rwhitby, Any other big changes coming up? Jul 11 07:31:19 nope, that's it. Dunno if it works, but it used to be 2.6.12.2 and it went backwards to 2.6.12 when we moved from nslu2-kernel to openslug-kernel. Jul 11 07:31:40 yeah, I was confused as to how to implement my private changes. Jul 11 07:31:54 So the nslu2-kernel stuff is obsolete now or? Jul 11 07:35:10 nslu2-kernel is the generic linux kernel, targetted for generic nslu2 hardware. Jul 11 07:35:27 openslug-kernel is a distro kernel which is derived from nslu2-kernel. Jul 11 07:35:48 VoodooSlug-kernel would be another distro kernel derived from nslu2-kernel Jul 11 07:37:29 I see. I like it. Jul 11 07:38:13 So If I wanted to change the nslu2.h file I would change it in nslu2-kernel/2.6.? right? Jul 11 07:38:24 Or I guess I could apply my own patches. Jul 11 07:38:42 depends if it's a change for everyone, or a change for voodooslug Jul 11 07:38:56 no, for myself only (dead I/O pin) Jul 11 07:40:13 but why is the nslu2.h file (and others) in nslu2-kernel/2.6.11.2 and not a general place? Jul 11 07:40:44 dunno, I didn't put them there. Jul 11 07:41:19 a transitional thing I guess. Jul 11 07:42:32 <[g2]> uh-oh I'm in emacs Jul 11 07:42:49 heh Jul 11 07:43:23 not emacs!! :) Jul 11 07:44:39 [g2]: at last you're using a real operating system. Emacs :-) Jul 11 07:44:46 hehe Jul 11 07:45:11 [g2]: merging, are we? Jul 11 07:45:42 <[g2]> A merge came up. I'm staring at the emacs screen Jul 11 07:45:48 reminds me I haven't had the time to put up that howto yet Jul 11 07:45:59 rwhitby-away> left window is A, right window is B, bottom window is the merged one. Jul 11 07:46:06 rwhitby-away> n for next, p for prev, a or b to select which one to become the merged output, q to quit. Jul 11 07:46:13 rwhitby> ok, once it is finished, hit C-x C-c Jul 11 07:46:23 That got me all the way ;) Jul 11 07:46:54 Can we change it to Vim instead? Jul 11 07:46:56 <[g2]> well a and b n and p just beep Jul 11 07:47:08 <[g2]> I know how to get around VIM Jul 11 07:47:23 vim can do multiple windows too. Jul 11 07:47:45 <[g2]> and where's the highlighting ? Jul 11 07:47:53 There's got to me an env. var. like $EDITOR or something. Jul 11 07:48:03 syntax highlighting you mean? Jul 11 07:48:12 <[g2]> yeah for the diffs Jul 11 07:48:20 in emacs? Jul 11 07:48:29 [g2]: there's no highlighting at all? Have you pressed anything after you entered ediff (or whatever its called)? Jul 11 07:48:54 <[g2]> I closed a little window Jul 11 07:49:12 ah, ok. Then I think you've done the same as I did the first time Jul 11 07:49:16 <[g2]> there are highlights further down, but the key working Jul 11 07:49:37 I solved it this way: I opened another terminal, killed emacs, restarted :P Jul 11 07:49:51 then a/b/n/p/q worked Jul 11 07:50:51 not the nicest solution, but it works Jul 11 07:51:05 Gotta go eat dinner Jul 11 07:52:25 [g2], Would you say it's "totally" safe for me to flash apex on my slug at this point? Jul 11 07:52:58 You got JTAG? Jul 11 07:53:04 nope. Jul 11 07:53:06 Then no Jul 11 07:53:10 hehehe. Jul 11 07:53:11 <[g2]> VoodooZ_work, it's never "totally" safe doing anything with computers Jul 11 07:54:19 Get a JTAG adapter, install the software, make it work and then it'll be "safe" as you can make it Jul 11 07:55:11 ok, then, let me rephrase the question: would it work fine with the current openslug after it's flashed? Jul 11 07:55:49 apex has no interdependency with either openslug or unslung (or stock firmware for that matter) Jul 11 07:55:52 I just need to boot to the internal JFFS2 FLASH. no more. And in less time that redboot Jul 11 07:56:22 true but It might not support the FS in question. Just making sure. Jul 11 07:57:02 I just want to speed up the boot and save Flash in the process potentially. Jul 11 07:57:12 It does support JFFS2 the last time I heard Jul 11 07:57:28 judging from the page it does. Jul 11 07:57:50 night all Jul 11 07:57:52 night Jul 11 07:58:01 Night Jul 11 07:58:41 [g2], you still boot off your openslug with apex right? Jul 11 07:58:57 Aargh, 2.6.12 Jul 11 07:59:12 what's wrong with it? Jul 11 07:59:13 <[g2]> I don't have a 64MB Redboot so my TurboFat slug runs it almost all the time Jul 11 07:59:13 Sorry, dumb error. Jul 11 07:59:22 jbowler-zzz: np - it's fixed now Jul 11 07:59:32 "2.6.12" was the old one - I meant to check in "2.6.12.2" Jul 11 08:00:03 jbowler-zzz: are the actual files that are there correct, and just incorrectly named? Jul 11 08:00:09 VoodooZ_work: the nslu2 files should be in the 'oldset' place where they worked. Jul 11 08:00:28 [g2], I see. I'll read the apex page further to see what's involved. I just don't know how to configure it yet. Jul 11 08:00:38 <[g2]> VoodooZ_work, booting from the jffs2 is slower because it reads the nodes in the fs and has some decompression to do Jul 11 08:00:39 jbowler-zzz, oldset? Jul 11 08:00:48 rwhitby-asleep: yes, there was no dependency on the PV in the files Jul 11 08:01:08 [g2], ummm. still faster than redboot though right? Jul 11 08:01:24 VoodooZ_work: the versions which worked in openslug-kernel-2.6.11.2 are still used in later kernels. Jul 11 08:01:37 ok. Jul 11 08:01:42 VoodooZ_work: s/oldset/oldest/ Jul 11 08:01:51 :) Jul 11 08:02:17 Ah, fingers wroking out of sequence :) Jul 11 08:02:23 <[g2]> rwhitby-asleep, jbowler-zzz it appears that we are still building 2.6.12.2 Jul 11 08:02:23 hehe Jul 11 08:02:48 [g2]: that's correct. jbowler-zzz vouched for it :-) Jul 11 08:02:52 [g2]: no I glitched it, my build is 2.6.12 (I just checked), rwhitby fixed it to build 2.6.12.2 again Jul 11 08:03:26 [g2], rwhitby-asleep: it shows up in the build because ${S} is actually (should be) 2.6.12.2 Jul 11 08:03:48 03rwhitby 07org.nslu2-linux.dev * r2e199969... 10/Makefile: Broke some superfluous dependencies Jul 11 08:05:19 Need to check that this change is in: nail org.openembedded.nslu2-linux * ree06322d... /packages/linux/openslug-kernel-2.6.12/defconfig: Added CONFIG_VLAN_8021Q, required for vlan package Jul 11 08:06:12 <[g2]> so we are building 2.6.12.2 with the old 11.2 patches with the broken serial I/O and possible flash corruption ? Jul 11 08:07:53 NAiL's VLAN change is there (the defconfigs are identical for 2.6.12 and 2.6.12.2, as they should be.) Jul 11 08:08:37 [g2]: no, the 11.2 patches don't all apply - it's building with the patches I have had since before OpenSlug-2.0 was released. Jul 11 08:09:43 [g2]: I've no evidence of memory corruption (i.e. I've been running with this kernel since before OpenSlug-2.0 release without any problems.) Jul 11 08:10:05 * rwhitby-asleep lets jbowler-zzz and [g2] discuss the virtues of 2.6.11.2 vs 2.6.12.2 and goes to bed ... Jul 11 08:10:19 <[g2]> sweet dreams rwhitby-asleep Jul 11 08:11:20 [g2]: according to rwhitby the serial problem is still there in 2.6.12.2 (i.e no output until pivot_root, which I believe means that the kernel kprintf doesn't output to the serial line) - someone with a serial line needs to debug it. I suspect a config change (I would look at how kprintf puts the messages on both the serial and in the memory log). Jul 11 08:12:05 <[g2]> jbowler-zzz, it's a well known issue I think things need to be changed from uart to buart or something like that Jul 11 08:14:28 <[g2]> Before I was all up for moving to 2.6.12, however, after the potentential jffs2 mtd error was mentioned in #oe I think it we need to seriously its possible effects Jul 11 08:15:57 [g2]: shutdown -r also fails, both because the watchdog reset hangs (most of the time with a disk attached) and because I couldn't get soft reboot to work. This is slugbug 145. Jul 11 08:16:30 <[g2]> nod... that's still an open issue Jul 11 08:16:52 [g2]: openslug disk errors (irq26 - nobody cared) still happens, the slugbug (109) is marked as critical but I don't believe it (or I wouldn't be running OpenSlug-2.1+ on my central file server ;-) Jul 11 08:17:18 <[g2]> it's actually a "Significant" error Jul 11 08:17:38 <[g2]> but as you noted it's not critical Jul 11 08:17:58 'severity' on 109 is 'critical', not 'significant'. Hum, I'm going to fix that, it isn't critical. Jul 11 08:18:50 <[g2]> I'd been hoping Lennert would have that fixed by now Jul 11 08:19:38 I was wondering why we don't build the kernel with preempt Jul 11 08:19:59 jacques: I tried, it crashed, but that was 2.6.11.8-mm4 Jul 11 08:20:39 <[g2]> iirc there was some mention of issues with preempt on lak Jul 11 08:21:06 jacques, good question. I just modded my defconfig to enable it. Jul 11 08:21:29 Mine worked fine for 2 months with preempt. Jul 11 08:21:37 Under heavy load? Jul 11 08:22:21 VoodooZ_work: which kernel? Jul 11 08:22:24 reading a usb2.0 camera at full speed and processing the image. and other stuff. Jul 11 08:22:51 probably 2.6.11.2 as I just switched to monotone last week and thus 2.6.12. Jul 11 08:23:05 My x86 gentoo system with preempt (2.6.11.3-mm4) fills /var/log/messages with whines about the device drivers not being preemptible Jul 11 08:23:07 ah I forgot there were posts about it on lak Jul 11 08:23:09 I'll be doing more test this week with 2.6.12 and I'll let you know Jul 11 08:23:33 Eh? I didn't think preempty was in the kernel main line in 2.6.11 Jul 11 08:24:04 Unfortunately a lot of this stuff is only ever properly tested on x86 Jul 11 08:25:37 CONFIG_PREEMPT is there in both 2.6.11.* and 2.6.12.*, but maybe the -mm stuff had something else. Also when I say 'crashed' I mean I didn't get to a dropbear prompt... Jul 11 08:26:00 Let me flash my new kernel and I'll run some tests. Jul 11 08:27:41 We've got a fixed set of device drivers, so getting it working safely is potentially a lot easier than on x86. Jul 11 08:28:02 For the core stuff anyway. If you want other USB devices then you're on your own :) Jul 11 08:28:15 I'm just worried about the IXP driver Jul 11 08:28:46 Most of that is just one huge blackbox because of the lib Jul 11 08:29:14 * VoodooZ_work is flashing... Jul 11 08:29:21 <[g2]> I think lennert had a debugging interface setup for the NPEs Jul 11 08:30:31 [g2]: what systems have MTD problems on 2.6.12.2? (I'm thinking that if it is a driver issue it's almost certainly a non-issue for NSLU2) Jul 11 08:31:21 iirc, either pb_ or mickeyl mentioned it Jul 11 08:31:43 My log doesn't go back that far Jul 11 08:31:45 * VoodooZ_work is booting openslug blind Jul 11 08:31:45 <[g2]> jbowler-zzz, mickeyl is in #openslug -- he was the one mentioning it was at the MTD layer Jul 11 08:32:01 <[g2]> we should ask him Jul 11 08:32:07 VoodooZ_work: which kernel, and why blind? Jul 11 08:32:23 <[g2]> because kernel serial output is busted in 2.6.12 Jul 11 08:32:27 the latest and greatest. 2.6.12.2 w/ preempt. Jul 11 08:32:28 no serial until pivot_root? Jul 11 08:32:28 nod Jul 11 08:33:05 yep. no serial from "done booting the kernel" to "Openslug Booting.." Jul 11 08:33:25 well.. doesn't matter very often, does it? ;) Jul 11 08:33:40 only if you can't mount a root fs, or have done changes to the kernel ? Jul 11 08:33:42 <[g2]> that's very important to me Jul 11 08:33:56 I want serial as well Jul 11 08:34:16 yep. Jul 11 08:34:23 [g2]: edit your local.conf as described on the wiki to choose the kernel you want. Jul 11 08:34:27 booted fine with preempt. Jul 11 08:34:46 VoodooZ_work: do you have ethernet? Jul 11 08:34:53 yep. Jul 11 08:35:20 would uname -a mention the preempt option like modinfo does? Jul 11 08:35:23 Hum... Annoying that it can't be set in local.conf. Jul 11 08:35:44 jacques, VoodooZ_work: what are the arguments for CONFIG_PREEMPT? Jul 11 08:36:00 (The one against is that something may stop working...) Jul 11 08:37:08 jbowler-zzz, uh? in defconfig you mean? I only put CONFIG_PREEMPT=y. Jul 11 08:38:57 damn, I think my build got dirty or something as it won't load my preempt modules. Probably should have bb cleaned the packages. dope! Jul 11 08:38:59 brb Jul 11 08:39:22 I mean, are then any arguments for changing the OpenSlug release to use CONFIG_PREEMPT=y? Jul 11 08:42:01 <[g2]> Is anticipatory I/O in 2.6.x ? or just in the -mm series ? Jul 11 08:44:23 jbowler-zzz, no clue. Jul 11 08:44:48 [g2], yeah, that could be useful too. Jul 11 08:45:27 <[g2]> I would think we'd get our biggest gains from anticiptory I/O and USB/Ethernet optimizations Jul 11 08:45:36 It looks like my openslug build inserted the preempt version of the modules but the old non-preempt version of the kernel! dang! Jul 11 08:46:19 that's not a good thing Jul 11 08:46:57 I messed up during the build Jul 11 08:46:59 <[g2]> there was mention in the Intel manual about locking data storage for the stack of an ISR to use the mini data cache area Jul 11 08:47:22 we don't use that at all do we at the moment? Jul 11 08:47:41 <[g2]> I thought of the stack for the USB disk or for packet data Jul 11 08:47:47 No, only the standard kernel locking is done. Jul 11 08:48:24 The standard stuff only knows about locking vm pages, if I read it correctly. Jul 11 08:48:25 <[g2]> Also using the second NPE for DMA x-fers would might help offload the processor Jul 11 08:49:07 <[g2]> s/would might/might Jul 11 08:51:41 <[g2]> IMHO my guess is the biggest gain would come from the anticipatory I/O on a HD with a cache buffer Jul 11 08:52:10 <[g2]> or using the minicache for storing some of that data Jul 11 08:52:22 <[g2]> or actually the main data cache Jul 11 08:53:40 <[g2]> with a block size of 512 that'd be 4 blocks per 2K. Jul 11 08:54:18 NAiL: sounds like we need a generic solution to the 'install/strip' problem. I'm thinking that I can make a really dumb install-native which drops the correct shell script into staging/host/bin. Jul 11 08:55:49 * VoodooZ_work is starting a build from scratch while he's out for lunch.. Jul 11 08:57:52 NAiL: slugbug 190, assigned to me - feel free to grab it if you (or anyone) wants to fix it before I do... Jul 11 08:59:05 <[g2]> The default in the kernel 2.6 is the anticipatory scheduler. Jul 11 09:00:42 rwhitby-asleep: are you aware of a problem with default DHCP? I only have very limited environments in which to test it (so if anyone encounters any problems please enter a slugbug) Jul 11 09:05:00 <[g2]> grep "io scheduler" m Jul 11 09:05:00 <[g2]> io scheduler noop registered Jul 11 09:05:00 <[g2]> io scheduler deadline registered Jul 11 09:05:17 <[g2]> that's from the 2.6.11.2 kernel dmesg Jul 11 09:09:51 As I remember it there is little point having more than one scheduler and little point in having anything other than the smallest. I think we're using noop at present, but deadline is built too. Jul 11 09:13:47 Yes, it looks like the primary method of changing the scheduler is the kernel command line, though I thought there was a user space way to do it too. Jul 11 09:16:26 NAiL: there may be a bug in the priority handling for the kernels, because it looks like kernel 2.6.12 is being build in preference to 2.6.12.2, might just be me... Jul 11 09:23:57 NAiL: false alarm, it's me Jul 11 09:28:05 <[g2]> I don't think we are building AS right now Jul 11 09:28:22 <[g2]> so we can't try it out without a rebuild Jul 11 09:30:34 That's correct. Jul 11 10:00:28 jbowler-zzz, yeah, would this explain why I had the wrong kernel in my final image? Jul 11 10:01:01 oops. sorry I didn't fast enough. Jul 11 10:46:30 my roboslug seems to work fine with preempt right now. Jul 11 12:40:01 jbowler-zzz: yes, the generic solution sounds nice Jul 11 12:42:37 jbowler-zzz: At the moment I'm not gonna take on any more bugs. My ToDo list kills whichever wiki I try to put it on :-P Jul 11 12:42:54 What do you mean by kill the wiki? Jul 11 12:43:01 I've been meaning to ask you about that Jul 11 12:43:25 Well, not kill it, but it cuts off the text and adds an � Jul 11 12:43:34 "�" Jul 11 12:43:45 Email me what you're trying to add Jul 11 12:43:47 It apparently only happens to me though Jul 11 12:45:41 Tiersten: mail sent Jul 11 12:46:18 uhm... trying to send, actually Jul 11 12:46:31 How big IS this? Jul 11 12:47:12 not *that* big.. like a couple of pages Jul 11 12:47:19 It should be fine then Jul 11 12:47:20 There are bigger pages on the wiki afaik Jul 11 12:48:20 What happens is basically this: Jul 11 12:48:31 I have a page that is somewhat long, and add a few lines Jul 11 12:48:45 When I press save, it doesn't save, and just returns me to the edit page Jul 11 12:48:55 It should print out something Jul 11 12:48:59 but the text in the textbox has now been cut off Jul 11 12:49:12 and been replaced by a single "�" Jul 11 12:49:13 You sent the email yet? Jul 11 12:49:24 I seem to be having mail server woes Jul 11 12:49:39 It doesn't say *anything* about any error either... that's what's bugging me Jul 11 12:49:54 you got weird 8 bit chars in this or something? Jul 11 12:49:55 hmm.. I think I'm gonna restart the mailserver Jul 11 12:50:00 it's definitely something on your end... Jul 11 12:50:05 yes Jul 11 12:50:07 well Jul 11 12:50:20 I can't understand *what* Jul 11 12:51:09 aha Jul 11 12:51:14 * NAiL starts to get a clue Jul 11 12:51:34 The mail was sent when I switched off the hardware firewall Jul 11 12:52:05 You doing some sort of transparent filtering? Jul 11 12:52:12 Not so transparent anyway Jul 11 12:52:29 Apparently not very good Jul 11 12:52:39 The wiki thing works now Jul 11 12:53:05 Did you get the mail? Jul 11 12:53:57 yes Jul 11 12:54:03 ok Jul 11 12:54:16 what firewall you using? Jul 11 12:54:36 It's the onboard NVidia stuff... I just hadn't switched it off Jul 11 12:55:24 Sounds a bit crap if it can't even pass basic HTML through... Jul 11 12:55:45 yes Jul 11 12:56:11 Wonder if a BIOS update or something fixes it... no idea how it actually works Jul 11 12:57:50 Might want to try the drivers as well Jul 11 12:59:03 yeah, only I updated them yesterday :P Jul 11 12:59:51 oh well.. Integrated stuff is bad. It's not exactly the reason I bought this card. I only found out after installing that it had a firewall-thingy Jul 11 13:01:52 (and it'n not like I'm the one that paid for it) Jul 11 13:05:24 Wasn't the integrated firewall supposed to be one of the big features of the nvidia range? :) Jul 11 13:05:38 heh Jul 11 13:05:42 Possibly Jul 11 13:05:46 Any way of configuring it not to fiddle with HTML content as it goes past? Jul 11 13:06:10 No idea.. I've barely looked at it.. It's actually using apache for configuration ;) Jul 11 13:06:22 <[cc]smart> hi all... somebody willing ot care about /var on tmpfs issue ? Jul 11 13:06:26 that's interesting Jul 11 13:07:08 <[cc]smart> and all got quiet :) Jul 11 13:07:39 [cc]smart: I'm very interested. I vote for most of var on disk, but log, run and tmp on ramfs. :) Jul 11 13:08:00 <[cc]smart> me,too. just a hero needed to do it :) Jul 11 13:08:03 That'll leave spool, mysql, and other stuff like that persistent Jul 11 13:08:53 <[cc]smart> sigh Jul 11 13:09:00 <[cc]smart> will think about it tomorrow Jul 11 13:09:10 <[cc]smart> in any case, it seems cyrus stuff wil come out nice. Jul 11 13:09:15 yeah Jul 11 13:09:25 btw, what user does cyrus need? Jul 11 13:09:46 <[cc]smart> typicaly it's caled cyrus and simply is any non prioviledged account Jul 11 13:09:59 <[cc]smart> dunno if there is an informal agreed uid used Jul 11 13:10:13 I talked to mickeyl about creating users in ipkg, and he gave me a small task.. Standardize the way to do it for all of OE :P Jul 11 13:10:51 <[cc]smart> i would have thought there should be a "sample" passwd file Jul 11 13:11:08 <[cc]smart> where all the system accounts possible form a default setting Jul 11 13:11:27 yeah, I had started doing that with base-passwd Jul 11 13:11:29 <[cc]smart> then in .bb scripts there would be like a new variable that would name the accounts needed by the package Jul 11 13:11:58 <[cc]smart> so oyu still in the ned only have accounts you need Jul 11 13:12:20 but it looks like the better way is to create a new .bbclass that makes it easy to implement in each ipk that is installed Jul 11 13:12:46 how'd you plan on updating the passwd on an already installed system? Jul 11 13:13:22 <[cc]smart> no different. you could prepend a wite out of a passwd line by the deletion of that same one Jul 11 13:14:42 I'm not sure I get what you're talking about. If that solution was in place now, and you have just flashed openslug. If I then add a package in OE, and do the passwd mumbo-jumbo, will my change end up in your passwd file when you install the ipkg? Jul 11 13:14:54 Just using adduser is simpler, no? Jul 11 13:15:10 <[cc]smart> like grep -v cyrus: /etc/passwd >/tmp/passwd; echo "cyrus:...." >> /tmp/passwd; cp /tmp/passwd /etc/passwd Jul 11 13:15:27 not good idea Jul 11 13:15:29 well Jul 11 13:15:35 I *thought* about that Jul 11 13:15:46 But how do you *guarantee* that the UID isn't already taken? Jul 11 13:16:04 <[cc]smart> that's the job of the master passwd file in OE repository Jul 11 13:16:21 How can that file know what users have been added to the local system? Jul 11 13:16:46 <[cc]smart> nonono Jul 11 13:16:50 <[cc]smart> exampl: Jul 11 13:17:09 <[cc]smart> you add a new file to OE: openembedded/files/passwd.master Jul 11 13:17:30 <[cc]smart> passwd.master contains all possible system accounts as goes: Jul 11 13:17:47 <[cc]smart> root:x:0:0:root:/root:/bin/bash Jul 11 13:17:54 I get that part Jul 11 13:17:56 <[cc]smart> squid:x:31:31:Squid:/var/cache/squid:/bin/false Jul 11 13:17:58 <[cc]smart> and so on Jul 11 13:18:03 <[cc]smart> including cyrus and whatnot Jul 11 13:18:14 mysql, ntpd, postfix, you name it Jul 11 13:18:25 <[cc]smart> then in your package dependent .bb file you introduce a new variable: Jul 11 13:18:40 <[cc]smart> USED_SYSTEM_ACCOUNTS = "cyrus" Jul 11 13:18:43 cyrus is more complicated. I think you need to set a password for the cyrus user since it's the default admin. Jul 11 13:19:15 Tiersten: Ok, that can be handled by the ipkg. If not automatically, prompt the user to do so. Jul 11 13:19:29 <[cc]smart> then the pipkg adds a line in passwd, a line in shadow and possibly lines in group which are associated with that user Jul 11 13:19:53 <[cc]smart> you can per se let this system write a default passwd Jul 11 13:20:05 <[cc]smart> it's alread in the passwd.master shadow.master group.master Jul 11 13:20:15 hm.. should I write something in the wiki about using sane and vlans? Jul 11 13:20:51 <[cc]smart> DaKa2: if there is OE/slug specific information Jul 11 13:20:52 <[cc]smart> sure Jul 11 13:21:07 <[cc]smart> if general only, i like the ease of linking :) Jul 11 13:21:33 well, atleast a bit oe/slug specific Jul 11 13:21:53 <[cc]smart> then you'll prolly have happy readers on the page Jul 11 13:22:20 hm.. ok, bb when I have something on "paper" Jul 11 13:22:53 <[cc]smart> now what i don't know is iif you guys think the master file approach makes sense Jul 11 13:22:57 <[cc]smart> dunno really either Jul 11 13:23:03 <[cc]smart> but at least its an idea .) Jul 11 13:23:32 does it matter if the info is specific for openslug? dont have unslung on anything.. Jul 11 13:23:53 <[cc]smart> you can decide where to introduce the page :) Jul 11 13:24:00 good Jul 11 13:26:07 <[cc]smart> goto(bed); sleep(8) Jul 11 13:26:19 8 seconds? Jul 11 13:26:22 damn.. Jul 11 13:26:24 <[cc]smart> hrhr Jul 11 13:26:38 [cc]smart: goto(bed); sleep(`echo $(8*3600)`) Jul 11 13:26:45 thats better Jul 11 13:27:05 gn Jul 11 13:27:07 or, cd /dev/bed && sleep(`echo $(8*3600)`) Jul 11 13:27:25 <[cc]smart> cat consciousnes >/dev/null Jul 11 13:27:58 cat /dev/urandom > /dev/brain Jul 11 13:28:23 <[cc]smart> dd if=/dev/random/alcohol of=/dev/bloodstream bs=kettle count=50 Jul 11 13:28:58 ... not good.. Jul 11 13:29:06 mv [cc]smart /dev/hospital Jul 11 13:29:14 <[cc]smart> :) Jul 11 13:29:53 <[cc]smart> nuff crazy stuff for today. need some left for tomorrow... Jul 11 13:29:58 <[cc]smart> n8 Jul 11 13:30:01 :) hehe. nite Jul 11 13:30:55 I have a small question about busybox? Jul 11 13:32:02 I have a script which is a factor slower in busybox then in bash. Jul 11 13:32:29 I think the problem is that busybox forks for [ ... ] and bash not. Jul 11 14:01:31 What is the configuration for /bin/busybox (not the 1.0 /opt/bin/busybox) Jul 11 14:13:48 hi people. I had problems building cron using the openslug-2.0 tarball. I made a patch that fixes the problem. Who/where can I usefully send this patch to? Jul 11 14:15:45 hannesr: you can send it to me Jul 11 14:15:57 but I think I might already have fixed this Jul 11 14:16:02 What caused it not to compile? Jul 11 14:16:43 error from GNU strip Jul 11 14:16:47 aha Jul 11 14:16:56 in do_install, I assume? Jul 11 14:16:59 it was calling the build machine's strip on a target binary Jul 11 14:17:02 that's the one Jul 11 14:17:11 exactly.. how did you fix it? :) Jul 11 14:17:46 I added a patch to cron's makefile that makes it use an explicit "strip" instead of "install -s". Jul 11 14:18:38 then I passed the "make" a "STRIP=${TARGET_SYS}-strip" from the .bb file, so it uses the right strip binary Jul 11 14:19:01 (the makefile calls it as "$(STRIP)") Jul 11 14:19:13 aha Jul 11 14:19:37 jbowler-zzz: ping? Jul 11 14:20:16 we have an alternate solution, but I'd say yours is better Jul 11 14:20:21 <[g2]> hannesr, good find Jul 11 14:20:57 okay, can I mail someone a patch? Jul 11 14:21:09 just send it to nail@nslu2-linux.org Jul 11 14:21:18 <[g2]> if I may ask, what are you planning on using openslug for ? Jul 11 14:21:35 <[g2]> (other that some periodic task) :) Jul 11 14:21:52 at the moment, I have a slug as a backup server for a few unix boxes Jul 11 14:22:02 hence the need for cron :=) Jul 11 14:22:15 hannesr, NAiL, sorry if you have already thought of this or if it's irrelevant, but autotooled packages should use install-sh to cross-strip. if you are working on an autotooled package, the right move may be to autoreconf with automake 1.6 or greater Jul 11 14:22:17 <[g2]> so you periodically pull stuff Jul 11 14:22:22 once that's sorted, I want it to play mp3s... Jul 11 14:22:42 g2, yeah, using rsnapshot (perl script around rsync) Jul 11 14:23:03 <[g2]> I'm guessing you built perl natively Jul 11 14:23:03 jp30-work: I'm not that much into this really, gonna talk to jbowler about it. He's been adding install-sh scripts to a few packages that break this way. Jul 11 14:23:09 jp30-work, cron doesn't use autotools, unfortunately Jul 11 14:24:07 <[g2]> hannesr, rsync from the feed ? Jul 11 14:24:37 g2, nope, got a workaround for the cross-compiled version (see http://www.nslu2-linux.org/wiki/OpenSlug/ReleaseNotes2x0) Jul 11 14:24:58 hannesr: I'd be very interested in hearing what packages you are using that works, and what doesn't work. I'm trying to clean up the stable feed a bit. Jul 11 14:25:05 g2, yes rsync from the feed... why, is it broken? Jul 11 14:25:15 <[g2]> I don't think so Jul 11 14:25:46 NAiL, I was working with an old bitkeeper checkout until recently, I'm just in the process of upgrading to the 2.0 version Jul 11 14:26:20 hannesr: ok, I'd be happy for bugreports if you run into problems ;) Jul 11 14:26:28 So far I haven't seen anything broken other than perl and the cron build. Jul 11 14:26:43 I'll be sure to let you know if I see other badness. Jul 11 14:27:09 thanks Jul 11 14:27:29 <[g2]> and you are aware of #openslug right ? :) Jul 11 14:28:05 g2, I wasn't - is this channel supposed to be unslung-focused, or what's the diff? Jul 11 14:28:23 <[g2]> hannesr, so you did the "echo /lib/libgcc_s.so.1 > /etc/ld.so.preload" Jul 11 14:28:29 yep Jul 11 14:28:55 I can't say I understand what it side-effects it may have, but it's been working for me :) Jul 11 14:29:03 <[g2]> We talk openslug in both Jul 11 14:29:06 (i just googled it up) Jul 11 14:30:13 <[g2]> but if you're openslug only I wanted you to be aware of the channel Jul 11 14:32:40 <[g2]> hannesr, are you using nfs something else to pull the files ? Jul 11 14:33:14 rsync over ssh, with passwordless keys (ugh) Jul 11 14:33:49 <[g2]> ugh because it's slow with ssh ? Jul 11 14:34:21 more ugh because of the security implications of the passwordless keys. Jul 11 14:34:22 <[g2]> or ugh because you don't like using the passwordless keys Jul 11 14:34:39 although I have a script to at least limit them to running "only" rsync Jul 11 14:35:16 slowness isn't a problem for pulling backups, unless there's a _lot_ to back up... Jul 11 14:35:38 <[g2]> well different ppl backup different things :) Jul 11 14:35:40 anyway, rsync makes it fast again, as it intelligently checks for changes Jul 11 14:35:50 <[g2]> nod. there Jul 11 14:35:58 <[g2]> after the initial sync Jul 11 14:36:18 ... so even running it on 25 gigs of mp3s is tolerable :-) Jul 11 14:36:55 <[g2]> hmmm... I'd guess that'd be around 6K songs Jul 11 14:37:34 it's scary what builds up over time. Jul 11 14:38:15 <[g2]> it's scary that an 80G external HD is < $99 Jul 11 14:39:28 <[g2]> you'be probably got no plans to use samba or nfs then right ? Jul 11 14:39:45 g2, not anytime soon Jul 11 14:40:02 <[g2]> Ok... Well THX a bunch for all the answers Jul 11 14:40:16 there are some windows users in the house that could benefit from sambe, but i don't think I'll bother... Jul 11 14:40:26 <[g2]> Openslugger's are a special breed Jul 11 14:40:40 <[g2]> we don't hear a whole bunch from them Jul 11 14:40:42 "special" is right... Jul 11 14:41:29 I'll be sure to pop in if I come across / break anything interesting. Jul 11 14:41:29 <[g2]> as a dev I'm always curious what ppl are really doing with the box Jul 11 14:41:39 <[g2]> cool. thx Jul 11 14:41:54 (or when i'm finally looking for someone with a jtag ;-) Jul 11 14:42:05 JTAG is my friend. Jul 11 14:42:11 <[g2]> are you interested in JTAG at all ? Jul 11 14:42:21 <[g2]> JTAG is my friend too Jul 11 14:42:38 I meant "when I manage to accidentally erase redboot" Jul 11 14:42:52 it's only a question of time... Jul 11 14:42:53 <[g2]> probably unlikely Jul 11 14:42:53 hehe at least for you it would have been an accident.... Jul 11 14:43:12 for me it's, "When I intentionally blow away redboot...." Jul 11 14:43:15 <[g2]> APEX is quite nice though :) Jul 11 14:43:17 which reminds me - what's the difficulty in upgrading the kernel image without reflashing everything? Jul 11 14:43:48 <[g2]> it's gotten much better Jul 11 14:44:01 <[g2]> there are just minor tracking issues Jul 11 14:44:32 i had a quick go at stuffing it in there using redboot (fis write blah), but after I did that the "boot" fell down in a heap. Jul 11 14:44:50 gotta keep the modules straight. Jul 11 14:45:03 so I guess you can't just write a zImage into the flash partition and expect it to work Jul 11 14:45:16 <[g2]> I think jbowler-zzz added a kernel update command Jul 11 14:45:35 <[g2]> hannesr, actually that's what I do with APEX Jul 11 14:46:07 <[g2]> I can ipkg update the zImage and the kernel-modules that are installed with a dest of /initrd Jul 11 14:46:08 g2, which is where your fried jtag comes in Jul 11 14:46:18 nice Jul 11 14:46:38 <[g2]> actually, I've been running APEX for months on the SLUGs but I don't have JTAG there yet Jul 11 14:46:43 * [g2] is a bad example Jul 11 14:47:00 brave. Jul 11 14:47:04 <[g2]> however, I did test it out on an OEM board and worked with beewookie Jul 11 14:47:16 <[g2]> and the OEM board had JTAG Jul 11 14:47:41 <[g2]> so it was only after lots of testing and others like dyoung and run it on the NSLU2 Jul 11 14:48:08 <[g2]> heh bewookie Jul 11 14:48:14 <[g2]> that's beewoolie Jul 11 14:48:15 03marceln * 10unslung/sources/ipkg-web/package.cgi: A web front end for ipkg Jul 11 14:48:34 lol, a wookie with yellow and black stripes.. Jul 11 14:48:43 <[g2]> LOL Jul 11 14:49:05 <[g2]> depending on how security concious you are, APEX is a nice update in that regard Jul 11 14:49:45 why security? is it apex network-accessible? Jul 11 14:50:13 <[g2]> no network accessibilty and no back doors Jul 11 14:50:21 <[g2]> like upslug Jul 11 14:50:58 <[g2]> sometimes security and useability are opposing one another Jul 11 14:50:59 right, upslug is a bit worrying if you have h4X0rs on your local network Jul 11 14:51:17 <[g2]> or if some just gets on the box Jul 11 14:51:39 <[g2]> they can trash the kernel part and reboot and it comes up waiting for an image Jul 11 14:51:49 <[g2]> beewoolie-afk, HEY Jul 11 14:51:59 <[g2]> I guess your ears were burning :) Jul 11 14:52:06 or they can just reboot and telet to redboot for real fun Jul 11 14:52:08 Hey beewookie. Jul 11 14:52:24 <[g2]> hannesr, there's a timing issue there Jul 11 14:52:57 <[g2]> beewoolie-afk, I don't know if it's in the stable feed but samba is all worked out Jul 11 14:53:16 <[g2]> thanks to NAiL jbowler-zzz et. al. Jul 11 14:53:22 beewookie-afk, nfs should be workable too. Jul 11 14:53:31 <[g2]> nfs was fixed too Jul 11 14:53:36 <[g2]> nfs-utils this Jul 11 14:53:51 <[g2]> but I don't know if they are pushed to the stable feed or not Jul 11 14:54:04 <[g2]> NAiL, ??? Jul 11 14:54:20 <[g2]> hannesr, NAiL's the openslug package mgr Jul 11 14:54:33 <[g2]> morning sunshine! Jul 11 14:58:47 jbowler-zzz: no problem with openslug dhcp as far as I am aware - it worked fine - I just forgot :-) Jul 11 15:10:31 <[g2]> jbowler-zzz, so "reflash -j jffs2.image" is my friend correct ? Jul 11 15:10:36 [g2]: K. I'll see if it pulls. Jul 11 15:11:26 <[g2]> On samba I just needed to copy the samba.cnf example in /etc/samba and do a little editing Jul 11 15:11:37 <[g2]> and create a smbpaswd Jul 11 15:11:47 <[g2]> and /etc/init.d/samba start Jul 11 15:12:06 <[g2]> quite fun actually Jul 11 15:12:59 <[g2]> and with the gphoto2 stuff working I think I'll be able to setup something for automatic dl'ing of the digital camera Jul 11 15:14:04 [g2]: reflash -i foo.img is a better friend, cause then you get the right modules too Jul 11 15:16:28 <[g2]> rwhitby, I'm looking to go kernel partition less Jul 11 15:16:55 <[g2]> and the modules are in the jffs2 partition no ? Jul 11 15:16:55 ah, ok, - cool Jul 11 15:43:29 samba in the stable feed is worked out Jul 11 15:44:01 hannesr: It might be possible to implement the strip-stuff in a bbclass, so it's not so ugly Jul 11 15:44:52 hannesr: .ie btw? Do you live there? Lived in Cork for 13 months :) Jul 11 15:45:10 that would probably be the cleanest, if you could just do "inherit strip" or similar Jul 11 15:45:17 exactly Jul 11 15:45:24 I used to live in Limerick Jul 11 15:45:40 Ah, the place I never got around to going to :P Jul 11 15:45:58 you missed an opportunity to get stabbed... Jul 11 15:46:30 I'm in .de now. Jul 11 15:47:33 [g2]: nfs-utils and samba are pushed to stable Jul 11 15:47:36 so they work Jul 11 15:47:41 just so you know Jul 11 15:47:43 <[g2]-away> Na Jul 11 15:47:47 <[g2]-away> NAiL, cool Jul 11 15:47:51 <[g2]-away> beewoolie-afk, will be happy Jul 11 15:47:53 <[g2]-away> bbl Jul 11 15:48:36 NAiL, what would the bbclass do? putting a symlink to armeb_linux-strip on the path is the easiest thing that comes to mind. Jul 11 15:49:31 hannesr: to be honest, I haven't started looking at bbclasses yet, but from what I've understood, it shouldn't be too hard to do it that way Jul 11 15:50:47 I have to start learning though, because I'm going to do a adding-users-thing through a bbclass Jul 11 15:52:39 I've only been doing this oe-stuff for a few weeks Jul 11 15:52:51 learning as I go :) Jul 11 15:53:04 adding-users-thing? like having a user to own the files belonging to a particular package? Jul 11 15:54:25 adding a user when an ipkg is installed Jul 11 15:54:49