**** BEGIN LOGGING AT Mon Mar 06 10:59:56 2006 Mar 06 12:09:01 03oleon * 10unslung/Makefile: optware: wl500g many packages promoted. - Promote ALL python pkgs, appweb, cdargs, cogito, typd-topfield, jikes, mutt, nmap, nload, nrpe Mar 06 13:46:25 03oleon * 10unslung/make/nrpe.mk: nrpe: Upstream upgrade from 2.0 to 2.4 Mar 06 13:49:20 03oleon * 10unslung/ (Makefile make/nagios-plugins.mk): nagios-plugins: add -lm for wl500g Mar 06 14:16:58 03oleon * 10unslung/make/asterisk.mk: asterisk: Upstream upgrade to 1.2.5 Mar 06 15:17:24 rwhitby: nagios-plugins got stucked by configure on wl500g. Need make nagios-plugins-dirclean on nudi Mar 06 15:45:24 This state should not happen. I suggest to update template.mk with $(_BUILD_DIR)/.configured: $(DL_DIR)/$(_SOURCE) $(_PATCHES) make/.mk so whenever .mk is changed package is reconfigured. Mar 06 15:47:59 03oleon * 10unslung/make/nagios-plugins.mk: nagios-plugins: add .mk depend on .configured Mar 06 15:52:59 03oleon * 10unslung/make/template.mk: template.mk: force package reconfigure whenever .mk changes Mar 06 17:19:11 oleo, thx for the lua fix Mar 06 17:35:42 03oleon * 10unslung/make/python.mk: python: add .configured depend on .mk Mar 06 17:36:32 [g2]: do you compile madwifi native or cross? Mar 06 18:46:34 What is openslug's default tick rate? I just read that pretty much all kernels since 2.5 are at 1000Hz as opposed to the old default of 100Hz. Mar 06 18:47:20 I just read an old kerneltrap posting that states that they wanted to change 2.6.13 to have a default of 250Hz. I'm wondering because it's important for my embedded app. Mar 06 19:04:00 I've found the defconfig option to be CONFIG_HZ_1000, CONFIG_HZ_250 or CONFIG_HZ_100. I guess the default is 250 now. I'll try some tests at 1000Hz. Mar 06 19:11:31 Hi, what is the fastest file sharing method with the nslu2 so far? Mar 06 19:11:44 I would like ot stream HDTV from the SLUG.. Mar 06 19:15:34 DVB Transport Streams are about 30000 symbols/second. DVD 8kbit/s. HDTV is a stream? I think, any method will do. Mar 06 19:19:01 I've got stuck with python build for wl500g, again. If rwhitby or jp30-work will not clean staging dir, I need python-stage-clean! Mar 06 19:22:08 oleo: what needs to be cleaned, and how? Mar 06 19:23:02 I think that rm add Mar 06 19:23:13 oops Mar 06 19:24:36 I think that removing /home/slug/optware/wl500g/staging/opt/lib/libpython2.4.so will do. It looks like python is linking its own libs from previous stage, which is broken. Mar 06 19:24:47 aha Mar 06 19:25:10 I can look at in a little while. Just need to get the box with my ssh-key online so I can log in ;) Mar 06 19:32:05 [g2]: debian:~# modprobe ath_pci Mar 06 19:32:05 ath_rate_sample: 1.2 (svn 2006-03-06) Mar 06 19:32:05 ath_pci: 0.9.4.5 (svn 2006-03-06) Mar 06 19:32:05 PCI: enabling device 0000:00:02.0 (0140 -> 0142) Mar 06 19:32:05 wifi0: 11b rates: 1Mbps 2Mbps 5.5Mbps 11Mbps Mar 06 19:32:05 wifi0: 11g rates: 1Mbps 2Mbps 5.5Mbps 11Mbps 6Mbps 9Mbps 12Mbps 18Mbps 24Mbps 36Mbps 48Mbps 54Mbps Mar 06 19:32:07 wifi0: turboG rates: 6Mbps 12Mbps 18Mbps 24Mbps 36Mbps 48Mbps 54Mbps Mar 06 19:32:09 wifi0: H/W encryption support: WEP AES AES_CCM TKIP Mar 06 19:32:11 wifi0: mac 5.9 phy 4.3 radio 4.6 Mar 06 19:32:13 wifi0: Use hw queue 1 for WME_AC_BE traffic Mar 06 19:32:15 wifi0: Use hw queue 0 for WME_AC_BK traffic Mar 06 19:32:17 wifi0: Use hw queue 2 for WME_AC_VI traffic Mar 06 19:32:19 wifi0: Use hw queue 3 for WME_AC_VO traffic Mar 06 19:32:21 wifi0: Use hw queue 8 for CAB traffic Mar 06 19:32:23 wifi0: Use hw queue 9 for beacons Mar 06 19:32:25 wifi0: Atheros 5212: mem=0x48010000, irq=27 Mar 06 19:32:37 looks nice :) Mar 06 19:33:23 yep.. i don't have the time to test it but I'd say that it works Mar 06 19:43:23 <[g2]> dwery awesome! Mar 06 19:43:40 [g2]: I just compiled it... Mar 06 19:44:26 <[g2]> dwery is that LE ? Mar 06 19:44:36 yes Mar 06 19:44:45 dwery: Did you look at the nice microstubler? :) Mar 06 19:44:48 you know, I have no BE rootfs :) Mar 06 19:45:20 <[g2]> dwery heh :) Mar 06 19:45:28 there's any package for debian I can use to test it? Mar 06 19:45:38 <[g2]> dwery but you didn't have any madwifi cards either :) Mar 06 19:45:57 <[g2]> it's simple to test Mar 06 19:45:58 that's the nas100d :) Mar 06 19:46:26 <[g2]> Ah... well then you can pull that card for your loft :) Mar 06 19:46:31 yes :) Mar 06 19:46:48 btw, anyone know where I might find a minipci-socket? Mar 06 19:46:54 <[g2]> dwery which svn version ? Mar 06 19:47:22 1440 Mar 06 19:47:40 <[g2]> hmmmm Mar 06 19:47:43 <[g2]> that's quite odd Mar 06 19:47:52 <[g2]> as the SVN says 2006-03-06 Mar 06 19:48:04 <[g2]> and 1440 fails for me Mar 06 19:48:08 <[g2]> in LE Mar 06 19:48:12 <[g2]> but works fine in BE Mar 06 19:48:19 strange. same card? Mar 06 19:48:30 <[g2]> same everything Mar 06 19:48:36 <[g2]> just different sw Mar 06 19:48:55 <[g2]> fails for jacques too Mar 06 19:49:04 is there any quick test you want me to do before I need to go? Mar 06 19:49:15 <[g2]> do a svn info Mar 06 19:49:33 I used the snapshot Mar 06 19:49:49 compiled with: Mar 06 19:49:50 make ARCH=arm KERNELPATH=/usr/src/linux-ixp4xx TARGET=xscale-le-elf TOOLPREFIX=armv5b-softfloat-linux- CROSS_COMPILE=armv5b-softfloat-linux- Mar 06 19:50:05 but had to modify hal/public/xscale-le-elf.inc Mar 06 19:50:09 to add -EL Mar 06 19:50:17 <[g2]> Ahh... Mar 06 19:50:17 in the line that starts with LD= Mar 06 19:50:24 using LDOPTS did not work Mar 06 19:50:35 <[g2]> did you remove the mshort-load-bytes thingy ? Mar 06 19:50:43 I did no other mods Mar 06 19:51:11 <[g2]> dwery UMMMM Mar 06 19:51:23 <[g2]> you're command lines says armv5B Mar 06 19:51:52 yes, it is the same I use for the other ARM related work Mar 06 19:51:54 armv5b-softfloat-linux-gcc -Wp,-MD,/srv/xtr/src/madwifi-ng-r1440-20060206/ath/.ath_pci.mod.o.d -nostdinc -isystem /opt/crosstool/armv5b-softfloat-linux/gcc-3.3.4-glibc-2.2.5/lib/gcc-lib/armv5b-softfloat-linux/3.3.4/include -D__KERNEL__ -Iinclude -include include/linux/autoconf.h -mlittle-endian -Wall -Wundef -Wstrict-prototypes -Wno-trigraphs -fno-strict-aliasing -fno-common -ffreestanding -Os -fno-omit-frame-pointer -fno-optimize-sibling-calls Mar 06 19:51:59 (example output) Mar 06 19:53:48 <[g2]> hmmm this is interesting Mar 06 19:54:12 ath0 23 channels in total; available frequencies : Mar 06 19:54:12 Channel 01 : 2.412 GHz Mar 06 19:54:12 Channel 02 : 2.417 GHz Mar 06 19:54:13 Channel 03 : 2.422 GHz Mar 06 19:54:13 Channel 04 : 2.427 GHz Mar 06 19:54:13 Channel 05 : 2.432 GHz Mar 06 19:54:15 Channel 06 : 2.437 GHz Mar 06 19:54:17 Channel 07 : 2.442 GHz Mar 06 19:54:19 Channel 08 : 2.447 GHz Mar 06 19:54:21 Channel 09 : 2.452 GHz Mar 06 19:54:23 Channel 10 : 2.457 GHz Mar 06 19:54:25 Channel 11 : 2.462 GHz Mar 06 19:54:27 Current Channel:0 Mar 06 19:55:13 do you have a console enabled sniffer tool? Mar 06 19:56:47 <[g2]> there's a monitor mode Mar 06 19:57:03 tell me Mar 06 19:57:15 <[g2]> 1 sec Mar 06 19:58:41 <[g2-lap]> modprobe wlan_scan_sta Mar 06 19:59:06 ok Mar 06 19:59:24 <[g2-lap]> wlanconfig ath0 list scan Mar 06 19:59:58 I don't have wl;anconfig Mar 06 20:00:00 <[g2-lap]> and if you haven't already Mar 06 20:00:02 wich package? Mar 06 20:00:05 which* Mar 06 20:00:11 <[g2-lap]> ifconfig ath0 up Mar 06 20:00:39 <[g2-lap]> look in the tools dir of the madwifi build Mar 06 20:00:59 <[g2-lap]> or maybe it's wireless-tools Mar 06 20:02:55 it is in madwifi Mar 06 20:02:58 but it got compiled BE :) Mar 06 20:03:41 <[g2-lap]> right the madwifi is probably compiled BE :) Mar 06 20:06:33 recompiling Mar 06 20:07:05 hi philpem Mar 06 20:07:11 hi pepijn Mar 06 20:07:16 what is the status on your freezing slug? Mar 06 20:07:56 [g2-lap]: debian:/tmp/madwifi-ng-r1440-20060206/tools# ./wlanconfig ath0 list scan Mar 06 20:07:56 debian:/tmp/madwifi-ng-r1440-20060206/tools# Mar 06 20:08:01 it says nothing... Mar 06 20:08:03 it hung again today in the middle of an scp of a huge file. kiwi syslog wasn't running at the time (%#$*ing windoze box hung) so i didn't get any logs Mar 06 20:08:20 :( Mar 06 20:08:24 <[g2-lap]> dwery did you ifconfig ath0 up ? Mar 06 20:08:35 too bad Mar 06 20:08:37 yes Mar 06 20:08:47 pepijn: even left the laptop going overnight to syslog it.. nada, just a load of firewall logs. Mar 06 20:08:56 <[g2-lap]> are there any APs around ? Mar 06 20:09:00 19 hours uptime. it's beginning to look like either an eciadsl bug or the usb hard drive... Mar 06 20:09:11 it's been a long time since I used wifi here :) Mar 06 20:09:31 I should have a wifi access point somewhere.. will test it tomorrow Mar 06 20:09:37 it's an 15gig ibm 75gxp in a vantec nexstar2 casing... Mar 06 20:09:39 and then not related to my ordeal Mar 06 20:10:20 still runing with the openslug 2.7 kernel? Mar 06 20:10:26 pepijn: yep Mar 06 20:10:50 bbl Mar 06 20:10:59 i've got debian sarge installed on vmware. the plan is to xcompile eciadsl-0.12 tonight and install that. 0.11 has "massive bugfix bonanza" in its CHANGELOG, 0.12 apparently has fixes that reduce cpu usage. Mar 06 20:11:29 what's REALLY annoying is that i can't consistently make it fail Mar 06 20:11:40 if i could, it would make debugging a hell of a lot easier Mar 06 20:12:17 yes, and then when it does not happen, you really don't know if it was fixed or not Mar 06 20:12:33 that is why i am interested Mar 06 20:12:42 yeah. i'm a sucker for hard projects :) Mar 06 20:12:44 <[g2-lap]> philpem what does eciadsl do ? Mar 06 20:12:54 g2-lap: http://eciadsl.flashtux.org Mar 06 20:13:12 it's a driver for the GlobeSpan USB ADSL modems. i'm using it with a fujitsu fdx310. Mar 06 20:13:39 you need to hang it off of your slug? Mar 06 20:13:44 yup Mar 06 20:14:05 usb2 hdd is on DISK1, adsl modem is on DISK2. Mar 06 20:14:19 <[g2-lap]> the fujitsu is a usb adsl modem right ? Mar 06 20:14:23 yup Mar 06 20:14:29 just sounds risky to have a fileserver directly connected to the net Mar 06 20:14:49 i've got iptables installed on it, and ssh is set to disallow root logins Mar 06 20:14:53 well, had that for year, just recently got it split up using the NSLU2 and WRT54G ;) Mar 06 20:14:57 openslug is a bit more bolted down than unslung though Mar 06 20:15:02 years even Mar 06 20:15:03 :) Mar 06 20:15:25 * NAiL usually has a slug internet-accessible Mar 06 20:15:26 <[g2-lap]> philpem you're running openslug right ? Mar 06 20:15:33 g2-lap: opendebianslug Mar 06 20:15:50 openslug 2.7 kernel+initrd flashimage, debian rootfs. Mar 06 20:15:54 <[g2-lap]> that's the BE one right ? Mar 06 20:15:57 yup Mar 06 20:16:16 philpem@wolf:~$ uname -a Mar 06 20:16:16 Linux wolf.philpem.me.uk 2.6.12.2 #1 Tue Sep 27 20:21:18 CDT 2005 armv5teb GNU/Linux Mar 06 20:16:26 <[g2-lap]> ah... Mar 06 20:16:37 <[g2-lap]> you're still on the old kernel and CSR ? Mar 06 20:16:46 CSR? Mar 06 20:16:52 <[g2-lap]> the intel NPE drivers Mar 06 20:17:00 <[g2-lap]> IAL/CSR and OSAL Mar 06 20:17:17 that would be the "ixp400" and "ixp425_eth" modules, right? Mar 06 20:17:23 yup Mar 06 20:17:25 <[g2-lap]> right Mar 06 20:17:33 <[g2-lap]> if it's ixp425_eth it's old Mar 06 20:17:54 so which kernel image should i upgrade to, and will i need to do anything to the debian fs? Mar 06 20:18:18 i thought 2.7 was the latest - thats what openslug.org says anyway Mar 06 20:18:20 just make sure you copy /lib/modules from the root fs the debian drive Mar 06 20:18:30 latest published Mar 06 20:18:53 philpem: the 3.6-beta release is somewhere around the corner Mar 06 20:19:03 <[g2-lap]> philpem here are my thoughts on the subject... take'm for what they are worth Mar 06 20:19:22 <[g2-lap]> 1) I'd build the latest openslug from the Makefile Mar 06 20:19:33 g2-lap: 1) the mastermakefile? Mar 06 20:19:37 <[g2-lap]> yeah Mar 06 20:19:37 mm.... Mar 06 20:19:51 I'm not so sure about that... Mar 06 20:20:11 might be better to check out the soon-to-be-release Mar 06 20:20:17 <[g2-lap]> NAiL is the Makefile borked right now ? Mar 06 20:20:22 no Mar 06 20:20:32 <[g2-lap]> are the images it builds borked ? Mar 06 20:20:36 no Mar 06 20:20:59 what i REALLY don't want to do it brick my slug to the point where i need to jtag it. Mar 06 20:21:00 <[g2-lap]> ok was what are you not sure about Mar 06 20:21:03 I think it should work, unless the libc6 dependency strikes Mar 06 20:21:19 libc6 deps shouldn't affect the kernel though... ? Mar 06 20:21:28 because iirc the kernel and kmods are built static Mar 06 20:21:29 [g2-lap]: whether it's best to build HEAD, or the almost-finished release Mar 06 20:21:55 philpem: yes, but some stuff you need to build the kernel depends on libc6 IIRC Mar 06 20:22:04 <[g2-lap]> well anyway I'm not gonna comment on that Mar 06 20:22:11 <[g2-lap]> what I will say is this Mar 06 20:22:17 philpem: we have never had a confirmed report of someone bricking to the point of jtag from an image built by the master makefile (even when the build was broken, you just can't boot and have to use upgrade mode to reflash with a good image) Mar 06 20:22:40 <[g2-lap]> using the make file build an openslug image Mar 06 20:23:07 [g2-lap]: yes, you can build both head and the stable releases with the makefile though :) Mar 06 20:23:10 rwhitby: so the only way to fry a slug to the point of needing jtag would be to replace redboot. Mar 06 20:23:21 philpem: yep Mar 06 20:23:38 that's our experience so far over the last almost two years. Mar 06 20:23:48 rwhitby: which there is precisely zero chance of me doing. ever. unless i get another slug and give it a ram upgrade that is :) Mar 06 20:24:11 philpem: [g2-lap] will try and convince you to replace redboot ;-) Mar 06 20:24:19 <[g2-lap]> heh Mar 06 20:24:21 hehe Mar 06 20:24:28 <[g2-lap]> no I'll try and convince him to buy a Loft Mar 06 20:24:30 <[g2-lap]> silly :) Mar 06 20:24:34 my mantra: "it works so don't mess with it" Mar 06 20:24:47 <[g2-lap]> Ok I'm gonna wrap this up Mar 06 20:24:48 g2-lap: loft? Mar 06 20:25:06 <[g2-lap]> http://www.giantshoulderinc.com Mar 06 20:25:13 <[g2-lap]> http://www.giantshoulderinc.com/hw-4533 Mar 06 20:25:23 <[g2-lap]> http://www.giantshoulderinc.com/ab3/case.jpg Mar 06 20:25:38 <[g2-lap]> I'm selling real hw I had made Mar 06 20:25:42 my mantra: if it works, mess with it Mar 06 20:25:48 ;) Mar 06 20:25:49 philpem: anyway, if you're gonna build a new image, I'd recommend you build the almost-finished release (So you'll end up with correct feeds and stuff) and you can provide useful testing info :) Mar 06 20:26:07 <[g2-lap]> philpem here's my point Mar 06 20:26:16 NAiL: howto on building from branch for OE? Mar 06 20:26:17 <[g2-lap]> building an openslug with just your adsl driver Mar 06 20:26:28 pepijn: with mastermakefile? Mar 06 20:26:32 <[g2-lap]> you can remove the usb hd issues by running from flash Mar 06 20:26:40 <[g2-lap]> it's be just the driver Mar 06 20:27:09 <[g2-lap]> furthermore, the old drivers which you are running had stability problems in the 8-20 hour range Mar 06 20:27:16 <[g2-lap]> I had lockups and so did others Mar 06 20:27:18 g2-lap: eciadsl is usermode Mar 06 20:27:38 it just needs n_hdlc and ppp-sync support in the kernel. Mar 06 20:27:50 <[g2-lap]> if you have serial or run a log of UDP there's an error message that gets printed Mar 06 20:27:52 s/ppp-sync/ppp-synctty/ Mar 06 20:27:52 philpem meant: it just needs n_hdlc and ppp-synctty support in the kernel. Mar 06 20:28:35 ugh, the targets for slugos aren't set up in the master makefile Mar 06 20:28:40 * NAiL thought they were Mar 06 20:29:02 NAiL: which targets? Mar 06 20:29:04 g2-lap: i think we have a suspect... 8 to 20 hours is exactly the timespan it's taking for wolf to fall flat. Mar 06 20:29:12 NAiL: yes, with MMM Mar 06 20:29:14 slugos-*-3.6-beta Mar 06 20:29:42 g2-lap: that sounds familiar to me too Mar 06 20:29:44 <[g2-lap]> philpem are you running dhcp on the ethernet ? Mar 06 20:29:49 g2-lap: yup Mar 06 20:29:59 i was blaming openntpd for stability in the 10 hour range Mar 06 20:30:04 <[g2-lap]> philpem bring it up statically and your probably will go away Mar 06 20:30:24 <[g2-lap]> change /etc/network/interfaces Mar 06 20:30:27 but.. i did first experience problems with debianslug Mar 06 20:30:30 2.6.15 Mar 06 20:30:43 be back in 2 mins. just rebooting the "athlon64 uberputer from hell" (aka cheetah) into fedora so i can have a go at building openslug on a reasonably fast box. Mar 06 20:30:48 rwhitby: since openslug/debianslug/ucslugc/slugos-ltu is now just slugos in svn, the old openslug-release target doesn't work. Mar 06 20:31:15 NAiL: feel free to update it then Mar 06 20:31:51 Making debianslug/ucslugc/slugos-ltu targets aren't a problem, but I don't know how to make an openslug target that acts correctly when building 2.7-beta/3.6-beta... Mar 06 20:31:58 since those two have different paths :( Mar 06 20:32:09 * NAiL produces laptop Mar 06 20:36:06 Hi, what is the fastest file sharing method with the nslu2 so far? Mar 06 20:36:26 netcat ;-) Mar 06 20:36:29 hehe Mar 06 20:36:32 ;) Mar 06 20:36:36 no shit eh? :P Mar 06 20:36:38 fastest IIRC is nfs Mar 06 20:36:52 * Daluv time to change I guess Mar 06 20:36:54 nfs-kernel I beleive Mar 06 20:36:58 <[g2-lap]> netcat and ttcp :) Mar 06 20:37:32 I hope I will be able to stream HD TV :( Mar 06 20:37:38 from my nslu2 that is. Mar 06 20:37:59 I think you should get a mb or two more with NFS actually Mar 06 20:38:11 sound good to me :) Mar 06 20:38:17 IIRC the difference was quite noticeably Mar 06 20:38:22 s/bly/ble Mar 06 20:38:56 I would love to be able to capture and strea m to my nslu2 from my HTPC Mar 06 20:39:37 right, i'm back Mar 06 20:40:51 philpem: the slugos-* targets aren't in the makefile quite yet Mar 06 20:41:27 i wasn't going to use slugos, i was going to use openslug Mar 06 20:41:45 to build the almost-finished 3.6-beta release, svn checkout svn://svn.berlios.de/openslug/releases/slugos-3.6-beta/ openslug Mar 06 20:41:49 openslug is based on slugos Mar 06 20:42:00 it's actually slugos-bag Mar 06 20:42:04 ok Mar 06 20:42:06 (with lots of packages) Mar 06 20:42:15 what about the LE slug ? Mar 06 20:42:28 LeSlug Mar 06 20:42:31 slugos-lag/debianslug Mar 06 20:43:58 then there's ucslugc/slugos-btu and slugos-ltu. Mar 06 20:44:50 so all i need to do is: 1) make openslug-image, 2) flash the slug with upslug, 3) boot with no drives connected, ssh in, connect the drive, mount it in /mnt and "cp /lib/modules/* /mnt/lib/modules -r", 4) reboot ? Mar 06 20:45:58 You should also do an ipkg upgrade Mar 06 20:46:01 offline one Mar 06 20:46:07 ipkg -o IIRC Mar 06 20:46:15 philpem: /releases/slugos-3.6-beta/ repvik 2 days Add eciadsl and HDLC support Mar 06 20:46:30 something like ipkg -o /tmp update && ipkg -o /tmp upgrade Mar 06 20:46:35 you don't need upslug Mar 06 20:46:39 pepijn: he knows I committed that ;) Mar 06 20:46:45 ok, now i'm lost. Mar 06 20:47:04 * philpem reads the "openslug-to-opendebianslug howto" on the wiki again... Mar 06 20:47:23 philpem: hang on, and I'll try to sort out the upgrading routine again ;) Mar 06 20:47:32 i would upgrade as follows: 1) touch /.recovery 2) reboot 3) turnup ram 4) reboot 5) reflash -i ... Mar 06 20:47:51 pepijn: yes, that's the way Mar 06 20:48:06 I am trying to get setup quilt to do patch management of nslu2 patches, but I can't figure out how you initialize quilt to use the existing series patches/ixp400/series files Mar 06 20:48:07 <[g2-lap]> bbl Mar 06 20:48:51 kaenat: quilt push -a ? Mar 06 20:49:05 after (5) you'll boot into flash and you have to fixup your USB drive to work with the newer kernel (cp /lib/modules) Mar 06 20:49:14 then turnup disk .. Mar 06 20:49:21 pepijn: sounds easy enough Mar 06 20:49:31 pepijn: one more step though Mar 06 20:49:35 before turnup disk Mar 06 20:49:35 if all fails, you need to fallback on upslug Mar 06 20:49:49 just waiting for subversion to get its act together. Mar 06 20:50:15 edit the feed addresses in /etc/ipkg (on the hdd) to point to the 3.6-beta feed, then do the ipkg -o /mnt update && ipkg -o /mnt upgrade Mar 06 20:50:21 *then* turnup disk Mar 06 20:50:31 but ipkg isn't installed on debian... Mar 06 20:50:43 philpem: this is from the flash Mar 06 20:50:51 NAiL: also for OpenDebianSlug ? Mar 06 20:50:52 (or, actually, from ram) Mar 06 20:51:00 oh, dang, no Mar 06 20:51:04 Just ignore me Mar 06 20:51:09 ;) Mar 06 20:51:11 :) Mar 06 20:51:39 as far as i can see the only thing that is effected by upgrading (after turnup disk) is the kernel Mar 06 20:51:53 which means you need to fix /lib/modules on the USB disk Mar 06 20:52:12 which means it would be a Good Idea (tm) to make a backup Mar 06 20:52:13 nothing else should need changing Mar 06 20:53:46 redboot loads kernel, kernel loads ramdisk, ramdisk mounts /dev/sdax, calls pivot_root and runs init from thereon Mar 06 20:53:54 thats what i thought. Mar 06 20:54:25 modules (except for basic interfacing to USB) are loaded from disk Mar 06 20:54:26 redboot -(bootload)-> kernel -(load initrd)-> initrd -(mount & pivot)-> hdd Mar 06 20:54:51 networking is initialized from disk, therefore you need the IXP modules on your hdd Mar 06 20:55:10 (correct me if i am wrong) Mar 06 20:55:12 woohoo, svn checkout complete :) Mar 06 20:56:03 so now i need to do a "make slugos-bag"? Mar 06 20:56:10 NAiL: quilt doesn't know about the series file yet. I get "No patches applied" I thought the quilt setup command was supposed to setup the quilt environment from a series file. Mar 06 20:56:40 hmm Mar 06 20:56:49 the patches are in the patches/ dir? Mar 06 20:56:54 I've never used setup ;) Mar 06 20:56:56 03orepvik * 10unslung/ (3 files in 2 dirs): Add ds101j target Mar 06 20:57:08 is there a way to find out if your disk can be put in "sleep" mode when it is not active? Mar 06 20:57:13 patches/ixp400/* is where the series file and the patches Mar 06 20:57:22 are Mar 06 20:57:33 dunno if quilt looks in the subdirs of patches... Mar 06 20:57:55 Daluv: Trying sdparm, I guess. Few usb-ide chipsets support it though Mar 06 20:58:20 Daluv: There's a lot of discussions about that on the mailinglist... You might wanna look there. Mar 06 20:58:35 ok do you know the name of the features? Mar 06 20:59:00 nope Mar 06 21:00:16 Nail:thanx. Mar 06 21:00:21 np Mar 06 21:00:44 hdparm or sdparm? Mar 06 21:01:15 sdparm Mar 06 21:01:20 03orepvik * 10unslung/Makefile: Oops, reset OPTWARE_TARGET back to nslu2 Mar 06 21:01:24 hdparm won't do much good with /dev/sdX Mar 06 21:01:31 uhm uhm Mar 06 21:01:45 hdparm - get/set hard disk parameters - version v6.5 Mar 06 21:01:51 doestn say much uhm Mar 06 21:02:03 yes. It sets/gets IDE-disk paramaters mostly. Mar 06 21:02:16 it does works :/ Mar 06 21:02:22 dparm /dev/sda Mar 06 21:02:32 s/works/read Mar 06 21:02:52 yeah, you can do speed testing and such without much problems Mar 06 21:03:03 but adjusting the disk settings will probably fail Mar 06 21:03:07 I guess it is time to build sdparm Mar 06 21:03:28 okay, i've got the IXP420accesslibrary and NPE zips, where do i put them? Mar 06 21:03:36 in the downloads dir Mar 06 21:03:48 the .bb says "openembedded downloads directory", but i don't see one in openembedded/ Mar 06 21:04:04 no, it's in the same dir as the openembedded/ dir Mar 06 21:04:47 not fwict. i've got bitbake[dir], conf[dir], Makefile, openembedded[dir], README, scripts[dir] and setup-env. Mar 06 21:05:19 make one ;) Mar 06 21:05:23 in that dir Mar 06 21:05:23 done :) Mar 06 21:05:39 You'll need to make the md5sums too Mar 06 21:07:31 done. Mar 06 21:07:42 nail it is call "SpinDown" I guess Mar 06 21:07:48 then you just do make openslug-image :) Mar 06 21:07:56 Daluv: ah, yeah Mar 06 21:08:04 I wonder if "Cypress Semiconductor Corp USB-2.0 IDE Adapter" is compliant. Mar 06 21:08:12 "handling bitbake files" Mar 06 21:08:25 Daluv: I don't think so. Mar 06 21:08:36 Daluv: depends which chip and what level of compliance you're talking about Mar 06 21:08:59 Daluv: I've got one of those, and I haven't succeeded getting it to spin down my chainsaw Mar 06 21:09:11 Nail: ok Mar 06 21:09:27 my nexstar2 has a cypress chip in it. "EZ-USB AT2LP" iirc. Mar 06 21:09:43 yeah this is a nextstar3 Mar 06 21:10:08 Vendor 0x04b4, Product 0x6830, rev 0.01. Mar 06 21:10:27 "Cypress Semiconductor" / "USB2.0 Storage Device" Mar 06 21:10:44 same here: ID 04b4:6830 Mar 06 21:10:45 philpem: same here, IcyBox with same id Mar 06 21:11:00 okay now this is interesting.. :) Mar 06 21:11:25 three different people, three different ide-usb boxes, all the same chipset. Mar 06 21:11:34 no surprise here Mar 06 21:11:41 that said, the cypress ones nail prolific's "offerings" to the floor. Mar 06 21:11:43 I mean there is not a lot of that chipset floating around. Mar 06 21:12:06 prolific, cypress, genesys.. Mar 06 21:12:07 on the grounds that cypress's chips actually work, while prolific's kinda don't. Mar 06 21:12:24 I had some problems with genesys and unslung 55 ... Mar 06 21:12:34 doesnt work flawlessly.. Mar 06 21:12:42 but genesys support 1000MB Mar 06 21:12:48 uh Mar 06 21:12:51 but genesys support 1000G Mar 06 21:13:02 who the hell would want to put 1000G of storage on a slug? Mar 06 21:13:13 dunno Mar 06 21:13:24 wait.. they'd be limited to 1 drive anyway, and the biggest hdd i've seen was (iirc) a 320. Mar 06 21:13:44 biggest hdd 320gb? 500gb ;) Mar 06 21:13:58 my, they're getting big now. Mar 06 21:14:07 note i said "biggest one i've *seen*". Mar 06 21:14:28 1000G is a to loose ;) Mar 06 21:14:35 is a lot to loose Mar 06 21:14:35 i've heard of stupidly big hds that exist in research labs, but the biggest one i've held in my hands was a maxtor 320gig. Mar 06 21:14:42 a maxtor 320gig that was doa :P Mar 06 21:15:18 not mine, belonged to a friend. he saw red when it came up with "SMART DISK FAILURE ALERT" on its first run. Mar 06 21:15:38 you can do multiple drives on unslung 6.x and openslug ... Mar 06 21:15:42 * NAiL mumbles bathtub curve Mar 06 21:16:36 HD TV fileserver uhm. Mar 06 21:16:49 * NAiL wants another 2.5" enclosure Mar 06 21:16:49 2x 1000G uhm. Mar 06 21:16:50 i used to recommend maxtors but their failure rates have been pretty stupid. one of my friends in the it trade quoted a 18% doa rate but he said that was "mostly from the dm8+ drives and we've stopped buying them anyway". Mar 06 21:17:42 I bought 8 seagate disks a bit more than a year ago. AFAIK, only two of them are alive today Mar 06 21:18:20 hmm, i've got a 40gig momentus in my laptop that seems fine. better than the fu-shit-su (sorry, fujitsu) i had in it before. Mar 06 21:18:21 eww Mar 06 21:18:28 talk about reliability... Mar 06 21:18:30 NAiL: you can say the same thing about your nslu2 hardware too, can't you ;-) Mar 06 21:18:46 I have WD for no particular reason Mar 06 21:18:55 got a mix here. Mar 06 21:19:03 rwhitby: Nah, only one out of three has failed ;) Mar 06 21:19:08 2.6.12.2-r17.4 uhm. Mar 06 21:19:27 maxtor 80gig d740x in cheetah, 40gig seagate momentus in the laptop, 4gb seagate u4 in the backup server, 6gb seagate in the riscpc. Mar 06 21:19:42 wow, didn't know i had that many seagates... Mar 06 21:20:15 ime, the biggest killer of hard drives is poor cooling. shove a d740x in an uncooled usb case and it'll fry. Mar 06 21:21:21 i've got a 2-slot gap between the 20gig ibm 35gxp and the 80gig maxtor. never had a problem. the golden rules are - 1) lots of air space, 2) a fan to cool the the hdd is a good idea Mar 06 21:21:24 I wonder how my nexstar3 + WD will do in a rather small closed area :/ Mar 06 21:22:09 hmm, 14940 hours on this drive?! Mar 06 21:22:54 ah, about 2 years. sounds about right. Mar 06 21:23:21 * NAiL has given up returning dead harddrives. I just buy new ones instead. Mar 06 21:23:51 start_stop_count: 2245, reallocated_sector_count: 0, power_cycle_count: 2109 Mar 06 21:24:01 temperature_celsius: 47 Mar 06 21:26:26 the big problem with maxtors is they run hot. also, the fdb bearings suffer badly from jitter - the current diamondmax drives tend to suffer from "firmware fudgeing syndrome" Mar 06 21:26:49 i.e. they think the firmware track is unreadable, so they read it then rewrite it.. while the motor is still not quite up to speed. Mar 06 21:27:21 eventually it ends up trashing the firmware track and the drive gets bricked. Mar 06 21:27:31 that's neat Mar 06 21:27:49 and only maxtors suffer from it, based on experience :) Mar 06 21:27:50 nice feature I must say ;) Mar 06 21:28:09 there's a reason i ordered a drive with a ball-bearing motor :) Mar 06 21:30:06 ideally there should be two or three firmware tracks. boot, then try and re-read it an hour or so later if there were any errors with the firmware. if it reads back bad, do a majority vote, fix the bad copy and rewrite it. if one of them is bad on boot, use the other two. Mar 06 21:30:18 that kind of thinking seems to escape most hdd firmware developers though :P Mar 06 21:31:22 but by far the worst design mistake i've ever seen is the ibm 75gxp's tendency to gouge pits in the lubricant. head rides over the pit and it causes a headcrash. Mar 06 21:31:42 the firmware patch makes the drive move the heads around a bit if it thinks they've been in one place too long. Mar 06 21:37:42 03oleon * 10unslung/sources/python/Makefile.pre.in.patch: python: change link search path to cuurent dir then staging. - This prevents to link python against shared lib from previous staging Mar 06 21:52:22 um, "build no workie"... Mar 06 21:52:49 it gets to glibc-.3.5+cvs20050627 then barfs at do_unpack Mar 06 21:53:17 gzip: stdin: decompression OK, trailing garbage ignored / tar: Child returned status 2 Mar 06 22:05:43 anyone got any ideas? Mar 06 22:08:26 ports_sources.redhat.com__20050627.tar.gz seems to have some garbage tacked on to the end Mar 06 22:09:32 try deleting the tarball/md5 and starting over Mar 06 22:22:59 It wont work Mar 06 22:23:39 I went through that excercise around 6 times before deciding the file really was busted. Mar 06 22:23:53 I got a better one from sources.nslu2-linux.org Mar 06 22:25:34 whereabouts on sources.nslu2-linux.org? Mar 06 22:25:50 i just gunzipped it and re-gzipped it to get rid of the "garbage data" Mar 06 22:36:29 ok that works too. :-) Mar 06 22:37:54 http://sources.nslu2-linux.org/sources/ports_sources.redhat.com__20050627.tar.gz Mar 06 23:00:15 03oleon * 10unslung/make/python.mk: python: bump ipk version Mar 06 23:19:37 03bzhou * 10unslung/make/memcached.mk: added memcached Mar 06 23:22:48 03bzhou * 10unslung/sources/lighttpd/configure.in.patch: need to patch configure.in to get PCRE going Mar 06 23:23:45 03bzhou * 10unslung/sources/lighttpd/rc.lighttpd: explicitly specify lighttpd.conf Mar 06 23:28:03 03bzhou * 10unslung/ (3 files in 2 dirs): added a default lighttpd.conf points to /opt/share/www/lighttpd/ Mar 06 23:29:04 03bzhou * 10unslung/Makefile: lighttpd & memcached ready for testing Mar 06 23:36:49 03bzhou * 10unslung/ (3 files in 2 dirs): added postinst and prerm, bumped IPK_VERSION Mar 06 23:38:00 03bzhou * 10unslung/sources/cherokee/rc.cherokee: some cosmetic change Mar 07 01:39:32 ok, now i've got a compile error on ncurses Mar 07 01:40:19 "TaskFailed event exception" Mar 07 07:13:25 03oleon * 10unslung/make/py-cheetah.mk: add .mk depend Mar 07 07:24:31 Hi Mar 07 08:19:01 eno, I am thinkg of adding .mk to all py-* packages. But I am unshure if it is beter to be dependent on python.mk or py-*.mk Mar 07 08:19:57 Or both Mar 07 08:21:43 or change scripts/optware-autoclean.pl Mar 07 08:51:48 Bumping ipk version does not force package reconfigure and thus this is no solution when package is not in Packages. Autoclean helps only for packages already in Packages. :( Mar 07 10:05:26 03oleon * 10unslung/Makefile: Mar 07 10:05:26 Demote all python-dependent packages until python is in packages. - Current Mar 07 10:05:26 packaging design prevents propagating rebuild to dependent packages. Mar 07 10:05:26 optware-autoclean.pl needs to updated to check timestamps of dependent package Mar 07 10:05:26