**** BEGIN LOGGING AT Sun Dec 17 02:59:59 2006 **** BEGIN LOGGING AT Sun Dec 17 05:46:20 2006 **** BEGIN LOGGING AT Sun Dec 17 05:47:40 2006 Dec 17 07:15:00 Does a standard slug handle read/write to NTFS... if so, does it mount the ntfs drive some how through samba ? Dec 17 08:08:46 sime: yes and yes Dec 17 08:09:32 although it is reported to not be 100% stable Dec 17 08:12:43 rwhitby, ic, then do the third party firmwares do similar ? Dec 17 08:45:49 sime: Unslung uses the vendor NTFS kernel module directly, so it has the same stability. Dec 17 08:46:10 SlugOS and Debian have standard Linux 2.6.x kernel NTFS support. Dec 17 08:48:41 i.e. r/o (with very limited r/w) Dec 17 11:48:14 03rwhitby * r627 10kernel/trunk/patches/2.6.19/05-mtdpart-redboot-partition-truncate.patch: Added the description and signed off - this one has been accepted upstream Dec 17 11:49:02 03rwhitby * r628 10kernel/trunk/ (8 files in 2 dirs): Additional configuration files for apex on different devices Dec 17 11:59:46 ~seen philpem Dec 17 12:00:03 philpem was last seen on IRC in channel #classiccmp, 72d 14h 8m 8s ago, saying: 'nyahaha, behold vmware and virtual network interfaces, the malware reverser's friends!'. Dec 17 12:04:00 hello Dec 17 12:04:25 morning Dec 17 12:08:40 does someone know how to use the nslu2's beeper ? Dec 17 12:09:01 I found a module, ixp4xx-beeper, loaded it, but don't know how to use it Dec 17 12:09:03 what firmware are you running? Dec 17 12:09:08 Try "beep" Dec 17 12:09:09 debian Dec 17 12:09:23 semial@anaelle:~$ beep Dec 17 12:09:23 -bash: beep: command not found Dec 17 12:09:36 apt-cache search nslu2 Dec 17 12:09:58 I think there's a package called nslu2-utils Dec 17 12:11:19 ok Dec 17 12:12:07 I installed the beep package Dec 17 12:12:17 when I call beep, the nslu2 emits a strange sound Dec 17 12:12:30 like "shkrr" Dec 17 12:13:18 mh, nslu2-utils is already installed Dec 17 12:14:01 beep --help Dec 17 12:14:11 I think it should be possible to adjust freq Dec 17 12:14:21 I see Dec 17 12:14:52 morning likewise :) Dec 17 12:15:03 morning :-) Dec 17 12:16:17 sudo beep -f 4 -l 200 Dec 17 12:16:19 makes a better sound Dec 17 12:55:43 03blaster8 * r629 10kernel/trunk/ (3 files in 2 dirs): Update STATUS file with latest upstream submission info Dec 17 14:57:46 blaster8: tnx for fixing http://bugs.openembedded.org/show_bug.cgi?id=1675, can I close it? Dec 17 14:58:28 didn't spot that Dec 17 14:59:02 yeah, close it Dec 17 15:06:27 done Dec 17 15:06:56 thanks Dec 17 15:07:37 I'm not entirely sure we should be shipping reiserfs, to be honest Dec 17 15:07:37 not sufficiently reliable IMO, even on x86 Dec 17 15:07:59 still, it's there and now it's fixed Dec 17 15:30:14 blaster8: I *used* to use it on my x86 boxen as it was fast for small files (read: source code) but I do not trust it with regards to regression over several versions. Dec 17 15:30:27 same here Dec 17 15:31:06 migrated to XFS on the one partition which *needs* to be fast, and used ext3 on the others (UPS-backed btw) Dec 17 15:32:12 blaster8: Yup, I like XFS in terms of design, but ext3 is a very nice default. ext4 should be very good in terms of design as well. Dec 17 15:32:42 XFS is only possible if you have a UPS, really Dec 17 15:33:03 any interruption, be it a crash or a power 'event' can leave you with serious data loss Dec 17 15:33:51 but I got a ridiculous second hand 1000VA UPS with new batteries to power all of my kit Dec 17 15:34:28 blaster8: I would be happy with a 5 minute UPS that just allows the systems to shutdown properly. Only have 1 powerfail per year here so far. Dec 17 15:34:57 blaster8: 1000 VA for how long? Dec 17 15:35:09 over half an hour Dec 17 15:35:30 we get too many failures here, about 6 over 30 secs every year Dec 17 15:35:56 blaster8: half an hour, that allows you to email all your online friends to tell you are going offline. Dec 17 15:36:06 :-) Dec 17 15:36:21 with the big computers off, it powers my network infrastructure for about 4 hours Dec 17 15:36:34 so I can still use wireless and ADSL ;) Dec 17 16:07:50 hey. could anyone suggest a method to backup my root filesystem (Debian/NSLU2) to another partition or to a file? i tried cpio and dd so far, and i was never able to restore a functionnal system. Dec 17 16:12:15 join #openwrt Dec 17 16:17:50 <[g2]> RobNC I think you want a / on that :) Dec 17 16:22:29 yeah sorry my fault... touchpads stink Dec 17 16:26:22 hi, made an optware ipk for fuse for the ds101g, to make it perfect I have to stage the kernel sources from my kernel modules ipk Dec 17 16:26:44 what do you think, simply copy the kernel sources to staging/src/linux? Dec 17 16:27:06 what's this on the wiki about UART2 Tx pins not "brought out"... is this really true? Dec 17 16:28:35 yes Dec 17 16:29:14 so the pads don't go to vias on the bottom of the 420? Dec 17 16:29:34 no Dec 17 16:30:05 or at least, not that anyone has ever found them Dec 17 16:30:07 But Rx works, so it could potentially be used for things like, UPS daemons, etc. Dec 17 16:30:25 I have an x-ray machine at work, I might try to confirm it. Dec 17 16:30:28 well, for a dumb ups Dec 17 16:30:39 yeah I dont have $$ for smart ones. Dec 17 16:30:40 RobN1: makes sense, if you can Dec 17 16:30:54 RobN1: neither do I, until you buy them second hand Dec 17 16:31:59 oops sorry my nick should be RobNC... true. Some ppl get rid of them b/c the batteries are bad but the batteries are pretty cheap compared to the electronics Dec 17 16:33:22 RobN1: if you want more serial ports, USB-Serial seems to be the best way to go Dec 17 16:34:25 yeah it's just more peripherals and larger size. I see that there are three more ports available. It is kinda funny how only one port supports thumbdrives but actually all ports should work. Linksys didn't connect the power on the other port. In fact, they have a ferrite bead instead of a fuse, on the powered one. Dec 17 16:34:53 strange Dec 17 16:41:42 sorry - lost connection (WRT issues) Dec 17 16:46:44 FreeBSD Wireless Router Dec 17 16:46:58 Wireless network issues a thing of the past :) Dec 17 16:47:44 75 days uptime flawless without a reboot until I decided to update the system Dec 17 16:49:29 heh I am working on that but have to first hack the heck out of a BHR-HP-54G. I've added JTAG and serial port, but it won't boot just yet b/c config registers probably need updating. I replaced 16MB with 128MB. Dec 17 16:49:46 (my current router is WRT54G v1.0) Dec 17 16:50:50 WRAP - 266MHz Geode x86 running pfSense Dec 17 16:51:14 yeah I thought of that but that's almost like having an ONT outside your house. Dec 17 16:51:53 ? Dec 17 16:52:06 WPA2/802.11g Dec 17 16:52:27 um, I mean like one of those optical network terminations (i.e., Verizon FiOS) Dec 17 16:52:42 UK, don't know about those things Dec 17 16:54:55 ah okay sorry about that confusion. Those BCM5352 should be good enough, 200MHz (o/c to 250MHz I hear), but flash and RAM are too small. Dec 17 16:55:11 I mean from factory, much like nslu2 :-) Dec 17 16:55:52 it's good, but too much work to replace for me Dec 17 16:56:05 especially then having to run Linux/MIPS on them... Dec 17 16:56:49 ah I have access to the necessary tools. Got a fatslug now, gotta replace redboot with APEX but I think APEX doesn't have recovery modes, like RedBoot, right? Dec 17 16:57:58 I wouldn't attempt it w/o microscope and SMT soldering stations. Dec 17 17:00:30 maybe it Dec 17 17:01:30 maybe it's better to just rebuild Redboot. We use uboot at work, and that seems to be pretty flexible (mostly environment options for config). Dec 17 17:04:22 APEX does have a recovery mode Dec 17 17:04:32 but no networking on ixp4xx Dec 17 17:04:39 so you need to use serial Dec 17 17:06:35 ah okay so it can support xmodem for transfer mode, like apex? Dec 17 17:06:39 I mean like redboot Dec 17 17:06:43 indeed Dec 17 17:06:50 the website is offline for some reason atm Dec 17 17:07:06 it's a damn site easier to build than a new redboot as well Dec 17 17:07:50 Thing I didn't like about redboot. You can receive a file, etc. to the NSLU2 but no way to (over serial) to send a file? Dec 17 17:08:18 I mean, i.e., send the contents of memory location A to B via xmodem. I guess you have to use JTAG for that, and that's a bit messy for me (yet). Dec 17 17:08:38 JTAG is best for full flash backup Dec 17 17:09:01 I wouldn't think that trying to build a new Redboot for the NSLU2 is a good idea Dec 17 17:09:06 Got the nslu2 with JTAG hard wired up, but haven't tested openwince JTAG yet. I did confirm the interface for the BHR-HP-WRT (detects bcm5352 and flash, and hairydairymaid works) Dec 17 17:10:01 yeah it can be risky. I haven't done a cross-compiler before but I do have linux running (albeit Fedora) and compiler/libraries installed. Dec 17 17:10:31 Wonder if it'd be better to install debian for this purpose. Or some ppl say Knoppix (also debian based) Dec 17 17:11:12 RobNC: I would use Debian Etch Dec 17 17:11:26 I have a dedicated virtualised build machine using OpenVZ Dec 17 17:12:05 Building a cross-compiler is trivial with our build system Dec 17 17:12:49 etch - is that an older version of debian (not sarge)? Dec 17 17:13:06 no Dec 17 17:13:07 03gda * r4722 10optware/trunk/make/ds101g-kernel-modules.mk: ds101g-kernel-modules: added stage target Dec 17 17:13:16 it's the testing version Dec 17 17:14:17 03blaster8 * r630 10kernel/trunk/patches/2.6.20/KERNEL: Update to 2.6.20-rc1-git4 Dec 17 17:14:22 ah gotcha... it seems the debian testing is probably more stable than some of these fedora cores. I'd have to re-learn all the config stuff though (i.e., no /etc/rc.d and possibly problems with alien and rpms) Dec 17 17:15:20 03gda * r4723 10optware/trunk/ (5 files in 3 dirs): ds101g-kernel-module-fuse: fuse module with utils for ds101g Dec 17 17:15:32 03blaster8 * r631 10kernel/trunk/patches/2.6.20/ (7 files): Preliminary cleanup of ixp4xx for 2.6.20, workstruct api fixups pending Dec 17 17:15:32 I've got less problems with debian unstable than fedora core :) Dec 17 17:16:34 mial: that is what I am thinking... I had some issues with FC5 - everytime logging into desktop, the whole machine would hang. Video card was Radeon 7000, and using Dell SC400 (stable platform!). Dec 17 17:17:16 are there no DVD installs for etch? Dec 17 17:17:33 hmm, not sure Dec 17 17:17:33 or do you first install sarge then upgrade to etch Dec 17 17:17:41 I'd use the debian beta installer Dec 17 17:18:04 I think it's better to do a netinstall and install all the stuff you need from internet Dec 17 17:18:17 http://www.debian.org/devel/debian-installer/ Dec 17 17:18:41 http://cdimage.debian.org/cdimage/etch_di_rc1/i386/iso-cd/debian-testing-i386-netinst.iso Dec 17 17:18:45 (btw, sorry for my english, I'm only 14 and I'm from switzerland) Dec 17 17:19:36 hey mial: I wish my schweizerdeutsch was as good as your english Dec 17 17:20:11 I'm from the french part of switzerland :) Dec 17 17:20:35 hehe, oops Dec 17 17:21:14 mial: heh no probleml thanks for tip and thanks blaster8 Dec 17 17:46:35 blaster8 so with that CD net install, is it command-line based or some kind of easy to traverse GUI for installs? I.e., during the install, I hate not knowing whatsome of the packages do Dec 17 17:46:49 there's a GUI now Dec 17 17:47:10 Are you installing this on a primary work machine? Dec 17 17:47:19 yeah, there is a GTK GUI now Dec 17 17:47:31 as replacement for the legacy ncurses GUI Dec 17 17:49:35 personally I would set up a headless machine with a minimal command line debian and ssh in from your favourite distro/OS Dec 17 17:50:45 mial: thanks, will have to try that in etch. Downloading CD image now but very slowly 500Bytes/sec. Dec 17 17:50:56 blaster8: yes that is my plan, and VNC in as needed Dec 17 17:50:56 aah Dec 17 17:50:58 mirror time Dec 17 17:51:21 blaster8 yeah true... gotta find a faster mirror Dec 17 17:53:33 mial: netinst CD has the GTK GUI install? Dec 17 17:55:56 probably not Dec 17 17:56:31 that stinks... it is a pain to use keyboard only to select things, esp with as many packages in debian Dec 17 17:56:51 but the ncurses GUI is not as bad Dec 17 17:57:12 and you can install all the packages you need later Dec 17 17:57:25 hi. after restoring from a backup (using dd), my slug seems to boot correctly, except that eth0 does not come up. it seems to be detected (ixp4xx_mac driver 0.2.1: eth0 on NPE-B with PHY[1] initialized) but it won't come up and i can't access the slug. i'm clueless. if anyone has an idea... (Debian/NSLU2) Dec 17 17:58:24 RobNC: install the minimum Dec 17 17:58:30 then add things later Dec 17 18:03:09 blaster8 I see your point but lotta times you don't know what you need until after you don't have it. Then it's a ratrace (with at least Fedora), find this RPM, oops, deps fail, need another package, yada yada Dec 17 18:04:07 there is a lot of packages in the officials debian repositories Dec 17 18:04:10 I mean you may end up spending hours meeting all the dependencies, finding packages, etc. before you can install the package you want. This is a good thing as it adds stability but it can be a headache when you don't know where to get this dependency, etc. Dec 17 18:04:29 I've heard there is about 15k packages in it Dec 17 18:05:50 wow... that's actually a good thing but then it can be hard to decide which package to choose (that do similar things). Dec 17 18:07:35 $ aptitude install[tabulation][tabulation gave that: Dec 17 18:08:00 Then I didn't even mention the problems of installing a package and then the desktop manager expects another to be installed, or worse that one program doesn't know about another b/c they're not the default apps installed. I dunno - perhaps too many previous problems with redhat and fedora installs! :-) Dec 17 18:08:00 display all 23228 possibilities ? (y or n) Dec 17 18:08:27 cool - so can apt-get automatically get the dependencies as needed, much like yum? Dec 17 18:08:27 yes Dec 17 18:08:38 ah that makes it significantly easier. So just one repository then? Dec 17 18:08:46 apt is probably the most powerful package manager Dec 17 18:08:52 yes Dec 17 18:09:55 yeah I thought so, after my limited use of it with nslu2 (running debian older version - needs to be upgraded) Dec 17 18:10:20 and if you install a package, who need a ton of dependances, and later you want to remove this package and all the deps who came with, apt remember which packages were installed as dependances, and remove them Dec 17 18:10:51 s/dependances/dependencies/ Dec 17 18:10:52 mial meant: and if you install a package, who need a ton of dependencies, and later you want to remove this package and all the deps who came with, apt remember which packages were installed as dependances, and remove them Dec 17 18:10:52 oh now that's nice. no stale files, etc. lying around? Dec 17 18:12:04 yes, exactly Dec 17 18:12:10 * mial loves apt Dec 17 18:13:00 gotcha mial. maybe it's better to just get the 2 DVD installs and install that versus the minimal CD install for etch RC1 Dec 17 18:13:31 it's better to use aptitude install instead of apt-get, because it better handles the dependencies as apt-get Dec 17 18:14:04 RobNC: do you think you'll use all the packages there is in the dvd ? Dec 17 18:14:19 making a netinstall will save a lot of bandwith Dec 17 18:15:19 mial: yeah I see your point in regards to aptitude. True, I probably won't need all those language stuff, Tek, none of the X-devel, KDE/Gnome devel, so indeed netinstall would be the fastest. Dec 17 18:15:40 yes Dec 17 18:16:20 gotta find a faster source for netinstall blaster8 pointed me to Dec 17 18:16:43 how fast are you downloading? I got 100KB/s from there Dec 17 18:17:06 http://ftp.acc.umu.se/cdimage/etch_di_rc1/i386/iso-cd/debian-testing-i386-netinst.iso Dec 17 18:17:39 I'm in USA so that might be slow for me Dec 17 18:18:20 gatech prolly fastest for me - at mirror site now Dec 17 18:19:59 seems like not many mirrors have etch only have sarge (3.1r4) Dec 17 18:20:04 mm Dec 17 18:22:42 http://ftp.ticklers.org/cdimage.debian.org-cdimage/etch_di_rc1/i386/iso-cd/debian-testing-i386-netinst.iso Dec 17 18:23:47 yeah significantly faster... 300kB/sec Dec 17 18:23:48 thanks! Dec 17 18:28:29 blaster8 you know where I can get an MD5 of that RC1? Dec 17 18:28:43 http://ftp.ticklers.org/cdimage.debian.org-cdimage/etch_di_rc1/i386/iso-cd/MD5SUMS Dec 17 18:29:06 thanks, it's not on the website for debian for some reason (not even a link). Dec 17 18:29:42 it's in the directory with the files Dec 17 18:31:54 yeah figured as much - after I saw that I did a "d'oh" Dec 17 18:43:05 better to ask than to spend ages looking :) Dec 17 18:43:32 yeah this is true! Dec 17 18:46:15 * RobNC is burning CD, etc. Dec 17 18:46:42 what computer are you putting this on? the one you are typing on now? Dec 17 18:49:59 blaster8: no, have separate PC for that. Dec 17 18:50:10 good, just checking Dec 17 18:50:19 heh!! Dec 17 18:50:24 debian is not necessarily what you want for a primary desktop PC Dec 17 18:50:36 shh Dec 17 18:50:51 Debian etch runs very nicely as a my principal desktop. Dec 17 18:51:16 hehe Dec 17 18:51:50 blaster8 mial: I've used icewm over VNC over ssh2 - that works pretty well Dec 17 18:52:07 not surprised, it's a good idea Dec 17 18:52:45 but KDE over VNC is a DOG Dec 17 18:53:39 gotta back up my /etc /home /root directories from fedora (re-use SSH keys) Dec 17 18:55:25 I also run boinc - presumably that's supported in debian. Do you guys use the nslu2 as-is w/o mods? Dec 17 18:56:28 serial/overclock is a must Dec 17 18:59:00 indeed - the easiest mod is the de-underclock. Dec 17 18:59:30 I am really impressed with Intel's documentation available. JUST the opposite for Broadcom though Dec 17 18:59:58 too bad their firmware redistribution policy is really nasty Dec 17 19:00:54 yeah I can't understand their paranoia but then again Intel probably has more $$ to go after those copycats in China and India Dec 17 19:01:24 not that this helps even the slightest bit to hinder copycats... Dec 17 19:01:48 it's not like product pirates pay attention to EULAs ... Dec 17 19:02:36 Intel's documentation is ok, shame about the software AND the license Dec 17 19:02:41 yeah but Intel has sites in china and india so they could say they'll pull their sites (bluff!) Dec 17 19:03:54 yeah I hear that IX420 ethernet driver opensource will be avail soon? Dec 17 19:04:38 well, there is a driver Dec 17 19:05:16 03bzhou * r4724 10optware/trunk/make/py-nose.mk: py-nose: added py25 subpackage Dec 17 19:06:50 I read somewhere that the OS driver is more stable than the Intel driver? Dec 17 19:07:34 it is indeed Dec 17 19:07:34 but it still has one major unresolved issue with heavy uploads Dec 17 19:07:55 and it some code location changes are required before it can go upstream Dec 17 19:08:26 I ran into a problem when I did rsync -rsac over NFS mount that the NSLU2 would hang network access. Unplugging and reinserting ethernet cable, the network was restored. Verified via serial console. Dec 17 19:08:37 post details to mlj28@cam.ac.uk and I will forward to the developer Dec 17 19:08:46 rsync run this way does a CRC integrity check over the network, which would run incredibly slow. Dec 17 19:09:00 ok, no problem... I didn't know if it was a known issue or not. Dec 17 19:09:27 This was with 170+ GB of data so it may be related to already known issues. Dec 17 19:09:28 always assume it's not a known issue, especially with a driver this new and hardware this bloody complicated Dec 17 19:09:44 there are only 2 known issues at all Dec 17 19:09:47 and even if we k Dec 17 19:09:50 oh I'm not using the new driver, using the one from OpenDebian Dec 17 19:09:59 ? Dec 17 19:10:03 the Intel driver? Dec 17 19:10:13 with kernel 2.6.17? Dec 17 19:10:25 yeah intel - gimme a sec to find details Dec 17 19:10:34 aah, in that case it's not a big deal Dec 17 19:10:45 the intel driver is broken, I thought you meant the OSS driver Dec 17 19:10:53 http://www.cyrius.com/debian/nslu2/ Dec 17 19:11:04 Debian 2.6.18 contains the OSS driver Dec 17 19:12:18 ah no I'm not using OSS driver but I will soon, as I am using the older 3.1 debian with 2.6.16 (can't recall but think its BE) Dec 17 19:12:45 ok Dec 17 19:13:22 "uname -a" shows Linux NSLU2 2.6.16 #1 PREEMPT Thu Jun 8 23:38:13 PDT 2006 armv5tel GNU/Linux Dec 17 19:13:37 little-endian Dec 17 19:13:59 ah ok... and the APEX is only LE? Dec 17 19:14:28 APEX? Dec 17 19:14:57 yeah I thought that the bootloader had to boot up in one mode and then the kernel, etc. had to be the same mode? Dec 17 19:15:03 APEX is the bootloader we use to load a larger kernel, it is open source and can be compiled as big or little endian Dec 17 19:15:16 your current install is LE, as shown by armv5tel Dec 17 19:15:36 ah ok, yeah it was a netinstall Dec 17 19:16:11 I'm hesitant to upgrade/reinstall b/c things are working now (NFS, samba) and it took a while to figure out deps etc. b/c it was unstable build Dec 17 19:16:25 don't upgrade if it's a working production system Dec 17 19:17:41 oh no it's for home use (poor man's mirroring raid), using 2 drives and rsync over NFS Dec 17 19:18:31 that counts as production Dec 17 19:18:39 rsync over NFS! eek Dec 17 19:18:55 but I think it's worth it to upgrade if it fixes some annoying problems I've noticed with intel driver Dec 17 19:19:05 rsync over NFS is reportedly faster than rsync over ssh Dec 17 19:19:29 rsync over rsync is the best, though Dec 17 19:19:29 yeah, it's true Dec 17 19:19:33 if rather insecure Dec 17 19:20:44 the problem I've seen most recently is if you use DHCP, and no DHCP server is found, or if DHCP lease expires and no DHCP server responds, the Debian install of NSLU2 never re-tries to acquire a lease Dec 17 19:21:42 03bzhou * r4725 10optware/trunk/make/py-configobj.mk: py-configobj: added py25 subpackage Dec 17 19:21:46 you have to restart the network to get ifconfig configured properly Dec 17 19:23:59 and there's some problem with local dns - not sure if it's b/c I may not have bind configured or installed correctly? Dec 17 19:24:22 bind is a monster Dec 17 19:24:45 I just use dnsmasq for local dns Dec 17 19:26:01 seems that isn't installed on my system, that could be the issue. But mostly I use numeric IPs so it's not really a biggie but I'd prefer to use DNS so if I change the network topology Dec 17 19:26:36 ? Dec 17 19:26:43 what is your router running on? Dec 17 19:26:58 router is wrt54g right now Dec 17 19:27:03 OS? Dec 17 19:27:33 oh sorry router is WRT54G running tomato (hack of linksys firmware) Dec 17 19:27:51 ok, so where are you running bind? Dec 17 19:28:36 oh yeah bind is not running, dnsmasq is running on the router, as server, but NSLU2 is client and is getting /etc/resolv.conf correct but can't resolve DNS names Dec 17 19:29:25 no idea why that would be, unfortunately Dec 17 19:29:46 might be a router issue, might be debian Dec 17 19:30:10 it's a problem on the NSLU2 for sure... other machines (WinXP / linux) work fine in getting DHCP info from router Dec 17 19:31:38 ah there is no nslookup? Dec 17 19:40:28 03blaster8 * r632 10kernel/trunk/Makefile: Update Makefile to cope with snapshots properly Dec 17 19:41:19 03blaster8 * r633 10kernel/trunk/Makefile: Revert to building 2.6.19 by default Dec 17 19:45:54 bbl/// Dec 17 20:09:58 03bzhou * r4726 10optware/trunk/make/ (4 files): py-paste*: added py25 subpackages Dec 17 20:15:18 RobNC: if you have access to an X-Ray machine, an X-ray of the internal traces of the NSLU2 board would be *very* useful. Dec 17 20:38:24 I'm not surprised rsync over ssh is slow - lots of cpu needed for ssh encryption Dec 17 20:38:28 03blaster8 * r634 10kernel/trunk/patches/2.6.20/KERNEL: Bump to 2.6.20-rc1-git5 to fix mincore error Dec 17 20:40:03 03bzhou * r4727 10optware/trunk/make/py-protocols.mk: py-protocools: added py25 subpackage Dec 17 20:40:34 03bzhou * r4728 10optware/trunk/make/py-ruledispatch.mk: py-ruledispatch: added py25 subpackage Dec 17 20:42:04 03bzhou * r4729 10optware/trunk/make/py-simplejson.mk: py-simplejson: added py25 subpackage Dec 17 20:44:01 03bzhou * r4730 10optware/trunk/make/py-protocols.mk: py-protocols: fixed DEPENDS Dec 17 20:55:28 03bzhou * r4731 10optware/trunk/make/ (py-protocols.mk py-ruledispatch.mk): py-protocols & py-ruledispatch: fixed -dirclean Dec 17 21:37:17 03bzhou * r4735 10optware/trunk/make/tor.mk: tor: 0.1.1.25 -> 0.1.1.26 Dec 17 22:00:42 03bzhou * r4736 10optware/trunk/make/py-turbogears.mk: py-turbogears: added py25 subpackage Dec 17 22:03:41 morning rwhitby Dec 17 22:04:25 morning Dec 17 22:09:38 03bzhou * r4737 10optware/trunk/make/nanoblogger.mk: nanoblogger: added bsdmainutils for cal command now that cal has been fixed Dec 17 22:25:02 rwhitby got ur message about xray machine Dec 17 22:25:47 the whole board or just certain parts? Dec 17 22:25:49 cool - can you add to http://www.nslu2-linux.org/wiki/Info/PhotosOfTheInternals with some *real* internal images? Dec 17 22:26:17 the whole board if you think you can get resolution to make out individual tracks Dec 17 22:26:55 e.g. resolution like http://www.nslu2-linux.org/wiki/uploads/NSLU2-board-front-bare.jpg Dec 17 22:27:07 yeah probably but hopefully easy to decipher. Might take some time though but better than nothing. Too bad we can't get the BRD file from Linksys. Dec 17 22:28:40 the X-ray machine can be blurry unless it's close and high resolution. It's mainly used for detection of cold solder joints for BGAs. Dec 17 22:30:47 I know the output goes to a CCD camera but not sure if there is a scanner built-in. I.e., dump it to a file directly versus printing and then scanning the output file. Dec 17 22:32:51 anyone tried the UIO stuff from gregkh? Dec 17 22:33:17 I've patched my kernel with the uio patches, but for some reason that is beyond me, it doesn't show up in make config Dec 17 22:34:43 rwhitby: Does Redboot have recover functionality that APEX lacks? Dec 17 22:35:38 Linksys redboot has the upgrade protocol, which allows recovery without serial console access Dec 17 22:35:54 oh that's it? Since I already have serial port, it's moot then Dec 17 22:36:17 I was wondering why NSLU2 movement went with APEX versus uboot (what we use at work) Dec 17 22:36:19 well, it also has networking, so you can tftp down a new image to flash Dec 17 22:36:37 yeah networking is kinda essential :-) Dec 17 22:36:52 (that's Linksys Redboot I was referring to) Dec 17 22:37:03 Apex doesn't have networking yet, but the author is working on it. Dec 17 22:37:22 OH I didn't know that. So no tftp recovery of flash image? Dec 17 22:37:36 As for why Apex vs U-Boot - no real reason other than the Apex author was around :-) Dec 17 22:37:44 Apex has no networking yet. Dec 17 22:37:52 That's why we put it in as a second stage. Dec 17 22:38:38 ah okay. That helps! I gotcha but doesn't that eat up flash space? I guess it prolly doesn't matter if you're booting from external image but I thought the kernel was in flash? Dec 17 22:39:02 the main reason for Apex is to get past the 1MB kernel limit Dec 17 22:39:09 And it's easier to use APEX than to re-make Redboot to support varying sizes? Dec 17 22:39:16 Apex now has specific code that knows about the quirks of the nslu2. Dec 17 22:39:36 varying size what? Dec 17 22:40:08 varying size in regards to SDRAM and flash sizes and populations Dec 17 22:40:31 RobNC: you need to realise that these decisions are based on the premise that we will never expect the end-users of the 50,000 downloads on nslu2-linux firmware to ever reflash the bootloader. Dec 17 22:40:48 which is a good thing Dec 17 22:40:50 this is true! Only necessary if there is a problem with the boot loader. Dec 17 22:40:57 wouldn't want end users to mess up their boot loader Dec 17 22:41:10 yes, not everyone has access to SMT soldering tools!! Dec 17 22:41:54 so we needed a second-stage bootloader which could help us get past the 1MB limit (so we could use the generic ixp4xx kernel in Debian), and was under the 128KB erase block size granularity. Dec 17 22:42:07 so the idea is use redboot to boot Apex, then boot a kernel in flash with different offset than when booting the kernel from redboot? Dec 17 22:42:21 yes Dec 17 22:42:24 rwhitby: how big is that kernel? Dec 17 22:42:38 nbd: about 1.2MB I think Dec 17 22:42:52 hmm... maybe it could fit under 1 MB with lzma Dec 17 22:42:54 <[g2]> rwhitby just doesn't like reflashing the original bootloader :) Dec 17 22:43:02 so it's prolly not a good idea then to replace redboot with apex, just yet. Dec 17 22:43:07 -rw-r--r-- 1 root root 1213948 Dec 12 09:58 vmlinuz-2.6.19-1-ixp4xx Dec 17 22:43:10 until networking is working Dec 17 22:44:09 [g2]: yeah, I've done it once (unintended), and it's not much fun and would be beyond the capabilities of at least 99% of nslu2-linux firmware users. Dec 17 22:44:23 we had some issues at work with uboot and networking. Some switches were picky and would not do auto-negotiate correctly, and sometimes the TFTP transfer of kernel image would stall. Dec 17 22:45:11 rwhitby: on one of our mipsel targets, the gzipped kernel was 1 MB, while the lzma-compressed one was only 700k Dec 17 22:45:20 <[g2]> RobNC u-boot on which platform ? Dec 17 22:45:28 nbd: the nslu2 boots slow enough already ;-) Dec 17 22:45:34 rwhitby you would suggest backing up the flash from the application software versus through redboot? I don't know or believe that redboot can "tftp put" a file. Dec 17 22:45:39 g2: PPC Dec 17 22:45:49 RobNC: cat /dev/mtdblock* in Linux Dec 17 22:46:01 rwhitby: i think it won't take much longer than 2-3 seconds Dec 17 22:46:04 <[g2]> or dd Dec 17 22:46:25 nbd: yeah, I was just poking you. Dec 17 22:46:25 rwhitby: and you save all the flash space that would otherwise be used by apex :) Dec 17 22:46:28 :) Dec 17 22:46:50 part of me is kinda stuck in the mindset of making backup of flash with JTAG or maybe I am just masochistic! :-) Dec 17 22:46:53 the main reason for Apex is the future promise of being able to telnet into the bootloader, instead of having to solder on a serial port. Dec 17 22:47:13 and being able to change the bootloader command line from userland Dec 17 22:47:14 <[g2]> RobNC that's a great idea of you want to verify your JTAT Dec 17 22:47:19 <[g2]> s/JTAT/JTAG/ Dec 17 22:47:21 [g2] meant: RobNC that's a great idea of you want to verify your JTAG Dec 17 22:47:33 g2: exactly!! I don't want to go down a path I can't get out. Dec 17 22:47:46 <[g2]> but unless you are running from the mtdblock dd works well Dec 17 22:47:54 RobNC: as long as you don't overwrite the vendor redboot, you'll always be able to get out. Dec 17 22:48:15 the flash shouldn't be mounted once the kernel boots up, right? Dec 17 22:48:29 depends where your rootfs is mounted from Dec 17 22:48:43 <[g2]> nbd got 5 minutes ? Dec 17 22:49:04 /dev/sda1 on / type ext3 (rw) Dec 17 22:49:35 but unsure where it boots from if no USB HDDs attached - isn't that flash? Dec 17 22:49:37 <[g2]> RobNC if you boot straight to sda1 then nothing will be mounted Dec 17 22:49:56 [g2]: sure Dec 17 22:50:22 <[g2]> nbd a month or two ago we briefly chatted about verifing/kernel configs Dec 17 22:50:34 but isn't the kernel simply copied from flash to ram, and then run from ram? No linux kernel should be running from flash, I would hope! Dec 17 22:50:48 [g2]: yes Dec 17 22:51:03 <[g2]> nbd I'm just about at that point for the XScale with setting up automated regression testing Dec 17 22:51:44 interesting Dec 17 22:51:46 <[g2]> My current setup is for gcc 4.1.1 and LE Dec 17 22:52:23 <[g2]> I've got two main config I'm gonna verfiy. The avila platform and loft. The avila is basically a USBless Loft Dec 17 22:53:23 <[g2]> So I'm looking to verify, NFS, jffs2, squashfs, ramdisk, CF, for the Avila and those + USB for the Loft Dec 17 22:53:38 <[g2]> booting to all those and testing automated Dec 17 22:53:42 RobNC: correct - kernel and modules are always in ram Dec 17 22:54:47 [g2]: hi, interesting, all fully automated? Dec 17 22:54:51 <[g2]> nbd and I build the madwifi drivers from svn too Dec 17 22:55:06 <[g2]> likewise I'm pretty close Dec 17 22:55:09 so is it correct to say that the CFE variables are in flash, and that redboot might not read it, thus increasing NSLU2 RAM wouldn't be detected b/c of redboot or b/c CFE is not read? I thought that CFE is first thing read by processor. Dec 17 22:55:12 nbd: do you have lzma kernel decompression for ARM? Dec 17 22:55:55 RobNC: Linksys RedBoot has RAM size hard-coded to 32MB. Dec 17 22:56:11 likewise: not yet. but i don't think that this is hard to do Dec 17 22:56:33 <[g2]> likewise right now I can build the full kernel and drivers from scratch and copy them to the DUT (Device Under Test) Dec 17 22:56:49 nbd: http://trac.nslu2-linux.org/kernel/browser/trunk/patches/2.6.19/squashfs Dec 17 22:57:09 (I believe that's for lzma) Dec 17 22:57:19 rwhitby: and the kernel reads this from redboot? you'd think that if the CFE was changed, then the kernel should be able to work with it, irregardless of the boot loader? Dec 17 22:57:45 RobNC: ARM kernels get RAM size from ATAGs. Dec 17 22:57:48 rwhitby: no that's (my) lzma patch for squashFS Dec 17 22:57:49 rwhitby: we already have squashfs lzma'd Dec 17 22:58:02 rwhitby: i'm talking about the kernel image Dec 17 22:58:11 ok, ignore my ignorance :-) Dec 17 22:59:10 rwhitby: does that mean via pullups/pulldowns? Dec 17 22:59:37 RobNC: no, the bootloader passes the information to the kernel. Dec 17 22:59:50 <[g2]> RobNC whether a tri-stated line is pulled to VCC or GND Dec 17 23:00:24 g2 yep that would be pullup/pulldown :-) Dec 17 23:00:26 http://www.arm.linux.org.uk/developer/booting.php Dec 17 23:00:48 ATAG_MEM Dec 17 23:01:05 <[g2]> nbd are you guys still on .17 ? Dec 17 23:01:16 no, .19 now Dec 17 23:01:34 on all 2.6 targets Dec 17 23:01:36 <[g2]> nbd Ok excellent. We should compare kernel configs Dec 17 23:01:39 thanks guys ... brb ... dinnertime Dec 17 23:01:59 <[g2]> I'd like to compare the avila non-usb config Dec 17 23:02:04 [g2]: you know that I booted OpenWRT on the nas100d yesterday? Dec 17 23:02:22 <[g2]> rwhitby no, but that's great news Dec 17 23:02:26 <[g2]> congrats! Dec 17 23:02:32 good everning all, happy hacking. Dec 17 23:02:42 <[g2]> rwhitby did you built from openwrt SVN ? Dec 17 23:02:59 yep, and are pushing patches back in Dec 17 23:04:20 <[g2]> rwhitby pushing patches to which repe ? Dec 17 23:04:25 OpenWRT. Dec 17 23:04:40 <[g2]> which changes ? Dec 17 23:04:45 <[g2]> or patches ? Dec 17 23:04:58 stuff for NAS100d, then NSLU2 and DSMG600. Dec 17 23:05:25 mostly just copying stuff from our kernel svn repo Dec 17 23:05:53 the aim is to get the NSLU2 as a supported machine in OpenWRT. Dec 17 23:06:23 <[g2]> rwhitby Ok. so what's the plan with the config area and wireless support ? Dec 17 23:06:35 [g2]: no plan yet Dec 17 23:07:12 <[g2]> rwhitby and you build BE I'm guessing Dec 17 23:07:15 (i.e. I don't understand that stuff well enough to contribute to a plan) Dec 17 23:07:16 <[g2]> s/build/built/ Dec 17 23:07:17 [g2] meant: rwhitby and you built BE I'm guessing Dec 17 23:07:34 yes, but I did push a site defaults file so that LE will build :-) Dec 17 23:07:47 (not that you would run openwrt in LE mode) Dec 17 23:09:26 <[g2]> rwhitby well if someone does get DC (data coherency mode) running then there would be no penalty for running in LE Dec 17 23:09:50 <[g2]> plus all the Debian stuff would come for "free" Dec 17 23:10:14 * rwhitby wonders why someone running OpenWRT would want to use Debian stuff, instead of just using Debian. Dec 17 23:10:27 03blaster8 * r635 10kernel/trunk/patches/2.6.20/ (4 files): Refresh/Enable DSM-G600 support Dec 17 23:15:54 is LE really slower than BE? Dec 17 23:16:17 for networking, when running in Address Coherent mode, yes. Dec 17 23:16:36 ah I see, so that could make NFS faster, etc. too running in BE Dec 17 23:16:46 although I've never seen numbers. Dec 17 23:17:36 so it could be FUD Dec 17 23:17:48 <[g2]> rough estimates are about at 20% CPU loss Dec 17 23:18:15 is it possible to run BE when bootloader is LE? Dec 17 23:18:28 <[g2]> when running multiple interfaces 3+ I hear things top out at 75-80Mbs Dec 17 23:18:59 <[g2]> RobNC it's pretty easy to switch the XScale between endianness Dec 17 23:19:18 <[g2]> iirc the box boots BE by default Dec 17 23:19:38 * [g2] is not aware of any LE IXP4xx bootloaders Dec 17 23:19:55 g2: yeah but no telling what's going on inside the processor. I.e., does it have to switch anything internally, thus causing some processing or latency penalties Dec 17 23:21:29 Didn't know that... thought redboot and apex were LE b/c kernel was LE but it sounds like it can be toggled anytime? Dec 17 23:22:05 [g2]: Apex is an LE IXP4xx bootloader, if configured to be so :-) Dec 17 23:22:25 <[g2]> well it can be toggled nearly anytime, but unless you are prepared for it it's probably a bad idea Dec 17 23:22:38 (and I do configure Apex LE for the nas100d running Debian) Dec 17 23:23:27 <[g2]> rwhitby nod. I know it can be configured LE. I'm just saying except for the PXA2{5|7}X like all the IXP4xx bootloaders I know of are BE Dec 17 23:23:34 aren't ARM and MIPS processors not binary compatible? so if this is the case, what would be gained in merging OpenWRT with NSLU2 Dec 17 23:24:00 <[g2]> rwhitby that's cool that you actually build APEX in LE mode for the nas100d Dec 17 23:24:08 RobNC: they're not binary compatible at all Dec 17 23:24:13 RobNC: but openwrt is not mips specific either Dec 17 23:24:31 RobNC: it's not a merge. it's adding NSLU2 hardware support to OpenWRT Dec 17 23:24:48 ndb: rwhitby: ok gotcha Dec 17 23:25:08 (nothing more than that. I don't want to hear anyone saying "SlugOS is dead, nslu2-linux is moving to OpenWRT" or anything like that. Dec 17 23:25:17 <[g2]> nbd so the WRT54Gs are 2.6 now ? Dec 17 23:25:30 [g2]: we do support 2.6 and 2.4 on those Dec 17 23:25:34 <[g2]> the BRCM drivers are all worked out ? Dec 17 23:25:34 [g2]: but no wifi on 2.6 yet Dec 17 23:25:59 i'm working together with mb from bcm43xx to finish the rewrite of the system code based on their ssb bus code Dec 17 23:26:09 and then we have a chance of making bcm43xx with d80211 work Dec 17 23:26:26 I got a BCM FAE to get me app notes and data sheets for BCM5352 but it stinks how their NDA prohibits redistribution. Dec 17 23:28:18 wish they were a little more open like Intel but I guess it's not too surprising coming from BCM. Most other fabless vendors are the same way. Dec 17 23:33:07 so, in regards to Fat SLUG, is it possible to use Redboot to boot APEX, to then boot into Debian per:http://www.cyrius.com/debian/nslu2/ Dec 17 23:33:49 <[g2]> RobNC are you the guy with the 256MB slug ? Dec 17 23:34:20 g2: um, think so... let's see... 4 @ 32x16 yeah I guess that's right Dec 17 23:35:25 RobNC: not sure that Apex will set the memory up if it's running from RAM as a second stage bootloader. Dec 17 23:35:43 think that's right... too many other problems (mainly WRT-related). I did that mod a couple of weeks ago. Dec 17 23:36:08 <[g2]> RobNC so you are running chip stacked right ? Dec 17 23:36:28 rwhitby: is it possible to modify redboot to get rid of the 1MB limit, or perhaps this isn't so easy. Dec 17 23:36:48 rwhitby; yeah, took them off some boards at work that were headed to the scrap bin. (!!) Dec 17 23:37:35 took probably 2 hours with microscope but mainly I didn't have a good solder tip so I had to be extra slow/careful. Luckily no solder bridges. Dec 17 23:37:52 RobNC: it's possible, but no-one has done it. Dec 17 23:38:35 Not even sure if all the address lines and bank select lines are routed; didn't think of this until now. Dec 17 23:38:36 RobNC: [g2] is your RedBoot recompile man - he does it as part of his business. Dec 17 23:39:01 rwhitby: cool. *BigGrin* Dec 17 23:40:05 <[g2]> RobNC on the slug at least the extra address line to 64MB is there Dec 17 23:40:23 how large is redboot - how much space does it take? I mean, I'm not afraid of JTAG but I haven't proven out my interface yet. DB25-to-2x6 header works though (used it for BHR-HP-G54) Dec 17 23:40:28 <[g2]> so I think the worst case you'd have 128MB Dec 17 23:40:39 <[g2]> with 128MB wasted by the holes Dec 17 23:41:08 g2: that's not so bad. It would be nearly impossible to route the line from BGA pad to pin, unless there are vias already there. Dec 17 23:41:30 <[g2]> I think the extra line is there also Dec 17 23:41:40 <[g2]> RobNC you've seen the "salted slug" pics right ? Dec 17 23:42:02 <[g2]> wow :) those are probably coming up on 2.5 years old now Dec 17 23:42:43 g2: what's that salted slug? Dec 17 23:43:33 yeah just looked it up g2 - I used 512Mb SDRAM, 32x16 4 each Dec 17 23:43:34 <[g2]> RobNC http://www.nslu2-linux.org/wiki/uploads/NSLU2-board-front-bare.jpg Dec 17 23:44:02 makes a nice desktop background Dec 17 23:44:04 oh yeah I saw that... salted? that's like the bare board - hard to get pics of that except where they do the PCB manufacture Dec 17 23:44:26 <[g2]> I sent a board to some guy that power sanded it :) Dec 17 23:44:33 <[g2]> ~g2 Dec 17 23:44:35 Easy to use, portable and powerful 2D graphics library. URL: http://www.ap.univie.ac.at/users/ljubo/g2/g2.shtml Dec 17 23:44:40 <[g2]> ~[g2] Dec 17 23:44:42 you are probably really cool, and an an all around nice guy! Also he is a real funny guy, the master of NSLU2 Buildroot and a known Slug Sacrificer. And editor of SlugNews. Dec 17 23:44:53 yeah I saw that - solder on it. Can't get the solder off. Dec 17 23:44:57 <[g2]> that's where the "Slug Sacrificer" part comes from Dec 17 23:45:22 wow... man... that's a shame!! talk about a contribution! Dec 17 23:45:26 <[g2]> he baked off the components first Dec 17 23:45:57 <[g2]> Oh it was totally AWESOME and so worth it Dec 17 23:46:11 <[g2]> there was a big question about JTAG lines Dec 17 23:46:42 g2: you saw that I have access to x-ray machine. I can't use it directly (I'm not qualified) but there are a couple of technicians that wouldn't mind doing it. Dec 17 23:47:32 <[g2]> RobNC I didn't hear that. I think these days for like $35-$50 you can send stuff about for a day or two turn-around on x-rays Dec 17 23:47:48 we used to do manufacturing at our facility but all that went to Mexico. Luckily, they kept the x-ray machine and it has come in very handy for BGA soldering issues. RoHS can be such a pain when not everyone has RoHS parts! Dec 17 23:48:01 <[g2]> it's really quite amazing how much things have changed in the last couple years Dec 17 23:48:18 <[g2]> ah.. you're in CM ? Dec 17 23:48:49 CM = ? Dec 17 23:49:02 <[g2]> Contract Manufacturing Dec 17 23:49:32 Nah, technically I'm a hardware engineer doing board design but I do more systems-like work these days. Dec 17 23:50:04 <[g2]> RobNC you're nicks not registered right ? Dec 17 23:50:12 <[g2]> s/you're/your/ Dec 17 23:50:13 [g2] meant: RobNC your nicks not registered right ? Dec 17 23:51:03 g2 unsure... prolly should register it. Used to be into IRC years / decades ago but hadn't done it recently b/c of time. These linux projects have motivated its re-use.. Dec 17 23:51:47 I just picked RobNC b/c Rob in NC (really lame) but I'm not really into l33t Dec 17 23:52:58 <[g2]> RobNC I think you can see my Private Messages, but can't send me any Dec 17 23:53:16 oh I just saw that... oops Dec 17 23:53:24 <[g2]> np Dec 17 23:53:30 <[g2]> it was a FYI Dec 17 23:53:41 <[g2]> not any big secret on my end Dec 17 23:54:05 yeah gotcha. I used to use mIRC but I didn't want all the bells/whistles, etc. Dec 17 23:54:22 so I picked portable gaim Dec 17 23:55:31 so sounds like (for FAT Slug) before I go too far, best to verify the pins are wired for 256MB access (otherwise I'll waste lots of time). Dec 17 23:56:14 that's prolly not too hard but I might wait until next year b/c if I'm at work ppl will ask me questions (and I'm technically on vacation) Dec 17 23:56:18 <[g2]> RobNC what's the real purpose of the FatSlug ? Dec 17 23:57:01 <[g2]> meaning you can probably just buy a board from Gateworks with 256MB on, a JTAG connector and Redboot setup for 256MB Dec 17 23:57:25 <[g2]> Gateworks makes the boards I sell Dec 17 23:57:25 well, while doing rsync over NFS, I noticed that it consumes large amounts of memory (as there are approx 33k files). It is my hope to use FAT Slug to keep from using swap on the USB HDDs. Dec 17 23:58:55 <[g2]> I'd imagine rsync would be quite memory intensive Dec 17 23:59:30 <[g2]> IMHO the real question is how much it really matters Dec 17 23:59:35 yes exactly that's why I figured I didn't really need the speed, I just need more memory. It has to keep track of all the files, sizes, times/dates, permissions, etc. Dec 17 23:59:56 <[g2]> as you are bandwidth limited on both the Network and disk Dec 18 00:00:22 <[g2]> the network is capped around 10MBs and 8-10MBs to the disk Dec 18 00:00:38 <[g2]> you'll probably at best see about 6MB for each Dec 18 00:00:40 network seems to be SLUG limited, as I only get 2MB/sec throughput approx. That is funny how some guy said he'd get GigE working for NSLU2. That won't make any difference at all. Dec 18 00:01:20 <[g2]> you can do 10-12MBs for serving up http Dec 18 00:01:27 <[g2]> scp is limited to around 2MBs Dec 18 00:01:47 On my 2GHz laptop to a 2.4GHz linux server, over 100Mbps LAN, I only get about 4Mbps for rsync over ssh using TrueCrypt (encrypted file system) to external USBHD. Dec 18 00:02:01 * [g2] has a gigE networks setup and I can scp at 25-30MBs on it Dec 18 00:02:47 I would set up GigE but I really don't need it yet. Not too many large files going across network. No HDTV setup just yet. Dec 18 00:02:52 <[g2]> it's really quite amazing as it's single disk accesses on both ends Dec 18 00:03:41 yeah I think you're right - I got about 1MB/sec with rsync over ssh. Thus I went rsync over NFS. I don't want to even think rsync over smb (as that is a pretty dumb protocol) Dec 18 00:04:36 <[g2]> except for cost/low-power apps, it's mostly end-of-game for the 100Mbs stuff Dec 18 00:05:07 yeah exactly... that's why I want to stick with WRT54G and NSLU2 b/c of cheap attached UPSes. Dec 18 00:05:12 <[g2]> for basically 2x the cost $200 verus $100ish you can get gigE Dec 18 00:05:48 True... NSLU2 would never benefit from it. Dec 18 00:06:21 <[g2]> my boxes run POE and could run a HD off the USB port Dec 18 00:06:38 <[g2]> you could actually consolidate to one box Dec 18 00:06:50 ah thats cool... you have 48V on the pins, and have PoE endpoints to strip off power? Dec 18 00:07:04 <[g2]> sure it supports 48V POE :) Dec 18 00:07:16 <[g2]> on the Ethernet Dec 18 00:07:40 <[g2]> but it also supports 48V at the input connector too iirc Dec 18 00:08:07 <[g2]> there's 2 ethernet ports too Dec 18 00:08:18 <[g2]> and it runs at 533Mhz Dec 18 00:08:48 <[g2]> and although I've never used it there's MD5/SHA1/other support Dec 18 00:08:55 cool, I have been thinking about making up some of that stuff. I work for an ASAM company and all our stuff works off -48V Dec 18 00:09:51 is that one of those VIA devices? Dec 18 00:10:38 I think VIA made some of those low-power chips, C3 I think. Dec 18 00:12:37 <[g2]> ASAM ? Dec 18 00:13:20 sorry - subscriber access module (i.e., telecom equipment - all telecom equipment in the USA uses -48V whereas in europe it's -60V) Dec 18 00:14:21 <[g2]> ah. [g2] was in telecom for 10 years or so Dec 18 00:14:22 The point being is that there are a ton of these bricks being thrown away that convert -48V to 5V etc. Problem though is that I'd have to use some baluns on the transceivers to get the 48V on the line and taken off the line Dec 18 00:14:45 and there's supposed to be some kind of signature-type circuit for official PoE Dec 18 00:15:02 <[g2]> this is PPOE, passive POE Dec 18 00:15:15 <[g2]> the POE spec limits power to around 16W Dec 18 00:15:26 <[g2]> iirc it's .385A @ 48V Dec 18 00:15:40 <[g2]> these boxes support a little more than that :) Dec 18 00:15:53 16W isn't much. True, more of a worry about the current than voltage (I^2 * R losses) Dec 18 00:17:42 <[g2]> well for typical network gear 16W is a bit Dec 18 00:18:12 <[g2]> and for a 24/48 port hub putting out an extra amount of power per port becomes a factor Dec 18 00:18:24 true, for telecom this is butter... most our stuff is like 50-80W. Dec 18 00:18:50 <[g2]> I'd imagine SIP/VOIP phone are the main app Dec 18 00:19:16 yep, the ringing requirement hasn't changed - still must do 5REN per line! Dec 18 00:19:48 can't quite cost-reduce the power... Dec 18 00:41:03 anyone know why Linksys didn't populate the ferrite bead on one of the USB ports? Dec 18 00:48:26 03osas * r4738 10optware/trunk/make/asterisk-sounds.mk: asterisk-sounds: fix ASTERISK-SOUNDS_SITE Dec 18 00:49:25 RobNC, I'm not sure we ever figured out why Linksys underclocked the thing, so that would be another who knows question... Dec 18 00:52:37 yeah, dunno b/c 5V 2A - is 1A enough for the processor and board (5V 500mA each port) Dec 18 00:58:20 03osas * r4739 10optware/trunk/Makefile: asterisk-sounds: ready for testing after fix Dec 18 01:05:30 anyone own a Linksys RVL200 ? Dec 18 02:11:04 FYI: RVS4000 device looks like it's by SerComm too (like the NSLU2) Dec 18 02:36:01 03bzhou * r4740 10optware/trunk/ (Makefile make/asterisk-sounds.mk): asterisk-sounds: promoted Dec 18 02:45:15 03bzhou * r4741 10optware/trunk/make/py-pastedeploy.mk: py-pastedeploy: try to fix endless build Dec 18 02:45:36 03bzhou * r4742 10optware/trunk/make/py-protocols.mk: py-protocols: try to fix endless autobuild Dec 18 02:45:53 03bzhou * r4743 10optware/trunk/make/py-ruledispatch.mk: py-ruledispatch: try to fix endless autobuild **** ENDING LOGGING AT Mon Dec 18 03:00:05 2006