**** BEGIN LOGGING AT Thu Aug 09 02:59:57 2007 Aug 09 04:14:38 03rwhitby * r6646 10optware/trunk/platforms/packages-openwrt-ixp4xx.mk: Demoted iptables and quickie for openwrt-ixp4xx Aug 09 04:33:31 03bzhou * r6647 10optware/trunk/make/py-tailor.mk: py-tailor: 0.9.26 -> 0.9.27 Aug 09 09:45:27 hello Aug 09 09:45:41 anyone with openwrt on nslu2 experience here ? Aug 09 09:47:38 I'm trying to migrate a ucslugc box to openwrt, but it doesn't appear on the network after flash/reboot, gets stuck on a red status led after upslugging :( Aug 09 09:54:35 :/ Aug 09 09:54:40 sorry, not me at least :P Aug 09 09:55:46 "For assistance, look for rwhitby in the #nslu2-linux or #openwrt-devel IRC channels on Freenode." Aug 09 09:55:48 :) Aug 09 09:56:53 rwhitby: have you seen such openwrt behaviour perchance ? Aug 09 10:37:52 a77ila: look for it on 192.168.1.77 Aug 09 10:58:34 ah, the wiki seems to be a bit misleading, it suggests 1.1, dhcp or previous addr... it really was on 1.77 Aug 09 10:58:45 tnx Aug 09 10:59:07 I'll go and update the wiki Aug 09 11:00:33 we made it .77 cause that what sluggers expect, and you're likely to have a router on .1.1 already Aug 09 11:03:45 makes sense, wiki updated Aug 09 11:08:05 thx Aug 09 11:48:27 03bzhou * r6648 10optware/trunk/make/py-tailor.mk: py-tailor: 0.9.27 -> 0.9.29 Aug 09 17:14:32 Hi, guys. I'm porting a boot loader to the nslu2. Everything goes fine, except I can't get the network up. It seems, the PHY needs some additional kick to be enabled, as I see no PHYs when scanning the mdio, Any ideas what it could be? Aug 09 17:15:27 The same setup works on an ixdp425-like board, with two PHYs detected on the mdio. Aug 09 21:07:50 xl0: which boot loader? Aug 09 21:28:05 yo! Aug 09 21:28:25 oops wrong chan :) Aug 09 21:44:41 rwhitby: gPXE (former Etherboot). I'm porting it as my Google Summer Of Code project. Aug 09 21:46:20 rwhitby: Mostly done, and I get Linux booted on na ixdp425-like board, but for some unknown reason I can't get the phy up on the NSLU2. Aug 09 22:10:50 xl0: have you looked at the nslu2 board setup code in the kernel for the phy information> Aug 09 22:10:51 ? Aug 09 22:12:12 xl0: what would be really good is if you added NPE support to the Apex bootloader. Aug 09 22:12:58 if you did that, we would have over 10,000 users for your efforts immediately .. Aug 09 22:13:00 rwhitby: You mean, the phy id? Yes, it's different from my board, but as I told, the phy does not appear even when scanning all the 0-31 range. Aug 09 22:13:14 rwhitby: Why apex and not u-boot? Aug 09 22:13:40 cause the debian support of nslu2 is already using apex Aug 09 22:14:07 and the apex author has already put in specific code for the nslu2 (including working around the hard-coded kernel size limit in the stock redboot) Aug 09 22:14:32 what we really need for the nslu2 is a bootloader which: Aug 09 22:14:46 1) can work around the kernel size limit by using skip regions like apex Aug 09 22:15:12 2) can work in usb host mode for a usb->serial cable so you can interact with the bootloader console Aug 09 22:15:29 3) supports the NPEs so you can tftp from the bootloader to flash new images Aug 09 22:15:46 show me a bootloader that does all that, and we will use it. Aug 09 22:16:08 (that's over 50,000 users for the nslu2 alone) Aug 09 22:17:06 rwhitby: This is possible. I've heard the Apex has some sort of ip stack. But are you sure we can't add the needed pieces to u-boot? I'm not sure about usb, but it should satisfy all the other requirements. Aug 09 22:18:04 Do I get it right that Apex is started from RedBoot on nslu2? Aug 09 22:32:22 xl0: yes Aug 09 22:32:32 4) it has to be less than 128KiB long too Aug 09 22:32:53 if you can get u-boot to meet all those requirements, even better. Aug 09 22:33:01 I'd love to be able to use u-boot on the nslu2 Aug 09 22:33:04 bbiab Aug 09 22:33:10 rwhitby: Why can't we just remove RedBoot entirely? Aug 09 22:33:29 cause you can't ask 50,000 people who don't have JTAG to replace their bootloader Aug 09 22:34:08 gotta run - off to work Aug 09 22:34:29 See you. Aug 09 22:39:11 xl0, some users replaced reboot with apex.. like me, because i extended my nslu2 ram from 32MB to 128MB and 2 years ago there was no other way to use mor memory Aug 09 22:40:23 but we need serial access to our slugs and have no tftp support Aug 09 22:40:41 Is the RedBood source witn nslu support publicly available? Aug 09 22:41:21 i think not Aug 09 22:42:59 Not good. I think I need to power on the phy chip somehow, but no idea how.. Aug 09 22:56:34 Could someone please point me to a RedBoot partition's backup binary? ;) Aug 09 23:02:23 you need the binany code, xl0? Aug 09 23:03:16 Yes, I did not care to back up it. ;) Aug 09 23:03:35 i backup it before installing apex Aug 09 23:04:00 http://nslu2.dyndns.info/temp/ Aug 09 23:04:29 tell me is u download it, so i delete it again from the server Aug 09 23:04:37 Thanks a lot. Aug 09 23:04:39 is=if Aug 09 23:04:49 Done. Aug 09 23:07:47 j-tagging 256k should take a year on my ByteBlaster. ;) Aug 09 23:08:17 thats right Aug 09 23:10:56 there http://www.nslu2-linux.org/wiki/HowTo/RecoverFromABadFlashUsingJTAG it says about 45minutes Aug 09 23:12:08 xl0: redboot for nslu2 is available Aug 09 23:53:06 xl0: ping Aug 09 23:54:08 xl0: http://nslu.cvs.sourceforge.net/nslu/sluggo/ for a Makefile to build nslu2 redboot source (actual file is at http://www.linksys.com/support/opensourcecode/nslu2/2.3r25/nslu2-2.3R25.zip) Aug 10 01:22:57 hi !! Aug 10 01:23:38 has anyone tried to compile perl-dbd-mysql.mk (see here: http://trac.nslu2-linux.org/optware/browser/trunk/make/perl-dbd-mysql.mk) from the nslu2 svn for mipsel target ? I followed thoses instructions (http://www.nslu2-linux.org/wiki/FAQ/Optware-uClibcBuild) with OPTWARE_TARGET=ddwrt and I cant' obtain a mipsel package. Can you help me please ? Aug 10 01:46:29 I get the error: Can't locate DBI/DBD.pm in @INC Aug 10 01:47:04 toboz: you'll need to talk to eno about this (he is away on holidays at the moment I think) Aug 10 02:08:43 rwhitby: many thanks for your help. I haven't obtained the same error. do you know what is the value to give to OPTWARE_TARGET in order to obtain mipsel package ? Aug 10 02:09:25 toboz: which firmware/device are you targetting? (haven't we had this conversation before?) Aug 10 02:10:30 yes we had, an dyou said to me to try the master makefile but when reading the wiki I believed it would be easier to follow the wiki for optware building Aug 10 02:11:05 but for answering you: firmware=openwrt / device=rb532 Aug 10 02:13:45 I ve been able to build this package but I've obtained an armeb package (perl-dbd-mysql_3.0003-1_armeb.ipk); argh !! : ) Aug 10 02:14:10 I would like the same but for mipsel Aug 10 02:17:49 rwhitby: Thanks a lot. Aug 10 02:18:37 xl0: do you have a reference to the details of your GSoC project? Aug 10 02:18:55 toboz: I told you "openwrt-brcm24: Aug 10 02:19:02 "openwrt-brcm24" Aug 10 02:19:30 rwhitby: http://code.google.com/soc/2007/etherboot/appinfo.html?csaid=1384445EBBE20CF9 git.etherboot.org Aug 10 02:19:32 right ! Aug 10 02:20:17 xl0: any reason why gPXE was chosen? Aug 10 02:22:15 xl0: ok, I see that it's a project driven by EtherBoot, not just a generic bootloader project. Aug 10 02:24:16 I see little sense in porting this particular boot loader to arm, but it was a lot of fun, and the guys from Etherboot say that they needed to know how portable their code was. Aug 10 02:24:53 PING - received 10 of 10 expected Aug 10 02:24:59 Yes, RedBoot works. Aug 10 02:25:47 So etherboot on the nslu2 is interesting in it's own right. Aug 10 02:26:29 I could use it to test new images more easily ... Aug 10 02:27:06 I just like low-level system programming, and there were little such projects for the GSoC. Aug 10 02:27:26 rwhitby: You could use RedBoot directly as well. Aug 10 02:27:52 You need a serial console in both cases. Aug 10 02:28:01 right. we do that already. Aug 10 02:28:39 I'm thinking of the 50K users who don't have jtag or serial :-) Aug 10 02:30:24 rwhitby: What is a skip region btw? Aug 10 02:31:11 it's an extension to the FIS directory, which define a set of bytes in an MTD partition which should be skipped when loading that partition into memory. Aug 10 02:31:59 for example, on the NSLU2, there *must* be a length header at a certain location in flash (hardcoded into redboot) or else it won't boot. this particular location restricts the kernel to 1MiB unless you can skip that length header with a skip region. Aug 10 02:32:05 Apex is the only bootloader that supports them. Aug 10 02:32:42 To port Debian to the NSLU2, we needed a kernel larger than 1 MiB, so we needed a bootloader that could handle 16 dead bytes somewhere in the middle of the kernel image in flash. Aug 10 02:37:15 Is the fis directory standardized somewhere, or is it something RedBoot-specific? Aug 10 02:42:00 rwhitby: Btw the url in the Makefile should be changed to ftp://ftp.linksys.com/opensourcecode/nslu2/2.3R25/nslu2-2.3R25.zip. Aug 10 02:44:37 thx Aug 10 02:44:48 xl0: fis directory is redboot specific Aug 10 02:44:54 03rwhitby * 10sluggo/Makefile: Fixed the source download URL Aug 10 02:45:05 xl0: Makefile fixed **** ENDING LOGGING AT Fri Aug 10 02:59:57 2007