**** BEGIN LOGGING AT Tue Jan 29 02:59:57 2008 Jan 29 03:01:51 flashing... Jan 29 03:21:34 didn't work.. the wiki mentions flashing 0x50000000 OR 0x0... I'll try 0x0 ... Jan 29 03:27:20 03rwhitby * r1010 10kernel/trunk/Makefile: Added support for using git to manage the kernel Jan 29 03:28:14 03rwhitby * r1011 10kernel/trunk/patches/2.6.25/ (15 files): Updated latest development to bleeding edge upstream and local patches. Jan 29 04:10:47 03bzhou * r7671 10optware/trunk/make/lftp.mk: lftp: 3.6.1 -> 3.6.3 Jan 29 04:15:01 03rwhitby * r1012 10kernel/trunk/Makefile: Fixed non-git operation Jan 29 04:16:05 03bzhou * r7672 10optware/trunk/make/yougrabber.mk: yougrabber: 0.29.3 -> 0.29.4 Jan 29 12:42:06 03rwhitby * r1013 10kernel/trunk/patches/2.6.25/ (ixp4xx-fsg-board-support.patch ixp4xx-fsg-led-driver.patch): Split the fsg patch Jan 29 13:08:32 03rwhitby * r1014 10kernel/trunk/patches/2.6.25/ (10 files): Refreshed patches against latest upstream Jan 29 17:05:59 03bzhou * r7673 10optware/trunk/make/py-decoratortools.mk: py-decoratortools: 1.6 -> 1.7 Jan 29 17:12:43 03bzhou * r7674 10optware/trunk/make/py-simpy.mk: py-simpy: 1.8 -> 1.9 Jan 29 17:27:12 So I'm trying to build a new flash image for my (now bootable) slugos/le + debian NSLU2. I intend to load the new image onto it via upslug2, so I'm using slugimage to build a new image. I assume that I need the apex loader off of the debian-installer image. I'll use the vmlinuz from the kernel-image package. That leaves me with the ramdisk left to build. What do I need to put in the ramdisk? Jan 29 18:09:51 I've tried flashing an LE flavour of Apex (using jtag) and it just doesn't work.. tried flashing BE and it does so it's not my procedure/jtag.. Jan 29 18:11:02 I tried flashing it with both endianness using jtag's endian command.. one doesn't do anything and the other boot but outputs what looks like byte swapped text.. Jan 29 18:11:41 anybody upgraded their bootloader to LE who can help? Jan 29 18:23:24 03osas 07slugos-4.8-beta * r458 10slugos/openembedded/packages/meta/slugos-packages.bb: slugos-packages.bb: stunnel promoted Jan 29 18:40:08 :( Jan 29 18:40:21 sorry, I don't think I can do anything for you Jan 29 18:40:49 I've just discovered that I don't have a working copoy of devio Jan 29 18:42:22 and my new slug image is *nearly* ready too... Jan 29 18:42:48 Not that I know whether it will work or not. Jan 29 18:42:51 03osas 07slugos-4.8-beta * r459 10slugos/openembedded/packages/ (4 files in 4 dirs): asterisk-core-sounds-en: 1.4.7 -> 1.4.8 Jan 29 18:53:38 03osas 07slugos-4.8-beta * r460 10slugos/openembedded/packages/ (4 files in 4 dirs): asterisk-extra-sounds-en: 1.4.6 -> 1.4.7 Jan 29 19:41:51 03bzhou * r7675 10optware/trunk/make/libcurl.mk: libcurl: 7.17.0 -> 7.18.0 Jan 29 19:45:42 Annirak: thanks anyways. I'm trying various byte swap methods (as per wiki) on the apex.bin file before flashing it.. long but covers all possiblilities.. Jan 29 19:46:26 I did open a can of worms by changing endianess! :) Jan 29 20:51:26 VoodooZ: any luck? Jan 29 23:05:44 Annirak: nope. The only le apex that worked was an old one I found on the wiki (apex.flash). Jan 29 23:06:38 so I'm at the point where I'm comparing hexdumps of the variations iterations (cross-compilers, byte-ordering, etc..) for differences.. Jan 29 23:06:57 Is the SERCOMM header crap necessary if you are not using redboot? Jan 29 23:07:33 VoodooZ_: no Jan 29 23:07:58 VoodooZ_: I use LE apex as second stage all the time (the Debian distro's apex is LE). Jan 29 23:08:18 I guess your problems is using Apex LE as first stage bootloader. I doubt that has ever been tested. Jan 29 23:08:33 ok. Jan 29 23:08:44 I'm a pioneer! ;-) Jan 29 23:09:22 The wiki mentions of two ways of byte-swapping the bin files and they both give different results so I'm a bit confused... Jan 29 23:10:01 I would think the instruction to set the endianess would be the first thing in apex right? Jan 29 23:11:53 rwhitby: Is the OE built cross compiler the same for both slugosbe and le? Jan 29 23:12:29 I noticed that there's a slugos, slugosbe and slugosle folders... the cross compiler is in the common slugos folder... Jan 29 23:13:14 the slugosle and slugosbe folders are deprecated - it's all built in the slugos directory now Jan 29 23:13:21 in fact, the only apex one I was able to flash successfully is the one designed for debian on the wiki... Jan 29 23:13:28 ah ok.. Jan 29 23:13:56 but the compiler is the same right? they are just called differently? Jan 29 23:14:15 no, there is arm-linux- and armeb-linux- ... Jan 29 23:15:20 ok. I noticed it for angstrom but not slugos... I'll compare the hexdump of both binaries of apex then.. Jan 29 23:15:57 Isn't the ixp defaulting to LE on reset anyways ? Jan 29 23:18:24 I'll try building apex with my freshly built slugosle-made compiler then.. Jan 29 23:26:04 VoodooZ_: I have no idea what it defaults to, cause I've never replaced RedBoot. Jan 29 23:26:53 I think apex is attempting to load my kernel from the wrong location. What do I do? Jan 29 23:27:21 Annirak: use the correct apex config file during the build Jan 29 23:27:57 are you building apex with the Makefile at http://svn.nslu2-linux.org/svnroot/kernel/trunk/ ? Jan 29 23:30:05 rwhitby: does OE's apex package build differently for le and be? I guess I'll have to try it instead of building my own.. Although it worked before.. Jan 29 23:30:39 rwhitby: No, I was using a prepackaged apex file, not a built apex Jan 29 23:30:59 Annirak: ok, so you can change the kernel load location on the apex cmdline Jan 29 23:32:05 rwhitby: when I use setenv on the apex commandline, I get "environment region is unwritable Jan 29 23:32:14 is the makefile you just mentioned the same as the master makefile or? Which one should I use for building apex? Jan 29 23:32:25 VoodooZ_: it's the kernel svn repo Makefile Jan 29 23:32:38 Annirak: that is configurable item in apex.. Jan 29 23:32:47 Annirak: you need to erase the whole partition first, and pad apex with 0xFFs. Jan 29 23:33:29 rwhitby: I wrote apex and my kernel image using upslug2 after building the image with slugimage Jan 29 23:34:09 how big was the apex binary? if it was more than about 40K then it's probably padded with zeros and won't work for changing the environment. Jan 29 23:34:20 It was 48k Jan 29 23:34:44 ok, pastebin the apex bootloader serial output then Jan 29 23:35:01 hold on Jan 29 23:35:12 I think this might have been my other apex bin Jan 29 23:35:16 which was 127k Jan 29 23:35:22 and would explain the error Jan 29 23:35:54 early debian images had zero padding in the apex partition for some reason Jan 29 23:36:07 Yeah, that'd do it. I'll try the 48k version and that should fix it Jan 29 23:39:45 perhaps I'll try one of those debian apex... Jan 29 23:40:29 rwhitby: do you remember marc singer's alias? does he still hang around here? Jan 29 23:40:49 he doesn't. you need to email him. Jan 29 23:41:16 ok. thanks Jan 29 23:41:23 VoodooZ_: none of the differences between the debian and slugos/le apex binaries would affect whether it boots or not Jan 29 23:42:53 ok. I could also try flashing back a be build of apex and using the -s option of apex's copy to byte swap the kernel.. unless the kernel doesn't change the endian bit.. Jan 29 23:46:15 I guess all I want is to get an EABI OS really.. angstrom or not.. I doubt slugos does yet.. Jan 29 23:52:52 03rwhitby * r1015 10kernel/trunk/patches/2.6.25/ (3 files): fsg led support can be included in the initial board patch Jan 30 00:12:10 03rwhitby * r1016 10kernel/trunk/patches/2.6.25/ (3 files): Submitted the leds-gpio patch upstream Jan 30 00:24:02 rwhitby: so is angstrom using a second stage apex or a shim to run in LE mode? Jan 30 00:24:49 apex Jan 30 00:25:01 ok. Jan 30 00:25:47 I guess I've lost enough time on this stuff... I need to start working on my project so I'll go with whatever gets me angstrom working.. Jan 30 00:27:07 so redboot is BE and loads an apex LE, which in turns starts an LE kernel right? Jan 30 00:51:19 yep Jan 30 00:55:17 so technically I could do the same using apex->apex->linux... ok... but in angstrom's case I have no clue how things are arranged... Jan 30 00:55:36 angstrom flash layout is identical to slugos Jan 30 00:55:40 (for the nslu2) Jan 30 00:56:43 ah ok.. Jan 30 00:57:17 But my angstrom-le build doesn't have a 8Meg image... only a rootfs and kernel... Jan 30 00:57:56 do you know which angstrom (le/be) has issues with libusb? (because of EABI) Jan 30 01:07:32 for angstrom, you need to inherit nslu2-image in your local.conf to get an image. Jan 30 01:07:43 I wasn't aware of libusb issues. Jan 30 01:08:03 ok. thanks. I'll add that to my TOTRY list.. Jan 30 01:14:40 VoodooZ_, they are NOT issues with EABI but rather unresolved libusb issues in BE. Jan 30 01:15:41 ah. ok.. Just couldn't remember which one to avoid... thanks... My darn irssi log was off so and my memory is lacking... Jan 30 01:16:25 ka6sox-laptop: btw, do you know of anybody ever getting apex built and flash for LE? Jan 30 01:16:46 I am using apex for LE on a Lude. Jan 30 01:16:53 nice Jan 30 01:16:55 but not on my slugs. Jan 30 01:17:06 but isn't it the same arch? Jan 30 01:17:31 I'm only wondering about the apex part as I can't get it to start correctly.. Jan 30 01:17:35 yes, but different machine and flash layout. Jan 30 01:18:07 flash layout shouldn't matter though right.. I mean I'm jtagging the bin file straight to 0x0... Jan 30 01:18:25 so what is lude then? not an ixp42x? Jan 30 01:19:01 yes, ixp420 with 16MB of Flash Jan 30 01:19:07 using redboot Jan 30 01:20:09 oh... so you are using apex only as second stage though? Jan 30 01:20:13 yes Jan 30 01:20:30 ah ok.. that won't help me.. I thought you were using only apex... Jan 30 01:20:48 no, using apex as 2ond stage only. Jan 30 01:21:29 I'm pretty sure it's possible as I got one load to boot but the normal apex messages were garbled... probably in the wrong order... Jan 30 01:22:11 then again, who knows if the processor was really in LE mode... I'm working in the dark here... Jan 30 01:22:31 interesting stuff nonetheless.. Jan 30 01:22:59 beewoolie would be helpful for this... Jan 30 01:23:08 you can email beewoolie about this. Jan 30 01:23:27 marc's a good guy Jan 30 01:23:44 I know but I couldn't find his email on his page!.. Jan 30 01:24:35 I mean I could just use slugosbe but I'd love to get the performance boost that EABI provides... Jan 30 01:24:56 do you happen to know his email? Jan 30 01:26:39 VoodooZ_: many people build and flash apex for LE, but not as first stage bootloader Jan 30 01:27:43 yeah. I know but it should technically work so it's going to bug me if I don't find out... Jan 30 01:28:27 VoodooZ_: google for "marc singer apex @buici.com" Jan 30 01:28:35 I understand that most slug users shy away from changing it but apex was designed for other people.. Jan 30 01:28:46 ok thanks Jan 30 01:29:01 worth a shot.. Jan 30 01:29:14 well, apex doesn't have network support, so most people don't want to wait for serial loads when you can tftp with redboot Jan 30 01:29:37 true.. but for a custom app it's not an issue once everything works.. Jan 30 01:30:05 nothing's ever an issue once everthing works ... Jan 30 01:30:27 hehehe Jan 30 01:31:06 well, once apex works I'll be fine.. I was for the last 2 years... Jan 30 01:37:23 anyways, I emailed marc so I'll keep my fingers crossed... **** ENDING LOGGING AT Wed Jan 30 02:59:56 2008