**** BEGIN LOGGING AT Sat Oct 11 02:59:57 2008 Oct 11 03:28:26 [florian] or nbd: ping Oct 11 05:45:12 hmm, a fresh checkout failed to build... patching uclib http://pastebin.ca/1225155 Oct 11 06:07:31 In ~/trunk/build_dir/toolchain-mipsel_gcc3.4.6 there are 4 items. ~linux ./linux-2.4.35.4 ./uClibc and uClibc-0.9.29 and uClibc-0.9.29 has the test that borked because it couldn't find the kernel headers in ./uClibc/include Oct 11 06:44:06 I guess my next question would be... if my system is adequate... then shouldn't the uClibc be a link to uClibc-0.9.29 ? Oct 11 07:30:56 ok... well it looks broken when using uClibc 0.9.29 in a bcm 2.4 kernel build Oct 11 07:45:55 I'll try switching to 0.9.28 Oct 11 11:04:39 nbd * r12952 /branches/8.09/ (5 files in 4 dirs): merge build system changes from trunk to 8.09 Oct 11 13:20:07 florian * r12953 /packages/ipv6/6scripts/ (Makefile files/6tunnel.init): Fix ip path checking and make 6scripts depend on kmod-ipv6 (#4085) Oct 11 13:45:14 florian * r12954 /packages/libs/libtheora/patches/ (. 001-no_docs_tests.patch): Disable docs and tests for libtheora (#4080) Oct 11 14:20:27 MateIn4: ping Oct 11 14:27:53 noz: pong Oct 11 14:33:28 noz: have you looked at the GPL code (http://www.usr.com/support/s-gpl-code.asp?loc=unst%22)? it's usually good for finding out the HW setup. Oct 11 14:37:43 MateIn4: Well, I've downloaded the redboot sources, but the NPE stuff seems to be licence-encumbered, in a way that means I can't link it against redboot Oct 11 14:38:02 What did yo udo? Oct 11 14:38:06 *you do Oct 11 14:39:11 Doesn't really bother me. It's for personal use. Oct 11 14:39:43 Do note that the dd-wrt uses those sources. Their redboot is based on an older version of the intel souces. Oct 11 14:41:54 There's an option to separate the binary blob into a separate partition. Oct 11 14:43:01 You can also take a look at the apex bootloader. It also has support for the npe based on the intel library. Oct 11 14:46:55 Sure I could, but if I've done the work, I'd like to make it available for other people, and the licence seems to suggest that's not possible Oct 11 14:49:33 I'm not so clear on that. I've distributed a .epk file that contain my changes that goes on top of the intel sources. Oct 11 14:49:55 I suppose I could provide a set of instructions on how to compile your own, but that will raise the bar rather a lot in terms of the number of people that will attempt it. Oct 11 14:50:40 noz: As far as I know if you accept the license then you are allowed to distribute it - ask Kaloz he distributes the crypto enabled microcode for ixp4xx Oct 11 14:50:41 you should ask intel. many vendors are distributing redboot based on those sources. Oct 11 14:51:06 Maybe I've read it incorrectly Oct 11 14:51:34 MateIn4: That's what confused me Oct 11 14:52:01 noz: I can't disagree that it's confusing. Oct 11 14:52:53 noz: let me know if you want my .epk file. you may be able to use it as a template for your changes. Oct 11 14:53:17 "You may reproduce, create derivative works of, and distribute ... in executable or source form ... subject to the restrictions set forth..." Oct 11 14:54:09 IANAL Oct 11 14:54:43 "C: ... You may distribute ... subject to the condition that each End User enters into a written agreement ... that (i) prohibits copying and distribution" Oct 11 14:57:48 The main problem seems to be that you've got to have a click-through or written agreement for distribution, so you can't just put up a compiled redboot image on a web page Oct 11 14:59:47 maybe I should add a note to the wiki that by downloading the user agrees to the end use agreement Oct 11 15:02:01 noz: where did you find this license? I'm searching the ecos source code... Oct 11 15:02:58 It appears as part of the download process when you click on "RedBoot NPE microcode" from http://www.intel.com/design/network/products/npfamily/download_ixp400.htm Oct 11 15:03:50 The really bizarre thing is that they provide source code, and then say that you can't use it in any way that enforces OpenSource lvences Oct 11 15:03:56 *licences Oct 11 15:04:20 just get the ecos sources. you don't need anything else Oct 11 15:04:56 Then it says that "The terms of the software license agreement included with any software you download will control your use of the software" Oct 11 15:05:24 Then you need to log in / register Oct 11 15:05:55 don't download that. just get the intel modified ecos sources from here: ftp://ftp.ges.redhat.com/private/gnupro-xscale-030422/redboot-intel-ixp4xx-070320.tar.gz Oct 11 15:06:03 Thanks Oct 11 15:07:18 I have, but the instructions say I need to copy over the IxNpeMicrocode.c file from the link I gave Oct 11 15:07:39 it includes the microcode Oct 11 15:07:45 Oh. Oct 11 15:08:26 Oh, yes, so it does.... Oct 11 15:08:27 oops, I take that back. it does not Oct 11 15:08:44 ;) Well it has that file - I didn't look inside Oct 11 15:09:24 You're right - that file has: #error "Replace this file..." Oct 11 15:12:21 use the dd-wrt sources instead. they include the file ;) Oct 11 15:47:45 MateIn4: Do they? Not in the v23-final tarball I just downloaded (at least not called the tight thing if they are) Oct 11 15:48:02 I've only used the svn sources Oct 11 15:49:43 [florian] or nbd: ping Oct 11 15:50:34 Where are they? The website doesn't seem to help you to find the source very easily... Oct 11 15:55:41 ...got it Oct 11 15:57:18 noz: maybe a better approach is to use openwrt's binary blob instead Oct 11 15:57:38 (or not! http://svn.dd-wrt.com/DD-WRT doesn't seem to be a subversion archive) Oct 11 15:58:45 use svn://svn.dd-wrt.com/DD_WRT/redboot Oct 11 15:59:20 OK - that at least is a repo Oct 11 16:00:59 Now really got it - thanks Oct 11 16:06:02 MateIn4: Now I'm in a position to look at the changes you made... Oct 11 16:10:23 how do you want me to send it to you? Oct 11 16:14:09 website would be easiest Oct 11 16:16:40 noz: http://mysite.verizon.net/jvasco/mi424wr/mi424wr.epk Oct 11 16:23:11 Forgive my ignorance - how do I add the epk to the redboot distribution? Oct 11 16:24:56 an epk file is just a tar file. use: tclsh $ECOS_REPOSITORY/ecosadmin.tcl add Oct 11 16:33:13 MateIn4: What did you add to main.c? Oct 11 16:34:12 that was for the reset button. you can ignore that. for details see this: http://ecos.sourceware.org/ml/ecos-patches/2008-08/msg00054.html Oct 11 16:35:24 OK. I will, except that it's overwritten the file - I'll get it back out of the original tar Oct 11 16:35:42 I should remove it from the epk file... Oct 11 16:47:16 What's the "default ESA" listed in the ethernet definitions? Oct 11 16:48:26 it's a default ethernet address. it's overridden using fconfig Oct 11 16:49:08 Do you think I should change it from yours (for the MI424), or leave it? Oct 11 16:50:01 don't bother. It's from the original sources. once ecos is up use fconfig to override it. Oct 11 16:50:08 k Oct 11 16:50:29 Need to work out the PHy numbers now Oct 11 16:51:15 did you check to see if the switch needs any initialization? Oct 11 16:53:55 Not yet, but the Jungo firmware doesn't load anything for it, so I suspect not. I have a list of 12 (!) PHYs to choose from Oct 11 16:55:02 my jungo didn't do anything either but it initialized at boot time. Keep that in mind if it doesn't work. Oct 11 16:55:41 regarding the PHYs, I suspect that they don't do proper decoding so the same phy is found in multiple addresses. Oct 11 16:56:02 OK, but of course if I don't get it right first time, then I'm into JTAG territory! Oct 11 16:56:37 It lists 9, 16-20 and 24-29 Oct 11 16:56:44 I was lucky because my HW used a separate phy for the WAN port. Oct 11 16:57:27 Ditto - the 4-port switch is separate from the WAN, which is handled by an Altima AC101 Oct 11 16:58:07 In that case, just use the phy corresponding to the WAN port. redboot can only use one interface anyway. Oct 11 17:16:47 What's the "Platform Machine Type" number - so I need to find out some kind of hardware registration? Oct 11 17:39:18 noz: that's the same number that's found in mach-types.h Oct 11 17:40:14 I've registered the mi424wr and got that added to mach-types.h. Looks like for you it should be 1762 Oct 11 18:47:53 MateIn4: mi424wr help request in #openwrt Oct 11 19:03:42 MateIn4: Presumably you had to change the interrupt allocation in hal_plf_ints.h? Oct 11 19:05:50 noz: I'll have take a look. It's been a while and I don't remember. Oct 11 19:11:35 noz: I did not change hal_plf_ints.h. I don't think redboot uses interrupts. Oct 11 19:11:57 OK. That helps! Oct 11 19:13:19 I dunno if anyone noticed earlier... but I had a error with uClibc 0.9.29... on a fresh pull. The patches wouldn't apply. Oct 11 19:14:13 I could have done something wrong, I admit... but I don't think so this time. Oct 11 19:17:34 Here is my build script... http://pastebin.com/m75e59de9 Oct 11 19:22:57 MateIn4: What was the MI424WR_LATCH - something particular to that board? Oct 11 19:24:01 noz: yes, the mi424wr has a bunch of LEDs that are controlled by a latch (not enough GPIOs) Oct 11 19:24:43 OK. I guess I need to work out what this board has (don't quite know how though!) Oct 11 19:25:29 noz: take a look at the GPL source. it might have some hints in the include files or init code. Oct 11 19:25:35 k **** ENDING LOGGING AT Sun Oct 12 02:59:57 2008