**** BEGIN LOGGING AT Thu Jul 28 23:59:56 2005 Jul 29 00:33:05 my slug is 128Mb! Jul 29 00:33:11 i just need to hack the firmware. Jul 29 00:38:36 the ram worked fine for the default 32mb after the upgrade. Jul 29 00:39:08 stacked up with 4 x 256mbit samsung chips. Jul 29 01:55:55 anyone know what i need to do to the kernel and bootloader to get 128mb ram working? Jul 29 01:57:20 piggypacked? Jul 29 01:57:44 it usually works on the second boot, but if it doesn't add mem=128M to linuxargs in /boot/params Jul 29 01:57:52 aaargh Jul 29 01:57:57 ha Jul 29 01:58:14 * koen confuses this with #handhelds.org and troubles with 128M on a h5550 Jul 29 02:01:23 landslide: I've been thinking about taking a sdram stick and upgrading to 128, you beat me too it :-) Jul 29 02:04:34 Well, does now someone know, what package has been patched to allow swap over nfs, and where i can find that patch? Jul 29 02:08:44 morlenxu1: dunno Jul 29 03:28:20 koen: is that all? Jul 29 03:28:54 landslide: for the 128M, no. I was confusing channels Jul 29 03:29:12 oh ;) Jul 29 03:31:43 koen: I've fixed the perl thing, I think, I have to test it, and then I'll upload Jul 29 03:31:56 DaKa2: cool Jul 29 03:32:15 simple problem of armeb and arm.. Jul 29 03:32:20 :-) Jul 29 03:34:40 :) Jul 29 03:37:57 so where do i start to create my own kernel and development environment so I can share packages etc. Jul 29 03:41:08 landslide: http://www.nslu2-linux.org/wiki/HowTo/UseTheMasterMakefile Jul 29 03:46:09 thanks ;) Jul 29 03:53:00 daka2: i get this Jul 29 03:53:06 --16:21:42-- http://ipkg-us-dyoung.nslu2-linux.org/monotone/nslu2-linux.db.gz Jul 29 03:53:07 => `monotone/nslu2-linux.db.gz' Jul 29 03:53:07 Resolving ipkg-us-dyoung.nslu2-linux.org... Jul 29 03:53:07 205.196.212.100 Jul 29 03:53:08 Connecting to ipkg-us-dyoung.nslu2-linux.org|205.196.212.100|:80... Jul 29 03:53:09 connected. Jul 29 03:53:09 HTTP request sent, awaiting response... 404 Not Found Jul 29 03:53:11 16:21:43 ERROR 404: Not Found. Jul 29 03:53:15 any idea what that is? Jul 29 03:54:16 uhh.. I think someoene done something wrong.. seems like the initial db is not on that mirror Jul 29 03:54:23 lets see where you can get it-- Jul 29 03:54:27 s/--/../ Jul 29 03:54:49 wait, I can fix that. Jul 29 03:54:54 ok Jul 29 03:56:34 how do i upload images to the wiki? Jul 29 03:56:41 i have some photos of my stacked ram. Jul 29 03:56:44 128mb slug ;) Jul 29 03:56:57 cool Jul 29 03:57:58 landslidevm: can you wait 10 minutes? I want to verify the database... Jul 29 03:58:11 yeah no probs. Jul 29 03:58:15 i've got all night. Jul 29 04:04:34 okay, try that again. Jul 29 04:04:49 I really dunno what happened; the directory got turned into a file for some reason. Jul 29 04:04:55 weird. Jul 29 04:04:57 it's working now. Jul 29 04:05:04 53mb coming down. Jul 29 04:05:09 i'm shaped at the moment.. spewing... ;) Jul 29 04:05:56 ok, now what about uploading to the wiki? Jul 29 04:06:04 i've got 3 images. Jul 29 04:06:08 landslide: can you put them somewhere I can download them from? Jul 29 04:06:17 yep sure can. Jul 29 04:06:22 will be slow though. Jul 29 04:07:01 http://thornley.afraid.org/downloads/LandslideFatSlug_Sml.JPG Jul 29 04:07:07 http://thornley.afraid.org/downloads/LandslideFatSlug_Big.JPG Jul 29 04:07:15 http://thornley.afraid.org/downloads/LandslideJTag.JPG Jul 29 04:13:28 is there a way to test my ram? Jul 29 04:14:15 landslide, landslidevm - are you two different people? Jul 29 04:14:27 vm = vmware, I guess Jul 29 04:15:02 ah, ok, cool. Jul 29 04:17:01 landslide: downloading the photos to the wiki uploads dir now. You will be able to access them by the base filename by using: Attach:LandslideFatSlug_Sml.JPG Jul 29 04:17:08 sweet. Jul 29 04:17:14 thanks rwhitby. Jul 29 04:17:21 daka: yeah i'm lazy :( Jul 29 04:17:31 np - thanks for contributing Jul 29 04:19:43 landslide, you want to test your 128MB of ram? Jul 29 04:20:19 yep. Jul 29 04:20:32 do you have JTAG yet? Jul 29 04:20:34 i'm downloading the build environment now. Jul 29 04:20:49 yeah i have jtag.. but it's not working through vmware. Jul 29 04:20:51 koen: the perl fix is uploaded to nslu2 now Jul 29 04:21:03 DaKa2: many thanks Jul 29 04:21:05 i could get it working by installing debian natively. Jul 29 04:21:08 which i'm planning to do. Jul 29 04:21:59 koen: hehe, I dont know if it works yet :-) Works for both a generic arm and armeb distro, but I dont know about others Jul 29 04:22:21 DaKa2: I secretly don't care about others ;) Jul 29 04:22:36 :) Jul 29 04:23:09 * koen is surrounded by arm devices Jul 29 04:25:19 there are others? Jul 29 04:26:53 landslide: if you have JTAG, you could step through the ram and read/write stuff. Jul 29 04:27:18 landslide: if not, you should read the Apex Wiki page. Are you going to be openslug or unslung/ Jul 29 04:27:21 ? Jul 29 04:27:31 i'm openslug at the moment. Jul 29 04:28:12 ok i might try get the jtag cable working... got another hour until that monotone package comes down. Jul 29 04:28:14 bbl. Jul 29 04:29:12 Openslug is known to work with apex and 64MB. Jul 29 04:29:40 >64MB there was some talk about possible issues with PCI Windows; but youre really going to be trailblazing there... Jul 29 04:29:58 so the only option for bootloader is apex? Jul 29 04:30:32 I have srcipts that make a redboot thats supposed to work with 64MB; but I dont have a fatslug, so I havnt been able to test it. Jul 29 04:31:18 ok. Jul 29 04:32:02 its in the nslu CVS. I think I added a target setup-sluggo-developer for the sluggo scriptps in the master makefile... Jul 29 04:49:25 dyoung: where abouts do i connect vcc from the jtag programmer? Jul 29 04:49:30 is there a vcc point you know of? Jul 29 04:50:12 landslide: http://www.nslu2-linux.org/wiki/HowTo/AddAThirtyFourPinUniversalConnector :-) Jul 29 04:50:27 lovely ;) Jul 29 04:50:41 I borrowed 3.3V from pin1 of the serial connector. Jul 29 04:51:12 http://nslu2-linux.thestuffguy.com/gallery/jtag Jul 29 04:51:16 is exactly what I did. Jul 29 05:03:05 do i still need to power up the slug to use jtag? Jul 29 05:06:34 yes, Jul 29 05:07:29 and with jtag connected it will continue to work until you start shifting stuff in; so make sure you interrupt the boot early so you dont risk a hard shutdown. Jul 29 05:12:19 yep i forced it into reboot Jul 29 05:12:22 redboot sorry Jul 29 05:12:27 anyway it's working sweeeeeet. Jul 29 05:12:49 i can't believe the jtag programmer came in use, it's been sitting in my drawer at work for 3 years. Jul 29 05:13:05 i had it to program some xilinx fpga's for a mobile computer we manufactured. Jul 29 05:13:08 working as in you replaced redboot and can see 128MB? Or working as it, "It still turns on!" Jul 29 05:13:32 working as in it still turns on and i can detect the jtag chain through vmware running debian. Jul 29 05:13:33 landslide: I use a Xilinx DLC5 cable too. Jul 29 05:13:33 that is cool. Jul 29 05:14:33 can i break it with the dlc5 ? Jul 29 05:14:50 like do something in software and fry a chip? or am i free to play around and experiement. Jul 29 05:15:40 as long as you are powering your cable from the DUT you should be fine. Jul 29 05:16:15 dut? Jul 29 05:16:22 what is that? Jul 29 05:17:32 device under test Jul 29 05:18:30 ah ok... i understand. Jul 29 05:18:54 how did you trace back all the pins in the first place? Jul 29 05:19:05 it must be SUPER difficult with a BGA device. Jul 29 05:19:28 salted slug Jul 29 05:19:53 http://www.nslu2-linux.org/wiki/uploads/NSLU2-board-front-bare.jpg Jul 29 05:19:59 http://www.nslu2-linux.org/wiki/uploads/NSLU2-board-rear-bare.jpg Jul 29 05:20:04 800Kb pictures Jul 29 05:21:27 haha salted slug. Jul 29 05:21:44 you guys are CRAZY :) Jul 29 05:23:36 Well, does now someone know, what package has been patched to allow swap over nfs, and where i can find that patch? Jul 29 05:34:34 dyoung: can i use the jtag tools to determine if the ram chips are working properly? Jul 29 05:35:03 which "the jtag tools" ? Jul 29 05:35:26 openwince-jtag Jul 29 05:35:28 the openwince stuff? Jul 29 05:35:31 ah.. Jul 29 05:35:32 yep Jul 29 05:35:40 Maybe; but I dunno the steps to do it. Jul 29 05:36:16 because you need to tell the ixp you have more memory than redboot reported. Jul 29 05:36:30 THEn you need to poke the stuff in. Jul 29 05:36:37 and read it back Jul 29 05:36:50 honestly, apex is going to be the easiest... Jul 29 05:37:46 yeah, I think we need to say that if you want a fat slug, then you use apex Jul 29 05:39:17 ok Jul 29 05:41:17 i'm going to learn about apex... Jul 29 05:41:19 morlenxu1: I just tried the instructions http://www.nslu2-linux.org/wiki/HowTo/MountNFSSwapPartition on my regular linux workstation and they work without a flaw. dmesg says Adding 131064k swap on /dev/loop0. Priority:-1 extents:1 Jul 29 05:42:14 malfi: Yes, but it want to use swap over nfs on a normal linux distribution, and there it don't work. Jul 29 05:42:22 huh? Jul 29 05:42:29 So it interests me, how it is done in openslug. Jul 29 05:43:06 you cant swap DIRECTLY with openslug, you need to do the loopback mount trick. Jul 29 05:43:06 morlenxu1: nfs-server is on gentoo, workstation using that swapfile via nfs is slackware 10.1, I think they qualify as "normal linux distributions" Jul 29 05:43:51 we didnt patch anything. Jul 29 05:44:00 malfi: Well, but there need to be a patch for swapon, look what it says when doing this: Jul 29 05:45:12 morlenxu1: if you want to use a direct swapon /mnt/nfs/swapfile, you might need a patch, but going this losetup route should work out of the box Jul 29 05:45:59 hm well, now it works. Jul 29 05:46:05 * malfi giggles. Jul 29 05:46:08 :) Jul 29 05:46:11 wtf Jul 29 05:46:24 I'm waiting since yesterday... :) Jul 29 05:46:31 Now it works. Jul 29 05:46:34 What a f. Jul 29 05:46:38 Thanks for help. :) Jul 29 05:47:17 morlenxu1: just the mere presence of dyoung makes things work Jul 29 05:47:17 thats the demonstration effect. Jul 29 05:48:03 :) Jul 29 05:48:58 morlenxu1: but remember that there are huge security implications as you are sending memory pages unencrypted over the network Jul 29 05:49:43 dyoung: jtag is REALLLY slow is that normal. Jul 29 05:49:51 i'm just trying to backup the flash. Jul 29 05:50:00 it's like 10K every 2 minutes or something Jul 29 05:50:17 Well, that doesn't metter much. The employes at my company run a linux-distribution, which is running from a cd, so some extra swap is nice, when starting openoffice or something else. Jul 29 05:51:23 landslide: yep. it's really slow Jul 29 05:52:08 yeah, thats sounds about right. Jul 29 05:52:17 the BSR is 498bits Jul 29 05:52:29 YUCK! Jul 29 05:52:32 morlenxu1: I don't think that network-swap improves performance as long as you are not using a 1000mbit network Jul 29 05:52:32 hehe. Jul 29 05:52:57 how big is the bootloader? Jul 29 05:53:00 There is a fairly detailed analysis of the slowness someplace on www.openjtag.net Jul 29 05:53:01 couple of hundred k? Jul 29 05:53:42 there are 2 blocks allocated for the bootloader. I think redboot is 220k Apex is like 50k or so but it doesnt have the IAL on it. Jul 29 05:54:06 IAL? Jul 29 05:54:23 sorry for all the acronyms... Jul 29 05:54:31 Intel Access Library. Jul 29 05:54:34 no probs i weild them myself as a programmer ;) Jul 29 05:54:38 ok Jul 29 05:54:48 what was BSR? Jul 29 05:54:50 bit speed rate? Jul 29 05:54:56 Bit Shift Register Jul 29 05:54:59 ah ok. Jul 29 05:55:01 basically the packet size. Jul 29 05:55:15 so anytime it needs to do anything; it needs to shift 498 bits. Jul 29 05:56:31 big register. Jul 29 05:58:01 its Effing Huge actually.... Jul 29 05:59:49 i suppose it' Jul 29 06:00:04 s easier to implement if wider. Jul 29 06:00:28 number operations are minimised. Jul 29 06:01:06 nah, its a ploy by JTAG equipment manufacturers to make you buy their stuff because its faster than a parallel port for large BSR's. Jul 29 06:01:09 anyway, apex has a special flagged 1.0.11 version for the ixp425.. should i use that. Jul 29 06:01:19 um Jul 29 06:01:21 dyoung: bastards. Jul 29 06:02:11 thats old! Jul 29 06:02:17 http://wiki.buici.com/twiki/bin/view/Main/ApexBootloader Jul 29 06:02:40 latest I got is 1.2.19 Jul 29 06:03:00 it's in OE now Jul 29 06:03:09 bb apex Jul 29 06:03:35 1.2.19 Jul 29 06:03:37 oh yeah I keep forgetting baou that. Jul 29 06:03:41 about that. Jul 29 06:03:50 but that link is a good read anyways for howto info.. Jul 29 06:04:20 landslide: beewoolie is the man in charge of Apex. Hes in here sometimes, or sometimes in #openjtag. Jul 29 06:08:32 ok thanks! Jul 29 06:16:23 bb apex works now? Jul 29 06:16:42 good. That will bring me one step closer to trying it. Jul 29 06:17:25 VoodooZ_Work: bb apex builds. I make no guarantee about it working. Jul 29 06:17:38 * VoodooZ_Work nods Jul 29 06:17:50 don't worry, I'm not flashing it quite yet! Jul 29 06:18:19 but does apex come with ethernet connectivity or not? Jul 29 06:18:50 not yet Jul 29 06:18:59 ... Jul 29 06:19:04 it doesnt? Jul 29 06:20:00 I thought the beewoolie was just finishing up tftp in the last couple of days Jul 29 06:20:25 so it's close, if not there already Jul 29 06:20:33 Last time I popped in there I thought I saw it was working? Jul 29 06:20:50 it's not in 1.2.19 Jul 29 06:21:02 Nope; that sfor sure. Jul 29 06:21:07 "if it's not in the repo, then it doesn't exist" :-) Jul 29 06:21:14 okay fine. ;-) Jul 29 06:21:20 No network for you! Jul 29 06:22:59 well, I guess I'll wait then as no network is no good for me. Jul 29 06:30:41 that monotone makefile is it pulling down EVERYTHING i'll ever need? Jul 29 06:30:48 does it do incremental updates? Jul 29 06:31:10 landslide: what do you mean? Jul 29 06:31:32 there is no monotone makefile - do you mean the "master makefile" ? Jul 29 06:32:57 yeah that's it. Jul 29 06:33:10 it uses monotone... i didn't explain myself right. Jul 29 06:33:45 in that case, yes and yes. Jul 29 06:34:07 if you're running debian, then make setup-host-debian will be useful too Jul 29 06:35:24 ok thanks. Jul 29 06:35:36 what does the debian one do? Jul 29 06:36:02 make sure you have all the required tools for building stuff Jul 29 06:46:46 it seems to have stopped here. Jul 29 06:46:47 NOTE: package libtool-native-1.5.10-r3: task do_configure: started Jul 29 07:06:36 I'm trying to understand the build process of make build-openslug. Does it automatically package each kernel module on its own, once it's built, or is there a module-package script somewhere, as it might be clever to package some modules in groups... Jul 29 07:07:06 each on it's own. Look in openembedded/classes/... Jul 29 07:07:40 ah tzanks Jul 29 07:07:42 thanks Jul 29 07:12:02 it's ok just taking ages to build. Jul 29 07:12:16 oops, the build aborted at this patch http://pastebin.ca/18779 Jul 29 07:20:20 the indenting differs, that's why the hunk fails Jul 29 07:32:30 okay, I fixed that patch, but if I overwrite the old patch ./OpenSlug-head/openslug/tmp/work/ixp425-eth-1.1-r5/patches/ethhdr.patch with my fixed one. It will still use the broken patch, once I restart make build-openslug. Where is the right place to put my fixed patch in? This makefile isn't very verbose... Jul 29 07:33:34 the fixed patch can be found here: http://home.fhtw-berlin.de/~s0502837/ethhdr.patch Jul 29 07:34:01 malfi: you need to read up on the OpenEmbedded build system, which we use. we didn't write it, just the makefile that we wrap around it Jul 29 07:34:48 I'm surprised it failed - we've built that package hundreds of times Jul 29 07:35:55 which package? I just did a full build without error. Jul 29 07:36:41 malfi got a patch failure on ixp425_eth Jul 29 07:37:00 strange. mine built fine. Could be his setup. Jul 29 07:37:17 mine builds fine, and it also builds fine on the official build machine Jul 29 07:37:42 $ patch --version Jul 29 07:37:42 patch 2.5.4 Jul 29 07:38:37 there is definitly an identing difference Jul 29 07:40:49 indenting... Jul 29 08:23:56 strange, I upgraded to patch 2.5.9 but that doesn't help Jul 29 08:39:27 My FC4 machine uses patch 2.5.4 Jul 29 08:39:37 So it has to be something else. Jul 29 09:14:42 malfi: Still there? Jul 29 09:14:47 * malfi nods. Jul 29 09:15:55 malfi: Have a look: Jul 29 09:16:07 FWIW, everything builds fine to the end, if I exclude this ixp425_eth package, so my setup can't be that broken. Jul 29 09:16:11 root@T_192168115121#/mnt: mkswap swapfile_121 Jul 29 09:16:11 Setting up swapspace version 1, size = 134213 kB Jul 29 09:16:11 root@T_192168115121#/mnt: losetup /dev/loop0 /mnt/swapfile_121 Jul 29 09:16:11 root@T_192168115121#/mnt: swapon /dev/loop0 Jul 29 09:16:11 swapon: /dev/loop0: Function not implemented Jul 29 09:16:33 Do i need to enable something in the kernel or why doesn't that work? Jul 29 09:17:27 what does losetup /dev/loop0 say? Jul 29 09:18:50 /dev/loop0: [0012]:14 (/mnt/swapfile_121) Jul 29 09:20:23 try a losetup /dev/loop1 /mnt/swapfile, and then swapon /dev/loop1, just to be sure that you're using a fresh loop device Jul 29 09:22:30 Same Jul 29 09:22:33 :-/ Jul 29 09:23:02 what happens, if you do a mkswap /dev/loop1 ? Jul 29 09:23:37 Setting up swapspace version 1, size = 134213 kB Jul 29 09:24:48 strace swapon /dev/loop1 Jul 29 09:26:24 Well, you wouldn't belive me :) Jul 29 09:26:50 Because the cd normal hasn't a swap i deactived swap support in the kernel... Jul 29 09:26:52 omg Jul 29 09:27:36 Well, it's fine, that tomorrow begins weekend... Jul 29 09:27:43 thanks for your hints Jul 29 09:29:49 :) Jul 29 09:30:03 :D Jul 29 09:49:36 <[g2]> rwhitby-asleep, FYI .54 A @ 5v with flash stick Jul 29 09:50:00 <[g2]> that's a turbo Jul 29 09:50:33 <[g2]> doh... that's the non-turbo Jul 29 09:52:26 I cannot find any problem in my setup. I pulled all files freshly with monotone but 'make build-openslug' always fails at one patch of ixp425_eth. I tried this patch on like 10 machines and it always fails. http://pastebin.ca/18789 Jul 29 09:53:15 download failure? Jul 29 09:54:30 what about bb openslug-image? Jul 29 09:55:10 <[g2]> malfi, what's the error at line 1403 ? Jul 29 09:58:00 NAiL: I did a monotone co -b org.openembedded.nslu2-linux openembedded so that this ethhdr.patch was checked out again, but it's the same Jul 29 09:58:51 VoodooZ_Work: I'll try that, even though I don't really understand, what difference that makes to the master-makefile Jul 29 09:59:21 [g2]: well, what do you mean with error? Jul 29 09:59:49 <[g2]> malfi, the patch file lines should match up with the source file Jul 29 09:59:57 <[g2]> apparently they don't Jul 29 10:00:08 <[g2]> the did for the first part of the patch (Hunk #1) Jul 29 10:00:08 malfi: delete the .c file from downloads and retry? Jul 29 10:00:13 <[g2]> but not the second Jul 29 10:00:19 <[g2]> at line 1403 in the the file Jul 29 10:00:19 NAiL: I did that Jul 29 10:01:04 [g2]: they said that this patch file is working for quite some time and it just worked for VoodooZ... Jul 29 10:01:55 <[g2]> malfi, it's worked for probably hundreds of ppl, however it's failing for you and I'm trying to help you find out why and how to fix stuff in the future Jul 29 10:02:51 [g2]: okay, I really appreciate that. I could be a download failure in ether the patch or the sourcefile. Jul 29 10:03:16 <[g2]> md5sum the patch and the tarball Jul 29 10:03:26 <[g2]> look in downloads for the source tarball Jul 29 10:03:57 <[g2]> look in oe-symlinks/packages/ixp4xx-csr/... something for the patch Jul 29 10:04:25 <[g2]> or it might be in the packages/linux/openslug-kernel-2.6.11.2/ dir Jul 29 10:04:26 $ md5sum downloads/ixp400LinuxEthernetDriverPatch-1_1.zip ccd8bc1b53c083764b50b895f757178e Jul 29 10:04:41 $ md5sum openembedded/packages/ixp425-eth/files/ethhdr.patch 9d577de2ea472fb3363188fb9925754b Jul 29 10:04:58 ccd8bc1b53c083764b50b895f757178e Jul 29 10:05:12 9d577de2ea472fb3363188fb9925754b Jul 29 10:05:16 looks the same to me Jul 29 10:05:49 <[g2]> ditto here Jul 29 10:06:19 <[g2]> which version of patch ? Jul 29 10:06:37 <[g2]> and which OS ? Jul 29 10:06:44 <[g2]> or distro Jul 29 10:07:01 if you extract that ixp400LinuxEthernetDriverPatch-1_1.zip manually and try to apply that patch to the ixp425_eth.c, the problem also occurs Jul 29 10:07:18 slackware 10.1, just upgraded patch from 2.5.4 to 2.5.9 Jul 29 10:07:20 <[g2]> @ line 1403 ? Jul 29 10:07:50 no, at 1395, because some other patches aren't applied in that archive Jul 29 10:08:27 <[g2]> nod. Jul 29 10:08:59 <[g2]> you should be able to see how far off the lines are Jul 29 10:11:20 I'm looking at the source and I don't see a good reason, why this hunk fails at this point. Jul 29 10:11:39 <[g2]> tabbing ? Jul 29 10:15:10 you're right, in the source file it's indented with tabs, in the patch with spaces Jul 29 10:15:49 but why does only my patch complain about that and not yours? Jul 29 10:22:28 <[g2]> I'd guess it's somehow a difference with slack Jul 29 10:22:34 <[g2]> I dunno how Jul 29 10:23:58 if you extract that file from this archive and try to apply this patch, it works? Jul 29 10:26:50 <[g2]> malfi, I'm the original openslug developer I've been running this stuff for 10 months+ -- I've probably extracted and build 500-1000 timse Jul 29 10:26:59 <[g2]> maybe more Jul 29 10:27:38 * malfi nods and sighs. Jul 29 10:28:21 <[g2]> it is very curious how stuff works with one distro and not another and getting to the root of the issue Jul 29 10:28:43 <[g2]> in your case it appears to be related to the tabbing/spacing Jul 29 10:28:58 <[g2]> now *why* that occurs is a big mystery to me Jul 29 10:29:34 doesn't patch have a --ignore-whitespace? Jul 29 10:30:42 yeeeeeeees, that did the trick Jul 29 10:31:18 * malfi cheers. Jul 29 10:32:16 obviously your patch do -l by default. Jul 29 10:33:01 <[g2]> my understanding was that OE built all it's own tools and used them so I don't know why your patch would be any different from my patch Jul 29 10:34:14 yes, aren't we supposed to be using quilt? Unless old packages still refer to patch. Jul 29 10:34:41 unless you do ASSUME_PROVIDED on a bunch of stuff Jul 29 10:37:52 I'll search the invocation of patch and add a "-l" or is there any other recommanded way? Jul 29 10:41:36 [g2]: actually OE doesn't build patch Jul 29 10:43:14 there is a perl script in openslug/tmp/staging/i686-linux/bin/patcher, which does $res = system("patch... Jul 29 10:46:02 malfi: Filesystem 1k-blocks Used Available Use% Mounted on Jul 29 10:46:06 /dev/mtdblock4 6528 4820 1708 74% / Jul 29 10:46:27 nice Jul 29 10:46:35 That's got extra stuff. Jul 29 10:47:56 Ok... that has (in the image), ntp, the ext2 file system programs and the reiserfs programs. Jul 29 10:50:10 malfi: set TARGET_OS_local="linux-uclibc" and OPENSLUG_EXTRA_RDEPENDS = "" Jul 29 10:50:19 * malfi notes that down. Jul 29 10:50:38 The result will not have any way to mkfs a disk file system, but you can mount one that is already mkfs'ed Jul 29 10:51:32 Then go in take a look at OPENSLUG_EXTRA_FILESYSTEMS in conf/distro/openslug.conf - you can strip all the modules you don't need... Jul 29 10:52:03 (I.e. the system doesn't need any of those to boot - just to mount external file systems). Jul 29 10:52:45 okay Jul 29 10:52:48 Some time in the near future I will write 'ucslugc.conf' which will be uclibc with absolutely nothing in the default image. Jul 29 10:53:04 (Except the stuff needed to boot) Jul 29 10:54:17 I'll also tidy up openslug.conf so that it is obvious what 'openslug' really is - because the default local.conf zaps OPENSLUG_EXTRA_RDEPENDS openslug.conf is misleading. Jul 29 10:55:08 I think it's possible to get to about 2MByte free on the flash. Jul 29 10:57:33 I know a way of squeezing another 400kByte (approx) out by putting the ethernet driver modules 'somewhere else', but I'm not sure that this is worth doing. Jul 29 11:02:02 <[g2]> malfi cool.. now we have our answer and we can all sleep better at night Jul 29 11:04:21 malfi: the ixp425-eth patching problem happens because of the 'do_ixp425_c_patch_fetch' stuff in the .bb file Jul 29 11:05:57 Check the checksum for ${DL_DIR}/ixp425_eth.c.patch: Jul 29 11:06:01 6257dce29cb2cbc9c65eff7356c0d365 ixp425_eth.c.patch Jul 29 11:06:09 After I modified all invocations of /usr/bin/patch to /usr/bin/patch -l in openslug/tmp/staging/i686-linux/share/quilt/push the ethhdr.patch goes through fine Jul 29 11:06:27 Well, yes, it would - there's no new line at the end of that file. Jul 29 11:06:39 oops Jul 29 11:06:41 So what is the md5sum of that file on your system? Jul 29 11:06:44 4ec0f7f7edd3bf35263ed09ad5e7db57 downloads/ixp425_eth.c.patch Jul 29 11:07:17 And how many bytes? Mine has 43320 Jul 29 11:07:33 43319 Jul 29 11:07:47 so I delete that file and download it again Jul 29 11:07:55 <[g2]> and the .md5 Jul 29 11:08:02 <[g2]> or .md5sum Jul 29 11:08:06 You can, but it depends on which sourceforge mirror you get Jul 29 11:08:33 I feel kinda stupid now, why are they two different versions of that file floating around? Jul 29 11:08:54 I don't think there are, I think sourceforge has one with no new line and some mirrors helpful append a newline. Jul 29 11:09:16 This happens to someone every couple of months.... Jul 29 11:09:36 Try echo >>downloads/ixp425_eth.c.patch Jul 29 11:09:47 (i.e. append a new line), then redo the md5sum Jul 29 11:10:08 oh, I just deleted it Jul 29 11:10:54 okay Jul 29 11:10:58 dd if=ixp425_eth.c.patch of=test bs=1 count=43319 Jul 29 11:11:07 4ec0f7f7edd3bf35263ed09ad5e7db57 test Jul 29 11:11:19 aha Jul 29 11:11:20 after that echo >> it turned from 4ec0f7f7edd3bf35263ed09ad5e7db57 to 6257dce29cb2cbc9c65eff7356c0d365 Jul 29 11:11:37 * NAiL__ sees brilliance at work ;) Jul 29 11:11:50 malfi: Yep. Can you enter a slugbug please? Jul 29 11:12:34 * [g2] starts a a list of slugbugs as a jbowler-away x-mas present :) Jul 29 11:12:37 NAiL: it took me about 4 hours to track it down last time, this time, fortunately, I was able to remember enough. Jul 29 11:13:08 <[g2]> IMHO that's *really* bad if the mirrors are doing that Jul 29 11:13:11 4hours seem like a fair deal, it took me 5h by now und I was not even close to the solution ;) Jul 29 11:13:12 [g2]: these things keep happening, like the postfix/pcre stuff. We know the solution but there is no documentation of it... Jul 29 11:13:34 <[g2]> what's the postfix/pcre issue ? Jul 29 11:13:44 I don't know, I can't remember. Jul 29 11:13:58 <[g2]> I've been trying to play with postfix recently Jul 29 11:14:15 <[g2]> I think it wants a MX record for my domain Jul 29 11:14:57 That's ok, so long as you have a pre-nuptial agreement. Jul 29 11:15:29 * malfi cheers for jbowler. Jul 29 11:17:25 should I file a slugbug under OpenSlug Firmware or OpenSlug Packages? Jul 29 11:20:53 <[g2]> ~praise jbowler-away Jul 29 11:22:40 <[g2]> jbowler-away, is there a way t psuedo turnup ? I've got a different slug that's reflashed and setup and I want to boot onto an external flash drive, but I don't really want to actually turnup Jul 29 11:26:33 malfi: openslug firmware Jul 29 11:27:24 [g2]: just do what turnup does, it's just a matter of making a bootable file system turnup doesn't do much and it's a shell script. Jul 29 11:28:03 <[g2]> after I get back from holiday I'll add a "pretend" option Jul 29 11:28:09 Once you have a bootable system turnup memstick... will make the correct linuxrc (but you have to do everything that turnup does - or write your own linuxrc) Jul 29 11:28:12 <[g2]> psuedo pretend Jul 29 11:28:58 <[g2]> I just "turned-up" to a different flash stick :) Jul 29 11:30:54 <[g2]> BTW these are *really* low power devices :) Jul 29 11:31:04 <[g2]> I'm measuring power on the 5V side Jul 29 11:31:26 [g2], quick question: if I have ext2 formatted flashdisk with data on it and I do turnup memstick, is it going to erase the data? Jul 29 11:32:08 I've never played with it and I might start. Jul 29 11:32:52 [g2], how low-power are we talking here? Jul 29 11:33:08 <[g2]> VoodooZ_Work, a bunch of stuff gets copied over so it depends on what you have on the disk Jul 29 11:33:12 <[g2]> memstick Jul 29 11:33:37 ok, but is it destructive in any way like a format? Jul 29 11:33:46 I only have files Jul 29 11:34:58 <[g2]> Ok... an idling slug with a memstick 256MB a ssh session clocks in at .54 and .59A for non-turbo and turbo (same physical memstick, different slugs) Jul 29 11:35:06 <[g2]> on the 5V side Jul 29 11:35:21 that's very nice Jul 29 11:36:09 <[g2]> or 2.7 and 2.95W Jul 29 11:38:57 <[g2]> given that I measure around 8W before about 2/3 of the power is just in the brick Jul 29 11:39:24 <[g2]> 8W on the 120 side Jul 29 11:44:47 yeah, that's about what I got on my roboslug. I got ~1amp with a huge sensor array Jul 29 11:48:33 that sounds great: roboslug Jul 29 11:51:06 any pictures of that roboslug around? :) Jul 29 11:51:44 <[g2]> VoodooZ_Work, 1A on the 5V side ? Jul 29 11:54:27 <[g2]> it's kinda amazing to me that you could run 5 turbo slugs with flash sticks for the price of 1 15W bulb :) Jul 29 11:54:31 sorry, I had somebody at my desk... Jul 29 11:54:41 http://robotics.no-ip.org Jul 29 11:54:56 Not much on that page page but check out SRX2 under My Robots Jul 29 11:55:45 [g2], yeah, on the battery pack side with the regulation loss. Jul 29 11:56:01 I'm more worried about total demand as batteries are not perfect. Jul 29 11:56:40 <[g2]> my units will runs 24V which will be nice Jul 29 11:57:44 24V, off batteries? Jul 29 11:58:03 <[g2]> batteries, wall, POE Jul 29 11:58:11 <[g2]> 5-24V+ Jul 29 11:58:20 i see. Jul 29 12:06:12 <[g2]> Ok... so the turbo slug is clock idling around .45A on the 5V with no flash stick attached Jul 29 16:51:28 They draw a explicit line between natively compiled and cross compiled packages. Does the cross compiler binary quality differ that much? Jul 29 17:16:59 The cross and native compilers are identical. Jul 29 17:17:32 (And the libraries are identical too). Jul 29 17:17:41 why are there two feed at least for unslung? Jul 29 17:20:34 For unslung and openslug both some packages can't be cross-compiled (or don't work if they are), so there is one cross feed and one native (that's a convenience I believe, since the builds of packages for those feeds necessarily happen on different machines). Jul 29 17:21:03 Oh, I see. Jul 29 17:22:45 I searched for http://ipkgfind.nslu2-linux.org/details.php?package=asterisk and this is a cvs-asterisk, while the asterisk package in openembedded is fairly old (but the latest release). The packager of this cvs-asterisk probably used some patches and a bitbake-file to build this package. Are those bitbake-makefiles stored anywhere? Jul 29 17:23:38 If it's an unslung package the master makefile will build it in the 'optware' build - everything will be there. Jul 29 17:24:21 Oh I didn't think about that, I will have a look. Thank you. :) Jul 29 17:24:54 It's probably easier to build a CVS asterisk from OE - indeed it is probably just a matter of renaming the .bb file and editing the SRC_URI line. Jul 29 17:26:29 I don't think so, because most of the codecs are heavily depending of floating point aritmetics and so are useless on the slug. They shouldn't be included, but if I think about it, they probably are, because space is a non-issue as I experienced... Jul 29 17:30:31 The original (Acorn) ARM fp emulator was pretty good, I don't know what the current ones are like though (the ARM FP instruction set has particularly good rounding handling, but it's difficult to exploit from C). Jul 29 19:55:35 ucslugc.conf - nice name jbowler-away :-) Jul 29 19:56:41 Re the ixp425_eth.c patch - we shoul just mirror that ourselves, and download it from our mirror - then we can be sure everyone gets the correct file contents. Jul 29 20:16:23 rwhitby-away: then we may as well simply check it in to the OE source tree - I don't think the license prevents us from doing that. Jul 29 20:17:01 It would simplify the .bb file - it current does an explicit wget and patch. Jul 29 20:28:35 jbowler-away: OE has previously said they don't want to enlarge the tree content size with patches. Jul 29 20:29:27 jbowler-away: it's now at http://nslu.sourceforge.net/downloads/ixp425_eth.c.patch Jul 29 20:34:42 Ok, I'll try a build with that URL and without the explict wget/patch - I can't see why this wouldn't work. Jul 29 20:36:31 Hum... maybe I do - the pre_patch stuff Jul 29 20:36:31 ok, back later (saturday arvo here) Jul 29 20:38:05 jbowler-away: just list it as a normal file, and not a patch, in the SRC_URI Jul 29 20:58:35 Ok, that works. **** ENDING LOGGING AT Fri Jul 29 23:59:56 2005