**** BEGIN LOGGING AT Sun Aug 07 23:59:56 2005 **** BEGIN LOGGING AT Mon Aug 08 00:57:20 2005 **** BEGIN LOGGING AT Mon Aug 08 01:05:35 2005 **** BEGIN LOGGING AT Mon Aug 08 01:28:04 2005 Aug 08 02:45:54 meh Aug 08 02:49:14 Gah Aug 08 03:04:05 almost there with the 128m slug. Aug 08 03:04:22 i can get apex to boot with the 'go' command but can't get it boot with the 'boot' command Aug 08 03:04:23 nice Aug 08 03:04:51 i was wondering, where in the kernel do they store the MEM=32M@0x000000 default command line? Aug 08 03:05:00 i grepped it and found a squillion instances. Aug 08 03:05:26 * koen starts working on his soldering skills Aug 08 03:10:24 landstide: I can get it to boot with boot mem=32M@0x00000000 console=ttyS0,115200 root=/dev/mtdblock4 rw rootfstype=jffs2 init=/linuxrc , but with 64Mb it panics Aug 08 03:17:29 landslide, openslug.conf iirc. Aug 08 03:21:45 DaKa2: at what point does it panic? Can you pastebin it? Aug 08 03:25:32 dyoung: "Kernel panic - not syncing: No init found. Try passing init= option to kernel.", but there is a init=, Im using the same boot line as with 32M, could probably just be me doing something wrong, or that the ram I soldered on is bad Aug 08 03:27:16 oh right, forgot, apex reports 32Mb in version, but if you look at 0xcc000000 its set to the right 4x8Mx16 cas3 Aug 08 03:31:55 * NAiL goes off to sleep a bit Aug 08 03:32:39 so the kernel actually boots. The problem you described it resolvable. Aug 08 03:38:01 ah.. Aug 08 03:38:14 u using apex? Aug 08 03:38:27 not recently. Aug 08 03:38:50 i've been hacking into apex trying to get it to boot Aug 08 03:38:52 but I was involved in early debugging. Aug 08 03:39:07 i've got apex booting though i can't pass atags.. Aug 08 03:39:12 dyoung: :) how to solve it? :-) Aug 08 03:39:18 it seems the build_atags() function is not even compiled in. Aug 08 03:39:25 i have no idea how it links. Aug 08 03:39:43 my apex works great with 32Mb Aug 08 03:40:07 DaKa2: take out the mem= line. [g2] took his out and it automagically discovered 64MB Aug 08 03:40:27 for starters. Aug 08 03:40:32 daka: are you using the boot command for apex? Aug 08 03:40:35 dyoung: it finds 32MB if I dont use mem:.. I think... Aug 08 03:40:40 landstide: yes Aug 08 03:40:46 have to check. Aug 08 03:40:58 then, if you can patsebin your bootlog that may help. Aug 08 03:41:32 ok, will do Aug 08 03:42:00 daka2: can you also paste your apex configuration? Aug 08 03:42:11 landstide: ok, in a minute Aug 08 03:42:19 thanks :) we can compare notes. Aug 08 03:45:53 dyoung, landstide: http://pastebin.com/331871 Aug 08 03:46:42 I don't like that apex says I have 32MiB memory.. **** BEGIN LOGGING AT Mon Aug 08 03:47:50 2005 Aug 08 03:53:34 postinst script returned status 127 Aug 08 03:53:34 ERROR: busybox.postinst returned 127 Aug 08 03:53:46 "//usr/lib/ipkg/info/busybox.postinst: 14: update-rc.d: not found" Aug 08 04:16:26 ps -ef Aug 08 04:17:04 daka: which compiler did you build apex with? Aug 08 04:17:26 test Aug 08 04:17:35 Package perl-module-xsloader wants to install file /usr/lib/perl5/5.8.7/armeb-linux/XSLoader.pm Aug 08 04:17:36 But that file is already provided by package perl-module-armeb-linux-xsloader Aug 08 04:17:40 and so on ... Aug 08 04:18:58 landslide: crosstools gcc-3.4.0-glibc-2.3.2 armv5b-softfloat-linux Aug 08 04:19:42 ahhuh. Aug 08 04:19:44 rwhitby: yes, just noticed that.. its a bit hard to fix if ipkg doesn't use replaces and provides... Aug 08 04:20:22 i'm using crosstools gcc-3.3.4-glibc-2.2.5 Aug 08 04:20:54 rwhitby: landslide beewoolie said something about 3.4.4 and not working in the same sentence Aug 08 04:20:58 argh.. Aug 08 04:20:58 DaKa2: yeah, just working out now how to ipkg remove all the perl-module packages :-) Aug 08 04:21:05 yeah i saw that too Aug 08 04:21:15 is it hard to grab the build your running with? Aug 08 04:21:16 skip the rwhitby part on the last one :-) Aug 08 04:21:23 ipkg remove perl-module-* ? :) Aug 08 04:21:40 ipkg list | grep perl | awk '{print $1}' | xargs ipkg remove Aug 08 04:21:52 rwhitby: ipkg list_installed | grep perl-module-armeb-linux | xargs ipkg remove --force-depends Aug 08 04:22:01 regexp could be something like 'pkgname*' '*file*' or similar Aug 08 04:22:01 eg: ipkg info 'libstd*' or ipkg search '*libop*' or ipkg remove 'libncur*' Aug 08 04:22:21 right, a cut -f1 -d" " in there too Aug 08 04:23:40 ipkg remove 'perl-module-armeb*' then.. Aug 08 04:23:47 test Aug 08 04:24:10 landslide: grab build, running with? Aug 08 04:24:39 sudo ipkg remove --force-depends perl-module-armeb-linux-* Aug 08 04:24:52 ah, yes Aug 08 04:25:04 DaKa2: what do you think about renaming openslug-native to openslug-devel ? Aug 08 04:25:29 daka: I used http://www.nslu2-linux.org/wiki/HowTo/CompileCrossTool to build my cross tool. Aug 08 04:25:53 landslide: what do i do to get the crosstool chain you are working with? Aug 08 04:26:01 hehe. Aug 08 04:26:27 rwhitby: well, that works Aug 08 04:27:39 landslide: well, download latest crosstols, modify demo-armv5b-softfloat.sh to use gcc-3.4.0-glibc-2.3.2.dat, run demo-armv5b-softfloat.sh Aug 08 04:28:26 oh that's simple ;) thanks! Aug 08 04:28:52 "//usr/lib/ipkg/info/busybox.postinst: 14: update-rc.d: not found" Aug 08 04:29:02 "//usr/lib/ipkg/info/libperl5.postinst: 4: ldconfig: not found" Aug 08 04:29:11 "//usr/lib/ipkg/info/openssh-sshd.postinst: 7: update-rc.d: not found" Aug 08 04:29:24 rwhitby: path issue, depends issue? Aug 08 04:29:41 koen: dunno yet, haven't debugged them Aug 08 04:36:55 rwhitby: you have sbin in your path? Aug 08 04:37:04 ie, 'su -' not 'su' Aug 08 04:37:21 DaKa2: looks like "sudo ipkg install ..." doesn't work :-( Aug 08 04:37:26 (due to the path) Aug 08 04:37:53 rwhitby: can't you do something like 'su -' with sudo? Aug 08 04:38:26 dunno Aug 08 04:39:42 sudo -i whatevercmd should work Aug 08 04:40:06 -i The -i (simulate initial login) option runs the shell specified in the passwd(5) entry of the user that the command is being run as. The comâ Aug 08 04:40:09 mand name argument given to the shell begins with a - to tell the shell to run as a login shell. sudo attempts to change to that userâs home Aug 08 04:40:12 directory before running the shell. It also initializes the environment, leaving TERM unchanged, setting HOME, SHELL, USER, LOGNAME, and PATH, Aug 08 04:40:17 * rwhitby feels an 'alias sudo="sudo -i"' in his future ... Aug 08 04:40:29 That seems like a very good idea :) Aug 08 04:43:47 hmm Aug 08 04:43:59 I wonder if I'm going to try upgrading 2.3-beta to 2.5-beta now Aug 08 04:44:13 the 2.5-beta feed is finished pushing, so it's the way it's supposed to be Aug 08 04:44:22 I upgraded to unstable :-) Aug 08 04:44:36 from 2.3? Aug 08 04:44:39 or 2.5? Aug 08 04:44:52 2.4-ish Aug 08 04:45:04 ok, because current unstable has new kernel modules Aug 08 04:45:10 yep Aug 08 04:45:18 -r7 breaks compatibility with everything before :\ Aug 08 04:45:30 reflash handles that fine Aug 08 04:45:35 ah ;) Aug 08 04:45:49 jbowler-zzz is a genius Aug 08 04:46:12 ~hail jbowler Aug 08 04:46:17 ALL HAIL jbowler ;) Aug 08 04:46:48 * NAiL notices he somehow flicked capslock twice Aug 08 04:47:38 Ok, I'm doing an ipkg upgrade to the 2.5-beta feed now Aug 08 04:48:37 Hopefully, that is all that is required to upgrade from 2.3-beta to 2.5-beta Aug 08 04:51:38 now, the perl stuff looks like it's ok. Is there anything I should be aware of? Aug 08 04:52:58 well, the perl-module-* internal rdepends are still not ok, but I'll fix that later today hopfully Aug 08 04:53:11 test Aug 08 04:53:23 poor slugbot ... Aug 08 04:53:40 DaKa2: Ok, was thinking of sending a "how to upgrade to 2.5-beta" mail on the ml Aug 08 04:53:54 * dyoung kill -9 slugbot Aug 08 04:55:01 Is there an official binary out yet? Aug 08 04:56:15 test Aug 08 04:59:17 test Aug 08 05:00:18 test Aug 08 05:02:41 Okay I give up. Aug 08 05:02:51 I'll need to look at it tommrow Aug 08 05:09:35 * NAiL waits for ingeba-zzz____________ Aug 08 05:10:21 foo Aug 08 05:10:25 test Aug 08 05:10:30 test Aug 08 05:10:37 ~lart nslu2bot Aug 08 05:10:44 * NAiL beats nslu2bot Aug 08 05:10:55 into a bloody pulp Aug 08 05:11:16 Okay, I see the messages being sent to loglibrary in the log Aug 08 05:11:23 so I'm going to assume that this isnt my problem. Aug 08 05:12:32 foo Aug 08 05:12:37 bar Aug 08 05:12:41 DaKa2: the unsatisfied recommendations are still nothing to worry about? Aug 08 05:12:42 baz Aug 08 05:12:52 qux Aug 08 05:13:08 Okay folks this isnt my fault. Aug 08 05:13:24 I see the stuff being sent. Aug 08 05:13:27 bat Aug 08 05:15:03 NAiL: nope, but Im about to fix that.. Aug 08 05:15:17 DaKa2: ok, just checking ;) Aug 08 05:15:18 daka: how did you go with your fat slug? Aug 08 05:15:43 ipkg-utils requires python, which ipkg doesn't find by default Aug 08 05:15:49 Collected errors: Aug 08 05:15:49 ERROR: Cannot satisfy the following dependencies for ipkg-utils: Aug 08 05:15:49 python Aug 08 05:16:11 NAiL: ipkg-utils doesn't need to be installed for openslug-native anymore.. but yes, that should probably be fixed.. Aug 08 05:16:26 ah, yes, it's compiled natively Aug 08 05:16:27 forgot Aug 08 05:17:23 * NAiL continues to upgrade Aug 08 05:17:41 * dyoung grumpy Aug 08 05:17:43 landslide: my added two 8M16 seems to work with fill/dump in flashed apex, but it says it's 32Mb in version, and the kernel fails.. with mem=64M Aug 08 05:18:17 root@maimedslug:/var/log# ipkg update && ipkg upgrade Aug 08 05:18:20 Nothing to be done Aug 08 05:18:24 :) Aug 08 05:18:40 landslide: have you finished the crosstools and made a working apex yet? Aug 08 05:18:46 So, now my slug is as close to 2.5-beta I can get without actually flashing Aug 08 05:18:53 landslide: I really want to see it your 128Mb slug works Aug 08 05:20:47 DaKa2: Ok, when you've done your next magic to perl, will it be necessary for users to do anything prior to ipkg upgrade? Or will ipkg handle everything just fine? Aug 08 05:21:16 there are some things I love about my job (i.e. everything that involvs the serverroom), and there are soem things I hate (for example, troubleshooting why an activex plugin doesn't work in Internet Explorer...) Aug 08 05:21:50 troubleshooting anything IE sucks Aug 08 05:21:57 NAiL: well, the name change still gives problems (removed armeb-linux-) Aug 08 05:22:49 DaKa2: so the exact commands the users should use before ipkg upgrade is "ipkg remove --force-depends perl-module-armeb-linux-*" ? Aug 08 05:23:13 NAiL: yes.. think so... Aug 08 05:23:16 * NAiL doesn't have *anything* left of < -r13 after ipkg upgrade. (From -r9) Aug 08 05:25:53 DaKa2: The perl-module-armeb-linux-* isn't supposed to be there at all? Aug 08 05:26:16 I don't think I've quite followed the discussion closely enough Aug 08 05:26:53 There's 41 of those modules installed on my system after upgrading to -r13 Aug 08 05:28:07 aha... -r14 is the latest in -HEAD Aug 08 05:28:12 just ignore me ;) Aug 08 05:29:46 :) Aug 08 05:30:00 I always do Aug 08 05:30:06 heey! :( Aug 08 05:30:12 nahh :-) Aug 08 05:30:16 I'll wait with propagating whatever changes you have in store until it's been tested a bit on unstable Aug 08 05:30:57 Well, the changes will maybe come sometime tonight.. I'm at work now, trying to actually work Aug 08 05:31:01 :-) Aug 08 05:31:11 No hurry ;-) Aug 08 05:31:27 I really want it fixed... Aug 08 05:31:51 Yeah, do it when you're back home ;) Aug 08 05:31:51 and I really want my fatslug to actually use the 64Mb Aug 08 05:33:30 Aww, what the eff!?! Aug 08 05:33:47 how many minutes did I waste? Aug 08 05:33:50 "Loglibrary.com will be up and down over the next 24 hours due to a major server upgrade. " Aug 08 05:36:38 The upgrade instructions are rather simple for 2.3-beta to 2.5-beta :) Aug 08 05:37:50 change the feed address, remove ipkg-utils and some perl modules, ipkg update && upgrade, ipkg install ntp-tickadj if you want "accurate" clock. Aug 08 05:41:24 kolla_: back in Norway? Aug 08 05:43:57 * NAiL wants an ipkg dist-upgrade :P Aug 08 05:51:18 NAiL: yap Aug 08 05:52:32 sigh.. damn "posten norge as" Aug 08 05:52:39 and tollen Aug 08 05:52:45 oh yes Aug 08 05:52:56 my package hasn't arrived from Hawaii yet Aug 08 05:53:06 I'm awaiting a parkage from .. how ironic.. sweden Aug 08 05:53:11 haha Aug 08 05:53:17 I should have brought it with me myself Aug 08 05:54:01 anyways, they send me two identical letters about this packge, they descrive it as "rekommandert brev fra sverige" and "vekt: 1" - whatever that means Aug 08 05:54:10 * NAiL restarts his slug to see if the "dist-upgrade" was successful Aug 08 05:54:59 it's a replacemtn case for the nintendo gamecube, I dont think it fits in a "brev" Aug 08 05:55:21 haha Aug 08 05:55:22 no Aug 08 05:55:29 yay Aug 08 05:55:55 The 2.3-beta -> "pseudo-2.5-beta" upgrade worked like a charm Aug 08 05:56:06 modchip and modcase for GC, so now I'll have something to watch movies with even when I hack down the xbox :) Aug 08 06:31:34 anyone know much about atags? Aug 08 06:46:46 i've traced my problems with apex and openslug.. the kernel doesn't boot until i remove the atag ATAG_CMDLINE Aug 08 06:46:49 once it it removed it boots fine. Aug 08 06:46:59 i can fill it with nothing and it still fails to boot. Aug 08 06:47:26 landslide: will it boot with 128Mb? Aug 08 06:47:48 i can't work out wher eto change the default command line in the kernel Aug 08 06:48:11 apex boots fine and I can use the extra ram fine. Aug 08 06:48:37 the kernel will boot but only 32m of ram.. because that's the default kernel command line. MEM=32M@0x0000000 Aug 08 06:49:21 daka: i'm in the middle of getting gcc.3.4.0-glib-2.3.2 Aug 08 06:49:35 crosstool that is.. i'll try compile apex and see if it makes a difference. Aug 08 06:49:38 how about this? boot init=/linuxrc mem=128M console=ttyS0,115200 root=/dev/mtdblock4 rw rootfstype=jffs2 Aug 08 06:50:22 daka: which file do I set it in openslug directory structure? and how do i recompile just the kernel? Aug 08 06:51:19 i found some really questionable stuff in apex that I need to ask beewoolie about. Aug 08 06:51:28 the atag structure is not dword aligned. Aug 08 06:51:36 i aligned it and it didn't help. Aug 08 06:52:11 landslide: file in openslug? huh? just apex> boot init=/linuxrc mem=128M console=ttyS0,115200 root=/dev/mtdblock4 rw rootfstype=jffs2 ? Aug 08 06:52:24 yeah it's not working. Aug 08 06:52:34 ah.. what does it say? Aug 08 06:52:41 as soon as apex sets the atag cmdline structure it crashes the boot. Aug 08 06:52:46 remove the atag structure and it boots. Aug 08 06:52:58 kernel just locks after uncompressing. Aug 08 06:53:34 strange... because boot init=/linuxrc mem=32M console=ttyS0,115200 root=/dev/mtdblock4 rw rootfstype=jffs2 works for me, but mem=64M fails after booting initrd.. Aug 08 06:53:51 lemme try... Aug 08 06:54:26 128M craches right after Uncompressing Linux.......................................................... done, booting the kernel Aug 08 06:54:35 but I don't have 128Mb so :-= Aug 08 06:54:36 :-) Aug 08 06:54:43 oh shit. Aug 08 06:54:45 it works. Aug 08 06:54:52 i've been just trying 'boot' Aug 08 06:54:56 I just found some 16M16 :-) Aug 08 06:55:01 by itself... the environment doesn't work. Aug 08 06:55:07 ah sweeeeet Aug 08 06:55:07 does it work with 128M? Aug 08 06:55:11 lemme try Aug 08 06:55:41 boot mem=something doesn't work... for some reason you have to use it all Aug 08 06:56:06 strlcpy (p->u.cmdline.cmdline, szEnv, COMMAND_LINE_SIZE); Aug 08 06:56:06 cb = strlen (szEnv) + 1; Aug 08 06:56:11 sorry. ;) Aug 08 06:56:30 does 128M work?? :-) /me want to know! Aug 08 06:56:57 no it dies. Aug 08 06:57:01 maybe this is related... Aug 08 06:57:07 it crashes right after uncompress. Aug 08 06:58:25 :( Aug 08 06:58:28 64Mb booted for me though. Aug 08 06:58:33 init=/linuxrc mem=64M@0x00000000 console=ttyS0,115200 root=/dev/mtdblock4 rw rootfstype=jffs2 Aug 08 06:58:43 hm.. and it completes everything? Aug 08 06:58:50 yep Aug 08 06:58:57 dammit.. :-) Aug 08 06:59:03 and cat /proc/meminfo identifies it. Aug 08 06:59:05 why not 128 ? Aug 08 06:59:28 what happens with mem=96M? Aug 08 06:59:31 128 dies. Aug 08 06:59:45 yes, I was just going to ask about 96M Aug 08 07:00:00 nope. Aug 08 07:00:07 only 64 Aug 08 07:00:17 maybe my top 2 64 chips are screwed or something Aug 08 07:00:25 but i tested in the bootloader it worked ok. Aug 08 07:00:27 hm.. I'd be happy with just 64... Aug 08 07:00:44 but my top 32M doesn't seem to work.. Aug 08 07:00:47 same for you btw.. Aug 08 07:00:55 doesn't the bootloader also send an ATAG_MEM32 entry too Aug 08 07:00:57 your top 64 doesn't work.. Aug 08 07:01:02 maybe something is scrwed with that? Aug 08 07:01:11 daka: same with me. Aug 08 07:01:30 should it be 2 regions or 1 continuous region? Aug 08 07:01:31 so.. we cannot make the second bank work.. Aug 08 07:01:54 what if some parts of the address range are reserved though? Aug 08 07:02:03 I thought the max was 64? Aug 08 07:02:05 I tried mem=0x02000000@0x0 mem=0x02000000@0x02000000 and that didn't work for me.. but.. Aug 08 07:02:34 what about the atag_mem32 structure? it gets sent down when you boot. Aug 08 07:02:44 hm.. dunno.. Aug 08 07:03:01 i have my suspicions about apex.. i think there are some issues there. Aug 08 07:03:22 48M booted ;) Aug 08 07:03:42 well it looks like i was chasing the same problem. Aug 08 07:03:43 doh. Aug 08 07:03:56 but totally in the wrong area. Aug 08 07:04:28 65M failed. :( Aug 08 07:04:31 uh.. with 48M I get: Aug 08 07:04:32 /bin/sh: error while loading shared libraries: /lib/libgcc_s.so.1: ELF file version does not match current one Aug 08 07:04:35 Kernel panic - not syncing: Attempted to kill init! Aug 08 07:04:38 but scrambled.. Aug 08 07:04:39 i don't even get that far. Aug 08 07:04:52 I just get this: 'Uncompressing Linux.......................................................... done, booting the kernel.' Aug 08 07:04:56 nothing after that Aug 08 07:05:29 how did you test the memory in apex, and, does version report your entire memory? Aug 08 07:05:40 is there a way to interrogate the atag structures from the commandline? Aug 08 07:05:48 daka: yeah it does. Aug 08 07:06:03 daka: all memory reported, I just filled some bytes into the higher areas of memory Aug 08 07:06:07 they stayed there correctly. Aug 08 07:06:16 hm.. landslide what do you get from: du -4 0xcc000000#32 Aug 08 07:06:33 i had some weird stuff occuring with the openslug compiler.. but after changing to cross tool. they went away Aug 08 07:07:06 du: invalid option -- 4 Aug 08 07:07:08 ;) Aug 08 07:07:28 uh.. in apex :-) Aug 08 07:07:54 cc000000: 0000001b 00000081 00000006 80000828 ........ .......( Aug 08 07:07:54 cc000010: 00001e26 00001e26 00001e26 00001e26 ...&...& ...&...& Aug 08 07:08:14 u can tell i'm a linux noob. Aug 08 07:08:38 version shows this Aug 08 07:08:38 memory: Aug 08 07:08:39 0x0 0x08000000 (128 MiB) Aug 08 07:08:52 hm.. 1b, correct for your configuration. Aug 08 07:09:02 and I have 19, correct for my configuration.. Aug 08 07:09:07 what is it? Aug 08 07:09:18 but: Aug 08 07:09:19 memory: Aug 08 07:09:19 0x0 0x02000000 (32 MiB) Aug 08 07:09:22 ah the memory register. Aug 08 07:09:23 i see. Aug 08 07:09:25 yep Aug 08 07:09:44 there is some memory scan code responsible for that. Aug 08 07:09:55 the memory must be configured absolutely correctly. Aug 08 07:10:19 and i found the memory scan code the problematic area when compiling apex with the openslug compiler. Aug 08 07:10:26 hm.. oh? Aug 08 07:10:48 yep. so I changed to crosstool. Aug 08 07:10:59 hm.. well. Im using crosstool :-) Aug 08 07:11:16 actually why don't you try crosstool gcc-3.3.4-glib-2.2.5? Aug 08 07:11:26 yeh.. probably should.. Aug 08 07:11:34 i can compile for you if you want. Aug 08 07:11:46 actually don't do that. Aug 08 07:11:52 u got jtag ? ;) Aug 08 07:11:56 hehe :-= Aug 08 07:11:57 :-) Aug 08 07:12:01 * DaKa2 hides Aug 08 07:12:24 the other thing is if you run the apex from sdram the memory will NOT be configured properly. Aug 08 07:12:37 that got me caught out big time. Aug 08 07:12:39 I know, thats why I flashed it Aug 08 07:13:00 why can't we use redboot? Aug 08 07:13:03 strange thing is version showed the correct 64Mb when run from sdram :-) Aug 08 07:13:07 it has ethernet and everything.... Aug 08 07:13:16 beewoolie didn't understand it at all Aug 08 07:13:17 daka: that is DARN weird. Aug 08 07:13:37 i think the memory scan code is flakey... Aug 08 07:13:50 or my extra 32M is broken Aug 08 07:13:50 i debugged it and found cases where it screwed up big time. Aug 08 07:14:15 like if the memory is not configured correctly it get weird overruns like the memory is mapped in segments. Aug 08 07:14:25 daka: possible but I doubt it. Aug 08 07:14:36 have you tried copying the bootloader to the upper area ? Aug 08 07:14:39 and check sum it? Aug 08 07:15:01 i have tried some fill/dump, and actually found a loose pin on a chip, so it seems to work Aug 08 07:15:11 should be ok. Aug 08 07:15:28 if we are both having probs.. I highly suspect something with the chipselect is dicky... Aug 08 07:15:34 soldered that one correctly, and looked at the rest and now fill/dump works great on a few testareas Aug 08 07:15:42 and possibly apex is not handling the regions correctly. Aug 08 07:16:01 well. apex can fill/dump on the top bank.. Aug 08 07:16:08 but the kernel.. Aug 08 07:17:36 ah sheesh.. i think my problem is the ram. Aug 08 07:17:41 i just tested it again.. no good. Aug 08 07:17:43 I should probably look a bit at the memory detection code in apex... since it doesn't get my extra 32M Aug 08 07:17:48 can't even dump 0x040000000 Aug 08 07:17:59 thats bad.. Aug 08 07:18:17 hang on.. too many zeros. Aug 08 07:18:26 heh Aug 08 07:18:28 yep Aug 08 07:18:30 didn't see that Aug 08 07:18:33 doh Aug 08 07:21:23 actually the copy doesn't work reliably on high memory. Aug 08 07:21:26 i'm sure of it now. Aug 08 07:21:36 daka: is yours an overclocked slug? Aug 08 07:21:51 landslide: yep, but I resoldered the resistor and tried that too Aug 08 07:21:57 hmm Aug 08 07:22:13 maybe that's my problem now. Aug 08 07:22:24 the top chips need to be fixed up Aug 08 07:22:54 need to reflow it. Aug 08 07:23:06 are you sure? Aug 08 07:23:47 yep posititive Aug 08 07:24:03 you cannot write to the top bank? Aug 08 07:24:14 could have been a shitty joint before.. now over time it's faulted.. or the chips are screwed. Aug 08 07:24:24 fill 0xAA 0x05200000+1000 Aug 08 07:24:25 du 0x05200000 Aug 08 07:24:31 does that work? Aug 08 07:24:33 i can write it's just some addresses are unreliable. Aug 08 07:24:59 yep that works Aug 08 07:25:02 probably a bad/floating address pin then. Aug 08 07:25:02 but if I do this Aug 08 07:25:14 hm.. if you identify what adresses doesn't work you might be able to figure out what pin Aug 08 07:25:19 copy 0x8000+5000 0x4000000 Aug 08 07:25:30 then compare du 0x8000 to 0x4000000 Aug 08 07:25:35 they are different. Aug 08 07:25:39 what about 55 not aa Aug 08 07:26:08 I'm not sure but perhaps trying address like 0x0000000001 and then 10, etc.. shifting the 1 up until you get an error? Aug 08 07:26:19 ok Aug 08 07:26:33 but then again, I have no idea if it's that simple. Aug 08 07:26:47 There's no virtual memory mapping crap at this point right? Aug 08 07:26:54 landslide: hm.. does 0x55 work too? Aug 08 07:27:06 yep Aug 08 07:27:34 well, then all the data pins are correct.. then just test every address pin.. Aug 08 07:27:49 55 and aa work on the 0x4000000 too Aug 08 07:27:57 but copy doesn't. Aug 08 07:28:02 i suspect copy might be to blame. Aug 08 07:28:12 hm.. let me try.. Aug 08 07:28:12 memory timing ? Aug 08 07:29:30 apex> comp -c 5000 0x8000 0x02000000 Aug 08 07:29:31 0 bytes the same Aug 08 07:29:46 what? if I dump both areas they are the same... Aug 08 07:30:29 ah, what if the area we are copying is data. Aug 08 07:30:37 the copy command modifies it itself. Aug 08 07:30:45 is that possible? Aug 08 07:31:25 uhm.. but.. how is compare supposed to work? I mean, lots of bytes are the same.. Aug 08 07:32:03 ifcomp 0x8000+1000 0x2000000 Aug 08 07:33:39 i think i have shorting pins. Aug 08 07:33:46 shorting address pins Aug 08 07:34:03 yep i'm sure of it now. Aug 08 07:34:17 but if they're shorting they'd short on the bottom chips too Aug 08 07:35:28 i might go look under a magnifying glass brb. Aug 08 07:37:48 ka6sox-away: ping? Aug 08 07:37:55 NAiL, pong Aug 08 07:37:57 * NAiL can't login to nudi? Aug 08 07:38:06 huh? Aug 08 07:38:08 login as: repvik Aug 08 07:38:08 Authenticating with public key "oyvind@repvik.org" Aug 08 07:38:08 Server refused to allocate pty Aug 08 07:38:22 ugh... Aug 08 07:38:46 going in to the lab in about 1hr...dentist appt first. Aug 08 07:38:50 ok Aug 08 07:39:04 It wasn't urgent, just wondered you were aware Aug 08 07:39:13 ka6sox-away, lucky you!! I have mine tommorow too. Not looking forward to it! Aug 08 07:42:40 VoodooZ_Work, yeah...this should be quite fast...just re-glueing a bit o'broken tooth. Aug 08 07:43:20 good then. I have a broken one too. It's due for a change but I have no insurance so i keep pushing it back. uh. Aug 08 07:43:43 yeah...I understand that one! Aug 08 07:53:37 daka2: you still around? Aug 08 07:53:43 landslide: yep Aug 08 07:54:30 if you copy 0x00008000 to 0x2000000 Aug 08 07:54:37 and dump both, are they exactly the same? Aug 08 07:54:46 they look the same.. Aug 08 07:54:52 hmmm. Aug 08 07:55:02 atleast the first 50.. Aug 08 07:55:15 actually no.. Aug 08 07:55:34 i'm seeing a pattern in where i fails to copy data to the top chips. Aug 08 07:55:35 some are wrong.. Aug 08 07:55:43 yeah and look 2 lines up Aug 08 07:55:52 you'll notice there is a kind of pattern there. Aug 08 07:55:56 yep.. Aug 08 07:56:12 i think the chip select line is not working Aug 08 07:56:26 or something is screwed in teh bootloader copying between memory Aug 08 07:56:40 but I did manage to dump/fill the top bank.. Aug 08 07:57:14 yep Aug 08 07:57:16 same here. Aug 08 07:57:18 now get this. Aug 08 07:57:35 if I copy on the lower chips I get no problem Aug 08 07:57:47 however i copy to the upper chips there are losses. Aug 08 07:57:57 but if I copy from the the upper chip to the upper chip.. there is no loss Aug 08 07:58:13 I think that [g2] only had 64M with only lower bank filled.. Aug 08 07:58:19 iirc Aug 08 07:58:23 ah. Aug 08 07:58:32 well my 64 mg is working fine on the lower bank. Aug 08 07:58:45 the upper bank is exhibiting the same problems as you Aug 08 07:58:51 I was under the impression that 64Megs was the max you could upgrade to. Aug 08 07:58:55 i doubt it's the memory chips or a bad connection. Aug 08 07:59:18 voodoo: a chip select line is available on one of the resistor arrays. Aug 08 07:59:24 landslide: exactly.. so.. hardware, apex, or something else... Aug 08 07:59:25 it was traced back with that salted slug. Aug 08 07:59:53 I'm not sure. I would check the Wiki for the CS stuff Aug 08 07:59:55 VoodooZ_Work: you should be able to get 256Mb.. not that I know if anything else screws that up.. Aug 08 08:00:25 yeah, but I though some blocks were reserved for other devices. Aug 08 08:00:37 maybe.. i'm not sure.. Aug 08 08:00:39 Not according to: http://www.nslu2-linux.org/wiki/Info/MemoryMap Aug 08 08:00:48 i think there is plenty of area left for sdram though Aug 08 08:00:54 it should take 256 mb Aug 08 08:01:12 yep.. that first block between 0x00000000 and 0x10000000 is 256Mb.. Aug 08 08:01:22 and the ixp should work with that Aug 08 08:01:45 wiki: "The bank enable line SDM_CS_N1 is connected to the current bank which leaves SDM_CS_N0 free. SDM_CS_N0 is available from the unconnected pin on PR12 directly below the SDRAM chips." Aug 08 08:01:51 oh i see now. Aug 08 08:02:04 how do I do a pastebin Aug 08 08:02:07 i have something show. Aug 08 08:02:12 www.pastebin.com? Aug 08 08:02:42 sorry, you are right Aug 08 08:03:00 I just checked it. But then again, nobody ever tested it. Aug 08 08:03:02 pastebin.ca Aug 08 08:03:58 thanks Aug 08 08:04:43 daka: http://pastebin.ca/19545 Aug 08 08:05:08 check out 2 lines above the UUUUUUUU Aug 08 08:05:12 it screwed with it. Aug 08 08:05:32 ugh.. lets see.. Aug 08 08:05:46 does yours do that? Aug 08 08:06:28 it happens everywhere. Aug 08 08:06:53 hm.. nope.. mine doesnt.. Aug 08 08:06:56 i think my address lines are shorted. Aug 08 08:09:17 uh.. this got wrong... Aug 08 08:09:47 things screwed up for me too actually.. Aug 08 08:10:07 02000070: a5 b8 39 1e 6f b9 55 55 55 55 55 55 55 55 5a 14 ¥¸9.o¹UU UUUUUUZ. Aug 08 08:10:19 with: fill 0x55 0x02000078+8 Aug 08 08:10:27 two off.. Aug 08 08:10:49 hmmm, it's strange this is only the top bank though. Aug 08 08:11:03 yep.. if it was shorted it should mess that up too Aug 08 08:11:37 yep i've found a definite pattern Aug 08 08:13:01 hm.. I must have wired something wrong... this is just too strange.. Aug 08 08:14:02 omg Aug 08 08:14:09 There's ages since I did a make update on my slug Aug 08 08:14:15 :) Aug 08 08:14:35 232 revs to ./ Aug 08 08:14:44 Dunno how many revs to openembedded yet Aug 08 08:14:51 argh what have I done to deserve this.. things just get wrong... Aug 08 08:15:06 You're swedish, isn't that enough? ;) Aug 08 08:15:10 grrrrrr Aug 08 08:16:31 nice pattern: Aug 08 08:16:32 02000060: a7 a8 3b 13 67 59 92 0f 27 94 2b b3 6e 19 db 45 §¨;.gY.. '.+³n.ÛE Aug 08 08:16:32 apex> fill 0x55 0x02000066+7 Aug 08 08:16:32 02000060: a7 a8 55 55 67 59 55 55 55 55 2b b3 55 19 db 45 §¨UUgYUU UU+³U.ÛE Aug 08 08:17:00 it looks like your top chips are not connected properly Aug 08 08:17:03 i think that's all it is. Aug 08 08:17:15 yep.. Aug 08 08:17:18 i worked out the pattern on mine. Aug 08 08:17:27 oh? Aug 08 08:18:11 if i write to bytes 2 3 6 7 10 11 14 15 they get duplicated 20 bytes before. Aug 08 08:18:37 so if I write to 5000026 the byte is written to 5000006 aswell Aug 08 08:18:56 same pattern through all the stacked ram. Aug 08 08:20:15 that means to me that bits 0 and 1 are some how tied to bit 5 (on the address bus) Aug 08 08:23:22 ok time for bed. Aug 08 08:24:36 landslide: ok, nite :-) Aug 08 08:25:40 daka will you be on when bewoolie is on? Aug 08 08:25:47 maybe.. Aug 08 08:25:56 if so let him know about the prob I had too. Aug 08 08:26:21 ok Aug 08 08:26:55 he might be able to help... not sure.. i don't know how the sdram is layed out in memory as far as the 2 chips are concerned. Aug 08 08:27:01 he may have some ideas... Aug 08 08:27:10 anyway have a goodday . Aug 08 08:29:27 send him these 2 pastebins... they'll help diagnose it Aug 08 08:29:28 FORWARD: http://pastebin.ca/19545 BACKWARD: http://pastebin.ca/19546 Aug 08 08:40:26 DaKa2: Is there a way to recreate destroyed u-a links? Aug 08 08:42:26 I installed a package that destroyed a bunch of links :( Aug 08 08:46:06 NAiL: well... just manually run the correct update-alternatives --install Aug 08 08:46:32 hehe, yeah. It's just that I don't know which were destroyed :P Aug 08 08:46:42 ah.. Aug 08 08:46:55 well.. make a small script that looks in /usr/lib/ipkg/alternatives then :-) Aug 08 08:48:03 woho! Aug 08 08:48:03 memory: Aug 08 08:48:04 0x0 0x04000000 (64 MiB) Aug 08 08:48:27 and.. Aug 08 08:48:29 linux.. Aug 08 08:48:33 boots!!! Aug 08 08:48:36 yay Aug 08 08:48:39 what was wrong? Aug 08 08:48:48 lets see.. Aug 08 08:49:05 dammit.. should have noted what pin I fixed.. Aug 08 08:49:11 not an addess or data pin Aug 08 08:49:14 some of the other ones Aug 08 08:51:03 MemTotal: 63140 kB Aug 08 08:51:03 MemFree: 54112 kB Aug 08 08:51:20 hehe.. now I have to do a 128MB one :) Aug 08 08:51:25 this was easy Aug 08 08:51:32 (almost... not...) Aug 08 08:52:58 damn.. now I want some 32M16 to make a 256Mb slug :-) *drools* Aug 08 08:54:24 * DaKa2 can't stop smiling Aug 08 08:56:12 * koen starts drinking less coffee and saving for a better soldering iron Aug 08 08:57:01 :) Aug 08 08:57:47 Im using a really old soldering station, with a 0,8mm tip.. Aug 08 08:58:03 and that is really not good for this.. Aug 08 08:58:34 * koen looks at the 2mm tip Aug 08 08:58:47 :) Aug 08 08:59:24 I could clamp in a needle Aug 08 08:59:37 2mm.. that covers three pins on the memory chips ;-) Aug 08 08:59:51 but it would be faster to use a hammer to kill the slug Aug 08 09:00:15 probably, yes :) Aug 08 09:00:35 I wonder if soldering paste would work Aug 08 09:00:49 Im going to bring my unmodified slug tomorrow and make that 128MiB... Aug 08 09:01:23 and if that works I'll bring my third and make that 128MiB too Aug 08 09:01:28 heh Aug 08 09:01:37 then I'm out of 16M16.. Aug 08 09:02:11 I have 8M16 to make 10 64MiB slugs.. Aug 08 09:02:28 12 actually Aug 08 09:02:46 how much does a 16M16 chip cost? Aug 08 09:03:00 no idea.. Im using old sdram stics Aug 08 09:03:03 sticks* Aug 08 09:03:54 4 chip 128mb cas3 pc133 sticks seems to have 16M16 Aug 08 09:04:22 so just take the four chips and replace the two in the slug Aug 08 09:52:37 There's two ways right? piggy-backing or replacing the chips right? Aug 08 09:54:42 ya Aug 08 09:54:48 but stacking is UGLY Aug 08 09:56:01 yeah, but i'm looking for the easiest solution as I don't really need the memory yet. I'm just curious. Aug 08 09:56:47 So to get 64Meg I would piggy-back two more chips right? Aug 08 09:59:38 VoodooZ_Work: yep.. thats what I did Aug 08 10:00:09 ugly, but works, has to be two 8M16 Aug 08 10:00:55 I'll have to dig for old SDRAM I guess. What DIMM size would those be on? Aug 08 10:02:28 dimm size? 8Mx16? Aug 08 10:04:41 I mean if I was to steal the ICs off an old DIMM stick. Aug 08 10:05:03 Then again, I'd need a better soldering iron first. Aug 08 10:07:55 VoodooZ_Work: ahh, well.. biggest chance: 4 chip 64MiB sdram sticks for the 8Mx16 or 4 chip 128MiB for the 16Mx16, but ytou can found them on other stuff too Aug 08 10:08:26 just remember that only somethingMx16 works, not x8 Aug 08 10:08:35 yeah. Thanks. I'll keep an eye out. Aug 08 10:09:18 and it might be a good idea to test the stick in a computer with some nice memtest program first Aug 08 10:10:53 definitely! Aug 08 10:17:25 I've been having at least 5 computers on the testbench all day testing ram.. tested atleast 40 sticks, 6 broken.. Aug 08 15:43:31 foo Aug 08 18:04:36 openslug autobuilds are broken: ERROR: [Errno 2] No such file or directory: '/home/slug/openslug/openembedded/packages/softdog/*.bb' while parsing /home/slug/openslug/openembedded/packages/softdog/*.bb Aug 08 22:48:32 NAiL: ping Aug 08 22:49:18 NAiL: revision bb045a5a17fc33253d0065173055ac1656458613 says "Added diffstat to feed" but adds a softdog/*.bb directory to conf/distro/openslug-packages.conf, which is killing the autobuilds. **** ENDING LOGGING AT Mon Aug 08 23:59:57 2005