IRC log for #nslu2-linux on 20070416

00:00.34rwhitbyis apex passing the correct atags?
00:00.43rwhitby(it prints them out)
00:01.08RobNCyeah but network is still working.. I can ping, etc and things still work.  Now, to turnup, get USB drive, etc. and verify that after that "crash", the USB dev on PCI is no longer accessible.
00:01.38rwhitbyok, I'll be busy for a while now (at work) - keep me informed here and I'll look between meetings.
00:01.48RobNCok, thanks!  have fun :-)
00:03.47rwhitbyRobNC: don't you need to run memlimit to set the atags properly?
00:03.55rwhitbythe atags being passed in your pastebin are only for 32mb
00:04.03rwhitbyyou shouldn't need the mem= on the cmdline
00:04.30RobNCrwhitby: with D-I, I didn't see that it make a difference.  I'll try it though.
00:05.23rwhitbyunless you run memlimit in apex, I don't think it will change the atags from 32mb
00:05.39rwhitbyI don't think memscan does what you think it does
00:05.54rwhitbyyou should run sdram-init, then memlimit
00:06.09rwhitbymemscan just looks for aliases, it doesn't change the boot params.
00:06.09RobNCno need for memscan?
00:06.19rwhitbyI don't think so.
00:06.21RobNCok gotcha.
00:06.27RobNCwill try and let you know.
00:06.34rwhitbyok, bbiab
00:21.41RobNCfyi, rwhitby, 256MB only detected if cmdline parameter appended.  "memscan" didn't do it (I had right params).
01:21.09RobNCnote that same params for Apex ->debian image didn't work (using just sdram-init and memscan, without memlimit).  *dumb looks*
01:21.32rwhitbywhat atags are passed when you use memlimit?
01:22.44RobNClemme look - will have to reboot
01:27.10RobNCu mean ATAT_MEM I presume - it is "ATAG_MEM: start 0x00000000 size 0x02000000"
01:27.37RobNCthat is with "sdram-init; memlimit 256m; etc.. " other startup params
01:28.58RobNCweird, seems that is the same as when 256MB is detected?!
01:29.27RobNCso it seems that kernel ignores ATAG_MEM and only looks at ATAG_CMDLINE for "mem=256MB@0x0"
01:30.49rwhitbyATAG_MEM 2000000 is only 32MB, right?
01:31.09rwhitbyyou need to get apex to send 0x10000000 for 256MB, no?
01:31.10RobNCyep.  256MB would be 0x10000000
01:32.09RobNCok thks
01:32.41rwhitbyjust work out which apex commands get the atags right, and then the kernel should see it
01:46.28RobNCgonna re-flash old debian image to compare atags
01:51.15RobNCAH I think I see the problem.  need "memscan -u 0+256m" -> -u passes to kernel
01:55.24*** join/#nslu2-linux Sinclair73de (
01:56.17RobNCyup that was it.  memscan -u 0+256m  => no need to meminit
01:59.36RobNCATAG_MEM shows 0x10000000
02:18.29RobNCrwhitby: ok reproduced problem via rsync of >200GB data - USB mounts die and aren't accessible, but kernel doesn't die (i.e., ramfs still functional).
02:27.45*** join/#nslu2-linux kolla (
03:10.57rwhitbyok, -u makes sense
03:11.21rwhitbyRobNC: nice, so now we can go to linux-arm-kernel with a 2.6.20 kernel problem
03:16.48RobNCrwhitby: indeed!  just a matter now of waiting for subscription email from moderators.
03:18.28rwhitbyRobNC: can you test to see where the limit is?  Is it 192MiB? Is is 200MiB?  That might be interesting information ...
03:19.05RobNCI got 192MB to work w/o a problem.  I will keep trying to find where the limit is
03:19.38RobNCthing I am not sure - where in memory space is the free memory - is there any prerequisite that mem is filled from high to low or vice versa
03:31.13*** join/#nslu2-linux osas (
03:35.32rwhitbyRobNC: dunno
03:35.58rwhitby192/256 is still a nice memory size for a slug anyway :-)
03:38.01RobNCtrue!!  it seems to have a problem right around the free space left.
03:38.48RobNCAH HA it crashed before (would not shut down) because I had plugged in my HDD, and it had swap.  SlugOS detected it, and enabled it.  for shutting down, it couldn't "unmount"
03:43.59RobNCweird.. think Ifigured out the problem perhaps or at least how to recreate it
03:45.07RobNCdo "free", then convert that number for free to MB (/1024).  Then do "memtest X" where X = that number, rounded down.   If you do X-1, it works fine.  But X causes DMA crash
03:49.33rwhitbyall good info ...
03:49.35rwhitbybbiab (lunch)
03:49.49RobNCok ttyialw
03:53.50*** join/#nslu2-linux jipi (
03:54.57RobNCinteresting... the IXP420 doesn't use PCI for networking, right?  It is a MAC, and just need a PHY, right?  How many devices are on the PCI bus?
03:56.11RobNCanswered my own question - only NEC USB controller on PCI bus
04:10.37*** join/#nslu2-linux nSiclair73de (
04:14.46rwhitbynSiclair73de: you've got a fatslug, right>
04:15.25RobNCok, can't stay up so late (work tomorrow).  gonna bcc you on email rwhitby
04:15.41rwhitbyok, thx
04:17.15rwhitbyRobNC: I'll be on gprs for 10 days, so please forward me by email any replies you get.
04:17.32RobNCok will do; have fun :-)
04:18.49RobNCttyl guys have a good one.
04:24.38*** join/#nslu2-linux Tobbe (
05:08.19RobNCrwhitby btw problem seems to be with malloc - I got it to do this using memory config all the way down to 16M!
05:34.39rwhitbySinclair73de: ping
05:46.02*** join/#nslu2-linux DrZimmerman (
05:54.45*** join/#nslu2-linux pleemans (
06:03.00*** join/#nslu2-linux jacques (n=jacques@nslu2-linux/jacques)
06:08.35CIA-503bzhou * r5938 10optware/trunk/make/ dovecot: fixed older automake problem, specify 1.9
06:17.06*** join/#nslu2-linux eiffel_ (
08:10.05*** join/#nslu2-linux pleemans (
08:14.40*** join/#nslu2-linux jelle_ (
08:40.16*** join/#nslu2-linux pleemans (
08:43.46*** join/#nslu2-linux likewise (
08:44.19likewiseDoes anyone know how I can detect if a IXP42x platform was cold booted (i.e. from power-up)
08:56.17jelle_maybe you cat alter the reboot script to leave some file somewhere to let you know that it was a warm reboot, and have this file checked on boot to set a flag somewhere... But not out of the box, no.
08:56.47jelle_not fool proof, but a start.
09:00.43likewiserwhitby: I can only find a register bit telling me it was a watchdog reset.
10:21.33*** join/#nslu2-linux mithro (
10:47.57*** join/#nslu2-linux kolla (
11:35.30*** join/#nslu2-linux jipi_ (
11:53.32*** join/#nslu2-linux eiffel (
12:05.19*** join/#nslu2-linux likewise (
12:08.36*** join/#nslu2-linux eiffel (
12:30.59*** join/#nslu2-linux GPSFan (n=kenm@
12:41.37*** join/#nslu2-linux eiffel (
12:50.48*** join/#nslu2-linux eiffel (
14:54.36*** join/#nslu2-linux WesselVS (
15:29.13*** join/#nslu2-linux eiffel (
15:52.24*** join/#nslu2-linux HopsNBarley (n=hops@nslu2-linux/HopsNBarley)
16:38.39*** join/#nslu2-linux Cheef-Daniel (
17:17.57*** join/#nslu2-linux nSi (
17:22.20*** join/#nslu2-linux AdamBaker (
17:25.49*** join/#nslu2-linux Tobbe (
18:21.05*** join/#nslu2-linux hillct (
18:33.08*** part/#nslu2-linux hillct (
18:34.22*** join/#nslu2-linux pleemans (
18:39.49*** join/#nslu2-linux kolla (
18:53.12*** join/#nslu2-linux caplink811-log (
19:31.18*** join/#nslu2-linux [g2] (n=tom@nslu2-linux/g2)
19:47.05*** join/#nslu2-linux alump_ (
20:40.03*** join/#nslu2-linux HopsNBarley (
20:40.33*** join/#nslu2-linux jelle_ (
22:29.57*** join/#nslu2-linux drath_ (
22:32.58*** join/#nslu2-linux marceln (
22:33.27CIA-503marceln * r5939 10optware/trunk/make/ Dovecot: changed version number to 1.0.z0 so that it overrules the RC versions
23:15.11*** part/#nslu2-linux marceln (
23:15.48CIA-503bzhou * r5940 10optware/trunk/make/ py-genshi: 0.3.6 -> 0.4
23:38.09CIA-503bzhou * r5941 10optware/trunk/make/ elinks: 0.11.2 -> 0.11.3
23:41.07CIA-503bzhou * r5942 10optware/trunk/ (make/ platforms/ transcode: 1.0.2 -> 1.0.3; promoted for fsg3; dunno who is using it though, i am not

Generated by Modified by Tim Riker to work with infobot.