irclog2html for #nslu2-linux on 20050523

00:37.29*** join/#nslu2-linux jbowler (~jbowler@jbowler.nslu2-linux)
00:47.47*** join/#nslu2-linux [g2] (~g2@g2.nslu2-linux)
01:06.25*** join/#nslu2-linux sp4ce (
02:11.58*** join/#nslu2-linux Jacmet (
02:17.10*** join/#nslu2-linux Tiersten_ (tman@Tiersten.nslu2-linux)
02:23.36*** join/#nslu2-linux Ksilebo (
02:48.47*** join/#nslu2-linux ByronT (~byron-por@ByronT.nslu2-linux)
03:01.42*** join/#nslu2-linux jbot (
03:01.42*** topic/#nslu2-linux is Developer (not end-user) discussion about modifying the Linux operating system on the Linksys NSLU2. Use #nslu2-general for end-user questions. See for info, and be sure to read before asking questions.
03:46.03*** join/#nslu2-linux Ksilebo (
04:51.50*** join/#nslu2-linux Ommadawn (
05:04.52*** join/#nslu2-linux Jacmet (
05:06.28*** join/#nslu2-linux ruDelph (~delph@
07:13.21*** join/#nslu2-linux llagendijk (
07:26.36*** join/#nslu2-linux jf-work (~none@jeanfabrice.nslu2-linux)
07:30.08jbowlergood morning
07:47.39*** join/#nslu2-linux Jacmet (
09:05.05*** join/#nslu2-linux bob_tm-zzz (
09:15.38dyoungmorning bob_tm
09:16.05bob_tmdyoung: Morning
09:47.16*** join/#nslu2-linux eno-away (
10:53.47*** join/#nslu2-linux bob_tm (
11:51.26*** join/#nslu2-linux [g2] (
11:56.22*** join/#nslu2-linux sjmelhuish (
11:57.16*** part/#nslu2-linux sjmelhuish (
12:07.50*** join/#nslu2-linux jf-work (~none@jeanfabrice.nslu2-linux)
12:07.54*** join/#nslu2-linux phedny (
12:11.19rwhitbybk is back
12:11.50dyoung-zzzzyeah, I just finished pulling it.
12:11.53rwhitbyanything else which should go in 4.21 ?
12:12.06dyoung-zzzzOk you got the mnt/repair fix?
12:13.02dyoung-zzzzok you can push it.
12:13.17dyoung-zzzzas far as other repairs for 4.21, I cant think of any.
12:15.11dyoung-zzzzunless you think its prudent to leave out the oe feed until we have a way to automagically edit the Packages file.
12:16.17[g2]bk was back yesterday afternoon for me. I made a fresh clone and build OpenSlug and openslug-packages.  I haven't downloaded the firmware but all built fine
12:16.48dyoung-zzzzI've been slacking this weekend whilst fiddlign the Tivo.
12:17.06[g2]the S3 ?
12:17.27*** join/#nslu2-linux ruDelph (~delph@
12:17.51dyoung-zzzzI thought you might know about a S3 that I dont know about. :-)
12:17.53[g2]70wpm with only 60% errors
12:18.17[g2]Tivo is supposed to be getting out of the hw buisness right ?
12:18.46[g2]iirc that's what I read a month or two ago
12:19.12dyoung-zzzzI wonder who they'll source hardware to
12:20.02dyoung-zzzzThey should source hardware to [g2]'s House of IXP Hardware And Other Stuff.
12:21.19CIA-103rwhitby 07 * r1.3319 10openembedded/ (4 files in 4 dirs): Unslung 4.21 - added /mnt/repair fix. Also added /etc/inittab to conffiles in sysvinit
12:21.26[g2]I'll most likely be avoiding the DVR market (except for the NAS backend part)
12:21.31dyoung-zzzzI was intrigued by some cryptic error messages I was getting earlier.
12:22.07dyoung-zzzzIt seems to get some "stuff" from/through a FPGA on board.
12:22.08[g2]Totally Off topic but... Have you guys heard of davetv ?
12:22.22dyoung-zzzzSounds like a film.
12:22.59[g2]August is the hw launch $199. They've got 100,000 hours of content
12:24.25[g2]heh  comforts
12:24.26dyoung-zzzzI loaded 4.20 and did the basic test.
12:24.36dyoung-zzzzipkg update; ipkg install openssh
12:24.39dyoung-zzzzbut thats about it.
12:25.12rwhitbyI still don't have the required spare LAN port to hook up devslug.
12:25.12[g2]are you guys following the OE/SVN updates much ?  They're working out a tinderbox QA thing
12:25.37rwhitbyI'm also still waiting for the wl530g to arrive ...
12:25.55dyoung-zzzzNo, but since kergoth pointed out the bitbake-devel list, I've been trying to keep up there.
12:26.03[g2]I'd like to incorporate that in to the EmKnoppix release and then we can get massive regression testing
12:26.09dyoung-zzzztrying doesnt mean succeeding.
12:26.28[g2]well this is all a work-in-progress
12:26.49[g2]I thought rwhitby was the vortex ? :)
12:27.00dyoung-zzzzno, rwhitby was a machine.
12:27.03dyoung-zzzz* is
12:27.41dyoung-zzzzHmm, could be rwhitby is spinning real fast so he could be the vortex....
12:27.50dyoung-zzzzyou said Skynet.
12:27.54dyoung-zzzzIt should really be SonyNet.
12:28.38dyoung-zzzzThe PS3 with its wireless capability and Cell CPU with 2 TF....
12:29.01dyoung-zzzztheyre all going to talk to each other (ps3's that is) and the network will become self-aware within months.
12:30.12[g2]the Xbox 360 and Sony P3 will rock
12:30.31dyoung-zzzz(yeah, yeah, I know the Cell 2 Teraflop figure is for single precision)....
12:31.05[g2]while I don't agree with the closed nature of the platform, I highly applaud them pushing the price/performance envelope
12:33.03dyoung-zzzzrwhitby, should be remove the rw bit from the kernel cmdline?
12:33.13dyoung-zzzzreducing the risk of accident?
12:33.17[g2]rwhitby, are you going to sleep soon ?
12:33.22rwhitbywhat accident?
12:33.38rwhitby[g2]: nope, I have time right now to talk OpenSlug 2.x and futures ...
12:33.40dyoung-zzzzrm -rf /initrd type accident
12:34.05[g2]ok lets talk OpenSlug 2.x roadmap
12:34.21rwhitbydyoung-zzzz: they'd have to do that before unslinging, right?
12:34.34rwhitbyin which case, they're on their own
12:35.04dyoung-zzzzno, thats something that would happen after unslinging because the jffs2 is pivoted to there.
12:35.26dyoung-zzzzFrom there?  You get the idea....
12:35.26rwhitbywhy not just umount /initrd in the rc script instead?
12:35.34dyoung-zzzzor that.
12:35.45dyoung-zzzzbut noone ever figured out how.
12:36.23rwhitbyI think we could do that now in /etc/rc pretty easily, since the removal of switchbox revolution
12:36.57dyoung-zzzzthere was a lingering dev/console somwehre in there.
12:37.09dyoung-zzzzbut I didntlook closly into it.
12:37.40jacquessays a lot for the reliability of bkbits that it now goes down every weekend and has to be kicked on monday morning
12:37.53rwhitbyI thought that disappeared since we booted from jffs2 instead of switchbox
12:38.04rwhitby(dev/console that is)
12:38.17rwhitbyjacques: indeed
12:39.02dyoung-zzzzAFAIK the issue still exists.
12:39.16dyoung-zzzzbut I'm way to tired to think about it right now,
12:39.26rwhitbyyou're probably right - I haven't even flashed it :-)
12:40.02dyoung-zzzzand AFAIK "we" decided to ignore the issue because its not using RAM anymore.
12:40.22rwhitbyyeah, that sounds familiar
12:41.20rwhitbynight dyoung-zzzz
12:42.00rwhitbyjacques: what's so good about the t42?  (compared to the run-of-the-mill Dell Lattitude D600 that work provides me) ?
12:42.18jacquesrwhitby, it's the bestest :-)
12:43.18dyoung-zzzzThe lack of a serial port was an initial downer for me; but overall I really enjoy my T42.
12:43.20jacqueslooking at it, you can tell it's a serious piece of equipment - the R&D shows IMHO
12:43.44jacqueshardware is all supported by linux
12:43.52jacquesthird mouse button :-D
12:43.58jacquesS-IPS screen
12:44.09jacquesmagnesium case
12:44.11dyoung-zzzzokay, you're gonna have to tell me the secret to making that IPW2200 do soemthing sometimday.
12:44.28jacquesdyoung-zzzz, it works fine under ubuntu 5.04 :-D
12:44.44dyoung-zzzzyeah, it does. ;-)
12:44.45jacquesI'm just emerging gentoo now so I haven't tried building it myself
12:44.49[g2]don't they use the wrappers ?
12:45.00[g2]NDIS ?
12:45.05jacquesthat's something else
12:45.15dyoung-zzzzNDIS Wrappers can Bite Me.
12:45.23jacquesme too
12:45.34dyoung-zzzzipw2200.sourceforge.nget I believe.
12:45.46jacquesyeah that looks right (except for the nget)
12:46.53jacquesi love having the trackpad *and* the little eraser pointer thingie
12:47.00dyoung-zzzzMaybe I'll end up ubuntu-izing the machine.
12:47.10jacqueskeyboard is best in class
12:47.24dyoung-zzzzI like the keyboard too.
12:47.36dyoung-zzzzversus the last 2 dells I had, its solid.
12:48.20jacquesdell you never know what you're gonna get - it may be total crap they designed, or it may be a rebadged Sharp (nice)
12:48.25dyoung-zzzz(I went through 5 keyboards over 2 dells....)
12:49.18jacqueslol, T42 has a parallel port (!) so I can do JTAG with it
12:49.47jacquesoh, I forgot to mention - no windows key :-D
12:49.49dyoung-zzzzhehe, funny you should mention that.  Mmy old Inspiron 8200 parallel port couldnt talk to that DLC5.
12:50.07dyoung-zzzzI fought it for at least 3 days.
12:50.14dyoung-zzzzso I said forget it and got a IBM.
12:50.19rwhitbywhich T42 model?
12:50.42jacquesrwhitby, mine is a 2379DXU
12:50.57dyoung-zzzzI have a 2373.
12:51.18jacquespentium-M 1.7, 15" 1400x1050 flexview display
12:51.31rwhitbyhmm - can't see the 2379DXU here:
12:51.34dyoung-zzzz14" SXGA for me.  Same CPU
12:52.19jacquesrwhitby, yeah it's not on there right now for some reason - I got mine from newegg
12:53.19jacquesoh yeah, this model has 3 year parts and labor warrantee
12:53.23jacques(not all do)
12:53.37dyoung-zzzzthe GigE is nice.
12:54.23jacquesit will be at some point :-)
12:54.38jacquesI dont have any gige stuff yet
12:55.48jacquesrwhitby, the DXU disappeared from that ibm site when they switched over to Lenov, also apparently the switchover is a bit chaotic - wait times are long ordering through them - that's another reason I ordered from newegg - no waiting
12:57.12jacquesheh, last night I ordered 2GB ram for it, 9-cell battery, and extra power adapter
12:59.37jacquesnow I need a case
12:59.52rwhitby[g2]: so you were talking about OpenSlug 2.x ?
13:00.00[g2]rwhitby, nod.
13:00.12[g2]1 minute I'm just about down with a wiki update
13:01.17*** join/#nslu2-linux phedny (
13:06.20[g2]jacques, I'm guessing you want all that power to be portable right ?
13:06.23[g2]the T42
13:08.36jacques[g2], exactly
13:08.49rwhitby[g2]: General point 2 - why wouldn't those tweaks be OpenSlug distro conditionals in OE ?
13:09.55[g2]rwhitby, Good point! I think that would be the proper way of handling it.
13:10.15[g2]BTW: Are you guys familiar with OE overlays at all ?
13:11.52rwhitby[g2]: why NewRoadmap, and not just OpenSlug/Roadmap ?
13:12.04[g2]Collections actually 3.7
13:13.06[g2]rwhitby, I'll get rid of the New... the OpenSlug wiki pages need to be cleaned up a little
13:14.36rwhitby[g2]: re collections: we could use that to have an nslu2-linux repo which was using a different CM system from the base OE repo ....
13:14.53[g2]exactly that's the point :)
13:15.29[g2]or make nslu2 specific changes to global packages
13:15.50[g2]that are more than a conditional change
13:16.29rwhitbyyeah, that introduces a maintenance headache though ....
13:17.01rwhitby(if you keep your changes separate, then you don't get free version upgrades)
13:18.19[g2]so lets talk about CM systems.
13:19.44[g2]Since Unslung and Openslug have greatly stabilized wrt OE, how do you feel about bi-weekly/monthly based lines for normal users ?
13:20.51rwhitbywhy time-period-based instead of new-feature-based ?
13:21.04[g2]that's possible too
13:21.52rwhitbyless expectations that way ...
13:21.54[g2]we may want to include some interium releases if the feature set is lagre
13:23.21[g2]So what features would you like in the next OpenSlug release ?
13:25.19[g2]for bluetooth ?
13:25.52rwhitbyi.e. it does everything I need it to do, and all I need to do know if make sure that my local mods (like /etc/inittab) are in conffiles so that it stays working when I do ipkg upgrade on it
13:26.19rwhitbyBT and dial-in modem is what I mainly use AccessSlug (which runs OpenSlug) for ...
13:26.36rwhitbyoh, and miau
13:26.54rwhitbyand it only has 256MB flash attached to it, and no other disk
13:27.42[g2]At some point I'd like to learn about the dial-in changes
13:28.28[g2]so miau is running well for you ?
13:28.30rwhitbyvery straight-forward, just install the pppd package, and change the options file
13:29.04rwhitbyyep, miau is no-maintenance for me
13:29.23rwhitbyas is the bluetooth access
13:29.44rwhitbyand dial-in only stopped working cause /etc/inittab got overwritten on the last upgrade
13:30.09[g2]the other feature set I'll be playing with is ebtables/iptables and nf_bridge
13:30.09rwhitbywhich reminds me, I should rebuild openslug on nudi to update the feed
13:30.53[g2]I'll be setting up a transparent bridge for monitoring
13:31.03[g2]and saving off all the traffice
13:31.45[g2]so I'll have 3 IP interfaces on the slug
13:38.09rwhitby[g2]: why would you be using a slug for that when you have a revvy?
13:38.47[g2]I'll use the revvie if the slug can't hack it, but at < 10Mbs I think the slug works fine
13:39.11[g2]actually, I think the slug can handle 50+Mbs
13:39.59rwhitbyHmm.  Building on nudi I get: AttributeError: 'module' object has no attribute 'createCopy'
13:40.27[g2]I thought the lastest bk needed the latest SVN of bitbake right ?
13:41.04rwhitbyHmm.  Is bitbake installed globally on nudi or something?
13:41.35[g2]So on the OpenSlug 2.x features etc...
13:41.53rwhitbyyep, I have basically zero input on that
13:42.12[g2]we never did resolve who was going to drive stuff
13:42.25[g2]so you want me to continue with the feature development
13:42.57rwhitbyif you would like to, sure :-)
13:43.19rwhitbyMy push was just to get the beta out ...
13:44.17[g2]Well I'll be working on the list as time permits
13:45.56[g2]rwhitby, On a different subject... What makes up a "real" release i.e., what's triggers the transition from Beta or Release
13:47.11[g2]Ok.. so just like Google that runs Beta's for years we run Beta as long as things are changing
13:47.28[g2]then it's "Released"
13:47.44rwhitbydunno - it's a good question
13:48.13[g2]Ok.. I think 31 July should be the release of OpenSlug and Unslung version.
13:48.32[g2]That marks 1 year for the Yahoo group right ?
13:48.53[g2]or maybe 4 July as "Independence" from Linksys :)
13:49.03jacquesi like the scheme of feature freezes and only doing bug fixes on a "stable" track while going crazy on a "devel" track
13:49.27[g2]jacques, exactly... I think we are at the point where we can do that now
13:49.57rwhitby[g2]: Jan 28th has passed
13:50.15[g2]Jan 28th ?
13:50.36[g2]1 year Wrt54g independence ?
13:50.38rwhitbyJuly 4th ?
13:51.34rwhitby(oops, Jan 26, not Jan 28 - Jan 28 is my son's b'day)
13:52.11[g2]rwhitby, so which pet peeve did I run across ?
13:53.08[g2]a) stable vs. devel, b) "Independence day", c) anniversary of some kind, d) other ?
13:53.19rwhitbyWell, you said "or maybe 4 July as Independence ...", and I said "Jan 26"
13:53.38[g2]Ahh.... Ok
13:54.12[g2]So 31 July could be International Slug Independence day
13:54.39[g2]I think that's when you created the Yahoo group
13:55.11rwhitbythe pet peeve is using a single-nation local holiday as a milestone for a global project
13:55.43rwhitbyif you had said "this summer" you would have tripped over another
13:56.04[g2]"this summer, your winter"
13:56.18jacquesyeah be safe and just say "noon"  :-P
13:56.19rwhitbyand if you had said "how about 7/4/2005 ..." then you would have hit the jackpot :-)
13:57.05[g2]you mean on Independence Day this summer 7/4/2005 :)
13:57.17rwhitbyyeah :-)
13:57.57[g2]That's for sharing as I now know how to date all release event :)
13:59.04[g2]It truly is a great honor to be working on an International project
13:59.19jacquesnoon on Independence Day this summer 7/4/2005
14:03.42[g2]rwhitby, thx
14:03.51[g2]I'll be back later
14:07.35rwhitby-asleepnight all
14:17.34*** join/#nslu2-linux aardbei_ (
15:09.42ka6soxacutally guys its 4/7/2005
15:10.13ka6sox*most* of the world writes it that way...only North America writes it the other way.
15:19.35jbowler[g2]-away I need to talk about a set of changes I have for boot - I've got nfs, disk and ram (all the targets I know about) working fine, but I want to check the changes with you (since I changed turnup and openslug-init).
15:21.21jbowler[g2]-away I've verified that NTP tm_mon change to nslu2-io.c too - it doesn't crash, so I'm sure something is passing uninitialised data fields.
15:22.06jbowlerIf I make that change NTP runs and nothing stops working, but the uninitialised data is still there - haven't had time to track it down yet.
15:22.51jbowlerrwhitby-asleep dyoung-zzzz with my /linuxrc changes the openslug image is no longer vulnerable to rm -rf /initrd
15:23.05*** join/#nslu2-linux ByronT (byrontodd@ByronT.nslu2-linux)
15:24.17jbowlerI've got auto-magic unmount of the initrd to work *once*, but it seems like something manages to prevent ffs2 being unmounted for a few seconds after the pivot
15:25.15jbowlerSo my idea is to have an rcS script 'unmountmnt' to do the umount when it will work (it always seems to work fine after init boot).
15:27.12jbowlerBTW the reason a user might do rmdir /initrd is that it is the standard way with lilo (and other boot loaders?) of auto-umounting the initrd
15:29.37jbowler[g2]-away it would be good to have a known openslug 'next release' number.  Those boot (/linuxrc) changes I have should be documented on the wiki, but the documentation needs to say 'only applies to bitbake builds and openslug >=1.20' (or something like that).
15:58.47*** join/#nslu2-linux Jacmet (
16:01.56*** join/#nslu2-linux _chris_ (
16:12.42*** join/#nslu2-linux llagendijk (
16:14.56*** join/#nslu2-linux phedny (
16:59.44*** join/#nslu2-linux [cc]smart (
17:47.05[g2]-awayjbowler, Sounds all like Excellent work.  I'll have to look at the current release number and we should update it and make a new -rc (release candidate)
17:47.54jbowlerI'd prefer at least a few days review before the /linuxrc stuff goes live
17:48.04jbowlerI.e. review by dev, not users who might complain ;-)
17:48.40[g2]-awayjbowler, Ok... historically, the repo is up for new changes
17:49.30jbowlerRight.  There are a couple of significant bugs still in there - the 'shutdown -r now' doesn't work problem (it hangs, somewhere)
17:49.36[g2]We obviously test stuff, but for forward progress we don't want to wait too long or progress is impeeded
17:49.51jbowlerAnd the other (well, this is minor) is that some startup scripts aren't running when they should.
17:49.58[g2]the reboot and shutdown -r is already there.
17:50.30[g2]Someone even sees it with reboot on the avila mailing list which has nothing to do with our code
17:50.44[g2]the guess on that list we either busybox or the mtdcode on jffs2
17:51.20jbowlerYes: it's an OE problem. isn't running for example.  I just checked in a fix for the portmap-doesn't-get-killed problem.
17:52.18jbowlerOk.  I'll throw the big switch on the /linuxrc change now - it will be checked in in a few minutes...
17:52.34jbowlerWhat about the NTP problem?  My one liner hides it fine.
17:54.00[g2]jbowler, so with the NTP issue, is there a change required in the nslu2-io.c ?
17:54.17[g2]I missed the tm_mon part
17:55.28jbowlerYes: it's just in there.  Indeed, the underlying bug must be in there too.  I think I know where it is, but you seemed to need a work round ASAP!
17:56.00jbowlerIf I check in the one-liner (and if I'm right) the symptom will disappear, but I can still debug to find the real problem.
17:57.10[g2]Ok... so commenting out x1205-rtc in the OpenSlug config and leaving nslu2-io.c alone *doesn't* fix the NTP problem
17:57.48[g2]you got an idea of what is causing the problem in nslu2-io.c but haven't fully verified it yet
17:57.49jbowlerAargh, sorry, I just got that wrong - the one liner is in x1205-rtc.c, nslu2-io.c needs no change
17:58.29jbowlerStopping x1205 running certainly fixes the problem, but it stops the leds too and that's bad from my viewpoint
17:59.04jbowlerOh, I think I understand dyoung's original change too - it caused the 1 month problem, but it was probably necessary because of the same underlying bug.
17:59.40jbowlerI have x1205 without the 1 month problem and without the crash (uptime 28 hours)
17:59.42[g2]For just the LED code we could remove the timer connection right ?
18:00.10jbowlerYes, but we don't need too - I'm sure the change I have to x1205 is sufficient.
18:01.06[g2]Ok... You're doing great work getting to the root issues of several of these problems
18:02.01[g2]I've been up for 3 days 14 hours+ with the more drastic fix
18:02.10[g2]the drift does keep getting bigger
18:02.30[g2]I'm losing about 1ppm a day
18:03.15jbowlerI keep needing to reboot to test the changes ;-)
18:03.24jacqueswe all need more slugs
18:03.31jbowlerLosing 1ppm/day sounds bad.  My drift cycled around -10 to +6
18:03.43[g2]So I don't know if there will be a big adjustment sometime or it'll just take 500 more days until we get 512ppm limit
18:04.24jbowlerI don't think there is a practical limit - it should keep working fine.  If it hits 10000 it might give up...
18:04.48[g2]actually I think 512 is the limit
18:05.01[g2]but I may be confused
18:05.50jbowlerAt some point it stops using adjtime (to slow up/speed down the clock) and starts stepping.  512 is only +5 to tickadj, so that sounds small.
18:06.25jbowlerMaybe you are seeing temperature drift?  My slug hasn't been off for more than a few seconds in the last several months, so it might have reached a stable temp.
18:06.40jacqueswhy are these things so much less precise than a $3 digital watch?
18:07.12[g2]because you don't check your watch to 10 places past the decimal point :)
18:07.51jbowlerLast digital watch I had lost +/- 1s/day - thats 11ppm
18:08.08*** join/#nslu2-linux sp4ce (
18:08.22jacquesso slugs are infact better?
18:08.33jbowlerI gave up on watches, if I want to know the time I look at where the sun is.  Reliable, used for 10000's years, never failed yet.
18:08.58jacquesfor example, right now it's "day-time"
18:09.03jacquesI can tell by looking outside
18:09.51ByronTyou can see outside?!?!!?
18:10.14ByronTmy office has no such openings to the outside world.... :(
18:11.19ByronTjacques, my PS2 output looks incredible on that 2005FPW
18:12.05jbowlerI have a chainsaw.  It'll cut through 80% of the interior walls in US offices (i.e. the ones which aren't concrete.)
18:12.38jacquesByronT, excellent!
18:12.47jacquesByronT, s-video?
18:12.52ByronTbut my boss might not let me in the door with my Stihl chainsaw
18:13.31ByronTjacques, I haven't found the PS/2 accessory for s-vid out
18:13.42ByronTand no Beast Buys in town
18:14.16jacquesByronT, composite ??
18:14.35jacquesimagine what s-video would look like :-D
18:14.46ByronThey, I'm working up to that
18:15.23ByronTI let the wife play GT4 and she said "We really need the steering wheel add-on for this"
18:15.33ByronT"need" :D
18:15.38jacquesnice :-)
18:17.37jacquesI'm messing with gentoo on my t42
18:17.47jacquescpufreqd is cool
18:17.49ByronTwireless working yet?
18:18.06jacqueshaven't tried that yet
18:18.21jacquesit worked fine under ubunto 5.04 so I'm not really worried
18:18.21[g2]what cpufreqd ?
18:18.55jacques[g2], it changes the processor speed based on rules you can configure
18:19.25jacquesyou can tell it what speed to run when certain apps are running, on AC, batt, etc
18:20.03[g2]Hey I'm looking for a DVD writer for disks like the Sluggix release
18:20.25[g2]anyone know what the best format is or a good device ?  
18:20.44jacquesyou checked ?
18:20.49[g2]Is -R the most common ?  I imagine 4X is pretty popular
18:20.53[g2]jacques, not yet :)
18:20.54jbowlerThe latest writers support pretty much all formats
18:20.56ByronTDVD-R is the most common
18:21.22jbowlerFor releases DVD-R (it's write once, and tends to be cheaper)
18:22.00[g2]And does anyone have thoughts on which media they like ?
18:22.40jbowlerI used SONY writers, so I try to buy SONY media.  Compatibility seems to be a real issue...
18:23.07ByronTI've never had problems with imation media
18:26.21jbowler[g2] seems to work fine, but isn't in the build (i.e. doesn't get added to the ipkg.conf), is there a reason?  (like, maybe, the danger of overloading the server)
18:26.51[g2]jbowler, I was wonder the same thing. I think it should be there too
18:26.57[g2]setup as default
18:27.06[g2]we can check with rwhitby-asleep and ka6sox
18:27.23jbowlerI'm tempted to add it - I've got the change locally - but we do need to double check...
18:27.39jbowlerping ka6sox
18:27.43[g2]actually, just add it
18:27.54jbowlerWe can also remove it...
18:27.56[g2]I mean there are what 20-50 active OpenSluggers ?
18:28.04[g2]nod on the remove
18:30.20jbowlerOk, changesets up to 1.3327, which is the ipkg.conf add.  Phew, that's all my changes safely checked in...
18:31.44jbowlerAnd I'm running just fine off an NFS root (convenient 'cause the /var/log/messages file is remote readable!)
18:31.47[g2]jbowler, Ok so you've just checked it all in ?
18:32.22jbowlerYes, everything - NTP, /linuxrc, ipkg.conf, portmap fix
18:33.31jbowler8 separate commits, complete rebuild required (though it will probably boot ok without because the flash boot is fairly robust)
18:35.48[g2]jbowler, did you see the stuff on the OpenSlug roadmap ?
18:35.50jacquesanyone know where "laptop mode" hides in kernel config?
18:35.52[g2]jacques, too ?
18:36.08jacques[g2], yeah I saw it last night when you put it up - looked good
18:36.16jbowler[g2] - yes, I read that this morning.
18:36.42[g2]any suggestions or comments ?
18:37.19jbowlerThe nslu2-io driver has been worrying me.  I can't get the LEDs to flash in sync because of the separate timers.
18:37.47*** join/#nslu2-linux Jacmet (
18:38.02[g2]jbowler, I think it'll be fun to re-write that driver
18:38.03jbowlerBut the nslu2-io.c code is pretty good, the main thing is just missing functionality in the buttons
18:38.32[g2]or adapt/enhance would be a better choice of words
18:38.39[g2]the drive is a great start
18:38.59jbowlerAre you going to do that?  I changed leds to be able to futz with all the LED ioctls
18:39.37[g2]Is much lower on my TODO list, eventually I'll play with it but it's like 6-10
18:41.25[g2]The top 3 (not counting testing like jb's latest) are 1) The Knoppix CD/DVD, 2)Native compile environement fixups, and 3) APEX and (RedBoot bump to 64MB)
18:42.53[g2]do you guys know much about uniionfs ?
18:47.54[g2]jbowler, OK.. the changes from the latest pull just built.
18:48.11[g2]I haven't dl'd yet, but all builds
18:51.08jbowlerunionfs: it should work that way, you probably need to ensure that the DVD file time stamps remain correct (i.e. as in the original build)
18:51.57jbowlerProblem is bitbake isn't truely incremental - it tends to destroy the complete work/package and rebuild.
18:52.40jbowlerThings like unionfs can fix that if the file system is written to realise the file data is identical, but the file inode changes because of the timestamp.
18:55.08jbowler[g2] type 'turnup' (just on it's own) to get the help message ;-)  'turnup disk -i /dev/sda1' is probably what you need
18:55.40jbowlerLEDs: 'red' means inside the kernel boot
18:55.57jbowler'flashing amber' means running rcS (startup scripts)
18:56.13jbowler'flashing amber/green' should be next, but there's a problem inside sysvinit
18:56.27jbowler'green' means (as before) that the rc5 scripts have finished
18:57.58jbowlerIn general flashing means 'changing mode', 'amber' means S,0,1 or 6, 'green' means 2,3,4 or 5.  During a transition from one runlevel to another the LEDs should flash between the two settings, or flash on/off if they are the same.
18:58.45[g2]jbowler, SUPER! Makes a bunch of sense, rwhitby will be jealous
19:00.28jbowlerThanks, it really helps me with the bootstrap cause now I know where its failed (sometimes)
19:01.25jbowlerIn fact it's possible to output error codes on the disk1/2 LEDs, try 'leds +1 !2' for example...
19:01.49jacquesmorse code?
19:02.20jbowlerI thought about that, but it turns out that the power/disk LEDs have 90 different combinations, so static signals are fine.
19:04.33jbowlerOh, 'leds' (no arguments) puts them back in a sensible state.
19:04.34[g2]hey jbowler can you use that mega-brain of your to help me with another problem ?
19:04.57jbowlersure - what is it?
19:05.37[g2]when I do a shutdown -r now, the box doesn't power down like it should... That's because we probably are not pulling the GPIO8
19:06.03[g2]however, it seems to re initialize the USB driver which is wrong
19:06.36[g2]I'm curious what that is happening and we should fix that and add the GPIO 8  as the power down spot
19:06.53jbowlerWas the re-init of USB happening before my latest set of changes?
19:06.55[g2]I'm more curious about the re-initialization
19:07.09[g2]yeah.... this has been happending for ever
19:07.24[g2]this is unrelated to your changes
19:07.44jbowlerOk.  shutdown -r now works maybe 50% of the time with flash-only mounts, but only 10% with disk or nfs mounts.
19:08.06jbowlerIn the non-working cases the slug hangs - the kernel doesn't evern respond to the power button.
19:09.05jbowlerI don't have a serial, and the syslogd is shut down before this happens, so I didn't know about the USB re-init,  I suspect something hasn't died and re-accesses the USB when it shouldn't, maybe.
19:09.55[g2]you're not serial enabled at all ?
19:10.08jbowlerThe portmapper was never dying - two bugs, both now fixed - but that doesn't help.  The file systems are never properly umounted - bogus OS shutdown scripts.
19:10.35jbowlerserial: no - that's why I'd probably go for one of your slugs with the serial/JTAG (that an APEX)
19:10.39jacqueshow can I tell which scheduler the kernel is using?
19:11.18jbowlerjacques: it's in the kernel .config, there's a setting for the default scheduler.
19:11.36[g2]speaking of which I need to check on the status of the 1st articles
19:12.40jacquesjbowler, you;d think it would be in /proc or /sys somewhere
19:14.11[g2]there's all kinda stuff in there like you usb stick/disk serial number
19:14.16jbowlerThere's some way of changing it from user space, but I can't remember the details.  I can't seen anything in /proc, there may be a device
19:14.44*** join/#nslu2-linux kitno455 (
19:16.46[g2]kitno455, hey WB
19:16.51jbowler[g2] shutdown -r now: I just did the absolute minium boot/shutdown sequence (from flash, no disk attached) and it worked fine, but apparently if I do anything with mounts it hangs in the reboot sequence.
19:16.52kitno455been ages
19:17.11kitno455just downloaded openslug beta
19:17.26kitno455what do i need to be aware of before i load this :)
19:21.38[g2]kitno455, do you have a HD / flash attached ?
19:24.41kitno455just a 256 meg flash stick
19:25.01[g2]do you have anything important on it ?
19:25.44[g2]then I'd pull the flash stick and reflash the image and then you can ssh into the unit
19:26.20kitno455does it overwrite redboot?
19:26.29[g2]You can put the flash stick in and tweak "turnup" and it's allow you pivot on to the flash stick
19:26.51kitno455sorry, i keep hearing turnup mentioned, but i have no idea what it does
19:26.55[g2]no not that version, it's like the old stuff that just replaces the kernel and rootfs partitions
19:27.26[g2]we you get in, "which turnup" then cat "...path/turnup"
19:27.40kitno455shell script?
19:27.47[g2]nod. like 15 lines
19:27.56kitno455pivot root, etc?
19:28.24[g2]no it's the setup to allow the pivot root from the /linuxrc in the jffs2 partition
19:28.47[g2]jbowler, I'm up on the latest image
19:28.50kitno455ok, you cant use that if jffs is overmounted, right?
19:28.55[g2]I've got my green light back :)
19:29.18[g2]I'm not sure what you mean by overmounted
19:29.31kitno455uh, if turnup mods contents of jffs2root
19:29.45[g2]kitno455, and you know the default root pw right ?
19:29.47kitno455then you can only run turnup with usb devices disconnect
19:30.28[g2]kitno455, You plug in the flash stick / hd then turnup
19:30.42kitno455but after you have booted from internal flash
19:30.59kitno455if i want to re-run turnup later
19:31.05kitno455i have to unplug flash/hd?
19:31.25kitno455or is internal flash always mounted somewhere?
19:31.37[g2]no, You'll need to mount and remove the .sda1root from the jffs2 and reboot
19:32.40kitno455how does .sda1root relate to /linuxrc?
19:32.57jbowlercat /linuxrc - it just checks for .sda1root
19:33.15kitno455why not use turnup to re-write /linuxrc?
19:34.11jbowlerNow shakspeare said something about 'great minds think alike', I just did that ;-)  (hum, then he said 'fools seldom differ')
19:34.25kitno455yes- fools seldom differ :)
19:34.43jbowlerStill, that's what [g2] is trying at this very moment...
19:35.05kitno455had some trouble with BK, so i will try what is in the binary 1.12 now....
19:35.36jbowlerRight, the description above applies to the beta, bk is quite a lot different when booting
19:36.11kitno455ok, tearing it open now....
19:36.26[g2]jbowler, looks like turnup has changed a bit :)
19:39.40jbowler:-) help is now longer than the original...
19:39.44kitno455prefered flashing method for openslug? i usually tear it open and serial into redboot and fipswrite, etc, but i want to try something easier this time....
19:40.43jbowlerupslug is my preferred method - needs a network which doesn't drop packets though
19:41.01[g2]some ppl upslug from redboot, I tftp in an old linksys kernel and ramdisk and the flash from the UI
19:41.18[g2]that method won't be long for this world
19:42.27kitno455ok, i got a decent network, now to find a d*mn catV cable....
19:42.28jbowlerturnup ram (bk version) works now, so reflash from a command prompt is possible.
19:43.07[g2]so you automatically create a ramdisk copy stuff over then umount the jffs2 ?
19:43.20kitno455turnup ram (bk) runs from ramdisk of internal flash or external?
19:43.47[g2]I'd guess internal
19:44.13[g2]I like the addition of the -s BTW
19:44.25[g2]so for the flash stick it's
19:44.46jbowlerIt's as in unslung and the original openslug linuxrc (but without the unslung specific stuff) - it makes an ext2 ramdisk, copies the flash to it and pivot_root
19:45.11[g2]turnup memstick -i -s 2 /dev/sda1 ?
19:45.18jbowler-s: yes, my Maxtor disk doesn't need it.  Annoying to wait 10s just because someone has a slow disk ;-)
19:45.35kitno455makes the ext2 ramdisk from the contents of internal jffs2?
19:45.49kitno455at boot time?
19:45.52jbowlerNo space (sorry, cheap command line parsing) "turnup memstick -i -s2 /dev/sda1" (untested)
19:46.10[g2]np. sound good to me
19:47.50jbowlerIn fact the 2 second delay shouldn't be necessary (and, I just realised, I didn't test the memstick command at all, but it just prepends -m to the corresponding disk command)
19:48.24[g2]it ran
19:48.41[g2]should it create .sda1root or is that someplace else now ?
19:48.52kitno455in the /linuxrc :)
19:49.30[g2]ok so linuxrc is re-written and the old one is copied to linuxrc.sav
19:49.50kitno455wow, i actually knew what was going on for once :)
19:50.12jbowlerThe old one will be a symbolic link in this case.  The /linuxrc is only a few lines (about 2).
19:50.34[g2]I just read that :)
19:50.50[g2]but the link and the 82 bytes :)
19:51.22[g2]it's exec'ing /boot/disk though, should that be /boot/flash ?
19:52.03jbowlerThose names are maybe confusing, /boot/flash means, strictly, the Flashdisk partition, /boot/disk means something on the bus - which needs to be pivot_root'ed
19:52.51[g2]then just a shutdown -r now for the reboot  ?
19:52.54[g2]or do you reboot ?
19:53.58jbowlerEither, or pull the plug.  turnup mounts and unmounts the new partition - so unless you have it mounted somewhere else it is safe.
19:54.45jbowlerIt will probably hang on the shutdown -r now if you have done anything other than the turnup (and maybe even then), so you may want to just press the power button ;-)
19:54.46[g2]shutdown -r worked that time
19:56.07[g2]you're automatically loading the V/FAT modules right ?
19:56.46*** join/#nslu2-linux sjmelhuish (
19:56.46jbowlerYes - but if you don't want whiney messages for a non-fat disk try 'turnup disk -i /dev/sda1 -t ext2' (then it won't try fat before ext2)
19:57.22[g2]so close
19:57.33jbowlerwhat happened?
19:57.34[g2]rootfs / rootfs rw 0 0
19:57.34[g2]/dev/root /initrd jffs2 rw,noatime 0 0
19:57.34[g2]/dev/sda1 / ext2 rw 0 0
19:58.00[g2]the remount is before the pivot :)
19:58.30[g2]it's NBD (no big deal)
19:58.43jbowlerI don't understand
19:58.43[g2]I'm very pleased with all the improvements
19:59.36[g2]jbowler, I "turnup memstick -i -s2 /dev/sda1"
19:59.57[g2]which should have been turnup memstick -i -s2 /dev/sda1 -t ext2
20:00.04[g2]in my case
20:00.19CIA-103sjmelhuish * 10unslung/make/
20:00.19CIA-1Tidied up staging rules. moved to bin directory
20:00.23[g2]I was expecting that to mount the /dev/sda1 with the noatime option correct ?
20:01.12*** part/#nslu2-linux sjmelhuish (
20:01.17jbowler[g2] yes and ramfs on /dev etc
20:02.13jbowlerThat output looks like the memstick was totally ignored.  Can you cat the /linuxrc?
20:02.57[g2]there's no linuxrc on the memstick, you mean the one in the jffs2 right ?
20:03.24[g2]sleep='2' exec '/boot/disk' '/dev/sda1' '-o' 'noatime'
20:03.24[g2]exec /boot/flash
20:03.54jbowlerMaybe I've got a bug in boot/disk...
20:04.36jbowlerNope.  I think the noatime was passed in.  I've noticed it doesn't appear on NFS mounts either.
20:06.29[g2]Ok... that'll need to be looked into
20:06.40[g2]two other questions
20:06.44jbowlerTry 'mount -o remount,noatime /'
20:07.00[g2]I did and it does showup in /proc/mounts :)
20:07.20[g2]actually my command was mount / -o remount,noatime
20:07.58jbowlerYes, same for NFS.  This is weird because all the other options in an NFS /linuxrc are there (and they are not the default!)
20:08.15[g2]jb I'm really stoked about all these changes in turnup
20:08.41jbowlerThat's good - I was worried they might be excessive.
20:09.03[g2]I think there are some very well thought out changes
20:09.28[g2]great initiative and brain power
20:09.51kitno455dude, he's married
20:10.09[g2]I've never run expect with ssh (I'd imagine you can), but we can setup full regression testings
20:10.23[g2]which is a little down on the list but before 6-10
20:10.43kitno455ok, cant seem to do a cvs co from to get upslug
20:10.53kitno455alternative location?
20:14.57jbowler[g2] /etc/init.d/ is killing the noatime, for the busybox mount "mount -o remount,ro /" kills noatime
20:16.12jbowlerI need to futz the /etc/fstab too for the root partition so that it has noatime
20:18.28[g2]Ok.. so I re turnup now ? with turnup memstick /dev/sda1 -t ext2 ?
20:19.48jbowlerI entered it as slugbug 150, unslung has a similar problem (there's already a slugbug for that)
20:20.13[g2]hey shutdown -r worked again from the flash stick
20:20.39jbowler[g2] that re-turnup should be fine.  Edit /etc/fstab manually to have the 'noatime' option and that should workround the bug.
20:21.08[g2]I"ve already rebooted from the returnup :)  Will do on the /etc/fstab
20:21.35[g2]So does it matter at all whether the flashstick is ext2 or ext3 ?
20:21.50jbowlerStill, it looked like the tmpfs settings had got lost - what is in your /etc/fstab at present?
20:22.14[g2]cat /etc/fstab
20:22.14[g2]rootfs          /               auto    defaults        1  1
20:22.14[g2]proc            /proc           proc    defaults        0  0
20:22.14[g2]tmpfs           /var            tmpfs   defaults        0  0
20:22.14[g2]tmpfs           /media/ram      tmpfs   defaults        0  0
20:22.14[g2]usbfs           /proc/bus/usb   usbfs   defaults        0  0
20:22.38jbowlerYou want ext3 I believe - because the journalling will probably reduce the writes, but I don't know for sure.  You could try jffs2 on it, that should work too.
20:23.28jbowlerWacky, that /etc/fstab is fine - are they in /proc/mounts (/var and /media/ram?)
20:23.32[g2]mount: Mounting /dev/sda1 on /tmp/rootfs.842 failed: Device or resource busy
20:23.32[g2]turnup disk: /dev/sda1(-t ext3 -o noatime): unable to mount device on /tmp/rootfs.842
20:24.00kitno455dudes, i have the biggest smile on my face.
20:24.17[g2]kitno455, you're gonna marry jbowler  ?
20:24.26kitno455i just loaded openslug beta for the first time, and this rocks
20:24.38kitno455no more linksys bs
20:24.43jbowlerNo, it's you - he's using the beta ;-)
20:24.50kitno455real linux box
20:25.16kitno455ok, how dangerous is redboot in the stock config
20:25.28jbowlerre-turnup: yes, I guess it's already mounted as ext2 on /, so trying to remount ext3 won't work.
20:25.33kitno455would you put an openslug box on an untrusted net?
20:25.53[g2]kitno455, not yet... but soon
20:26.01[g2]APEX works
20:26.08kitno455what is current status of usb in apex?
20:26.19jbowlerOk, so turnup to a different parition, reboot, then try the ext3 thing - I wouldn't want to permit '-f' in this case because it might not mount -t ext3 because there is no journal.
20:26.33[g2]We're working on booting the kernel from the /boot/zImage in jffs2
20:26.44[g2]then all the legacy stuff goes *poof*
20:26.51kitno455one contiguous partition...
20:26.56[g2]and it's just APEX and the rootfs.jffs2
20:27.06jbowlerkitno455 - untrusted net - not until you've changed the root password!  Even then an NSLU2 is vulnerable during redboot init.
20:27.11kitno455how will apex get network params
20:27.12[g2]one 7.75MB partition :)
20:27.28kitno455or does apex not do net?
20:27.38[g2]apex is serial only
20:27.44kitno455right, fine with me
20:28.01kitno455ok- here's a question
20:28.03[g2]OpenSlug used to pickup the net config from sysconfig
20:28.20kitno455sysconf will be gone with apex
20:28.25[g2]I'll have to check out the uber turnup and see what's changed
20:28.26kitno455where is macadd stored
20:28.51[g2]Ok. there's only a couple unique bytes in Redboot.
20:29.05[g2]Actually they reside in the last 144 bytes
20:29.14jbowler[g2] - it reads SysConf once to make /etc/default/sysconf, then it reads /etc/default/sysconf.  That can be editted to change the boot.
20:29.31kitno455last bytes of flash, or of redboot partion?
20:29.44jbowlerRedboot partition.
20:29.54kitno455apex going to leav it there?
20:30.47[g2]jbowler, is the sysconf just the same sysconf less the 4 byte size at the beginning ?
20:31.24jbowler[g2] It gets cleanup up, I haven't tried writing it back yet (to the SysConf partition) but it should work...
20:31.48kitno455i could use another type of turnup option- want to hear it?
20:32.00jbowlerI don't think RedBoot every reads it.
20:32.02[g2]cleaned up how other than removing the length bytes at the beginning ?
20:32.15[g2]no RedBoot doesn't read it
20:32.19jbowlercleanup: remove blank lines, correct lan_interface to eth0
20:32.46[g2]so we could probably recreate a valid sysconf from it
20:33.29[g2]so I'd like to add the following where we save off the last unique 144 bytes of RedBoot and check the checksum on the first 265K
20:33.53kitno455how about vfat partition on memstick, with an ext2 fs on it, that gets ramdisked?
20:33.58[g2]Everyone I've asked has the same MD5SUM
20:34.21jbowler[g2] try this: devio '<</dev/mtdblock1;cpb;fb1,10' | diff - /etc/default/sysconf
20:35.18jbowler[g2] md5sum: that's excluding the last 144 bytes?
20:35.55[g2]yes dd if=/dev/mtdblock0 bs=1K count=256 | md5sum
20:36.27[g2]dd if=/dev/mtdblock0 bs=1K count=256 | md5sum
20:36.27[g2]dd: invalid number `1K'
20:36.27[g2]d41d8cd98f00b204e9800998ecf8427e  -
20:36.36kitno455lower case k
20:36.58[g2]dd if=/dev/mtdblock0 bs=1k count=256 | md5sum
20:36.58[g2]0a07200514fed35911bce0e777506629  -
20:37.02jbowler9f6e958a79ed01398cbfe1bfc1531860  -
20:37.25[g2]jbowler, that number look familiar
20:37.31kitno455e991b3b5540694eb04ce6d2b17914cc9  -
20:38.10jbowler[g2] try these two commands:
20:38.13jbowlerdevio '<</dev/mtdblock0;cp 256,1024*' | md5sum -
20:38.29jbowlerdevio '<</dev/mtdblock0;cp 256,1024*144-' | md5sum -
20:38.58[g2]ok I think I did this before
20:38.59kitno455jbowler, what do those do, just pull ranges from that device?
20:39.23[g2]dd if=/dev/mtdblock0 bs=1000 count=256 | md5sum
20:39.24jbowleryes, easier than dd because they do the correct blocking
20:39.48kitno455hence the 256?
20:39.59kitno455or the 1024?
20:40.03jbowlerRPN arithmetic, but it doesn't affect the block size
20:40.22kitno455right, ok
20:40.23jbowlergotta go... biab
20:41.10kitno455i am simply amazed at the amount of effort this group has expended to make this a really usable device.
20:41.25[g2]it's been a bunch of fun
20:41.32kitno455any company with half a brain would kill to have this done for free
20:41.47[g2]I've got to head out for a bit
20:41.52kitno455me too.
20:41.56[g2]actually, I don't think that's the case
20:42.18kitno455gotta sell more slugs
20:42.21[g2]I think we've got more ppl working on nslu2-linux than the company has employees
20:42.27kitno455thats what i mean
20:42.36[g2]they've only got 156 employees
20:42.47kitno455they use outside contractors
20:42.50kitno455for devel
20:42.54kitno455and they are part of cisco
20:43.01[g2]oh... My bad... I mis read you statement
20:43.04kitno455which takes over their hr, payroll, etc
20:43.29[g2]The reason they "can't" and "won't" is that this commoditizes their products"
20:43.44[g2]not to mention makes them open
20:43.54kitno455they see value add in their crappy softwar3
20:43.59[g2]that's why [g2] is starting to build HW :)
20:44.15kitno455you want your bare slug board back?
20:44.21kitno455hang on the wall?
20:44.27[g2]first the rework boards then full-on custom :)
20:44.44[g2]nope. But I do have some questions for you
20:45.04kitno455i will get the board and bring it here tomorrow if i cant answer now
20:45.05[g2]I must run now. I'll e-mail or catch you in the channel
20:45.11kitno455you got my email
20:45.16[g2]think so
20:45.16kitno455thats better than irc for me
20:45.31[g2]thx for stopping by and all your help in the past!
20:45.39kitno455more help coming now i think....
21:38.35*** join/#nslu2-linux sjmelhuish (
21:40.44*** join/#nslu2-linux Christopher (
21:42.46ChristopherI was thinking about ways to increase the amount of memory for the slug w/o soldering on more ram chips and thought about creating a RAM swap. This would be ram chips connected to one of the USB ports that would be the swap space. This would reduce wear and tear on the harddrive as well as decrease swap access times
21:43.07Christopheranybody have any ideas on this?
21:43.42dyoung-zzzzwe beat that one around a while ago.
21:43.54dyoung-zzzzthe general consensus was "doable".
21:45.11Christopherhas anyone attempted it?
21:48.02dyoung-zzzzI dont remember what the hangup was.
21:48.27dyoung-zzzzand afaik, noone actually attempted it.
21:48.50Christopherok, thanks for the info I might look into some if I get the motivation and time
21:50.37dyoung-zzzzWhens your internship start?
22:05.09*** join/#nslu2-linux nslu2-log_ (dyoung@dyoung.nslu2-linux)
22:09.24*** join/#nslu2-linux tking_ (
22:14.22[g2]Christopher, I'm having a professional company modify replace the memory on the slug. It's really straight forward work and the cost will be extremely reasonable
22:14.47[g2]for the performance gain, I'm sure it'll be worth it.
22:15.09[g2]I'll also be doing the same on the Linksys RV units that have faster processors
22:15.45[g2]the cost for the RAM chips is like $8.40 per chip in 25 piece quantity
22:16.28[g2]iirc the memory bus runs at 133Mhz x 32-bits (16 + 16 from the two chips)
22:17.27[g2]the USB 2.0 interface is first through the PCI bus which is 33 x 32 not to mention the USB controller can't handle full USB 2.0 rates.
22:17.51[g2]That would be further complicated because it'll be sharing with disk access.
22:19.46dyoung-zzzzheh, I never said it was practical... just doa ble.
22:20.23[g2]Well I've got a 300G HD that has a 16MB cache
22:20.51[g2]If I put 2 of them on there and only had a 16MB partition everything would reside in the cache
22:30.38*** part/#nslu2-linux sjmelhuish (
22:42.20*** join/#nslu2-linux louis_lagendijk (
22:42.33*** part/#nslu2-linux louis_lagendijk (
23:32.48jbowler[g2] Christopher: Lennert Buytenhek posted this message:
23:34.21jbowlerThose tests get 193 clocks for each uncache SDRAM read, ok that's a 600MHz ARM, but it's still scary, the IRQ handling stuff is, of course, more so
23:35.13jbowlerThe cache seems to manage 32bits in 2 clocks, so get the process to fit in a cache is a real win over everything else.
23:35.20[g2]jbowler, I had read that. You are basically talking about on-chip or off device access
23:35.52[g2]the on-board access to some areas in the drivers
23:36.10[g2]maybe a status somewhere
23:36.44[g2]It would be interesting to dump the virtual memory map tables and see where and how large those areas are for us
23:36.51[g2]and another data point.
23:37.27[g2]You know my OEM box an IXP422 calculates 266 BogoMIPS right  ?
23:37.58[g2]basically, the identical sw with some very minor tweaks
23:38.14jbowlerI don't believe that figure - I've seen it, but stop watch timing consistently gives 2 cycles per 32 bits.
23:38.51[g2]I don't have access to any high-speed equipment yet
23:39.30[g2]with some luck I'll be getting some high speed analysis equipement in the next couple months
23:40.02[g2]I've heard of a friend that has a cool usb 2.0 logic analyzer that does 132 pins @ 500Mhz
23:40.08[g2]I'm trying to get some details on it
23:40.40[g2]but basically, one code sit on the memory bus and watch the timing
23:42.53[g2]Hard to believe that was almost 8 years ago... the MPC860 B to C steppings ...
23:43.26jbowlerThe point about the 16MByte cache is well made - I'm sure that's enough to factor out disk speed (i.e. reduce it to the issue of USB speed)
23:45.09[g2]my understanding of the USB issues is that the NEC may have eratta that it can only run at half speed
23:45.45[g2]bacically, my understanding is that the chip is not dual ported enough to talk to both the PCI and USB bus at the same time
23:46.25[g2]therefore, it can only keep about half of the 13 microframes full during high-speed xfers
23:46.51[g2]I'd love to get some resolution on this
23:47.08[g2]I'm waiting on geting an IXP425 development board
23:47.31[g2]They have actual PCI slots so I'll be able to plug in PCI cards and really do some testing
23:47.47[g2]I'm not sure whether that PCI bus can also run at 66Mhz
23:48.09[g2]But I've got some very cheap SATA controllers I'm gonna try

Generated by by Jeff Waugh - find it at! Modified by Tim Riker to work with blootbot logs, split per channel, etc.