**** BEGIN LOGGING AT Fri Feb 11 23:59:56 2005 Feb 12 00:00:13 my dad on maui uses linux Feb 12 00:00:36 so does my mom most of the time now :) Feb 12 00:01:01 thats neat. Feb 12 00:01:38 Nah there are a bunch of linux folk here; but nothing is organized. Feb 12 00:03:05 hmmm... sounds like an opportunity Feb 12 00:04:48 hmm I'd go to an expo in HI Feb 12 00:05:30 Hmm Feb 12 00:05:42 Maybe when I'm feeling less lazy I'll talk to some folks and see if theres interest Feb 12 00:08:18 youd probably get lots of west coast ppl Feb 12 00:08:37 it's far and expensive for east coast and europe tho Feb 12 00:08:37 ka6sox, can you snag me a programme? Feb 12 00:09:07 I'm a little curious about what kind of vendor sponsorship they had/have. Feb 12 00:09:18 I'm determined to make it to HI while I'm on the west coast Feb 12 00:09:45 dyoung: programme from the Conference? Feb 12 00:09:48 yeah Feb 12 00:09:49 okay I will. Feb 12 00:10:12 jacques, when I was dyoung-at-sea, I got a LAX-HNL flight for $120. Feb 12 00:10:13 cpage_: remind me. Feb 12 00:10:24 so it can be super cheap. Feb 12 00:10:55 dyoung yep well within my range there Feb 12 00:11:13 heck I paid $400 to fly to OK for Christmas :-\ Feb 12 00:11:21 Ow. Feb 12 00:14:54 LinuxChix? Feb 12 00:15:12 (new rock band?) Feb 12 00:15:16 Are you SBLUG? Feb 12 00:15:21 yes Feb 12 00:15:45 Can I join? Feb 12 00:15:49 yes Feb 12 00:16:15 I am the InstallFest/LAN party Chairman. Feb 12 00:16:25 InstallFest? Feb 12 00:16:29 Sounds spooky. Feb 12 00:16:43 network installation server? Feb 12 00:16:52 you mean you haven't done an installfest? Feb 12 00:17:02 Please Define. Feb 12 00:17:37 ppl bring their machines and get linux installed on them Feb 12 00:17:47 people come to a defined place with their computers and we install linux on it. Feb 12 00:18:35 Geez, the sblug webpage seems to have a lot of grafitti. Feb 12 00:18:48 comments section in the installfest section. Feb 12 00:18:52 huh? Feb 12 00:21:55 okay fixing now. Feb 12 00:22:44 (actually I'm asking the site manager to fitz Feb 12 00:35:59 we now have $860 to spend on the build machine. I'm looking for recommendations for $800 or less (keeping the $60 for domain fees, etc) Feb 12 00:36:39 jacques, Tie, ka6sox-away, others: what's the plan for purchase? Feb 12 00:37:38 the machine I spec'ed was only about $500 - but the extra $300 could get us a lot more power I think Feb 12 00:38:03 for $500 we can get something that will build all cross packages in 90 minutes Feb 12 00:38:16 90min isnt horrible. Feb 12 00:38:32 what sort of memory does the jacques machine require> Feb 12 00:38:49 2x512MB DDR Feb 12 00:39:01 at least 2700 Feb 12 00:40:08 can we put a XP2700+ in there? Feb 12 00:40:15 we will make a decision in the next hour or three, so put your bids in now :-) Feb 12 00:40:47 sure, I spec'ed a XP2500+ anything up to XP3200+ will work in the same motherboard Feb 12 00:40:52 1GB is enough. Feb 12 00:41:07 I was thinking about something we might be able to bitbake on Feb 12 00:41:09 is that a dual-channel MotherBoard? Feb 12 00:41:10 but with the extra money we might wanna look at Athlon64 Feb 12 00:41:17 nod Feb 12 00:41:27 ka6sox-away, yes nforce2 ultra400 Feb 12 00:42:42 ka6sox-away, ah I see newegg has a good deal on the XP2700+ right now - must be what you meant Feb 12 00:43:27 yes, since we have $800 then we definitely want to be able to bitbake one unslung and one openslug and also build unslung packages. all at the same time. Feb 12 00:44:41 rwhitby-800bucks, uh I dont see that happening Feb 12 00:44:52 I hope you're kidding Feb 12 00:45:10 jacques, yes that was what I was thinking Feb 12 00:45:24 ka6sox-away, http://www.newegg.com/app/ViewProductDesc.asp?description=56-101-455&depa=0 Feb 12 00:45:42 I've built three of those Feb 12 00:45:54 okay would a faster proc really help us/ Feb 12 00:45:55 ? Feb 12 00:46:08 (enough to make it worth getting) Feb 12 00:46:12 yes Feb 12 00:46:21 ooh look for $100 more; http://www.newegg.com/app/ViewProductDesc.asp?description=56-101-457&depa=0 Feb 12 00:46:22 XP3200+? Feb 12 00:46:25 socket939 Feb 12 00:46:29 k Feb 12 00:46:42 that plus Athlon64 3500+ would ROCK Feb 12 00:47:18 what are the ramifications of settin up a 64bit vs 32 bit system. Feb 12 00:47:18 * dyoung hides his unused SATA Drives Feb 12 00:47:38 Actually... Feb 12 00:47:48 How much disk do we need? Feb 12 00:48:31 80gb Feb 12 00:48:37 Hmm Feb 12 00:49:36 hmm also bear in mind with the shuttle SFF we don't need boxed proc - OEM is fine and cheaper (doesn't come with HSF which would not work with shuttle anyway) Feb 12 00:49:57 for example OEM 2700+ is $89 while boxed is $105 Feb 12 00:50:43 we need OEM because of the Heat Pipe in the Shuttles. Feb 12 00:50:45 we want to have a system (32 bit vs 64 bit) which most people can replicate - remember this is the template system that anyone wanting to build cross packages should be able to replicate easily. So I think that means 32 bit. Feb 12 00:52:00 http://www.newegg.com/app/ViewProductDesc.asp?description=19-103-460&depa=0 Feb 12 00:52:26 so sweet Feb 12 00:53:26 even sweeter: http://www.newegg.com/app/ViewProductDesc.asp?description=19-103-498&depa=0 - winchester core :-) Feb 12 00:54:07 90nm vs 130 Feb 12 00:54:18 :) Feb 12 00:55:14 even running in 32bit mode the athlon64's rule Feb 12 00:56:33 rwhitby-800bucks, I can spec a system that meets your above requirement, but the motherboard alone will cost about $800 Feb 12 00:56:57 n-way opteron? Feb 12 00:57:06 yes Feb 12 00:57:09 the 4-way motherboard is pretty drn pricy Feb 12 00:57:09 NUMA Feb 12 00:57:18 probably a tyan pro board Feb 12 00:57:18 note I didn't say they had to all build in time period you previously specified. obviously it will take three times as long to build at the same time. Feb 12 00:57:52 rwhitby-800bucks, the contention for cpu/ram/disk would make that rather inefficiant Feb 12 00:57:58 ineffienent Feb 12 00:57:59 dammit Feb 12 00:58:01 inefficient Feb 12 00:58:08 yep Feb 12 00:58:24 now a SCSI/NUMA/SMP system... :-) Feb 12 00:58:29 we wouldn't do it often, but it has to handle it (at the expense of time) Feb 12 00:58:53 rwhitby-800bucks, let's just add up the RAM required to do two oe builds simultaneously Feb 12 00:59:00 2GB Feb 12 00:59:26 you guys settle on a system and then email me the specs. Feb 12 00:59:27 :) Feb 12 00:59:35 we could build a NUMA system with 2GB per proc - sweeet! Feb 12 01:00:01 is it really 1Gb for an OE build? I build it on a CoLinux session with 384Mb real memory and 1Gb swap Feb 12 01:00:39 well for me with psyco, the python process alone wants about 400MB, so maybe not 1GB Feb 12 01:00:56 ah, I don't use psycho Feb 12 01:00:57 but I have no idea how you do it with 384MB ram RAM Feb 12 01:01:20 512MB is generally accepted as the absolute bare minimum Feb 12 01:01:45 does it swap a lot? how long does it take? Feb 12 01:01:45 2 hours to build unslung with 384MB Feb 12 01:01:58 that's not bad at all Feb 12 01:02:09 and on colinux no less? shocking Feb 12 01:02:44 i gotta go cook dinner Feb 12 01:03:10 so, apart from the NUMA pipe dream, what is the current consensus? Feb 12 01:03:13 if my pork chops marinate too much longer they might fall apart Feb 12 01:03:33 * ka6sox-away waits with baited breath. Feb 12 01:04:02 I have 2 specs: $500 nforce2 XP2700+, about $800 also shuttle socket 939, athlon64 3500+ Feb 12 01:04:24 this is assuming the ram for the athlon64 would be about the same which I think is correct Feb 12 01:04:39 neither includes hard drive, cdrom, video, etc Feb 12 01:05:23 I can put a CDROM in there to get it built. Feb 12 01:05:36 I have spare cd drives too Feb 12 01:05:39 I have a cheap video card. Feb 12 01:05:39 well dunno if there should be an etc there - the shuttles include NIC sound etc Feb 12 01:05:57 it will do for these purposes. Feb 12 01:06:02 here are the two barebones for you guys to peruse (if you havent already) Feb 12 01:06:04 still 1GB ram? Feb 12 01:06:07 let's not skimp on the RAM Feb 12 01:06:09 http://www.newegg.com/app/ViewProductDesc.asp?description=56-101-455&depa=0 Feb 12 01:06:15 http://www.newegg.com/app/ViewProductDesc.asp?description=56-101-457&depa=0 Feb 12 01:06:45 I haven't looked closedly at the ram requirement for the athlon64 box Feb 12 01:07:04 I'm hesitant about the 64 bit compatibility ... Feb 12 01:07:17 ? Feb 12 01:07:43 if you read the reviews you will see that the athlon64 running 32bit windows still kills Feb 12 01:07:51 and it can run 32bit linux fine Feb 12 01:07:56 as well as 64bit Feb 12 01:07:59 dunno - it's new to me, so I don't know if there are any issues (we *don't* want this machine on the bleeding edge of linux compatibility) Feb 12 01:09:52 gotta go cook. Feb 12 01:10:22 "the meat just falls off the bone" Feb 12 01:10:50 is there Linux Drivers for the Marvell LAN chipset? Feb 12 01:11:05 Marvell 88E8001 Feb 12 02:06:36 jacques, did you eat yet? Feb 12 02:07:43 catch ya later... Feb 12 02:10:21 yeah Feb 12 02:19:09 so what's the highest spec that the http://www.newegg.com/app/ViewProductDesc.asp?description=56-101-455&depa=0 can run at without stress (e.g max RAM, max processor, max # RAID disks) ? Feb 12 02:21:25 realistically probably 1GB RAM because it only has 2 slots and 1GB DIMMs are $$$ Feb 12 02:22:05 max proc is the fastest athlon XP which is 3200+ Feb 12 02:22:21 nobody ever mentioned RAID as a requirement Feb 12 02:25:56 we can rsync the important bits to external USB disk. Feb 12 02:26:07 Or SlugDisk. Feb 12 02:26:52 it has 1 internal 3.5" bay, 1 accessible 3.5" and 1 accessible 5.25" Feb 12 02:28:23 3200+ has 400MHz FSB which means DDR3200 Feb 12 02:28:28 we dont need no stinkin floppy. Feb 12 02:28:36 yep I agree Feb 12 02:28:53 (although if we need one I have.. Numerous extra. Feb 12 02:31:36 so whats 3200 proc plus 1Gb ddr3200 ram going to total all up (with two disks) ? Feb 12 02:35:45 $180 for the proc Feb 12 02:36:59 memory depends on specs Feb 12 02:37:08 latency etc Feb 12 02:37:24 http://www.newegg.com/app/viewproductdesc.asp?description=20-145-450&DEPA=1 is good Feb 12 02:37:32 best stable crucial ..... Feb 12 02:38:10 huh? Feb 12 02:38:11 corsair is good too Feb 12 02:38:26 http://www.newegg.com/app/viewproductdesc.asp?description=20-145-440&DEPA=1 is cheaper also corsair DDR3200 Feb 12 02:39:42 this is also good (I've had great luck with corsair and geil) http://www.newegg.com/app/ViewProductDesc.asp?description=20-144-310&depa=1 Feb 12 02:40:17 anyway that gives a range of ram prices between $136 and $190 Feb 12 02:41:40 so what's that $190 + proc + eveything else + 2 x 80Gb (or larger) disks total? Feb 12 02:44:22 80GB 7200rpm ATA disks are ranging from $60-90 Feb 12 02:44:34 maybe two of these: http://www.newegg.com/app/viewproductdesc.asp?description=22-148-039&DEPA=1 ? Feb 12 02:44:37 i think 80GB may no longer be the sweet spot Feb 12 02:44:44 I think !60 probably is Feb 12 02:44:46 160 Feb 12 02:45:22 rwhitby-800bucks, that's SATA - no SATA here unless we add a card Feb 12 02:45:32 oh - ok. Feb 12 02:46:08 two of these then: http://www.newegg.com/app/ViewProductDesc.asp?description=22-148-021&depa=1 ? Feb 12 02:46:38 quiet to match the shuttle case Feb 12 02:46:41 as I suspected - 160's are in same price range as 80's now Feb 12 02:47:21 only 2MB buffer :-\ 8 is standard and 16 is becoming common Feb 12 02:47:38 personally I have had bad luck with seagate, but probably best not to start that war Feb 12 02:48:03 5 yr warranty .... Feb 12 02:48:13 but yeah, 8mb minimum Feb 12 02:48:45 http://www.newegg.com/app/ViewProductDesc.asp?description=22-148-022&depa=1 ? Feb 12 02:48:48 if you do the search on newegg you can get all the 160GB 7200rpm 3.5" ATA drives Feb 12 02:49:13 heh something tells me rwhitby-800bucks likes seagate Feb 12 02:49:41 all my desktop drives are seagate Feb 12 02:49:47 but I'm not fanatical about it. Feb 12 02:50:34 I figure with a 5 yr warranty, and everything in CVS ... Feb 12 02:50:37 i just find them to be slightly more expensive and in my individual experience not worth the extra cost (my two failed :-( Feb 12 02:51:00 here is Aus, they are usually cheaper than WD and maxtor Feb 12 02:51:05 ah Feb 12 02:52:00 anyway, maxtor, WD, seagate any should work OK - it's luck of the draw whether one gets a bad drive Feb 12 02:52:26 maxtor has 16MB buffer and has been using FDB for a while now so they are also quiet Feb 12 02:52:40 (the new maxtors have 16MB buffer - older have 8) Feb 12 02:53:05 just don't let peteru-home know if we decide to buy a maxtor :- Feb 12 02:54:14 I think we should stay in the drive size/price sweet spot - we don't need huge storage but it's silly to pay same for old tech 80GB as new tech 160GB Feb 12 02:55:59 agreed Feb 12 03:10:35 maxtor. Feb 12 03:10:44 I have an aversion to maxtor Feb 12 03:10:56 they use twice as much power as the same size seagate. Feb 12 03:11:04 and they run REAL hot. Feb 12 03:11:34 I too would prefer WD or Seagate to Maxtor Feb 12 03:12:18 get what you want Feb 12 03:12:39 I'm getting rather tired of arguing drive brand religion Feb 12 03:12:51 fair enough Feb 12 03:13:14 they are only disk drives :-) Feb 12 03:13:36 I have all of the above. I'm just noting my maxtors run Real hot. Feb 12 03:13:38 yep Feb 12 03:14:12 if there is anything else with 5 yr warranty, then I'm open to it. otherwise, all other things being equal, I go for the longer warranty on disks. Feb 12 03:14:19 And something tells me Hot and 24/7 operation dont get along. Feb 12 03:14:33 argh Feb 12 03:14:42 some of these things I can't resist responding to Feb 12 03:14:53 i run my maxtors 24/7 for years Feb 12 03:15:06 can we just flip a coin or something and stop talking abot it? Feb 12 03:16:05 I'm sure most of these drive come from the same conscription labour camp anyways. Feb 12 03:16:34 ok, if we were to vote, I suspect seagate would win - so let's get seagate Feb 12 03:18:04 maybe this: http://www.newegg.com/app/ViewProductDesc.asp?description=22-148-025&depa=1 Feb 12 03:18:17 only about $7 more than the 120GB (I think) Feb 12 03:18:34 Does newegg have decent shipping for the conus? Feb 12 03:18:40 They always jack me on shipping. Feb 12 03:18:45 but everyone jacks me on shipping. Feb 12 03:18:57 * dyoung grumble Feb 12 03:19:29 jacques: looks good to me Feb 12 03:19:40 cool Feb 12 03:20:28 so 200 for case & misc, 200 for proc, 200 for ram, 200 for disks. Feb 12 03:20:48 anything else we need? Feb 12 03:21:52 http://www.newegg.com/app/ViewProductDesc.asp?description=56-101-455&depa=0 Feb 12 03:21:58 http://www.newegg.com/app/viewproductdesc.asp?description=20-145-450&DEPA=1 Feb 12 03:22:14 http://www.newegg.com/app/ViewProductDesc.asp?description=22-148-025&depa=1 * 2 Feb 12 03:23:04 ka6sox-away said he has video card to use while installing so we don't need that Feb 12 03:23:08 link for processor? Feb 12 03:23:26 link ? Feb 12 03:23:36 the newegg link. Feb 12 03:23:41 oh sorry Feb 12 03:24:19 that was weird - they have a retail and OEM but the retail is cheaper(!) I think that's because conventional wisdom is the boxed procs are older revs Feb 12 03:24:25 and ppl who overclock want the latest Feb 12 03:24:34 difference was about $2 Feb 12 03:24:41 maybe $1 Feb 12 03:24:51 Whats the warantee policy on oem vs retail box these days? Feb 12 03:24:54 is there somewhere where ka6sox can just walk in and buy this stuff at these prices, or do we need to consider shipping too ? Feb 12 03:25:20 I would imagine that fry's would have similar pricing... Feb 12 03:25:59 but I havnt been to one for a few years Feb 12 03:26:27 I have no idea where ka6sox-away can walk into Feb 12 03:26:54 I presume there is a Fry's in LA ? Feb 12 03:27:17 http://www.newegg.com/app/ViewProductDesc.asp?description=19-103-390&depa=0 Feb 12 03:27:22 http://www.newegg.com/app/ViewProductDesc.asp?description=19-103-391&depa=0 Feb 12 03:27:27 the two procs Feb 12 03:27:45 yeah there's a Fry's there but LA is big - really big Feb 12 03:27:57 that's something we would have to ask him Feb 12 03:28:07 theres one near where he'll be in anaheim Feb 12 03:28:14 how much is shipping anyway? in CA there is also sales tax BTW Feb 12 03:28:37 ok, what is the warranties on all these parts (since it is a machine built from donations) ? Feb 12 03:29:05 I don;t understand the connection Feb 12 03:29:30 well, to replace a part we would have to go back to the community to get more funds. Feb 12 03:29:38 I dont think theres a fry's near his convention though. Feb 12 03:30:34 memory is easy - lifetime :-) Feb 12 03:30:43 other's probably pathetic as usual Feb 12 03:31:09 lifetime for memory, 5 yrs for disk (which is probably the lifetime of this project), .... Feb 12 03:31:36 CPU Failure is pretty rare. Feb 12 03:32:51 power supply? Feb 12 03:33:00 and motherboard? Feb 12 03:33:02 part of the case? Feb 12 03:33:14 yeah, how long is the warranty on it? Feb 12 03:33:18 the whole barebones is under same warantee Feb 12 03:34:38 which is ? Feb 12 03:36:18 interesting Feb 12 03:36:42 on shuttle's site they have a SN45GV3 (instead of V2) which has SATA RAID Feb 12 03:38:47 its 12month on barebones systems Feb 12 03:39:36 hi dyoung, how's life with yer mac mini? Feb 12 03:40:15 I'm pretty satified with the mini Feb 12 03:40:35 http://sys.us.shuttle.com/Scgsupport/RMAInquiry.aspx Feb 12 03:40:59 is it the 1.2 ghz? what about the ventilator inside..does it run when e.g. surfing the web, using word? Feb 12 03:41:27 jacques: is the newegg extended warranty worth it? Feb 12 03:41:27 No, it only runs when bitbaking, or when importing a big video file, etc. Feb 12 03:41:34 under normal use the fan stays off. Feb 12 03:41:39 or "slow". Feb 12 03:41:50 so its really damn quiet? :) Feb 12 03:42:15 Yes. I'm happy about the sound levels. Feb 12 03:42:22 rwhitby-800bucks, do they have one on the shuttle ? I usually don't get extended warranties Feb 12 03:42:37 now the loud thing is the crappy LaCie disk that Slug is plugged to. Feb 12 03:43:24 at what resolution are u using the mac mini with? is it true that running it at 1280*1024 will slow the cpu down due to little gfx mem? Feb 12 03:46:03 I run at that resolution. Theres 32MB of dedicated video memory. Feb 12 03:46:16 so i dont think theres a slowdown, but I havnt run at lower resoltions either. Feb 12 03:46:36 k Feb 12 03:48:25 jacques: http://www.newegg.com/app/ViewProductDesc.asp?description=56-101-455&depa=0 has the "qualifies for extended warranty" graphic on the side ... Feb 12 03:49:45 Extended Warranty: $ 16.32 Feb 12 03:49:50 for one additional year Feb 12 03:50:18 seems to contradict this tho: **This item is warranted through the product manufacturer only. Feb 12 03:50:36 unless extended warranty is through shuttle Feb 12 03:51:14 dunno Feb 12 03:53:01 Its probably like the compusa extended warantee. Feb 12 03:53:16 or any other big box extended warantee. Feb 12 03:57:02 yeah, the details indicate you send it to newegg, and they send you back a replacement. Feb 12 04:25:24 so how much over $800 does it all add up to? Do we only buy one disk to bring the cost down? Feb 12 04:28:02 what disk size was the decision? Feb 12 04:28:42 http://www.newegg.com/app/ViewProductDesc.asp?description=22-148-025&depa=1 Feb 12 04:28:46 160 Gb Feb 12 04:29:44 is that a raid device, or a backup device? Feb 12 04:30:08 if its a backup device we can back it up to something smaller Feb 12 04:30:38 or wait till theres an extra $100 in the kitty. Feb 12 04:31:26 I was figuring the second drive would be raid (or do you really need more than two)? Feb 12 04:31:40 but ka6sox can probably back it up somewhere else ... Feb 12 04:31:54 what kind of failure/oops do you want to protect against? Feb 12 04:32:11 just drive failure Feb 12 04:32:25 all the source will be in CVS/BB anyway ... Feb 12 05:19:03 rwhitby-800bucks: What was the reasons for not being able to use PayPal in EU? Feb 12 05:20:15 something about privacy laws - or that might have been responding to my suggestion of someone over there collecting donations in a bank account and then forwarding a single paypal donation. Feb 12 05:22:51 I've got a PayPal UK account. I think it might be something to do with Germany Feb 12 05:23:17 I seem to remember they've got some extra laws regarding privacy because eBay Germany works differently to everybody else Feb 12 05:23:22 *shrug* Feb 12 05:24:02 I don't know of any EU wide service similar to paypal however which is a bit annoying Feb 12 05:25:19 nochex is slightly cheaper if you're accepting credit card payments Feb 12 05:25:26 http://help.nochex.com/?Action=Q&ID=92 Feb 12 05:29:43 i heard something about worldpay being okay Feb 12 05:29:58 Looks quite expensive from that comparision page Feb 12 06:15:14 hi Feb 12 06:16:00 Hi Feb 12 06:18:22 i have a problem with bitbaking the unslung image Feb 12 06:18:40 ipkg-native fails Feb 12 06:19:16 all_read.c:45: error: conflicting types for 'full_read' Feb 12 06:19:16 | libbb.h:119: error: previous declaration of 'full_read' was here Feb 12 06:19:52 sorry. can't help you with that one! you'll have to wait for kergoth or one of the others to reappear Feb 12 08:54:44 hm... does anybody know how to fix compilation of nfs-utils for an uclibc-openslug? Feb 12 08:55:19 <[g2]> siddy, I didn't know it was broken Feb 12 08:55:45 <[g2]> I've got to run right now, but it doesn't compile ? Feb 12 08:56:12 it fails with "`NFS_STATEDIR' undeclared" Feb 12 08:56:37 rmtab.c: In function `setrmtabent': Feb 12 08:56:42 rmtab.c:27: error: `NFS_STATEDIR' undeclared (first use in this function) Feb 12 08:56:47 to be exact. Feb 12 09:01:05 <[g2]> siddy, it's broke in glibc too Feb 12 09:01:28 <[g2]> well I'm off Feb 12 09:01:32 yes, but it with a different error, if I remember correctly. Feb 12 09:08:37 rwhitby-away: Seems like appWeb is missing a library in the IPK - libmpr.so - it is needed to run e.g. htPassword. Feb 12 10:54:27 aha! Feb 12 10:54:35 i knew something completely obvious was amiss Feb 12 10:54:40 * giel hits himself Feb 12 10:56:26 ? Feb 12 11:01:37 i have no /dev/console on my hdd Feb 12 11:01:58 so my /bin/init failed Feb 12 11:02:06 are you still working on switchbox? Feb 12 11:03:43 uhu Feb 12 11:03:56 don't have much free time, so it's going slow Feb 12 11:04:02 but i'm almost where i want to be Feb 12 11:26:12 giel, which switchbox do you use? the one in the openembedded-repository or in unslung-4.x? Feb 12 11:26:28 i'm using switchbox 4.6 now Feb 12 11:28:04 k, I think, I'm also gonna use switchbox, the 8MB flash is just too small... ;) Feb 12 11:28:34 i just want to use hdd as root filesystem, just for flexibility Feb 12 11:28:51 and btw, you are already using switchbox Feb 12 11:29:00 it's only switching to the jffs2 instead of hdd Feb 12 11:29:22 I know, but is openslug's switchbox already able to use usb-storage-devices? Feb 12 11:29:34 no Feb 12 11:29:59 but switchbox 4.6 is? Feb 12 11:30:03 no Feb 12 11:30:08 you'll have to modify it Feb 12 11:30:22 not too heavily tho Feb 12 11:30:30 oh... so where's the advantage in using the newer version? Feb 12 11:30:51 oh, it has infrastructure for using various devices as root filesystem Feb 12 11:31:25 it would work if you'd use a kernel with usb drivers and scsi drivers in the kernel instead of modules Feb 12 11:31:42 but if you are using modules you'll have to copy them in switchbox and load them before you mount the hdd Feb 12 11:31:46 it's not too hard Feb 12 11:31:58 k, so I "simply" have to make switchbox load the usb-storage-modules, modify it to use /dev/sda1 and I'm done? Feb 12 11:33:30 yeah that's about it Feb 12 11:33:37 i can help you when you're going to do it Feb 12 11:33:46 but first i have to get it working for myself :) Feb 12 11:33:48 i'm close tho Feb 12 11:36:14 well, I just "bricked" my openslug once again, and I'm sick of not being able to access it's root-fs from another device. So, modifying switchbox to use my cardreader as root would be the way to go I guess. Feb 12 11:45:15 Is it possible to split up a firmware images into its components (kernel, rootfs etc)? Feb 12 11:45:58 I would like to loopback mount the rootfs Feb 12 11:51:09 yeah Feb 12 11:51:26 use slugimage Feb 12 11:52:03 it's in the CVS repo Feb 12 11:52:19 the module is called slugimage Feb 12 12:05:37 is there a reason why we have I2C-support in the defconfig of openslug's kernel? Feb 12 12:08:33 maybe just because the ixp4xx defconfig has it Feb 12 12:09:02 hm... k, so I'll better let it in there... :-\ Feb 12 12:09:37 giel: There is a package called slugtool, but no slugimage. You mean BK-repo? Feb 12 12:11:01 wiki/OpenEmbedded/OEKernelBuildWithStockRamdiskLog seems to suggest there's an rtc on the i2c bus Feb 12 12:12:28 ah, thx. so I'll just keep I2C and hope it's not too big. My kernel is getting close to the 1MB-boundary... :-( Feb 12 12:18:51 bob_tm: no, i mean sourceforge CVS Feb 12 12:19:54 Hmmm... no make/slugimage.mk.... Where is it? Feb 12 12:20:03 ...under unslung, that is Feb 12 12:20:09 <[g2]> hey lennert Feb 12 12:20:14 it is not under unslung Feb 12 12:20:20 it is a separate module Feb 12 12:20:27 do a cvs checkout slugimage Feb 12 12:20:33 Ok - will do. thnx Feb 12 12:20:47 <[g2]> giel, which mods did you do to switchbox Feb 12 12:20:54 <[g2]> to make it boot Feb 12 12:21:00 i copied the usb and scsi modules to it Feb 12 12:21:05 and did some insmodding Feb 12 12:21:14 but it still doesn't work Feb 12 12:21:15 <[g2]> and some waiting ? Feb 12 12:21:23 uh yeah Feb 12 12:21:30 <[g2]> can you e-mail me a copy ? Feb 12 12:21:32 my slug reboots now after chrooting i guess Feb 12 12:21:43 pbly a panic Feb 12 12:21:46 dunno why Feb 12 12:22:01 <[g2]> me either, but at least I've got serial :) Feb 12 12:22:22 well i can telnet in before booting now Feb 12 12:22:31 but from within that telnet shell i can't run init Feb 12 12:23:47 [g2]: i'll email you what i've got once i got it working, okay? Feb 12 12:23:55 <[g2]> I think init needs to be job 1 Feb 12 12:24:31 <[g2]> well I may get a chance in about 5-6 hours to play with it a bit Feb 12 12:24:49 <[g2]> I can just go straight to switchbox 4.6 Feb 12 12:25:26 <[g2]> or just play with it straight on the jffs2 partition Feb 12 12:25:50 <[g2]> maybe that's the thing to do Feb 12 12:26:10 <[g2]> lose switchbox, and boot straight to the jffs2 partition Feb 12 12:26:22 hm Feb 12 12:26:51 <[g2]> if the reset button is held down, then the linuxrc skips booting to external drives and loads a fails safe ext2 part or the jffs2 Feb 12 12:37:33 can anyone help with vsftpd? Feb 12 12:37:58 hello [g2] Feb 12 12:45:36 this is weirf Feb 12 12:45:52 why isn't my linuxrc executed as job 1 Feb 12 12:47:32 Can anyone help me with vsftpd? I can only log in as 'root'. Feb 12 12:47:45 ~nslu2-vsftpd Feb 12 12:47:46 have a look at http://www.nslu2-linux.org/wiki/Unslung/Vsftpd Feb 12 12:48:48 yes. Feb 12 12:49:20 Let me as this... should I be able to ssh as anyother user besides 'root'? Feb 12 12:51:32 childeb1: Out of, or into the slug Feb 12 12:51:34 do you hvae other users defined on your unit? Feb 12 12:51:49 out of yes of course. In to see above. Feb 12 12:52:35 Into the slug. Feb 12 12:53:20 Yes, I have myself setup as a user with admin privs and I cannot login to ssh or ftp (w/ vsftpd). Feb 12 12:53:30 I have to log in as root Feb 12 12:54:12 Can you su to the other user once logged in? Feb 12 12:54:59 what is his login shell Feb 12 12:55:42 Someone in #nslu2-general might be able to better help you. Feb 12 12:55:46 ~nslu2-general Feb 12 12:55:47 #nslu2-general is the IRC channel specifically set up for end-user questions about the NSLU2 device. Questions that are not related to the nslu2-linux development projects should not be asked in #nslu2-linux, but should be asked in #nslu2-general instead. Feb 12 12:56:14 ok. Thanks for the pointer Feb 12 13:00:08 ok, i now know at least what the problem is Feb 12 13:00:43 dyoung-zzzz: you were pivot_rooting? Feb 12 13:01:07 right now? Feb 12 13:01:11 or in gneral? Feb 12 13:01:15 in general Feb 12 13:01:18 Yes. Feb 12 13:01:31 which linuxrc are you using? Feb 12 13:01:36 standard switchbox 4 one? Feb 12 13:01:40 Yes Feb 12 13:01:47 weird Feb 12 13:01:57 Youre uclibc right? Feb 12 13:02:03 yeah Feb 12 13:02:05 what are you trying to do exactly? Feb 12 13:02:23 the reason i ask is we cheat with switcbhx by symlinking stuff form the jffs2. Feb 12 13:02:39 well i've made my own linuxrc atm Feb 12 13:02:51 Without serial? Feb 12 13:02:55 but the problem is i can't execute init, 'cause linuxrc isn't run as pid 1 Feb 12 13:02:58 yup Feb 12 13:03:02 Youre really Adventurous. Feb 12 13:03:24 Okay so when linuxrc tries to exec init, it grumbles about that? Feb 12 13:03:42 i have two debugging methods: starting telnetd at some point in linuxrc and logging in, and writing stuff to a file on a usb stick Feb 12 13:03:59 well, the kernel panics 'cause it can't execute init, 'cause init needs to be pid 1 Feb 12 13:04:14 and init gets the pid of linuxrc (by means of exec), and linuxrc isn't pid 1 Feb 12 13:04:18 what does your command line look like? Feb 12 13:04:22 and that i don't understand Feb 12 13:04:32 linuxrc is supposed to be 1 and it isn;t Feb 12 13:04:35 what command line? Feb 12 13:04:44 your kernel command line Feb 12 13:04:57 dunno, standard redboot one i guess Feb 12 13:05:15 i can boot into openslug with jffs2 and dmesg Feb 12 13:05:20 1 mom Feb 12 13:05:38 in fact, 'couple of moms, booting takes a while :) Feb 12 13:06:32 ok be back in a bit then, gotta glue my poor slug back together. Feb 12 13:06:42 ok Feb 12 13:06:47 while sleeping? Feb 12 13:07:20 sleepwalking perhaps Feb 12 13:09:29 Kernel command line: root=/dev/slug rootfstype=ext2,jffs2 initrd=0x01000000,10M mem=32M@0x00000000 console=ttyS0,115200n8 Feb 12 13:12:53 hm Feb 12 13:13:02 i think i know a workaround Feb 12 13:13:20 running init as linuxrc will work, it won't have to be pid 1 Feb 12 13:13:39 but it's still a mystery why linuxrc isn't pid 1 Feb 12 13:14:28 hmm thats weird. Feb 12 13:14:33 yeah, I dunno. Feb 12 13:14:39 But what you point out is interesting. Feb 12 13:15:16 because unslung-4.x is stalled for a couple of reasons both of which i think are related to "something" that happens at the switchbox level. Feb 12 13:16:00 PID Uid VmSize Stat Command Feb 12 13:16:01 1 0 SW [swapper] Feb 12 13:16:02 ... Feb 12 13:16:09 581 0 164 S /bin/sh /linuxrc Feb 12 13:16:13 okay, next test Feb 12 13:16:22 i've symlinked /linuxrc to sbin/init on my hdd Feb 12 13:16:27 and i'm trying to exec that one Feb 12 13:16:45 if init behaves, it skips the pid == 1 check if it's called linuxrc Feb 12 13:23:32 dyoung-zzzz: is there a way to change the kernel command line? Feb 12 13:24:21 giel, simply modify the kernel's defconfig and recompile Feb 12 13:25:15 siddy: thnx Feb 12 13:27:03 I have almost finished my usb-storage-enabled kernel, its size is 0.999MB, but it seems to fit... :-D Feb 12 13:29:33 [g2]-away: ping Feb 12 13:29:58 do we still have consensus on the build machine to buy ? Feb 12 13:30:34 rwhitby-away: I keep leaving msgs but then am never around to hear your response... Feb 12 13:30:45 Did you get msg about jamvm? Feb 12 13:30:47 hm Feb 12 13:30:59 is it possible to add init=/linuxrc somewhere to the kernel command line? Feb 12 13:31:08 giel: should be Feb 12 13:31:36 yeah, but where? isn't the kernel command line in redboot? Feb 12 13:31:53 i always specify those in the kernel config and recompile Feb 12 13:32:28 oh ofcourse Feb 12 13:32:35 lemme try that then Feb 12 13:32:55 openembedded/packages/linux/openslug-kernel_2.6.9.bb Feb 12 13:32:56 <[cc]smart> build machine ? Feb 12 13:37:23 <[g2]-away> rwhitby-away, pong Feb 12 13:37:58 hi [g2] Feb 12 13:38:12 * [g2] hugs lennert :) Hey Feb 12 13:38:19 uhm Feb 12 13:38:23 hi dear Feb 12 13:38:26 :P Feb 12 13:38:32 what's up? Feb 12 13:38:37 <[g2]> I'm glad we nailed that usbnet Feb 12 13:38:40 <[g2]> issue Feb 12 13:38:46 yeah Feb 12 13:38:50 how long were you stuck on it? Feb 12 13:39:27 <[g2]> we'll it's been a background issue for a couple months, but it took about a 1 or 2 of working on it Feb 12 13:40:32 'day or 1 or 2'? Feb 12 13:40:34 <[g2]> I was putting printk's in the IP stack to find the exact failure, then you guys came to the rescue and mopped up Feb 12 13:41:07 <[g2]> 5-10 hours of time Feb 12 13:41:35 as i said, 't was just some random idea i got Feb 12 13:41:51 <[g2]> well not really that random :) Feb 12 13:42:12 i mean, it wasn't exactly obvious that it was the alignment handler Feb 12 13:42:32 you'd expect a million other places that either forget to byteswap or byteswap twice, but not this Feb 12 13:43:08 but does this mean we have to have alignment set to fixup all the time ? Feb 12 13:43:27 fontenot: nope, i think it's unconditionally enabled for kernel code Feb 12 13:43:36 lennert: ah good Feb 12 13:43:40 <[g2]> nod. non-obvious actually Feb 12 13:43:43 the proc entry controls just userspace Feb 12 13:43:47 well Feb 12 13:44:05 the only reason i've heard why it's not enabled unconditionally is because apps might expect the 'wrong' behavior Feb 12 13:44:20 <[g2]> lennert, there are actually a handful of othere latent error in there that are fixed by this Feb 12 13:44:34 i think that's just a weak excuse -- the arm people probably _want_ unaligned accesses to fail Feb 12 13:44:41 [g2]: oh, like what? Feb 12 13:45:05 <[g2]> I'd don't know where they are (I could find them now by looking at the trap) but I see them in the trap statistics :) Feb 12 13:45:17 ah Feb 12 13:45:17 the best reason I have heard is the performance impact Feb 12 13:45:23 did you try 5 > /proc/cpu/alignment ? Feb 12 13:45:28 that should log a syslog message for each unaligned access Feb 12 13:45:32 (user accesses) Feb 12 13:45:36 fontenot: yeah Feb 12 13:46:05 <[g2]> they are system accesses, I could set that up in inittab and find who they are Feb 12 13:46:22 [g2]: can you dump_stack() in the alignment handler for system accesses? Feb 12 13:46:29 you'd get the backtrace then Feb 12 13:46:46 <[g2]> how do I dump_stack ? Feb 12 13:47:11 well Feb 12 13:47:22 find do_alignment in arch/arm/mm/alignment.c Feb 12 13:47:32 and then insert 'dump_stack();' right after 'ai_sys += 1;' Feb 12 13:47:36 recompile, boot, etc. Feb 12 13:47:47 <[g2]> nod. Feb 12 13:47:48 then watch your syslog Feb 12 13:49:40 <[g2]> root@nslu2:~# cat /proc/cpu/alignment Feb 12 13:49:40 <[g2]> User: 0 System: 6 Feb 12 13:49:40 <[g2]> Skipped: 0 Half: 6 Feb 12 13:49:40 <[g2]> Word: 0 Multi: 0 Feb 12 13:49:40 <[g2]> User faults: 0 (ignored) Feb 12 13:50:02 <[g2]> 6 half word alignment error in system is what happens Feb 12 13:50:13 that probably can't really hurt Feb 12 13:50:17 but i still wonder where that happens Feb 12 13:50:31 * [g2] too :) Feb 12 13:51:08 <[g2]> The interesting thing (to me) was that the alignment patch works automatically for sytem code but needs to be enabled for user code Feb 12 13:51:25 policy decision Feb 12 13:51:25 <[g2]> and also that it fixed the system issues Feb 12 13:51:40 <[g2]> I think that's a great policy Feb 12 13:51:50 <[g2]> it made total sense to me Feb 12 13:52:01 i think it should complain louder though Feb 12 13:52:04 <[g2]> along with the stats Feb 12 13:52:11 print a big fat warning if there's an unaligned access in kernel mode in any case Feb 12 13:52:24 make sure people fix their shit Feb 12 13:53:06 <[g2]> I'm okay with that, what surprises me is that a simple regression test like the alignment test you made isn't run and verified Feb 12 13:53:21 not many people run 2.6 on big-endian ARMs, it seems Feb 12 13:53:25 <[g2]> stuff just *shouldn't* be released be for simple stuff like that Feb 12 13:53:45 <[g2]> before Feb 12 13:53:57 most of these things are fixed for 2.4, in the -ds and various other trees Feb 12 13:54:06 but the patches never went upstream in 2.4 becuase big-endian is not supported in 2.4 Feb 12 13:54:21 so people have had to sort through the various 2.4 patches to find bugfixes Feb 12 13:54:28 and then resubmit them for 2.6 Feb 12 13:54:31 <[g2]> speaking of 2.4 upstream is that basically dead except for maintenance ? Feb 12 13:54:37 yeah Feb 12 13:54:48 but, most vendors are still on 2.4 Feb 12 13:54:54 <[g2]> Hmmmm Feb 12 13:55:07 the other ixp's.. Feb 12 13:55:15 ixp2000 support is in 2.6, but not supported by intel Feb 12 13:55:25 ixp1200 support never got ported to 2.6, too little interest in it Feb 12 13:55:57 <[g2]> I don't understand Intel or some of the Big Iron vendors Feb 12 13:56:08 when the 2.6 ixp4xx port was done, people naturally didn't check for other big-endian issues in the 2.4 trees Feb 12 13:56:18 i don't get intel either Feb 12 13:56:44 their stuff is _so_ hard to get Feb 12 13:56:54 march last year we decided we wanted an ixp2000 board Feb 12 13:56:58 it took like four months to get one Feb 12 13:57:10 they have all these products on their web page Feb 12 13:57:18 and then you expect a 'buy it now' button somewhere or something Feb 12 13:57:21 but noooooooo Feb 12 13:57:27 'contact our authorized resellers' Feb 12 13:57:39 so you contact the company that's listed as reseller for the netherlands Feb 12 13:57:44 'we sell only pentium processors' Feb 12 13:57:49 "do you have an ixp2000 board?" Feb 12 13:57:52 "what's that?" Feb 12 13:57:54 ok Feb 12 13:57:58 so you contact an american reseller Feb 12 13:58:06 "sorry, we can't sell stuff to you because you're not in the usa." Feb 12 13:58:10 so basically, forget it. Feb 12 13:58:34 <[g2]> it's gonna be interesting in a year with the IBM PPC initiative Feb 12 13:59:00 the only (!) importer of intel desktop and server mobos stopped importing them some months ago into .nl Feb 12 13:59:32 [g2]: did you see the specs on the build machine last night? Any comments, objections, etc? Feb 12 13:59:42 [g2]: i hope they come down in price Feb 12 14:00:04 <[g2]> rwhitby-away, you guys made lots of sense Feb 12 14:00:51 <[g2]> I didn't catch the final verdict on the SATA, I do like SATA, especially with my little Raptor Feb 12 14:01:09 <[g2]> but I don't think that's worth the trade off Feb 12 14:01:17 <[g2]> in price Feb 12 14:02:11 yeah, I'd like to see someone do a price comparision on the same system with SATA support. Jacques found the V2 box on the shuttle site. Feb 12 14:02:45 V3 Feb 12 14:03:17 <[g2]> lennert, you mean the PPC come down in price ? Feb 12 14:04:15 <[g2]> lennert, and have you looked at any of the TCP-offload hw, I think there an Nvidia mobo that's got Gigabit and TCP-offload Feb 12 14:05:52 * [g2] thinks gigabit with tcp-offload support will be a commodity by last '05 or in '06 Feb 12 14:06:14 <[g2]> doh!, late '05 or in '06 Feb 12 14:07:09 blegh weird things still Feb 12 14:08:02 [g2]: yeah, ppc are still really expensive (relatively) Feb 12 14:08:20 [g2]: are there specs for the board? Feb 12 14:09:45 <[g2]> I forget where I read about it, it had two Gigabit parts 1 something standard like 3Com, etc.. and one Nvidia with the tcp-offload enhancements Feb 12 14:10:18 why if i change the kernel command line in the defconfig in packages/linux/openslug-2.6.9 and i rebuild it doesn't seem to have changed the command line Feb 12 14:11:04 oh, it's in the bb Feb 12 14:11:15 <[cc]smart> g2: think that's an asus board Feb 12 14:11:24 I pasted the line for you... Feb 12 14:11:31 [g2]: nvidia are notoriously short on docs Feb 12 14:11:40 <[cc]smart> g2: nforce socket 939 i guess Feb 12 14:12:24 <[cc]smart> g2: they have this builtin firewall feature too Feb 12 14:12:24 <[g2]> If I has some money I'd start an open-source hw company Feb 12 14:12:42 <[cc]smart> just that nobody could download a mobo Feb 12 14:13:05 <[g2]> you just need to do what digium does Feb 12 14:13:22 <[cc]smart> do they aactually earn enough ? Feb 12 14:13:30 <[g2]> give away asterisk and sell hw to support at *somewhat* reasonable costs Feb 12 14:13:40 <[cc]smart> thought about buying a zaptel myself, but you'll find no vendor around here Feb 12 14:13:58 <[g2]> I have *no* idea, but I'm guessing they are makeing boatloads Feb 12 14:14:33 <[g2]> I used work at a company that made access equip. Feb 12 14:14:35 <[cc]smart> i don't think so, but i wish them some Feb 12 14:14:45 i'm still wondering what to do next Feb 12 14:14:56 <[g2]> Their 4-port T1 card goes for $1500 Feb 12 14:15:43 <[g2]> lennert, what do you want to do ? Feb 12 14:16:28 not sure yet Feb 12 14:16:51 <[cc]smart> earn money ? Feb 12 14:17:11 obviously, i've got to feed myself in some way Feb 12 14:17:37 <[cc]smart> sure, just wanted to set a start :) Feb 12 14:18:05 hehe :) Feb 12 14:18:45 we'll see Feb 12 14:18:59 i'm getting married first (three weeks from now), then i'll start looking for something again Feb 12 14:19:04 lennert: you're looking for a job? Feb 12 14:19:23 there're 6 vacancies at my group at philips Feb 12 14:19:45 giel: what kind of group? Feb 12 14:19:52 storage systems & applications Feb 12 14:19:56 research Feb 12 14:22:14 <[cc]smart> hmm... somebody a suggestion why upslug doesn't work for me ? Feb 12 14:22:16 research or "research"? Feb 12 14:22:40 well, somewhere in between :) Feb 12 14:22:40 <[g2]> lennert, congrats on the marriage, make sure to take a couple weeks off after the marriage too :) Feb 12 14:22:45 like.. when you work at "kpn research", you just paraphrase other people's work :) Feb 12 14:22:46 <[cc]smart> i "version control" output Feb 12 14:22:52 [g2]: thanks :) Feb 12 14:22:58 oh then it's pretty much real research Feb 12 14:23:10 anyway, we're getting rapidly offtopic Feb 12 14:26:22 <[cc]smart> is there an alternative source for upslug ? Feb 12 14:26:56 <[cc]smart> the main one linked doesn't work for last couple days at least Feb 12 14:39:21 okay, i'll be needing a different init Feb 12 14:39:39 the kernel is broken and the sysvinit standard in openslug can't cope with that Feb 12 14:39:45 http://lists.debian.org/debian-boot/2001/10/msg00269.html Feb 12 14:40:15 at least, that's what i think now Feb 12 14:40:38 but i'll ask the guy who wrote initrd monday, i happen to know him Feb 12 14:41:10 <[g2]> giel, how is the kernel broken ? Feb 12 14:41:35 linuxrc just isn't executed as pid 1 Feb 12 14:41:40 so no way init can exec Feb 12 14:43:23 <[g2]> root@nslu2:~# ps Feb 12 14:43:23 <[g2]> PID Uid VmSize Stat Command Feb 12 14:43:23 <[g2]> 1 root 528 S init [2] Feb 12 14:43:47 <[g2]> I don't understand how yours' could be different Feb 12 14:44:09 you don't do a pivot_root, do you? Feb 12 14:44:27 'cause if you use the proc real root dev method, the kernel starts init itself Feb 12 14:44:37 with pivot root, you start init from within linuxrc Feb 12 14:44:47 PID Uid VmSize Stat Command Feb 12 14:44:48 1 0 SW [swapper] Feb 12 14:44:49 ... Feb 12 14:44:55 581 0 164 S /bin/sh /linuxrc Feb 12 14:46:34 <[g2]> and you just exec where you continue the execution thread right ? Feb 12 14:46:54 yep Feb 12 14:49:08 why is root=/dev/slug on the command line? Feb 12 14:49:42 <[g2]> iirc it's a trigger to cause the linuxrc to run Feb 12 14:49:49 right. Feb 12 14:50:08 <[g2]> any non /dev/ram0 or ramN will do it Feb 12 14:50:17 aha! Feb 12 14:50:21 <[g2]> non root Feb 12 14:50:33 i think root=/dev/ram0 init=/linuxrc does the trick then Feb 12 14:50:37 to execute linuxrc as pid 1 Feb 12 14:50:38 <[g2]> root != /dev/ram0 or /dev/ramN Feb 12 14:51:20 <[g2]> with your ramdisk loaded right Feb 12 14:51:38 hm Feb 12 15:02:56 aha Feb 12 15:03:04 i may have found another solution Feb 12 15:03:18 by browsing through the sysinit source i've found an undocumented option Feb 12 15:03:31 init -i forces init to behave as if pid = 1 Feb 12 15:05:06 <[g2]> cool Feb 12 15:05:25 <[g2]> "Luke -- Be one with the source" Feb 12 15:05:26 boy, it sure is nice to have source Feb 12 15:05:48 <[g2]> jacques, OH That's *so* overrated :) Feb 12 15:27:13 Does anyone know a good sftp client for pocketpc? Feb 12 15:27:39 I setup my nslu2 box with ssh, and like to use sftp to access it Feb 12 15:28:00 or perhaps webdav is a better alternativ? Feb 12 15:28:40 ideally I would like the filesystem to be mounted Feb 12 15:51:57 giel, I just booted with my mmc-card as rootfs! switchbox worked flawless! Feb 12 15:53:39 how can I avoid filling up the conf partition? Feb 12 15:53:55 I insalled the toolkit from the "sanctioned" mirror Feb 12 15:53:59 and poof, there goes conf Feb 12 15:54:46 you could symlink /share/hdd/conf/opt to somewhere on data Feb 12 15:55:21 poor thing locked solid after I hit 100 percent Feb 12 15:58:21 siddy: congrats! Feb 12 15:58:48 i have something kind of working too, now Feb 12 15:58:57 but not really flawless Feb 12 16:00:35 I still have some prblems with a strange-looking /dev, but everything else looks pretty... :-D Feb 12 16:00:40 if i exec /sbin/init -i it works Feb 12 16:00:48 if i add a chroot before that it doesn't work Feb 12 16:00:51 dunno why Feb 12 16:01:09 and still don't know why my linuxrc isn't pid 1, it seems to be a known problem Feb 12 16:01:22 hm... I'm using old switchbox 3.9 without such problems. I just changed the mount-command and that's it. Feb 12 16:01:28 oh yeah Feb 12 16:01:32 that's the old method Feb 12 16:01:36 no pivot_rooting Feb 12 16:02:14 you changed only the mount command? Feb 12 16:02:16 hm... any disadvantages of that method? Feb 12 16:02:30 and the kernel to support the usb-storage-devices Feb 12 16:02:32 i think you have to change something with real root dev too Feb 12 16:02:41 which mount command did you change? Feb 12 16:03:23 in linuxrc I changed the mount /dev/mtdblock?? to mount /dev/sde1. Feb 12 16:03:45 uhm Feb 12 16:03:55 i don;t think you have it working the way you want it atm Feb 12 16:04:04 I also changed that real-root-dev-thing. Feb 12 16:04:07 oh ok Feb 12 16:04:08 :) Feb 12 16:04:19 and fstab i suppose? Feb 12 16:04:58 hm, forgot that one... ;) Feb 12 16:18:36 [g2] around? Feb 12 16:32:30 03garpinc * 10unslung/ (4 files in 3 dirs): Feb 12 16:32:30 Gift Ares team addressed issue that caused gift to crash. Also changed Feb 12 16:32:31 ares and opennap to check out from CVS with date Feb 12 17:19:42 03ptweety * 10unslung/ (4 files in 2 dirs): sasl commands renamed Feb 12 17:24:29 03ptweety * 10unslung/ (6 files in 2 dirs): initial check in. Not ready yet but you can preview the package. Feb 12 17:27:51 I've already managed to create mailboxes in cyrus-imap and am able to use them from my imap client. Feb 12 17:28:05 But I think that some more tests are needed ;-) Feb 12 17:28:46 It would be great if someone wants to help me testing the package Feb 12 17:31:24 night Feb 12 18:05:38 [g2] around? Feb 12 18:06:31 AHA Feb 12 18:06:35 nvm, found my problem. Feb 12 18:10:34 This is cool. Feb 12 18:11:02 <[g2]> dyoung, what's cool ? Feb 12 18:11:31 I can boot openslug or unslung with apex. Feb 12 18:12:22 My immediate issue is it drops into switchbox and cant/doesnt know how to get to the jffs2. Feb 12 18:12:41 and we need to work on enabling the ethernet on apex startup. Feb 12 18:12:55 (thats for openslug) Feb 12 18:13:17 it boots unslung flawlessly, except for maybe the ethernet. Feb 12 18:13:41 <[g2]> dyoung, THAT's awesome ! Feb 12 18:14:12 I discovered the cmdline is passed from apex. Feb 12 18:14:29 makes life easier. Feb 12 18:14:34 <[g2]> AHH, a real-boot loader :) Feb 12 18:15:03 <[g2]> That's excellent Feb 12 18:15:14 So thats pretty much where things are at now. Feb 12 18:15:25 <[g2]> wonderful! Feb 12 18:15:41 I ended up having to pay the 40min penalty once to get openslug back in there. Feb 12 18:16:03 <[g2]> beewoolie's been a great addition and you're always a great help Feb 12 18:16:30 <[g2]> there's alot to be said for clean small designs Feb 12 18:16:47 last night he restructured the way the config works. Feb 12 18:16:57 so now the memory configs are in a config file Feb 12 18:17:14 it'll make fatslug customization take like 20 seconds. Feb 12 18:17:22 <[g2]> that's the second block right ? Feb 12 18:17:39 I meant at boottime Feb 12 18:17:41 oop Feb 12 18:17:43 compile time rather. Feb 12 18:17:59 we didnt get as far as config block yet. ;-) Feb 12 18:18:08 <[g2]> nod. Feb 12 18:18:24 <[g2]> I'm not complaining, just wondering out loud Feb 12 18:18:37 * [g2] is quite pleased Feb 12 18:19:18 <[g2]> For openslug I'm thinking that with APEX we could have two full kernels and rootfs in the box. Feb 12 18:19:26 Yeah, bewoolie's great! Feb 12 18:19:45 <[g2]> and the first kernel/rootfs would be the failsafe one Feb 12 18:19:59 openslug kernel is supposed to know about jffs2 right? Feb 12 18:20:00 <[g2]> The second would be the user kernel/ rootfs Feb 12 18:20:15 <[g2]> Openslug knows jffs2 Feb 12 18:20:17 do you know the cmdline option offhand to tell it that the rootfs is a jffs2? Feb 12 18:20:32 like root=/dev/mtdblock4 fstype blahblah ??? Feb 12 18:20:48 <[g2]> I think we pass multiple fstypes ext2,.... Feb 12 18:21:04 Right, but I need to be explicit for this test. Feb 12 18:21:19 its trying to mount my jffs2 as a ext2. Feb 12 18:21:40 <[g2]> Kernel command line: root=/dev/slug rootfstype=ext2,jffs2 initrd=0x01000000,10M mem=32M@0x00000000 console=ttyS0,115200n8 Feb 12 18:21:50 Aha. Feb 12 18:21:51 okay Feb 12 18:22:00 see apex is passing something else. Its... Feb 12 18:22:21 cmdline *= console=ttyS0,115200 root=/dev/ram0 initrd=0x01000000,10m Feb 12 18:22:51 <[g2]> hmmm, no memory size :) Feb 12 18:23:12 <[g2]> I guess there kernel sets everything up :) Feb 12 18:23:14 Woo hoo! Feb 12 18:23:37 I pasted that (your) command line in, and it works! Feb 12 18:24:00 <[g2]> openslug ? Feb 12 18:24:05 yah! Feb 12 18:24:32 <[g2]> So you're booting OpenSlug directly from APEX correct ? Feb 12 18:24:34 no ethernet though; because we didnt get around to turning it on. Feb 12 18:24:35 Yes Feb 12 18:24:50 I flashed it in there. Feb 12 18:25:17 <[g2]> Looks like dyoung should start a new database on the yahoo group ! :) Feb 12 18:25:41 of..? Non-redboot-booted-openslugger or something? Feb 12 18:25:59 <[g2]> non-redboot bootloaders Feb 12 18:26:18 heh Feb 12 18:26:37 Its not for the faint of heart. Feb 12 18:26:38 <[g2]> You know with APEX and the usbnet fixes we can build static kernels that boot NON-IXP ethernet :) Feb 12 18:26:45 nod Feb 12 18:27:55 <[g2]> I'm *really* warming up to the thought of two complete kernel and rootfs, maybe in their own partitions or not. Feb 12 18:28:25 <[g2]> Certainly, APEX, and kernel+postpended initrd ext2 as the failsafe Feb 12 18:28:35 I like the idea of pushing the button a certain way to get to a rescue kernel Feb 12 18:28:48 and rootfs. Feb 12 18:28:52 <[g2]> That's what that kernel/rootfs would be Feb 12 18:29:08 <[g2]> All in one contiguous partition Feb 12 18:29:32 so whats the proposal for the boot sequence for when you dont push the button, ie: normal boot? Feb 12 18:29:58 <[g2]> It'd also like to see APEX learn jffs2 and then the next partition would be the normal JFFS2 partition with the kernel inside Feb 12 18:30:14 <[g2]> nod. normal boot to the second partition Feb 12 18:31:03 <[g2]> I'd actually like to put a watchdog of sorts in the "normal" partition Feb 12 18:31:20 <[g2]> although that might be stretching it too far Feb 12 18:31:46 <[g2]> but maybe not. Feb 12 18:31:54 to watch what? Feb 12 18:32:12 and if trigged do what? Feb 12 18:32:24 <[g2]> We'll at the end of init or maybe later, it would signal the dog Feb 12 18:32:41 arf arf Feb 12 18:32:51 <[g2]> probably kennel the dog until the next boot Feb 12 18:33:17 so basically if it doesnt finish a sucessful boot, do something differnt. Feb 12 18:33:35 <[g2]> that way if you botch the normal kernel/init, the dog bites and on the warm reboot the failsafe is automatically loaded Feb 12 18:33:42 <[g2]> nod. Feb 12 18:34:46 if apex knows jffs2 doing this via flagfile wouldnt be too much... Feb 12 18:34:49 <[g2]> Obviously will either need a VERY long (10 minutes) timeout or need to speed up init a bunch (or both) Feb 12 18:35:06 <[g2]> Oh yeah. Feb 12 18:35:29 Are you jtag enabled? Feb 12 18:35:39 <[g2]> not yet. Feb 12 18:36:03 <[g2]> I don't know when/if ka6sox has shipped back the FatSlug yet Feb 12 18:36:24 <[g2]> He was going to send it many days ago but I haven't received it or heard from him Feb 12 18:36:44 <[g2]> Maybe Monday :) Feb 12 18:37:17 <[g2]> So the three things to teach APEX are: 1) listen to the reset button during the boot, Feb 12 18:37:33 <[g2]> 2) usb support to allow usb-serial consoles Feb 12 18:37:36 <[g2]> 3) jffs2 Feb 12 18:38:10 <[g2]> and maybe 4) simple partition searching to find the next partitions after the failsafe partition Feb 12 18:39:06 theres a 0 there too. Feb 12 18:39:17 customizable config information would be nice. Feb 12 18:39:30 <[g2]> well that's already in the plans Feb 12 18:39:34 Right. Feb 12 18:40:02 <[g2]> I think beewoolie was planning on using block 2 (one relative) for the APEX config Feb 12 18:40:20 <[g2]> and leave block 3 (one relative) as the sysconf block Feb 12 18:47:27 I just sent you/bewoolie the bootlog and diagnostic. Feb 12 18:48:02 <[g2]> cool Feb 12 18:48:12 I forgot to put back my sysconf earlier so anything related to that will be incorrect. Feb 12 18:51:04 <[g2]> dyoung, so APEX isn't in there right now is it ? Feb 12 18:51:13 in where? Feb 12 18:51:17 <[g2]> FLASH Feb 12 18:51:20 it is Feb 12 18:51:30 <[g2]> can you try a quick test for me ? Feb 12 18:51:33 I dont have redboot anymore, thats why I dont have ethernet right now. Feb 12 18:51:42 sure. Feb 12 18:52:50 <[g2]> boot root=/dev/mtdblock4 rootfstype=jffs2 mem=32M@0x00000000 console=ttyS0,115200n8 Feb 12 18:54:02 thats actually why I was looking the option a bit earlier. trying now Feb 12 18:54:26 <[g2]> thx Feb 12 18:54:27 VFS: Mounted root (jffs2 filesystem). Feb 12 18:54:30 works. Feb 12 18:54:34 <[g2]> I thought it would Feb 12 18:54:45 I had to reason to think it wouldnt. Feb 12 18:55:07 had no reason..... Feb 12 18:55:13 <[g2]> ah.. Feb 12 18:56:12 <[g2]> If we keep the kernel separate (for the User kernel) then APEX wouldn't need to learn jffs2 yet Feb 12 18:56:29 unless we want it to read a flag file in the jffs2 Feb 12 18:56:50 <[g2]> no need to read a flag file when it's got a 128K config block Feb 12 18:57:03 <[g2]> everyone shares the config block Feb 12 18:57:19 <[g2]> and the failsafe kernel can boot without config Feb 12 18:57:35 <[g2]> or has rudundant config Feb 12 19:07:01 <[g2]> dyoung your ethernet should be up Feb 12 19:07:30 <[g2]> dyoung do you know what this marks ? Feb 12 19:07:45 <[g2]> besides a full replacment of all the firmware Feb 12 19:08:01 MY Ethernet never comes up. Feb 12 19:08:03 <[g2]> (if you use a usb-net adapter) Feb 12 19:08:11 the chip is disabled because it never got enabled. Feb 12 19:08:13 oh. Feb 12 19:08:18 <[g2]> dyoung, are you in OpenSlug Feb 12 19:08:28 yes, if I plug in my pegasus it should work fine. Feb 12 19:08:44 Yeah, I'm at the openslug console now Feb 12 19:08:47 <[g2]> nod. on the pegasus Feb 12 19:09:09 <[g2]> cat /etc/init.d/sysconfigsetup Feb 12 19:09:21 I should be more specific. The internal IXP Ethernet never gets turned on. Feb 12 19:09:56 That wont work for me right now. Feb 12 19:10:02 <[g2]> I knew what you meant, and its a great call. Something that happens in Redboot but not in APEX yet Feb 12 19:10:21 because I blew away my sysconf last night. Feb 12 19:10:28 I have it; but I cant get to it easily. Feb 12 19:10:35 <[g2]> don't need it Feb 12 19:10:45 <[g2]> is the ixp425_eth module installed Feb 12 19:10:54 or rather, I'm too l azy to figure out how to wedge it in using apex. ;-) Feb 12 19:11:03 lsmod tells me it is Feb 12 19:11:14 <[g2]> ifconfig eth0 192.168.1.1 up Feb 12 19:11:39 * [g2] predicts a WOOT! Feb 12 19:11:56 well, no. Feb 12 19:12:03 <[g2]> no woot ? Feb 12 19:12:04 because my ethernet still isnt enabled. ;-) Feb 12 19:12:13 The LED is off and all. Feb 12 19:12:24 <[g2]> is it plugged in Feb 12 19:12:29 openslug THINKS its up. Feb 12 19:12:38 of course its plugged in. Feb 12 19:12:46 <[g2]> mine's not :) Feb 12 19:14:33 <[g2]> so Redboot really does enable the hw for the ixp Feb 12 19:14:42 <[g2]> and the kernel doesn't Feb 12 19:14:46 <[g2]> or the modules Feb 12 19:14:49 Yes Feb 12 19:15:03 I knew this because the LED never comes on. Feb 12 19:21:21 btw, what does this mean other than being a complete replacement of all the firmware bits? Feb 12 19:40:28 <[g2]> complete replacement in < 6 months :) Feb 12 19:43:45 Woo. Feb 12 19:43:52 is that faster than expected? Feb 12 19:46:13 <[g2]> I think it's just a neat observation. Feb 12 19:46:33 <[g2]> I mean most of this stuff is all mind candy Feb 12 19:47:01 <[g2]> our version of the "New York Times Crossword Puzzle" Feb 12 19:49:19 I had over 100 inquires about the Slug, and over 1/2 of them were interested in OpenSlug. Feb 12 19:50:28 <[g2]> where are the inquires comming from ? Feb 12 19:50:44 http://www.socallinuxexpo.com/ Feb 12 19:51:06 <[g2]> I saw the expo stuff, did you talk about the slug ? Feb 12 19:52:17 yes Feb 12 19:52:30 I had people bringing their friends over to ask about it. Feb 12 19:52:40 <[g2]> excellent! Feb 12 19:53:45 somebody from Tom's Guides was there. Feb 12 19:54:01 <[g2]> Tom's Networking ? Feb 12 19:54:10 yes Feb 12 19:54:31 <[g2]> that's awesome Feb 12 19:55:03 he has some recording equiptment and come over but wasn't able to get up to the front....got dragged away by some of the organizers... Feb 12 19:56:30 the time scale is pretty compressed. 6mo isnt bad for "entertainment" project. Feb 12 19:56:50 are all the pieces in place now? Feb 12 19:59:54 <[g2]> ka6sox-away, umm what do you mean ? Feb 12 20:01:18 I guess I misunderstood what you were saying earlier. Feb 12 20:02:22 <[g2]> I'm unsure what you meant by all the pieces for what ? Feb 12 20:02:42 complete replacement in under 6 months Feb 12 20:04:17 <[g2]> yes the pieces are pretty much in place Feb 12 20:04:41 wonderful Feb 12 20:04:53 <[g2]> APEX is going to need a little work, but dyoung booted unslung and openslug from a flashed APEX Feb 12 20:05:05 did you guys come up with the spec for the build machine? Feb 12 20:08:32 <[g2]> they were talking about late last night iirc, things looked pretty good Feb 12 20:09:07 okay time for some Vittles. Feb 12 20:11:32 <[g2]> cool. Feb 12 20:11:37 <[g2]> enjoy Feb 12 20:11:44 <[g2]> time for some sleep :) Feb 12 20:13:47 nighty Feb 12 20:13:51 nite [g2] Feb 12 20:13:52 time for din. ;-) Feb 12 20:14:01 <[g2]> dyoung, congrats ! Feb 12 20:14:14 okay It'll work out...dyoung can you easily give me the specs? Feb 12 20:20:50 spec? Feb 12 20:21:06 oh the computer spec. Feb 12 20:21:57 http://www.newegg.com/app/ViewProductDesc.asp?description=56-101-455&depa=0 Feb 12 20:21:57 http://www.newegg.com/app/viewproductdesc.asp?description=20-145-450&DEPA=1 Feb 12 20:21:57 http://www.newegg.com/app/ViewProductDesc.asp?description=22-148-025&depa=1 * 2 Feb 12 20:22:12 http://www.newegg.com/app/ViewProductDesc.asp?description=19-103-390&depa=0 Feb 12 20:22:13 http://www.newegg.com/app/ViewProductDesc.asp?description=19-103-391&depa=0 Feb 12 20:22:13 the two procs Feb 12 20:22:22 that was the latest I think... Feb 12 20:23:00 I would like to have Crucial RAM...its the most reliable. Feb 12 21:58:34 ~jbot Feb 12 21:58:35 jbot is, like, the shipboard computer, but you may call me eddie if it helps you relax Feb 12 21:58:54 What's jbot again? Feb 12 21:59:00 What script that is. Feb 12 21:59:53 blootbot Feb 12 22:00:04 Got it, thanks :) Feb 12 22:00:46 np Feb 12 22:07:13 I'm trying to get unslung's openssh to forward X11. it gets as far as setting DISPLAY=localhost:10, but doesn't seem to be listening on port 6010. any advice? Feb 12 22:47:37 more progress on openssh... it *is* listening. the problem is domain name resolution. "localhost" doesn't resolve to 127.0.0.1 as it should, in spite of there being a listing for it in the hosts file. e.g. "telnet localhost" results in "localhost: Unknown host". odd. Feb 12 22:57:43 03jp30 * 10unslung/ (make/openssh.mk sources/openssh/control): configure openssh to look for xauth in /opt/bin, rather than wherever it was on the cross-compilation host Feb 12 23:23:14 03jp30 * 10unslung/ (make/x11.mk sources/x11/postinst): x11: insist on automake 1.9 at build time; ensure that there is an /etc/nsswitch.conf after installation. this latter is needed so that DISPLAY=localhost:0 can work **** ENDING LOGGING AT Sat Feb 12 23:59:56 2005