irclog2html for #nslu2-linux on 20041004

00:05.32gerrynjrinteresting... xscale dev boards are much more expensive... granted, they have more I/O accessible...
00:14.13ka6sox-awayI'm used to standalone assemblers: will the C call this little program and where will I pass the arguments back to the C?
00:14.24ka6sox-awayon the Stack?
00:15.33ka6sox-awaysri for such dumb questions.
00:19.49ka6sox-awaynm..figured it out
00:30.11gerrynjrah.. the gumstix are cheap 8)
01:23.31*** join/#nslu2-linux kergoth__ (~kergoth@ip71.churchill-park.dfw.ygnition.net)
01:25.43*** join/#nslu2-linux kergoth (~kergoth@ip71.churchill-park.dfw.ygnition.net)
01:26.19ka6sox-awaywb kergoth
01:31.36kergoththanks
01:35.57[g2]greets kergoth!
01:36.14kergothhey
01:37.58gerrynjrhey kergoth
02:00.45*** join/#nslu2-linux CodeWhacker (~CodeWhack@nr24-216-68-196-228.fuse.net)
02:27.51*** join/#nslu2-linux NAiL_ (~repvik@172.80-202-209.nextgentel.com)
02:40.58CodeWhackerping VoodooZ
02:43.57[g2]Hey CodeWhacker what's happening
02:45.05CodeWhackerhey, [g2].  eventful weekend
02:45.16[g2]really ? How so ?
02:45.42CodeWhackermoved my desktop to 2.6, which turned out non-trivial.
02:46.33[g2]I saw some stuff in the logs. Did you wind up on 2.6.6 ?
02:46.44CodeWhackergentoo 2.6.8 has usb issues, and nvidia's vid driver doesn't play well with 2.6.9-r3 vanilla
02:46.59CodeWhackeractually, I found the patch dance for nvidia and I'm on 2.6.9
02:47.25[g2]most excellent..  You're running gentoo right ?
02:47.40CodeWhackeryep
02:47.51CodeWhackerbut my flash reader died in the middle of this
02:47.57[g2]DOH!
02:48.22[g2]I know the pointer to the usbnet 2.6 bk repo :)
02:48.33CodeWhackerwhich made for some great frustration until I realized it was really dead
02:48.53[g2]wow... talk about unlucky
02:49.16[g2]You must have been making too much progress for one individual
02:49.32CodeWhackerso now my desktop has a floppy (because that was the only 3.5" internal flash reader I could find in town on short notice)
02:49.42[g2]:)
02:50.48[g2]So which ver of the nvidia drivers are you running ?
02:51.00CodeWhacker6111, patched
02:51.05[g2]And how much will it cost me to get a pointer to the patch ?
02:51.13CodeWhacker:)
02:51.17CodeWhackerhang on a sec
02:52.19VoodooZHey!!
02:56.44CodeWhackerhttp://00f.net/blogs/index.php?blog=1&cat=16&page=1&disp=posts&paged=2
02:57.18CodeWhackerYou might hit a missing __VMALLOC_RESERVE symbol
02:57.50CodeWhackerif you do, add 'unsigned int __VMALLOC_RESERVE = 128 << 20;' to nv.c
02:58.16VoodooZCodeWhacker: You pinged me master? :)
02:58.42CodeWhackerVoodooZ, yeah.  didn't you say you couldn't mount a drive in openslug?
02:59.07VoodooZYeah, the necessary modules weren't being installed.
02:59.26VoodooZI tried fixing it with [g2] but that caused more problems.
02:59.50CodeWhackerI just did a full build and I can mount a drive fine.
03:00.02CodeWhackerbut I found another small problem
03:00.06VoodooZI spliced them in the ramdisk and it works but there's still depmod problems.(modutils?)
03:00.59CodeWhackerhaven't made any headway on modutils yet.  2.6 needs module-init-tools, but I haven't figgered out how to pull modutils out yet
03:01.38CodeWhackerand the host-built version still coughs on the machine ID.  it's an endian thing
03:01.56[g2]Is that the ELF thing ?
03:02.06VoodooZSo what changes did you make to mount a drive and all?
03:02.51VoodooZIt's hard to understand the bootup sequence as the log is non sequential
03:02.52CodeWhackerVoodooZ, actually none.  I plugged it in, hotplug recognized it and gave it a /dev entry, and I did 'mount -t ext3 /dev/sda1 /mnt/sda1'
03:03.14[g2]after creating the mount point :)
03:03.18VoodooZreally? Is that with a fresh pull from the repo?
03:03.33*** join/#nslu2-linux Christopher (~Christoph@128-193-247-9.resnet.oregonstate.edu)
03:03.56CodeWhacker[g2], when depmod loads a module to check it out, it looks at some header info.  The machine ID is a short, and the i686 version doesn't know to flip it.
03:04.08CodeWhackerVoodooZ, yep
03:04.49VoodooZDarn! Mine doesn't seem to work. I'll try yet again. Unless it's because I'm using a flash drive?
03:04.52[g2]CodeWhacker, so basically depmod is endian borked on the i686
03:04.54CodeWhackerbut beware a full build!  that adjustment I made to the ramdisk size last Thurs nite turns out to be ill-advised
03:05.07CodeWhacker[g2], you are correct, sir!
03:05.26VoodooZreally? Do I need to adjust it?
03:06.20CodeWhackerwhen I did an everything pass this am 9rm -rf tmp), the rootfs build failed with a really strange error.  I finally tracked it down to the disk image filling up.
03:06.58VoodooZwhich error is that? the one you posted about earlier today? About a potentially corrupted fs? I got that too!
03:07.11CodeWhackerI jyst tried with 10240 and got a good boot, but a lot of interim complaints about 'can't access /etc/ld.so.conf'
03:07.37CodeWhackerso there's something lurking
03:08.16VoodooZJust to recap: that value is what? the reserver max size for the compressed ramdisk ?
03:08.26CodeWhacker[g2], btw... maybe it's just subjective, but 2.6 feels snappier than 2.4 did
03:08.53[g2]On the desktop ?  Mine definetly felt that way
03:08.57CodeWhackerIMAGE_ROOTFS_SIZE_ext2 = "10240"
03:08.57CodeWhackerIMAGE_ROOTFS_SIZE_ext2.gz = "10240"
03:09.11VoodooZANd it should! I noticed that the anticipatory I/O scheduler is loaded. I can't wait to try to preemptive stuff too.
03:09.14CodeWhackerfrom ~packages/conf/nslu2.conf
03:09.42VoodooZYeah, but is that the max allowable size or?
03:10.01CodeWhackerthe only thing I have left is 2.6 has borked my bluetooth.  but I think my drivers were pretty old, anyway
03:10.13[g2]We need to trace down why we get the error with openslug
03:10.32[g2]10MB shouldn't blow up
03:10.43[g2]unless it's related to ipkg
03:10.54CodeWhackerVoodooZ, that's the size of the disk that genext2fs creates.  It's not completely full, but it's the full size the kernel is prepared for
03:11.13[g2]I'm wondering if we can do and ipkg DEPENDS but leave it off RDEPENDS for now
03:11.16VoodooZAnd why are those the same size? (sorry for my ignorance)
03:11.31CodeWhacker[g2], it's not blowing up like it did when the size was 12288.
03:12.38VoodooZCodeWhacker: Can you post your BOOT_EXTRA_RDEPENDS line in openslug.conf please?
03:13.08CodeWhackerBOOTSTRAP_EXTRA_RDEPENDS += "kernel-module-usbnet kernel-module-usbserial  \
03:13.08CodeWhackerkernel-module-usbcore kernel-module-usb-storage  kernel-module-uhci-hcd \
03:13.08CodeWhackerkernel-module-pegasus  kernel-module-p8022 kernel-module-kaweth \
03:13.08CodeWhackerkernel-module-ehci-hcd \
03:13.08CodeWhackerkernel-module-scsi-mod kernel-module-sd-mod kernel-module-ext3 kernel-module-jbd \
03:13.09CodeWhackermodule-init-tools"
03:14.28VoodooZYeah, but the repo version of this file doesn't contain the scsi, jbd and ext3 stuff? You added those rigth? That's what I meant by it didn't work out-of-the-box.
03:15.12CodeWhackeroops.  you are correct.  I did add those
03:15.39VoodooZI guess you could push those. They are needed anyways.
03:16.19CodeWhackerINIT: version 2.85 booting
03:16.19CodeWhacker/bin/sh: can't open cache '/etc/ld.so.cache'
03:16.19CodeWhacker[: can't open cache '/etc/ld.so.cache'
03:16.30CodeWhacker[g2], that's what I'm getting now
03:16.44CodeWhackerand lots more of them
03:17.17[g2]CodeWhacker, I think there's an issue with how the ramdisk is mae
03:17.19[g2]made
03:17.24VoodooZI got lots of problems but never got that one. Let me build one, I probably will!
03:17.36[g2]I think it's truncated or there's a race during startup
03:17.52VoodooZCan't I just avoid all that and boot from my flash USB key?
03:18.09CodeWhackerVoodooZ, :)
03:18.34[g2]I was able to get *really* close on a custom 2.6 kernel to booting off my harddrive
03:19.01[g2]I was just a mount -t proc proc /proc away from directly booting onto the hd
03:19.02VoodooZWhat's the JBD module for anyways? Do I need to add one for Vfat?
03:19.08[g2]used by ext3
03:19.25CodeWhackerjournaling
03:19.34[g2]vfat for msdos flash sticks
03:19.49VoodooZI see. I noticed that they have built-in dependencies so just loading scsi for example will get the other ones.
03:19.53VoodooZ[g2]: Thanks.
03:20.45VoodooZI noticed that when they are all loaded the ramdisk overflows or something so I removed all the usbnet/serial which I don't use yet.
03:21.36VoodooZIf I can only get access to my flash drive then I can start experimenting with the I2C stuff.
03:21.44CodeWhackerspeaking of sticks, I ran across one of these at Target today: http://www.efunctional.com/pqifi128us.html
03:21.48VoodooZThen again, the i2c devices are not being created properly so..
03:22.18TierstenCodeWhacker: That's the USB1 version
03:22.21TierstenThe USB2 version is black
03:22.24VoodooZBuilding....
03:22.41VoodooZWhat are those CodeWhacker? (I can't copy and paste from windows it seems.
03:23.15hculverI'm having trouble getting nfs-utils to compile under oe and it seems to be related to linux-libc-headers.
03:23.23CodeWhackertrue.  I got a USB 2.0, but it actually is branded by GE
03:23.41TierstenI've got one of the USB 2 PQI Intelligent Sticks
03:23.47TierstenWorks fine and it fits nicely into my wallet
03:23.57VoodooZNewbie question of the day: When you 'ping' someone in IRC, is it just to get his attention?
03:24.02CodeWhackerVoodooZ, it's an Intelligent Stick USB drive.  
03:24.18VoodooZintelligent??? What does it do more than your plain one?
03:24.22TierstenNothing
03:24.23CodeWhackermine has some intensely funky partition on it that linux doesn't like
03:24.25TierstenIt's just a name
03:24.40hculverIt compiles OK with the snapgear toolchain, but I get lots of conflicting definitions from the headers in the oe toolchain.
03:24.40TierstenCodeWhacker: That's the useless private partition thing
03:25.02TierstenYou can defeat the "security" on them fairly easily
03:25.02CodeWhackerWin98 mounts it OK.  Going to try it on XP at work tomorrow, and then probably format it.
03:25.18TierstenDidn't know that PQI rebrand them for anybody else though
03:25.29CodeWhackerTiersten, is it the same bogus scheme Lexar uses?
03:25.42TierstenThe horrible PQI Windows app lets you create or delete the special partitioning
03:26.13TierstenIt's not encrypted or anything like that but they don't say it is anyway
03:26.34TierstenIt's just got a wacky partition near the end of the stick which doesn't get recognised by standard software
03:27.23CodeWhackerI should repartition mine with a "secure" area and then poke around
03:27.30TierstenThat's all their software is doing.  It's not quite as bad as Lexar since they don't claim encryption
03:27.50TierstenFor the average person out there it's secure enough anyway
03:27.58CodeWhacker'XOR flamingo' *is* encryption! :)
03:28.14TierstenYeah but so is ROT13 :)
03:28.30Tiersten*cough* Adobe eBooks
03:28.44TierstenIf you're that CueCat company then BASE64 is encryption ;)
03:29.43CodeWhackeractually, cuecats used XOR 0xCC
03:29.54TierstenDidn't they use base64 as well?
03:29.56CodeWhacker*then* they used base-64 on top of that
03:30.17Tierstenbecause that's secure...
03:30.28Tierstentriple DES is only actually using two keys anyway
03:30.53TierstenYou're just doing 3 operations but only with 2 keys
03:31.00CodeWhackerI know, but that sorta takes the steam outta the joke
03:31.12Tierstenheh
03:32.02TierstenUse Rijndael anyway :)
03:32.04TierstenMuch nicer
03:32.36CodeWhackerRC5
03:32.52CodeWhackerthat one never really caught on
03:32.59TierstenWhat?  RC5?
03:33.06CodeWhackeryeah.
03:33.35*** join/#nslu2-linux Neon12345 (~Neon12345@pD95308C5.dip.t-dialin.net)
03:34.02TierstenIt's used in a few things but not particularly much no
03:34.28VoodooZCodeWhacker: Told you! Now I'm getting hundreds of those ld.so.cache errors like you! That's luck!
03:34.39TierstenRC6 got turned down for AES
03:34.54VoodooZIt at least gave me a shell!
03:35.19CodeWhackerVoodooZ, that's not luck, it's meticulous attention to the details of a master.
03:35.38VoodooZheheheh. I'm more of a copy-cat!
03:36.27CIA-703CodeWhacker 07 * r1.2329 10packages/conf/openslug.conf: Add the modules needed to mount disks.
03:36.41VoodooZCodeWhacker: Thanks!
03:36.48CodeWhackernp
03:37.28VoodooZI already have a hard time understanding like that without messing up the whole OE structure trying to post my changesets! :)
03:38.27VoodooZThe weird thing though Is it doesn't tell me which /dev/sda entry is associated with my USB key and Hard disk. Shouldn't it?
03:40.55CodeWhackeryour key and drive will be separate entries.  /dev/sda and /dev/sdb
03:41.39VoodooZI know but shouldn't specify which one is which when it first enumerates them like on a PC?
03:43.13VoodooZdoing fdisk on any /dev/sdn entry of trying to mount doesn't work so I guess I'm still missing something. Maybe I'll try pulling your copy just in case.
03:43.35CodeWhackerhmmm.... won't mount my usb key
03:43.43hculverAnybody out there running 2.6 (with kernel headers) who can tell me whether nfs-utils-1.0.6 compiles OK?
03:44.11CodeWhackerI get a /dev/sdb1, but no 'mount -t vfat' joy
03:44.43VoodooZcould be a problem witht the vfat support. try formatting it to ext2/3.
03:44.59VoodooZI can't even get a /dev/sdn assigned!
03:45.10CodeWhackerI'll try that with another key.  this one has stuff I want to keep on it
03:46.02CodeWhackerhculver, stby... compiling
03:46.15VoodooZSo is there a hard limit to changing the ramdisk size constant?
03:47.37CodeWhackerhculver, nfs-utils-1.0.6-r4 from portage compiles against 2.6.9-rc3 with no complaints
03:48.25hculverhmmm, I've got no trouble with the snapgear toolchain which has 2.4.22 headers, but I've got trouble with oe which is using linux-libc-headers-2.6.8.
03:49.15hculverCodeWhacker: I'm getting conflicting types between linux/nfs.h and rpcsvc/nfs_prot.h.
03:49.38CodeWhackerhculver, is that an oe build giving you probs?
03:49.48hculverCodeWhacker: yeah.
03:50.07CodeWhackerlemme try building it in oe.  I just checked my desktop
03:50.56hculverCodeWhacker: I'm trying add a new package to oe (nfs-utils).
03:51.06CodeWhackergood man!
03:51.54CodeWhackerand that would explain why I get 'nothing provides nfs-utils' :)
03:52.35CodeWhackerhculver, want to send me your .oe file?
03:52.51hculverCodeWhacker: I'm gonna have to compare the linux-2.6.x headers for nfs.h and nfs_prot.h and those from linux-libc-headers.
03:53.39hculverCodeWhacker: to the address I see in irc?
03:57.24[g2]CodeWhacker, VoodooZ, hculver, jacques -- any others here's an openslug question
03:58.01[g2]I'm toying with the idea of just capturing a oe build snapshot.
03:58.03VoodooZIf we have an openslug question? Not a specific one no, but lots of problems :)
03:58.56[g2]And creating a shell script that more or less builds all the tools/toolchains to build the source.
03:58.58VoodooZAnything! As long as I can get a working setup. I'll even send paypal you a beer! :)
03:59.28[g2]Well VoodooZ we had a 2.6 kernel build shell script
03:59.46VoodooZI know and i had way more success with that one too!
03:59.47[g2]The real question is what does one want on the ramdisk
04:00.26[g2]long-term I think oe is great fit for openslug
04:00.40VoodooZI guess in my case because I'm aiming at a robot board, I just need it to support my usb key so I can boot off it and then I can modify the rest.
04:01.21[g2]short-term, we may want to stablize the kernel and fight the oe learning curve later
04:01.44[g2]I don't know, I'm wondering how you guys are feeling about that
04:02.35CodeWhacker[g2], I think it's a not-half-bad idea for more casual developers that don't want to climb the oe curve
04:03.12CodeWhackerbut it implies that someone will have to periodically refresh it... unless you make it into an oe target. :)
04:03.20kergothrefresh what?
04:03.34[g2]ahh... just the man we need
04:03.45*** join/#nslu2-linux quantin (~quantin@194-208-136-074.TELE.NET) [NETSPLIT VICTIM]
04:05.04[g2]kergoth, for openslug I seem to be running into some stumbling blocks because I'm an oe idiot
04:05.31[g2]that coupled with the 2.6.x issues is like fight a battle on two fronts
04:06.33[g2]it's gets compounded by the fact that oe runs the latest stuff and ever though our nslu2 specific stuff doesn't change our builds change from hour to hour
04:07.02kergothso create a .conf like we did for openzaurus-3.5.1
04:07.10kergothexert control over what versions of the packges you want.
04:07.25[g2]that would certainly help
04:07.31kergothits not hard to make builds from oe a reproducable element
04:08.21CodeWhackerhculver, this?
04:08.25CodeWhacker| rmtab.c:27: error: `NFS_STATEDIR' undeclared (first use in this function)
04:09.42VoodooZWhatever you guys come up with on the openslug front, post it here and I'll read it tommorow morning as I my 5km race and hockey game killed and I need to crash bad.
04:09.49VoodooZGood night/morning!
04:10.04hculverThat was probably one of them, but most of them were conflicting types for things like NFS_OK between linux/nfs.h and rpcsvc/nfs_prot.h
04:11.47CodeWhackerthat was the only error I got, and the compile crashed on it
04:12.27[g2]dyoung-brikmastr, we've got two issues 1) kernel issues, 2) process and rootfs issues -- With a shell script we can divide and conquer
04:12.34hculverCodeWhacker, I'm oemaking again (and I didn't save the errors from the last make).
04:13.23dyoungI guess I can stop being the brickmaster now.
04:13.28ka6sox-awayheh
04:13.39CodeWhackerhculver, tmp/work/nfs-utils-1.0.6-r0/temp/ keeps logs
04:14.35dyoungI have yet to make a workable rootfs that works with 2.6.7 properly.  I havnt given a whole lot of effort over the past 2 weeks though.
04:15.34hculverCodeWhacker: making all in nfs, compiling exports.c ...
04:16.05[g2]dyoung, define workable
04:16.14hculverexport/project/openembedded/build/tmp/staging/armeb-linux/include/linux/nfs.h:40: error: conflicting types for 'NFS_OK'
04:16.39hculverexport/project/openembedded/build/tmp/staging/armeb-linux/include/rpcsvc/nfs_prot.h:33: error: previous definition of 'NFS_OK' was here
04:16.56CodeWhackerhculver, this from 'oemake nfs-utils'?
04:17.09hculverCodeWhacker: yes.
04:17.25kergothheh, does it really need to pull in linux/nfs.h?
04:17.45CodeWhackerstrange, I don't see that at all
04:18.00CodeWhackerit blows up in rmtab.c here
04:18.39dyoungWell, I still have broken device nodes.  Like I said though, I havnt touched it in 2 weeks.
04:20.16CodeWhackergotta go, guys.  I'll check in from work tomorrow (oop... later today)
04:20.32hculverthanks CodeWhacker.  Have a good one.
04:21.03*** part/#nslu2-linux CodeWhacker (~CodeWhack@nr24-216-68-196-228.fuse.net)
04:23.19dyoungbbiab
04:24.13hculverkergoth, any advice.  my nfs-utils build fine with the snapgear toolchain (containing 2.4.22 kernel headers) but I'm having trouble with the oe linux-libc-headers (2.6.8).
04:24.51kergothdoes it build against 2.6 at all?
04:24.55kergothif not, it needs to be fixed
04:25.05kergoth2.6 is the current stable kernel series.
04:25.24hculverCodeWhacker just verified for me that yes nfs-utils-1.0.6 compiles fine against 2.6.
04:26.42hculverI'll keep digging, I just thought you may have seen similar things before and point me in the right direction.
04:28.32kergoththe sanitized headers are great, but they arent perfect. they're missing headers for mipsel and others, and it wouldnt surprise me if you've found a bug.
04:28.43kergothneeds investigation
04:29.21hculverok, thx
04:30.02*** join/#nslu2-linux quantin (~quantin@194-208-136-074.TELE.NET) [NETSPLIT VICTIM]
04:31.06hculvernite all
04:32.38*** join/#nslu2-linux nudot (~tking@adsl-66-159-198-116.dslextreme.com)
04:37.05[g2]nite all
04:46.27*** join/#nslu2-linux dyoung (~dyoung888@cpe-24-31-90-123.hawaii.rr.com)
05:25.33Christopheri am trying to compile the unslung-image with oe and keep getting an error that says that genext2fs couldn't allocate a block (no free space) any ideas?
06:58.46*** join/#nslu2-linux dyoung (~dyoung888@cpe-24-31-90-123.hawaii.rr.com)
07:11.36*** join/#nslu2-linux rwhitby-treo (~upirc@203.202.55.156)
07:12.38rwhitby-treoping
07:16.50*** part/#nslu2-linux rwhitby-treo (~upirc@203.202.55.156)
07:34.12*** join/#nslu2-linux Latent (~michael@203-114-158-250-dsl.inspire.net.nz)
08:11.12*** join/#nslu2-linux CIA-7 (~CIA@to.je.spocco.com)
08:39.42dyoungArr.
09:06.04*** join/#nslu2-linux pb_ (~pb@dsl-62-3-66-204.zen.co.uk)
09:48.00*** join/#nslu2-linux Latent` (~michael@203-114-158-188-dsl.inspire.net.nz)
09:52.19*** join/#nslu2-linux Latent`` (~michael@203-114-158-188-dsl.inspire.net.nz)
10:03.26dyoungSure is quiet today.
10:04.18Latent``yeap
10:04.24Latent``hows your s3 board going?
10:12.00dyoungIts a learning challenge.
10:12.27Latentwell learning is good :)
10:12.35dyoungYes it is!
10:12.42dyoungLearning is my friend.  usually.
10:22.45Latentgoodnight dyoung-sleep

Generated by irclog2html.pl by Jeff Waugh - find it at freshmeat.net! Modified by Tim Riker to work with blootbot logs, split per channel, etc.