| 00:00.15 | rwhitby-web | have you made a changeset for the new patch? |
| 00:00.32 | CodeWhacker | I want to see if it works first. Then I'll make a changeset. |
| 00:01.09 | rwhitby-web | making a changeset is the easiest way to get [g2] and I to test it. Then we can build our own from the changeset that you can email to us. |
| 00:01.37 | rwhitby-web | [well, maybe not the easiet, but the most repeatable and collaborative] |
| 00:01.42 | CodeWhacker | fair enough... brb |
| 00:02.08 | rwhitby-web | [g2] what do you think about posting oe changesets to the mailing list? Get some real content on there ... |
| 00:03.16 | Tiersten | I'm being lazy. Is the IXP425 USB port a master or slave? |
| 00:04.36 | [g2]-away-buildi | I'd lazily guess master since the devices are all slaves |
| 00:05.03 | Tiersten | Not the NEC USB2.0 one. The one built into the IXP425 |
| 00:05.24 | Tiersten | I've just loaded the datasheet and it just says "The integrated USB 1.1 interface is a device-only controller." |
| 00:05.50 | Tiersten | Whats that mean then? You can only attach devices to it i.e. it's a master or it only appears as a device i.e. it's a slave?? |
| 00:06.04 | [g2]-away-buildi | cool. then we could daisy-chain nslu2's :) |
| 00:06.15 | Tiersten | Only USB1.1 tho :) |
| 00:06.17 | [g2]-away-buildi | but only at 11Mbs |
| 00:06.18 | tking | dunno, I'll look at the semantics. |
| 00:06.47 | [g2]-away-buildi | We need the USB2.0-GigE adapters |
| 00:08.30 | Tiersten | The debug unit is quite interesting. Would make kernel dev much easier |
| 00:08.37 | tking | I'll settle for usb-100bt! |
| 00:08.46 | jacques | http://groups.yahoo.com/group/nslu2-linux/message/1562 how's this work with no 3.3V ? |
| 00:09.06 | Tiersten | It's powered from the PC |
| 00:09.31 | jacques | oooh I just read the last part |
| 00:09.33 | Tiersten | Ground is still connected so it's fine assuming it is a 3.3v output |
| 00:09.49 | jacques | Start minicom and set the port to /dev/ttyUSB0 |
| 00:09.55 | jacques | I hadn't noticed that before |
| 00:10.33 | jacques | that could be useful since I have a dearth of serial ports |
| 00:10.38 | CodeWhacker | rwhitby: how do I export a changeset for you guys? citool appears to have created it. |
| 00:11.34 | Tiersten | jacques: Yeah. It's a USB -> serial dongle that is made to only output 3.3V as that's what the mobile phone uses |
| 00:11.42 | rwhitby-web | cw: bk send |
| 00:13.27 | *** join/#nslu2-linux rwhitby-web (~88b602dd@CPE-144-136-196-72.sa.bigpond.net.au) |
| 00:14.25 | [g2]-away-buildi | CodeWhacker, that image hangs when uncompressing the kernel |
| 00:14.36 | [g2]-away-buildi | at least on the first attempt |
| 00:14.59 | CodeWhacker | bummer! |
| 00:16.07 | [g2]-away-buildi | I can try again is 20-30 minutes |
| 00:16.31 | [g2]-away-buildi | I've got to go for a few minutes |
| 00:17.01 | [g2]-away-buildi | bbiab 20-30 |
| 00:17.29 | *** join/#nslu2-linux rwhitby-web (~88b602dd@CPE-144-136-196-72.sa.bigpond.net.au) |
| 00:17.48 | CodeWhacker | rwhitby: msg me an email addy for the changeset |
| 00:18.20 | CodeWhacker | changes en route |
| 00:19.23 | Tiersten | Not sure the IXP425 USB is usable actually. The controller doesn't seem to be detected at all on the PCI bus. |
| 00:19.46 | Tiersten | You only see some sort of coprocessor which must be the network thing and the NEC USB chip |
| 00:20.00 | tking | I'm wondering if the Power Management of the chip has shut it down. |
| 00:20.26 | Tiersten | Yeah. You might need to tell it to turn it on |
| 00:20.28 | tking | there are registers that control what units are on/off |
| 00:20.50 | chris_ | re |
| 00:22.03 | Tiersten | Looks like it's a slave device |
| 00:22.20 | Tiersten | "The UDC support 16 endpoints and can operate half-duplex at a rate of 12 Mbps (as a slave only, not as a host or hub controller)." |
| 00:23.22 | tking | hmmm... |
| 00:24.37 | *** join/#nslu2-linux rwhitby-web (~88b602dd@CPE-144-136-196-72.sa.bigpond.net.au) |
| 00:25.32 | Tiersten | Have to set the UDE bit in the UDRCR to enable it. Unless Intel have written a driver for it, it'll mean a lot of work |
| 00:25.42 | tking | could be a standard usb device with a driver. |
| 00:25.57 | Tiersten | Makes sense that it's a USB slave actually if it's designed to go into routers and DSL modems |
| 00:25.58 | rwhitby-web | I've just created an nslu-changesets@lists.sourceforge.net mailing list where we can share changesets easily before we push them back up the tree to kergoth. |
| 00:26.04 | tking | being a slave limits its use |
| 00:26.19 | chris_ | stupid question: what are changesets? |
| 00:26.20 | rwhitby-web | [will let you know when sf.net actually creates it - it takes some time] |
| 00:26.34 | Tiersten | Some DSL modems let you plug it in directly via USB even though they've got ethernet |
| 00:26.51 | tking | yes, I've seen that. |
| 00:27.03 | tking | I saw a Satellite modem the same way. |
| 00:27.20 | Tiersten | One for when I'm really bored one weekend I think :) |
| 00:27.34 | tking | yes |
| 00:27.53 | chris_ | what're changesets |
| 00:27.54 | chris_ | ? |
| 00:28.02 | Tiersten | Bitkeeper diffs basically |
| 00:28.09 | chris_ | ok |
| 00:28.11 | tking | I understand that the second serial port is unusable. |
| 00:28.22 | Tiersten | You can receive with it but not transmit |
| 00:30.02 | tking | receive the UPS alarm then. |
| 00:30.18 | tking | (probably the best and highest use for it) |
| 00:30.55 | Tiersten | Yeah. Anything which just sends out data really. |
| 00:31.10 | Tiersten | You can get weather station kits which just send out data via their serial port |
| 00:31.21 | tking | that is a good use too. |
| 00:31.23 | Tiersten | Plugging one into a NSLU2 would be neat. Instant web interface |
| 00:31.40 | jacques | GPS ? |
| 00:31.44 | tking | is php too big to run on one of these? |
| 00:32.02 | Tiersten | jacques: That as well. NTP server |
| 00:32.04 | chris_ | has anyone played with the python that was ported? |
| 00:32.12 | jacques | Tiersten, yep my thoughts too |
| 00:32.25 | Tiersten | I think you can just about manage to get it running but you'll probably need to get rid of the ramdisk and use the HD as root |
| 00:32.29 | chris_ | dhcp would be boss |
| 00:33.58 | tking | with USB-> ethernet I would make this my router |
| 00:34.05 | tking | and dhcpd server |
| 00:34.15 | chris_ | tking: that's what i said: dhcp would be boss |
| 00:34.25 | Tiersten | My AP is my DHCP server |
| 00:34.31 | chris_ | tking: probably wouldn't be difficult too |
| 00:34.38 | jacques | tking, I am confident we will have USB->ethernet soon |
| 00:34.52 | tking | I think so too. looks like that is moving along |
| 00:35.02 | chris_ | jacques: what do you mean by "USB->ethernet"? |
| 00:35.10 | Tiersten | A USB ethernet dongle? |
| 00:35.21 | jacques | yeah |
| 00:35.43 | jacques | so we can have two net interfaces |
| 00:35.45 | tking | that gives us the second etherport. |
| 00:35.48 | Tiersten | damn. where was that link... |
| 00:35.58 | Tiersten | Ah. Found it http://www.xbitlabs.com/news/video/display/20040607023341.html |
| 00:36.01 | chris_ | jacques: why're people interested in doing that? |
| 00:36.05 | Tiersten | Now that would be interesting |
| 00:36.08 | tking | I'm watching carefully the USB flash in port one. |
| 00:36.11 | Tiersten | USB2.0 graphics card! :) |
| 00:36.16 | jacques | ooooooh@ |
| 00:36.36 | tking | this is NOT exactly Workstation class hardware! |
| 00:36.51 | tking | Xscale what....266mhz? |
| 00:36.53 | Tiersten | Yeah but would make a reasonable xterm :) |
| 00:36.57 | tking | true |
| 00:37.20 | tking | how much is the video adapter? |
| 00:37.25 | Tiersten | Says $80 in Tokyo |
| 00:37.31 | Tiersten | Unlikely to support Linux however *shrug* |
| 00:37.37 | chris_ | can someone explain to me why having 2 ethernet ports on the nslu2 is desirable? |
| 00:37.44 | tking | router |
| 00:38.15 | Biovore | lol |
| 00:38.45 | Biovore | tking: your right.. its not workstation system.. |
| 00:38.47 | tking | a replacement for the BIG, noisy, power hungry P1 I'mm running as a firewall/router |
| 00:39.02 | chris_ | tking: oh i get it. |
| 00:39.03 | Tiersten | Buy a WRT54G? |
| 00:39.06 | Biovore | yeah.. actualy it would be good for that.. |
| 00:39.16 | Tiersten | They've got two ethernet as standard and you get 802.11g for free |
| 00:39.19 | chris_ | tking: so you may not even attach a storage device to the nslu2 anymore - just use it as a firewall/nat |
| 00:39.23 | chris_ | tking: that's pretty clever :) |
| 00:40.20 | tking | well, Actually I will attach storage..what I'm looking for is USB Flash as /dev/hda and /dev/hdb to be mounted and dismounted (the 200gb USB2 Drive) |
| 00:42.11 | tking | bbs.. |
| 00:42.34 | rwhitby-web | nslu-changesets@lists.sf.net is now operational. Please send any changesets there so that we can stay in sync with each other to test things before submitting to kergoth. |
| 00:42.48 | rwhitby-web | http://lists.sourceforge.net/lists/listinfo/nslu-changesets |
| 00:47.50 | rwhitby-web | So if you make a change to your local oe repository, then bk send it to nslu-changesets@lists.sf.net and then we can all test it out for you. Make sure you give good comments so we know what we are testing. |
| 00:58.36 | chris_ | would anyone mind if i added a dhcp server to the nslu2? |
| 00:58.41 | chris_ | i just got one to cross compile |
| 00:59.01 | Biovore | give it a whirl.. |
| 00:59.09 | chris_ | i'm not going to add it to the top level Makefile |
| 00:59.19 | chris_ | but i'll cvs add the supporting files |
| 00:59.26 | Biovore | It could be a usefull feature.. if know one else has done it.. |
| 00:59.40 | Biovore | I havn't a clue whats in the CVS.. |
| 01:02.24 | chris_ | who put together the unslung cvs tree and the Makefile system? |
| 01:14.38 | rwhitby-web | chris_: that would be me. |
| 01:14.52 | rwhitby-web | it's a stop-gap measure until we get Unslung into OE |
| 01:16.40 | chris_ | rwhitby-web: i really like it |
| 01:17.09 | rwhitby-web | It's based on OpenWRT, which was based on buildroot from uClibc. |
| 01:17.25 | rwhitby-web | There is little original input from me :-) |
| 01:17.39 | CIA-9 | 03blunck2 * 10unslung/make/dhcp.mk: added. a makefile that will build a dhcp server |
| 01:18.11 | CIA-9 | 03blunck2 * 10unslung/sources/dhcp.control: |
| 01:18.11 | CIA-9 | added. control file used by the ipk utility. describes the dhcp server |
| 01:18.11 | CIA-9 | recently added |
| 01:18.40 | rwhitby-web | if you like it, then you'll love OE |
| 01:18.51 | chris_ | i like how easy it is for me to port |
| 01:19.20 | CIA-9 | 03blunck2 * 10unslung/sources/dhcp.rc: added. start/stop script for the dhcp server |
| 01:20.27 | [g2]-away-buildi | root@nslu2:/# rsync -av --dry-run /etc/ /etc2 |
| 01:20.27 | [g2]-away-buildi | building file list ... done |
| 01:20.27 | [g2]-away-buildi | .configured |
| 01:20.27 | [g2]-away-buildi | default/devpts |
| 01:20.27 | [g2]-away-buildi | default/rcS |
| 01:20.28 | [g2]-away-buildi | default/usbd |
| 01:20.43 | [g2]-away-buildi | you get the idea |
| 01:22.44 | [g2]-away-buildi | chris_, adding rsync to the nslu2 oe root fs was as easy as adding the following two lines to my local.conf configuration file |
| 01:22.47 | [g2]-away-buildi | BOOTSTRAP_EXTRA_DEPENDS = "rsync" |
| 01:22.47 | [g2]-away-buildi | BOOTSTRAP_EXTRA_RDEPENDS = "rsync" |
| 01:23.44 | rwhitby-web | [g2]: but that won't put it into /opt for an Unslung package yet ... |
| 01:24.45 | [g2]-away-buildi | rwhitby-web, Unslung has it's own set of issues. You know I'm a anyu2 guy at heart :) |
| 01:26.28 | tking-away | chris_ my current "silent" server is a Via Epia 533 cost me over $300 with memory and CF HD. |
| 01:26.47 | tking-away | I'm hoping this little box can do the same thing for about 1/2 that |
| 01:29.11 | tking-away | (and draw 1/2 the power |
| 01:29.24 | CodeWhacker | [g2]: got serial (ther hard way... http://www.lantronix.com/products/eds/xport/) |
| 01:30.20 | CodeWhacker | also got my kernel to boot, but it complains about the usbnet modules. "depmod: ELF file /lib/modules/2.4.22-xfs/kernel/drivers/usb/usbnet.o not for this architecture |
| 01:30.20 | CodeWhacker | " |
| 01:30.29 | Tiersten | Only serial and 3 IO though |
| 01:32.20 | [g2]-away-buildi | that's what jacques was saying the other day iirc, that the undefined references were from the 2.6 tree and it looked like a maix up |
| 01:32.34 | [g2]-away-buildi | mix |
| 01:33.09 | *** join/#nslu2-linux rwhitby-web (~88b602dd@CPE-144-136-196-72.sa.bigpond.net.au) |
| 01:33.17 | [g2]-away-buildi | wb |
| 01:33.38 | [g2]-away-buildi | that's what jacques was saying the other day iirc, that the undefined references were from the 2.6 tree and it looked like a mix up |
| 01:34.06 | Tiersten | Hi dyoung |
| 01:34.15 | dyoung | Ahoy matey! |
| 01:34.23 | tking-away | arr |
| 01:35.11 | [g2] | Avast! Will you guys quit being so *serious*! |
| 01:35.14 | dyoung | The build system is built. |
| 01:35.18 | CodeWhacker | well, now that I can do some testing, think I'll see what happens if I swap the virgin modules from 2.4.22 back in |
| 01:35.47 | CodeWhacker | dyoung: good... but will it build a build system? |
| 01:35.50 | CodeWhacker | :) |
| 01:36.04 | [g2] | dyoung, so bk and oe are running and you've built your image ? |
| 01:38.24 | dyoung | I didnt get that far yet. I did get bk now. |
| 01:39.06 | dyoung | Codewhacker, with some luck it will be building me a ARM based build system. ;) |
| 01:43.56 | joule360-away | Interesting finding. I believe you cannot create a symlink in a share mounted via samba. I have been working on finished the courier-imap build, and it has been reproducible crashing tonight trying to create a symlink. I realize that previously I had the nslu2 (where I store the whole unslung build tree) mounted via nfs, but today I mounted via samba, and now I am seeing this problem. Has anyone else noted this about samba? |
| 01:44.41 | Tiersten | Probably. SMB doesn't support symlinks as Windows doesn't (yes, I know you can sort of) |
| 01:45.29 | Tiersten | Samba will follow a symlink but from it's view, it's just another copy of that file. It doesn't tell the remote system that it's a symlink |
| 01:45.55 | *** join/#nslu2-linux rwhitby-web (~88b602dd@CPE-144-136-196-72.sa.bigpond.net.au) |
| 01:48.33 | rwhitby-web | CodeWhacker: just sent a reply on nslu-changesets to discuss your changeset - I think it should be split into a separate one for the usbnet.c patch, and then one for the openslug kernel. |
| 01:50.03 | [g2] | root@nslu2:/etc/rc2.d# ssh localhost |
| 01:50.28 | [g2] | that's funny with no ethernet but loopback it works |
| 01:51.18 | CodeWhacker | rwhitby: sounds reasonable. at the mo, the whole thing fails miserably. :) |
| 01:53.14 | [g2] | nod. to the split changeset idea |
| 01:55.13 | chris_ | damn i can't find a copy of libtermcap source. |
| 01:58.12 | rwhitby-web | [g2]: Do you want to create an openslug_2.4.22.oe file in packages/linux, and a corresponding openslug-2.4.22 directory in there too - then you guys can go for it on adding stuff to the 2.4.22 kernel, in preparation to getting an openslug-2.6.7.oe file in there later ... |
| 01:59.15 | rwhitby-web | [should openslug replace anyu2 or nslu2plus?] |
| 02:00.05 | dyoung | parsing .oe files. |
| 02:00.07 | [g2] | rwhitby-web, no. I don't beleive in 2.4.x. I'm playing with it now as I learn about oe. 2.6.7+ makes the most sense to me |
| 02:00.22 | [g2] | openslug==anyu2 |
| 02:00.29 | *** join/#nslu2-linux McTails (~mctails@dialup-4.237.251.248.Dial1.NewYork1.Level3.net) |
| 02:00.29 | dyoung | OpenSlug == AnyU2 |
| 02:00.57 | [g2] | tr/[A-Z]/[a-z]/g |
| 02:01.13 | dyoung | Okay fine. |
| 02:01.15 | [g2] | just playing :) |
| 02:01.22 | chris_ | it's surprising how difficult it is to obtain libtermcap source code |
| 02:01.23 | dyoung | openslug == anyu2. |
| 02:01.27 | chris_ | i can't find any to save the life of me! |
| 02:01.59 | dyoung | How long did it take you guys to get the oe build built? |
| 02:02.05 | rwhitby-web | I think OpenSlug (published name) / openslug (filename) is the incarnation of AnyU2 in OpenEmbedded. |
| 02:02.19 | [g2] | 3 hours 2 minutes from scratch today with downloads |
| 02:02.32 | Tiersten | chris_: Soooo... what do I get for telling you where to get termcap? :) |
| 02:02.44 | [g2] | rwhitby-web, agreed |
| 02:02.46 | chris_ | Tiersten: well if i'm good, then you get mysql ported ;-) |
| 02:03.00 | Tiersten | uh? why do you need termcap? |
| 02:03.01 | chris_ | i found termcap-compat from the slackware 8.1 distro - am trying to work with that now |
| 02:03.03 | Tiersten | You compiling the utils as well? |
| 02:03.16 | dyoung | openslug is a filename. It irritates me havnig FileNameLikeAWikiWord. |
| 02:03.28 | chris_ | Tiersten: you think if i just build mysqld that i won't need -ltermcap ? |
| 02:03.30 | Tiersten | chris_: ftp://ftp.gnu.org/gnu/termcap/ |
| 02:03.49 | Tiersten | Dunno. You could try |
| 02:03.59 | dyoung | Okay, since i have at least 3 hours to kill, I'm gonna go to the hardware store. |
| 02:04.08 | rwhitby-web | [g2]: agreed also that openslug_2.4.22 is a non-starter. But we should not be sending changesets to linux-nslu2_2.4.22.oe (or it's corresponding files) which move it away from the stock firmware (apart, perhaps for bug-fixes). |
| 02:04.10 | jacques | sigh |
| 02:04.21 | [g2] | is it a wiki limitation that things can't be names "Filename like a Wiki word" |
| 02:04.24 | dyoung | SlugPlus. |
| 02:04.39 | jacques | I used to live 3 miles from a cool Fry's, now it's about 60 |
| 02:04.49 | Tiersten | [g2]: You have to have a single word as the node name |
| 02:05.04 | [g2] | Tiersten, thx |
| 02:05.10 | rwhitby-web | Unslung, SlugPlus, OpenSlug - I like it. |
| 02:05.21 | [g2] | nod. |
| 02:05.25 | dyoung | It makes sense to me at least. |
| 02:05.26 | Tiersten | Wasn't there Snail or something? somebody mentioned it yesterday |
| 02:05.35 | rwhitby-web | [g2]: you can display it differently by using [[WikiWord Wiki Word]] |
| 02:05.41 | [g2] | yes a slug + a shell |
| 02:05.45 | dyoung | Yes, a slug with a shell. but I forget what it descrives |
| 02:06.30 | rwhitby-web | [g2]: I didn't like opennslu2 due to the repeated "n" :-) |
| 02:06.49 | [g2] | escargot ? |
| 02:06.58 | chris_ | this is ignorant: do you need to strip .a files? |
| 02:07.10 | rwhitby-web | OpenSnail |
| 02:07.24 | Tiersten | chris_: No |
| 02:07.31 | Tiersten | Stripping it breaks it |
| 02:07.32 | [g2] | I do like the OpenNSLU monikier |
| 02:07.38 | chris_ | Tiersten: thanks. i'm a java programmer |
| 02:08.37 | Tiersten | You're not going to try porting Kaffe? :) |
| 02:08.45 | chris_ | lol |
| 02:08.57 | chris_ | i've done the ports tree build of the jdk under freebsd. |
| 02:08.59 | chris_ | scary |
| 02:09.09 | [g2] | rwhitby-web, sorry about the distractions I was respond to the early post about 2.4.x being a non-starter |
| 02:09.35 | Tiersten | There are quite a few open source JVMs available |
| 02:09.37 | Tiersten | http://www.kaffe.org/links.shtml |
| 02:10.55 | rwhitby-web | Pity Unslung is not a WikiWord by default. I should have thought of that when naming it. |
| 02:11.09 | [g2] | rwhitby-web, I don't know how much help is available to get the 2.6.7+ ship-shape. Hardy-har-har |
| 02:11.23 | [g2] | kergoth may show up tomorrow with it mostly done |
| 02:11.30 | Tiersten | Damn. Can somebody kick him for that terrible pun? :) |
| 02:11.46 | [g2] | or it could take weeks for us to get into the oe groove and work it all out |
| 02:12.26 | [g2] | clearly some guys like jacques are already with the program. |
| 02:12.30 | rwhitby-web | I think we know just about what we need to do the OE stuff to get openslug-2.6.7 working |
| 02:12.50 | rwhitby-web | if we've got the patches as well, then we can do it |
| 02:13.30 | [g2] | I've looked a little at the root_fs for the nslu2 from oe and I kinda like the sysvinit style |
| 02:13.30 | rwhitby-web | [but I do want to keep the Unslung kernel very close to the Linksys kernel for stock applications compatibility reasons] |
| 02:13.50 | [g2] | I also like that dropbear is installed and brought up out of the box |
| 02:14.25 | [g2] | personally, with some mild tweaking, the box could be brought up with only serial enabled or only serial-usb |
| 02:14.44 | [g2] | That's the desired turn-up proceedure from my perspective |
| 02:15.28 | rwhitby-web | [g2]: [Responding to OpenNSLU] I think we should perhaps steer away from NSLU2 in distribution monikers, in case Linksys is not happy about it ? What do you think along those lines? |
| 02:15.30 | jacques | we can have a linux/linux-nslu2-2.6.7.oe no problem - we just need to force the unslung distro to use the 2.4 kernel (also not a problem) |
| 02:15.35 | [g2] | With some JTAG testing, it's possible to remove Redboot and boot directly to Linux |
| 02:16.02 | Tiersten | [g2]: Why would we want to do that? |
| 02:16.11 | rwhitby-web | jacques: I'm thinking "linux-nslu2" should be reserved to match Linksys kernels, and another name used for the OpenSlug kernels |
| 02:16.18 | dyoung | Jacques, are you using a standard nslu2_defconfig now? |
| 02:16.46 | [g2] | Tiersten, to secure the box. I'm planning on eventually putting mine on the Big-I |
| 02:16.48 | rwhitby-web | [remember that we want to have Unslung, SlugPlus and OpenSlug all come out of OE] |
| 02:17.28 | jacques | actually, looking at it again - we need linux/linux-nslu2-2.6_2.6.7 and linux/linux-nslu2-2.4_2.4.22-xfs |
| 02:17.28 | dyoung-scurvydog | Yes, but mrakes has the machineid of nslu2 set already; we ought to be able to come up with a nslu2_defconfig. |
| 02:17.31 | rwhitby-web | so we can't go changing the base linux-nslu2 stuff that kergoth has in there now, cause that needs to be kept for Unslung and SlugPlus |
| 02:17.38 | Tiersten | [g2]: I'm sure you could disable the network abilities of redboot or prevent it from listening |
| 02:17.40 | [g2] | rwhitby-away, right. OE already builds an 8MB flash image. |
| 02:17.49 | dyoung-scurvydog | I see. Okay. |
| 02:17.51 | Tiersten | If an outside attacker can reboot your NSLU2 then you've got bigger problems |
| 02:18.02 | jacques | dyoung-scurvydog, yes I am using the standard stuff now |
| 02:18.04 | rwhitby-web | jacques: what if Linksys release a 2.6 kernel in the next nslu2 firmware release - then we collide |
| 02:18.21 | [g2] | Tiersten, if anybody compromises a box on the local net it'd make an easy target |
| 02:18.48 | Tiersten | Yeah but you just need to stop it listening for those couple of seconds at the beginning |
| 02:18.49 | jacques | rwhitby-web, then we have linux/linux-nslu2-2.6_2.6.7-anyu2 ? |
| 02:18.50 | rwhitby-web | we need one name for Linksys-derived kernels, and another name for the open kernels. We can't use linux-nslu2 for both in my opinion. |
| 02:19.12 | rwhitby-web | jacques: I'm thinking openslug_2.6.7.oe |
| 02:19.12 | dyoung-scurvydog | Mmmm Honey.... |
| 02:19.14 | Tiersten | Once the kernel is loaded and running, redboot doesn't do anything |
| 02:19.16 | jacques | rwhitby-web, I'm just looking at the current policy which seems to be hardware name in the kernel |
| 02:19.30 | jacques | rwhitby-web, oops nevermind I just saw openzaurus |
| 02:19.36 | Tiersten | So to make it secure, we just need to stop it listening to port 9000 when it starts up |
| 02:19.38 | rwhitby-web | and opensimpad |
| 02:19.45 | jacques | rwhitby-web, so I agree then - openslug works nicely |
| 02:19.59 | Tiersten | Easier and safer than trying to remove redboot totally |
| 02:20.24 | [g2] | Tiersten: have you audited all the code running on your nslu2 ? Outsiders are running code on your box |
| 02:20.41 | rwhitby-web | ok, so the policy is that linux-nslu2_*.oe is reserved for replicating Linksys kernels, and openslug_*.oe is for complete replacements with no ties to Linksys kernels. Agreed? |
| 02:20.52 | Tiersten | Have you audited all the code running on your desktop PC? |
| 02:20.54 | jacques | so, are we agreed? we're adding linux/openslug-2.6_2.6.7.oe to oe ? |
| 02:21.08 | dyoung-scurvydog | Sounds like a plan. |
| 02:21.17 | [g2] | rwhity-web - on the openslug_*.oe nod. |
| 02:21.17 | rwhitby-web | jacques: do we need the first 2.6? |
| 02:21.35 | jacques | rwhitby-web, not really |
| 02:21.46 | jacques | rwhitby-web, not if we dont plan on having a 2.4 kernel in openslug |
| 02:21.59 | dyoung-scurvydog | By the time this initial build builds and I'm ready for 2.6.7, you guys should have sussed out the details. |
| 02:22.06 | dyoung-scurvydog | hardware store. BBIAB. |
| 02:22.14 | [g2] | lol |
| 02:22.25 | rwhitby-web | none of the others do, and I don't think any is proposing a 2.4 openslug kernel, and even if we wanted one, then we can still call it openslug_2.4.x.oe |
| 02:22.27 | [g2] | dyoung-scurvydog, goes to buy memory |
| 02:22.47 | dyoung-scurvydog | Aluminum rod actually. |
| 02:22.58 | jacques | rwhitby-web, lol I guess the only one that does it the first thing I worked on in oe - omap |
| 02:22.59 | dyoung-scurvydog | I got 768M in there now; that should be enough? |
| 02:23.28 | rwhitby-web | jacques: right - it's the odd one out now :-) |
| 02:24.00 | [g2] | dyoung-scurvydog, I returned that 256MB of swap space for you today |
| 02:24.02 | Tiersten | [g2]: I'm not seeing your point regarding totally removing RedBoot and making the kernel load directly. Obviously you'll have to secure the rest of the NSLU2 but disabling the small period where RedBoot listens for a telnet connection to port 9000 should be enough. Stock RedBoot doesn't have any backdoors and we should be able to compare the sercomm code against the stock code to see what they've done |
| 02:24.12 | jacques | rwhitby-web, and handhelds-pxa-2.6_ |
| 02:24.52 | rwhitby-web | jacques: can you create the openslug_2.6.7.oe file and openslug-2.6.7 directory, populate with what we have in the AnyU2 makefile, and bk send to both nslu-changesets@lists.sf.net (which you should subscribe to) and also to kergoth, referencing the policy we've decided? |
| 02:25.33 | jacques | rwhitby-web, OK but I assume no CSR for now? |
| 02:25.57 | [g2] | Tiersten, a code audit and reproduction is a very possible alternative. Costly, timely and of no real value long-run because the Redboot functionality is already crippled |
| 02:26.01 | rwhitby-web | jacques: right - looks like they put the 2.6 in there to save having to move the corresponding directory all the time. |
| 02:26.12 | rwhitby-web | jacques: correct, no CSR |
| 02:26.25 | jacques | not sure if you saw kergoth's comments on needing changes to oe to support the intel distrbution policy |
| 02:26.34 | [g2] | Tiersten, you reasoning is sounds and approach viable. economically I disagree |
| 02:26.44 | rwhitby-web | jacuqes: yep, saw and agree |
| 02:26.53 | jacques | ok cool |
| 02:26.55 | Tiersten | [g2]: One advantage of redboot that you'd lose is that reflashing the kernel is extremely painful via JTAG if you don't have a proper JTAG interface |
| 02:27.11 | Tiersten | It took about 20 minutes to reflash just 128KB |
| 02:27.39 | Tiersten | tking has tested it with a BDI2000 and that can do 100KB in about 1 second |
| 02:27.42 | rwhitby-web | jacques: perhaps we ping kergoth on whether it's openslug-2.6_2.6.7.oe or just openslug_2.6.7.oe - there may well be a good reason to use the former. |
| 02:28.05 | Tiersten | ah okay |
| 02:28.24 | [g2] | JFFS2/squashfs |
| 02:28.25 | Tiersten | Interesting idea though. Some people have implemented it in real products before. |
| 02:28.32 | rwhitby-lunch | bbiab 1 hour |
| 02:28.36 | Tiersten | Enjoy |
| 02:28.38 | [g2] | enjoy |
| 02:28.40 | chris_ | where is rwhitby located? |
| 02:28.46 | Tiersten | Australia |
| 02:28.49 | rwhitby-lunch | jacques: we're in sync? |
| 02:28.50 | chris_ | nice :) |
| 02:29.12 | jacques | rwhitby-web, the only reason I know is that by default oe uses the latest version of a package, so if there were an openslug-2.4.22 and openslug-2.6.7, it would pikc 2.6.7 - with the -2.4_ and -2.6_ you can specify the kernel as openslug-2.4 and oe will pick the latest 2.4 version |
| 02:30.05 | rwhitby-lunch | so if we don't forsee creating openslug-2.4, then we should be fine. We'd better check with kergoth or others on #oe first. I'll leave it up to you to get the consensus from them. |
| 02:30.06 | jacques | rwhitby-wI saw for now I do openslug-2.6.7.oe and we ask kergoth when we see him |
| 02:30.13 | Tiersten | [g2]: If you're looking to do that then I don't see any problems regarding dumping redboot totally as you'll be basically using Linux as a bootloader of sorts. It's just if you weren't then it'd be really annoying everytime you break it |
| 02:30.15 | [g2] | jacques, does oe produce a manifest of all sw installed ? |
| 02:30.25 | rwhitby-lunch | jacques: agreed. ciao |
| 02:30.49 | rwhitby-lunch | jacques - do you know the convention for underscores vs hyphens in the .oe file name? |
| 02:31.01 | rwhitby-lunch | shouldn't that be openslug_2.6.7.oe ? |
| 02:31.03 | [g2] | Tiersten, you're on the money again - agreed |
| 02:31.55 | *** join/#nslu2-linux rwhitby-treo (upirc@203.202.55.184) |
| 02:32.10 | jacques | rwhitby-lunch, looks like it separates the package name from version - so it should be openslug_2.6.7.oe (tho the dir is still openslug-2.6.7) |
| 02:32.26 | rwhitby-treo | agreed |
| 02:32.41 | jacques | [g2], I'm not sure about the manifest |
| 02:33.11 | [g2] | jacques, I was going to write a little perl script to generate that from the NOTE: info from the build |
| 02:33.43 | jacques | [g2], but almost all of your questions are what I would consider fine to ask in #oe |
| 02:33.46 | [g2] | somebody should put a gui front-end on oe and put the ptxdist stuff to shame |
| 02:34.28 | jacques | kergoth just didn't want 30 new nslu2 developers going to #oe and asking "how do I get started?" |
| 02:35.15 | [g2] | well he's probably gonna get about 50-100 in the next couple months |
| 02:35.49 | CodeWhacker | the perils of popularity |
| 02:35.50 | [g2] | god forbid buzbee mentions oe as the standard build system |
| 02:35.52 | rwhitby-treo | they should come here first, unless it's generic oe |
| 02:35.53 | *** join/#nslu2-linux McTails (~mctails@dialup-4.237.211.76.Dial1.NewYork1.Level3.net) |
| 02:36.37 | rwhitby-treo | he will in two or so articles time :-) |
| 02:36.57 | [g2] | rwhitby-treo, precisely that my point :) |
| 02:37.35 | kergoth`bbl | [g2]: qt and gtk2 frontend work has been done, but nobody bothered finishing them |
| 02:37.56 | [g2] | morning sunshine! |
| 02:38.06 | kergoth`bbl | hey |
| 02:38.48 | [g2] | IMHO oe and nslu2-linux and buzbee all have great synergy together |
| 02:39.51 | [g2] | buzbee does great articles, but it's really just a teaser to get started or a project for the weekend coding jockey |
| 02:40.46 | [g2] | nslu2-linux has a bunch of good momentum due to rwhitby's focus on process and tons of others helping out and driving that scalable vision |
| 02:40.50 | kergoth`bbl | glad to hear it. people starting to get more comfortable with oe? |
| 02:41.36 | [g2] | oe is a success all on it's own, but nslu2-linux and some press from buzbee just pours gasoline on the bondfire |
| 02:41.43 | jacques | I would say so kergoth |
| 02:42.01 | kergoth`bbl | [g2]: fyi, we've intentionally kept oe out of the limelight in the past |
| 02:42.10 | kergoth`bbl | no notifications, no announcements, no press of any kind |
| 02:42.22 | kergoth`bbl | due to the fact that its been in continual develpoment |
| 02:42.40 | kergoth`bbl | now is definately the time to increase visability, but we really need to get the OE core released 1.0 |
| 02:42.45 | [g2] | right. an old friend used to say timing is 90% of everything |
| 02:43.25 | kergoth`bbl | the oe core hitting 1.0 is pending docs. perhaps we can leverage the nslu2 community to help improve the oe documetnation |
| 02:43.56 | rwhitby-treo | good idea |
| 02:44.14 | [g2] | surely I think there's a great need for that. And the nslu2-linux community has lots of members just itching to help out |
| 02:44.45 | jacques | kergoth`bbl, you planning on pushing your nslu2 2.6.7 stuff soon? |
| 02:44.49 | rwhitby-treo | tking is a great example |
| 02:45.08 | [g2] | yes. tking is a roxor |
| 02:45.26 | kergoth`bbl | jacques: just needs someone to pester me periodically. at my parents house enjoying unproductivity at the moment |
| 02:45.56 | chris_ | i thought i was the only one in the world that used the term "unproductivity" |
| 02:46.12 | rwhitby-treo | can you check the policy we defined for oe nslu2 above? |
| 02:46.15 | kergoth`bbl | ~unproductivity |
| 02:46.16 | jbot | yay! |
| 02:46.18 | kergoth`bbl | :) |
| 02:46.22 | jacques | I was just about to start integrating the 2.6.7 from anyu2, but I will have to add an additional patch for the -load-shot-bytes thing since oe uses 3.4.2 for nslu2 iirc |
| 02:46.36 | jacques | --load-short-bytes |
| 02:47.01 | jacques | and I know from experience anyu2 does not build the kernel with 3.4.2 |
| 02:47.19 | rwhitby-treo | kergoth: just before you came in |
| 02:47.44 | kergoth`bbl | jacques: yep. the patch for that is in rmk's patch tracker |
| 02:47.45 | rwhitby-treo | we set some rules - can you check/confirm them? |
| 02:47.59 | kergoth`bbl | jacques: feel free to get it added yourself, i doubt i'll have time in the near future |
| 02:48.07 | kergoth`bbl | rwhitby-treo: certainly |
| 02:48.52 | jacques | rwhitby-treo, is there a target in the anyu2 makefile to just build the kernel? I don't see one |
| 02:49.49 | rwhitby-treo | away from desk |