| 09:37.23 | *** join/#arm-netbook fjp` (~fjp@ip4da6b463.direct-adsl.nl) |
| 10:01.27 | *** join/#arm-netbook lkcl (~lkcl@nat66.mia.three.co.uk) |
| 11:40.25 | fjp | starting to push kernel repo to alioth |
| 11:40.44 | fjp | will take a while... |
| 12:15.07 | mnemoc | :D |
| 12:15.22 | mnemoc | what's the name of the project "arm-netbook" ? |
| 12:16.50 | mnemoc | wonders why alioth doesn't use a cert from cacert |
| 12:17.29 | mnemoc | ca.debian.org |
| 12:18.45 | mnemoc | it's sad that not even debian is using cacert.org anymore |
| 12:19.40 | mnemoc | good, now chrome is happy connecting to alioth |
| 12:20.40 | mnemoc | https://alioth.debian.org/projects/arm-netbook/ <---- topic? :) |
| 12:24.38 | fjp | Sure. |
| 12:40.25 | fjp | Ah, they's upped my dandwidth again :-) |
| 12:40.44 | fjp | bandwidth that is |
| 12:44.30 | mnemoc | :) |
| 12:48.03 | fjp | mnemoc: Added |
| 12:48.06 | fjp | lkcl: Added |
| 12:48.13 | lkcl | aaa |
| 12:48.35 | fjp | lkcl: I've bombarded you too admin |
| 12:48.36 | lkcl | s/aaa/taaa |
| 12:48.42 | lkcl | lucky mee |
| 12:48.55 | mnemoc | thanks |
| 12:49.07 | lkcl | git? |
| 12:49.23 | fjp | linus is almost there |
| 12:49.33 | mnemoc | no wiki? |
| 12:49.35 | fjp | After that I need to push a few more branches |
| 12:49.42 | fjp | We can create one. |
| 12:49.47 | fjp | Using ikiwiki |
| 12:49.51 | lkcl | ahh... what's with the stupid account names? guest_lkcl or something stupid |
| 12:50.05 | mnemoc | fjp: i thought this forge had a built-in one |
| 12:50.07 | lkcl | i'm not putting up with that. how do i get that changed? |
| 12:50.11 | mnemoc | foo-guest |
| 12:50.18 | fjp | lkcl: By becomming a DD... |
| 12:50.24 | lkcl | oh fer xxxx's sake |
| 12:50.26 | lkcl | ok |
| 12:50.37 | fjp | the normal namespace is reserved for DDs. |
| 12:50.54 | mnemoc | the "elite" :) |
| 12:51.05 | fjp | If they did not do that then they'd have to check alioth before assigning new user IDs to DDs |
| 12:51.24 | fjp | This is the only way that can be avoided. |
| 12:51.34 | mnemoc | have you used ikiwiki with git/asciidoc ? |
| 12:51.51 | fjp | My own blog is in ikiwiki on alioth |
| 12:51.55 | fjp | (on got) |
| 12:51.58 | fjp | git |
| 12:52.05 | mnemoc | how does it go? |
| 12:52.22 | mnemoc | does it worth to leave dokuwiki for it? |
| 12:52.31 | fjp | It's fairly simple. There's a howto somewhere. |
| 12:52.40 | fjp | I don't know dokuwiki |
| 12:53.00 | fjp | http://git.debian.org/?p=arm-netbook/arm-netbook.git;a=summary |
| 12:53.00 | mnemoc | it's the most "famous" db-less wikis |
| 12:53.04 | fjp | Linus is up |
| 12:57.48 | fjp | I quite like using git as the "database" |
| 12:59.52 | mnemoc | any sample toy? |
| 13:00.03 | mnemoc | i mean, using libgit |
| 13:00.24 | mnemoc | or you mean for versioning the wiki? |
| 13:00.58 | mnemoc | i wanted to use git as *db* engine instead of tokyocabinet |
| 13:01.04 | mnemoc | in an app |
| 13:01.10 | mnemoc | but haven't had time to research |
| 13:03.02 | mnemoc | the snapshot-ness and consistency control is very attractive to be used as internat storage of an embedded controller |
| 13:05.12 | fjp | git is used to manage the complete source of the wiki |
| 13:12.01 | fjp | http://ikiwiki.info/rcs/git/ |
| 13:12.19 | fjp | You can subversion and others too if you prefer. |
| 13:12.47 | lkcl | mnemoc: do look at ikiwiki. the web interface version does git commits, so when you modify a page just like you would any other wiki, it goes into the git repository. |
| 13:13.06 | lkcl | then you can also just do a git clone and a git push if you prefer, bypassing the web editing interface |
| 13:13.09 | fjp | http://wiki.debian.org/Alioth/ikiwiki |
| 13:13.13 | lkcl | because... well... you're using git! |
| 13:14.49 | mnemoc | lkcl: does it `pull` under the hood when you push bypassing the web interface? |
| 13:15.14 | lkcl | if it's been configured correctly, with a git commit hook, then yes. |
| 13:15.21 | lkcl | sorry - git push hook |
| 13:15.29 | mnemoc | lkcl: when you said "database" instead of "backend" my brain switched from ikiwiki to something like http://code.google.com/p/gimd/ |
| 13:32.53 | mnemoc | s/lkcl/fjp/ |
| 14:09.54 | *** join/#arm-netbook SQlvpapir (~teis@188.177.95.62) |
| 14:24.46 | fjp | OK. Git repo is complete now. |
| 14:25.03 | fjp | The current branch to build from is ct-pc89e/2.4.27/from-smartq |
| 14:25.20 | fjp | I'll mail the list with explanations of the rest. |
| 14:25.42 | fjp | Please do not push to existing branches for now.. |
| 14:26.23 | fjp | lkcl: Can you check the contents of ct-pc89e/2.4.27/from-smartq please? |
| 14:26.30 | lkcl | fjp: ack |
| 14:26.36 | lkcl | ber... |
| 14:26.40 | fjp | It is based on the repo from mer. |
| 14:26.50 | lkcl | run the git commands by me again, i never get them right :) |
| 14:27.19 | fjp | To get the whole thing: git clone git+ssh://git.debian.org/git/arm-netbook/arm-netbook.git |
| 14:27.31 | fjp | Use gitk to view. |
| 14:27.47 | lkcl | yes i have that |
| 14:27.58 | fjp | Best for an overview is: gitk --all 2.4.27.7.. |
| 14:28.08 | lkcl | i hate gitk! |
| 14:28.20 | fjp | Use something else then. |
| 14:28.25 | lkcl | :) |
| 14:29.06 | lkcl | ok i'll manage - i get there in the end. somehow git branch-related commands don't sink in to my fingers |
| 14:29.54 | lkcl | wow, you done quite a few extra commits in the past hour or so methinks |
| 14:30.32 | fjp | No, I've cleaned up things a bit by rebasing :-) |
| 14:30.55 | fjp | And that needed to be done before pushing the current state. |
| 14:31.04 | lkcl | ohh |
| 14:31.04 | lkcl | ah. |
| 14:32.03 | fjp | lkcl: If you want to make changes I suggest to create a new branch using: git checkout -b ct-pc89e/2.4.27/lkcl ct-pc89e/2.4.27/from-smartq |
| 14:32.15 | lkcl | ack |
| 14:32.41 | lkcl | btw, for a MACHINE_NAME we need to submit MACHINE_CT_PC89E_S3C6410 |
| 14:33.15 | lkcl | because there is also an SO-DIMM from seatron containing an SIRF Prism IV |
| 14:33.51 | fjp | Best to wait a bit for ChiTech to respond. |
| 14:33.52 | lkcl | and they've taken on-board the suggestion of doing a samsung Arm Cortex A8 SO-DIMM, it looks like they'll pick the S5PV210 (1ghz) |
| 14:33.59 | lkcl | yes. |
| 14:34.04 | lkcl | true |
| 14:34.28 | fjp | Because they'll also need to request new machine IDs when they make changes in the design... |
| 14:34.46 | lkcl | ah? oh. |
| 14:35.43 | fjp | Sure: whenever you need to change kernel support that cannot be autodetected, but must be #defined or #ifed you need a new machine ID |
| 14:36.09 | fjp | In that respect x86 is vastly superior to arm me thinks. |
| 14:36.20 | lkcl | rueful |
| 14:36.39 | lkcl | ok. i let the git pull do its stuff. 250k/sec 50% way through, 84mb ... |
| 18:53.26 | lkcl | fjp: diff -ru --> http://lkcl.net/arm_systems/CT-PC89E/lkcl.fjp.diff |
| 18:55.10 | lkcl | also i found yet _another_ mer-smartq patch which changes the CPU speed increments, i believe it makes it possible for the CPU to run at 600mhz. not entirely sure what's going on: the increments stop at 533mhz at the moment, does that mean that the cpu can still run at 667? |
| 19:10.28 | fjp | The CPU freq is lower than with original kernel ATM |
| 19:13.23 | fjp | I don't think there's anything relevant in that diff. |
| 19:14.39 | fjp | Maybe the sounc stuff, but we haven't really gotten to that yet. |
| 19:15.04 | mnemoc | 00:00:38 < mnemoc> lkcl: do you have the "fantastic thoughtful analysis and business case" somewhere i could read it? :) |
| 19:15.10 | lkcl | ohh yeh. |
| 19:15.12 | mnemoc | :) |
| 19:15.14 | lkcl | what's your email address? |
| 19:15.31 | fjp | I'd be interested too |
| 19:15.43 | lkcl | fjp: no, there's nothing relevant, but.. it's.. bitty. CONFIG_LCD_7 is that still in there or did i miss that? |
| 19:15.52 | mnemoc | lkcl: amery geeks cl |
| 19:16.07 | lkcl | ack |
| 19:16.20 | fjp | I reverted that whole LCD_[47] stuff. |
| 19:17.14 | fjp | AFAICT it's specific to what HHTECH did and a huge ugly hack. |
| 19:17.38 | fjp | My impression was that it was better to work from cleaner code. |
| 19:17.54 | fjp | See one of my revert patches. |
| 19:36.53 | fjp | thinks his D-I initrd was still too large |
| 19:37.17 | fjp | Now building with a <1MB one, which should be OK (I think) |
| 19:39.55 | lkcl | fjp: yep, that's a fact, on the "ugly-hackness" :) |
| 19:40.51 | fjp | I also reverted the USB suspend stuff. Mostly only a distraction and it will have been implemented better in mainline. |
| 19:43.49 | fjp | Most important IMO is that we now have a common base. |
| 19:46.23 | fjp | lkcl: Can you send me that analysis too? |
| 20:08.55 | mnemoc | fjp: hey! i asked for it first :) |
| 20:09.31 | fjp | points mnemoc to the word "too" in his request |
| 20:14.15 | mnemoc | it was a subtile way to say i haven't received it :) |
| 20:36.42 | fjp | Yikes! I now have it running a whole load of init scripts! |
| 20:37.11 | fjp | And all failing because some stuff isn't there, but that's fine. |
| 20:37.48 | fjp | basically it keeps respawning |
| 20:37.54 | fjp | Progress! |
| 20:38.38 | fjp | But why don't I get a shell? |
| 20:38.57 | fjp | Why is udev not started (AFAICT). |
| 20:41.35 | mnemoc | missing dev/tty<n> ? |
| 20:41.41 | fjp | Maybe |
| 20:41.49 | fjp | How do I create those? |
| 20:42.02 | mnemoc | normally, with udev |
| 20:42.11 | fjp | And with makedev? |
| 20:42.28 | mnemoc | that should work :) |
| 20:42.38 | fjp | Think I've got it. |
| 20:44.10 | mnemoc | adding ls -al /dev/ in your init scripts can be "enlightening" |
| 20:54.25 | fjp | Except when ls does not produce any output... |
| 20:55.18 | fjp | wonders if there's a problem running the build on an NFS share... |
| 21:00.14 | fjp | Because I'm not allowed to create character devices there... |
| 21:00.31 | fjp | And thus end up with a non-working /dev/console and /dev/null... |
| 21:02.07 | fjp | Add to that that D-I does not like the init=/init that ChiTech passes (because it causes a loop). |
| 21:02.22 | fjp | And I think I may just have the root of the problems. |
| 21:08.00 | mnemoc | fjp: tried devtmpfs? |
| 21:08.44 | mnemoc | uhm... are you are still fighting .24 or switched to .33? |
| 21:08.50 | fjp | 24 |
| 21:08.59 | mnemoc | ok, no devtmpfs |
| 21:09.48 | mnemoc | a common trick to the NFS problem is to make a devices.tar.bz2, and extract that early on init |
| 21:11.30 | fjp | Woot! A shell! |
| 21:11.41 | fjp | But no keyboard. |
| 21:13.33 | fjp | I now get debug output from the early boot. |
| 21:14.16 | fjp | So the D-I builds have gotten broken somehow wrt creation of default devices in dev/ |
| 21:15.55 | fjp | lkcl: ^^^^^^^^^ |
| 21:16.29 | fjp | I have a D-I debug shell, but keyboard is not working. Any ideas why that could be? |
| 21:16.36 | fjp | mnemoc: You too. |
| 21:17.46 | mnemoc | lsmod ? |
| 21:18.00 | fjp | No, this is WAY before modules |
| 21:18.14 | mnemoc | and how do you expect to have keyboard? ;-) |
| 21:18.32 | fjp | USB and HID compiled in. |
| 21:18.58 | fjp | Let's see if I actually have that. |
| 21:20.07 | fjp | Nope. Was modular. |
| 21:20.09 | fjp | Oops |
| 21:20.13 | mnemoc | :) |
| 21:20.32 | mnemoc | add the modprobe calls to your init script |
| 21:20.43 | mnemoc | easier than recompiling |
| 21:20.52 | fjp | I have nowhere to read the modules from. |
| 21:21.07 | fjp | recompiling is way easier. |
| 21:21.10 | mnemoc | not in the initramfs? |
| 21:21.13 | fjp | 10 seconds or so. |
| 21:21.14 | fjp | No |
| 21:21.17 | mnemoc | ok |
| 21:21.30 | mnemoc | built-in is safer |
| 21:22.19 | fjp | Yikes! It works. |
| 21:22.26 | fjp | And udev has run correctly too! |
| 21:23.09 | fjp | We're in business now. |
| 21:24.06 | lkcl | wha-heey! |
| 21:25.18 | mnemoc | \o/ |
| 21:25.28 | fjp | $ fakeroot mknod xqxq c 5 1 |
| 21:25.37 | fjp | -rw-r--r-- 1 fjp fjp 0 Mar 20 22:25 xqxq |
| 21:26.03 | fjp | sudo mknod xqxq c 5 1 |
| 21:26.09 | fjp | crw-r--r-- 1 root root 5, 1 Mar 20 22:25 xqxq |
| 21:26.17 | mnemoc | uh |
| 21:26.52 | fjp | Normally udev will fix that. |
| 21:27.33 | fjp | But my real problem was not udev, but the fact that init=/init was causing D-I's init script to loop. |
| 21:27.53 | fjp | So, let's try something more challenging... |
| 21:49.13 | fjp | http://people.debian.org/~fjp/tmp/CT-PC89E/vlcsnap-2010-03-20-22h45m10s164.png |
| 21:58.16 | fjp | http://people.debian.org/~fjp/tmp/CT-PC89E/vlcsnap-2010-03-20-22h57m12s218.png |
| 22:00.36 | lkcl | wha-heeeey! |
| 22:00.59 | lkcl | brilliant stuff. |
| 22:08.11 | mnemoc | kudos fjp! |
| 23:02.49 | fjp | BTW, note that those images show D-I running with framebuffer. |
| 23:32.44 | SQlvpapir | smiles at the thought of freedom |
