| 00:27:42 | jacques | anyone awake? |
| 00:30:20 | Keiser_Soze | i am |
| 00:33:33 | jacques | Keiser_Soze: hi, are you running linux on your tuxscreen? |
| 00:33:42 | jacques | Keiser_Soze: I am having a problem with DNS |
| 00:37:16 | jacques | hi scanline |
| 00:49:01 | sammy_ | hello |
| 00:55:25 | jacques | anyone here know if DNS is broken in 0.4 ? |
| 00:56:13 | sammy_ | morning scanline :) |
| 00:56:22 | scanline | morning sammy_ :) |
| 00:56:54 | jacques | :-( |
| 00:57:05 | scanline | jacques: The 0.4 buildroot images? |
| 00:57:23 | jacques | scanline: yes |
| 00:57:44 | scanline | I don't think it is... |
| 00:57:46 | jacques | scanline: I get this: |
| 00:57:50 | jacques | # nslookup www.apple.com |
| 00:57:50 | jacques | Server: default |
| 00:57:50 | jacques | Address: default |
| 00:57:50 | jacques | nslookup: www.apple.com: Unknown error |
| 00:58:08 | scanline | check your /etc/resolv.conf for syntax errors? |
| 00:58:26 | jacques | scanline: I copied it from another (working) linux box |
| 00:58:40 | jacques | scanline: and have checkied it many times |
| 00:58:51 | scanline | hmm... |
| 00:58:54 | scanline | strange |
| 01:00:39 | jacques | ah crap |
| 01:00:43 | jacques | dammit |
| 01:00:47 | jacques | i found the problem |
| 01:01:12 | scanline | ethernet cable not plugged in? ;) |
| 01:01:14 | jacques | busybox does not like my dns cache |
| 01:01:18 | jacques | dammit |
| 01:01:46 | jacques | i changed to another dns server and it works |
| 01:01:52 | jacques | should have tried that a long time ago |
| 01:02:17 | jacques | maybe there's an option to pdnsd which will make it work |
| 01:08:01 | scanline | morning Sammy |
| 01:12:22 | Sammy | hello scanline :D |
| 01:31:02 | scanline | learns Xlib programming |
| 01:36:17 | Russ | morning sammy |
| 01:43:28 | Sammy | morning Russ |
| 01:58:27 | EGg0 | is back (gone 02:12:43) |
| 02:05:24 | EGg0 | is away: Second star on the right, straight on 'till morning. |
| 03:49:17 | jacques | it sure is quiet |
| 03:52:56 | scanline | I'm reading some X11 tutorials |
| 03:53:36 | scanline | Frighteningly over-engineered... |
| 03:53:49 | GPSFan | I'm trying to figure out which files I need to install debian. |
| 03:54:34 | scanline | Just download the iso image, burn it to a CD, and boot from it |
| 03:55:03 | scanline | You can use the potato .iso to install newer versions of debian, you can have the installer download packages from the internet |
| 03:56:56 | jacques | GPSFan: install debian on what? |
| 03:57:06 | GPSFan | scanline: that makes sense, I was looking for something like that but wasn't sure if potatoe was the right one. btw which iso |
| 03:57:22 | GPSFan | jacques: on my PC so I can xcompile tux stuff. |
| 03:57:46 | jacques | GPSFan: ah, then scanline sounds right to me - I have only installed debian on my ipaq |
| 03:58:32 | | blob is the Boot Loader OBject |
| 03:58:32 | jacques | ibot: blob |
| 04:01:24 | | no idea, jacques |
| 04:01:24 | jacques | ibot: I am less interested in what it stands for than in where to get it and read about it |
| 04:02:18 | jacques | flashes a kernel and ramdisk image he just built with buildroot. |
| 04:03:01 | scanline | GPSFan: ftp://download.sf.net/pub/mirrors/debian-cd/2.2_rev4/i386/binary-i386-1.iso |
| 04:03:30 | GPSFan | scanline: so it is iso #1 that does the main install. |
| 04:05:12 | scanline | GPSFan: Yes. It will partition the disk and install a base system. the base system then boots, and it will install packages from CD, http, or other sources |
| 04:05:36 | GPSFan | scanline: the sf archive is real slow, I'm getting it from planetmirror |
| 04:06:15 | scanline | GPSFan: ok. ibiblio is nice and fast, but i couldn't find the iso on there |
| 04:06:34 | jacques | waits forever for the ramdisk to flash. |
| 04:13:47 | jacques | anyone know why I get this when booting? Waiting for chip to read, status = 4 |
| 04:14:12 | scanline | jacques: That's the flash memory erasing. Perfectly normal, just a bit noisy |
| 04:15:20 | jacques | scanline: thanks! I was a bit worried |
| 04:15:32 | jacques | scanline: i did notice that it only seems to happen on first boot after a flash |
| 04:15:57 | scanline | yes. Sometimes it happens when saving large files to the flash |
| 04:17:28 | jacques | scanline: makes sense |
| 04:17:46 | GPSFan | scanline: is potato the newest, there are also iso's for woody & sid? |
| 04:18:08 | scanline | GPSFan: No, potato's pretty old. I just didn't know there were isos for the newer ones :) |
| 04:18:25 | scanline | GPSFan: Grab the sid iso, it should work nicely |
| 04:18:28 | GPSFan | scanline: so which would be "best" ;>) |
| 04:19:06 | scanline | I think the versions went potato -> woody -> sid -> umm... not sure what's after sid |
| 04:19:24 | scanline | Debian releases are all named after characters from the movie Toy Story :) |
| 04:19:53 | scanline | Before potato there was hamm |
| 04:20:01 | GPSFan | ok, so sid is unstable, woody testing, and potato stable? |
| 04:20:33 | scanline | I think... |
| 04:20:43 | scanline | You could always ask on #debian :) |
| 04:22:34 | GPSFan | scanline: you suggested using the unstable release yesterday, so i was looking around for that. I'm downloading both now, but 1200MB will take a looong time. ;>) |
| 04:23:20 | scanline | I think sid is the unstable release, but maybe it's testing... |
| 04:23:29 | scanline | Just don't use the stable release and you'll be fine :) |
| 04:24:00 | GPSFan | ok, so I'll be unstable, I can do that. ;>) |
| 04:29:54 | jacques | cool - I now have a udrive and network card working in my tux |
| 04:31:22 | sjm | jacques: which card have you got? |
| 04:32:02 | jacques | sjm: netgear FA411 - nice card 10/100 fits in either slot |
| 04:32:33 | sjm | jacques: okay. thanks.... I'm still waiting for my tux to arrive :( |
| 04:33:15 | jacques | sjm: I got mine on Friday |
| 04:33:45 | sjm | I expect mine about mid December :) |
| 04:33:51 | scanline | Hmm... once you get through the excruciatingly large number of options, this Xlib stuff is rather fun |
| 04:35:25 | jacques | sjm: where are you? far away? |
| 04:35:50 | sjm | jacques: yeah.... Melbourne, Australia |
| 04:40:59 | jacques | cool - just chrooted to intimate on udrive, started sshd, and ssh'd into the tux |
| 04:42:07 | jacques | wonder what will happen if I try to start X |
| 04:43:27 | jacques | oh, I dont have X on this image |
| 04:44:28 | collord | jacques: does the FA411 work Just Fine with the standard kernel, etc? |
| 04:46:29 | jacques | collord: yes, it worked straight out of the box for me - uses the NE2000 (pcnet) driver |
| 04:46:57 | jacques | collord: only downside is they arent the cheapest cards out there |
| 04:48:21 | collord | true, but they're available and they're 100Mb too. I heard there was some esoteric problem or they weren't favored for some reason... |
| 04:50:12 | sjm | collord: According to http://tuxscreen.nnet/wiki/view/Main/PCMCIA they are recommended |
| 04:50:23 | sjm | only one n in net :) |
| 04:53:17 | collord | hrm well my memory must be failing me... |
| 04:53:43 | jacques | collord: I think it's a different card that's having the esoteric problem |
| 04:54:01 | scanline | I've got a FA411 in my Tux and it works great |
| 04:54:39 | jacques | The Linksys PCM100 (not to be confused with PCMPC100, PCM200, PCM100H1 , etc) is 3.3v pcmcia and is ne2k compatible. There is some question over whether some versions of this card will work properly with linux, see the sourceforge link below for details. |
| 04:54:39 | jacques | * http://www.linksys.com/products/product.asp?prid=100&grid=11 |
| 04:54:40 | jacques | * http://sourceforge.net/forum/forum.php?thread_id=111766&forum_id=33427 |
| 04:55:00 | jacques | collord: maybe you are thinking of the Linksys |
| 04:59:24 | collord | perhaps, thanks. related issue, anyone get the DLink DWL650 working? |
| 05:05:19 | jacques | that a wireless card? |
| 05:05:27 | jacques | uses prism2 driver |
| 05:07:01 | collord | yup, issue is that it is supposedly cardbus... but isn't really |
| 05:07:45 | collord | so somebody was working on seeing if it was some voltage setup, etc. issue to work with pcmcia |
| 05:21:44 | jacques | a |
| 05:21:46 | jacques | ah |
| 05:28:04 | dirkx | I found both the PMC100 and FA411 to work fine. Wireless I am using the linksys wpc11 as I could not get the DWL650 from Dlink to work perfectly - crypt failed to work and all sort of interupt errors. |
| 05:46:35 | jacques | hi dirkx, do you know if anyone has tried the netgear MA401? |
| 05:58:28 | jacques | heh, i dismantled a tux and reassembled it and it still works! |
| 06:04:50 | thekernel | andersee: hey, I got my DSL up again. Next weekend I should be getting around to writing my GPS app for the tuxscreen |
| 06:05:22 | thekernel | Do you have that uClibc pthreads port source tree? |
| 06:10:12 | andersee | I do |
| 06:10:19 | andersee | I'll have to dig for it. |
| 06:10:20 | thekernel | ok |
| 06:10:32 | thekernel | Would you email it to me? |
| 06:10:36 | thekernel | when you find it of course |
| 06:10:39 | andersee | lemme go look |
| 06:10:47 | andersee | where should I send it? |
| 06:12:53 | BZFlag | andersee: email. |
| 06:13:06 | thekernel | Anyway, I'm heading to bed now. I gotta be consious for work in the morning. |
| 06:13:33 | BZFlag | the 166.* is what ns1 should be right? so just your sone file needs edits. |
| 06:13:34 | thekernel | I don't want to get up and try and pour a glass of cerial like I did this morning. |
| 06:14:07 | BZFlag | heh, I find I pour the milk in the bowl first when I'm tired. ;-) |
| 06:14:30 | thekernel | My brother stopped me and asked me what I was doing. |
| 06:14:58 | BZFlag | heh. ;-) |
| 06:15:02 | thekernel | Anyway, as I was saying... zzzzz.... |
| 06:17:29 | andersee | BZFlag: the 209 one is the server of a buddy of mine |
| 06:18:26 | andersee | I've been using his box as primary for a couple months (since departing Lineo) |
| 06:24:36 | collord | what is nick for PaulS ? |
| 06:26:58 | jacques | hi andersee - are you the maintainer of busybox? |
| 06:27:56 | andersee | jacques: yup |
| 06:28:09 | andersee | jacques: guilty as charged |
| 06:29:21 | jacques | andersee: heh, I like it a lot but have a little problem - busybox name resolution does not get along with pdnsd DNS cache |
| 06:29:51 | jacques | andersee: as this is on my tuxscreen - I haven't had a chance to try to debug the issue |
| 06:30:00 | andersee | I don't even know what pdnsd DNS cache is I'm afraid, so I'm not surprised |
| 06:30:11 | jacques | andersee: not surprised why? |
| 06:30:46 | andersee | since I don't know what pdnsd is doing.... |
| 06:30:51 | jacques | BTW, pdnsd is at http://home.t-online.de/home/Moestl/ |
| 06:30:58 | dirkx | andersee: pthreads - would not mind a copy either :-) |
| 06:31:05 | andersee | looks |
| 06:31:26 | andersee | dirkx: pthreads for uClibc isn't exactly ready.... |
| 06:31:31 | jacques | andersee: yeah. I know, but pdnsd wirks with everything else I've thrown at is (at least with version 1.1.6) |
| 06:32:16 | andersee | dirkx: I have a lame 20 minute hack that works in a limited sort of way for a very limited definition of works |
| 06:32:30 | jacques | andersee: it's possible that there could be an option to pdnsd which would make it work - one of the stricter compliance options, but I'm not sure what to try |
| 06:33:07 | andersee | jacques: which bb applets in particular fail and what are the symptoms? |
| 06:35:10 | dirkx | andersee: that is fine - I'd just be curious to see how much is there. Trying to gauge my planning for the next weeks - and how far linux@tux needs to be pushed to get a VoIP demo with some apache stuff thrown in - or if another path would get me there quicker. Such as just using pgui with a java remote display to the tuxscreen and some clever UDP to the speaker/microphone. |
| 06:35:14 | dirkx | I.e. doing it all remote. |
| 06:36:53 | jacques | andersee: ping and nslookup fail (havent tried others) |
| 06:36:59 | jacques | andersee: looks like this: |
| 06:37:01 | jacques | <jacques> # nslookup www.apple.com |
| 06:37:02 | jacques | <jacques> Server: default |
| 06:37:02 | jacques | <jacques> Address: default |
| 06:37:02 | jacques | <jacques> nslookup: www.apple.com: Unknown error |
| 06:37:15 | andersee | dirkx: my lame 30 min implementation is already in CVS. |
| 06:37:28 | andersee | dirkx: http://cvs.uclinux.org/cgi-bin/cvsweb/uClibc/libpthread/pthread.c?rev=1.1&content-type=text/x-cvsweb-markup |
| 06:38:08 | dirkx | andersee: - ah - so the tuxroot stuff is not a HEAD of CVS - no trouble - I can cvs update to -head. THanks for the hint ! |
| 06:38:42 | andersee | dirkx: still needs clone to be implemented properly on ARM too I believe. I've only tried it on x86 so far |
| 06:38:44 | scanline | dirkx: PicoGUI doesn't have good java support yet |
| 06:39:07 | andersee | jacques: what does your /etc/resolv.conf look like? |
| 06:39:37 | dirkx | andersee: I am an old arm/riscos/acorn hacker. No troubel there :-) 6502 and ARM RISC where the first languages I learned. Stil have an Acorn RiscPC here :-) |
| 06:39:48 | jacques | andersee: |
| 06:39:51 | jacques | search activebuddy.com |
| 06:39:51 | jacques | nameserver 192.168.1.2 |
| 06:40:13 | dirkx | scanline: that is fine - if you use the java2sdk source code it is very easy to AWT the output - they use only 5 real important primitives. |
| 06:40:18 | jacques | andersee: changing the 192.168.1.2 to the address of a non-pdnsd dns server causes it to work |
| 06:42:09 | dirkx | andersee: the only thing which really sucks with all this java stuff is that using native calls from java is so expensive. Often you are better of to do java->network->local-loopback->c-code->displaycode. |
| 06:42:54 | andersee | jacques: interesting. |
| 06:43:17 | scanline | dirkx: Well, if you use AWT as your widget set and just draw via picogui, you lose pretty much all the benefits of picogui's architecture |
| 06:43:17 | andersee | jacques: I suspect that you are seeing a uClibc problem, not a busybox one. |
| 06:44:06 | dirkx | jacques: big differens between the two lookup is the number of replies - i.e. multiple records. (Though I would not expect apple to go over the udp->tcp limit). |
| 06:44:40 | dirkx | scanline; yes ideally you'd want to do it in AWT and map the widgets for things like menu's across. |
| 06:44:46 | andersee | jacques: The uClibc resolver is not the most sophisticated in the world. Perhaps it is getting confused. |
| 06:44:50 | dirkx | scanline: you are quite right about that. |
| 06:45:24 | scanline | dirkx: You should probably take a look at http://www.minirl.com/pgui-libgcj.html |
| 06:45:34 | dirkx | scanline: and you do not want to make it as slow as NanoX :-) |
| 06:45:55 | jacques | andersee: dirkx: thanks, I will play with the pdnsd options then - I remember seeing something about multiple records in there |
| 06:45:56 | andersee | jacques: The resolver was recently updated with ipv6 support. How recent is you uClibc snapshot? |
| 06:46:08 | Russ | andersee: any info on the mount -a thing? |
| 06:46:34 | scanline | dirkx: Well, the architecture of picogui is very different from most GUIs. In PicoGUI, it's possible to do simple line-drawing and such from clients, but the focus is on separation of content and presentation, letting the server do as much as possible |
| 06:46:42 | andersee | Russ: I keep having good intentions about looking at it. But my tuxscreen phone has been unplugged for about 2 weeks now. |
| 06:46:42 | Russ | did you see the 'second "mount" hangs' thread? |
| 06:46:50 | jacques | andersee: d'oh I built a buildroot image just a coupld of hours ago and flashed it but I don't think I tested for the bug after that - I was seeing this with v0.4 |
| 06:46:50 | dirkx | scanline: yes - I looked at that - compiled it and got it to work. But the license might bite me in the long run. Without starting a holy war - i'd be more confortable by a more 'free' one. This one rules out anyone or any entity who has ever signed or entered into a nda or other agreement with SUN. I am using solaris and sun JDK's a lot. |
| 06:47:17 | andersee | Russ: no, missed that. where was it? |
| 06:47:54 | scanline | dirkx: Also, if native C calls are really that slow, you could just talk to pgserver directly without using the C client lib. The network protocol is pretty simple. |
| 06:48:23 | andersee | Russ: On the plus side, I'm trying to get busybox 0.60.2 released tonight... |
| 06:48:28 | Russ | linux-mtd |
| 06:48:41 | andersee | Russ: So if we can find and squash something, now would be a good time. |
| 06:48:52 | dirkx | scanline: yes - I've whipped up some perl (sorry) scripts this morning to try exactly that. And it seems really simple enough. But as you said - it makes more sense to send higher level than primitives. Plus if I also need to move mic/speaker UDP traffic then I want to keep latancy down. |
| 06:49:12 | andersee | Ahh. Not subscribed at the moment. I'll go check the archives. |
| 06:49:23 | scanline | dirkx: What's wrong with perl? ;) |
| 06:49:36 | andersee | Russ: BTW, congrats on getting the shannon patches in to the latest from rmk. :) |
| 06:50:05 | dirkx | scanline: nothing - except for the fact that a lot of the VoIP 'marshall'ing code and call mgnt -> voicemailbox mngt is in java ;-) the UI side is. |
| 06:50:52 | Russ | should I start a 2.4.15-pre6-rmk1 release? or wait something stabilizes |
| 06:51:01 | Russ | he never said much about the alan/linus switch |
| 06:51:15 | Russ | ew, marshalling |
| 06:52:15 | dirkx | russ: given the overloading they do in the voip world - I am loath to add 'ing' to it :-) |
| 06:53:26 | Russ | java, marshalling, bad |
| 06:53:27 | Russ | er, |
| 06:53:28 | Russ | bah |
| 06:54:06 | scanline | curses the X consortium for oddly naming GXor and GXxor |
| 06:54:17 | andersee | Russ: rmk did mention that since there would be no more -ac patches he was switching back to the Linus kernels. |
| 06:54:25 | dirkx | Russ: actually in the VoIP world it means something totally different - and has more to do with the responsibility for the call to be transfered and/or gatewayed. I.e. just closer to the original meaning of the world. |
| 06:55:17 | Russ | andersee: I just want to get as much as possible in before it switches to 2.5/2.4 |
| 06:55:18 | dirkx | Ha! Cudos to the microwindows folks. That works. |
| 06:55:31 | Russ | I have the keyboard stuff pending |
| 06:57:05 | Russ | waiting on gork to clear up some ucb1x00-audio stuff for me |
| 06:57:23 | Russ | also waitng on the spi/wheaties/atmel stuff to stabilize |
| 06:57:42 | Russ | and some working out of the backlight stuff |
| 06:58:05 | andersee | Russ: just read dwmw2's replies on the 'second "mount" hangs' thread. I have to agree it has got to be the kernel's fault -- bb mount is too simple to foul things up that way... Kernel deadlock for sure. |
| 06:58:53 | andersee | Russ: and they didn't mention busybox, so I expect it is util-linux mount doing the same thing... |
| 06:59:20 | andersee | Any other last minute busybox requests for 0.60.2? |
| 07:00:49 | jacques | phew, that's 2 unlocked tuxen |
| 07:04:06 | Russ | I doubt it |
| 07:04:16 | Russ | otherwise, there'd be more jabber about it |
| 07:04:51 | raybass | boots russ in the head |
| 07:05:04 | Russ | raybass: I need engineering paper... |
| 07:05:13 | raybass | you have dr pepper? |
| 07:05:18 | raybass | cold? |
| 07:05:38 | Russ | If you give me change, I'll snag you one |
| 07:05:43 | raybass | ooh |
| 07:05:46 | raybass | let me see |
| 07:05:52 | raybass | engineering paper not enough? heh |
| 07:06:03 | Russ | if you are poor, I'll grab you a drp |
| 07:06:09 | raybass | ya I got change |
| 07:06:11 | andersee | has Mtn. Dew for those nearby. |
| 07:06:15 | andersee | heh |
| 07:06:19 | raybass | come get your paper |
| 07:06:30 | Russ | andersee: it has caffine, good luck giving it away |
| 07:08:04 | jacques | heh, any way to get back to a blob prompt once it's done: Starting kernel ... |
| 07:08:21 | jacques | apparently it's not smart enough to figure out that there is no kernel yet |
| 07:08:56 | andersee | jacques: the little button on the back will do it. ;-) |
| 07:09:11 | jacques | andersee: heh, thanks, that worked |
| 07:09:33 | jacques | andersee: i had to press it and lunge across the room to hit a key on the terminal |
| 07:11:27 | jacques | I guess I should not be hard so blob - it's only 14kB after all |
| 07:11:28 | Russ | wish I had a faster drive |
| 07:11:34 | Russ | merging takes a long time |
| 07:12:32 | jacques | is blob protected like the hh.org bootldr on an ipaq? |
| 07:13:29 | Russ | what do you mean by protected? |
| 07:13:32 | scanline | shoots Xlib again |
| 07:13:58 | BZFlag | jacques: not on the tux |
| 07:14:00 | Xlib | Aiieeee!!!! |
| 07:14:02 | scanline | oh no!!! |
| 07:14:06 | jacques | Russ: I mean once the bootldr is installed on the ipaq, the flash locations are protected from write |
| 07:14:13 | BZFlag | jacques: nope. |
| 07:14:20 | jacques | BZFlag: thanks |
| 07:14:22 | scanline | andersee: sorry about that :) |
| 07:14:31 | Russ | easy to change form userspace with mtd tools |
| 07:14:34 | BZFlag | jacques: you can reprotect them, but the only way to do that at present is with jtag |
| 07:14:36 | andersee | heh |
| 07:14:47 | BZFlag | Russ: have you tried the mtd tools? |
| 07:14:59 | BZFlag | they should work too, but I never tried. |
| 07:15:08 | jacques | BZFlag: understood |
| 07:15:32 | Russ | I use them to do stuff, never to lock and unlock |
| 07:15:54 | jacques | I'm going to see if this second tux takes as long to flash the ramdisk as the first one |
| 07:15:55 | BZFlag | andersee: so which IP does ns1 want? |
| 07:16:25 | andersee | BZFlag: 166.70.14.212 |
| 07:16:31 | Russ | jacques: some flash bug |
| 07:16:38 | Russ | jacques: reseting it fixes it |
| 07:16:46 | andersee | BZFlag: everything look kosher now? |
| 07:16:46 | Russ | (flash code in blob, my flash code...) |
| 07:17:38 | jacques | Russ: resetting what? this is the second tux and it's taking just as long as the first - kernel goes really fast but ramdisk takes about 6 minutes |
| 07:18:14 | Russ | after you flash the kernel, hit reset |
| 07:18:30 | jacques | Russ: ah!! thanks!!! |
| 07:19:21 | jacques | Russ: that's going to save me a lot of time :-) |
| 07:19:50 | Russ | yah, its my bad, haven't got around to fixing it |
| 07:21:36 | jacques | learns the perils of using an IR keyboard in a room full of tuxen. |
| 07:22:01 | BZFlag | andersee: yep. dns is fine now. |
| 07:22:13 | andersee | BZFlag: cool. |
| 07:25:59 | scanline | grumbles at his program that just ate a couple hundred meg of memory and crashed X... |
| 07:29:26 | dirkx | Has anyone started building jtag dongles in some sort of bulk level. I am right now using an activewireinc.com board for it - but that only works if your host has USB (in which case it works very well). |
| 07:31:53 | dirkx | Hmm - weird - this screen flicker whenever heavy NFS traffic is going on. |
| 07:32:46 | jacques | dirkx: ipaq had similar issue (maybe still does) it was/is video driver |
| 07:35:02 | Russ | building 2.4.15-pre6-rmk1-tux1... |
| 07:41:22 | jacques | andersee: ok, the dns problem with pdnsd still occurs with the latest buildroot image |
| 07:43:23 | andersee | jacques: too bad. Was worth a try. |
| 07:43:34 | andersee | Hmm. |
| 07:43:55 | andersee | jacques: Do _all_ lookup fail? |
| 07:44:17 | jacques | andersee: all that I've tried - let me try dome more |
| 07:44:29 | andersee | jacques: like say, codepoet.org |
| 07:45:43 | jacques | andersee: yep. everything fails so far, including codepoet.org |
| 07:46:35 | andersee | I wonder what the deal is. |
| 07:47:23 | andersee | Just as a quick test, could you grab busybox, compile it on x86 vs glibc and test there, so we can be certain its a uClibc resolver problem. |
| 07:47:43 | jacques | andersee: np, I'll do that right now |
| 07:47:49 | andersee | i.e. not a busybox thinko.... |
| 07:48:01 | andersee | (esp since I'm about to be making a release) |
| 07:49:28 | jacques | andersee: I assume it's OK for me to use the busybox tarball from the sources dor of my latest buildroot? |
| 07:49:38 | jacques | minus the patch of course |
| 07:50:32 | andersee | sure |
| 07:50:54 | jacques | andersee: oh, BTW |
| 07:51:10 | jacques | andersee: tinylogin is not compiling in the latest cvs buildroot |
| 07:51:20 | jacques | andersee: it requires makefile mangling |
| 07:51:37 | jacques | andersee: so the first time through a buildroot make never works |
| 07:52:19 | BZFlag | jacques: so mangle and send in a patch? ;-) |
| 07:52:22 | andersee | really? |
| 07:52:38 | andersee | I did a little work on it the other day... |
| 07:53:16 | jacques | it's a trivial patch and I'm not sure I did it the best way - the makefile has a hardwired crossgcc, which happens to be very wrong |
| 07:55:27 | jacques | andersee: OK, how do I configure which applets get built into busybox? by default I don't get ping or nslookup |
| 07:55:34 | andersee | jacques: oops. the CC=/usr/i386.../gcc thing? just fixed it, thanks |
| 07:55:53 | jacques | andersee: yepper, that's the problem |
| 07:55:57 | andersee | jacques: with the stable tree, edit Config.h (see README for details |
| 07:56:00 | andersee | ) |
| 07:56:01 | jacques | andersee: and I just found Config.h :-) |
| 07:58:05 | jacques | andersee: OK, nslookup built on x86 with glibc works: |
| 07:58:08 | jacques | jacques% ./busybox nslookup www.apple.com |
| 07:58:08 | jacques | Name: madonna.activebuddy.com |
| 07:58:08 | jacques | Address: 192.168.1.2 |
| 07:58:08 | jacques | Name: www.apple.com |
| 07:58:08 | jacques | Address: 17.254.0.91 |
| 07:58:43 | jacques | andersee: so you have my permission to release 60.2 :-D |
| 07:58:55 | andersee | jacques: thanks. |
| 07:59:02 | andersee | Definatey a uClibc bug. |
| 07:59:29 | andersee | Mind submitting a bug on it, lest I get busy and forget? |
| 07:59:43 | andersee | has been terribly busy recently |
| 08:00:33 | andersee | I try to fix a bug every day or two, but the queue has gotten very long and the uClibc contributor list has recently been just me... :( |
| 08:12:38 | jacques | andersee: sure thing |
| 08:12:59 | Russ | Linux version 2.4.15-pre6-rmk1-tux1 (russ@russ) (gcc version 2.95.2 19991024 (release)) #5 Mon N |
| 08:12:59 | Russ | ov 19 02:09:58 MST 2001 |
| 08:13:07 | andersee | wohoo! |
| 08:13:34 | andersee | Russ: canyou run 'mount' for me? |
| 08:13:56 | Russ | same problem |
| 08:14:03 | andersee | Does /var show up? |
| 08:14:26 | jacques | i thought that was just a ramfs thing |
| 08:14:39 | jacques | or are we using tmpfs now? |
| 08:14:44 | andersee | Old 2.4.7 kernel had size 0 for the tmpfs, so it didn't show.... |
| 08:14:58 | andersee | should be tmpfs |
| 08:15:04 | jacques | ah |
| 08:15:19 | jacques | /proc/mounts shows it |
| 08:15:35 | jacques | andersee: but in 2.4.7 it's ramfs |
| 08:15:40 | Russ | dunno, I can't get into it atm, its stuck |
| 08:15:48 | jacques | # cat /proc/mounts |
| 08:15:48 | jacques | /dev/root / jffs2 rw 0 0 |
| 08:15:48 | jacques | proc /proc proc rw 0 0 |
| 08:15:48 | jacques | ramfs /var ramfs rw 0 0 |
| 08:16:12 | Russ | if you make a quick jffs2 with / commented out in fstab or the diagnostic shell, I can chekc |
| 08:16:34 | Russ | http://russ.dhs.org/files/tuxpatches/patch-2.4.15-pre6-rmk1-tux1.gz |
| 08:18:53 | jacques | andersee: is there a formal way to submit a bug report for uclibc? I can't find it on the home page |
| 08:19:21 | jacques | andersee: also, how do I use the microwindows makefile included in buildroot? |
| 08:19:21 | andersee | http://bugs.lineo.com/ |
| 08:19:35 | andersee | I forgot to mention that on the uClibc page I guess... |
| 08:19:58 | andersee | jacques: there is a knob at the top of buildroot-tux/Makefile |
| 08:20:17 | dirkx | Hmm - tinylogin does not want to rebuild after a make clean. Weird. 'cd tuxroot-buid/tinylogin; make clean; cd ..; make (or make tinylogin)' gives no rebuild. |
| 08:20:49 | dirkx | Hmm- what is 'CROSS=' doing in tinylogin/Makefile. |
| 08:21:11 | jacques | dirkx: heh, I was ranting about that earier today - seems that buildroot is overly optimistic about when a package has been successfully built |
| 08:21:29 | jacques | if there is any error on building a package, and you start the make again, it just starts with the next package |
| 08:21:31 | dirkx | Ah - well - in the ideal world ..... |
| 08:21:53 | jacques | it's almost like buildroot records when it starts building a package and thinks of it as'done' |
| 08:22:17 | jacques | dirkx: I found that make clean at the top level was required |
| 08:22:47 | andersee | dirkx: rm tinylogin/tinylogin |
| 08:22:54 | dirkx | Well - it has funcly @cmp's between its output and the ones in root. And if the first are not wacked by a make clean you end up in trouble. I.e. no true dependency. Plus it does not depend on makefiles either - hence a cvs update often causes a rebuild to be skiped :-) |
| 08:23:18 | dirkx | andersee: that is not quite enoug :-(; it seems to really want a make clean at top level. |
| 08:23:24 | andersee | dirx: yeah, the dependancies are very lame |
| 08:23:51 | andersee | dirx: I did it that way since otherwise it would recompile 3000 things unnecessarily |
| 08:24:03 | dirkx | andersee: well - if you want to see sheer overkill - check out a build in the netbsd or freebsd world - 80% of the make files are about that ! It is neat but a *lot* of work to maintain. |
| 08:24:06 | andersee | This sortof cheats... |
| 08:24:39 | dirkx | andersee: that is ok - it works well enough for playing in the evenings. |
| 08:24:54 | andersee | I just remember the right files to kill to make stuff rebuild... |
| 08:29:01 | dirkx | Andersee/Russ: did you not say a moment ago that pthread was in ucLibc ? Is that in head or on a specific branch |
| 08:30:03 | dirkx | Sorry - cvs up -D -p got it just fine. My bad, |
| 08:30:04 | dirkx | . |
| 08:30:09 | andersee | I have a minimalist lame pthread sortof implementation. |
| 08:30:40 | dirkx | that is ok - the J2ME stuff needs very little (I've got it working now with *no* code behind it - just empty stubs). |
| 08:30:41 | andersee | Poke at it very hard and you will find tons on non-posix-nesses. |
| 08:31:09 | jacques | starts a buildroot with microwindows |
| 08:31:23 | andersee | I think clone needs implementing on ARM as well to make it go. |
| 08:31:28 | jacques | hmmm, has anyone contributed a tinyx makefile? |
| 08:31:34 | dirkx | :-) now what we want is a nice 'make cvsudate' :-) |
| 08:31:49 | andersee | I've only tried it on x86 (I wrote the current implementation in about 30 minutes) |
| 08:31:53 | dirkx | andersee: I can look into that. Esp. if I need it. |
| 08:33:03 | jacques | andersee: is there a reason why a make clean at top level does not rebuild the kernel? |
| 08:33:53 | dirkx | Ah - I see what you mean - no arm code for that :-) |
| 08:33:57 | andersee | jacques: I added a seperate kclean, since I needed to rebuild user space often, the kernel rarely |
| 08:34:22 | andersee | dirkx: easy enough to swipe from glibc and adjust... |
| 08:34:26 | jacques | andersee: ic :-) |
| 08:34:59 | andersee | jacques: rebuilding the kernel takes forever on my lowly box... |
| 08:35:04 | andersee | :) |
| 08:35:13 | dirkx | Aye - just looking at the i386 code. |
| 08:35:47 | jacques | andersee: I know what you mean - it takes mine almost 2 minutes |
| 08:36:01 | dirkx | what is the debian apt-get magic to get source rather than binary ? |
| 08:36:07 | andersee | thwaps jacques with a soggy salmon |
| 08:36:08 | jacques | (actually longer with all those drivers) |
| 08:36:10 | dirkx | in this case for a fresh glibc(2) copy ? |
| 08:36:14 | jacques | :-D |
| 08:36:32 | andersee | my box is a lowly PII celeron 400.... |
| 08:37:00 | andersee | I could compile as fast on a sheet of paper... |
| 08:37:03 | andersee | :-) |
| 08:37:07 | jacques | andersee: if you like to compile fast, I have one word fopr you - athlon |
| 08:37:28 | andersee | has recently been proposing a new computer to his wife..... |
| 08:39:34 | jacques | andersee: I have two words for you - dual athlon |
| 08:40:00 | andersee | is drooling |
| 08:40:13 | andersee | and then we could turn off the furnace too. ;) |
| 08:40:23 | jacques | andersee: oops, tinylogin still has that problem - I just checked out buildroot - would you expect it to be fixed? |
| 08:40:31 | jacques | andersee: heh |
| 08:40:41 | andersee | did you do a cvs up in tinylogin |
| 08:40:44 | andersee | ? |
| 08:40:54 | jacques | oops |
| 08:40:57 | jacques | I just noticed |
| 08:41:11 | andersee | rm -rf tinylogin sources/tinylogin* |
| 08:41:12 | jacques | for some stupid reason I was assuming that buildroot was getting things like that from cvs |
| 08:41:51 | andersee | would be nice, but I never added that |
| 08:42:25 | jacques | jacques% cvs up |
| 08:42:25 | jacques | cvs update: authorization failed: server opensource.lineo.com rejected access to /var/cvs for user anonymous |
| 08:42:25 | jacques | cvs update: used empty password; try "cvs login" with a real password |
| 08:42:39 | dirkx | cvs login |
| 08:42:44 | jacques | I did the login |
| 08:42:49 | dirkx | 'anonymous' or 'anoncvs' |
| 08:42:56 | dirkx | yes but oyu need to do it on each repo level. |
| 08:43:06 | dirkx | or edit your .cvspass file and clone each line for each repo |
| 08:43:19 | jacques | yeah, I did but hjust noticed oss.lineo.com vs opensource.lineo.com |
| 08:43:23 | jacques | are they the same machine? |
| 08:43:26 | andersee | yup |
| 08:43:54 | dirkx | ucLibc has not yet a termcap or terminfo yet ? right ? |
| 08:43:59 | jacques | ok, got it |
| 08:45:58 | andersee | There used to be some stuff in there when I took it over, but it was GPLed so I axed it. |
| 08:46:55 | jacques | is that why it assumes I'm on a 80x24 terminal? |
| 08:47:00 | dirkx | andersee: what license is it under now ? apache/bsd ? mpl ? - I though it was gpled |
| 08:47:31 | andersee | LGPL |
| 08:47:37 | andersee | like glibc |
| 08:47:40 | dirkx | Ah - ok. |
| 08:49:34 | dirkx | Hmm - shame readline wants termcap. |
| 08:50:04 | andersee | What do you need readline for? |
| 08:50:09 | andersee | bash? |
| 08:50:19 | jacques | mmmmmm, bash |
| 08:50:21 | dirkx | arg - silly buggers. why does gdb have its *own* readline. That is bad. |
| 08:50:25 | dirkx | Noo - not bash :-) |
| 08:50:48 | dirkx | Actually - netbsd's bourne shell compiled fine (but is way too big). |
| 08:51:03 | andersee | For what? |
| 08:51:07 | andersee | The phone? |
| 08:51:15 | dirkx | yes - for on the phone. |
| 08:51:43 | andersee | busybox contains a port of freebsd's ash -- with lots of adjustments... |
| 08:51:58 | jacques | wow, microwindows compiles a lot faster than tinyx |
| 08:52:16 | dirkx | Aye - ash makes sense. There are some very nice bits you can take from the PicoBSD disk as well should you care. |
| 08:52:58 | andersee | I've looked over it before. Been a long while though |
| 08:54:14 | dirkx | It's made great progress. Is very reliable and rock solid now. And still fits on a single floppy with spare room to add things like IPsec/IKE, ssh or a stripped down version of apache. |
| 08:54:38 | dirkx | Oh -nice - gdb's configure mixes up host and target :-(. |
| 08:56:09 | andersee | dirkx: I seen some busybox + uClibc based floppy disks with unwholsome amount of stuff on them.... |
| 08:56:22 | andersee | I've that is |
| 08:56:31 | andersee | should learn to spell |
| 08:56:36 | andersee | or proofread |
| 08:57:46 | dirkx | :-) what I like about picoBSD is the build around it - single config file and a make - even if your OS is totally differnt. Very isolated and very easy to 'stick' a version to -i.e. commit a single (or a few) files to CVS and be able to say a year later to a customer what they have - or reproduce what you did two years ago with easy. |
| 08:59:30 | andersee | the new version of busybox in CVS uses a much cleaner build system. |
| 08:59:39 | andersee | Much nicer... |
| 08:59:59 | dirkx | actually - buildroot already works quite nice and clean. I've seen worse ! |
| 09:00:49 | dirkx | by the way - is there a file on debian which tells you what apt-get file you have and what version - to record/bookeep a snapshot ? |
| 09:02:42 | andersee | you mean like 'dpkg --set-selections' and 'dpkg -l' |
| 09:04:31 | dirkx | INSTALLBASE in buildroot-tux/kernel should point to the final runtime lib/modules ? |
| 09:06:19 | dirkx | Or is the idea that the files in buildroot-tux/kernel get copied into the linux kernel tree ? |
| 09:07:46 | dirkx | Or should it always get patched in with sources/wheaties.patch ? |
| 09:14:24 | dirkx | i.e. in other words - what is the 'right' way to patch in files from the kernel CVS repository ? |
| 09:15:24 | dirkx | Or is the patch always the most recent ? |
| 09:16:35 | andersee | You need a patch something like http://codepoet.org/kernel.patch |
| 09:16:56 | andersee | But Russ has a newer kernel since then |
| 09:17:35 | dirkx | andersee: thanks - trying to work out how to stay up to date - whilst keeping all the local config tweaks. |
| 09:19:31 | dirkx | folks - thanks for all the help - good night. |
| 09:20:33 | andersee | night |
| 11:14:10 | jacques | hmm, microwindows image is not so stable |
| 11:17:14 | jacques | anyone using latest buildroot microwindows image? |
| 11:17:14 | jacques | it just hangs on boot for me |
| 12:53:42 | jacques | hi Tangent |
| 12:53:51 | Tangent | jacques : Morning |
| 12:56:58 | jacques | Tangent: what GUI are you running on your `screens? |
| 12:57:11 | Tangent | jacques : Blackbox |
| 12:57:28 | jacques | Tangent: using udrive? NFS? |
| 12:57:57 | Tangent | jacques : using udrive |
| 12:58:23 | Tangent | jacques : I just plugged in my intimate-ready drive, and hacked the linuxrc on the buildroot-tux jffs2 image |
| 12:58:48 | jacques | Tangent: oooh, gimme |
| 12:59:01 | jacques | Tangent: I have one chrooted to the udrive right now |
| 12:59:02 | Tangent | jacques : OK.. sec... |
| 12:59:16 | jacques | Tangent: the other is running a freshly compiled microwindows image |
| 12:59:28 | jacques | Tangent: no rush |
| 13:00:04 | jacques | Tangent: heh, what is the deal with the keyboard? |
| 13:00:15 | jacques | Tangent: I know exactly what you were complaining about now |
| 13:00:17 | Tangent | jacques : How do you mean... It's a keyboard.. it works |
| 13:00:33 | jacques | Tangent: I mean the dropped characters |
| 13:00:53 | Tangent | jacques : try newer batteries... worked for mew |
| 13:01:06 | jacques | Tangent: really? hmmmmm |
| 13:01:23 | jacques | Tangent: you mean the batts that came with it are close to end of shelf life? |
| 13:01:29 | jacques | that would be cool |
| 13:01:41 | jacques | Tangent: you can type fast on it now? |
| 13:02:21 | Tangent | jacques : Yep.. I can type at full rate on it now... |
| 13:02:27 | Tangent | jacques : email address? |
| 13:02:42 | jacques | Tangent: excellent |
| 13:03:46 | Tangent | jacques : OK.. email on it's way... it contains a replacement linuxrc for flash, and a script (/root/devstuff) to go on the microdrive |
| 13:04:26 | Tangent | jacques : I don't run init in intimate... sshd2, syslog and all that stuff... the script just boots you to a prompt... I then /etc/init.d/x start to get X going, and still have some memory free |
| 13:10:13 | jacques | Tangent: thanks! |
| 13:10:43 | jacques | Tangent: so X starts up and recognizes the screen automatically? |
| 13:10:54 | Tangent | jacques : Heh..you do realise that we're gonna have to sort intimate to boot on Tux by default.. or on iPAQ.. wherever the drive is connected |
| 13:11:09 | Tangent | jacques : When you start X, it touchecreen etc just works.. |
| 13:11:16 | jacques | Tangent: cooool |
| 13:11:25 | Tangent | jacques : You want to get fbtuxcal from the CVS to calibrate the touchscreen |
| 13:11:41 | Tangent | jacques : and _DONT_ run it inside X.. just use it before you start X |
| 13:12:02 | jacques | Tangent: going to cvs checkout it now |
| 13:13:11 | jacques | interesting - is it meant to be built inside the buildroot dir? |
| 13:13:35 | Tangent | jacques : I compiled it on the Tux |
| 13:13:50 | Tangent | jacques : It's just sitting there in /usr/src for me :) |
| 13:14:16 | jacques | Tangent: I'm just going by the default makefile - looking for compiler in ../staging_dir |
| 13:16:00 | jacques | Tangent: I assume you installed the images (bmp's) somewhere for fbtuxcal? |
| 13:16:33 | Tangent | jacques : Yeah.. they go in /usr/include or something.. I straced it to see where it failed |
| 13:16:52 | jacques | Tangent: cool |
| 13:23:12 | jacques | wow cool |
| 13:23:39 | jacques | both amex and my bank have updated their sites to allow me to access them with mozilla(galeon) now |
| 13:27:02 | Tangent | jacques : Cool.. :) |
| 13:44:41 | jacques | any of you guys seeing these: |
| 13:44:45 | jacques | Fixing up bad data abort at 0000ae94 |
| 13:44:45 | jacques | pgd = d03cc000 |
| 13:44:45 | jacques | *pgd = d03f2001, *pmd = d03f2001, *pte = d00f501f, *ppte = d00f5aae |
| 13:49:52 | jacques | yes!!! fresh batteries are the answer! |
| 13:54:23 | jacques | anyone messing with fbdev? |
| 13:57:17 | jacques | Tangent: what's the xyswap in your devstuff script? |
| 14:14:30 | Tangent | jacquesZzZz : Ah.. yes.. my mate made that... swaps the touchscreen orientation around.. from before I was using fbtuxcal |
| 14:41:01 | SteevD | SpaceCoaster: (if you are there) |
| 14:41:12 | SpaceCoaster | SteevD: hi |
| 14:42:06 | SteevD | SpaceCoaster: you pointed me at "kernel" on CVS, with recent wheaties. This is setup to build modules. Should I reverse out the previous wheaties kernel patches? |
| 14:42:38 | SteevD | or, can I merge into the kernel source (you've moved things apart from how they were - serparate modules for serial-flash etc) |
| 14:43:07 | SpaceCoaster | SteveD: the files should build either way, just drop the wheaties.c into your kernel source in drivers/char. |
| 14:43:09 | SteevD | BTW, SteevD==SteveD - left my home machine on |
| 14:43:59 | SteevD | presumably the wheaties.c now doesn't do the serial flash - I'll have to adjust makefile to link all the parts? |
| 14:44:20 | CosmicPenguin | scanline: Whats the news with the new touchscreen? |
| 14:44:43 | CosmicPenguin | scanline: or rather, the new touchscreen driver? Did GoRK get anything together? |
| 14:45:11 | SpaceCoaster | hmm, true, or take out the call to the serial flash from the mtd stuff. then it won't be loaded. |
| 14:45:52 | SteevD | SpaceCoaster: let's see how I get along. what the "preferred" setup then - modules? |
| 14:46:07 | SpaceCoaster | CosmicPenguin: I did something for the touchscreen that you will probably hate. |
| 14:46:28 | CosmicPenguin | SpaceCoaster: And that is? |
| 14:46:40 | SpaceCoaster | SteevD: modules is much, much easier for testing, but not the best if the hardware is broken without the module. |
| 14:47:38 | SpaceCoaster | CosmicPenguin: I wrote a little transformation module called tscal.c which steals the function vectors from the new "improved" ucb1x00-ts.c and provides backwards compatible ioctls. |
| 14:48:17 | SteevD | SpaceCoaster: OK - I'll hoik out the other wheaties driver... |
| 14:48:19 | CosmicPenguin | SpaceCoaster: Why would i hate that? |
| 14:48:38 | CosmicPenguin | SpaceCoaster: Thats a great idea! |
| 14:49:21 | SpaceCoaster | CosmicPenguin: well it's one of those love, hate it things so I was setting my sights low to avoid disappointment :-) |
| 14:49:59 | CosmicPenguin | SpaceCoaster: Lemme see.... a quick module that means that we don't have to write any new code.... |
| 14:50:13 | CosmicPenguin | SpaceCoaster: Whats not to love? |
| 14:50:29 | SpaceCoaster | CosmicPenguin: and it works with X, MicroWindows, PicoGUI, QTE... |
| 14:50:38 | CosmicPenguin | That also saves me from taking RMKs name in vain every three seconds.... |
| 14:50:56 | CosmicPenguin | Speaking of QTE, has anyone gotten that up and running yet? |
| 14:51:33 | SpaceCoaster | Yes, I have 3.0.0 compiled works pretty well on the TuxScreen. |
| 14:51:49 | CosmicPenguin | Sweet.... you gonna toss that into the buildroot? |
| 14:52:31 | SpaceCoaster | Good idea, only a tiny patch required to enable the CustomTouchScreen code. |
| 14:52:47 | CosmicPenguin | CustomTouchScreen code? |
| 14:53:11 | CosmicPenguin | Also, GoRK and scanline need to get picogui into the buildroot.... |
| 14:53:22 | CosmicPenguin | And the X folks need to get something together as well.... |
| 14:53:25 | SpaceCoaster | Just to turn on the correct drivers in QTE. |
| 14:53:32 | CosmicPenguin | SpaceCoaster: IOC |
| 14:53:57 | CosmicPenguin | Then we will have 4 major graphics engines available by selecting a single flag in a Makefile.... |
| 14:55:51 | SpaceCoaster | CosmicPenguin: have a look at tscal.c, it is neat. The only hokey part is that I have to copy the coords out/in from userspace one more time. |
| 14:56:17 | CosmicPenguin | SpaceCoaster: *shrug* compatablity is the key..... it is in the CVS version of the kernel, right? |
| 14:56:57 | SpaceCoaster | CosmicPenguin: is there a CVS for our kernel? |
| 14:57:04 | CosmicPenguin | ?? I thought there was? |
| 14:57:46 | SpaceCoaster | ok it is in http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/tuxscreen/kernel/ |
| 14:59:09 | CosmicPenguin | SpaceCoaster: Ok... I'll look at it when i get ready to upgrade... I'm still messing around with 2.4.7 - I have to merge in a new Microwindows release, and I have a ton of code to put in.... |
| 14:59:36 | CosmicPenguin | Also, I still haven't written anything for tuxphone yet.... :( |
| 15:04:55 | SpaceCoaster | CosmicPenguin: it's one less thing to be done, we could use it to provide compatibility with other systems for quick porting, i.e. extend it to support stuff like "mode=ipaq" etc. |
| 15:05:49 | SpaceCoaster | CosmicPenguin: I didn't have any luck getting the latest 2.4.13 kernel to boot, gonna try again today. |
| 15:07:10 | CosmicPenguin | SpaceCoaster: I haven't even tried... it scares me a bit, actually... :) |
| 15:12:08 | Russ | morning SpaceCoaster |
| 15:12:19 | SpaceCoaster | Russ: hi |
| 15:17:52 | Russ | CosmicPenguin: any chance to look at the backlight stuff? |
| 15:31:41 | CosmicPenguin | Russ: Afraid not... I am totally embarssed though... I have been droping the ball big time lately... |
| 15:33:59 | Russ | morning _rmk_ |
| 15:34:16 | _rmk_ | morning Russ |
| 15:34:51 | Russ | _rmk_: thats for applying all that stuff so quickly |
| 15:35:15 | _rmk_ | the only thing that's left is the keyboard stuff. |
| 15:35:54 | Russ | I have backlight support, and spi/atml/dsp in my queue still, it will probably be a while on the latter |
| 15:37:02 | Russ | _rmk_: I'm looking at head-armv.S and it looks like it does an initial setup of the MMU assuming that there is 4M of contigous memory at RAM start |
| 15:37:42 | _rmk_ | Russ: it does, but you shouldn't need to concern yourself too much with that - it does that so that large-ish kernels work. |
| 15:38:21 | Russ | _rmk_: one of the patches I have pending is for 32M sodimm support, it does so by completely ignoring the first 8M because its in 1M chunks |
| 15:38:42 | erikm | Russ: hmm, this is somehow not right |
| 15:39:01 | Russ | I think the whole hack could be avoided if I got head-armv.S to map the first 4 1M chunks together |
| 15:39:04 | erikm | Russ: I'd expect the kernel to work if you tell it "here is 16MB of memory, go play with it" |
| 15:39:53 | Russ | right now its just giving the kernel the bank of contiguous memory at 0xd0000000, but that requires a different kernel |
| 15:40:02 | _rmk_ | Russ: please note that there may be delays in my responses - I'm normally on 3 channels at one time, now I'm on 4 8/ |
| 15:40:22 | Russ | erikm: I think the zone stuff does that |
| 15:40:32 | erikm | Russ: *nod*, and that's what we should avoid to do. |
| 15:40:59 | Russ | erikm: but before the zone stuff runs, there is an initial setup of the MMU |
| 15:41:25 | erikm | Russ: which should work in both 8MB and 32MB SODIMM case. |
| 15:42:02 | Russ | well, the memory at 0xc0000000 is in 1M chunks when the 32M SODIMM is in because a different row/col setting |
| 15:42:13 | erikm | Russ: if we just lie to the kernel that there is only 4MB at 0xd0000000, it should work in the initial case, and later on we can feed it the correct memory map |
| 15:42:35 | Russ | erikm: but then you need a different case for the kernel when there is no SODIMM |
| 15:43:05 | erikm | Russ: IMHO all tuxscreens BZFlag sold are *with* SODIMM |
| 15:43:19 | erikm | Russ: so we can safely ignore that |
| 15:44:04 | Russ | erikm: its still removable |
| 15:44:26 | Niddix | I think I found a slight problem with make distclean.. although I'm sure its just me. |
| 15:44:26 | erikm | Russ: of course, but why would you like to do that? |
| 15:46:25 | Russ | erikm: I don't know, but it seems like a more universal fix to change head-armv.S |
| 15:46:45 | Russ | esp if you come across an arch with memory in 1M or 2M hunks |
| 15:47:52 | Russ | _rmk_: btw, I made a standardized interface for backlights/frontlights. |
| 15:48:02 | EGg0 | is back (gone 13:42:28) |
| 15:48:15 | _rmk_ | Good - I've got a couple of those in the patch queue atm. |
| 15:48:22 | Russ | _rmk_: drivers register themselves with sa1100fb.c, and userspace interacts with the backlight though a /dev/fb ioctl |
| 15:48:40 | _rmk_ | the right way 8) |
| 15:48:45 | Russ | still hammering out some of the details, but it works, would you like to take a quick peek? |
| 15:48:56 | _rmk_ | I hope the ioctl numbers are defined using the _IOC_* macros in asm/ioctl.h ? |
| 15:49:37 | Russ | not something I really knew about, but its easy to change |
| 15:49:46 | _rmk_ | hmm, I ought to list all these things that cause patches to find their way to 'discarded' real quickly. |
| 15:52:02 | erikm | goes through the "incoming" patch queue |
| 15:52:48 | _rmk_ | hmm, not checking down_interruptible() return value. |
| 15:53:18 | Russ | all the docs I could find didn't say anything about a return value |
| 15:53:38 | Russ | but from your concern, I'm guessing it returns if there is an interrupt |
| 15:53:53 | _rmk_ | if (down_interruptible(&sem)) return -ERESTARTSYS; |
| 15:54:01 | _rmk_ | yep |
| 15:54:38 | Russ | I'm trying to protect from the module being removed while the settings are being changed through the ioctl |
| 15:55:50 | _rmk_ | it might be worth using the 'owner' trick that's used in struct file_operations there then. |
| 15:56:21 | Russ | whats that do exactly? |
| 15:56:29 | Russ | or what file is that done in |
| 15:56:35 | _rmk_ | many 8) |
| 15:56:48 | _rmk_ | basically though, you pass a pointer to the module structure in. |
| 15:57:18 | _rmk_ | when you want to reference something in the module, you increment the module use count before you call the module function. |
| 15:57:32 | Russ | ah |
| 15:57:38 | _rmk_ | tries to think where there's a suitable example. |
| 15:58:11 | _rmk_ | drivers/char/busmouse.c is pretty simple. |
| 15:58:50 | Russ | alright |
| 15:59:12 | _rmk_ | see the try_inc_mod_count and semaphore stuff |
| 16:01:55 | Russ | lock_kernel()? |
| 16:03:08 | _rmk_ | you should be able to ignore those as far as the backlight stuff goes. |
| 16:03:23 | _rmk_ | lock_kernel() and unlock_kernel() is otherwise referred to as the big kernel lock. |
| 16:04:42 | Russ | so its increasing the usage count of itself? |
| 16:05:00 | _rmk_ | not quite. |
| 16:05:01 | _rmk_ | ok. |
| 16:05:21 | _rmk_ | when you register/unregister a service, you protect it with a semaphore. |
| 16:05:46 | _rmk_ | in busmouse.c's case, its mouse_sem. |
| 16:06:11 | _rmk_ | this means that if mouse_sem is locked, nothing can be registered or unregistered. |
| 16:06:30 | _rmk_ | when you want to use a reference to a module, you take that semaphore. |
| 16:06:42 | _rmk_ | (this prevents your reference going stale) |
| 16:07:03 | _rmk_ | you then try_inc_mod_count() on it |
| 16:07:38 | _rmk_ | if the module is in the process of unloading (eg, waiting on your semaphore), then try_inc_mod_count will fail. |
| 16:08:02 | _rmk_ | otherwise it will succeed, and increase the child module use count by one. |
| 16:08:34 | _rmk_ | you can then release the semaphore safe in the knowledge that the module isn't going to disappear under you. |
| 16:08:54 | Russ | so you are only holding the semaphore momentarily |
| 16:09:05 | _rmk_ | yep. |
| 16:09:50 | Russ | and you can use down(&sem) as you know you aren't sleeping |
| 16:10:31 | _rmk_ | you might sleep, but for a short time. |
| 16:11:17 | _rmk_ | hmm, another problem I've spotted... |
| 16:11:27 | _rmk_ | not checking return values from copy_to_user/copy_from_user |
| 16:11:44 | _rmk_ | if (copy_from_user(to,from,len)) return -EFAULT; |
| 16:12:08 | Russ | think I'll need a goto unlock in there, its already starting to get messy |
| 16:12:20 | Russ | er, just do that before locking |
| 16:12:46 | Russ | (and after locking in the copy_to_user case) |
| 16:13:06 | _rmk_ | umm, you shouldn't call ucb1x00_enable() and leave it enabled. |
| 16:13:25 | _rmk_ | the ucb1x00 code is setup such that it'll enable the interface clock when needed. |
| 16:13:55 | Russ | I noticed the other modules call ucb1x00_disable() when they aren't using it |
| 16:14:19 | _rmk_ | the ucb1x00 chips don't require the clock to function luckily 8) |
| 16:14:47 | Russ | I do call ucb1x00_disable on the removing of the module |
| 16:15:04 | Russ | should I call ucb1x00_enable/disable for each function? |
| 16:16:19 | _rmk_ | I think I'd prefer it that way |
| 16:16:57 | Russ | k |
| 16:17:00 | Russ | anyway, off to work |
| 16:42:43 | Russ|werk | @work |
| 17:37:52 | EGg0 | is away: I'm reading a book. Do you remember books? |
| 18:13:31 | Russ|werk | afternoon tim |
| 18:44:20 | BZFlag | afternoon. |
| 18:45:35 | gjg | hi |
| 19:15:04 | Niddix | Ok this will sound like a stupid question. But what could cause an erroneous no space left on device error? |
| 19:15:53 | Niddix | Ok.. it was a stupid question.. I found the problem. |
| 19:54:07 | as_ | Evnin all. |
| 20:05:34 | Russ|werk | _rmk_: actually the fb checklist was slightly usefull, I had values for the tuxscreen I was using in the old format |
| 20:06:06 | Russ|werk | _rmk_: to I translated them into the new format, and copy and pasted the old one |
| 20:09:13 | _rmk_ | Russ: its a duplicate of the same information though, and something the new driver was always intended to get rid of. |
| 20:09:42 | Russ|werk | _rmk_: it was only usefull because it was done that way at some point |
| 20:12:57 | erikm | _rmk_: hmm, I was just about to send you a diff that removed the old LART entries |
| 20:13:53 | _rmk_ | Russ: its been there for people to verify the new values for 6 months. I think 6 months enough time to do that, and I think it is entirely reasonable to remove the values now. |
| 20:14:12 | Russ|werk | yah, trash em |
| 20:14:37 | _rmk_ | (they first appeared on the 28th May) |
| 20:14:59 | Russ|werk | if I'm moving my direction register stuff to shannon.c, should I also move the stuff where its the driver setting the direction? |
| 20:15:13 | Russ|werk | (ie, mcp/ucb driver) |
| 20:15:19 | Russ|werk | or leave it there |
| 20:16:16 | _rmk_ | just the GPDR bit - the GPCR/GPSR should stay in the ucb code. |
| 20:16:31 | Russ|werk | right |
| 20:16:46 | _rmk_ | ie, move GPDR into shannon_map_io() |
| 20:23:09 | _rmk_ | grr, pt_system3 people have yet-another-way-to-handle-backlight |
| 20:23:18 | _rmk_ | patch 780/1 |
| 20:23:46 | Russ|werk | _rmk_: any good? |
| 20:24:48 | _rmk_ | nope. |
| 20:24:51 | erikm | Russ|werk: just submit yours and have it declared "standard" by _rmk_ :) |
| 20:25:30 | Russ|werk | erikm: I need to clean it up a bit, then I'll submit it to linux-arm-kernel, take suggestions, then sumbit it |
| 20:25:56 | _rmk_ | Russ: might also be worth CC:'ing the fbdev developers as well. |
| 20:26:19 | Russ|werk | _rmk_: would be really easy to move the driver up a level so that its available to all arches |
| 20:27:21 | _rmk_ | agreed - otherwise we'll be going through the same exercise again with some other framebuffer driver. |
| 20:27:47 | _rmk_ | rotfls |
| 20:28:05 | _rmk_ | + bool 'Bloat the size of struct page.' CONFIG_PIGMEM |
| 20:28:06 | Russ|werk | _rmk_: I mean at a later date, it would just be copy and paste, and then change the calls to register in the indiviual backlights |
| 20:29:19 | _rmk_ | Russ: I can guarantee that a new driver will have their own special way of handling the backlight. |
| 20:29:46 | Russ|werk | what do you mean? |
| 20:31:29 | _rmk_ | in my experience, people don't reuse code in existing drivers, so I agree it should be moved up a level. |
| 20:32:30 | Russ|werk | alright, I'll make that change as well then |
| 20:33:20 | Russ|werk | brb |
| 21:16:03 | Squart | anyone awake? |
| 21:16:14 | erikm | no |
| 21:16:19 | Squart | ;-) |
| 21:17:16 | Squart | I am trying to find mscomm16.ocx somewhere on the web, but I can't get it. does someone have it on his windoze box? |
| 21:18:19 | erikm | google says http://www.atex.fi/services/listat/vbxfile.doc |
| 21:18:52 | Squart | I know, I don't need a .doc ... |
| 21:28:45 | _rmk_ | uploads 2.4.15-pre6-rmk2 |
| 21:35:10 | kernelwork | uint is 32 bit on both arm and x86 right? |
| 21:35:20 | _rmk_ | yep |
| 21:36:04 | kernelwork | Ok... I'm making a binary file on an x86 host that I want to read mmaped on an ARM host. |
| 21:36:20 | kernelwork | anything I should particularly watch out for? |
| 21:36:31 | SteveD | SpaceCoaster: built the modularized wheaties driver, installed latest tuxphone and my speaker button etc now appears really solid. Nice work, thanks! |
| 21:36:43 | erikm | kernelwork: yes, endianess, though that won't bite you for arm and i386 |
| 21:36:46 | SteveD | kernelwork: watch out for alignment issues on the arm. |
| 21:37:19 | kernelwork | Ok, I have been trying to keep everyting 32bit aligned |
| 21:38:07 | Russ|werk | just do ntohl, htonl for fun |
| 21:38:43 | _rmk_ | drops off openprojects. |
| 21:39:30 | kernelwork | hates endianess |
| 21:39:58 | kernelwork | isn't most everyone little endian and only a few are big endian? |
| 21:40:05 | kernelwork | or is it the other way around. |
| 21:40:41 | SteveD | kernelwork: most are little, sparc is big, newer arms can be either I believe |
| 21:40:46 | AlHazWork | kernelwork: I could swear I've heard the argument both ways |
| 21:40:53 | AlHazWork | mips are bi-endian as are arm |
| 21:41:19 | kernelwork | Little endian always made more sense to me. |
| 21:41:27 | AlHazWork | iirc m68k macs were big endian |
| 21:41:31 | AlHazWork | I forget what ppc is |
| 21:42:53 | erikm | AlHazWork: arm is big endian |
| 21:43:10 | erikm | AlHazWork: it *can* workj little endian, but that's not real little endian |
| 21:43:15 | SteveD | kernelwork: I did some code on the HP "Saturn" processor. nibble addressed, little-endian by nibbles. Ooo-err |
| 21:43:18 | AlHazWork | erikm: ah |
| 21:43:35 | erikm | AlHazWork: more like little endian on a big endian bus, which doesn't work too wel |
| 21:43:42 | AlHazWork | *nod* |
| 21:43:54 | erikm | AlHazWork: see the hooplas nicolas had to do in his latest -np patches |
| 21:43:56 | AlHazWork | most of the mips boards I've worked on could be either way, was a matter of swapping in different firmware |
| 21:44:01 | AlHazWork | *nod* |
| 21:45:04 | kernelwork | is now confused as to the difference and wishes he had his i386 reference to look at |
| 21:45:39 | kernelwork | Not that i386 is a good example mind you, but it's the only reference I really have for any architecture. |
| 21:47:14 | MacAtack | ot... anyone know how to map the escape key to finish commands like the tab key does (RH 7.1)? |
| 21:47:46 | BZFlag | kernelwork: for the number 12345678 hex x86 has 78563412 in memory |
| 21:48:45 | BZFlag | which is little-endian as the least significant byte is stored first. |
| 21:49:00 | BZFlag | better name would be little-beginian. |
| 21:49:26 | BZFlag | but hey, I'm not a liliputian. [sp?] |
| 21:50:36 | BZFlag | MacAtack: bad idea. |
| 21:51:20 | MacAtack | HP-UX does it? I'm researching this for a client (he's an HP-UX Luser) |
| 21:51:27 | BZFlag | MacAtack: esc is used as the beginning of many terminal codes, so it has timing issues. you might get esc just over a remote session by using func keys etc. |
| 21:53:34 | MacAtack | ok... then why can HP-UX get away with it? |
| 21:54:11 | BZFlag | MacAtack: "\e": complete in .inputrc should do it. |
| 21:54:25 | BZFlag | but I still say it's a bad idea. |
| 21:54:36 | MacAtack | ok thanks... i'll mention it to him that it may wakk things out on his linux box |
| 21:54:57 | BZFlag | it will mess things up everywhere. just not all the time. |
| 21:55:18 | MacAtack | ok, thanks again |
| 22:25:54 | BZFlag | collord: hey! looks nice. ;-) |
| 22:26:50 | AlHazWork | lunux? |
| 22:27:03 | BZFlag | eep. ;-) |
| 22:33:18 | kernelwork | needs to update his domain name registration. Should have done it last summer, the prices have gone up :( |
| 22:38:04 | AlHazWork | dotster is pretty cheap, if you can live with never getting email back from customer support |
| 22:38:32 | as_ | They havnt turned up yet BZ, but I wont start worrying till a week or so. |
| 22:39:51 | kernelwork | AlHazWork: The trouble here is there is only one registrar for .cc domains |
| 22:40:01 | AlHazWork | kernelwork: Ah |
| 22:40:20 | kernelwork | It's only another $15/year but that hurts when buying another 5 years. |
| 22:41:19 | AlHazWork | ah |
| 22:41:50 | kernelwork | is attached to his domain name because he gets email through it. |
| 23:49:09 | SpaceCoaster | spse__: hi |
| 23:51:08 | spse__ | watcha |
| 23:52:19 | spse__ | SpaceCoaster: |
| 23:52:32 | spse__ | how goes your translation thing for touchpad |
| 23:55:27 | SpaceCoaster | spse__: great, works, very simple too |
| 23:55:54 | spse__ | dod you add in calibration ioctls |
| 23:56:02 | bhima | yawns. |
| 23:56:11 | bhima | needs to find a 70 pin IDC plug. |
| 23:56:51 | SpaceCoaster | yes, they were always in the plan, otherwise it wouldn't have been very backwards compatible, but the insmod variables are probably as useful. |
| 23:57:51 | spse__ | what kernel func to you use to read the char device of the touchscreen ? |
| 23:57:56 | bhima | I made a little extension cable between the motheboard and daughterboard of my tuxcreen. |
| 23:58:11 | bhima | but the spacing between teh two 34 pin connectors is _small_. |
| 23:58:52 | Russ | SpaceCoaster: don't forget to move it to userspace eventually |
| 23:59:54 | SpaceCoaster | spse__: filp_open() |