**** BEGIN LOGGING AT Sun Dec 04 02:59:57 2005 Dec 04 13:19:11 marceln: ping Dec 04 13:23:01 Hi Dec 04 13:23:37 rwhitby: Did you find the monotome commit? Dec 04 13:23:56 pulling right now Dec 04 13:24:11 saw the email - any reason why you sent it to nslu2-linux instead of nslu2-developers ? Dec 04 13:24:33 (I read -developers in real-time, whereas I get -linux as a digest) Dec 04 13:24:44 Oke. I wasn't sure that i thit the monotome commit in the right way. Dec 04 13:25:39 wasn't really necessary to do a new branch. You're the only one working on Unslung at the moment, so you won't affect anyone else unless you change the preferred version of the kernel in the unslung.conf file Dec 04 13:25:45 Most of the traffic is on nslu2-linux. Dec 04 13:27:14 please send unslung 6.x stuff to -developers until we have an alpha version ready for the wider community to test ... Dec 04 13:27:38 Ok. Dec 04 13:28:17 and you will want to take down that image, cause you are technically violating the intel license agreement by not having a click-through for it Dec 04 13:28:57 (assuming it contains the ixp modules) Dec 04 13:29:28 Down. I will change the note on the nslu2 pages. Dec 04 13:29:35 when do you think you will have serial ? Dec 04 13:29:37 thx Dec 04 13:31:27 I haven't planned it. Dec 04 13:34:01 it will be much easier for you to debug with serial Dec 04 13:34:43 The problem is that i don't have the equipment or experience for soldering. Dec 04 13:35:14 nobody around that can do it for you? Dec 04 13:37:26 Maybe i could ask a colleague but at this moment he is moving to a new home. Dec 04 13:38:16 I don't think he has time for it now. Dec 04 13:41:11 I think you might have changed the wrong defconfig. Dec 04 13:42:11 You changed the one in packages/linux/nslu2-linksys-kernel-2.4.22/nslu2/ right? Dec 04 13:42:14 ./openembedded/packages/linux/nslu2-linksys-kernel-2.4.22/nslu2/defconfig Dec 04 13:42:29 Should be the one in packages/linux/unslung-kernel/ Dec 04 13:44:52 Your rigth. Dec 04 13:45:21 The current config is the unmodified "packages/linux/unslung-kernel/defconfig". Dec 04 13:45:38 right Dec 04 13:48:13 How stuppid. Thats the one with the patches. Why did i change the other one? Dec 04 13:48:50 I eventually want to remove that other nslu2-linksys-kernel dir ... Dec 04 13:49:11 It's never used and it just adds to confusion Dec 04 13:51:55 ok, building now ... Dec 04 13:52:12 One of the hunks failed. Dec 04 13:52:19 But that's the console line Dec 04 13:55:39 ok, I have to build from scratch, so it won't get up to the kernel for a good hour or maybe two. I'll report the results here then Dec 04 13:56:21 Well i have patched the right version and are commting that to the branch Dec 04 13:57:08 Well. It take me hours to build so i won't complain. Dec 04 13:57:31 I am not sure if i am still on the channel then. Dec 04 13:57:44 I will leave in two and a half hour. Dec 04 14:04:59 Good morning gentlemen - I'll give the new unslung a try; I assume I need to do something with a config file to cause the R63 flavor to be built? Dec 04 14:05:53 Well it's more then a the config file. Dec 04 14:07:06 mwester: I'm building now. if it is successful, then I will commit the changes to the main branch. Dec 04 14:07:57 Ok. Once committed will that become the default in monotone? Dec 04 14:08:37 yes Dec 04 14:09:00 well, I will make sure it boots before I change it in unslung.conf Dec 04 14:09:17 PREFERRED_PROVIDER_virtual/kernel ?= "unslung-kernel" Dec 04 14:09:17 PREFERRED_VERSION_unslung-kernel ?= "2.4.22.l2.3r63" Dec 04 14:09:17 PREFERRED_VERSION_unslung-rootfs ?= "2.3r63" Dec 04 14:09:17 PREFERRED_VERSION_nslu2-linksys-firmware ?= "2.3r63" Dec 04 14:09:17 PREFERRED_VERSION_nslu2-linksys-libs ?= "2.3r63" Dec 04 14:09:19 PREFERRED_VERSION_nslu2-linksys-ramdisk ?= "2.3r63" Dec 04 14:09:39 that is what you will need to do in unslung.conf to test it. Dec 04 14:10:20 Thank - noted! I'll give it shot. Dec 04 14:11:01 I'll check in the other stuff now, so you can test in parallel - do you have serial ? Dec 04 14:11:19 Yes Dec 04 14:14:03 mwester: pushing other changes now - they will have no effect without the changes to the conf file as above Dec 04 14:14:42 Hmm - I'll need to bring devslug back into service so I can test this new kernel and rootfs ... Dec 04 14:15:12 I checked the kernel build and it still works with the patched config. Dec 04 14:19:49 mwester: changes are pushed. there is something wrong with ciabot reporting from our monotone server. Dec 04 14:20:50 thanks all - I'll update and start building... :) Dec 04 14:23:14 this will be called Unslung 6.1-alpha Dec 04 14:23:52 rwhitby, not sure what is happening to me today, rebuilt from scratch yesterday, made a few small changes, committed them, pushed to monotone.nslu2-linux.org; however these do not show up in CIA Dec 04 14:24:03 aren't they pushed upstream? Dec 04 14:24:26 I also saw that currently there are two heads, but didn't dare yet to do a merge... Dec 04 14:25:04 ah, is this rleated to your msg to mwester from a few mins ago? Dec 04 14:30:55 I have just fixed CIA, so future messages should show Dec 04 14:31:20 koen's server pulls from us periodically, we don't push upstream Dec 04 14:32:13 ah ok, would it be ok to do an mt merge now ? the other head is in a completely different place. However, for now I'd rather take extreme caution Dec 04 14:32:20 don't want to mess things up again. Dec 04 14:32:40 also would appreciate it if you can verify if my changes are properly entered this time Dec 04 14:32:45 just to be sure Dec 04 14:33:17 do you use the master makefile targets to do your pulls and pushes? Dec 04 14:33:29 (i.e. "make push-openembedded") Dec 04 14:34:57 nope, fully unaware of those, when I started with this koen told me to do mt commit and mt push Dec 04 14:37:32 if you use the makefile targets, then it does the correct merge for you Dec 04 14:37:55 it is impossible to break things if you only use the makefile targets Dec 04 14:38:28 ah, ok. learned a new thing today Dec 04 14:39:12 better add it to that wiki page :-) Dec 04 14:39:34 yessss Dec 04 14:40:26 where is the page you are writing? Dec 04 14:40:54 http://www.nslu2-linux.org/wiki/Development/Development101 Dec 04 14:41:03 if you have a better name, let me know Dec 04 14:41:27 also any comment or improvement is of course apprecaited. Koen already went over it once Dec 04 14:42:20 ok, add the makefile targets Dec 04 14:42:36 it does only contain mt and bb info for now Dec 04 14:42:54 and require that developers use those for all pushes, pulls and updates Dec 04 14:43:18 (just the monotone related makefile targets, not the build targets as they are on a different page) Dec 04 14:44:46 ok Dec 04 14:46:25 saw you merged the head Dec 04 14:46:49 is update-openembedded the preferred command to sync to the head? Dec 04 14:47:14 yes, or just update Dec 04 14:51:38 eFfeM: org.openembedded.nslu2-linux no longer exists as a branch, so it should be removed from the wiki page Dec 04 14:53:30 ok, done, (but it is still listed if I do mt list branches Dec 04 14:54:04 that must be in your own local database, cause it shouldn't be on the server Dec 04 14:54:56 monotone.nslu2-linux.org will not serve up org.openembedded.nslu2-linux (intentionally, to try and get rid of that branch) Dec 04 14:55:37 puzzled. I started from scratch yesterday, in a clean dir, fetched the master makefile, only copied the downloads dir over, so I would expect this branch to be in the copy I fetched Dec 04 14:56:04 so I suspect it is in the server db (at least yesterday) Dec 04 14:56:25 hmm - ok, I will look at it Dec 04 14:56:25 does your mt list branches not list it? Dec 04 14:56:31 gr8 Dec 04 14:56:46 anyway it is away from the wiki Dec 04 14:57:45 Hmm - yes, my list branches does list it. We need to remove it from the database on monotone.nslu2-linux.org which is uploaded to sources.nslu2-linux.org Dec 04 14:57:48 calling it a day for now, spent a lot of time last night recovering from a slug crash (the current head does not seem to be as stable as the previous version I had) (not that I'm complaining) Dec 04 14:58:08 yep, that's why it's the head :-) Dec 04 14:58:11 rwhitby-away, this seems outside my capabilities or knowledge Dec 04 14:58:24 LOL, I knew you were going to say that .... Dec 04 14:58:24 eFfeM: I will be handling that Dec 04 14:58:29 ok Dec 04 15:34:10 flashing unslung 6.1-alpha now ... Dec 04 18:21:01 03rwhitby 07org.openembedded.dev * r9ed03027... 10/packages/linux/unslung-kernel_2.4.22.l2.3r63.bb: unslung-kernel: Removed modification of kernel (left-over from the binary version) Dec 04 19:32:40 mwester: seems that the unslung kernel is now too big Dec 04 19:32:47 will have to make something into modules Dec 04 19:50:03 :o Still building here, don't know why it's so slow today. What's left to turn into modules? Dec 04 22:25:21 mwester: I think it's all the codepages which have increased the size Dec 04 22:27:01 That could be, but the image still fits in flash as far as I can see Dec 04 22:28:44 kernel boots, can't find a RAM disk image starting at 0, or so it claims. Dec 04 22:30:14 oh, did you get the latest change? Dec 04 22:30:34 if it's complaining about /dev/slug, then you don't have the latest change Dec 04 22:32:13 Probably not! I shall do so, and rebuild. In any case, is there anything significant to turn into a module to claim some space back? Dec 04 22:32:30 well, maybe those code pages can be modules ... Dec 04 22:32:52 the reason why the stock kernel fits is cause it doesn't have jffs2 support. Dec 04 22:38:47 I expect that all the obvious data structures have been trimmed down already? Dec 04 22:39:04 which ones do you mean? Dec 04 22:39:37 I seem to recall tweaking numbers of buffers, max users, stuff like that back on old UNIX kernels. Dec 04 22:42:27 trying now with all codepages (which is the only new stuff) as modules ... Dec 04 22:49:36 What changed recently? make build_unslung didn't build anything after the make update -- so either I have the latest, or I need to remove some stamps to force a rebuild Dec 04 22:50:37 oh, the only thing that changed was the .bb file, but I forgot to bump the PR, so you need to rebuild the kernel Dec 04 22:51:09 rm -rf unslung/tmp/{work,stamps}/unslung-kernel* Dec 04 22:53:17 Thanks. That'll keep it busy for a bit. Dec 04 23:04:16 flashing now ... Dec 04 23:10:56 :D Well now, that's the sorta thing that should happen when a kernel's too big. Fails to build - and even tells me why. Not bad... Dec 04 23:11:38 pushing the fixes now ... Dec 04 23:15:58 03rwhitby 07org.openembedded.dev * r6217b755... 10/packages/linux/unslung-kernel/defconfig: unslung-kernel: Changed the defconfig to make all the new codepages as kernel modules. Dec 04 23:19:22 Hmm: "/bin/mount: error while loading shared libraries: libgcc_s.so.1: cannot open shared object file: No such file or directory" Dec 04 23:22:34 Well, progress anyway. Dec 04 23:24:45 libgcc_s.so.1 is missing execute privs on my rootfs, but it most certainly is present. Dec 04 23:27:18 does it appear in unslung/tmp/rootfs/lib/ on your build machine? or just on your existing unslung 5.5 rootfs Dec 04 23:29:39 on build machine. Dec 04 23:29:48 flashing... Dec 04 23:31:53 hmm - only 88 blocks left in the flashfs Dec 04 23:32:21 and adding libgcc would be 58 blocks ... Dec 04 23:33:13 Booted up and running. Dec 04 23:35:10 at least without drives, it's up and running Dec 04 23:42:40 how much space do you have in the flash? Dec 04 23:43:14 and what version of gcc is in unslung/tmp/work/gcc-* on your build machine? Dec 04 23:44:57 cause I don't think we used to require libgcc in the rootfs ... Dec 04 23:46:31 gcc-cross-3.4.4-r3 Dec 04 23:46:57 72 blocks left in flash after gcc lib is present Dec 04 23:47:48 I need to find an old version of unslung's nslu2.conf to see what version of gcc we used to use ... Dec 04 23:48:41 It boots, but does not start the httpd process. When adding a drive, it gets the ports backwards so that sdb1 is on the wrong port... I think there are other probs too, but it clearly runs. Dec 04 23:49:05 do you have the console boot log ? Dec 04 23:49:13 can you put it on pastebin ? Dec 04 23:49:38 I can get it; haven't used pastebin, I'll give it a shot. Dec 04 23:53:04 thx Dec 04 23:56:12 http://pastebin.com/449263 Dec 04 23:57:31 "/etc/rc: cannot create /etc/mtab: No space left on device" Dec 04 23:58:12 So we'll need to make some space somehow ... Dec 04 23:58:36 ? How big is mtab, though? Dec 04 23:58:41 1 block? Dec 04 23:59:02 jffs2 requires at least one erase block (128kb) free to work properly Dec 04 23:59:15 and it's best to have at least two (256kb) free Dec 04 23:59:27 Gulp! That's a lot... Dec 05 00:00:20 libgcc is only 58kb -- must be the modules for the code pages? Dec 05 00:01:30 I'm thinking that we're using a later version of gcc (which is causing the libgcc requirement) and it's making bigger exe's or something ... Dec 05 00:01:47 the modules for the code pages are not included in the rootfs Dec 05 00:02:36 I just rm'd the serial and net usb modules, boot is now continuing. Dec 05 00:03:30 Booted, web server started. Dec 05 00:08:19 ok, I'm going to try again from scratch Dec 05 00:12:56 I compared to a running 23 unslung - the new /lib contains lowcase.dat, upcase.dat, and valid.dat for a total of 320KB -- a spot check of the libraries seems to match in terms of sizes. Dec 05 00:13:14 -rw-r--r-- 1 root root 131072 Dec 31 1969 lowcase.dat Dec 05 00:13:22 -rw-r--r-- 1 root root 131072 Dec 31 1969 upcase.dat Dec 05 00:13:27 -rw-r--r-- 1 root root 65536 Dec 31 1969 valid.dat Dec 05 00:15:03 :) I need some sleep; I'll work on this a bit more tomorrow! Dec 05 01:07:08 mwester: good find. We'll have to get rid of them and put them in an installable ipk Dec 05 01:34:03 I got libgcc that time ... Dec 05 02:56:48 Flashed, removed /lib/*.dat, rebooted, unslung. Looks like 6.1 works if /lib/*.dat is not there. **** ENDING LOGGING AT Mon Dec 05 02:59:56 2005