**** BEGIN LOGGING AT Thu Sep 13 03:00:00 2012 Sep 13 09:04:47 morning ant_work Sep 13 09:05:20 ant_work: strangely, I couldn't get that kernel to flash - the LEDs just went off quickly Sep 13 09:05:29 maybe I should try another card Sep 13 09:45:58 hi, it's too big! Sep 13 09:46:27 btw there is still hope Sep 13 09:46:30 which, the card or the kernel? Sep 13 09:46:50 khem's patch for kernel compiler is in Saul queue today Sep 13 09:46:52 mised it Sep 13 09:47:32 that 2.6.31 kernel is way too big, 1,9 MiB iirc Sep 13 09:47:42 has to be kexec'ed Sep 13 09:48:06 ah ok, right Sep 13 09:48:21 I guess you were trying to check if my device was borked? :) Sep 13 09:59:43 heh Sep 13 10:00:09 really, Jay7 had a readable boot log. With more debugging that could rule out the doubts Sep 13 10:00:39 besides that, about the 'fixes' for armv4 the confusion is infernal :) Sep 13 10:00:56 ant_work: I'm busy these days Sep 13 10:01:07 let's hope Khem's patch goes in Sep 13 10:01:10 but Sat should be free enough Sep 13 10:01:10 hi Jay7 Sep 13 10:01:14 hi ppl Sep 13 10:02:10 bluelightning: in any case it's worth to insist toward 3.6 Sep 13 10:02:52 hi Jay7 Sep 13 10:02:57 ant_work: ok Sep 13 10:03:25 btw has anyone used kexecboot as a serious replacement for something like grub on x86? Sep 13 10:03:44 I'm not a huge fan of grub especially with what they've done to it in grub2 Sep 13 10:05:34 iirc Peter Chubb / tiny project Sep 13 10:05:44 but maybe only on arm Sep 13 10:08:32 just kind of seems like everything has to be shoehorned into the bootloader eventually, especially filesystem support Sep 13 10:08:56 surely it's easier to just rely on Linux to support those things... Sep 13 11:22:12 regarding the transition towards 3.4 or 3.6 kernel series: Sep 13 11:23:12 every kernel past 3.3-rc1 will outright fail to boot Sep 13 11:24:01 I have a fix for that works for pxa2xx based zaurii devices Sep 13 11:24:21 collie is still untested, it will be when jay7 finds some free time Sep 13 11:24:26 there are other bugs Sep 13 11:25:09 for instance, you cannot have a working ads7846 touchscreen driver and a working pxamci driver at the same time Sep 13 11:25:35 the problem is the ads7846 driver requires a regulator Sep 13 11:25:43 which zaurii devices don't have Sep 13 11:25:57 that means we have to enable CONFIG_REGULATOR_DUMMY Sep 13 11:26:04 this works for ads7846 Sep 13 11:26:16 however this option will break the pxamci driver Sep 13 11:26:45 since it's regulator support is written in such a way that it expects a working regulator or no regulator at all Sep 13 11:27:12 and if you enable CONFIG_REGULATOR_DUMMY it will break since it doesn't handle this situation Sep 13 11:27:52 i have a patch for that too, i posted in on ALKML, got no response so far Sep 13 11:28:07 probably because i CC-ed only Eric Sep 13 11:28:39 also, trying to remove the pxa2[57]x_udc driver will oops the kernel Sep 13 11:28:55 this is due to broken usb transceiver support and a NULL pointer dereference Sep 13 11:29:02 this is also a trivial fix Sep 13 11:30:22 my point being, if you want to move to newer kernels, you will need these patches in meta-handheld Sep 13 11:32:02 and these are only the bug i encountered so far Sep 13 11:32:10 *bugs Sep 13 11:38:27 hi dromede Sep 13 11:38:43 yes, I do remember your 3.6 patch for collie Sep 13 11:56:15 hi Sep 13 11:56:42 as for the pm bug, this is beyond me Sep 13 11:56:54 i'm gonna need some help with that one Sep 13 11:57:15 even if i completely disable the AC_IN interrupt, the kernel will still fail to wake up from suspend Sep 13 11:57:40 i just can't explain that Sep 13 12:00:35 can't help you in that level, sorry Sep 13 12:01:01 still trying to get a partition parser for Z Kernel Sep 13 12:01:19 I've talked with a polish dev working on BCM and doing similar things Sep 13 12:02:03 in fact he did commit a partition parser as kernel module Sep 13 12:02:26 yeah i saw that Sep 13 12:03:14 at first sight seems we'd just need a scaled-down driver like its one. Read only Sep 13 12:04:40 once we have the structure (see kexecboot userspace code) we just pass it to the platform driver and on init all is magically set Sep 13 12:05:02 well, you have to remove the hatdcodings done in machine conf Sep 13 15:14:23 bluelightning: stil lwaiting for kernel.bbclass, module-base.bbclass: Use CC to form KERNEL_CC Sep 13 15:14:49 dromede: I think this evening I'll test ZRom on 3200/terrier Sep 13 15:15:06 (while building kernels...;) Sep 13 15:15:13 great! Sep 13 15:15:16 ZGrom even Sep 13 15:15:47 ant_work: if it's been more than a few days, just ping it... Sep 13 15:15:47 now there's 4 testers Sep 13 15:16:19 bluelightning: Saul sent it yesterday as RFC. Maybe this evening Sep 13 15:16:39 dromede: do we have the links in the logs here? Sep 13 15:16:53 ant_work: oh, in that case it wouldn't be merged immediately... Sep 13 15:17:02 no, the links are sent privately for now Sep 13 15:17:05 just a minute... Sep 13 15:17:09 [OE-core] [CONSOLIDATED REQUEST 00/10] Misc Fixes (cover only) Sep 13 15:17:15 sounds harmless Sep 13 15:17:53 ant_work: that's not an RFC then... Sep 13 15:18:35 heh Sep 13 15:19:03 it's a CR Sep 13 15:19:11 ^_^ Sep 13 15:20:26 dromede: any other fixes apart slob/slab and re-enabling cpufreq in your 3.2 kernels? Sep 13 15:21:07 ah, kemaps Sep 13 15:21:16 custom keymapd iirc Sep 13 15:21:57 just slob/slab as far as i remember Sep 13 15:22:06 and the kernel size maybe Sep 13 15:22:21 mine is 1.5 the yocto kernel is 2.0 Sep 13 15:22:24 quite large Sep 13 15:22:41 I'm working to make it in fragments Sep 13 15:22:51 how about i post my defconfigs for the kernels i released with ZGrom, would that help? Sep 13 15:22:53 we need 3.4 though, where the framework is complete Sep 13 15:23:28 dromede: we took one spuitz config and extended to other machines for conformity Sep 13 15:24:38 now we'd like to have snd, usb, *stuff modules well defined in text files = fragments Sep 13 15:26:16 and we can play with KERNEL_FEATURES_append Sep 13 15:26:31 see http://cgit.openembedded.org/openembedded-core/tree/meta/recipes-kernel/linux/linux-yocto_3.4.bb Sep 13 15:31:02 bbl Sep 13 20:03:54 dromede: downloading image Sep 13 20:04:19 dromede: I think I'll install it on spitz internal hdd Sep 13 20:13:24 it shouldn't matter, anywhere where kexecboot can find it Sep 13 20:19:42 I plan to upload my mame roms collection Sep 13 20:19:49 there Sep 13 20:20:08 do you happen to remember which romset is expected? Sep 13 20:21:37 emulators are placed in /home/root/emulators Sep 13 20:21:50 the mame4all directory contains complete documentation Sep 13 20:22:07 hopefully is the same set I've on ipad Sep 13 20:22:27 0.37b5 Sep 13 20:22:43 mame4all was very lightly tested Sep 13 20:23:14 I have to rebuild one set for 0.139u1 Sep 13 20:23:26 (MAME4IOS Reloaded) Sep 13 20:23:29 when you run it, gmenu2x will open a file selector Sep 13 20:23:41 ah, nice Sep 13 20:23:50 use it to navigate to your roms dir and then select any file in that dir Sep 13 20:24:01 mame will then open that dir and parse every rom Sep 13 20:24:19 this is a bit strange, gmenu2x will let you select files but not directories Sep 13 20:24:20 ok, I suppose all will just works like on akita Sep 13 20:24:32 it should Sep 13 20:24:41 the only thing that i'm worried about is the kernel Sep 13 20:24:59 nah, i'ts just one defconfig line... Sep 13 20:25:04 as this is the only thing different Sep 13 20:25:07 yes i know Sep 13 20:25:10 but still... Sep 13 20:25:13 one never knows Sep 13 20:25:15 he he Sep 13 20:25:21 righto Sep 13 20:25:35 d/l at 63% now Sep 13 20:25:44 damn rapidshare Sep 13 20:25:48 should've used mediafire Sep 13 20:26:03 or github :) Sep 13 20:26:25 can github do binary uploads? Sep 13 20:26:40 i thought you can only push git repo's Sep 13 20:26:51 https://github.com/andrea-adami/z-prom Sep 13 20:27:41 ah Sep 13 20:27:56 it never occured to me that i could place tarballs in git repo's ;) Sep 13 20:28:15 same here but I hate those public hosting sites Sep 13 20:28:23 so I tried :) Sep 13 20:28:59 maybe there is some sort of size-limit, dunno Sep 13 20:29:06 hopefully any Sep 13 20:31:38 79% Sep 13 20:32:16 flying at ~30kb/s :) Sep 13 20:32:19 "Romsets have to be MAME 0.37b5 ones (July 2000)." Sep 13 20:32:23 quote from the readme Sep 13 20:33:15 fine, seems same as imame4all Sep 13 20:33:33 and MAME4droid Sep 13 20:35:14 dromede: d/l chart is a spaced triangle wave, peaks at 15'KB and 0 Sep 13 20:35:21 *150 Sep 13 20:35:42 seems heavily managed Sep 13 20:35:46 yeah, i read somewhere that rapidshare does that Sep 13 20:36:07 i think it throttles in 10 sec intervals, something like that Sep 13 20:36:33 maybe 12 Sep 13 20:36:58 anyway, almost done at 94% :) Sep 13 20:38:50 got it Sep 13 21:15:47 unfortunately you have packed it under /rootfs Sep 13 21:16:00 need manual intervention :) Sep 13 21:16:18 yes, that was intentional since i have to provide two kernels Sep 13 21:16:48 just make a /kernel dir and all in / :) Sep 13 21:17:05 or even /boot/kernel Sep 13 21:17:39 in that way we can rename it as hdimage1.tgz and flash it with updater.sh Sep 13 21:17:51 and just add a symlink for zImage Sep 13 21:18:31 hmm, i never thought about using updater.sh for installation Sep 13 21:18:43 zImage -> mtd1 Sep 13 21:18:54 initrd.bin -> mtd2 Sep 13 21:19:01 home.bin -> mtd3 Sep 13 21:19:11 plus hdimage1.tgz for spitz Sep 13 21:19:25 check updater sources Sep 13 21:21:06 it seems much simpler to me to have the entire rootfs in a single partition Sep 13 21:28:26 it all ends in /rootfs.. seems a bit unconvenient for me Sep 13 21:28:58 anyway, almost ready to boot Sep 13 21:39:15 hm.. krnel panic Sep 13 21:41:32 of course it can't be simple Sep 13 21:41:35 :/ Sep 13 21:41:45 is ext3 compiled in kernel? Sep 13 21:41:51 I'm booting from CF Sep 13 21:42:12 No init found Sep 13 21:42:39 yes, both ext3 and ext2 are included Sep 13 21:43:14 any mention of "unknown block" or something like that? Sep 13 21:43:17 let's retry with an older card Sep 13 21:43:28 this was SanDisk 4GB Sep 13 21:44:06 i do remember making ext3 as a module Sep 13 21:44:12 and then i reverted that change Sep 13 21:44:25 but it could be that i didn't recompile the spitz kernel Sep 13 21:45:17 I'll reformaton ext2 then Sep 13 21:53:23 well, now I have ext2_lookup errors Sep 13 21:53:40 maybe my cards did die during my holidays :p Sep 13 21:54:27 damn Sep 13 21:54:32 got any sd cards? Sep 13 21:54:32 is the kernel overclocked? Sep 13 21:54:34 no Sep 13 21:54:44 now it was a 1GB SD, tested on other Z Sep 13 21:55:38 sd cards will also work, the kernels have CONFIG_MMC_UNSAFE_RESUME=y Sep 13 21:56:19 i'll test my cf card with ext2 Sep 13 21:56:26 i have an APACER 1GB Sep 13 21:59:11 argh, .. wrong file permissions Sep 13 21:59:40 http://paste.debian.net/189738/ Sep 13 22:00:14 sudo chown -R root * Sep 13 22:01:07 i just did a fresh rootfs on a cf ext2 card Sep 13 22:01:10 it works fine Sep 13 22:01:17 and i didn't have to change permissions Sep 13 22:04:29 http://paste.debian.net/189740/ Sep 13 22:05:44 my bad, forgot -p in tar -xjfv Sep 13 22:07:36 hm, no... Sep 13 22:08:25 what now? Sep 13 22:09:35 still kernel paniuc Sep 13 22:10:56 Attempt to kill init...you know... Sep 13 22:11:10 ant__: indeed I do :( Sep 13 22:11:20 i would need the entire output before i know what the problem is Sep 13 22:11:21 :D Sep 13 22:11:39 could you enable serial console and grab the output? Sep 13 22:11:53 bluelightning: should I try one more time with kernel.bbclass patch for collie? Sep 13 22:15:09 ant__: was there a problem with the last one? Sep 13 22:15:22 oh maybe I am confused :( Sep 13 22:17:43 dromede: http://paste.debian.net/189742/ Sep 13 22:18:27 bluelightning: http://patches.openembedded.org/patch/36399/ Sep 13 22:21:38 dromede: OE/Yocto built images have right root:root persmissions Sep 13 22:22:00 smthg is wrong there , I get strange id's Sep 13 22:22:37 dromede: I've noticed you fixed audio as well, just missing modules.. Sep 13 22:22:47 the rootfs is built with OE/Yocto Sep 13 22:23:12 you have wrong permissions in the tar image then Sep 13 22:41:22 i dont understand "strange id's", what exactly do you mean by that? Sep 13 22:43:08 bluelightning: andrea@mizar /oe/oe-core $ patch -p1 rebuilding now Sep 13 22:43:21 dromede: one mom Sep 13 22:43:30 no rush Sep 13 22:44:19 you know, desesperate try Sep 13 22:44:50 ? Sep 13 22:45:12 i'm compiling qemu-native, gonna try testing the spitz kernel that way Sep 13 22:46:00 kernel seems ok Sep 13 22:46:15 works on a different rootfs? Sep 13 22:48:56 about UID and GID, here andrea is 1000 and ssmtp is 1000 Sep 13 22:49:12 so I guess dromede is 1000 as well Sep 13 22:50:47 while I'd expect 0:0 for a rootfs Sep 13 22:52:01 for the skeleton Sep 13 22:52:18 yes, uid is 1000 Sep 13 22:52:25 but still, it work on my machine Sep 13 22:52:46 could you try chowning the entire rootfs to root, just to be sure ? Sep 13 22:53:28 bluelightning: pls try /collie/zImage-14-09.bin Sep 13 22:57:03 dromede: same after chown Sep 13 22:57:11 I'll try another rootfs now Sep 13 22:57:28 then the permission aren't the problem Sep 13 22:58:04 i would have been happier if they were :) Sep 13 23:01:34 it's almost the same failure as collie :/ Sep 13 23:02:05 I've tried with init=/bin/sh and it still paniced Sep 13 23:02:43 this is bas Sep 13 23:02:46 *bad Sep 13 23:02:49 very bad Sep 13 23:02:58 I'll take a log now Sep 13 23:03:38 ouch .. /bin/bash here :) Sep 13 23:04:12 yes, bash does that Sep 13 23:04:14 hm, the symlink is there... Sep 13 23:04:27 bash will overwrite the busybox sh symlink Sep 13 23:05:15 here sh points to bash Sep 13 23:05:42 I'm used to use klibc (da)sh Sep 13 23:05:54 anyway Sep 13 23:06:14 it doesn't matter, it shouldn't panic like that Sep 13 23:06:29 have you tried my kernel on a different rootfs? Sep 13 23:07:30 I don't have images for spitz here, let me search Sep 13 23:07:54 or even better, just try the linux-yocto kernel on my rootfs Sep 13 23:08:13 heh, I'm still building for armv4 Sep 13 23:08:29 I'll do tomorrow pxa builds Sep 13 23:09:06 btw framebuffer is 'scrambled' Sep 13 23:09:06 just a thought, my oe-core is setup for akita machine type, could this be the problem? i doubt it, but still Sep 13 23:09:23 some packages will differ Sep 13 23:09:37 scrambled framebuffer? Sep 13 23:09:57 yes, color pixel texture Sep 13 23:10:13 rotation is good Sep 13 23:10:48 maybe it was crashed during fb refresh :) Sep 13 23:11:12 ZGrom comes with psplash, maybe that's what caused it Sep 13 23:12:14 psplash is working beautifully in oe-core images for Z Sep 13 23:12:40 yes, that's why i included it :) Sep 13 23:16:59 I'll use another kernel from ubi image in nand Sep 13 23:19:44 just to be sure, i downloaded the rapidshare link again Sep 13 23:19:47 and it works fine Sep 13 23:23:48 i got qemu up and running Sep 13 23:23:54 and the spitz kernel + rootfs work fine Sep 13 23:24:32 dromede: http://paste.debian.net/189749/ Sep 13 23:24:42 now I'll let it boot the rootfs Sep 13 23:27:28 yes Sep 13 23:27:35 I'm at ts-calibrate Sep 13 23:28:19 hm Sep 13 23:28:25 i'm at a loss Sep 13 23:28:41 the kernel and the rootfs from the rapidshare link work fine under qemu Sep 13 23:28:41 now... FATAL: Module snd-pcm-oss not found Sep 13 23:28:53 that's normal the modules are for 3.2.24 Sep 13 23:29:16 anyway seems you'll have to rebuild the kernel Sep 13 23:29:37 just a rebuild? no changes? Sep 13 23:30:51 can't say, just remove that CONFIG_GPIO_PCA953X=y Sep 13 23:31:28 and build for machine = spitz Sep 13 23:32:29 ooooooohhhhh Sep 13 23:32:31 :D Sep 13 23:32:36 forgot about that Sep 13 23:33:11 both kernel were built with akita/borzoi/spitz enables Sep 13 23:33:14 *enabled Sep 13 23:33:42 i disabled akita support for spitz kernel, lets see if that does the trick Sep 13 23:33:49 akita in OE was long broken because of the missing ^^ Sep 13 23:33:58 (my bad :p) Sep 13 23:34:49 hmm, no Sep 13 23:34:52 actually no Sep 13 23:35:15 spitz defconfig had CONFIG_MACH_AKITA=y Sep 13 23:35:15 CONFIG_MACH_SPITZ=y Sep 13 23:35:15 CONFIG_MACH_BORZOI=y enabled Sep 13 23:35:23 that is correct right? Sep 13 23:35:58 ant__: no difference with that latest kernel image Sep 13 23:36:46 dromede: is ok Sep 13 23:37:35 bluelightning: I somehow had that feeling ;) Sep 13 23:37:53 collie still fails after init? Sep 13 23:38:11 seems so Sep 13 23:38:34 https://rapidshare.com/files/3414140209/zImage Sep 13 23:38:41 try this one and let's call it a night Sep 13 23:40:49 booting Sep 13 23:40:55 it works? Sep 13 23:41:26 no, same panic Sep 13 23:41:31 :/ Sep 13 23:41:51 dammit Sep 13 23:41:51 I'll try on other SD/CF cards, I have just to find them Sep 13 23:42:11 ok, tomorrow i'll send you a kernel built with meta-handheld defconfig Sep 13 23:42:17 but I doubt it's that Sep 13 23:42:41 ok, thx Sep 13 23:42:51 time to sleep Sep 13 23:42:53 gn Sep 14 00:00:12 gn **** ENDING LOGGING AT Fri Sep 14 02:59:59 2012