**** BEGIN LOGGING AT Sat May 26 02:59:58 2012 May 26 04:11:17 nerdboy! It works, son! May 26 04:12:01 ${TERMCMDRUN}=${XTERM_CMDRUN}; ${TERMCMD}=${XTERMCMD}, I mean. May 26 04:12:09 Now on to breaking the kernel! May 26 04:15:06 Oops, I must be tired. I mean TERMCMDRUN = ${XTERMCMDRUN} May 26 04:15:48 Wait, not even that (no way to C-&-P): May 26 04:16:29 TERMCMDRUN = "${XTERMCMDRUN}"; TERMCMD = "${XTERMCMD}" May 26 04:17:04 TERMCMD = ${XTERM_TERMCMD} maybe? May 26 04:18:04 Oh yeah, the typing part of my brain seems to frozen. May 26 04:18:34 I'm so overjoyed by the announcement of http://lwn.net/Articles/498877/ , which I knew was coming, that I can hardly bear it! May 26 04:20:30 then you should be ecstatic about the ARM camera project May 26 04:20:39 and we're hiring... May 26 04:29:32 ARM camera project sounds interesting. I worked on a colonoscopy camera with RTSP on gumstix once. May 26 04:29:45 Thank goodness. no humans were involved except for me. May 26 09:21:46 Alison_Chaiken: arthrex.com if you're interested May 26 12:44:11 :-) May 26 13:29:06 hi otavio May 26 13:31:09 woglinde: :) how are you doing? May 26 13:32:13 fixing icedtea builds May 26 13:32:43 for gcc-4.7 May 26 13:33:40 nice! May 26 13:34:07 ahh May 26 13:34:15 I have a pending patch for you May 26 13:34:16 hold May 26 13:35:27 http://paste.debian.net/171250/ May 26 13:35:49 woglinde: this is a gcc-4.7 I did locally but didn't yet cleaned up to send May 26 13:36:42 forget llvm so far May 26 13:36:52 doesnt work May 26 13:38:21 Ok :) May 26 13:38:33 how are you doing it? May 26 14:02:45 how do you mean it? May 26 14:03:03 and I am uncertain to support openjdk6 further May 26 14:03:16 want to move to 7? May 26 14:03:31 I already did May 26 14:03:44 look at the icedtea7-branch May 26 14:03:52 so merge it :P May 26 14:04:02 I would May 26 14:04:10 but you need my relocation patch May 26 14:04:19 ? May 26 14:04:22 otherwise chrpath will not work correctly May 26 14:05:19 look for this patch on oe-core patchqueue May 26 14:05:21 relocatable.bbclass: do not erease already existing ORIGIN paths and sanity resulting rpath May 26 14:06:26 I see May 26 14:06:31 did people commented on it? May 26 14:06:39 yes May 26 14:07:00 but I didn't have to time to answer proberly May 26 14:07:20 please try to; so it can go to denzil too May 26 14:08:13 you could try too May 26 14:32:28 i don't have the full background of the patch ... May 26 14:32:41 and didn't yet even try it ... May 26 14:32:48 i am not a good one to push for it May 26 16:59:06 moin May 26 17:19:08 someone using repo to manage multiple git trees for oe? May 26 19:35:49 bluelightning, are you around ? May 26 19:35:57 lautriv: hi, yep May 26 19:36:53 bluelightning, hello ;) i did a manual change of 3.4 release because the patches for -rc won't match, anything but one part went fine...... May 26 19:37:37 lautriv: all of those patches seeemed to apply for me... May 26 19:38:30 bluelightning, i had one broweser open to show you where i miss something. unfortunately it's closed and i must dig a bit ... May 26 19:39:04 bluelightning, you patched some 3.4-rc but not release, i assume ? May 26 19:41:44 ... found it May 26 19:42:18 bluelightning, here --> https://github.com/bluelightning/meta-handheld/blob/paule/hx4700-update/recipes-kernel/linux/linux-3.4/hx4700/0003-mfd-asic3-Add-PCMCIA-CF-support.patch the lines from 19 to 27 has no match May 26 19:54:57 lautriv: I can only say, when I applied them via bitbake in the order they are listed in the recipe, there were no errorts May 26 19:55:01 errors May 26 19:55:06 this is with 3.4 release May 26 19:56:18 bluelightning, that can be but i did it manually and a bunch beginning with 0001 however for the given part is only one occurence, regardless of the order of patches. May 26 19:57:00 well, I'm not sure what to sugges May 26 19:57:01 t May 26 19:57:07 I'm not the author of the patches... May 26 19:57:26 bluelightning, however, even without PCMCIA-CF there is a more important thing i would have your opinion. May 26 19:58:44 i found a new git-repo for acx100 which is considered to be not for production, where the old acx100 is obviously abandonded. May 26 20:00:01 tested the source on a kernel for x86_64 and it compiled. but nott on arm, where the caveeat seems to be /proc ...... is there something special/different for arm ? May 26 20:00:37 lautriv: not that I'm aware of, but I'm no expert May 26 20:03:56 bluelightning, from my point of view it should not matter what arch but this source defines a strucht depending on some preprocessor decisions, where the matching rule is "CONFIG_PROC_FS && ACX_NEEDS_PROC_ANYWAY" now, if i CONFIG_PROC_FS which is rather usual, i run into implicit declaration, if i define the latter i get a redefinition. looks like a chicken-egg problem in this code. May 26 20:04:43 lautriv: I guess if I was hitting this I would probably mail the acx100 devs May 26 20:05:15 they're much more likely to understand the implications of that code... May 26 20:05:51 bluelightning, sure, just would have another opinion, at least because the same code works on non-arm. May 26 20:10:03 bluelightning, ok, now a rather simple thing and the last for now ;) gnutoo mentioned a patch to correct the alignment of kernel-code because it was not possible to produce a working kernel on oe/hx4700 last time. may i assume a git-pull;bitbake some-image will actually take that into account ? May 26 20:12:51 nvm, it did not even touch the kernel :( May 26 20:25:10 lautriv: I'd certainly like to know about that patch... gnutoo, what's that about? May 26 20:33:39 ah the alignement patch? May 26 20:33:41 it's a hack May 26 20:34:03 newer toolchain broke kernel booting for old kernels May 26 20:34:19 that hack fixes it May 26 20:34:23 it's not the correct fix May 26 20:34:32 the correct fix is to import the mainline kernel patch May 26 20:34:38 but it requires a bit more work May 26 20:35:39 khem, ping May 26 20:36:13 khem, without ld-is-gold as distro features for arm....it still segfaults/illegal instructions May 26 20:41:40 gnutoo: hmm, well I'm all for getting things working until we get the proper fix... May 26 20:42:56 try the hack then May 26 20:43:02 just try if it fix booting May 26 20:43:10 like locally May 26 20:43:33 gnutoo: what is it though? May 26 20:43:54 sorry I don't think I was around when this was discussed earlier May 26 20:45:19 do you have serial console on theses machines? May 26 20:45:29 I guess not May 26 20:45:44 but I had serial on crespo(nexus S) May 26 20:45:51 bluelightning, if you lsusb does your hx4700 always appear via this Netchip Technology, Inc. Linux-USB Ethernet/RNDIS Gadget ? May 26 20:45:52 and there was alignement issues during boot May 26 20:46:05 during early boot of the kernel May 26 20:46:52 gnutoo: what I mean is, what hack did you apply to fix it in your case? May 26 20:46:58 let me find it May 26 20:47:01 thanks May 26 20:47:58 http://git.freesmartphone.org/?p=linux-2.6.git;a=commit;h=35f4ab25c5e2a4ad0c0cda2d4a87131930e4b8d8 May 26 20:48:24 basically the hack was to add that in the kernel Makefile: -mno-unaligned-access May 26 20:48:33 else it failed very early during boot May 26 20:48:55 and without a serial cable it's impossible to debug May 26 20:49:04 so just try adding that flag to the kernel Maekfile May 26 20:55:59 ok, this image is supposed to give me serial via USB with 115200,8N1 where lsusb tells me ipaq is Linux-USB Ethernet/RNDIS Gadget. modprobing usbserial and ipaq on the host does not care about ttyUSB0 ? May 26 20:56:29 serial console means: May 26 20:56:50 device<---serial---->usb-to-serial<->computer May 26 20:57:06 so it's a real serial port on the device May 26 20:57:10 like the SOC serial port May 26 20:57:24 not some USB or modprobe something May 26 20:57:33 that's already too late for most stuff May 26 20:57:37 gnutoo, module ipaq is meant to connect on the craddles port. May 26 20:57:42 ok May 26 20:57:49 then it should be compiled in the kernel May 26 20:57:55 so you see early boot failures May 26 20:58:19 else it has to mount the filesystem etc.... to load a module May 26 20:58:27 * otavio hates old kernel! May 26 20:58:32 and at that point the kernel is already booted May 26 20:58:42 so you could inject something in the init scripts May 26 20:58:43 gnutoo, image console-image is supposed to have console=ttyUSB0,115200 8N1 May 26 20:58:47 ok May 26 20:58:51 :P fixes bugs all around to you kill them once again ... crap! May 26 20:59:03 ttyUSB0 is for usb serial adapters May 26 20:59:17 and usually the console image takes the serial port from the machine config May 26 20:59:59 gnutoo, THAT is what i mean ...... how else should i get the console ? ( irmember it worked a year ago) May 26 21:00:17 gnutoo, module ipaq is meant to connect on the craddles port. May 26 21:00:25 maybe trough the craddle??? May 26 21:00:31 what is your device? May 26 21:00:47 ipaq something but I don't remember the number off hand May 26 21:01:21 gnutoo, ok, fineprint : iapq->on_the craddle-->via craddles USB-->to the host May 26 21:01:50 what is the model of the ipaq already? May 26 21:01:54 I don't remember off hand May 26 21:02:05 gnutoo, BUT it reports always this etherne-gadged on lsusb. May 26 21:02:25 maybe there is a serial-to-usb adapter in the craddle? May 26 21:02:31 it's an hx4700 once a flavored target May 26 21:03:01 ok May 26 21:03:07 let me look May 26 21:03:16 if there were some converter, it would deny any other use and report its existence May 26 21:03:53 let me lok May 26 21:03:57 *look May 26 21:06:02 sidenote : it may happen the ipaq istself did not load correctly his part of the driver but this display is to small to see anything on CLI and it's touchscreen only. ( furthermore i checked also usbnet which defaults to be on 192.168.7.2 because this usb_ether on the bus but no response) May 26 21:06:57 lautriv, what is the SOC of your device May 26 21:07:07 PXA27x May 26 21:07:15 ok thanks May 26 21:08:06 afaik, they were sold to marvell later on ....... May 26 21:08:17 yes I'm looking at the kernel source May 26 21:10:23 it seems that there is no serial May 26 21:11:03 GNUtoo-desktop, i remember to use it in the past, where this thingie was on 2.6.21 May 26 21:11:13 I looked at 3.4 May 26 21:11:28 so you got boot messages? May 26 21:12:00 like for instance [ 0.000000] Linux version ... May 26 21:12:37 GNUtoo-desktop, i saw a whole bootup with (newish green) ok levels but they are a quarter millimeter in size because framebuffer is 800x600 on CLI :( May 26 21:12:41 brb May 26 21:24:16 back May 26 21:25:26 so last line (as supposed ) hx4700 login : (which should be also on usb/serial) May 26 21:25:32 lautriv, basically your device doesn't boot right? May 26 21:25:41 and it has 3.0 kernel right? May 26 21:26:47 GNUtoo-desktop, nah, boots successful but can't reach it, cdc_ether offline, usbserial somehow not present, commandline ok but no keyboard, wireless off due to bug in oe/acx May 26 21:27:00 ahh ok May 26 21:27:05 so my patch is useless then May 26 21:27:10 I tought it didn't boot May 26 21:27:26 and that you had no way to see what was happening May 26 21:28:08 GNUtoo-desktop, i did a git pull /rebuild and that kernel does not boot but the same sources compiled on a qemu-arm booting fine. May 26 21:29:16 GNUtoo-desktop, i guess it will be the easiest way to take over the media and analyse init. May 26 21:29:34 yes May 26 21:29:47 if the kenrel doesn't boot May 26 21:29:51 you may need my hack May 26 21:30:01 so try patching the kernel May 26 21:30:18 *kernel Makefile May 26 21:31:02 GNUtoo-desktop, will do both, actual state seems more it won't/can't load neccessary modules but no spygalss around ;) May 26 21:32:50 and even bader, i was used to SysV-init but looks like angstrom switched to this newish systemd :( May 26 21:38:36 sure May 26 21:38:40 systemd is more fun May 26 21:38:54 booting to shell under 10 seconds May 26 21:39:40 it however requires a recent kernel and matching configuration May 26 21:39:51 *matching kernel configuration May 26 21:40:13 woglinde, i had a read on systemd because it seems to have a few good features but seriously, i don't care if i need 10 or 12 seconds to boot-up ;) May 26 21:42:43 GNUtoo-desktop, what may i check on the kernels config regarding systemd ? i used the defconfig but right now it looks like sub-optimal ( found HTC EGPIO enabled which makes no sense on this target) May 26 21:43:25 lautriv, you need at least the following(not sure if it's sufficent): May 26 21:43:33 * devtmpfs + devtmpfs mount May 26 21:43:54 * cgroups May 26 21:44:01 for the rest I don't know May 26 21:45:28 ok, those 2 are reasonable (matching my reading about the subject) however using tmpfs on embedded is not so nice if you take into account RAM is mostly the smallest part . May 26 21:45:48 devtmpfs May 26 21:45:50 not tmpfs May 26 21:50:14 any *tmpfs would take some RAM however i found a good reason why it van't work --> cdc-ethernet was built-in instead a module wich denies usbserial by definition. i will investigate in all options and build my own defconfig (fond several needless config, simply not present on this PDA) May 26 21:53:17 and ipaq on the host is useless because it's for propietary use ( winCE and such crap ) May 26 21:53:54 ok May 26 21:59:09 ... rebuilding kernel, will take approx. 5 years on this qemu-arm ;) May 26 22:53:11 hello all, I am trying to figure out how to see the arch/arm/mach-davinci/Kconfig file from the xconfig utility May 26 23:03:56 this Kconfig thing is crap **** ENDING LOGGING AT Sun May 27 02:59:59 2012