**** BEGIN LOGGING AT Mon Dec 15 14:38:29 2008 Dec 15 14:40:02 Jay7: zaurus initrd kexec is WORKING !! Dec 15 14:40:40 pwgen: that image which you create instead of pdaxrom's emergency image? Dec 15 14:42:06 Jay7: yes , i can extract some tar image with /boot/zimage to an sd card instert that card . and with pressong the ok button while per on an nice screen appears wher i can select that image Dec 15 14:42:26 s/per/power/ Dec 15 14:42:29 ant_work: we have new installer :) Dec 15 14:43:05 s/new/another/ Dec 15 14:43:24 and installed that image with d+c pressing ... Dec 15 14:45:35 hi, i build one of my .bb but the binary is for intel, not arm now - why that? did something changed around here recently? Dec 15 14:45:59 yeah we abandoned arm support after getting some intel sponsorship Dec 15 14:46:36 :D Dec 15 14:48:01 Dammn that OH buyout :) Dec 15 14:52:33 it says TARGET_ARCH = "arm" but if i do a file ... it says ELF 32-bit LSB executable, Intel 80386... any idea? the bb build fine last time i tried - i changed only the bb name from p.bb to p_git.bb but PV is set in the file. any ideas? Dec 15 14:53:39 emdete: I'm suprised... there is a check that tests the elf header on files... Dec 15 14:53:51 emdete: which package, which binary, which location? Dec 15 14:54:38 zecke_: yes, it seems that's the check that fails: ERROR: QA Issue: Architecture did not match (40 to 3) on /work/armv4t-angstrom-linux-gnueabi/gsm0710muxd-1.6-r6/install/gsm0710muxd/usr/sbin/gsm0710muxd Dec 15 14:55:07 zecke_: its the actual version of the muxer that i wanted to release today... :/ Dec 15 14:55:53 zecke_: but why does it build an i386 anyway? Dec 15 14:57:23 pwgen: you can hack u-boot setenv bootargs and automatically have kexecboot Dec 15 14:58:08 broken makefile? Dec 15 14:58:18 anyway, off to the airport Dec 15 14:58:35 , that may be my net plan .. Dec 15 14:58:50 grpmmf s/net/next/ Dec 15 14:58:52 Jay7: pwgen has a broken nand and cannot flash on standard address. Otherwise I don't see why having kexecboot in emergency... Dec 15 14:59:30 ant_work: we can combine emergency and kexecboot :) Dec 15 14:59:36 there are a lot more reasons. updateing the kernel will be very easy, Dec 15 15:00:26 and for example is use on my c-3200 debian, and if i do an update for the kernel i will work directly withou any changes to the nand . Dec 15 15:01:01 and its smaller then the pdaxrom emergency Dec 15 15:01:12 okay, found it, my problem ... thanks :) Dec 15 15:01:18 and i can boot emergency from sd . and and and ... Dec 15 15:01:39 I miss the point... Dec 15 15:01:52 2: Emergency 0x00540000 0x00060000 0 3: Kernel 0x00160000 0x005a0000 0 Dec 15 15:02:39 pwgen: sharp bootloader and u-boot default are doing nand read.logical 0xa0800000 0x005a0000 Dec 15 15:02:51 i will get rid of the extra kernel partiton .. Dec 15 15:03:32 iirc your nand is broken after 0xa005a0000 Dec 15 15:03:44 and you cannot flash with updater.sh Dec 15 15:03:48 right ? Dec 15 15:03:49 32e000 Dec 15 15:04:01 ah, ok Dec 15 15:04:05 i cannont use an jffs2 image . Dec 15 15:04:37 i had to copy it from some emergecy image Dec 15 15:05:12 this system is something like an user interacting bootloader Dec 15 15:08:31 hrw: found any clues to the alignment faults and segfaults with the gpe apps? Dec 15 15:11:21 zecke_: pong Dec 15 15:16:39 prpplague: sorry - had some other stuff to do. release time Dec 15 15:17:03 hrw: np, gonna look at it today myself Dec 15 15:21:10 i someone will test the kexec stuff: look at http://eric-weiss.de/trac/wiki/zaurus Dec 15 15:26:41 hrw: ping Dec 15 15:36:17 png Dec 15 15:36:43 hrw: nm, kergoth helped me out Dec 15 15:37:05 ok Dec 15 15:37:37 prpplague: I saw some alignment faults on armv5te Dec 15 15:37:55 prpplague: and can confirm gpe-sysinfo is broken Dec 15 15:38:35 probably segfaults Dec 15 15:38:35 ant_work: what version of gpe-conf are you using (gpe-info is built part of that) Dec 15 15:38:35 can someone tell me for what the last number in a row is doing from the updater.pro file Dec 15 15:39:52 ant_work: i think i'm going to use the openmoko prefered versions for all of GPE right now Dec 15 15:40:08 U-BOOT ,BIN,0x00000000,320 < what is 320 describing ? Dec 15 15:40:30 prpplague: I build angstrom from .dev Dec 15 15:40:42 possibly different ver Dec 15 15:42:48 ant_work: yea, i'm building from .dev as well Dec 15 15:43:20 ant_work: 0.2.6 of gpe-conf appears to be what is used for the pref-ver on several other build Dec 15 15:43:22 builds Dec 15 15:46:28 prpplague: I'll branch serial and track some logs Dec 15 15:48:09 pwgen: this '320' is the size Dec 15 15:48:48 ant_work: in hex i guess Dec 15 15:48:55 EMERGENC,IMG,0x00060000,5376 - > mtdparts=mtdparts=laze-0:256k(U-Boot)ro,128k(U-BootEnv),5376k(Emergency),1408k(Kernel),-(JFFS2) Dec 15 15:49:14 (this was the olfd layout, copyand paste ;-)) Dec 15 15:49:25 ahh 256+126 = 320 Dec 15 15:49:42 ahh 256+126 = 320 in kilobytes .. Dec 15 15:49:49 no, now there is anymore a dedicated partition Dec 15 15:49:54 for u-boot env Dec 15 15:52:49 kexec,bin,0x005a0000,1400 in updater.pro should install my image in kernelspace , im am right ? Dec 15 15:54:50 wasn't broken nand? Dec 15 15:55:45 : i can install the base stuff ( kernel,bootloader ) but when i will install the jffs2 image it fail. Dec 15 15:57:37 but that's not my target , i will have something like an graphical bootloader, and an very easy way to install Dec 15 15:58:30 and maybe the posibility to choose another nand layout ( no home/root partition ) Dec 15 15:59:45 pwgen: build nandlogical and do some experiments. You can flash from console Dec 15 16:10:05 pwgen: in case you missed it http://www.h5.dion.ne.jp/~rimemoon/zaurus/memo_006.htm Dec 15 16:11:11 indeed your nand is bad at 32e000, where initrd lives Dec 15 16:12:33 but still IMHO you don't need u-boot necessary then, you can flash the kexec-kernel at 00220000 as ususal Dec 15 16:12:56 for booting CF / SD Dec 15 16:13:40 ah..wait..I'm I blind...what was the address? Dec 15 16:14:31 ahh that will be find , this ( the japaneese link ) that helps me a lot Dec 15 16:15:11 grmp s/find/fine/ Dec 15 16:17:48 Bad eraseblock 465 at 0x03a20000 Dec 15 16:18:43 pwgen: it's at the end of 'normal kernel' space...bad bad Dec 15 16:19:07 you could try a minimal linux-kexecboot-image, below 1.2 mb Dec 15 16:19:34 thats exactly what i am doing ! ... Dec 15 16:20:15 my image has a size of 1.4 mb , Dec 15 16:31:44 ant_work: ping Dec 15 16:31:47 ant_work: how is the definition of the mtdlayout done , can i change it by kernel bootparams ? Dec 15 16:33:24 pwgen: it's a bit complicated story...a sane bootloader should pass this parameter Dec 15 16:34:49 and a modern kernel > 2.6.23 can receive these Dec 15 16:35:07 ant_work: what kind of device are you testing gpe-info on? Dec 15 16:35:14 zaurus c7x0 Dec 15 16:35:31 my idea is tho extend the initrdkexec to read some file with the kernel bootbarams , so i can held the bootconfiguration on some externel device Dec 15 16:36:14 pwgen: initramfs-kexec-kernel is aware of custom mtdparts (provided by u-boot) Dec 15 16:36:42 you don't need to specify it in kernel cmdline Dec 15 16:37:24 ant_work: don't suppose you could run gpe-info from a cmdline and tell me what kind of error you get? Dec 15 16:37:40 yes, np Dec 15 16:37:50 later Dec 15 16:44:14 ant_work: ok i will play a bit around with it .. Dec 15 16:44:24 pwgen: look with attention at the u-boot-git patches for Zaurus in OE Dec 15 16:44:49 these 'undo' the pdaxrom partitioning Dec 15 16:45:28 and give right mtdparts to the emergency kernel Dec 15 16:46:17 this will be a long night for me . ;P Dec 15 16:46:20 btw..if you happen to write a framebuffer-console for u-boot, well, sweet Dec 15 16:46:39 :) Dec 15 16:46:51 *G* that was anoter idea, but an kexec intrd will be a lot more simple Dec 15 16:47:20 ant_work: looks like there might be a bug in gpe-info that doesn't handle unkown devices properly Dec 15 16:48:17 gpeinfo was previously working Dec 15 16:48:37 same hardware ^_^ Dec 15 16:49:41 ant_work: hmm Dec 15 16:54:55 prpplague: gpe-info was probably developed on ipaq and zaurus :-) Dec 15 16:55:36 ant_work: yea Dec 15 16:55:51 emerge -auD system --newuse Dec 15 16:55:59 oops..sorry Dec 15 16:56:20 too many windows... Dec 15 17:00:23 ant_work: looks like the problem extends beyound gpe-info, several of the mb-panel-applets are segfaulting too Dec 15 17:01:18 prpplague: too bad Dec 15 17:01:23 some regression Dec 15 17:01:48 I'll try to find some time for testing gpe during next holidays Dec 15 17:03:00 actually what I did for the last months was: see if it boots, check if it resumes after suspend, reboot Dec 15 17:03:03 :-/ Dec 15 17:03:41 prpplague: btw is there any benchmark for gpe? Dec 15 17:04:24 with the one in opie I had some strange findings: it seems uclibcgnueabi is 40% faster than glibc Dec 15 17:04:38 *only for floating point* Dec 15 17:04:55 read softfloat Dec 15 17:05:31 ant_work: sorry i do not know, i normally don't do userland stuff, mainly kernel Dec 15 17:07:41 FWIW gpe-info crashes with both libc Dec 15 17:09:11 prpplague: are the alignment issue due to miscompilation? Dec 15 17:09:54 ant_work: that would be my guess, i noticed that the packages i'm having trouble with were compiled with -O2 Dec 15 17:10:19 you see, I use gcc-4.2.4 Dec 15 17:16:02 ant_work: i'm using 4.2.4 as well Dec 15 17:24:51 cu guys Dec 15 17:46:35 before going, crazy link : http://jerryleecooper.com/ Dec 15 17:46:37 bbl Dec 15 17:52:28 microsoft is giving you money if you use m$ ?? Dec 15 17:56:06 sounds like a winning business model to me! Dec 15 17:58:26 what a crap .. if you use m$ you had to renew every 4 years your complete hardware, because thy cancel the support, add this to your tco and the you will use some AIX or IBM hardware that runs more the the livetime of some crappy m$ gameing plattform Dec 15 18:04:20 nice Dec 15 18:04:29 angstrom 2008.1 breaks my rootfs of my new machine :) Dec 15 18:04:43 using a really old strip down of angstrom 2007 seems to work :~ Dec 15 18:06:36 will PREFERRED_VERSION_linux-davinci = "2.6.25" do what I expect? select linux-davinci_2.6.25? Dec 15 18:08:06 It's supposed to. Dec 15 18:08:24 Crofton|work: it sould Dec 15 18:08:42 yeah, waiting for a build to finish before I test Dec 15 18:08:58 the _virtual/kernel is involved :) Dec 15 18:09:41 Right, but the PREFERRED_VERSION_* is applied after the virtual is resolved. Dec 15 18:39:57 ! [rejected] org.openembedded.dev -> org.openembedded.dev (non-fast forward) Dec 15 18:40:08 any thoughts on why I see this when I try to push? Dec 15 18:40:25 Crofton: You need to merge with the changes other people have pushed. Dec 15 18:41:10 possibly by doing 'git pull --rebase' or 'git rebase' to rebase your changes onto theirs, possibly by merging origin/org.openembedded.dev into your branch. Dec 15 18:41:52 thanks Dec 15 18:42:09 I tried git add -i and seem to have got a file in a weord state Dec 15 18:42:23 # packages/libnfnetlink/libnfnetlink_0.0.39.bb Dec 15 18:42:26 is unteacked now Dec 15 18:42:51 I need to remove the git add -i as a "cool way" to make changes .... Dec 15 18:43:13 Or learn how to drive it :) Dec 15 18:43:20 right Dec 15 18:43:30 but it shouldn't be i nthe phraseboo the way it is Dec 15 18:43:37 git gui can be helpful there - it's got the pointy clicky going on, and does support -i style commiting. Dec 15 18:43:37 with 0 instructions Dec 15 18:44:10 my go with git add -i seems to have got libnfnetlink in a weird state Dec 15 18:44:12 hello Dec 15 18:44:16 question Dec 15 18:44:19 and trying to check it out again does not help Dec 15 18:44:41 I just set up my bitbake/oe environment again following the info at http://wiki.navit-project.org/index.php/Navit_on_%C3%85ngstr%C3%B6m Dec 15 18:45:00 bitbake fails building already at step 6. Dec 15 18:45:24 whic is shasum-native Dec 15 18:45:29 which is shasum-native Dec 15 18:45:41 it can't find stddef.h Dec 15 18:45:55 and stdarg.h Dec 15 18:46:07 but they both are in /usr/include/linux Dec 15 18:46:10 as they have been Dec 15 18:46:15 what went wrong? Dec 15 18:46:22 and how can I correct this? Dec 15 18:47:25 Crofton: manually git add it? Or look at the commit in git gui? Dec 15 18:48:08 udovdh, what distro are you using on the build machine? Dec 15 18:48:15 fedora 10 Dec 15 18:48:20 it used to work on fedora 9 Dec 15 18:48:23 so you think git adding it back will solve the problem? Dec 15 18:48:30 tahts why I wrote the small howto-ish wiki thing Dec 15 18:48:48 Crofton: Dunno, depends on exactly what you've done. Dec 15 18:48:48 i think you need to try diabsling CCACHE Dec 15 18:49:00 try adding CCACHE = "" in local.conf Dec 15 18:49:21 CCache was coment for me? Dec 15 18:49:21 broonie, I don't know what I did either :) Dec 15 18:49:56 CCache is already in local.conf here Dec 15 18:50:06 Crofton: Quite :) adding it back then looking at teh overall effect of the commit might be a good starting point. Dec 15 18:50:10 s/commit/index/ Dec 15 18:50:45 now git status says it is new, but git diff is empty Dec 15 18:50:49 bitbake is 1.8.12 if that matters Dec 15 18:50:56 I need to horse around with it some Dec 15 18:51:05 udovdh, try CCACHE = "" in local.conf Dec 15 18:51:18 Crofton, it is alread yin there Dec 15 18:51:24 Crofton, it is already in there Dec 15 18:53:30 udovdh, I don't know Dec 15 18:53:40 I need to take a break for a bit alsi Dec 15 18:54:40 conf: http://pastebin.com/dc21065 Dec 15 18:54:50 hmmm Dec 15 18:54:59 bitbake 1.8.10 fixes this issue!? Dec 15 18:55:02 it does more now Dec 15 18:55:48 Crofton: git diff will show you changes not in the index by default, you need to diff against your HEAD to see the changes that will be committed (I do recommend gui for visualisation in this sort of situation). Dec 15 18:56:25 Crofton: You might also use 'reset' to forget the index state (*don't* use --hard or you'll be depressed, though). Dec 15 18:59:32 git diff --cached is what will be committed. git diff HEAD is both what is currently staged int he index for commit, and any additional changes in the working copy that aren't yet in the index Dec 15 19:03:42 thanks for reading, later... Dec 15 19:20:50 while bitbaking I see: Dec 15 19:20:51 NOTE: Handling BitBake files: \ (2831/4977) [56 %]http://svn.projects.openmoko.org/svnroot/python-openmoko/python-gsmd: (Not a valid URL) Dec 15 19:20:56 who can fix this? Dec 15 19:34:41 udovdh: What's the problem? Dec 15 19:46:01 udovdh: are ypu using latest git ? Dec 15 19:46:46 have a look at package/python/python-gsmd_svn.bb on latest head Dec 15 20:15:36 why doesn't autobuilder produce armv6-novp packages? :( Dec 15 21:17:48 mickey|zzZZzz: hello; i'm working at a basic lxde set of packages and pcmanfm is one of them. The only user looks to be OM but I doubt it is being used. I've updated it to 0.5 and would like to drop the other versions, any objection? Dec 15 21:21:00 morning all, Dec 15 21:21:17 is anybody doing ppc builds? Dec 15 21:44:06 thesing, I do some ppc405, but I can't test them .... Dec 15 22:00:06 * * OE Bug 4935 has been created by david.hagood(AT)aeroflex.com Dec 15 22:00:07 * * build of flex-native fails Dec 15 22:00:10 * * http://bugs.openembedded.net/show_bug.cgi?id=4935 Dec 15 22:04:31 mickey|zzZZzz: there is a new muxer for you at http://pyneo.org/downloads/ - version 1.6 fixes the coredump at ^C problem you talked about Dec 15 22:20:28 thesing: ping Dec 15 22:20:48 ant_: pong Dec 15 22:21:08 hello, I'm looking at the defconfigs again... Dec 15 22:21:58 Crofton|work: which machine do you use? Dec 15 22:22:09 did you follow some special criteria doing kexecboot-defconfig for collie? Dec 15 22:22:39 ant_: no. I tried to throw everything possible out. Dec 15 22:22:56 ok, I was lazy and only disabled networking... Dec 15 22:23:43 if the kernel is small enough thats totally ok. Dec 15 22:23:47 I use xilinx-ml403 Dec 15 22:25:28 thesing: referring to koen's patch (+echo 'CONFIG_BLK_DEV_INITRD=y Dec 15 22:25:28 +CONFIG_INITRAMFS_SOURCE="initramfs.cpio.gz"' >> ${S}/.config) Dec 15 22:25:53 don't you think CONFIG_KEXEC=y Dec 15 22:25:59 should be forced too? Dec 15 22:26:09 yes it should Dec 15 22:26:32 (I hate this rewrite of defconfigs done by the recipe) Dec 15 22:27:01 Crofton|work: if you do a clean build some time in the future, could you try if you can build module-init-tools-cross without building perl-native first? Dec 15 22:30:39 thesing: do you know about power management changes between 2.6.24 and 2.6.26? Dec 15 22:31:23 there were probably some. Dec 15 22:31:35 but for collie I have pm disabled in kexecboot kernel. Dec 15 22:31:41 I have that minor issue..no recharging if suspended Dec 15 22:32:05 ant_: if you see Jay can you tell him that he can get git access to kexecboot if he wants? Dec 15 22:32:20 btw I have pm enabled in kexecboot for 2.6.26-c7x0 and suspend/resumes happily Dec 15 22:32:27 ant_: he has to register at linuxtogo and request to join collie project Dec 15 22:32:41 thesing: seems very good idea Dec 15 22:33:20 ant_: pm is kexecboot has nothing to do with pm in real kernel. Unless you want to suspend in the boot menu. Dec 15 22:34:13 thesing: all kexecboot kernels I tested have that same issue: the ac-power get disconnected immediatly before scanning Dec 15 22:34:26 the boot media Dec 15 22:34:51 kernels < 2.6.24 re-initiate the power-charge Dec 15 22:34:57 2.6.26 doesn't Dec 15 22:35:12 (after kexecboot) Dec 15 22:35:27 strange... Dec 15 22:35:38 same defconfig Dec 15 22:35:57 but different patchset Dec 15 22:37:45 thesing: btw resuming goes back to console..pressing a key gives back the framebuffer Dec 15 22:37:50 maybe its because of the gpio lib Dec 15 22:38:28 sorry, I meant <= 2.6.24 Dec 15 22:40:42 thesing: the only difference I see, 2.6.26 has > CONFIG_INPUT_APMPOWER=y Dec 15 22:41:34 yes but that is only a config change. Under the hood there were probably many changes. Dec 15 22:41:54 the fix is to check for ac at the startup of the pm driver. Dec 15 22:41:57 yes, 2.6.26 has some 15 patches less Dec 15 22:42:54 thesing: this is the only issue...if you branch AC *before* suspending, all is fine... Dec 15 22:43:22 dunno on other machines of the Z class Dec 15 22:49:50 thesing: one last thing...do you want commit tosa-kexecboot or do you prefer that Dmitry does it himself? Dec 15 22:51:12 I don't care. I wanted to do it yesterday, but I didn't find your patch ;) Dec 15 22:51:22 not mine... Dec 15 22:51:40 the on that dmitry based his on. Dec 15 22:51:59 ah ok...errors included :-) Dec 15 22:52:19 please purge that # uncomment... Dec 15 22:53:18 ant_: its best if dmitry commits it if he has time Dec 15 22:53:53 I'll have to bother him about kernel defconfigs too! Dec 15 22:54:05 RP is practically absent nowadays Dec 15 22:55:05 * * OE Bug 4936 has been created by david.hagood(AT)aeroflex.com Dec 15 22:55:08 * * Build of openssl-native fails Dec 15 22:55:10 * * http://bugs.openembedded.net/show_bug.cgi?id=4936 Dec 15 22:57:22 ant_: practically but not always ;-) Dec 15 22:59:00 :) Dec 15 23:00:49 RP: are you still maintaining the clamshell kernels or did you pass that to the pxa maintainer? Dec 15 23:01:25 thesing: Officially I'm still the maintainer Dec 15 23:02:24 RP: do you plan to resume work on them? Dec 15 23:02:34 thesing: I'd very much like to, yes Dec 15 23:02:35 RP: Dmitry was talking about 2.6.28...how far is it? Dec 15 23:02:52 thesing: Hopefully I'll get some time over the holidays to catch up with that Dec 15 23:03:03 very good Dec 15 23:03:35 RP: what do you think about moving from sharpsl-pm to the power class? Dec 15 23:04:09 thesing: Will probably need to be done sooner or later Dec 15 23:05:24 good to know. I will use it for collie then. Dec 15 23:06:28 thesing: RP: and about these messages? Dec 15 23:06:30 sharpsl-pm sharpsl-pm: Error: AC check failed. Dec 15 23:06:31 sharpsl-pm sharpsl-pm: Offline Charger: Error occurred. Dec 15 23:06:31 sharpsl-pm sharpsl-pm: Charging Error! Dec 15 23:06:50 in linux-rp-2.6.26 Dec 15 23:07:36 ant_: have a look at arch/arm/common/sharpsl_pm.c ;) Dec 15 23:07:55 well, ok, some missing patch? Dec 15 23:08:17 no some debug messages. Dec 15 23:10:02 these happen inserting AC-plug when suspended (no recharge, no led) Dec 15 23:13:06 * * OE Bug 4937 has been created by david.hagood(AT)aeroflex.com Dec 15 23:13:08 * * Build of db-native fails Dec 15 23:13:10 * * http://bugs.openembedded.net/show_bug.cgi?id=4937 Dec 15 23:17:07 thesing: I've seen..smthg goes wrong if ((temp > sharpsl_pm.machinfo->charge_acin_high) || (temp < sharpsl_pm.machinfo->charge_acin_low)) Dec 15 23:17:27 dev_err(sharpsl_pm.dev, "Error: AC check failed.\n"); Dec 15 23:18:47 so its too cold or too warm. Or the sensor reading is wrong. There hase been some move from zaurus specific drivers to generic drivers for some chips. Dec 15 23:21:14 how are the readings fine if not suspended then? Dec 15 23:21:25 no idea. Dec 15 23:21:44 if suspended there are no reading. Dec 15 23:21:47 the cpu sleeps. Dec 15 23:22:05 When suspended and charging it periodically wakes up Dec 15 23:22:18 remember, it's a 2.6.26 only problem... Dec 15 23:22:39 I'd suspect all the corgi_ssp changes :( Dec 15 23:23:01 on OESF there are reports of Zaurus in berzerk mode, auto-activating in the night ... Dec 15 23:26:03 seriously, the reset done by pulling the battery out reset the device to a state where the insertion of the AC plug is recognized Dec 15 23:26:20 same happens with linux-rp-2.6.23 and 24 kernels Dec 15 23:27:38 so possibly is a GPIO thing Dec 15 23:27:44 possibly Dec 15 23:39:03 night all Dec 15 23:47:31 03Khem Raj  07org.openembedded.dev * r8cb93959b7 10openembedded.git/ (2 files in 2 dirs): Dec 15 23:47:31 sane-srcrevs.inc: Move sane srcrev for eglibc out of here. Dec 15 23:47:31 eglibc_svn.bb: Move sane srcrev for eglibc to here. Dec 15 23:50:47 03Khem Raj  07org.openembedded.dev * r55ba8c22a8 10openembedded.git/packages/images/native-sdk-console-image.bb: Dec 15 23:50:47 native-sdk-console-image.bb: Add new. Dec 15 23:50:47 * This target includes console image plus everything thats needed to test Dec 15 23:50:47 the toolchain on the target. Dec 15 23:50:48 khem: Why move the srcrev to there? Dec 15 23:53:36 RP: I playa lot with bleeding libc and every time I change it it forces bb parse Dec 15 23:53:50 which is sort of annoying Dec 15 23:54:11 khem: so change it locally, not globally :/ Dec 15 23:55:08 Dont you think its more accessible if it is in recipe Dec 15 23:55:10 khem: any value set in the file will override the conf/inc file Dec 15 23:56:10 RP: yes but once its in the recipe then why do we need the global conf/inc one Dec 15 23:56:57 The point being distros can select their versions Dec 15 23:57:42 is sane-srcrev unique for distros ? Dec 15 23:58:33 distros are allowed to override it Dec 16 00:01:28 i see Dec 16 00:02:06 I think it is less of work to maintain it in recipe in this case Dec 16 00:03:07 I thought svn recipe is more of experimental Dec 16 00:03:39 and distribution should be more concerned about the releases Dec 16 00:04:11 khem: Just do what you want but one rule for you and one for everyone else never goes down well Dec 16 00:04:18 I don't have time to discuss it :( Dec 16 00:05:17 RP I will restore it in sane-srcrev.inc my only motivation is to save the bb parse time Dec 16 00:05:51 which I think I could do by keeping the recipe change local Dec 16 00:05:58 khem: You might find it more useful to write a good BBMASK... Dec 16 00:06:25 I tried once and failed Dec 16 00:11:10 03Khem Raj  07org.openembedded.dev * r5d02d38a97 10openembedded.git/conf/distro/include/sane-srcrevs.inc: ane-srcrevs.inc: Reinstate eglibc sane srcrevs. Dec 16 00:12:39 khem: A GUI would be useful for that... Dec 16 00:12:45 hmmm so armv5t is also thumb now Dec 16 00:13:05 its a PITA to develop thumb in my experience Dec 16 00:13:15 tools dont go so well. Dec 16 00:13:23 gdb, strace .. Dec 16 00:14:57 well Angstrom is thumb ... Dec 16 00:15:34 yeah I have to move to a arm distro :) Dec 16 00:21:11 khem: it's a one-line-hack ;-) Dec 16 00:21:39 khem, if you can provide examples of why thumb is bad, send them to the Angstrom list Dec 16 00:21:41 Crofton|work: Does it mean that you will compile glibc in thumb mode too Dec 16 00:21:46 and also kernel ? Dec 16 00:21:50 not cure Dec 16 00:21:52 er sure Dec 16 00:22:14 I suspect so Dec 16 00:22:18 Crofton|work: I have made a uclibc/thumb distro in past Dec 16 00:22:38 Crofton|work: Glibc was not compilable in thumb mode and kernel too Dec 16 00:22:43 there seems to be some data that says thumb is a performance won Dec 16 00:22:50 may be Angstrom guys fixed it Dec 16 00:22:52 I dont know Dec 16 00:23:12 eglibc can be compiled in thumb2 but thats no way thumb Dec 16 00:23:37 yeah, I get confused by thumb and thumb2 Dec 16 00:23:43 thumb2 is armv7? Dec 16 00:23:47 Crofton|work: yes Dec 16 00:24:41 I considers tools to be important. You might get a system working in thumb but once you want to debug something on it then you will stumble upon many things Dec 16 00:24:51 like debug info Dec 16 00:24:57 and incomplete support in tools Dec 16 00:25:02 for thumbness Dec 16 00:25:15 well, this came up on the Angstrom list, but the problem is no one knows this there Dec 16 00:25:21 but I guess now we can fix them :) Dec 16 00:25:30 so we all sit around and don't say anything Dec 16 00:25:33 heh Dec 16 00:25:36 that is good also Dec 16 00:25:45 thumb needed a scapegoat Dec 16 00:25:46 khem: andrea@mizar /oe/org.openembedded.dev/packages $ grep -R ARM_INSTRUCTION_SET . | grep "arm" Dec 16 00:26:00 the users versus developers war will never end Dec 16 00:26:21 ok, I need to do my shoulder exercises Dec 16 00:26:38 Crofton|work: are you a pitcher :) Dec 16 00:27:06 no, just getting old Dec 16 00:27:12 anyway, that is bowler Dec 16 00:27:35 bowler like in cricket bowler Dec 16 00:27:42 right Dec 16 00:31:59 http://lists.linuxtogo.org/pipermail/openembedded-devel/2008-October/006718.html Dec 16 00:32:59 ARM-state vs Thumb-state userspace Dec 16 00:33:22 so kernel is probably arm mode for sure but libc not sure Dec 16 00:34:09 I think one tool that helps quite a lot is prelinker I dont know if we have that as part of OE **** ENDING LOGGING AT Tue Dec 16 02:59:57 2008