**** BEGIN LOGGING AT Tue Aug 07 02:59:59 2012 Aug 07 06:25:06 hello Aug 07 06:25:18 bitbake -c fetchall succeedded Aug 07 06:25:45 but if I try to build minimal image bitbake fails here Aug 07 06:26:03 STDERR: fatal: unable to connect to git.yoctoproject.org: Aug 07 06:26:03 git.yoctoproject.org[0: 140.211.169.122]: errno=No route to host Aug 07 06:26:14 build machine is without internet connection Aug 07 06:26:31 why is not this downloaded with previous fetchall ? Aug 07 06:48:55 CruX|: seems more a question for #yocto Aug 07 06:49:05 I'd be interested in the answer too Aug 07 06:51:03 thx Aug 07 06:52:16 probably because of trying git ls-remote for AUTOREV or SRCPV Aug 07 06:54:47 JaMa|Wrk: JaMa|Wrk can be that easy fixed ? affected .bb recipe is virtual:native:/oe/openembedded-core/meta/recipes-devtools/opkg-utils/opkg-utils_git.bb Aug 07 06:57:09 does your config set AUTOREV for opkg-utils or empty BB_LOCALCOUNT_OVERRIDE? Aug 07 06:57:38 or even BB_GIT_CLONE_FOR_SRCREV? Aug 07 06:58:36 no, i have set only BB_GENERATE_MIRROR_TARBALLS = "True" Aug 07 07:00:21 hmm git2_git.yoctoproject.org.opkg-utils.tar.gz exist in my source dir Aug 07 07:00:27 then in theory it shouldn't try to access that, check from where it's calling git Aug 07 07:07:34 can I turn off git update/prune... in SRC_URI ? Aug 07 07:08:38 because mirror tarball is extracted but git fetch -f --prune git://git.yoctoproject.org/opkg-utils refs/*:refs/* is executed Aug 07 07:12:09 do you have latest bitbake? http://git.openembedded.org/bitbake/commit/?id=2ba23df5fad4b94d38a6aed97f7822226d72eb89 Aug 07 07:12:34 ah this is different prune Aug 07 08:43:18 JaMa|Wrk: have you recently tested armv4t ? Aug 07 08:45:47 hi all Aug 07 08:46:37 hi Noor, all Aug 07 08:56:57 ant_work: yes I'm still using om-gta02 Aug 07 08:58:17 we have issues on collie and I was wondering about Aug 07 08:58:18 http://repository.timesys.com/buildsources/g/gcc/gcc-4.7.0/gcc-4.7.0-arm-unbreak-eabi-armv4t.patch Aug 07 08:58:34 do you have gold enabled? Aug 07 08:58:46 using defaults Aug 07 08:58:56 no distro Aug 07 08:59:26 it's broken with gold for sure, don't know if gold is enabled by default Aug 07 08:59:31 fwiw I see th eyocto kernel config scripts complain that THUMB is not in the final config Aug 07 08:59:36 :) Aug 07 08:59:51 I'm not using yocto kernel Aug 07 09:03:54 we have panic after Freeing init memory, could also be the kernel Aug 07 09:15:30 JaMa|Wrk: fwiw that patch still apply, gcc master does #define SUBTARGET_CPU_DEFAULT TARGET_CPU_arm10tdmi Aug 07 09:16:19 khem: once you wake up, ^^^ Aug 07 09:39:48 guys I am facing issue while compiling docbook-utils-native http://pastebin.ca/2178131 Aug 07 09:40:36 Can somebody guide me what is going wrong here Aug 07 10:17:37 morning all Aug 07 10:20:16 hi pb_ Aug 07 10:25:02 hi bluelightning Aug 07 10:57:00 hi pb_ Aug 07 11:09:21 pb_: correct me if I'm wrong: ARM and Thumb mode of the kernel and userspace are independent Aug 07 11:10:36 ant_work: maybe you're missing http://git.freesmartphone.org/?p=linux-2.6.git;a=commit;h=0333777365bef05b336d4f4a06bb3b37f4ada838 Aug 07 11:13:32 hm, thi sis for meta-toolchain Aug 07 11:14:41 ok, I see, in toolchain-klayer sits gcc-4.6 Aug 07 11:20:51 the same does apply to 4.7 Aug 07 11:38:47 guys any pointer on docbook-utils-native failure Aug 07 11:52:53 JaMa|Wrk: ok, th, will try Aug 07 11:53:19 linker is not gold btw Aug 07 12:39:50 What is The Correct Way™ to make cronie run the scripts in /etc/cron.hourly, daily, etc? Aug 07 12:41:09 Do I make a new package that contains /etc/crontab, or do I somehow "import" the cronie package and add modified /etc/crontab? Aug 07 12:41:58 In the current version, 1.4.8-r4, /etc/crontab doesn't run anything. Aug 07 12:45:22 And same question on a more generic level: Aug 07 12:46:54 If I want to add or modify a file in a package, so that my package is just like the upstream package, but one file added or modified, how do I do that so that changes from upstream will automagickally go to my version as well? Aug 07 12:49:03 hi ant_work Aug 07 12:49:07 ant_work: yes, correct Aug 07 12:49:33 pb_: unfortunately he's gone... Aug 07 12:49:40 ah yes, whoops Aug 07 12:49:50 oh well, that's what he gets for asking questions when I am at the pub :-} Aug 07 12:49:55 heh Aug 07 12:51:10 ibot, botmail for ant_work: correct Aug 07 12:52:59 inch: it's easy to add additional files; you just need to put the file in a files (or PN, or PN-PV, etc.) subdir next to the recipe, then edit the recipe and add the file to SRC_URI and install it within do_install Aug 07 12:54:09 inch: if you're using OE-Core then you probably want to do that in a bbappend in your own layer rather than modifying the original recipe Aug 07 12:55:15 Ok, googling "bbappend" and "oe" wasn't exactly helpful :-) Aug 07 12:56:16 the most up-to-date documentation regarding the current OE-Core based OE is probably the Yocto Project documentation: https://www.yoctoproject.org/documentation Aug 07 12:56:25 the dev manual has a short section on bbappends Aug 07 12:57:29 bluelightning: Thanks. Aug 07 13:04:17 if there a way to disable sstate-cache Aug 07 13:05:51 i'm having trouble to share sstate-cache among different builds, a single build works fine though Aug 07 13:37:52 xxiao_: not really, no... what do you mean exactly by share among different builds? Aug 07 13:45:40 bluelightning: i have many similar boards(hardware) that is sharing one sstate-cache Aug 07 13:46:03 trying to change it to each board with its own sstate-cache, though those boards are very much the same Aug 07 13:46:32 basically move SSTATE-CACHE from TOPDIR to each board's BUILD dir now Aug 07 13:46:37 xxiao_: ok, there should not be a problem with that... what kind of failures are you seeing? Aug 07 13:46:41 that should fix it, but will take lots of disk space Aug 07 13:47:05 bluelightning: first board compiles fine, second board in its own build directory failed perl build Aug 07 13:47:39 xxiao_: presumably you mean perl for the target rather than perl-native? Aug 07 13:48:26 bluelightning: it's under meta/recipes-devtools/perl/perl_5.12.3.bb, not sure if that's native or target Aug 07 13:50:22 http://pastebin.com/fugtMxzU Aug 07 13:51:03 this works fine for my first board, second board sharing the same sstate-cache will fail perl Aug 07 14:00:08 xxiao_: are those real paths in that log or have you sanitised them? Aug 07 14:00:32 bluelightning: i sanitised the path a bit Aug 07 14:01:08 it's the TMPDIR/work/linux/ that has me confused, as long as that's actually armv7-distro-linux or similar Aug 07 14:01:20 i tried to debug it...i can re-run that pastebin commands and it errs, Aug 07 14:01:45 which tar version is on your host? Aug 07 14:02:07 1.22, ubuntu 10.04 64bit Aug 07 14:02:27 again the first build passes always, anything following will fail perl Aug 07 14:02:51 i can even swap the board build order, always the second board will fail Aug 07 14:03:28 can the sstate-cache cope with various version/board well? or should I use sstate-cache for each board Aug 07 14:03:33 that version of tar has a bug in it that prevents it from correctly overwriting symlinks Aug 07 14:03:53 (fixed in 1.24) Aug 07 14:03:58 now, we're supposed to be working around that by building tar-replacement-native Aug 07 14:04:04 bluelightning: is this a known bug? why did it not happen for the first build, curious Aug 07 14:04:11 ic Aug 07 14:04:31 because we never really hit the overwriting symlinks case unless we're unpacking on top of an existing directory which can happen for sstate Aug 07 14:04:47 let me build a 1.24 tar and give it a shot Aug 07 14:04:55 xxiao_: hang on a sec Aug 07 14:05:05 xxiao_: is tar-replacement-native being built for you? Aug 07 14:05:15 let me check Aug 07 14:05:29 it should be, the bitbake wrapper script checks the version of tar you have and builds it if it's 1.23 or earlier Aug 07 14:06:02 how to find out tar-replacement-native is used Aug 07 14:07:55 xxiao_: have a look if there's a tar executable under TMPDIR/sysroots//usr/bin Aug 07 14:09:23 ./work/x86_64-linux/tar-replacement-native-1.26-r0 Aug 07 14:09:49 ./work/x86_64-linux/tar-replacement-native-1.26-r0 Aug 07 14:10:02 xxiao_: ok, but did it actually put a tar binary in the sysroot? Aug 07 14:11:32 ./sysroots/x86_64-linux/usr/bin/tar Aug 07 14:11:55 by native here do we mean the target or the host? i normally think native means the target sysroot Aug 07 14:12:08 here it says x86_64...that's host to me Aug 07 14:12:27 right, by native I mean host Aug 07 14:12:32 ./sysroots/x86_64-linux/usr/bin/tar is 1.26 Aug 07 14:12:52 so it did install tar into the sysroot...ok Aug 07 14:13:44 xxiao_: if you can reproduce the error with that command line externally does it still happen if you substitute a direct path to the tar 1.26 executable instead of just "tar"? Aug 07 14:18:46 bluelightning: that stays after i replace tar with a fixed path to this 1.26 tar from sysroot Aug 07 14:18:52 same err Aug 07 14:19:18 hmm, ok... I guess it's not the same issue... Aug 07 14:19:47 xxiao_: both tar commands I guess? Aug 07 14:20:04 yes Aug 07 14:20:21 tar-1.26 cf | tar-1.26 xf Aug 07 14:22:11 hmm, it's not behaving the way that the tar manual suggests that it should... Aug 07 14:22:59 this is edison/yocto Aug 07 14:23:27 that's fine Aug 07 14:23:29 anyway thanks for the help! i will debug a little more Aug 07 14:23:46 xxiao_: ok, I'll continue to think about it, let me know if you find anything Aug 07 14:23:57 sure. thanks again Aug 07 14:41:06 pb_: got your answer, thx. Now trying with gcc patches for armv4t Aug 07 14:43:24 very good Aug 07 14:50:26 bluelightning: it might not be a tar issue at all, the perl bb seems wrong Aug 07 14:50:45 xxiao_: oh, what did you find? Aug 07 14:51:25 tar cf - | tar xf /tmp will always work, one after another so overwriting with tar is fine Aug 07 14:51:55 i then compared the two perl directories, the dest perl has a directory perl5, the src perl has a symlink perl5 Aug 07 14:52:23 a symlink is not supposed to replace a real directory of the same name i guess, the content is also different Aug 07 14:52:30 not sure why it happened though Aug 07 15:56:06 bitbake hello -c compile -f still works? i found unless PR is bumped up, local changes will not take effect even -f is used, then why do we still have -f option Aug 07 15:56:31 all because of sstate-cache i think Aug 07 15:59:04 isn't there something like ~/.bitbakerc these days? I can't find the exact name Aug 07 16:02:04 hollisb: no. Aug 07 16:02:09 never has been Aug 07 16:02:19 kergoth: oh, I thought you mentioned something like that the other day Aug 07 16:02:32 for a rough equivalent, add ~/.oe to BBPATH in bblayers.conf and create ~/.oe/conf/site.conf Aug 07 16:02:38 that's what i do Aug 07 16:02:42 oh, I see Aug 07 16:02:46 xxiao_: yes, known issue fixed recently Aug 07 16:03:32 kergoth: works until I create a new build dir 3 months for now and forget to do it again... but thanks Aug 07 16:04:39 ideally we would have a mechanism that allowed people to customise the conf templates used by the oe-init-build-env script Aug 07 16:05:04 I'm not entirely sure how it would work though Aug 07 16:05:16 given the lack of context when running the script... Aug 07 17:27:57 hollisb: you can make your own script to create your build dir ;) Aug 07 17:32:28 mmm hmm Aug 07 17:59:45 is saul around? Aug 07 18:00:08 hollisb: heh, there's a reason i don't use oe-init-build-env :) Aug 07 18:15:39 I would expect OE's init-build-env to work more or less like it does.. for other distribution providers, it's highly likely they will tailor or replace it to meet their needs.. Aug 07 18:15:50 I wouldn't be surprised if alterantive versions for oe-core existed either.. Aug 07 18:19:43 fray: "highly likely" meaning you happen to know that they do? :) Aug 07 18:19:52 means it was designed that way Aug 07 18:20:16 when you have your own layers, and such.. using the existing oe-init is likely not what you want Aug 07 18:38:32 hello Aug 07 18:42:47 i would like to lean OpenEmbedded, but i haven't device now. can anyone suggest an embedded device please? Aug 07 18:43:02 qemuarm Aug 07 18:43:26 fray: it would be nice to have the default oe-init-build-env overridable or extensible via something in BSP layer Aug 07 18:43:40 i prefer a physical device :) Aug 07 18:43:41 JaMa is correct, use a QEMU emulated system to start.. then move on to a "real" system.. Aug 07 18:43:56 msm -- bsp layer is minor compared to distribution and functional layers.. Aug 07 18:44:25 but the problem is until oe-init-build-env runs, there is no bblayers.conf, and without that, it can't look at the layers to figure out what files to copy in.. Aug 07 18:45:05 so once something says what layers you want (on a new configuration) the rest of the logic is pretty easy.. Aug 07 18:46:08 JaMa: can you suggest a physical device? Aug 07 18:47:07 fray: did not think about that issue… Aug 07 18:47:35 thats why it becomes custom for a distribution (or even layers) Aug 07 18:48:30 fray: well we have something, angstrom has something, it would be nice to have a standard convention Aug 07 18:49:13 the best I can tell, a "configure like" syntax of --enable- or --with- is likely needed to pass the layers.. then the logic for parsing them for default configuration elements (or fragments) becomes pretty clear.. Aug 07 18:49:29 combine the fragments/elements to creat the local.conf and bblayers.conf Aug 07 18:49:51 that would work Aug 07 18:50:22 and to make it easier, you'd want to support directory globbing/wild carding.. Aug 07 18:50:29 something like --with-layers=meta-* Aug 07 18:57:19 peppux: depends what you want from it, beagleboard,pandaboard,rpi,freerunner,samsung phones,zaurus Aug 07 18:57:52 there are anumber of Intel Atom systems as well.. PPC and MIPS is harder to get these days.. Aug 07 18:58:08 (PPC and MIPS seem to have become very specialized, so the devices are similarly more expensive) Aug 07 18:59:53 (specific information about a few Atom boards: http://boardswithbenefits.com/) Aug 07 19:27:51 fray: i feel ppc and mips are easy to get, why do you think it's hard Aug 07 19:28:18 I haven't been able to find a place to buy an individual board for less then US$1000 Aug 07 19:28:28 fray: what's the spec Aug 07 19:28:31 and usually then they require qunatities > 3 to get anything.. Aug 07 19:28:54 MIPS32R2 or better.. PowerPC pretty much anything modern.. Aug 07 19:29:05 Atom (and IA) is easy to aquire.. same w/ ARM.. Aug 07 19:29:21 on MIPS it used to be easy to get a RouterStation Pro for little cost.. but that board has been discontinued Aug 07 19:29:48 * xxiao_ owns 6 Routerstation Pro Aug 07 19:29:58 ya, I have one as well as a test system.. Aug 07 19:30:17 i think you can still buy mips/ppc boards chip from vendors like routerboard.com Aug 07 19:30:29 s/chip/cheap/ Aug 07 19:30:49 gosh, for $1000 there are so many options Aug 07 19:30:50 Broadcom told me their stuff was cheap as well.. they pointed me at a few places.. and they wouldn't sell me boards in less then 10 units.. Aug 07 19:31:07 pricing for what I consider a reasonable hobbiest board is between $100 and $250 (max) Aug 07 19:31:15 those are only good for routing. to have a decent testbed you'd need keyboard and lcd attached Aug 07 19:31:17 i never touch any broadcom boards, don't feel they're OSS friendly Aug 07 19:31:32 ant__, none of my systems have keyboard or lcd (including IA) :) Aug 07 19:31:48 xxiao__ BCM unfortunately keeps buying up semi vendors who have been more "open" Aug 07 19:32:12 fray: I was used to routers until I bought my first Zaurus Aug 07 19:33:28 stuff I do is all low level.. serial port, ethernet and preferrably USB or some other way to have storage on-board is really all I need.. Aug 07 19:33:37 for network stuff ppc/mips still win ARM Aug 07 19:33:51 yes it is.. but that is also part of hte reason for the cost premium.. Aug 07 19:34:06 the vendors know you are buying stuff for networking, and feel they can charge more because of that Aug 07 19:34:17 for portable device, even industrial control or generic embedded device, with/without display, ARM seems to be choice Aug 07 19:34:45 ya, it's come down to cost.. PPC used to be a good choice, but the costs just haven't kept up Aug 07 19:34:59 look at freescale's p10xx and 8308 devices, they're cheap and powerful(1Gbps, PCIe, etc) Aug 07 19:35:11 Some of the Intel Atom parts are getting to the point (power/performance) to be challenging to arm in industrial control and automation uses.. Aug 07 19:36:07 fray: right, they're going to meet each other somewhere in the middle, very soon Aug 07 19:36:17 (I'm replacing my person DNS backup and email gateway with a cheap Intel atom board.. I couldn't find anything small form faxctor and no moving parts any cheaper) Aug 07 19:36:37 how much is that atom thingy Aug 07 19:36:47 $125 Aug 07 19:36:53 plus $30 for an SSD.. Aug 07 19:37:02 Thats an N450 "Blacksands" board Aug 07 19:37:20 came w/ either 512 MB or 1 GB of ram.. can't remember anymore.. Aug 07 19:37:29 but it's plenty for that machine.. the SSD is a 30GB model.. Aug 07 19:37:30 not bad Aug 07 19:37:39 the $125 included the power brick as well.. Aug 07 19:37:57 meanwhile i would think a $50 ARM board with 8GB SD will do fine for that Aug 07 19:37:58 (it's going onto a shelf in my basement.. hopefully never to be touched again..) ;) Aug 07 19:38:16 it's the mail forwarding and storage that i wanted the higher performance for.. Aug 07 19:38:24 ic Aug 07 19:38:44 I looked at ARM systems, but couldn't find any that used an SSD.. they were all SD.. and that stuff (in my experience) burns out way too quickly.. Aug 07 19:38:53 fanless is good, SSD is cool, i would use atom in that case Aug 07 19:39:03 even the SSD is a bit low on the MTBF in my experience.. but larger capacity hopefully means better wear levling Aug 07 19:39:40 I stressed 8 Vertex(?) SSD a while ago, 2 of them suffered sudden death and never came back to life Aug 07 19:39:56 ya.. builds and swap are not a good idea on SSDs.. still Aug 07 19:40:03 but general usage seems to be fine.. Aug 07 19:41:17 the power supply that comes with it is a 150 watt brick.. Aug 07 19:41:24 (12v & 12.5 amps max) Aug 07 19:41:34 that's overkill why Aug 07 19:41:36 when I checked in a few weeks ago.. it was drawing about 50 wats Aug 07 19:41:46 it's to support hard drivers and other rotating media Aug 07 19:42:05 but it worked well with the DVD-ROM I had hooked to it initially.. ;) Aug 07 19:42:12 ic Aug 07 20:07:56 fray: with headless machines you're ignoring the dark side of *x Aug 07 20:08:06 :p Aug 07 20:10:05 that is very much on-purpose.. I'm not a UI guy Aug 07 20:11:02 heh, that old excuse I'm using since 20 years... Aug 07 20:11:48 after working on a product installer, and working with a UI designer.. I've thrown in the towel.. I have no desire to do that work.. since doing it right takes a proper UI designer (which I'm not), and none of my employers want to get me one.. ;) Aug 07 20:12:06 any UI I come up with will be half-assed at best Aug 07 20:12:44 best UI makes people dumber, actually bad for evolution Aug 07 20:13:15 na.. it's like any other kind of tool.. do I want to use it or do I want to know how it "works", so I know why it does what it does.. Aug 07 20:13:31 most of the time I just want to use it.... sometimes I want to know why it works.. :) Aug 07 21:59:26 what's the difference between do_stage and do_install Aug 07 21:59:30 ? Aug 07 22:00:34 do_stage is obsolete. Aug 07 22:00:42 perhaps you mean do_populate_sysroot? Aug 07 22:00:56 I'm looking at old recipes from oe_classic Aug 07 22:01:24 Specifically, I'm looking at one that has both do_stage and do_install, and wondering what the difference is Aug 07 22:01:47 close your eyes, look away Aug 07 22:03:15 ant__: there was some message from you Aug 07 22:03:16 for me Aug 07 22:03:20 what was it about Aug 07 22:03:39 some armv4t Aug 07 22:04:11 yes, kernel panic on collie Aug 07 22:04:24 http://paste.debian.net/182631/ Aug 07 22:05:24 khem: are you aware of http://repository.timesys.com/buildsources/g/gcc/gcc-4.7.0/ Aug 07 22:05:35 yes Aug 07 22:05:47 well no Aug 07 22:06:29 ok explain your problem and I can give you an answer Aug 07 22:06:46 kernel doesn't boot the initramfs Aug 07 22:07:08 but, same armv4t binary (init) can run on armv5te Aug 07 22:07:15 so is not miscompiled Aug 07 22:07:23 I see Aug 07 22:07:35 now, we could have 1) kernel misconfigured/buggy 2) toolchain issues Aug 07 22:07:42 can you post the log.do_compile of sysvinit ? Aug 07 22:08:02 if you get illegal intruction traps then its easy Aug 07 22:08:16 but if you get other sort of breakage then its hard to tell Aug 07 22:09:17 khem: add to this we use linux-yocto and we get a warning like Aug 07 22:09:20 Value requested for CONFIG_ARM_THUMB not in final ".config" Aug 07 22:09:30 hmmm Aug 07 22:09:36 from # Begin: meta/cfg/kernel-cache/arch/arm/arm.cfg Aug 07 22:09:43 is userspace compiled thumb ? Aug 07 22:09:59 heh, no, klibc still has it disabled... Aug 07 22:10:05 In do_install, is the root directory special or magic in some way? "install -m 0755 ${WORKDIR}/${HEXFILE} ${D}/${HEXFILE}" isn't doing what I expect it to. Aug 07 22:10:16 then it should not matter Aug 07 22:10:23 (initramfs is klibc-static) Aug 07 22:10:39 ok, that was my conclusion :) Aug 07 22:10:51 so the hproblem is in kernel Aug 07 22:10:59 awozniak: ${D} is not where you want it Aug 07 22:11:08 ${D} is essentially / on your target Aug 07 22:11:22 khem: yes, that is where I want it. Aug 07 22:11:48 awozniak: install -m 0755 ${WORKDIR}/${HEXFILE} ${D}/ Aug 07 22:11:52 should be it Aug 07 22:11:54 khem: JaMa pointed us to http://git.freesmartphone.org/?p=linux-2.6.git;a=commit;h=0333777365bef05b336d4f4a06bb3b37f4ada838 Aug 07 22:12:00 khem: than you Aug 07 22:12:04 thank you Aug 07 22:12:26 khem: but the kernel does boot, up to the init Aug 07 22:12:49 ant__: ok then it could be a different problem Aug 07 22:12:57 ant__: it means your kernel is ok Aug 07 22:13:13 strangely it cannot open console... Aug 07 22:13:32 ant__: as I said first if you post the log.do_compile of sysvinit it will give me better idea of whats going Aug 07 22:13:34 is collie already in qemu? Aug 07 22:13:44 was in -dev Aug 07 22:13:45 ant__: yes I think so Aug 07 22:14:02 ant__: oh but I use bleeding edge qemu not the OE-Core one Aug 07 22:14:26 khem: yeah, it's not getting installed. What else could be wrong? Aug 07 22:14:53 awozniak: installing is one thing, are you packaging it as well or not ? Aug 07 22:15:00 if not then add it to some package Aug 07 22:15:10 may be FILES_${PN} += Aug 07 22:15:25 thanks Aug 07 22:15:26 "/${HEXFILE}" Aug 07 22:16:01 khem: http://paste.debian.net/182648/ Aug 07 22:16:41 ant__: -march=armv4 -mthumb-interwork -mtune=strongarm1100 Aug 07 22:16:47 khem: I've applied the armv4 patch and the kernel patch above to gcc 4.7 Aug 07 22:17:27 khem: I've applied the armv4 patch to gcc 4.7 and the kernel patch (see above) Aug 07 22:17:34 :) Aug 07 22:20:52 khem: vanilla oe-core was http://paste.debian.net/182649/ Aug 07 22:21:37 so the changes are only under the hood Aug 07 22:36:26 khem: http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.dui0491e/CIHGEBDH.html Aug 07 22:36:36 4 ARMv4 without Thumb SA-1100 Aug 07 22:37:37 that's the arm compiler Aug 07 22:47:53 khem: possibly the bx instruction? **** ENDING LOGGING AT Wed Aug 08 02:59:58 2012