/srv/irclogs.linaro.org/2012/01/03/#linaro-android.txt

nicksydneyhi all...i'm trying to build the android linaro toolchain and i've successfully build gmp-5.0.2 but i'm having problem building ppl-0.10.200:09
nicksydneyi'm getting error --> "configure: error: Cannot find GMP version 4.1.3 or higher."00:09
nicksydneyi tried googling but can't find any solution that works at the moment00:10
nicksydneyi'm using Ubuntu 11.10 64bit00:10
doanacnicksydney: you might ask that question on #linaro or send an email to linaro-dev. a lot of guys are still on vacation today, but should start showing up online tomorrow00:17
nicksydneydoanac: oh ok ...cool..thanks for that00:17
asacnicksydney: bero would know more as well... he is the one doing our binary android toolchain builds.00:34
beronicksydney: What exactly are you doing? Are you using the linaro-build.sh script?00:35
beronicksydney: ppl does require gmp, but gmp is built first and should be found... Works when I'm doing my builds. If you're using the linaro-build script, try "find . -name config.log" to find the configure log files w/ more information on why it thought it couldn't find gmp00:36
asaci guess https://wiki.linaro.org/Platform/Android/Toolchain is the right place?00:41
asacbero: is that still up-to-date?00:41
beroLet me check...00:41
* asac wonders if we really need to keep the 00:42
asac"Alternative method (deprecated)" section00:42
asacmaybe good for education if thats still true00:42
beroSomewhat true00:42
beroThese days we should build arm-linux-androideabi- and not arm-eabi- compilers, ICS can't actually build with the bare metal compilers00:43
* bero updates00:43
asacnot sure if it justifies keeping it inline on this top level page though :)00:43
asacbero: thanks.00:43
beroTrue, I think we shouldn't encourage people to build it the old way00:45
beroAt least not people who don't know exactly what they're doing00:45
nicksydneyi will try from that link and see if i can built it00:45
asacnicksydney: bero said he has to do one update run to update instructions for our ICS toolchain00:46
asacso better wait a bit00:46
asac:)00:46
nicksydneyok thanks asac :D00:46
beroasac: Is this bit about the prebuilt toolchain binaries still true: "Note that they are provided as a preview and still undergo validation"? (Not sure what our official position on the stability is...)00:46
asachmm00:47
asacdifficult questions late at night00:47
asacwe dont mark them as preview anymore00:47
asacon the downloads page00:47
asacso i guess thats void00:47
asacand can be removed00:47
asacbero: at the bottom we reference jserv--'s benchmarks from really early times00:49
asaccan we add or move that link to the new place for doanac's benchmarks?00:49
asachmm00:50
beroSure, and we don't currently have a link to the bzr toolchains either, adding that too00:50
asacdoanac: i guess moving the Benchmark home from Platform/Android/AndroidToolchainBenchmarking/ to Platform/Android/Toolchain/Benchmarking/... would make sense but would be painful to do?00:51
asaci think we would need to add redirects00:51
doanacasac: wouldn't be too hard. like you said, we'd just need to keep redirects00:52
doanacI'll open a bug and assign it to myself00:53
asac:)00:53
beroOk, changes done00:59
asacnicksydney: so try https://wiki.linaro.org/Platform/Android/Toolchain now01:01
* asac off01:01
nicksydneythanks01:41
=== 18WAALUCN is now known as jkridner_
vk41hi all06:55
pundiramitbhoj, is it still open or valid for ICS? https://bugs.launchpad.net/linaro-android/+bug/86525808:07
ubot5Ubuntu bug 865258 in Linaro Android "adb on panda and staging-panda hangs when device goes to suspend mode" [High,Confirmed]08:07
bhojpundiramit, yes it is valid.08:07
pundiramitbhoj, ok08:08
nicksydneyhi all...i tried  to build the toolchain using the following page --> https://wiki.linaro.org/Platform/Android/Toolchain and i'm getting error configure: error: could not find a working compiler, see config.log for details08:18
nicksydneymake: *** [stmp-config-host-gmp] Error 108:18
bero|2nicksydney: So what does your config.log say?09:02
nicksydneythe config.log inside the /build directory ?09:02
nicksydneyhere is the config.log from the /build folder09:04
nicksydneyhttp://pastebin.com/eY7C4Pv309:04
bhojbero|2, ping09:25
nicksydneyany idea ?09:27
nicksydneyok found the problem...looks like there are missing libraries that need to be installed after doing the following from source.android.com it works09:39
nicksydneythis is the command that i execute for the libraries09:40
nicksydneysudo apt-get install git-core gnupg flex bison gperf build-essential \09:40
nicksydney  zip curl zlib1g-dev libc6-dev lib32ncurses5-dev ia32-libs \09:40
nicksydney  x11proto-core-dev libx11-dev lib32readline5-dev lib32z-dev \09:40
nicksydney  libgl1-mesa-dev g++-multilib mingw32 tofrodos python-markdown \09:40
nicksydney  libxml2-utils xsltproc09:40
nicksydneyit's able to move forward in the building process09:40
nicksydneyok hit another problem...09:57
nicksydneyit hit error when building PPL09:57
nicksydneychecking for version 0.10 of PPL... configure: error: Can't find correct version of PPL09:57
nicksydneynotice the PPL version being compiled and installed is 0.11.209:57
nicksydneylooks like the error occured while building cloog-ppl-0.15.909:58
nicksydneyfound an IRC logs with this problem http://irclogs.linaro.org/2011/06/22/%23linaro-android.html10:07
asachow can i easiest clone from a pinned manifest.xml that i have downloaded as file locally?10:20
bhojrepo init -u git://android.git.linaro.org/platform/manifest.git  -b linaro_android_4.0.3 -m tracking-panda.xml10:26
bhojcp pinnedmanifest.xml  .repo/manifests/tracking-panda.xml10:27
bhojrepo sync10:27
asacbhoj: ok. feels a bit hackish, but is probably right10:27
asacthanks!10:28
bhojasac,  or ... you can host a git and upload the pinnedmanifest there and then repo init with that git .10:28
manssonHi bhoj. Have you tried building Snowball with 4.03?10:31
asacbhoj: yeah. not sure if thats more convenient if you just want to reproduce a pinned manifest source10:31
bhojmansson, no I have not . but pundiramit  had tried it. are you running into any issue ?10:32
manssonYes, a bout a hundred errors in external/llvm/lib10:32
bhojmansson, okay . will run a trigger a build on hackbox to see if i can reproduce ... bero|2 has build for 4.0.3 snowball at https://android-build.linaro.org/builds/~berolinux/landing-snowball-4.0.3/10:35
manssonThanks. I'll look at beros. The 11.12 release toolchain  is OK to use?10:36
bhojyes. I guess . bero|2 's builds are based on 11.12 toolchain10:37
manssonI see that his is a landing-snowball. Mine is staging-10:38
bhojmansson, landing-snowball is the one with accelerated graphics .10:39
manssonYes. But that is no reason for staging- to fail, is it?10:40
bhojmansson, true . staging should still build. only difference between the 2 would be kernel .10:41
kenwsHi there, is the linaro_android_2.3.7 branch supposed to work (default.xml)?11:51
kenwsI'm not sure if it's an issue on my end (since just I migrated my build env to another machine)11:52
asackenws: what issue do you see?11:52
kenwsI did a repo init + sync on the 2.3.7 branch and kicked off a build using the android-toolchain-eabi-linaro-4.6-2011.12-5-2011-12-12_14-40-40-linux-x8611:52
asacthe idea is that it would still work. would be interesting to see why it doesnt11:52
asacah11:52
asackenws: 2.3.7 is gingerbread and we are on ICS now ... e.g. you have to use our 11.11 toolchain i guess11:53
asackenws: why is gingerbread so interesting for u?11:53
kenwsI'd like to look into the old debuggerd again11:53
kenws(this tiny patch I made to make use of libunwind)11:53
kenwshttp://paste.ubuntu.com/791550/11:54
kenwsthis looks quite similar to the issue we had with my Android.mk11:54
kenws(a missing slash)11:55
nicksydneyok after changing the ppl_minor_version variable inside the configure script according to the IRC log --> http://irclogs.linaro.org/2011/06/22/%23linaro-android.html ... the build process can move forward....can the documentation be modified to add this info so other devs know what to do if they find the same kind of problem ? ..i meant this page  https://wiki.linaro.org/Platform/Android/Toolcha11:55
* patrikryd remembers the day(s) when the slash was missing....11:56
kenwsIf I remember correct it was caused by using TOP_DIR in the Android.mk11:56
asackenws: yeah that looks odd.11:57
asackenws: if you have changed something i would guess that your changes are causing this :)11:58
kenwsis there a repo pendant to "git status"?11:58
kenwsI'm not aware of any changes...11:59
kenws... but what the best way to double check?11:59
kenws+is11:59
bhoj doing a repo sync  would show uncommited changes.12:00
kenwsok, thanks. repo sync didn't mention any changes12:00
kenwsso, I guess I'm on the head of the 2.3.7 branch12:01
kenwsis there something like a "git log" for the whole repo?12:01
kenwshm, maybe the gerrit webgui thingy...12:02
asaclooking at the manifest i only see the wallpaper thing not being fixed to a branch http://android.git.linaro.org/gitweb?p=platform/manifest.git;a=blob;f=default.xml;h=6e9d7f9c5d92a27803031b3d1228f6b55428649f;hb=refs/heads/linaro_android_2.3.712:04
asacso   32   <project path="packages/wallpapers/LinaroWallpaper" name="platform/packages/wallpapers/LinaroWallpaper" revision="master" />12:04
asachmm12:04
asacthere are more12:04
asacuboot, mmtest, 0xbench, glmark2 etc.12:04
asackenws: you could try starting with the 11.11 releaese pinned manifest instead of the tip 2.3.712:05
asacand see if that helps12:05
kenwsIs one of these the compiled using the autotools (calling configure and friends from the Android.mk)?12:06
kenws11.11 release - ok, I'll try12:06
kenwshm, same thing with the linaro-android-11.11-release - I guess It's something on my end then12:12
* kenws checks out the sources from scratch12:14
patrikrydkenws, there is 'repo status'12:35
kenwspatrikryd: ah, thanks!12:38
kenwshm12:38
kenwshttp://paste.ubuntu.com/791586/12:38
kenws"NO BRANCH"?12:38
bhojmansson_,  landing-snowball built fine on hackbox12:46
mansson_ok.12:47
mansson_My landing- build is soon done..12:48
bhojok12:48
mansson_I am suffering from unexpected shutdowns when building lately. I suspect it's some overheating monitor that kicks in. Very annoying.12:50
bhojmansson_, make -j4 is the max what my machine can sustain. i5 2nd gen proc with 4 gb ram.  anything above -j4 makes my laptop to reboot12:51
mansson_bhoj: Aha. I have i5 and 4GB too. I normally run -j4.12:54
patrikrydkenws, No branch is normal for repo. after a repo sync all git will be checkout on no branch.12:54
kenwspatrikryd: Do you know why this happens - is repo checking out a tree using a commit id?12:58
kenws(detached head)12:58
patrikrydkenws, yes12:58
patrikryda repo sync is almost like git init; git remote add linaro; git fetch linaro; git checkout some-sha-id for every git in the manifest.13:00
kenwsI find it quite awkward to work with a git tree with a detached head13:01
patrikrydthe first thing to do when you start modifying a git is to check out a branch :)13:03
patrikrydYou can use repo start or git branch for each branch13:03
kenwsrepo start or git branc? I'd think that one could just do a git checkout to a branch on a certain git tree13:04
patrikrydkenws, if you inside a git you can do whatever you are use to. :)13:08
kenws: )13:08
patrikrydrepo start is just a way of creating git branches on many gits13:09
kenwsyep13:09
kenwsthank you for the explanations patrikryd!13:09
=== nytowl_away is now known as nytowl
_Thomasnytowl: hi :)13:55
kenwspatrikryd: asac: A clean repo init/sync on the linaro_android_2.3.7 works fine. The diff -rupN between my old repo and the clean checkout is 154MB (yes, I did a make clean).14:43
kenwsI'm not sure what really went wrong but somehow a repo init + sync on my old code base was not able to get me what I get from a clean ceckeout14:44
kenws+to14:45
nytowlHi _Thomas14:57
_Thomasnytowl: What14:57
_Thomasargh14:57
nytowlHi :)14:58
_Thomasnytowl: What's the status of the OMX-port for Origen?14:58
nytowlIn android or Ubuntu ?14:58
_Thomasboth ;)14:58
_Thomas(yeah, I know this is the android chan...)14:58
asackenws: cool. thanks for confirm15:00
nytowl_Thomas, otp back in couple15:00
_Thomasok15:01
fabopundiramit: about bugs set to wontfix, how do we determine if something is gingerbread specific? or is that usually obvious?15:07
faboor have tried to reproduce the bug on ICS?15:07
faboasac: ^^^15:07
pundiramitfabo, I looked at the last release being used to verify the status and make sure that it is not a LT Kernel issue15:08
pundiramitfabo, if it is anything older than 11.11 it is put under "wont fix"15:09
nytowl_Thomas,  Annamalai is handling the MFC codec in Android so he will have to upadte you15:10
pundiramitfabo, Zach said if the bug still exist on ICS it would be open/reopen again15:11
asacpundiramit: hmm. are there no bugs we still want to fix?15:11
nytowl_Thomas,  for Ubuntu We have the firmware required now it will be a matter of just putting the pieces together15:11
asacpundiramit: ok i see. though i think its really unlikely someone will go a second time through won't fix15:11
asacbugs ... so the decision if thats still valid probably would need to be done on first (and only) cleanup run15:12
pundiramityes but most bugs were related to feature enablement15:13
pundiramitso instead of enabling them on GB15:13
pundiramiteffort will be put on ICS15:13
asacright. i assume you apply some common sense on what still makes sense so i guess things are OK :)15:15
asacthanks for doing the bug clean up!!15:15
pundiramitasac, no problem15:15
* asac feels android bug process in general probably deserves some fine tuning15:16
asacmore continuous ownership/triaging15:16
_Thomasnytowl: so you have the omx lib as well?15:32
nytowl_Thomas, I have a version for Ubuntu15:32
_ThomasYou do?15:34
_ThomasUsing the mfc version in lk 3.2?15:34
nytowl_Thomas, yeah, currently I', not sure of the distribution status other than the LT15:35
nytowl_Thomas, don't know haven't teasted with 3.2 yet15:35
tgall_foohmm so I guess ICS doesn't run on the panda es15:35
_Thomasnytowl: what did you test it with?15:35
nytowl_Thomas,  the last time I test it was 2.6.3815:36
_ThomasOk, I think they changed the mfc-driver radically at some point, and I think it's closer to 3.2 than 2.6.3815:37
_ThomasMeaning you don't have a working omx-driver in that case15:37
nytowl_Thomas, probably not15:37
=== vishal__ is now known as bhoj
BlInK311any specific reason for discontinueing support for the beagleboard xM?16:11
=== ramana is now known as ramana`
tgall_foohow/where does one get your hands on skia-bench for android?18:22
tgall_foois there something you have to turn on during the build or ?18:22
doanactgall_foo: I just enabled that in builds!18:28
tgall_foodoanac, cool ?  what's the magic?18:28
doanacI think it should be in today's daily build. here's the actual commit: http://review.android.git.linaro.org/#change,131518:29
tgall_fooI'll have a look. I've got a local build for my ltj stuff and was wanting to run that18:29
doanactgall_foo: i bet you can do a "repo sync" and pull in my change.18:30
doanactgall_foo: let me know if you do anything very interesting with skia_bench or have certain tests that you find particularly interesting.18:31
tgall_foodoanac, I figure it's the first layer of the onion, I'd like to see what effect ltj has on skia-bench18:32
doanacI'm working on "lava-tizing" it all. Here's an example run: http://validation.linaro.org/lava-server/dashboard/streams/anonymous/doanac/bundles/bf7550641ed6c11a338ede777ddf2c5872e28587/18:32
tgall_fooit SHOULD have an improvement18:32
doanactgall_foo: its probably easiest by hand. But I do have some scripts (not quite yet merged). that can take multiple lava runs and then produce charts to demonstrate performance changes18:33
doanaclet me know if that might help you.18:34
tgall_foodoanac, hmm  so it would appear that for the panda branch I'm on,  there isn't a skia_bench module at all18:39
tgall_foodoanac, I'd been pointed at : repo init -u git://android.git.linaro.org/platform/manifest.git -b linaro_android_4.0.3 -m landing-panda.xml    for the panda ? is that perhaps not the best branch for ICS on panda ?18:39
doanactgall_foo: crud. let me take a look. I thought I verified all builds would have skia.18:40
doanactgall_foo: I see landing-panda.xml has the skia repo. It should be under external/skia (with the benchmarks under external/skia/bench)18:42
tgall_fooah ok ? I was doing a find on the wrong term18:42
doanacFYI - if you just want to build skia_bench you can do something like "make <params.....> skia_bench18:44
=== jkridner_ is now known as jkridner

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!