- Welcome Fahad Kunnathadi
- 12.08 Release
- Review all BPs that are not done
- Next cycle
- 64-bit clean patches
- Result aggregation
- turn everything, benchmarks, tests, etc... into bundles
- Galaxy Nexus in LAVA
- Snowball enablement
- Using test results to make AOSP improvements
- Anything else?
Past Action items
- Action items that came out of this weeks meeting.
Zach Pfeffer <pfefferz>
- Virtual Connect
- VC Session 6: Android and Traditional Linux: Differences and Convergence Plans
- VC Session 9: A closer look at Dalvik and native performance on the Cortex A9 and A15
- VC Session 11: Adding support for new Android platforms in Linaro, Summer Edition (Part 1)
- VC Session 11: Adding support for new Android platforms in Linaro, Summer Edition (Part 2)
- Wrap up cards
- Start planning 12.09
- Power meter hacking
Write TINY_ANDROID wiki
- Plan 12.09
- Prepare for 12.08
Bernhard Rosenkraenzer <bero>
- Released merge-gnexus-blobs script
- Restored all strict-aliasing patches from ICS
- Updated docs for installing on gnexus
- Updated toolchain components: mpfr 3.1.1, mpc 1.0, Bionic from JB
- Built 12.08 toolchain
- Fix remaining blockers for 12.08
- Figure out why -O3 breaks libdex
Still need to properly housetrain the dog
Your Name <irc nick>
- Short bullet points of work you've done that week which convey
progress and highlights which can be used to report on how the team is progressing as a whole.
- Your individual plans for the coming week(s).
- Your individual plans for the coming week(s).
Zygmunt Krynicki <zyga>
I've wrapped my work on android-lava-test. It is a standalone tool that run entirely on the Android system and produces LAVA-compatible result bundles. The code is on github (github.com/zyga/android-lava-wrapper) and is tracked by my blueprint: https://blueprints.launchpad.net/linaro-android/+spec/aosp-tests-on-jb I'm working (as of this writing) to integrate it into our builds.
- I had a perfect holiday that has earned me the status on an Open Water Diver!
- See the android-lava-wrapper BP to completion so that we have daily results from the -tests build.
- I'm going back to Canonical next Month. I will will see you as a Community member the next time!
Fahad Kunnathadi <fahadkdi>
Going up with the ramp up stuffs with the Linaro build system.
And getting up a Android Icecream setup working with Origin board looking into the current code.
Doing some experiments with build stuffs in Linaro server.
- Got a linux ubuntu PC installed and accound with Linaro server.
- Need to check with current pending blueprint
- Have some issues in ports not enabled so coudent connect to hackbox account checking out with it.
Patrik Ryd <patrikryd>
Created the [[https://android-build.linaro.org/builds/~linaro-android/snowball-jb-gcc47-linaro-tracking-open/| snowball-jb-gcc47-linaro-tracking-open ] build.
- Have a almost working MALI accelerated JB for Snowball internally at ST-Ericsson. Need to fix some crashes and deliver.
- Fix the MALI (EGL crashes) problem in the ST-Ericsson internal snowball build
- Deliver patches and new vendor tarball
- Get the tracking build for snowball to actually build.
Need stronger coffee.
Kejun Zhou <kejun-zhou>
- Updated ECM patches.
- Help Hongbo (PMWG) to fix the new version pm-qa issue in ICS.
- Fix the bug in ECM (if there was).
Vishal Bhoj <bhoj>
- Worked on capturing benchmark results in console for 4 apps.
- Virtual Connect
- Complete capturing benchmark results and work with liuq to get them on android-build page.
Yongqin Liu <liuyq>
- Worked on integrated methanol test into android-build and lava production instances
- Virtual Connect
- Did a little investigation on CTS test
- Created lava-android-test branch for bhoj's benchmark test
- submit and deploy the lava-android-test branch for bhoj's benchmark test
- displayed the methanol test results on android-build page of each build
- reproduce the CTS problem that consumes CPU
- All the builds are broken for build from android-build, so can't verify the methanol test on android-build, and try the latest CTS test on latest JB builds
Android Team Meeting IRC log
<pfefferz> #startmeeting <linarobot> Meeting started Wed Aug 22 12:58:54 2012 UTC. The chair is pfefferz. <linarobot> Useful Commands: #action #agreed #help #info #idea #link #topic. <pfefferz> hey everyone <bhoj> hi <fahadkdi> hi <patrikryd> Hej! * mpoirier (~quassel@S0106002369de4dac.cg.shawcable.net) has joined #linaro-meeting <liuyq> hi <akgraner_> hi <pfefferz> hey its fahadkdi ! * jackh (~firstname.lastname@example.org) has joined #linaro-meeting <fahadkdi> hi yes pfefferz :) <pundiramit> hello o/ <pfefferz> hey pundiramit <pfefferz> bero, around? <pfefferz> no kejun? <pfefferz> okay.... <pfefferz> how is everyone? <pfefferz> zyga, ? <fahadkdi> fine going :) pfefferz <zyga> here <zyga> busy but ok :) <pfefferz> hey zyga <pfefferz> :) * patrikryd is fine, but needs more coffee. :) * pfefferz agrees <pfefferz> #agenda https://wiki.linaro.org/Platform/Android/Meetings/2012-08-22 <pfefferz> #topic Welcome Fahad Kunnathadi <pfefferz> hey fahadkdi, welcome <patrikryd> fahadkdi, welcome :) <pfefferz> would you introduce yourself? <fahadkdi> that was supposed to be last .. but i missed last week because it was a holiday over here :) thank you for the intro . <pfefferz> fahadkdi, would you share where your from, etc... <fahadkdi> Actually in bangalore . <fahadkdi> I ve but soemthing about me in the Meet me in the team <pfefferz> cool <fahadkdi> well I have done some contributions to Linaro too before <pfefferz> you're coming to us form Samsung right? <fahadkdi> yes in samsung <pfefferz> #link http://www.linaro.org/linux-on-arm/meet-the-team/fahad-kunnathadi/ <pfefferz> well welcome <fahadkdi> thank you <pfefferz> its good to have you here <fahadkdi> yup <jackh> fahadkdi: hey <fahadkdi> hi all and jach :) <pfefferz> okay <pfefferz> #topic 12.08 Release <pfefferz> #topic Baselines <pfefferz> bhoj, hows the bootloader issue going? <bhoj> pfefferz, jcrigby has reverted the commit and it looks good for stable-panda now . <pfefferz> nice! <pfefferz> bhoj, what commit blew things up? <bhoj> pfefferz, http://git.linaro.org/gitweb?p=boot/u-boot-linaro-stable.git;a=commit;h=f3f98bb0b8cc520e08ea2bdfc3f9cbe4e4ac29f5 <bero> pfefferz: Around now, just had a little doggy emergency ;) <pfefferz> :) <pfefferz> np bero <pfefferz> OMG <pfefferz> look at that commit message <pfefferz> ARM: OMAP4/5: Do not configure non essential pads, clocks, dplls. <pfefferz> Currently on OMAP4/5 platforms, many kernel drivers are dependent <pfefferz> upon the bootloaders for mux, dpll and clock configurations. <pfefferz> This should not be the case and bootloaders should set only the <pfefferz> minimum required for the uboot functionality and kernel boot. <pfefferz> Note that this is going to break the kernel drivers. But this <pfefferz> is the only way to get things fixed in the kernel. <pfefferz> Signed-off-by: R Sricharan <email@example.com> <bero> We should probably backport the kernel fixes at some point <pfefferz> aye <bhoj> do we really want to invest time in stabilizing things on 3.2 kernel . <pfefferz> bhoj, did jcriby send a mail to Andy and the TI guys? <pfefferz> bhoj, probably not at this point * kejun-zhou (~firstname.lastname@example.org) has joined #linaro-meeting <bhoj> pfefferz, no . this failure was seen only on 3.2 kernel and not on 3.4 kernel <pfefferz> bhoj, but its the only thing we've got that kind of stable <pfefferz> bhoj, cool <pfefferz> hmm... <pfefferz> sounds like there's no follow up then <pfefferz> ? <bhoj> pfefferz, yes . I think if things start breaking due to latest u-boot we should stick to older u-boot for our stable branch. <pfefferz> sounds good <bhoj> pfefferz, yes . I think we are good for now . jcrigby is going to keep that patch reverted in the master branch . <pfefferz> okay, and is he going to talk to the kernel devs at all <pfefferz> or just leave it <bhoj> pfefferz, he had a discussion with ndec on this and decided on reverting that patch . <pfefferz> okay <pfefferz> cool <pfefferz> thanks for following up <patrikryd> Have you seen https://android-build.linaro.org/index today? <pfefferz> bero, how are the toolchain patches looking <bero> Things are looking good <pfefferz> cool <pfefferz> patrikryd, what's up with it? <patrikryd> I don't there are any JB builds that are ok <pfefferz> looks like an infrastructure issue <pfefferz> or a LAVA outtage <pfefferz> one has <bero> Last night there was a problem because -fno-strict-aliasing broke the obsolete 4.5 host compiler we're using in the build roots <pfefferz> frameworks/base/libs/androidfw/AssetManager.cpp:319:40: error: dereferencing type-punned pointer will break strict-aliasing rules <bero> I just pushed a fix for that <pfefferz> :) <pfefferz> bero <pfefferz> would you kick all the baselines? <patrikryd> The few I hjave checked are strick aliasing related... <bero> Sure <bero> I think I'll just add -fno-strict-aliasing to the HOST compiler flags <pfefferz> thanks! <bero> It's not like we care about performance there a lot <pfefferz> hehe <bero> And we never know what sort of broken HOST compilers people come up with <bero> With 4.7-2012.08 as HOST compiler, everything is fine already <pfefferz> :) * akira_t has quit (Quit: Ex-Chat) <pfefferz> pundiramit, bhoj everything looking good with out vexpress stuff? <bhoj> yes . integrated switcher build for rtsm/fastmodel look good . <pfefferz> nice <pfefferz> thats, vexpress-rtsm-jb-gcc47-armlt-stable-open, right? <pundiramit> thats reference switcher build <pundiramit> integrated switcher build is the private one * pundiramit looking for link <pundiramit> pfefferz, here it is https://android-build.linaro.org/builds/~linaro-android-restricted/vexpress-rtsm-isw-jb-gcc47-armlt-stable-open/ <pfefferz> cool <bhoj> pfefferz, that should boot since kernel and boot loader have not been updated for months . will let you know <pfefferz> bhoj, thanks! <pfefferz> please take a look at the instructions <pfefferz> for both <pfefferz> vexpress-jb-gcc47-armlt-tracking-open <pfefferz> and <pfefferz> vexpress-rtsm-jb-gcc47-armlt-stable-open <bhoj> ok <pfefferz> and for the integrated switcher as well :) <bhoj> pfefferz, I can't verify this vexpress-jb-gcc47-armlt-tracking-open since I don't have a vexpress board . <pfefferz> sure <pfefferz> I think some one got it <pfefferz> let me look quick <bhoj> pfefferz, will check amit khare who has the board . <bhoj> pfefferz, will check amit_khare who has the board . <pfefferz> botao got, https://android-build.linaro.org/builds/~linaro-android/vexpress-jb-gcc47-armlt-tracking-open/#build=16 <bhoj> ok . <pfefferz> bhoj, would you put some instrucitons up for it though <bhoj> ok <pfefferz> thanks <pfefferz> cool, that should cover us for the release <pfefferz> bero, did you have a chance to write a blog post about the Galaxy Nexus <bero> Not yet <bero> I wanted to make sure we have a 100% build first <bero> Will probably do it later today <pfefferz> k <pfefferz> cool <pfefferz> akgraner, ^^^^ <pfefferz> that'll be an easy to use <pfefferz> Galaxy Nexus <pfefferz> with linaro bits <pfefferz> patrikryd, hows our snowball stuff going <pfefferz> staging ready to release? <patrikryd> except that it did not build today. :9 <patrikryd> s/9/) <patrikryd> I think it is ok <patrikryd> Nothing has really been changed for a while, except the changes that are for all boards <pfefferz> right <patrikryd> There will not by any MALI code in this release. <pfefferz> did ushit get back to you with the tracking build stuff <patrikryd> Only software rendering. <pfefferz> right <patrikryd> no <bero> What's the problem w/ MALI? Incompatibility with JB or something else? <patrikryd> I don't think we should release (or mention) the tracking build yet since it does not even build. <pfefferz> k <pfefferz> please send him the mails we mentioned in the con call <bero> I can take a look at the build failures once the other bits are fixed <pfefferz> to gfet him started <pfefferz> aye <pfefferz> and send the build failures around <pfefferz> if you spend some time getting ushit ready to go <pfefferz> then he'll be able to work on his own a bit <patrikryd> bero, we have some crashes that probably but they were probably there in ICS as well and then I need some time to navigate around the obstacles in the delivery process to get stuff out in the open . <patrikryd> All contact ushit <pfefferz> cool <patrikryd> I'll contact ushit <pfefferz> please get him going ASAP <pfefferz> okay <pfefferz> BPs <pfefferz> https://launchpad.net/linaro-android/+milestone/12.08 <pfefferz> first <pfefferz> everyones really doing a good job keeping these thigns updated <pfefferz> thank you * vireshk (~email@example.com) has joined #linaro-meeting <pfefferz> fahadkdi, have you been able to update, https://blueprints.launchpad.net/linaro-android/+spec/new-starter-tasks-fahad-k <fahadkdi> yes i was checking it <pfefferz> cool <pfefferz> see if you can mark it <pfefferz> pundiramit, would you update, https://blueprints.launchpad.net/linaro-android/+spec/versatile-fastmodel-jellybean <pfefferz> I think thats Implemented <pfefferz> once bhoj puts the doc in place <jackh> fahadkdi: you are the expert of hdmi, especially the android implementation of hdmi? <pfefferz> patrikryd, I think https://blueprints.launchpad.net/linaro-android/+spec/snowball-jellybean is probably done, would you spin up a BP with anything thats not gonna get done? <fahadkdi> i do know somethings about HDMI , but the HDMI ip is bit complex <patrikryd> I have a few minor things, but I can close it today... <pfefferz> liuyq, is https://blueprints.launchpad.net/linaro-android/+spec/android-arm-browser-benchmarks done? <patrikryd> bero, can you have a look at http://review.android.git.linaro.org/2659 <patrikryd> ? <fahadkdi> jackh :) <pfefferz> kejun-zhou, hey <kejun-zhou> hi <fahadkdi> also ppl are tending to go to DRM HDMI sort now leaving the V4L drivers <pfefferz> kejun-zhou, were you able to rebase ECM? <liuyq> pfefferz: patrikryd, will the wifi be supported on snowball-jb in this cycle? <patrikryd> pfefferz, I have move everything except the things I expect to close today. <bero> patrikryd: Sure, looking right now <pfefferz> patrikryd, did you create another BP? <kejun-zhou> yes. and patrikryd commented the gerrit <patrikryd> liuyq, No. <patrikryd> pfefferz, No. You did. Two actually. :) <liuyq> pfefferz: no, still wait the build on android-build to work for verifying the display <pfefferz> liuyq, you don't need to block finishing yours until Snowball works <pfefferz> liuyq, cool <kejun-zhou> the ECM patches are all on the linaro_android_4.1.1 <pfefferz> kejun-zhou, right, but were you able to rebase them so they'll merge <pfefferz> ? <pfefferz> bero, is there anything to do for https://bugs.launchpad.net/bugs/1031231 <ubot2`> Ubuntu bug 1031231 in linaro-android "Need binary blob overlay for Galaxy Nexus" [Critical,New] <patrikryd> kejun-zhou, My comment in gerrit was on white space. Ignore my comments. :) It's better to get ECM support than to have a few tabs converted to space. <pfefferz> liuyq, I'll move https://bugs.launchpad.net/bugs/1034218 to 12.09 <ubot2`> Ubuntu bug 1034218 in linaro-android "CTS Java process consume all the CPU on control and cause lab outages" [High,New] <kejun-zhou> I don't know why the rebase is needed and the ECM just on the linaro_android_4.1.1 <pfefferz> Gerrit says its needed <pfefferz> so you need to pull a clean tree <pfefferz> rebase the patches on it <pfefferz> fix any conflicts <pfefferz> and repush <pfefferz> after testing <kejun-zhou> is it the ECM should be on the tag linaro_android_4.1.1? <kejun-zhou> should the ECM be started from the tag linaro_android_4.1.1? <pfefferz> kejun-zhou, it should be on that branch, yes <pfefferz> but <pfefferz> that branch has moved on <pfefferz> soe the patches you sent <pfefferz> no longer apply cleanly <pfefferz> Gerrit tells you this <pfefferz> which is why <pfefferz> you need to pull a clean tree <kejun-zhou> And I started the three ECM all the latest tag linaro_android_4.1.1 <pfefferz> or do a repo sync <kejun-zhou> Yes <pfefferz> right <kejun-zhou> I did that <pfefferz> k <pfefferz> lets take a look <kejun-zhou> the gerrit also have the tag information is tag linaro_android_4.1.1 <pfefferz> okay I see the rebase <pfefferz> and the new push <pfefferz> once the builds work again <pfefferz> I'll try merging them again <pfefferz> kejun-zhou, would you rebase the other changes <pfefferz> http://review.android.git.linaro.org/#change,2598 <pfefferz> and <pfefferz> http://review.android.git.linaro.org/#change,2554 <pfefferz> just to be sure <pfefferz> ? <kejun-zhou> Can't be merged? <pfefferz> well if the first one couldn't be merged <pfefferz> the dependent patches may not be mergable either <pfefferz> and since we can't merge until the builds are working again * linarobot has quit (Remote host closed the connection) <pfefferz> may as well rebase the other patches <kejun-zhou> I think I need the right to merge <kejun-zhou> When my patches need to be reviewed, other patches are mergeing <pfefferz> kejun-zhou, that's just how concurrent development work kejun-zhou <pfefferz> kejun-zhou, that's just how concurrent development works kejun-zhou <kejun-zhou> I have not chance to make my patches be involed <pfefferz> some patchset can be rebased 20 times before they make it in <pfefferz> okay, please rebase the other changes <pfefferz> in preperation <pfefferz> then hopefully everything will just go in <zyga> heh, linaro certainly taught me about history-rewriting parts of git <pfefferz> #topic TINY_ANDROID <pfefferz> I wrote up, https://wiki.linaro.org/Platform/Android/UseTinyAndroid <pfefferz> thanks for spinning up the TINY_ANDROID baseline benonsoftware <zyga> pfefferz, that's cool stuff zach! <pfefferz> thanks for spinning up the TINY_ANDROID baseline bero <zyga> pfefferz, maybe we could build tiny android too? <pfefferz> expect to see more about TINY <pfefferz> yeah <pfefferz> the one thing we need to do <pfefferz> is create stripped down manifests <pfefferz> so the check out is fast * rajagopal (~firstname.lastname@example.org) has joined #linaro-meeting <zyga> ++! <pfefferz> we're creating these <patrikryd> How to you build it locally? Do you set TARGET_PRODUCT? <zyga> on a separate note I'd love to know how to keep a some kind of cache so that clean repo sync finishes faster <pfefferz> for kernel devs and other low level people <pfefferz> patrikryd, take a look at the wiki <jackh> tiny android is great for debugging some apps <pfefferz> zyga, we have it <pfefferz> zyga, if you download the tarball <jackh> but it can only run on linaro boards? <patrikryd> pfefferz, I did :) <pfefferz> then repo sync it <jackh> it should be universe <patrikryd> The I checked the build page. <pfefferz> thats as fast as you can get <zyga> pfefferz, which tarball? <patrikryd> s/The/Then <pfefferz> REPO_SEED_URL=http://android-build.linaro.org/seed/uniseed.tar.gz <pfefferz> okay, almost out of time <zyga> pfefferz, yeah but I mean a general solution for that <pfefferz> next cycle is going to be awesome <zyga> for git clones <pfefferz> ight <pfefferz> right <pfefferz> 64-bit Android <pfefferz> finishing up benchmarking <pfefferz> 64-bit clean patches <pfefferz> Result aggregation <pfefferz> turn everything, benchmarks, tests, etc... into bundles <pfefferz> Galaxy Nexus in LAVA <pfefferz> Snowball enablement <pfefferz> Bugs <pfefferz> system_server <pfefferz> CTS <pfefferz> Using test results to make AOSP improvements <pfefferz> :) <pfefferz> #topic Anything else? <zyga> what do you mean by bundles? <zyga> anyway I guess that's next month anyway * rnayak (~a0131687@nat/ti/x-ajbwjnltvusyxxfu) has joined #linaro-meeting <pfefferz> zyga, I mean <pfefferz> that all the results <pfefferz> get aggrgated <pfefferz> and to do that <pfefferz> doanac, tells me we need to get all the results into LAVA bundles <pfefferz> okay <pfefferz> thanks for a great meeting guys <pfefferz> #endmeeting
Platform/Android/Meetings/2012-08-22 (last modified 2012-08-22 14:02:21)