- Welcome back Amit!
- Vishal and Amit to talk about meeting at ARM
- Please make sure BPs are marked Implemented and all remaining work is in bugs linked to the BP or talk to Zach to split
- 12.09 is shaping up. Please look through your BPs and add any WIs or other usful info.
- Will be getting more things scheduled this week.
- We need to get all baselines booting linux-linaro kernels
- maybe we shoudl specifically call out the builds as such?
- Vishal to give an update on future variance work
- Patrik and Kejun to give an update on linux-linaro snowball
YongQin to talk about browser work
- Amit to talk about CPU idle test
- Fahad to talk about how his starter tasks are going
- big.LITTLE status
Past Action items
- Action items that came out of this weeks meeting.
Zach Pfeffer <pfefferz>
- Helped wrap up 12.08
- Started planning 12.09
- Research 64-bit Android Planning
- Helped get ECM in
- Made sure build usage instructions were acurate
- Plan 12.09
- Attend LinuxCon/Plumbers
- Finish research on preliminary 64-bit Android draft plan
- Work on power measurement system
- Verified the methanol test on android-build
- Helped to modify and verify the lava-android-test for display monkeyrunner result on android-build
- Talked with zyga about the blackbox test
- Work on the CTS consuming bug
- work on the BP "Aggregate Browser Benchmark Data"
Zygmunt Krynicki <zyga>
- Wrapping up blackbox tests for android. I finally have a working merge proposal for lava-android-test integration. Everything else is ready for adding to our manifests. I hope to see some results this week. * Last meeting as a Linaro assignee!
- Hop back in as community next week
- All good
Fahad Kunnathadi <fahadkdi>
- Got hackbox.linaro.org accound ready and able to loggin to it.
- Did download, android platform from linaro repository and builded it for origen .
- Flashed tar.bz2 to SD card, and run on origen.
- Follow the BP remaining tasks
- Android not booting up, getting some zygote exit in logcat.
Vishal Bhoj <bhoj>
- Benchmark results for Antutu,AndEBench,Linpack,Caffeinemark now show up on android-build page.
- Meeting with ARM on current benchmarking status and future work.
- Parse GLBenchmark,nBench results.
- Characterization of benchmarks.
Amit Pundir <pundiramit>
- Ported cyclictest on Android
- Meeting at ARM.
- Sysbench Android port
- Update VE-A9 firmware
- Update b.L scripts for clean exit of bl-agitator
- DS-5 integration in LAVA
- look into lava-android-test wrapper of Zyga
- look into b.L-MP kernel based Android builds
Patrik Ryd <patrikryd>
- Working on the MALI for snowball
- Have something working
- Have successfully navigated the mine field called the FOSS process to be able to deliver
- Currently stuck in the delivery process. (Large companies have processes for everything)
- Get mali for snowball out
- Get the snowabll tracking build to build.
- Stuck in a process because the internal snowball team have not done what the had planned to do. I have to do their work or invent shortcuts or workarounds to get stuff out in the open.
Kejun Zhou <kejun-zhou>
Fixing the https://bugs.launchpad.net/linaro-android/+bug/1042239. The dhcpcd crashed every time when the cable was not plugged from power on.
- Fix the bug 1042239.
Android Team Meeting IRC log
12:58:58 <bhoj> #startmeeting 12:58:58 <linarobot> Meeting started Wed Aug 29 12:58:58 2012 UTC. The chair is bhoj. 12:58:58 <linarobot> Useful Commands: #action #agreed #help #info #idea #link #topic. 12:58:58 <bhoj> Hi everyone 12:58:58 <pundiramit> hello o/ 12:58:58 <patrikryd> Hej! 12:58:58 <fahadkdi> Hi 12:58:58 <zyga> hey 12:58:58 <zyga> :-) 12:58:58 <fahadkdi> :-) 12:58:58 <bhoj> no kejun ,liuyq ? 12:58:58 <liuyq> hi 12:58:58 <bhoj> hi :) 12:58:58 <bhoj> since pfefferz is at plumbers I will be running todays meeting 12:58:58 <bhoj> #agenda https://wiki.linaro.org/Platform/Android/Meetings/2012-08-29 12:58:58 <bhoj> #topic Welcome back Amit! 12:58:58 <pundiramit> :) 12:58:58 <bhoj> welcome back pundiramit :) 12:58:58 <bhoj> we are excited to have you back in android team 12:58:58 <zyga> so you've moved back :-) ? cool 12:58:58 <pundiramit> same here :) 13:03:01 <pundiramit> will still be doing some lava work in the background 13:03:01 <bhoj> cool . I hope you had an exciting time working in LAVA team . 13:03:01 <pundiramit> have some BPs aligned towards lava integration 13:03:01 <pundiramit> yes I had a good run there :) 13:03:01 <bhoj> ok . I think you will be able to help the lava team while working in android team. 13:03:01 <pundiramit> I'll try my best :) 13:03:01 <bhoj> #agenda Vishal and Amit to talk about meeting at ARM 13:03:01 <bhoj> We had a meeting with the ARM's benchmarking team at Bangalore to discuss the current work and future plan on benchmarks 13:03:01 <bhoj> It was a very useful meeting . 13:03:01 <bhoj> we gave a demo of our automated thirdparty benchmark suite 13:03:01 <bhoj> and they plan to use it internally in their TC2 setup. 13:03:01 <bhoj> We spoke on the current status on the automation and the plan to get the results on android build page using hooks . 13:03:01 <bhoj> as part of the next step they want us to characterize the benchmarks 13:03:01 <bhoj> to classify them as which benchmark stresses which part of the system 13:03:01 <bhoj> as in SMP, bionic, dalvik and other android components 13:03:01 <bhoj> which will then help us to demonstrate improvements wrt to optimizations in those components 13:03:01 <bhoj> pundiramit, do you want to add anything ? 13:03:01 <pundiramit> They also shared their benchmarking setup and analysis of few benchmarks, which was very useful. 13:03:01 <zyga> pundiramit: do they do things differently? 13:03:01 <pundiramit> differently as in, they are doing manual iterations. 13:03:01 <zyga> what are manual iterations? 13:03:01 <pundiramit> running tests manually 13:10:46 <bhoj> they explained the methodologies they used in characterizing the benchmarks . 13:10:46 <pundiramit> and then collecting rsults 13:10:46 <pundiramit> and then collecting results 13:10:46 <liuyq> pundiramit: could we share it in linaro? 13:10:46 <zyga> pundiramit: did they explain how do they collect and gather results from manual testing? 13:10:46 <pundiramit> they plan to use our lava setup to carryout these tests via automation 13:10:46 <bhoj> zyga, they ran the benchmarks manually on different end devices and then used them to characterize the benchmarks . 13:10:46 <pundiramit> liuyq, zyga methodology is open but actual benchmarking numbers were not disclosed 13:10:46 <bhoj> liuyq, they cannot share their results . They want us to repeat the exercise to have the characterization in the open . 13:10:46 <pundiramit> liuyq, numbers can be shared privately I assume 13:10:46 <zyga> pundiramit: are they using anything related to lava to store data? 13:10:46 <pundiramit> zyga, no. they manually collect results via screenshots, logcat dump etc 13:10:46 <zyga> I see 13:10:46 <zyga> thanks 13:10:46 <pundiramit> thats where linaro's role kick in 13:10:46 <pundiramit> thats where bhoj's benchmarking automation work comes in. 13:10:46 <bhoj> zyga,they don't plan to use specific to lava . but they plan to use lava-android-test to run these benchmarks . 13:13:59 <pundiramit> they plan to use lava to automate their benchmarking exercise. 13:13:59 <zyga> bhoj: I was interested in their attitude towards storing the data in the dashboard 13:13:59 <bhoj> zyga, since they plan to use only the automation of benchmarks and tools to extract the results from bundles lava-android-test was sufficient to do the process 13:13:59 <pundiramit> zyga, bhoj I doubt if they do it openly on dashboard, but plan is to share the results with Linaro and work together to improve them 13:13:59 <zyga> pundiramit: it does not have to be open, it can be a private dashboard internal to arm 13:13:59 <pundiramit> zyga, right but as bhoj mentioned their primary goal is to run the tests automatically, they may not be interested in complete lava suit. 13:13:59 <pundiramit> but they might end up using it. who knows :) 13:13:59 <zyga> sure I was only asking 13:13:59 <bhoj> zyga, good point . we will propose it to them on this in the next meeting . 13:13:59 <bhoj> next meeting with ARM 13:13:59 <zyga> especially as we approach data processing doing it in an uniform way could be beneficial 13:13:59 <bhoj> moving to the next topic 13:13:59 <bhoj> #topic 12.08 release 13:13:59 <bhoj> Please make sure BPs are marked Implemented and all remaining work is in bugs linked to the BP or talk to Zach to split 13:13:59 <bhoj> #link https://launchpad.net/linaro-android/+milestone/12.08 13:13:59 <zyga> ok 13:13:59 <bhoj> all the BPs look in good state 13:13:59 <bhoj> zyga, is there anything to be done on this BP https://blueprints.launchpad.net/linaro-android/+spec/aosp-tests-on-jb ? 13:13:59 <zyga> well, kind of 13:13:59 <zyga> I'd like to move that to backlog 13:13:59 <zyga> as this blueprint is the leftover of the split 13:13:59 <zyga> and it's not likely to get finished entirely this cycle 13:13:59 <bhoj> zyga, cool . Talk to pfefferz on that . 13:13:59 <zyga> as most of my work went into the part that automated the tests I've discovered 13:13:59 <zyga> I will 13:22:32 <zyga> (whooping 100+ tests with many more test cases :-) 13:22:32 <bhoj> it was a humongous task :) 13:22:32 <bhoj> anything else on 12.08 ? 13:22:32 <bhoj> ok moving to the next topic 13:22:32 <zyga> not from me 13:22:32 <bhoj> #topic 12.09 planning 13:22:32 <bhoj> #link https://launchpad.net/linaro-android/+milestone/12.09 13:22:32 <bhoj> Few of the BPs have been filed and I think Zach will file few more 13:22:32 <bhoj> please fill in the relevant WI in the BPs 13:22:32 <bhoj> We need to get all baselines booting linux-linaro kernels 13:22:32 <bhoj> I think we will be adding build for all the boards based on linux-linaro kernel 13:22:32 <bhoj> any inputs on this ? 13:22:32 <zyga> what kind of baselines? 13:22:32 <zyga> performance? 13:22:32 <bhoj> these are the tracking builds 13:22:32 <patrikryd> The snowball tracking build is based on linux-linaro 13:22:32 <patrikryd> but it does not build at the moment. :( 13:22:32 <bhoj> we are missing origen-jb tracking build 13:22:32 <bhoj> the target is to have the builds ready for this cycle. Even if they don't have display . 13:22:32 <fahadkdi> what is tracking builds basically ? 13:22:32 <liuyq> bhoj: sorry, linux-linaro kernel? is that a kernel that with support for all the boards(incluing panda/origen/snowball..)? 13:22:32 <bhoj> fahadkdi, tracking builds are the build which are based on the tracking kernel . i.e the latest kernel the LT are working on 13:22:32 <fahadkdi> okey 13:22:32 <bhoj> its usually bases on linux-linaro tree 13:30:04 <bhoj> fahadkdi, here is some info https://wiki.linaro.org/Platform/Android/LandingStagingTracking 13:30:04 <fahadkdi> so who all are working with origen in the team 13:30:04 <fahadkdi> origen and android in the team 13:30:04 <bhoj> fahadkdi, previously Annamalai and pundiramit looked at the origen build 13:30:04 <fahadkdi> okey, 13:30:04 <bhoj> fahadkdi, I had setup the stable-jb origen build based on the stable branch . 13:30:04 <bhoj> tracking-jb origen build was pending . 13:30:04 <fahadkdi> i did build one origen setup downloaded 13:30:04 <fahadkdi> today 13:30:04 <bhoj> liuyq, linux-linaro is the kernel with upstream effort in linaro 13:30:04 <fahadkdi> okey 13:30:04 <bhoj> liuyq, it should have the basic support for all the boards/platforms supported by linaro . 13:30:04 <bhoj> anything else on 12.09 topic ? 13:30:04 <bhoj> moving on to the next topic 13:30:04 <bhoj> #topic Vishal to give an update on future variance work 13:30:04 <bhoj> nothing much of an update here 13:30:04 <bhoj> I am currently analyzing benchmarks with the high variance to understand what is the reason behind it . 13:35:29 <bhoj> moving ahead in the future the plan is to add benchmarks 13:35:29 <bhoj> from toolchain team 13:35:29 <bhoj> which should showcase the toolchain improvements . 13:35:29 <bhoj> and as previously said 13:35:29 <bhoj> the next step with benchmarks is to characterize them and classify them 13:35:29 <patrikryd> bhoj, Is there any summary somewhere on what you have done so far to reduce the variance for benchmarks? 13:35:29 <bhoj> as which benchmarks stresses which part of the system it stresses . 13:35:29 <bhoj> patrikryd, yes . I had presented it during the dalvik session in VC. 13:35:29 <bhoj> patrikryd, the video is here https://plus.google.com/u/0/events/c9r67l6t01cgvgo6sjclfccqo00/116754366033915823792 13:35:29 <bhoj> patrikryd, the presentation link is here https://docs.google.com/a/linaro.org/file/d/1dpb_WLVe4hizFwkc1zcOueqnYW8BrTRXROdxBiKt5ywBHR_wLdpbNkfACdk8/edit 13:35:29 <bhoj> patrikryd, here is the right link https://docs.google.com/open?id=1dpb_WLVe4hizFwkc1zcOueqnYW8BrTRXROdxBiKt5ywBHR_wLdpbNkfACdk8 13:35:29 <patrikryd> bhoj, Found https://docs.google.com/a/linaro.org/file/d/1dpb_WLVe4hizFwkc1zcOueqnYW8BrTRXROdxBiKt5ywBHR_wLdpbNkfACdk8/edit 13:35:29 <patrikryd> ok same link :) 13:35:29 <bhoj> cool :) 13:35:29 <patrikryd> Thanks 13:35:29 <bhoj> moving on to the next topic 13:35:29 <bhoj> #topic Patrik and Kejun to give an update on linux-linaro snowball 13:35:29 <bhoj> patrikryd, can you update us since kejun is not around ? 13:35:29 <patrikryd> We have two builds. The tracking and the stable. 13:35:29 <patrikryd> The tracking does not build at the moment. The kernel does not compile. 13:35:29 <patrikryd> Build 36 of the stable build went into the release. It does not have mali support. 13:35:29 <patrikryd> I have an internal build at ST-Ericsson with mali support. 13:35:29 <bhoj> patrikryd, what is the reason for the failure for tracking ? 13:35:29 <patrikryd> net/built-in.o: In function `tcp_is_local6': 13:35:29 <patrikryd> /home/patrik-ryd/src/tracking/kernel/net/ipv4/tcp.c:3667: undefined reference to `rt6_lookup' 13:35:29 <bhoj> patrikryd, is it broken in linux-linaro tree itself ? 13:35:29 <patrikryd> Have not looked at it yet 13:35:29 <patrikryd> maybe 13:35:29 <patrikryd> I have asked kejun to help me, but he was busy with some ECM bug. 13:35:29 <bhoj> patrikryd, cool . please follow up with ushit and Andrey if that is the case . 13:35:29 <patrikryd> ok 13:35:29 <bhoj> patrikryd, any ETA on mali deliveries ? 13:44:53 <patrikryd> I have run into some problems internally when delivering. 13:45:05 <bhoj> ok . 13:45:05 <bhoj> running late . 13:45:05 <bhoj> moving to the next topic 13:45:05 <bhoj> #topic Amit to talk about CPU idle test 13:45:05 <patrikryd> I will find a workarond somehow. :) 13:45:05 <bhoj> pundiramit, how is it looking ? 13:45:05 <bhoj> patrikryd, you rock :) 13:45:05 <pundiramit> i think pfefferz meant CPU hotplug latency test and not CPU idle test 13:45:05 <bhoj> pundiramit, may be :) 13:45:05 <pundiramit> I made good progress on CPU hotplug test 13:45:05 <pundiramit> ported cyclictest to Android 13:45:05 <bhoj> cool :) 13:45:05 <pundiramit> I just need to find correct baseline to run this test on 13:45:05 <pundiramit> and sysbench need to be ported to android as well 13:45:05 <pundiramit> thats in to do list for this BP 13:47:42 <pundiramit> thats all from my side on this 13:47:42 <bhoj> ok . pundiramit pls check with amit kucheria on which platform has hotplugging support 13:47:42 <pundiramit> yeah already ping him 13:47:42 <bhoj> cool 13:47:42 <bhoj> moving to next topic 13:47:42 <bhoj> #topic Fahad to talk about how his starter tasks are going 13:47:42 <fahadkdi> hi yes 13:47:42 <bhoj> fahadkdi, how are things at your end :) ? 13:47:42 <fahadkdi> I did a origen build 13:47:42 <bhoj> looks like you are making good progress there . 13:47:42 <fahadkdi> and flashed on sd card and it started booting. 13:47:42 <bhoj> did you run into any issue while deploying ? 13:47:42 <fahadkdi> some issue for not seeing android 13:47:42 <bhoj> it could be the missing graphics libraries 13:47:42 <fahadkdi> i think some audi library not getting up 13:47:42 <bhoj> did you extract the vendor tarball in your source ? 13:47:42 <fahadkdi> i used the command 13:47:42 <fahadkdi> ./linaro_android_build_cmds.sh -m /full/path/to/pinned-manifest.xml -o 13:47:42 <fahadkdi> vendor.tar.bz2 13:47:42 <fahadkdi> with the paths given 13:47:42 <fahadkdi> properly 13:47:42 <fahadkdi> poped up with this issue in logcat /AudioFlinger( 2289): loadHwModule() error -22 loading module primary 13:47:42 <fahadkdi> W/AudioPolicyManagerBase( 2289): could not open HW module primary 13:51:58 <bhoj> fahadkdi, please share the crash log after the meeting we can help you with it :) 13:51:58 <fahadkdi> thats the crash log up .Okey 13:51:58 <bhoj> fahadkdi, that log not harmful . the versioning is wrong since the vendor tarball has audio libraries from ICS . 13:51:58 <fahadkdi> okey 13:51:58 <bhoj> fahadkdi, please share the complete logcat in a pastebin 13:51:58 <fahadkdi> okey will do it 13:51:58 <bhoj> moving to next topic 13:51:58 <bhoj> #topic big.LITTLE status 13:51:58 <bhoj> tixy has been maintaining the build for big.LITTLE MP for TC2 13:51:58 <bhoj> and it seems to be in good shape . 13:51:58 <bhoj> and for the big.LITTLE system project 13:51:58 <bhoj> we are yet to get a booting kernel for TC2 hardware to run our android builds . 13:51:58 <pundiramit> i have not looked into bL-MP for RTSMs for sometime now. I'll look into it as well 13:51:58 <bhoj> pundiramit, thanks :) 13:55:22 <bhoj> thats it from my side . 13:55:22 <bhoj> #topic AOB 13:55:22 <zyga> hmm :) 13:55:22 <zyga> I'll be adding blackbox tests to one build 13:55:22 <zyga> if it breaks please chase me 13:55:22 <bhoj> cool zyga :) 13:55:22 <bhoj> zyga, btw it was great having you in our team :) 13:55:22 <zyga> thanks, I always felt very welcome! 13:55:22 <zyga> and I'm not really going away that much 13:55:22 <zyga> I will stick around 13:55:22 <bhoj> thanks for all the efforts towards the blackbox test :) 13:55:22 <zyga> I guess I'll apply for that linaro email address 13:55:22 <bhoj> zyga, yes we look forward to continue working with you :) 13:55:22 <bhoj> zyga, yes it will be great to have you continue as a community developer :) 13:55:22 <zyga> I'm looking forward to it 13:55:22 <pundiramit> thanks for your help zyga on lava :) 13:55:22 <zyga> did you get FM into shape? 13:55:22 <bhoj> ok guys times up . lets continue the discussion on #linaro-android 13:55:22 <pundiramit> zyga, Andy is doing most of the integration. I was merely doing bL test integration and test runs 13:55:22 <bhoj> closing the meeting 13:55:22 <bhoj> #endmeeting
Platform/Android/Meetings/2012-08-29 (last modified 2012-09-07 13:10:34)