08:33:06 <bero> #startmeeting
08:33:06 <linarobot> Meeting started Thu Nov 26 08:33:06 2015 UTC.  The chair is bero. Information about MeetBot at http://wiki.debian.org/MeetBot.
08:33:06 <linarobot> Useful Commands: #action #agreed #help #info #idea #link #topic.
08:33:11 <yspan> hi
08:33:18 <bero> One piece of news today:
08:33:26 * liuyq1 is here
08:33:39 <bero> We might have a team sprint soon, probably in the 2nd week of December
08:34:04 <Qian__> That's a good news
08:34:34 <xruxa> hi all
08:34:37 <bero> Some details still need to be sorted out (including if it's the entire team or only people officially involved with the Spreadtrum/Tinno project, and location
08:34:55 <bero> if it happens it'll definitely be somewhere in China
08:35:18 <bero> so not a long trip for most of you
08:36:14 <bero> Details still need to be sorted out, but please try to keep the week free of other stuff for now
08:37:02 <zhizhoutian__> bero: good news. welcome all of you
08:37:26 <bero> Let's go to status updates...
08:38:02 <bero> Qian_: do you want to go first?
08:38:08 <Qian__> What did you do?
08:38:08 <Qian__> Merge libc patches on MX5 and PRO 5, Fix the conflictions
08:38:08 <Qian__> What are you going to do?
08:38:10 <Qian__> Test the performance data before and after
08:38:12 <Qian__> What problems do you have?
08:38:14 <Qian__> Are lib-bench.sh and stringbench.sh both used for bionic?
08:39:14 <bero> yes, libc-bench and stringbench both test the speed of functions in Bionic
08:39:29 <bero> they aren't needed, but can be helpful to look at some performance data
08:39:33 <Qian__> bero: Thank you
08:39:37 <sunao> Qian__, Hi  where can we find the lib-bench.sh ?
08:39:43 <bero> yspan: do you want to go next?
08:40:01 <Qian__> sunao, i found it in lave-server
08:40:07 <Qian__> lava-server
08:40:19 <sunao> thanks Qian__
08:40:43 <yspan> What did you do?
08:40:44 <yspan> Refactor libpng patch. Already sent v2.
08:40:44 <yspan> What are you going to do?
08:40:44 <yspan> Keep profiling any possible functions that could be segmented.
08:40:44 <yspan> What problems do you have?
08:40:45 <yspan> none
08:41:16 <xruxa> Qian__, is that part of LMG-914?
08:41:45 <bero2> thanks
08:42:06 <bero2> xavierhsu: do you want to go next?
08:42:11 <Qian__> xruxa: No, it's work of my company, will take one day or two
08:42:21 <xruxa> Qian__, ack
08:42:22 <xavierhsu> Bero, ok
08:42:44 <xavierhsu> What did you do? 1. The build-in app (Gallery) can not be used but others jpeg related app that can be used.   2. Testing the system's stable, the screen of system sometimes will only display black.
08:42:58 <xavierhsu> What are you going to do? 1. Debug the program.
08:43:14 <xavierhsu> What problems do you have? 1. Sometimes system will display black.  1. The build-in app (Gallery) can not be used.
08:43:26 <Qian__> xruxa: I will feed back some test results, I think both my company and linaro like it
08:44:07 <bero2> Qian__: yes, test results are always good
08:44:17 <bero2> xavierhsu: thanks
08:44:39 <xavierhsu> Bero, haveing lots of problem >"<
08:44:47 <xruxa> xavierhsu, you did not have time for getting rid of hardcoded -O flags yet? LMG-924
08:45:08 <bero2> liuyq: do you want to go next?
08:45:23 <bero2> xavierhsu: do you need help?
08:46:18 <xavierhsu> xruxa, yes@@
08:46:36 <xruxa> xavierhsu, bero2 -  ok, moving it out to next sprint.
08:46:48 <xruxa> unless someone one else did that and wants to do it still this week :)
08:47:49 <xavierhsu> bero2, I will try to find the problem, if I need the help, I will cry for help @@
08:48:11 <xavierhsu> Bero2, thanks.
08:48:13 <Qian__> xavierhsu: sorry, forgot to send the demo, i will send it right nwo
08:48:32 <bero2> Release is done and 924 doesn't have a delivery deadline, so it's not super important to do it this week - but if anyone is looking for extra stuff to do, that's one option...
08:48:39 <xavierhsu> Qian__, it's ok. Thanks.
08:49:13 <bero2> zhizhoutian__: do you want to go next?
08:49:32 <zhizhoutian__> What did you do?
08:49:32 <zhizhoutian__> - talk with sumit, ask for some advice, his reponse:
08:49:32 <zhizhoutian__> * swapping pages 1 page at a time (v/s default 8) by changing init.rc: `write /proc/sys/vm/page-cluster 0`
08:49:32 <zhizhoutian__> kernel swaps default 8 pages at a time to swap space; this will make it swap 1 page at a time, and can help when system is under memory pressure
08:49:32 <zhizhoutian__> * use CMA(continue memory allocate)
08:50:02 <zhizhoutian__> bero2: Some of them i do not know. So i need to investigate them
08:50:30 <zhizhoutian__> bero2: expecialy the second one.
08:51:12 <zhizhoutian__> bero2: everytime zram compress data, it will continue them first.
08:51:17 <bero2> that could be interesting... Sure, take some time to investigate
08:51:33 <zhizhoutian__> bero2: i mean continuously pages
08:51:45 <zhizhoutian__> bero2: yes, thanks
08:51:52 <xruxa> zhizhoutian__, LMG-930? Done or not going to be done? :)
08:52:04 <zhizhoutian__> xruxa: let me see
08:52:15 <bero2> yes, me too - heard of CMA before but that's pretty much all I know about it
08:52:35 <zhizhoutian__> xruxa: Sorry, they are all blocked
08:53:06 <zhizhoutian__> xruxa: except 939 and 937
08:53:49 <zhizhoutian__> bero2: yes, no one can know everything;)
08:53:53 <xruxa> is the C-reduce work blocked or just not being done because you got other tasks now?
08:54:48 <zhizhoutian__> xruxa: i think toolchains issue will resolved by google soon or later
08:55:09 <zhizhoutian__> xruxa: so maybe i do not need to push time into this task;)
08:55:35 <xruxa> zhizhoutian__, ok. Unless bero2 disagrees lets cancel that tasks
08:55:57 <zhizhoutian__> xruxa: ok, thanks
08:56:50 <zhizhoutian__> xruxa: about 896, i have know the reason, but i have not resolved it yet.
08:57:16 <bero> Google doesn't care about gcc 5.x -- the last gcc they support is 4.9, from then on they'll only support clang
08:57:16 <xruxa> zhizhoutian__, ok
08:57:29 <xruxa> bero, so we still want this work done
08:58:12 <Qian__> xavierhsu: have sent the demo to you, please check
08:58:14 <bero> yes, it's not highest priority, but we should look at it. I can take it though, I have creduce installed and ready to go
08:58:43 <zhizhoutian__> xruxa: what about 897? I indeed have little time to push on this
08:59:00 <bero> zhizhoutian__: could you remind me what exactly the problem was? I remember you got a compile time error, but that's all I remember about this one (and there's nothing in the card's description)
08:59:29 <zhizhoutian__> bero: let me see
08:59:43 <zhizhoutian__> bero: i have a log, let me check them
09:00:47 <bero2> OK, please add the relevant part of the log to LMG-930
09:01:22 <bero2> wuhai: do you want to go next?
09:01:32 <wuhai> OK
09:01:36 <wuhai> What did you do?
09:01:36 <wuhai> 1.Analysis LocationManagerService architecture code, first learning to this service, so they need more time to understand the architecture of code and services, currently in progress
09:01:36 <wuhai> What are you going to do?
09:01:36 <wuhai> 1.Continue to analyze the structure of code LocationManagerService,try to find the code that can be converted to C/C++ through JNI
09:01:36 <wuhai> What problems do you have?
09:01:37 <wuhai> None
09:01:39 <zhizhoutian__> bero2: got it, it is a "internal compiler error:"
09:02:16 <zhizhoutian__> bero2:  if i change the compiler's branch(platform/prebuilts/gcc/linux-x86/aarch64/aarch64-linux-android-4.9) from master to marshmallow-release
09:02:21 <zhizhoutian__> bero2: it will be ok
09:08:21 <bero2> wangjian: do you want to go next?
09:08:24 <zhizhoutian__> bero2: ok, i have added those relevant informations into this task.
09:08:49 <bero2> zhizhoutian__: thanks
09:09:10 <wangjian> What did you do?
09:09:10 <wangjian> Analyze the launcher code to found unnecessary resource.
09:09:10 <wangjian> What are you going to do?
09:09:10 <wangjian> Continue analyze the launcher code.I think this work will take a long time because the code is too much and too complicate。
09:09:10 <wangjian> What problems do you have?
09:09:10 <wangjian> none till now.
09:11:35 <bero2> yes, it's not expected to be a quick task
09:12:11 <bero2> sunao: do you want to go next?
09:12:20 <sunao> What did you do?
09:12:22 <sunao> 1. Define "BOARD_MALLOC_ALIGNMENT := 8" in BoardConfig.mk build and test. The system can boot up to homescreen and have 1~2 MB free memory than default dlmalloc. But hard to identify where doest it come from.
09:12:23 <sunao> what are you going to do?
09:12:25 <sunao> 1.  continue work on dlmalloc.
09:12:26 <sunao> what problems do you have?.
09:12:28 <sunao> 1.  None.
09:13:27 <bero2> OK...
09:13:51 <bero2> luther: still busy with company stuff?
09:14:01 <luther> bero2: yes :-)
09:14:22 <sunao> bero2: what's meaning of Lost RAM in the outpu of dumpsys meminfo?
09:15:29 <bero2> sunao: not entirely sure, got to double-check the docs
09:15:38 <sunao> okay, I will
09:15:59 <bero2> It could be memory lost from alignment etc.
09:16:29 <sunao> thanks,bero2
09:17:29 <bero2> liuyq: do you want to go next?
09:17:49 <liuyq1> What did you do?
09:17:49 <liuyq1> created release notes for HiKey, please review.
09:17:49 <liuyq1> Investigated about adb connection not stable problem.(no result yet)
09:17:49 <liuyq1> Tried build with linaro gcc5.2, has problem with member builds HiKey and juno.(https://android-review.linaro.org/#/c/16346/)
09:17:52 <liuyq1> Sorted out bugs for release notes and test report.
09:17:54 <liuyq1> What are you going to do?
09:17:58 <liuyq1> Continue investigation about the adb not stable problem for hikey on LAVA
09:18:00 <liuyq1> Finish release things.
09:18:02 <liuyq1> grab bootchart data for juno(LMG-962)
09:18:04 <liuyq1> What problems do you have?
09:18:06 <liuyq1> none
09:18:31 <zhizhoutian__> liuyq1: about adb connect not stable problem , try those command:
09:18:37 <liuyq1> bero2, busy with release and hikey these days:(
09:18:46 <zhizhoutian__> echo 1 > /sys/class/power_supply/battery/stop_charge
09:18:46 <zhizhoutian__> echo aaa > /sys/power/wake_lock
09:19:22 <zhizhoutian__> liuyq1: they works on sprd platform, you can try if they work on hikey
09:20:02 <bero2> yes, might be a more generic bug
09:21:35 <xruxa> liuyq1, about thumb2. Are we done or do we need to investigate more on the floating point regressions?
09:21:48 <bero2> eric___: do you want to go next?
09:21:55 <liuyq1> zhizhoutian__, no /sys/class/power_supply/battery/stop_charge file on Hikey
09:22:40 <zhizhoutian__> liuyq1: what about the other one? that command can stop machine from deep sleep
09:22:48 <liuyq1> bero2, how do you think about the  floating point regressions under thumb2 mode?
09:23:05 <liuyq1> xruxa, I will submit some changes and close that card after that
09:23:05 <zhizhoutian__> liuyq1: deep sleep may cause usb lost connect.
09:23:08 <Qian__> liuyq1: seems zhizhoutian__ wants to keep it awake always
09:23:30 <zhizhoutian__> Qian__: yes, you are right, always keep it awake.
09:24:16 <liuyq1> root@hikey:/ # cat /sys/power/wake_lock
09:24:16 <liuyq1> PowerManagerService.Display
09:24:16 <liuyq1> root@hikey:/ #
09:24:31 <liuyq1> zhizhoutian__, this is the default content in /sys/power/wake_lock
09:25:53 <zhizhoutian__> liuyq1: yes. it will disappear in serveral seconds.
09:26:07 <liuyq1> zhizhoutian__, Qian_, we have a script calls "svc power stayon true" to make it not go to sleep on LAVA
09:26:54 <liuyq1> bero2, should we consider to use linaro gcc5.2 in 15.12? you just said google will use clang in the future
09:28:15 <Qian__> liuyq1: I am not sure, you can check whether it's stable when screen is on or off locally
09:29:48 <liuyq1> Qian_, the problem is not easy to reproduce on local side.
09:30:29 <Qian_> liuyq1: ack
09:34:32 <xruxa> do we have anyone else here who did not report status yet?
09:34:53 <xruxa> eric____,  ?
09:35:30 <xruxa> bero2, looks like we are done (?)
09:40:03 <eric____> yes, nothing update today
09:40:21 <xruxa> ok, thanks all, lets call it a day. See you next time.
09:40:37 <eric____> I think clang static analyzer doesn't suite for android
09:43:30 <eric____> maybe lint is a better tool for android
09:46:38 <bero> #endmeeting