07:32:01 <xruxa> #startmeeting
07:32:01 <linarobot> Meeting started Wed Oct 21 07:32:01 2015 UTC.  The chair is xruxa. Information about MeetBot at http://wiki.debian.org/MeetBot.
07:32:01 <linarobot> Useful Commands: #action #agreed #help #info #idea #link #topic.
07:32:05 <xruxa> #chair bero
07:32:05 <linarobot> Current chairs: bero xruxa
07:32:26 <xruxa> #info Document to edit prior to the meeting is still the same https://docs.google.com/document/d/1CgwSksOXW5r6V3M7EPxv5ot-IZNBuTo_h2_OHfOi3aA/edit
07:32:47 <xruxa> is luther here?
07:32:52 <xruxa> please go first
07:32:58 * luther is here
07:33:13 <luther> 2015.10.21
07:33:13 * luther What did you do?
07:33:13 <luther> 0. Failed to compile apache-harmony org code
07:33:13 <luther> $ svn co http://svn.apache.org/repos/asf/harmony/enhanced/java/trunk
07:33:13 <luther> $ ant -Dauto.fetch=true all
07:33:13 <luther> $ failed info is http://cs.anu.edu.au/people/Robin.Garner/vmmagic-20070207.jar
07:33:17 <luther> can't find under http://cs.anu.edu.au/people, because people 'Robin.Garner' dose not exist now
07:33:19 <luther> 1. study what is apache-harmony, how to use it
07:33:21 <luther> 2. Apache Harmony Project has Discontinued since 2010 after IBM shifted its resources to OpenJDK
07:33:23 <luther> 3. Many of Android core libraries derived from Apache Harmony
07:33:25 <luther> 4. Dalvik VM uses a subset of Harmony until Android 4.4,
07:33:27 <luther> from Android 5.0 a totally new designed ART replaces Dalvik as Android default VM
07:33:33 <luther> ART core code derives from Classpath which is same with OpenJDK,
07:33:35 <luther> and TCK (Technology Compatibility Kit) is
07:33:37 <luther> used to check its compatibility with the Java SE 5 specification, so it will be accepted more commonly
07:33:40 <luther> than Dalvik VM which mostly based on Apache Harmony
07:33:42 <luther> 5. apache-harmony Architecture
07:33:46 <luther> The Dynamic Runtime Layer virtual machine consists of the following components:
07:33:48 <luther> 5.1 The VM core: with its subcomponents concentrates most of the JVM control functions.
07:33:50 <luther> 5.2 The porting layer: hides platform-specific details from other VM components behind a single interface and is based on the Apache Portable Runtime layer.
07:33:53 <luther> 5.3 The garbage collector: allocates Java objects in the heap memory and reclaims unreachable objects using various algorithms
07:33:56 <luther> 5.4 Execution Manager: selects the execution engine for compiling a method, handles profiles and the dynamic recompilation logic.
07:34:03 <luther> 5.5 Class Library: is a Java standard library.
07:34:05 <luther> 5.6 The thread manager that handle operating system threading
07:34:07 <luther> 5.7 The execution engine: consists of the following:
07:34:09 <luther> 5.7.1 The just-in-time compiler for compilation and execution of method code.
07:34:11 <luther> 5.7.2 The interpreter for easier debugging.
07:34:13 * luther What are you going to do?
07:34:17 <luther> 1. study what is apache-harmony, how to use it
07:34:19 <luther> 2. Compile it on Android 5.1
07:34:21 * luther What problems do you have?
07:34:23 <luther> 1. None till now
07:34:50 <xruxa> luther, so back to your statement "Dalvik VM uses a subset of Harmony until Android 4.4" - does it mean it is not used anymore in L and M?
07:35:29 <luther> xruxa, yes, ART does not base on apache-harmony
07:36:14 <xruxa> luther, ok, I will need to delegate this one to bero when he gets back
07:36:27 <luther> xruxa: ok :-)
07:36:44 <xruxa> Qian__, you are next please
07:37:11 <Qian__> What did you do?
07:37:11 <Qian__> investigate LMG-916, see the table of the comment
07:37:11 <Qian__> What are you going to do?
07:37:13 <Qian__> investigate core part of external, framework
07:37:15 <Qian__> What problems do you have?
07:37:17 <Qian__> I just list the modules, the more details should be disscussion.
07:37:35 <xruxa> Qian__, and going good or you have questions?
07:38:02 <Qian__> my question is i don't know the details
07:38:40 <xruxa> Qian__, what details? Of what to do or about the SW you are investigating?
07:39:10 <Qian__> So, I just list the modules java bin used, the further step should be discussed
07:39:35 <xruxa> Qian__, ok, good start, sync with bero when you have something ready.
07:39:38 <xruxa> yspan, your status next please.
07:39:47 <Qian__> LMG-916, because it's a table
07:39:49 <yspan> What did you do?
07:39:49 <yspan> Completed a simple program, proof of concept OK!
07:40:05 <yspan> codes are already reviewed by bero
07:40:11 <yspan> What are you going to do?
07:40:11 <yspan> Apply this idea to libpng
07:40:13 <xruxa> yspan, nice, so that proved the dlopen approach works, right?
07:40:20 <yspan> right!
07:40:51 <yspan> I think the approach could be also applied to other lib, such as libjpeg
07:41:00 <xruxa> yspan, please have the code shared somewhere (online, preferably in git repo if possible)
07:41:29 <yspan> OK
07:41:31 <xruxa> #action yspan to share his libpng and test app code
07:41:35 <xruxa> yspan, thanks.
07:41:52 <xruxa> sunao, you are next
07:42:06 <sunao> What did you do?
07:42:08 <sunao> 1. Libopus.so is linked by libstagefright.so, this causes the libopus.so loaded by zygote and shared by all its child process. It may stay in memory all the time .  But actually libstagefright.so has nothing to do with libopus.so.  libopus doesn't need to be loaded by zygote.
07:42:09 <sunao> modify code and testing.
07:42:11 <sunao> detailed information please refer to https://projects.linaro.org/browse/LMG-913
07:42:12 <sunao> 
07:42:14 <sunao> what are you going to do?
07:42:15 <sunao> 1. continue working on LMG-913.
07:42:17 <sunao> what problems do you have?
07:42:18 <sunao> None.
07:42:20 <sunao> that's all
07:42:25 <bero> sorry, got to get back to luther's task for a second (only reading it now because I'm in the Tinno meeting)
07:42:33 <bero> I didn't realize harmony had already been dumped
07:42:38 <xruxa> sunao, thansk. Good speed by the way - you investigated few libs already.
07:42:41 <bero> so the work needs to move on to its replacement
07:42:56 <bero> So all of the optimization work should be done with its classpath replacements...
07:42:58 <xruxa> bero, ack. luther you understood?
07:43:05 <bero> No need to optimize stuff that is dead already
07:43:10 * bero shifts back to Tinno meeting
07:43:27 <luther> bero, which is next? art or something else?
07:43:30 <xruxa> #action bero to update the task for luther https://projects.linaro.org/browse/LMG-909 to target replacement for apache harmony
07:44:01 <bero> luther: I think most of the core stuff moved to libcore/luni, some more is inside art
07:44:10 <bero> will update LMG-909
07:44:17 <luther> bero: ok, thanks:-)
07:44:27 <xruxa> thanks
07:44:55 <xruxa> sunao, anything else before we move on? Can you get 96board from your company?
07:45:11 <sunao> I can't get a board
07:45:38 <sunao> I work on the emulator and a phone.
07:45:41 <xruxa> sunao, ok, lets take it offline
07:45:47 <sunao> OK
07:46:01 <xruxa> #action xruxa and bero to check with all assignees about getting some development device
07:46:11 <xruxa> I do not see Xavier here ...
07:46:36 <xruxa> #info Xavier missing
07:46:46 <xruxa> zhizhoutian_, please your status next.
07:47:11 <zhizhoutian_> a minite pls
07:47:40 <zhizhoutian_> What did you do?
07:47:40 <zhizhoutian_> fix juno aosp master build, and find a toolchain bug
07:47:40 <zhizhoutian_> investigate C-Reduce tools,but C-Reduce tools require llvm 3.7 or later. The latest version in Ubuntu is 2.9. so i tried earlier version, but still can not work until now.
07:47:40 <zhizhoutian_> ready for Tinno code-base
07:47:40 <zhizhoutian_> What are you going to do?
07:47:40 <zhizhoutian_> investigate C-Reduce tools,but C-Reduce tools require llvm 3.7 or later. The latest version in Ubuntu is 2.9. so i tried earlier version, but still can not work until now.
07:47:40 <zhizhoutian_> ready for Tinno code-base
07:47:41 <zhizhoutian_> What problems do you have?
07:47:41 <zhizhoutian_> none
07:48:13 <xruxa> zhizhoutian_, to Juno master build works now?
07:48:47 <zhizhoutian_> yes, juno master have a toolchain bug, bero said that we should report it.
07:48:56 <xruxa> zhizhoutian_, I guess we need tasks for you in JIRA, will you make them yourself?
07:49:18 <xruxa> zhizhoutian_, expecially the Tinno one :)
07:49:28 <zhizhoutian_> aosp maintance?
07:49:48 <zhizhoutian_> oh, yes
07:49:49 <xruxa> zhizhoutian_, C-Reduce tools and Tinno work
07:50:14 <xruxa> #action zhizhoutian_ make JIRA tasks for current work (C-Reduce tools and Tinno work, etc). Ask xruxa if not sure
07:50:21 <zhizhoutian_> but i still do not know how to release Tinno code-base
07:50:41 <zhizhoutian_> it based on TShark devices
07:50:53 <zhizhoutian_> by the way, i am maintancing the TShark lava devices.
07:51:09 <xruxa> zhizhoutian_, do thsarks work now? Is there more than one?
07:51:20 <zhizhoutian_> Tinno will use TShark devices this time
07:51:56 <zhizhoutian_> yes, there are five tshark-hd devices and two tshark-qdd
07:51:58 <zhizhoutian_> qhd
07:51:58 * xruxa wondering how that works out with bero 's work on Tinno phones on the low memory
07:52:26 <xruxa> zhizhoutian_, ok, thanks.
07:52:30 <xruxa> liuyq, you are next
07:52:51 <liuyq> What did you do?
07:52:51 <liuyq> enabled selinux for fvp and juno build for masrshmallow
07:52:51 <liuyq> Made sure that CaffeineMark test is enabled on LAVA
07:52:51 <liuyq> Made sure that Reduced benchmark tests are not submitted to LAVA
07:52:51 <liuyq> Fixed problems for lcr-member-fvp-m build: version and toolchain
07:52:52 <liuyq> 
07:52:53 <liuyq> What are you going to do?
07:52:59 <liuyq> investigating about the selinux problem for vexpress marshmallow build.
07:53:01 <liuyq> investigating on created of the CTS6.0 package for FVP build.
07:53:03 <liuyq> making sure all builds can boot on LAVA and run jobs successfully on LAVA
07:53:05 <liuyq> What problems do you have?
07:53:08 <liuyq> might need more time to create the cts package for fvp build.
07:53:37 <xruxa> liuyq, lets add one more - investigate with QAS and B&B why we do not have CTS results on Juno L M-LCR
07:54:13 <liuyq> xruxa, OK.
07:54:13 <xruxa> ok, did I forget someone (else than bero)?
07:54:29 <xruxa> bero, you do not happen to be around now?
07:54:34 <liuyq> xruxa, I will take this card from Bero: https://projects.linaro.org/browse/LMG-889
07:54:54 <bero> xruxa: I am, just not caught up with everything
07:55:02 <bero> Tinno meeting is still on
07:55:13 <xruxa> bero, ok. lets end the meeting now then
07:55:32 <xruxa> any more topics before end the meeting?
07:55:45 <bero> News from my side:
07:55:48 <bero> What did you do?
07:55:48 <bero> More toolchain bisection to find M boot failure with gcc 5.2 (still in progress)
07:55:48 <bero> Started NDK build, currently fixing issues
07:55:48 <bero> What are you going to do?
07:55:48 <bero> Fix NDK build issues
07:55:48 <bero> Continue toolchain bisection
07:55:48 <bero> Tinno meeting
07:55:49 <bero> What problems do you have?
07:55:49 <bero> Need more than 24 hours in a day
07:56:03 <bero> Also, looks like we may get to move memory optimization work to M after all
07:56:04 <xruxa> ;)
07:56:09 <bero> but there's no formal decision yet
07:56:16 <bero> so for now it's still focusing on L
07:56:22 <xruxa> bero, of, that would be change (L->M)
07:57:05 <luther> many manufacturer like M
07:57:05 <bero> Task started on L because that's what members asked for, but in the Tinno meeting they're actually asking about M
07:57:12 <xruxa> #info We will very likely not get Mali drivers for Marshmallow LCR this month - releasing SW rendering means no need to benchmarks and test the usual way
07:57:14 <bero> so may or may not move soon
07:57:56 <bero> I don't have anything else right now
07:58:38 <xruxa> bero, thanks.
07:58:47 <xruxa> #endmeeting