Hi guys,
Re. caffinemark Stuart did some investigation and found the following. Not sure if it applies to your builds too.
* Investigated why CaffeineMark wasn't working. For Juno, we use a small heapsize (48MB), which wasn't sufficient for caffeinemark to run successfully against. Looks like the allocator is subject to fragmentation issues. Using flounder's configuration (a 2GB 10inch tablet, apparently) works fine
Regards,
Steve
On 20/03/2015 14:16, "linaro-android-request@lists.linaro.org" linaro-android-request@lists.linaro.org wrote:
Send linaro-android mailing list submissions to linaro-android@lists.linaro.org
To subscribe or unsubscribe via the World Wide Web, visit http://lists.linaro.org/mailman/listinfo/linaro-android or, via email, send a message with subject or body 'help' to linaro-android-request@lists.linaro.org
You can reach the person managing the list at linaro-android-owner@lists.linaro.org
When replying, please edit your Subject line so it is more specific than "Re: Contents of linaro-android digest..."
Today's Topics:
- Re: Update v1 : Jobs for build#335 (Mahanth Gouda)
- Re: Update v1 : Jobs for build#335 (Botao Sun)
Message: 1 Date: Thu, 19 Mar 2015 16:28:44 +0000 From: Mahanth Gouda mahanth.gouda@linaro.org To: YongQin Liu yongqin.liu@linaro.org Cc: Linaro Android Mailman List linaro-android@lists.linaro.org, Jakub Pavelek jakub.pavelek@linaro.org Subject: Re: Update v1 : Jobs for build#335 Message-ID: CAKhO2P9rPYOf+Z8GNNrYMfoCqsJats=kJr0ZJO2Ug_BP=mBTbQ@mail.gmail.com Content-Type: text/plain; charset="utf-8"
Pleas find my observations in line:
nbench - Looks ok.
GFXBench3 - CRASH - (I get this msg on the display - Your version of
the GFXBench is outdated, please refresh the app and start again. Let me know if you need to the log cat msg for this ?)
- gearses2eclair - Looks good to me. Tried a couple of times and it
looks OK.
Thanks,
Mahanth
On 19 March 2015 at 16:10, YongQin Liu yongqin.liu@linaro.org wrote:
Hi, Mahanth
Another 3 apks:
- nbench
- GFXBench3
- gearses2eclair, not finished yet, seems to be timed out
Could you please help to check these 3 apks as well? They are failed to get the result at my jobs.
Thanks, Yongqin Liu On 19 March 2015 at 23:49, Mahanth Gouda mahanth.gouda@linaro.org wrote:
Hi all, Just wanted to let you know that I've picked up the build 335 and started manual testing this morning on Juno. I'm half the way running these tests so far, hoping to finish the rest in next few hours time. So, thought of giving you a quick results update:
I'm testing only those benchmarks that've been marked by Yongqin as failed. Please let me know if I've missed something?
Benchmark job 1:
- antutu 5 - *HUNG/CRASH *(No log as the test didn't complete )
- caffeinemark - *CRASH *- (Every other time this test fails) - logcat
& snapshot attached 3. cfbench - *PASS*
Benchmark job 2:
- smartbench2014 - *PASS*
- gearsES1eclair - *CRASH* (There's no log for this because the script
does not complete, it's hung in between) It took long time investigating this, but could not get to the root cause of this issue. So, I've left it open for now.
Benchmark job 3:
- vellamo - testing in progress (currently the benchmark sounds like
it's paused on Networking Loader, since a while)
Benchmark job 4:
- andebench2014 -
- javawhetstone -
- jbench -
- linpackjava -
- scimark -
Please note - I've repeated all these tests at least 5 times in order to check the consistency.
Thanks & regards, Mahanth
On 19 March 2015 at 10:59, Mahanth Gouda mahanth.gouda@linaro.org wrote:
Hi Jakub, Thanks for the help offer, I've started with this, will surely seek your help when required.
Thanks, Mahanth
On 19 March 2015 at 09:19, Jakub Pavelek jakub.pavelek@linaro.org wrote:
Hi Botao and Mananth,
if you need anything from us ping me or YongQin. We would like to investigate the failing benchmark automation ASAP, otherwise we need QAS to run those benchmarks manually and we will lose trends tracking in ImageReports.
On 18 March 2015 at 19:01, Milosz Wasilewski < milosz.wasilewski@linaro.org> wrote:
Botao and Mahanth can help. Mahanth has access to the hw. As far as I remember:
- antutu 5 crashes (was crashing before) before it finishes
- caffeinemark crashes 4 out of 5 attempts
- nbench produces empty results for some reason (only zeros)
- GFXBench takes very long to run, we might have too short timeout
- GearsES1 crashes always and should be abandoned
I don't know about the others. This is what was happening before moving to 5.1. Mahanth, please take this as high priority to investigate. Botao migh help you with Gears as he did the automation.
milosz
On 18 March 2015 at 15:56, Jakub Pavelek jakub.pavelek@linaro.org wrote: > Milosz is out of office all this week. We need to find another way to check > things out :-/ > > On 18 March 2015 at 10:14, YongQin Liu yongqin.liu@linaro.org wrote: >> >> Hi, Misolz >> >> Could you please help to check if the benchmark tests mentioned below >> can work on the Juno with the 5.1 build#335? >> >> Thanks, >> Yongqin Liu >> >> On 18 March 2015 at 17:11, YongQin Liu yongqin.liu@linaro.org wrote: >>> >>> Hi, All >>> >>> The job for the benchmarks failed again. >>> Seems there is a problem to run the vellamo on Juno, the test timed out >>> and caused the sync problem. >>> So I split the job into 3 small jobs: >>> >>> Benchmark Job#1 >>> https://validation.linaro.org/scheduler/job/287238.0 finished. >>> but there is no benchmark result for benchmarks application of antutu5, >>> Caffeinemark, and CF-Bench. >>> And also all the benchmark result for nbench are zero. >>> >>> Benchmark Job#2 >>> https://validation.linaro.org/scheduler/job/287240.0 >>> Smartbench2012 seems not started correctly. >>> And the gearses1eclair test was timed out, and caused the sync problem. >>> >>> >>> Benchmark Job#3: vellamo job, timed out again. >>> https://validation.linaro.org/scheduler/job/287242.0 >>> >>> >>> Benchmark Job#4: tests for andebench2014, Javawhetstone, JBench, >>> LinpackJava, and Scimark >>> https://validation.linaro.org/scheduler/job/287503.0 >>> >>> >>> Thanks, >>> Yongqin Liu >>> On 17 March 2015 at 21:30, YongQin Liu yongqin.liu@linaro.org wrote: >>>> >>>> Hi, Jakub >>>> >>>> The 2 cts jobs finished. >>>> >>>> Thanks, >>>> Yongqin Liu >>>> On 17 March 2015 at 15:47, YongQin Liu yongqin.liu@linaro.org wrote: >>>>> >>>>> Hi, All >>>>> >>>>> Following are job status for Juno build #335 on LAVA: >>>>> The 3 resubmitted jobs are running on LAVA now. >>>>> >>>>> Normal Job: >>>>> https://validation.linaro.org/scheduler/job/282224 finished >>>>> >>>>> benchmark-lite: >>>>> https://validation.linaro.org/scheduler/job/282229.0 finished >>>>> >>>>> >>>>> linaro-android_armv8-android-juno-lsk-benchmark-application >>>>> https://validation.linaro.org/scheduler/job/282231.0 finished >>>>> >>>>> >>>>> linaro-android_armv8-android-juno-lsk_cts_32bit: >>>>> https://validation.linaro.org/scheduler/job/282225.1 Canceled >>>>> https://validation.linaro.org/scheduler/job/284313.0 Incomplete >>>>> https://validation.linaro.org/scheduler/job/286312.0 resubmitted >>>> >>>> Finished >>>>> >>>>> >>>>> linaro-android_armv8-android-juno-lsk_cts_64bit: >>>>> https://validation.linaro.org/scheduler/job/282227.0 Canceled >>>>> https://validation.linaro.org/scheduler/job/284315.0 Incomplete >>>>> https://validation.linaro.org/scheduler/job/286310.0 resubmitted >>>> >>>> Finished. >>>>> >>>>> >>>>> Benchmarks: >>>>> https://validation.linaro.org/scheduler/job/282233.0 Canceled >>>>> https://validation.linaro.org/scheduler/job/284317.0 Incomplete >>>>> https://validation.linaro.org/scheduler/job/286314.0 resubmitted >>>>> >>>>> >>>>> -- >>>>> Best Regards, >>>>> Yongqin Liu
>>>>>
>>>>> #mailing list >>>>> linaro-android@lists.linaro.org >>>>> http://lists.linaro.org/mailman/listinfo/linaro-android >>>> >>>> >>>> >>>> >>>> -- >>>> Best Regards, >>>> Yongqin Liu >>>> --------------------------------------------------------------- >>>> #mailing list >>>> linaro-android@lists.linaro.org >>>> http://lists.linaro.org/mailman/listinfo/linaro-android >>> >>> >>> >>> >>> -- >>> Best Regards, >>> Yongqin Liu >>> --------------------------------------------------------------- >>> #mailing list >>> linaro-android@lists.linaro.org >>> http://lists.linaro.org/mailman/listinfo/linaro-android >> >> >> >> >> -- >> Best Regards, >> Yongqin Liu >> --------------------------------------------------------------- >> #mailing list >> linaro-android@lists.linaro.org >> http://lists.linaro.org/mailman/listinfo/linaro-android > > > > > -- > With best regards, > > Jakub Pavelek > > Linaro Mobile Group project manager > Linaro.org?Open source software for ARM SoCs >
-- With best regards,
Jakub Pavelek
Linaro Mobile Group project manager Linaro.org?Open source software for ARM SoCs
-- Best Regards, Yongqin Liu
#mailing list linaro-android@lists.linaro.org linaro-dev@lists.linaro.org http://lists.linaro.org/mailman/listinfo/linaro-android
-------------- next part -------------- An HTML attachment was scrubbed... URL: http://lists.linaro.org/pipermail/linaro-android/attachments/20150319/2ac ebd8e/attachment-0001.html
Message: 2 Date: Fri, 20 Mar 2015 17:16:38 +1100 From: Botao Sun botao.sun@linaro.org To: Mahanth Gouda mahanth.gouda@linaro.org Cc: Linaro Android Mailman List linaro-android@lists.linaro.org, Jakub Pavelek jakub.pavelek@linaro.org Subject: Re: Update v1 : Jobs for build#335 Message-ID: CAMGVPmUryFQoPs+RF4M4P+nFm524HdTm=g7Y1GP6Ed3yOHyMaw@mail.gmail.com Content-Type: text/plain; charset="utf-8"
Hi Mahanth,
Glad to hear that you managed to run gearses2eclair successfully, I'm afraid there is nothing more I can do on my side - I only have a TI Panda 4430 which the latest available image is 14.10, and obviously it works well on that build as I wrote the code based on it.
Thanks goodness you have the access to Juno board in Cambridge. ;)
Let me know if you need any extra explanation about the code.
Botao
On 20 March 2015 at 03:28, Mahanth Gouda mahanth.gouda@linaro.org wrote:
Pleas find my observations in line:
nbench - Looks ok.
GFXBench3 - CRASH - (I get this msg on the display - Your version
of the GFXBench is outdated, please refresh the app and start again. Let me know if you need to the log cat msg for this ?)
- gearses2eclair - Looks good to me. Tried a couple of times and it
looks OK.
Thanks,
Mahanth
On 19 March 2015 at 16:10, YongQin Liu yongqin.liu@linaro.org wrote:
Hi, Mahanth
Another 3 apks:
- nbench
- GFXBench3
- gearses2eclair, not finished yet, seems to be timed out
Could you please help to check these 3 apks as well? They are failed to get the result at my jobs.
Thanks, Yongqin Liu On 19 March 2015 at 23:49, Mahanth Gouda mahanth.gouda@linaro.org wrote:
Hi all, Just wanted to let you know that I've picked up the build 335 and started manual testing this morning on Juno. I'm half the way running these tests so far, hoping to finish the rest in next few hours time. So, thought of giving you a quick results update:
I'm testing only those benchmarks that've been marked by Yongqin as failed. Please let me know if I've missed something?
Benchmark job 1:
- antutu 5 - *HUNG/CRASH *(No log as the test didn't complete )
- caffeinemark - *CRASH *- (Every other time this test fails) -
logcat & snapshot attached 3. cfbench - *PASS*
Benchmark job 2:
- smartbench2014 - *PASS*
- gearsES1eclair - *CRASH* (There's no log for this because the
script does not complete, it's hung in between) It took long time investigating this, but could not get to the root cause of this issue. So, I've left it open for now.
Benchmark job 3:
- vellamo - testing in progress (currently the benchmark sounds like
it's paused on Networking Loader, since a while)
Benchmark job 4:
- andebench2014 -
- javawhetstone -
- jbench -
- linpackjava -
- scimark -
Please note - I've repeated all these tests at least 5 times in order to check the consistency.
Thanks & regards, Mahanth
On 19 March 2015 at 10:59, Mahanth Gouda mahanth.gouda@linaro.org wrote:
Hi Jakub, Thanks for the help offer, I've started with this, will surely seek your help when required.
Thanks, Mahanth
On 19 March 2015 at 09:19, Jakub Pavelek jakub.pavelek@linaro.org wrote:
Hi Botao and Mananth,
if you need anything from us ping me or YongQin. We would like to investigate the failing benchmark automation ASAP, otherwise we need QAS to run those benchmarks manually and we will lose trends tracking in ImageReports.
On 18 March 2015 at 19:01, Milosz Wasilewski < milosz.wasilewski@linaro.org> wrote:
> Botao and Mahanth can help. Mahanth has access to the hw. As far >as I > remember: > - antutu 5 crashes (was crashing before) before it finishes > - caffeinemark crashes 4 out of 5 attempts > - nbench produces empty results for some reason (only zeros) > - GFXBench takes very long to run, we might have too short timeout > - GearsES1 crashes always and should be abandoned > > I don't know about the others. This is what was happening before > moving to 5.1. Mahanth, please take this as high priority to > investigate. Botao migh help you with Gears as he did the >automation. > > milosz > > > On 18 March 2015 at 15:56, Jakub Pavelek jakub.pavelek@linaro.org > wrote: > > Milosz is out of office all this week. We need to find another >way > to check > > things out :-/ > > > > On 18 March 2015 at 10:14, YongQin Liu yongqin.liu@linaro.org > wrote: > >> > >> Hi, Misolz > >> > >> Could you please help to check if the benchmark tests mentioned > below > >> can work on the Juno with the 5.1 build#335? > >> > >> Thanks, > >> Yongqin Liu > >> > >> On 18 March 2015 at 17:11, YongQin Liu yongqin.liu@linaro.org > wrote: > >>> > >>> Hi, All > >>> > >>> The job for the benchmarks failed again. > >>> Seems there is a problem to run the vellamo on Juno, the test > timed out > >>> and caused the sync problem. > >>> So I split the job into 3 small jobs: > >>> > >>> Benchmark Job#1 > >>> https://validation.linaro.org/scheduler/job/287238.0 finished. > >>> but there is no benchmark result for benchmarks application of > antutu5, > >>> Caffeinemark, and CF-Bench. > >>> And also all the benchmark result for nbench are zero. > >>> > >>> Benchmark Job#2 > >>> https://validation.linaro.org/scheduler/job/287240.0 > >>> Smartbench2012 seems not started correctly. > >>> And the gearses1eclair test was timed out, and caused the sync > problem. > >>> > >>> > >>> Benchmark Job#3: vellamo job, timed out again. > >>> https://validation.linaro.org/scheduler/job/287242.0 > >>> > >>> > >>> Benchmark Job#4: tests for andebench2014, Javawhetstone, >JBench, > >>> LinpackJava, and Scimark > >>> https://validation.linaro.org/scheduler/job/287503.0 > >>> > >>> > >>> Thanks, > >>> Yongqin Liu > >>> On 17 March 2015 at 21:30, YongQin Liu yongqin.liu@linaro.org > wrote: > >>>> > >>>> Hi, Jakub > >>>> > >>>> The 2 cts jobs finished. > >>>> > >>>> Thanks, > >>>> Yongqin Liu > >>>> On 17 March 2015 at 15:47, YongQin Liu >yongqin.liu@linaro.org > wrote: > >>>>> > >>>>> Hi, All > >>>>> > >>>>> Following are job status for Juno build #335 on LAVA: > >>>>> The 3 resubmitted jobs are running on LAVA now. > >>>>> > >>>>> Normal Job: > >>>>> https://validation.linaro.org/scheduler/job/282224 finished > >>>>> > >>>>> benchmark-lite: > >>>>> https://validation.linaro.org/scheduler/job/282229.0 finished > >>>>> > >>>>> > >>>>> linaro-android_armv8-android-juno-lsk-benchmark-application > >>>>> https://validation.linaro.org/scheduler/job/282231.0 finished > >>>>> > >>>>> > >>>>> linaro-android_armv8-android-juno-lsk_cts_32bit: > >>>>> https://validation.linaro.org/scheduler/job/282225.1 Canceled > >>>>> https://validation.linaro.org/scheduler/job/284313.0 >Incomplete > >>>>> https://validation.linaro.org/scheduler/job/286312.0 > resubmitted > >>>> > >>>> Finished > >>>>> > >>>>> > >>>>> linaro-android_armv8-android-juno-lsk_cts_64bit: > >>>>> https://validation.linaro.org/scheduler/job/282227.0 Canceled > >>>>> https://validation.linaro.org/scheduler/job/284315.0 > Incomplete > >>>>> https://validation.linaro.org/scheduler/job/286310.0 > resubmitted > >>>> > >>>> Finished. > >>>>> > >>>>> > >>>>> Benchmarks: > >>>>> https://validation.linaro.org/scheduler/job/282233.0 Canceled > >>>>> https://validation.linaro.org/scheduler/job/284317.0 >Incomplete > >>>>> https://validation.linaro.org/scheduler/job/286314.0 > resubmitted > >>>>> > >>>>> > >>>>> -- > >>>>> Best Regards, > >>>>> Yongqin Liu > >>>>> >--------------------------------------------------------------- > >>>>> #mailing list > >>>>> linaro-android@lists.linaro.org > >>>>> http://lists.linaro.org/mailman/listinfo/linaro-android > >>>> > >>>> > >>>> > >>>> > >>>> -- > >>>> Best Regards, > >>>> Yongqin Liu > >>>> >--------------------------------------------------------------- > >>>> #mailing list > >>>> linaro-android@lists.linaro.org > >>>> http://lists.linaro.org/mailman/listinfo/linaro-android > >>> > >>> > >>> > >>> > >>> -- > >>> Best Regards, > >>> Yongqin Liu > >>> --------------------------------------------------------------- > >>> #mailing list > >>> linaro-android@lists.linaro.org > >>> http://lists.linaro.org/mailman/listinfo/linaro-android > >> > >> > >> > >> > >> -- > >> Best Regards, > >> Yongqin Liu > >> --------------------------------------------------------------- > >> #mailing list > >> linaro-android@lists.linaro.org > >> http://lists.linaro.org/mailman/listinfo/linaro-android > > > > > > > > > > -- > > With best regards, > > > > Jakub Pavelek > > > > Linaro Mobile Group project manager > > Linaro.org?Open source software for ARM SoCs > > >
-- With best regards,
Jakub Pavelek
Linaro Mobile Group project manager Linaro.org?Open source software for ARM SoCs
-- Best Regards, Yongqin Liu
#mailing list linaro-android@lists.linaro.org linaro-dev@lists.linaro.org http://lists.linaro.org/mailman/listinfo/linaro-android
-------------- next part -------------- An HTML attachment was scrubbed... URL: http://lists.linaro.org/pipermail/linaro-android/attachments/20150320/989 687c9/attachment.html
linaro-android mailing list linaro-android@lists.linaro.org http://lists.linaro.org/mailman/listinfo/linaro-android
End of linaro-android Digest, Vol 40, Issue 13
-- IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you.
ARM Limited, Registered office 110 Fulbourn Road, Cambridge CB1 9NJ, Registered in England & Wales, Company No: 2557590 ARM Holdings plc, Registered office 110 Fulbourn Road, Cambridge CB1 9NJ, Registered in England & Wales, Company No: 2548782
Hi, Steve, Stuart
On 24 March 2015 at 18:56, Steve Bannister Steve.Bannister@arm.com wrote:
Hi guys,
Re. caffinemark Stuart did some investigation and found the following. Not sure if it applies to your builds too.
- Investigated why CaffeineMark wasn't working. For Juno, we use a small
heapsize (48MB), which wasn't sufficient for caffeinemark to run successfully against. Looks like the allocator is subject to fragmentation issues. Using flounder's configuration (a 2GB 10inch tablet, apparently) works fine
Just want to be clear about the problem. From what I investigated, the problem about this CaffeineMark is that
it will get crashed on the result display window after waiting some time(like 10 minutes) because of the memory leak problem. We can see the result page on Juno and on other platforms(like the Nexus9)(which means the test itself should be finished), but the crash will cause problem for us to extract the test result with automation scripts.
So is your above solution to fix that memory leak problem on the result display window? Or your above solution is to fix another problem I did not realize?
Thanks, Yongqin Liu
Regards,
Steve
On 20/03/2015 14:16, "linaro-android-request@lists.linaro.org" linaro-android-request@lists.linaro.org wrote:
Send linaro-android mailing list submissions to linaro-android@lists.linaro.org
To subscribe or unsubscribe via the World Wide Web, visit http://lists.linaro.org/mailman/listinfo/linaro-android or, via email, send a message with subject or body 'help' to linaro-android-request@lists.linaro.org
You can reach the person managing the list at linaro-android-owner@lists.linaro.org
When replying, please edit your Subject line so it is more specific than "Re: Contents of linaro-android digest..."
Today's Topics:
- Re: Update v1 : Jobs for build#335 (Mahanth Gouda)
- Re: Update v1 : Jobs for build#335 (Botao Sun)
Message: 1 Date: Thu, 19 Mar 2015 16:28:44 +0000 From: Mahanth Gouda mahanth.gouda@linaro.org To: YongQin Liu yongqin.liu@linaro.org Cc: Linaro Android Mailman List linaro-android@lists.linaro.org, Jakub Pavelek jakub.pavelek@linaro.org Subject: Re: Update v1 : Jobs for build#335 Message-ID: <CAKhO2P9rPYOf+Z8GNNrYMfoCqsJats=kJr0ZJO2Ug_BP=
mBTbQ@mail.gmail.com>
Content-Type: text/plain; charset="utf-8"
Pleas find my observations in line:
nbench - Looks ok.
GFXBench3 - CRASH - (I get this msg on the display - Your version of
the GFXBench is outdated, please refresh the app and start again. Let me know if you need to the log cat msg for this ?)
- gearses2eclair - Looks good to me. Tried a couple of times and it
looks OK.
Thanks,
Mahanth
On 19 March 2015 at 16:10, YongQin Liu yongqin.liu@linaro.org wrote:
Hi, Mahanth
Another 3 apks:
- nbench
- GFXBench3
- gearses2eclair, not finished yet, seems to be timed out
Could you please help to check these 3 apks as well? They are failed to get the result at my jobs.
Thanks, Yongqin Liu On 19 March 2015 at 23:49, Mahanth Gouda mahanth.gouda@linaro.org wrote:
Hi all, Just wanted to let you know that I've picked up the build 335 and started manual testing this morning on Juno. I'm half the way running these tests so far, hoping to finish the rest in next few hours time. So, thought of giving you a quick results update:
I'm testing only those benchmarks that've been marked by Yongqin as failed. Please let me know if I've missed something?
Benchmark job 1:
- antutu 5 - *HUNG/CRASH *(No log as the test didn't complete )
- caffeinemark - *CRASH *- (Every other time this test fails) - logcat
& snapshot attached 3. cfbench - *PASS*
Benchmark job 2:
- smartbench2014 - *PASS*
- gearsES1eclair - *CRASH* (There's no log for this because the script
does not complete, it's hung in between) It took long time investigating this, but could not get to the root cause of this issue. So, I've left it open for now.
Benchmark job 3:
- vellamo - testing in progress (currently the benchmark sounds like
it's paused on Networking Loader, since a while)
Benchmark job 4:
- andebench2014 -
- javawhetstone -
- jbench -
- linpackjava -
- scimark -
Please note - I've repeated all these tests at least 5 times in order to check the consistency.
Thanks & regards, Mahanth
On 19 March 2015 at 10:59, Mahanth Gouda mahanth.gouda@linaro.org wrote:
Hi Jakub, Thanks for the help offer, I've started with this, will surely seek your help when required.
Thanks, Mahanth
On 19 March 2015 at 09:19, Jakub Pavelek jakub.pavelek@linaro.org wrote:
Hi Botao and Mananth,
if you need anything from us ping me or YongQin. We would like to investigate the failing benchmark automation ASAP, otherwise we need QAS to run those benchmarks manually and we will lose trends tracking in ImageReports.
On 18 March 2015 at 19:01, Milosz Wasilewski < milosz.wasilewski@linaro.org> wrote:
> Botao and Mahanth can help. Mahanth has access to the hw. As far as >I > remember: > - antutu 5 crashes (was crashing before) before it finishes > - caffeinemark crashes 4 out of 5 attempts > - nbench produces empty results for some reason (only zeros) > - GFXBench takes very long to run, we might have too short timeout > - GearsES1 crashes always and should be abandoned > > I don't know about the others. This is what was happening before > moving to 5.1. Mahanth, please take this as high priority to > investigate. Botao migh help you with Gears as he did the >automation. > > milosz > > > On 18 March 2015 at 15:56, Jakub Pavelek jakub.pavelek@linaro.org > wrote: > > Milosz is out of office all this week. We need to find another way > to check > > things out :-/ > > > > On 18 March 2015 at 10:14, YongQin Liu yongqin.liu@linaro.org > wrote: > >> > >> Hi, Misolz > >> > >> Could you please help to check if the benchmark tests mentioned > below > >> can work on the Juno with the 5.1 build#335? > >> > >> Thanks, > >> Yongqin Liu > >> > >> On 18 March 2015 at 17:11, YongQin Liu yongqin.liu@linaro.org > wrote: > >>> > >>> Hi, All > >>> > >>> The job for the benchmarks failed again. > >>> Seems there is a problem to run the vellamo on Juno, the test > timed out > >>> and caused the sync problem. > >>> So I split the job into 3 small jobs: > >>> > >>> Benchmark Job#1 > >>> https://validation.linaro.org/scheduler/job/287238.0 finished. > >>> but there is no benchmark result for benchmarks application of > antutu5, > >>> Caffeinemark, and CF-Bench. > >>> And also all the benchmark result for nbench are zero. > >>> > >>> Benchmark Job#2 > >>> https://validation.linaro.org/scheduler/job/287240.0 > >>> Smartbench2012 seems not started correctly. > >>> And the gearses1eclair test was timed out, and caused the sync > problem. > >>> > >>> > >>> Benchmark Job#3: vellamo job, timed out again. > >>> https://validation.linaro.org/scheduler/job/287242.0 > >>> > >>> > >>> Benchmark Job#4: tests for andebench2014, Javawhetstone, JBench, > >>> LinpackJava, and Scimark > >>> https://validation.linaro.org/scheduler/job/287503.0 > >>> > >>> > >>> Thanks, > >>> Yongqin Liu > >>> On 17 March 2015 at 21:30, YongQin Liu yongqin.liu@linaro.org > wrote: > >>>> > >>>> Hi, Jakub > >>>> > >>>> The 2 cts jobs finished. > >>>> > >>>> Thanks, > >>>> Yongqin Liu > >>>> On 17 March 2015 at 15:47, YongQin Liu <yongqin.liu@linaro.org
> wrote: > >>>>> > >>>>> Hi, All > >>>>> > >>>>> Following are job status for Juno build #335 on LAVA: > >>>>> The 3 resubmitted jobs are running on LAVA now. > >>>>> > >>>>> Normal Job: > >>>>> https://validation.linaro.org/scheduler/job/282224 finished > >>>>> > >>>>> benchmark-lite: > >>>>> https://validation.linaro.org/scheduler/job/282229.0 finished > >>>>> > >>>>> > >>>>> linaro-android_armv8-android-juno-lsk-benchmark-application > >>>>> https://validation.linaro.org/scheduler/job/282231.0 finished > >>>>> > >>>>> > >>>>> linaro-android_armv8-android-juno-lsk_cts_32bit: > >>>>> https://validation.linaro.org/scheduler/job/282225.1 Canceled > >>>>> https://validation.linaro.org/scheduler/job/284313.0 >Incomplete > >>>>> https://validation.linaro.org/scheduler/job/286312.0 >resubmitted > >>>> > >>>> Finished > >>>>> > >>>>> > >>>>> linaro-android_armv8-android-juno-lsk_cts_64bit: > >>>>> https://validation.linaro.org/scheduler/job/282227.0 Canceled > >>>>> https://validation.linaro.org/scheduler/job/284315.0 >Incomplete > >>>>> https://validation.linaro.org/scheduler/job/286310.0 > resubmitted > >>>> > >>>> Finished. > >>>>> > >>>>> > >>>>> Benchmarks: > >>>>> https://validation.linaro.org/scheduler/job/282233.0 Canceled > >>>>> https://validation.linaro.org/scheduler/job/284317.0 >Incomplete > >>>>> https://validation.linaro.org/scheduler/job/286314.0 >resubmitted > >>>>> > >>>>> > >>>>> -- > >>>>> Best Regards, > >>>>> Yongqin Liu > >>>>> >--------------------------------------------------------------- > >>>>> #mailing list > >>>>> linaro-android@lists.linaro.org > >>>>> http://lists.linaro.org/mailman/listinfo/linaro-android > >>>> > >>>> > >>>> > >>>> > >>>> -- > >>>> Best Regards, > >>>> Yongqin Liu > >>>> --------------------------------------------------------------- > >>>> #mailing list > >>>> linaro-android@lists.linaro.org > >>>> http://lists.linaro.org/mailman/listinfo/linaro-android > >>> > >>> > >>> > >>> > >>> -- > >>> Best Regards, > >>> Yongqin Liu > >>> --------------------------------------------------------------- > >>> #mailing list > >>> linaro-android@lists.linaro.org > >>> http://lists.linaro.org/mailman/listinfo/linaro-android > >> > >> > >> > >> > >> -- > >> Best Regards, > >> Yongqin Liu > >> --------------------------------------------------------------- > >> #mailing list > >> linaro-android@lists.linaro.org > >> http://lists.linaro.org/mailman/listinfo/linaro-android > > > > > > > > > > -- > > With best regards, > > > > Jakub Pavelek > > > > Linaro Mobile Group project manager > > Linaro.org?Open source software for ARM SoCs > > >
-- With best regards,
Jakub Pavelek
Linaro Mobile Group project manager Linaro.org?Open source software for ARM SoCs
-- Best Regards, Yongqin Liu
#mailing list linaro-android@lists.linaro.org linaro-dev@lists.linaro.org http://lists.linaro.org/mailman/listinfo/linaro-android
Hello, There isn't a memory leak - there is fragmentation such that it can't allocate sufficient memory. Increasing the maximum heap growth from 48MB to 128MB+ means that caffeine can succeed for the same reason it does on the Nexus 9, which uses the same configuration.
Regards, Stuart
On 10/04/15 16:35, YongQin Liu wrote:
Hi, Steve, Stuart
On 24 March 2015 at 18:56, Steve Bannister <Steve.Bannister@arm.com mailto:Steve.Bannister@arm.com> wrote:
Hi guys, Re. caffinemark Stuart did some investigation and found the following. Not sure if it applies to your builds too. * Investigated why CaffeineMark wasn't working. For Juno, we use a small heapsize (48MB), which wasn't sufficient for caffeinemark to run successfully against. Looks like the allocator is subject to fragmentation issues. Using flounder's configuration (a 2GB 10inch tablet, apparently) works fineJust want to be clear about the problem. From what I investigated, the problem about this CaffeineMark is that it will get crashed on the result display window after waiting some time(like 10 minutes) because of the memory leak problem. We can see the result page on Juno and on other platforms(like the Nexus9)(which means the test itself should be finished), but the crash will cause problem for us to extract the test result with automation scripts.
So is your above solution to fix that memory leak problem on the result display window? Or your above solution is to fix another problem I did not realize?
Thanks, Yongqin Liu
Regards, Steve On 20/03/2015 14:16, "linaro-android-request@lists.linaro.org <mailto:linaro-android-request@lists.linaro.org>" <linaro-android-request@lists.linaro.org <mailto:linaro-android-request@lists.linaro.org>> wrote: >Send linaro-android mailing list submissions to > linaro-android@lists.linaro.org <mailto:linaro-android@lists.linaro.org> > >To subscribe or unsubscribe via the World Wide Web, visit > http://lists.linaro.org/mailman/listinfo/linaro-android >or, via email, send a message with subject or body 'help' to > linaro-android-request@lists.linaro.org <mailto:linaro-android-request@lists.linaro.org> > >You can reach the person managing the list at > linaro-android-owner@lists.linaro.org <mailto:linaro-android-owner@lists.linaro.org> > >When replying, please edit your Subject line so it is more specific >than "Re: Contents of linaro-android digest..." > > >Today's Topics: > > 1. Re: Update v1 : Jobs for build#335 (Mahanth Gouda) > 2. Re: Update v1 : Jobs for build#335 (Botao Sun) > > >---------------------------------------------------------------------- > >Message: 1 >Date: Thu, 19 Mar 2015 16:28:44 +0000 >From: Mahanth Gouda <mahanth.gouda@linaro.org <mailto:mahanth.gouda@linaro.org>> >To: YongQin Liu <yongqin.liu@linaro.org <mailto:yongqin.liu@linaro.org>> >Cc: Linaro Android Mailman List <linaro-android@lists.linaro.org <mailto:linaro-android@lists.linaro.org>>, > Jakub Pavelek <jakub.pavelek@linaro.org <mailto:jakub.pavelek@linaro.org>> >Subject: Re: Update v1 : Jobs for build#335 >Message-ID: > <CAKhO2P9rPYOf+Z8GNNrYMfoCqsJats=kJr0ZJO2Ug_BP=mBTbQ@mail.gmail.com <mailto:mBTbQ@mail.gmail.com>> >Content-Type: text/plain; charset="utf-8" > >Pleas find my observations in line: > > - nbench - Looks ok. > > > - GFXBench3 - CRASH - (I get this msg on the display - Your version of > the GFXBench is outdated, please refresh the app and start again. Let >me > know if you need to the log cat msg for this ?) > > > - gearses2eclair - Looks good to me. Tried a couple of times and it > looks OK. > > >Thanks, > >Mahanth > >On 19 March 2015 at 16:10, YongQin Liu <yongqin.liu@linaro.org <mailto:yongqin.liu@linaro.org>> wrote: > >> Hi, Mahanth >> >> Another 3 apks: >> >> - nbench >> - GFXBench3 >> - gearses2eclair, not finished yet, seems to be timed out >> >> Could you please help to check these 3 apks as well? >> They are failed to get the result at my jobs. >> >> >> Thanks, >> Yongqin Liu >> On 19 March 2015 at 23:49, Mahanth Gouda <mahanth.gouda@linaro.org <mailto:mahanth.gouda@linaro.org>> >>wrote: >> >>> Hi all, >>> Just wanted to let you know that I've picked up the build 335 and >>>started >>> manual testing this morning on Juno. >>> I'm half the way running these tests so far, hoping to finish the rest >>>in >>> next few hours time. >>> So, thought of giving you a quick results update: >>> >>> I'm testing only those benchmarks that've been marked by Yongqin as >>> failed. >>> Please let me know if I've missed something? >>> >>> Benchmark job 1: >>> --------------- >>> 1. antutu 5 - *HUNG/CRASH *(No log as the test didn't complete ) >>> 2. caffeinemark - *CRASH *- (Every other time this test fails) - logcat >>> & snapshot attached >>> 3. cfbench - *PASS* >>> >>> Benchmark job 2: >>> --------------- >>> 4. smartbench2014 - *PASS* >>> 5. gearsES1eclair - *CRASH* (There's no log for this because the script >>> does not complete, it's hung in between) It took long time >>>investigating >>> this, but could not get to the root cause of this issue. So, I've left >>>it >>> open for now. >>> >>> Benchmark job 3: >>> --------------- >>> 6. vellamo - testing in progress (currently the benchmark sounds like >>> it's paused on Networking Loader, since a while) >>> >>> >>> Benchmark job 4: >>> --------------- >>> 7. andebench2014 - >>> 8. javawhetstone - >>> 9. jbench - >>> 10. linpackjava - >>> 11. scimark - >>> >>> Please note - I've repeated all these tests at least 5 times in order >>>to >>> check the consistency. >>> >>> >>> Thanks & regards, >>> Mahanth >>> >>> >>> >>> >>> On 19 March 2015 at 10:59, Mahanth Gouda <mahanth.gouda@linaro.org <mailto:mahanth.gouda@linaro.org>> >>> wrote: >>> >>>> Hi Jakub, >>>> Thanks for the help offer, I've started with this, will surely seek >>>>your >>>> help when required. >>>> >>>> >>>> Thanks, >>>> Mahanth >>>> >>>> On 19 March 2015 at 09:19, Jakub Pavelek <jakub.pavelek@linaro.org <mailto:jakub.pavelek@linaro.org>> >>>> wrote: >>>> >>>>> Hi Botao and Mananth, >>>>> >>>>> if you need anything from us ping me or YongQin. We would like to >>>>> investigate the failing benchmark automation ASAP, otherwise we need >>>>>QAS to >>>>> run those benchmarks manually and we will lose trends tracking in >>>>> ImageReports. >>>>> >>>>> >>>>> >>>>> On 18 March 2015 at 19:01, Milosz Wasilewski < >>>>> milosz.wasilewski@linaro.org <mailto:milosz.wasilewski@linaro.org>> wrote: >>>>> >>>>>> Botao and Mahanth can help. Mahanth has access to the hw. As far as >>>>>>I >>>>>> remember: >>>>>> - antutu 5 crashes (was crashing before) before it finishes >>>>>> - caffeinemark crashes 4 out of 5 attempts >>>>>> - nbench produces empty results for some reason (only zeros) >>>>>> - GFXBench takes very long to run, we might have too short timeout >>>>>> - GearsES1 crashes always and should be abandoned >>>>>> >>>>>> I don't know about the others. This is what was happening before >>>>>> moving to 5.1. Mahanth, please take this as high priority to >>>>>> investigate. Botao migh help you with Gears as he did the >>>>>>automation. >>>>>> >>>>>> milosz >>>>>> >>>>>> >>>>>> On 18 March 2015 at 15:56, Jakub Pavelek <jakub.pavelek@linaro.org <mailto:jakub.pavelek@linaro.org>> >>>>>> wrote: >>>>>> > Milosz is out of office all this week. We need to find another way >>>>>> to check >>>>>> > things out :-/ >>>>>> > >>>>>> > On 18 March 2015 at 10:14, YongQin Liu <yongqin.liu@linaro.org <mailto:yongqin.liu@linaro.org>> >>>>>> wrote: >>>>>> >> >>>>>> >> Hi, Misolz >>>>>> >> >>>>>> >> Could you please help to check if the benchmark tests mentioned >>>>>> below >>>>>> >> can work on the Juno with the 5.1 build#335? >>>>>> >> >>>>>> >> Thanks, >>>>>> >> Yongqin Liu >>>>>> >> >>>>>> >> On 18 March 2015 at 17:11, YongQin Liu <yongqin.liu@linaro.org <mailto:yongqin.liu@linaro.org>> >>>>>> wrote: >>>>>> >>> >>>>>> >>> Hi, All >>>>>> >>> >>>>>> >>> The job for the benchmarks failed again. >>>>>> >>> Seems there is a problem to run the vellamo on Juno, the test >>>>>> timed out >>>>>> >>> and caused the sync problem. >>>>>> >>> So I split the job into 3 small jobs: >>>>>> >>> >>>>>> >>> Benchmark Job#1 >>>>>> >>> https://validation.linaro.org/scheduler/job/287238.0 finished. >>>>>> >>> but there is no benchmark result for benchmarks application of >>>>>> antutu5, >>>>>> >>> Caffeinemark, and CF-Bench. >>>>>> >>> And also all the benchmark result for nbench are zero. >>>>>> >>> >>>>>> >>> Benchmark Job#2 >>>>>> >>> https://validation.linaro.org/scheduler/job/287240.0 >>>>>> >>> Smartbench2012 seems not started correctly. >>>>>> >>> And the gearses1eclair test was timed out, and caused the sync >>>>>> problem. >>>>>> >>> >>>>>> >>> >>>>>> >>> Benchmark Job#3: vellamo job, timed out again. >>>>>> >>> https://validation.linaro.org/scheduler/job/287242.0 >>>>>> >>> >>>>>> >>> >>>>>> >>> Benchmark Job#4: tests for andebench2014, Javawhetstone, JBench, >>>>>> >>> LinpackJava, and Scimark >>>>>> >>> https://validation.linaro.org/scheduler/job/287503.0 >>>>>> >>> >>>>>> >>> >>>>>> >>> Thanks, >>>>>> >>> Yongqin Liu >>>>>> >>> On 17 March 2015 at 21:30, YongQin Liu <yongqin.liu@linaro.org <mailto:yongqin.liu@linaro.org>> >>>>>> wrote: >>>>>> >>>> >>>>>> >>>> Hi, Jakub >>>>>> >>>> >>>>>> >>>> The 2 cts jobs finished. >>>>>> >>>> >>>>>> >>>> Thanks, >>>>>> >>>> Yongqin Liu >>>>>> >>>> On 17 March 2015 at 15:47, YongQin Liu <yongqin.liu@linaro.org <mailto:yongqin.liu@linaro.org>> >>>>>> wrote: >>>>>> >>>>> >>>>>> >>>>> Hi, All >>>>>> >>>>> >>>>>> >>>>> Following are job status for Juno build #335 on LAVA: >>>>>> >>>>> The 3 resubmitted jobs are running on LAVA now. >>>>>> >>>>> >>>>>> >>>>> Normal Job: >>>>>> >>>>> https://validation.linaro.org/scheduler/job/282224 finished >>>>>> >>>>> >>>>>> >>>>> benchmark-lite: >>>>>> >>>>> https://validation.linaro.org/scheduler/job/282229.0 finished >>>>>> >>>>> >>>>>> >>>>> >>>>>> >>>>> linaro-android_armv8-android-juno-lsk-benchmark-application >>>>>> >>>>> https://validation.linaro.org/scheduler/job/282231.0 finished >>>>>> >>>>> >>>>>> >>>>> >>>>>> >>>>> linaro-android_armv8-android-juno-lsk_cts_32bit: >>>>>> >>>>> https://validation.linaro.org/scheduler/job/282225.1 Canceled >>>>>> >>>>> https://validation.linaro.org/scheduler/job/284313.0 >>>>>>Incomplete >>>>>> >>>>> https://validation.linaro.org/scheduler/job/286312.0 >>>>>>resubmitted >>>>>> >>>> >>>>>> >>>> Finished >>>>>> >>>>> >>>>>> >>>>> >>>>>> >>>>> linaro-android_armv8-android-juno-lsk_cts_64bit: >>>>>> >>>>> https://validation.linaro.org/scheduler/job/282227.0 Canceled >>>>>> >>>>> https://validation.linaro.org/scheduler/job/284315.0 >>>>>>Incomplete >>>>>> >>>>> https://validation.linaro.org/scheduler/job/286310.0 >>>>>> resubmitted >>>>>> >>>> >>>>>> >>>> Finished. >>>>>> >>>>> >>>>>> >>>>> >>>>>> >>>>> Benchmarks: >>>>>> >>>>> https://validation.linaro.org/scheduler/job/282233.0 Canceled >>>>>> >>>>> https://validation.linaro.org/scheduler/job/284317.0 >>>>>>Incomplete >>>>>> >>>>> https://validation.linaro.org/scheduler/job/286314.0 >>>>>>resubmitted >>>>>> >>>>> >>>>>> >>>>> >>>>>> >>>>> -- >>>>>> >>>>> Best Regards, >>>>>> >>>>> Yongqin Liu >>>>>> >>>>> >>>>>>--------------------------------------------------------------- >>>>>> >>>>> #mailing list >>>>>> >>>>> linaro-android@lists.linaro.org <mailto:linaro-android@lists.linaro.org> >>>>>> >>>>> http://lists.linaro.org/mailman/listinfo/linaro-android >>>>>> >>>> >>>>>> >>>> >>>>>> >>>> >>>>>> >>>> >>>>>> >>>> -- >>>>>> >>>> Best Regards, >>>>>> >>>> Yongqin Liu >>>>>> >>>> --------------------------------------------------------------- >>>>>> >>>> #mailing list >>>>>> >>>> linaro-android@lists.linaro.org <mailto:linaro-android@lists.linaro.org> >>>>>> >>>> http://lists.linaro.org/mailman/listinfo/linaro-android >>>>>> >>> >>>>>> >>> >>>>>> >>> >>>>>> >>> >>>>>> >>> -- >>>>>> >>> Best Regards, >>>>>> >>> Yongqin Liu >>>>>> >>> --------------------------------------------------------------- >>>>>> >>> #mailing list >>>>>> >>> linaro-android@lists.linaro.org <mailto:linaro-android@lists.linaro.org> >>>>>> >>> http://lists.linaro.org/mailman/listinfo/linaro-android >>>>>> >> >>>>>> >> >>>>>> >> >>>>>> >> >>>>>> >> -- >>>>>> >> Best Regards, >>>>>> >> Yongqin Liu >>>>>> >> --------------------------------------------------------------- >>>>>> >> #mailing list >>>>>> >> linaro-android@lists.linaro.org <mailto:linaro-android@lists.linaro.org> >>>>>> >> http://lists.linaro.org/mailman/listinfo/linaro-android >>>>>> > >>>>>> > >>>>>> > >>>>>> > >>>>>> > -- >>>>>> > With best regards, >>>>>> > >>>>>> > Jakub Pavelek >>>>>> > >>>>>> > Linaro Mobile Group project manager >>>>>> > Linaro.org?Open source software for ARM SoCs >>>>>> > >>>>>> >>>>> >>>>> >>>>> >>>>> -- >>>>> With best regards, >>>>> >>>>> Jakub Pavelek >>>>> >>>>> Linaro Mobile Group project manager >>>>> Linaro.org?Open source software for ARM SoCs >>>>> >>>>> >>>> >>> >> >> >> -- >> Best Regards, >> Yongqin Liu >> --------------------------------------------------------------- >> #mailing list >> linaro-android@lists.linaro.org <mailto:linaro-android@lists.linaro.org> <linaro-dev@lists.linaro.org <mailto:linaro-dev@lists.linaro.org>> >> http://lists.linaro.org/mailman/listinfo/linaro-android >> >
Hi, Stuart
Following are some information about the error I got with Nexus9(Google LRX22C) on the Caffeine result window after wait for several minutes(less than 10 minutes). It seems that the problem is on the result show window, not on during the execution of the benchmark.
I got the stopped dialog at the result window like the attached caffeine.png file.
Gott following exception in the logcat, for details you can check the attached caffeine.txt file
I/art ( 4509): WaitForGcToComplete blocked for 133.895ms for cause Alloc I/art ( 4509): Clamp target GC heap from 207MB to 192MB I/art ( 4509): Alloc concurrent mark sweep GC freed 6(192B) AllocSpace objects, 0(0B) LOS objects, 0% free, 191MB/192MB, paused 402us total 152.397ms I/art ( 4509): Forcing collection of SoftReferences for 16KB allocation I/art ( 4509): Clamp target GC heap from 207MB to 192MB I/art ( 4509): Alloc concurrent mark sweep GC freed 3(96B) AllocSpace objects, 0(0B) LOS objects, 0% free, 191MB/192MB, paused 758us total 152.635ms I/art ( 4509): WaitForGcToComplete blocked for 304.111ms for cause Background E/art ( 4509): Throwing OutOfMemoryError "Failed to allocate a 16396 byte allocation with 10304 free bytes and 10KB until OOM" I/art ( 4509): WaitForGcToComplete blocked for 53.314ms for cause Alloc E/AndroidRuntime( 4509): FATAL EXCEPTION: Thread-366 E/AndroidRuntime( 4509): Process: com.flexycore.caffeinemark, PID: 4509 E/AndroidRuntime( 4509): java.lang.OutOfMemoryError: Failed to allocate a 16396 byte allocation with 10304 free bytes and 10KB until OOM E/AndroidRuntime( 4509): at java.io.BufferedReader.<init>(BufferedReader.java:112) E/AndroidRuntime( 4509): at java.io.BufferedReader.<init>(BufferedReader.java:95)
And here is what I got from the ps command line, you can see that the RSS was increasing till it was force killed
u0_a82 4509 599 2097660 111404 ffffffff 00000000 S com.flexycore.caffeinemark u0_a82 4509 599 2097660 117772 ffffffff 00000000 S com.flexycore.caffeinemark u0_a82 4509 599 2096588 111368 ffffffff 00000000 S com.flexycore.caffeinemark u0_a82 4509 599 2096588 122160 ffffffff 00000000 S com.flexycore.caffeinemark u0_a82 4509 599 2096716 128836 ffffffff 00000000 S com.flexycore.caffeinemark u0_a82 4509 599 2096972 237420 ffffffff 00000000 S com.flexycore.caffeinemark u0_a82 4509 599 2097020 181848 ffffffff 00000000 S com.flexycore.caffeinemark u0_a82 4509 599 2099040 188032 ffffffff 00000000 S com.flexycore.caffeinemark u0_a82 4509 599 2097968 188260 ffffffff 00000000 S com.flexycore.caffeinemark u0_a82 4509 599 2097968 189096 ffffffff 00000000 S com.flexycore.caffeinemark u0_a82 4509 599 2097968 186928 ffffffff 00000000 S com.flexycore.caffeinemark u0_a82 4509 599 2097968 192560 ffffffff 00000000 S com.flexycore.caffeinemark u0_a82 4509 599 2097968 195332 ffffffff 00000000 S com.flexycore.caffeinemark u0_a82 4509 599 2097968 206860 ffffffff 00000000 S com.flexycore.caffeinemark u0_a82 4509 599 2097968 213164 ffffffff 00000000 S com.flexycore.caffeinemark u0_a82 4509 599 2097968 224904 ffffffff 00000000 S com.flexycore.caffeinemark u0_a82 4509 599 2098480 231828 ffffffff 00000000 S com.flexycore.caffeinemark u0_a82 4509 599 2098480 241944 ffffffff 00000000 S com.flexycore.caffeinemark u0_a82 4509 599 2098480 250816 ffffffff 00000000 S com.flexycore.caffeinemark u0_a82 4509 599 2098480 259696 ffffffff 00000000 S com.flexycore.caffeinemark u0_a82 4509 599 2098480 266236 ffffffff 00000000 S com.flexycore.caffeinemark u0_a82 4509 599 2098480 264624 ffffffff 00000000 S com.flexycore.caffeinemark u0_a82 4509 599 2098480 267888 ffffffff 00000000 S com.flexycore.caffeinemark u0_a82 4509 599 2098480 267848 ffffffff 00000000 S com.flexycore.caffeinemark u0_a82 4509 599 2098480 267844 ffffffff 00000000 S com.flexycore.caffeinemark u0_a82 4509 599 2099552 262680 ffffffff 00000000 S com.flexycore.caffeinemark u0_a82 4509 599 2099552 95960 ffffffff 00000000 S com.flexycore.caffeinemark <-----------------------Force killed here. u0_a82 4509 599 2099552 102676 ffffffff 00000000 S com.flexycore.caffeinemark u0_a82 4509 599 2099552 108388 ffffffff 00000000 S com.flexycore.caffeinemark u0_a82 4509 599 2099552 119948 ffffffff 00000000 S com.flexycore.caffeinemark u0_a82 4509 599 2099552 129204 ffffffff 00000000 S com.flexycore.caffeinemark u0_a82 4509 599 2099552 139728 ffffffff 00000000 S com.flexycore.caffeinemark u0_a82 4509 599 2099552 144680 ffffffff 00000000 S com.flexycore.caffeinemark
Here is the heap related property:
11:30:11 liuyq: lollipop$ adb shell getprop|grep heap [dalvik.vm.heapgrowthlimit]: [192m] [dalvik.vm.heapmaxfree]: [8m] [dalvik.vm.heapminfree]: [512k] [dalvik.vm.heapsize]: [512m] [dalvik.vm.heapstartsize]: [16m] [dalvik.vm.heaptargetutilization]: [0.75] 11:30:28 liuyq: lollipop$
Thanks, Yongqin Liu
On 10 April 2015 at 23:48, Stuart Monteith Stuart.Monteith@arm.com wrote:
Hello, There isn't a memory leak - there is fragmentation such that it can't allocate sufficient memory. Increasing the maximum heap growth from 48MB to 128MB+ means that caffeine can succeed for the same reason it does on the Nexus 9, which uses the same configuration.
Regards, Stuart
On 10/04/15 16:35, YongQin Liu wrote:
Hi, Steve, Stuart
On 24 March 2015 at 18:56, Steve Bannister <Steve.Bannister@arm.com mailto:Steve.Bannister@arm.com> wrote:
Hi guys, Re. caffinemark Stuart did some investigation and found thefollowing. Not sure if it applies to your builds too.
* Investigated why CaffeineMark wasn't working. For Juno, we use asmall heapsize (48MB), which wasn't sufficient for caffeinemark to run successfully against. Looks like the allocator is subject to fragmentation issues. Using flounder's configuration (a 2GB 10inch tablet, apparently) works fine
Just want to be clear about the problem. From what I investigated, the problem about this CaffeineMark is that it will get crashed on the result display window after waiting some time(like 10 minutes) because of the memory leak problem. We can see the result page on Juno and on other platforms(like the Nexus9)(which means the test itself should be finished), but the crash will cause problem for us to extract the test result with automation scripts.
So is your above solution to fix that memory leak problem on the result display window? Or your above solution is to fix another problem I did not realize?
Thanks, Yongqin Liu
Regards, Steve On 20/03/2015 14:16, "linaro-android-request@lists.linaro.orgmailto:linaro-android-request@lists.linaro.org" <linaro-android-request@lists.linaro.org mailto:linaro-android- request@lists.linaro.org> wrote:
>Send linaro-android mailing list submissions to > linaro-android@lists.linaro.org <mailto:linaro-android@lists.linaro.org> > >To subscribe or unsubscribe via the World Wide Web, visit > http://lists.linaro.org/mailman/listinfo/linaro-android >or, via email, send a message with subject or body 'help' to > linaro-android-request@lists.linaro.org mailto:linaro-android- request@lists.linaro.org > >You can reach the person managing the list at > linaro-android-owner@lists.linaro.org mailto: linaro-android-owner@lists.linaro.org > >When replying, please edit your Subject line so it is more specific >than "Re: Contents of linaro-android digest..." > > >Today's Topics: > > 1. Re: Update v1 : Jobs for build#335 (Mahanth Gouda) > 2. Re: Update v1 : Jobs for build#335 (Botao Sun) > > >-----------------------------------------------------------
> >Message: 1 >Date: Thu, 19 Mar 2015 16:28:44 +0000 >From: Mahanth Gouda <mahanth.gouda@linaro.org <mailto:mahanth.gouda@linaro.org>> >To: YongQin Liu <yongqin.liu@linaro.org mailto: yongqin.liu@linaro.org> >Cc: Linaro Android Mailman List <linaro-android@lists.linaro.org mailto:linaro-android@lists.linaro.org>, > Jakub Pavelek <jakub.pavelek@linaro.org mailto: jakub.pavelek@linaro.org> >Subject: Re: Update v1 : Jobs for build#335 >Message-ID: > <CAKhO2P9rPYOf+Z8GNNrYMfoCqsJats=kJr0ZJO2Ug_BP= mBTbQ@mail.gmail.com mailto:mBTbQ@mail.gmail.com> >Content-Type: text/plain; charset="utf-8" > >Pleas find my observations in line: > > - nbench - Looks ok. > > > - GFXBench3 - CRASH - (I get this msg on the display - Your version of > the GFXBench is outdated, please refresh the app and start again. Let >me > know if you need to the log cat msg for this ?) > > > - gearses2eclair - Looks good to me. Tried a couple of times and it > looks OK. > > >Thanks, > >Mahanth > >On 19 March 2015 at 16:10, YongQin Liu <yongqin.liu@linaro.org mailto:yongqin.liu@linaro.org> wrote: > >> Hi, Mahanth >> >> Another 3 apks: >> >> - nbench >> - GFXBench3 >> - gearses2eclair, not finished yet, seems to be timed out >> >> Could you please help to check these 3 apks as well? >> They are failed to get the result at my jobs. >> >> >> Thanks, >> Yongqin Liu >> On 19 March 2015 at 23:49, Mahanth Gouda < mahanth.gouda@linaro.org mailto:mahanth.gouda@linaro.org>
>>wrote: >> >>> Hi all, >>> Just wanted to let you know that I've picked up the build 335 and >>>started >>> manual testing this morning on Juno. >>> I'm half the way running these tests so far, hoping to finishthe rest >>>in >>> next few hours time. >>> So, thought of giving you a quick results update: >>> >>> I'm testing only those benchmarks that've been marked by Yongqin as >>> failed. >>> Please let me know if I've missed something? >>> >>> Benchmark job 1: >>> --------------- >>> 1. antutu 5 - *HUNG/CRASH *(No log as the test didn't complete ) >>> 2. caffeinemark - *CRASH *- (Every other time this test fails) - logcat >>> & snapshot attached >>> 3. cfbench - *PASS* >>> >>> Benchmark job 2: >>> --------------- >>> 4. smartbench2014 - *PASS* >>> 5. gearsES1eclair - *CRASH* (There's no log for this because the script >>> does not complete, it's hung in between) It took long time >>>investigating >>> this, but could not get to the root cause of this issue. So, I've left >>>it >>> open for now. >>> >>> Benchmark job 3: >>> --------------- >>> 6. vellamo - testing in progress (currently the benchmark sounds like >>> it's paused on Networking Loader, since a while) >>> >>> >>> Benchmark job 4: >>> --------------- >>> 7. andebench2014 - >>> 8. javawhetstone - >>> 9. jbench - >>> 10. linpackjava - >>> 11. scimark - >>> >>> Please note - I've repeated all these tests at least 5 times in order >>>to >>> check the consistency. >>> >>> >>> Thanks & regards, >>> Mahanth >>> >>> >>> >>> >>> On 19 March 2015 at 10:59, Mahanth Gouda < mahanth.gouda@linaro.org mailto:mahanth.gouda@linaro.org> >>> wrote: >>> >>>> Hi Jakub, >>>> Thanks for the help offer, I've started with this, will surely seek >>>>your >>>> help when required. >>>> >>>> >>>> Thanks, >>>> Mahanth >>>> >>>> On 19 March 2015 at 09:19, Jakub Pavelek < jakub.pavelek@linaro.org mailto:jakub.pavelek@linaro.org> >>>> wrote: >>>> >>>>> Hi Botao and Mananth, >>>>> >>>>> if you need anything from us ping me or YongQin. We would like to >>>>> investigate the failing benchmark automation ASAP, otherwise we need >>>>>QAS to >>>>> run those benchmarks manually and we will lose trends tracking in >>>>> ImageReports. >>>>> >>>>> >>>>> >>>>> On 18 March 2015 at 19:01, Milosz Wasilewski < >>>>> milosz.wasilewski@linaro.org mailto:milosz.wasilewski@ linaro.org> wrote: >>>>> >>>>>> Botao and Mahanth can help. Mahanth has access to the hw. As far as >>>>>>I >>>>>> remember: >>>>>> - antutu 5 crashes (was crashing before) before it finishes >>>>>> - caffeinemark crashes 4 out of 5 attempts >>>>>> - nbench produces empty results for some reason (only zeros) >>>>>> - GFXBench takes very long to run, we might have too short timeout >>>>>> - GearsES1 crashes always and should be abandoned >>>>>> >>>>>> I don't know about the others. This is what was happening before >>>>>> moving to 5.1. Mahanth, please take this as high priority to >>>>>> investigate. Botao migh help you with Gears as he did the >>>>>>automation. >>>>>> >>>>>> milosz >>>>>> >>>>>> >>>>>> On 18 March 2015 at 15:56, Jakub Pavelek < jakub.pavelek@linaro.org mailto:jakub.pavelek@linaro.org> >>>>>> wrote: >>>>>> > Milosz is out of office all this week. We need to find another way >>>>>> to check >>>>>> > things out :-/ >>>>>> > >>>>>> > On 18 March 2015 at 10:14, YongQin Liu < yongqin.liu@linaro.org mailto:yongqin.liu@linaro.org> >>>>>> wrote: >>>>>> >> >>>>>> >> Hi, Misolz >>>>>> >> >>>>>> >> Could you please help to check if the benchmark tests mentioned >>>>>> below >>>>>> >> can work on the Juno with the 5.1 build#335? >>>>>> >> >>>>>> >> Thanks, >>>>>> >> Yongqin Liu >>>>>> >> >>>>>> >> On 18 March 2015 at 17:11, YongQin Liu < yongqin.liu@linaro.org mailto:yongqin.liu@linaro.org>
>>>>>> wrote: >>>>>> >>> >>>>>> >>> Hi, All >>>>>> >>> >>>>>> >>> The job for the benchmarks failed again. >>>>>> >>> Seems there is a problem to run the vellamo on Juno, thetest >>>>>> timed out >>>>>> >>> and caused the sync problem. >>>>>> >>> So I split the job into 3 small jobs: >>>>>> >>> >>>>>> >>> Benchmark Job#1 >>>>>> >>> https://validation.linaro.org/scheduler/job/287238.0 finished. >>>>>> >>> but there is no benchmark result for benchmarks application of >>>>>> antutu5, >>>>>> >>> Caffeinemark, and CF-Bench. >>>>>> >>> And also all the benchmark result for nbench are zero. >>>>>> >>> >>>>>> >>> Benchmark Job#2 >>>>>> >>> https://validation.linaro.org/scheduler/job/287240.0 >>>>>> >>> Smartbench2012 seems not started correctly. >>>>>> >>> And the gearses1eclair test was timed out, and caused the sync >>>>>> problem. >>>>>> >>> >>>>>> >>> >>>>>> >>> Benchmark Job#3: vellamo job, timed out again. >>>>>> >>> https://validation.linaro.org/scheduler/job/287242.0 >>>>>> >>> >>>>>> >>> >>>>>> >>> Benchmark Job#4: tests for andebench2014, Javawhetstone, JBench, >>>>>> >>> LinpackJava, and Scimark >>>>>> >>> https://validation.linaro.org/scheduler/job/287503.0 >>>>>> >>> >>>>>> >>> >>>>>> >>> Thanks, >>>>>> >>> Yongqin Liu >>>>>> >>> On 17 March 2015 at 21:30, YongQin Liu < yongqin.liu@linaro.org mailto:yongqin.liu@linaro.org> >>>>>> wrote: >>>>>> >>>> >>>>>> >>>> Hi, Jakub >>>>>> >>>> >>>>>> >>>> The 2 cts jobs finished. >>>>>> >>>> >>>>>> >>>> Thanks, >>>>>> >>>> Yongqin Liu >>>>>> >>>> On 17 March 2015 at 15:47, YongQin Liu < yongqin.liu@linaro.org mailto:yongqin.liu@linaro.org>
>>>>>> wrote: >>>>>> >>>>> >>>>>> >>>>> Hi, All >>>>>> >>>>> >>>>>> >>>>> Following are job status for Juno build #335 on LAVA: >>>>>> >>>>> The 3 resubmitted jobs are running on LAVA now. >>>>>> >>>>> >>>>>> >>>>> Normal Job: >>>>>> >>>>> https://validation.linaro.org/scheduler/job/282224finished >>>>>> >>>>> >>>>>> >>>>> benchmark-lite: >>>>>> >>>>> https://validation.linaro.org/scheduler/job/282229.0 finished >>>>>> >>>>> >>>>>> >>>>> >>>>>> >>>>> linaro-android_armv8-android- juno-lsk-benchmark-application >>>>>> >>>>> https://validation.linaro.org/scheduler/job/282231.0 finished >>>>>> >>>>> >>>>>> >>>>> >>>>>> >>>>> linaro-android_armv8-android-juno-lsk_cts_32bit: >>>>>> >>>>> https://validation.linaro.org/scheduler/job/282225.1 Canceled >>>>>> >>>>> https://validation.linaro.org/scheduler/job/284313.0 >>>>>>Incomplete >>>>>> >>>>> https://validation.linaro.org/scheduler/job/286312.0 >>>>>>resubmitted >>>>>> >>>> >>>>>> >>>> Finished >>>>>> >>>>> >>>>>> >>>>> >>>>>> >>>>> linaro-android_armv8-android-juno-lsk_cts_64bit: >>>>>> >>>>> https://validation.linaro.org/scheduler/job/282227.0 Canceled >>>>>> >>>>> https://validation.linaro.org/scheduler/job/284315.0 >>>>>>Incomplete >>>>>> >>>>> https://validation.linaro.org/scheduler/job/286310.0 >>>>>> resubmitted >>>>>> >>>> >>>>>> >>>> Finished. >>>>>> >>>>> >>>>>> >>>>> >>>>>> >>>>> Benchmarks: >>>>>> >>>>> https://validation.linaro.org/scheduler/job/282233.0 Canceled >>>>>> >>>>> https://validation.linaro.org/scheduler/job/284317.0 >>>>>>Incomplete >>>>>> >>>>> https://validation.linaro.org/scheduler/job/286314.0 >>>>>>resubmitted >>>>>> >>>>> >>>>>> >>>>> >>>>>> >>>>> -- >>>>>> >>>>> Best Regards, >>>>>> >>>>> Yongqin Liu >>>>>> >>>>> >>>>>>------------------------------------------------------
>>>>>> >>>>> #mailing list >>>>>> >>>>> linaro-android@lists.linaro.org <mailto:linaro-android@lists.linaro.org> >>>>>> >>>>> http://lists.linaro.org/mailman/listinfo/linaro-android >>>>>> >>>> >>>>>> >>>> >>>>>> >>>> >>>>>> >>>> >>>>>> >>>> -- >>>>>> >>>> Best Regards, >>>>>> >>>> Yongqin Liu >>>>>> >>>> ------------------------------
>>>>>> >>>> #mailing list >>>>>> >>>> linaro-android@lists.linaro.org <mailto:linaro-android@lists.linaro.org> >>>>>> >>>> http://lists.linaro.org/mailman/listinfo/linaro-android >>>>>> >>> >>>>>> >>> >>>>>> >>> >>>>>> >>> >>>>>> >>> -- >>>>>> >>> Best Regards, >>>>>> >>> Yongqin Liu >>>>>> >>> ------------------------------
>>>>>> >>> #mailing list >>>>>> >>> linaro-android@lists.linaro.org <mailto:linaro-android@lists.linaro.org> >>>>>> >>> http://lists.linaro.org/mailman/listinfo/linaro-android >>>>>> >> >>>>>> >> >>>>>> >> >>>>>> >> >>>>>> >> -- >>>>>> >> Best Regards, >>>>>> >> Yongqin Liu >>>>>> >> ------------------------------
>>>>>> >> #mailing list >>>>>> >> linaro-android@lists.linaro.org <mailto:linaro-android@lists.linaro.org> >>>>>> >> http://lists.linaro.org/mailman/listinfo/linaro-android >>>>>> > >>>>>> > >>>>>> > >>>>>> > >>>>>> > -- >>>>>> > With best regards, >>>>>> > >>>>>> > Jakub Pavelek >>>>>> > >>>>>> > Linaro Mobile Group project manager >>>>>> > Linaro.org?Open source software for ARM SoCs >>>>>> > >>>>>> >>>>> >>>>> >>>>> >>>>> -- >>>>> With best regards, >>>>> >>>>> Jakub Pavelek >>>>> >>>>> Linaro Mobile Group project manager >>>>> Linaro.org?Open source software for ARM SoCs >>>>> >>>>> >>>> >>> >> >> >> -- >> Best Regards, >> Yongqin Liu >> --------------------------------------------------------------- >> #mailing list >> linaro-android@lists.linaro.org mailto:linaro-android@lists. linaro.org <linaro-dev@lists.linaro.org mailto:linaro-dev@lists.linaro.org> >> http://lists.linaro.org/mailman/listinfo/linaro-android >> >
linaro-android@lists.linaro.org
-
Steve Bannister -
Stuart Monteith -
YongQin Liu