Urgent! Android version crashes like hell

Solar2D General Questions/Discussion
#1

I didn’t do any change to my app recently but my users started to complain my app is crashing very often.

I looked up Google report, there are indeed many crashes that I’ve never seen before like below:

#00 pc 000117cc /system/lib/libc.so (dlfree+1191) #01 pc 0000dd23 /system/lib/libc.so (free+10) #02 pc 000823ad /system/lib/libcrypto.so (CRYPTO\_free+24) #03 pc 0002ab09 /system/lib/libssl.so (ssl\_parse\_serverhello\_tlsext+244) #04 pc 00016b3d /system/lib/libssl.so (ssl3\_get\_server\_hello+924) #05 pc 00019633 /system/lib/libssl.so (ssl3\_connect+642) #06 pc 00024fd1 /system/lib/libssl.so (SSL\_do\_handshake+72) #07 pc 0000c5f9 /system/lib/libjavacrypto.so #08 pc 0001eacc /system/lib/libdvm.so (dvmPlatformInvoke+112) #09 pc 0004efab /system/lib/libdvm.so (dvmCallJNIMethod(unsigned int const\*, JValue\*, Method const\*, Thread\*)+398) #10 pc 00027ee0 /system/lib/libdvm.so #11 pc 0002ee88 /system/lib/libdvm.so (dvmMterpStd(Thread\*)+76) #12 pc 0002c534 /system/lib/libdvm.so (dvmInterpret(Thread\*, Method const\*, JValue\*)+184) #13 pc 000613b3 /system/lib/libdvm.so (dvmCallMethodV(Thread\*, Method const\*, Object\*, bool, JValue\*, std::\_\_va\_list)+338) #14 pc 000613d7 /system/lib/libdvm.so (dvmCallMethod(Thread\*, Method const\*, Object\*, JValue\*, ...)+20) #15 pc 000560bb /system/lib/libdvm.so #16 pc 0000d280 /system/lib/libc.so (\_\_thread\_entry+72) #17 pc 0000d41c /system/lib/libc.so (pthread\_create+240)

I think it might be related to webView or http request because my app uses webView to open many dynamic web pages and also makes a lot of http requests for images dynamically (based on the data polling from the Internet) and the stack shows something about “ssl” etc.

Corona please help to identify this problem. The crash really was not there but started to happen since April. There are just too many crashes reported by our users and we can see the crash reports is mounting in the Google report. It’s not just a random/forgivable crash but something that needs immediate attention.

#2

Found an online article having similar issue: http://stackoverflow.com/questions/16950492/abstractgoogleclient-fatal-signal-11-sigsegv

Is it the same problem & possible fix for CoronaSDK?

Corona, please help. I am bombarded with user complaints and 1-star reviews.

#3

Here others discussion with same problem: https://github.com/square/okhttp/issues/1338

#4

If you haven’t changed anything, then there is nothing magical that we do behind the scene that would fix this.  It sounds like Google broke something.  I’ll alert Engineering.

Rob

#5

There are two possibilities:

(1) I did update my app with newer-version CoronaSDK and CoronaSDK might have changed something

(2) Google broke something

Even for case (2) that Google broke something, the crash has been discussed online with some possible fix, I hope Corona can try to work around it even it’s Google’s fault.

#6

You said in the top post that you did not make any changes.  Did you update the app?  What version of Corona SDK did you use to update your app with?

Rob

#7

I wasn’t sure about the exact time this crash started. I didn’t make any change in last two months but the crash report might started to happen earlier (and I wasn’t paying attention until recently that too many users are complaining so I think it’s not just a random crash due to rare cases but some serious one) and might be involved with my last update in March. Besides, even for my last update, I didn’t change anything related to this kind of http requests. I was just trying to isolate the problem to point out that my code is not causing this problem.

Anyway, the CoronaSDK version is 2564.

#8

Engineering looked at the log and there is nothing Corona related in it.  It’s coming from somewhere in the system and there isn’t much useful in the crash log.  Since the other links above are not Corona related, we can only deduce that Google broke something.

Rob

#9

@Rob

Does CoronaSDK use “OkHttp” ?  Could you check this with the engineer?

#10

Just wondering, are you using HTTPS or any SSL in your app ?

#11

I am just using webView and http downloading. I am not doing something special.

#12

@joe528,

Corona SDK doesn’t use OkHttp, so those links about solutions for OkHttp aren’t the most helpful for us.

Based on the fact that products unrelated to Corona are reporting similar issues, that’s more evidence that the problem lies with Google and not Corona SDK.

That being said, it may be worthwhile to poke around Android forums for any news on the issue.

#13

Despite this leaning towards a Google issue, we want to see if there’s anything we can do to help mitigate the issue from occurring within Corona.

That being said, to everyone who’s commented on the issue we’re curious about the following:

* Can you reproduce this issue on your own device (not user crash report)?

* What OS versions is this crashing on?

* What devices is it crashing on?

* What plugins are you guys using in the apps that exhibit this behavior?

#14

Thank you for your further investigation. It’s really appreciated.

  1. I cannot reproduce it. It seems the crash does not not happen universally but it happens for certain device & Android version consistently. So the users with the certain configuration keeps experiencing this crash.

  2. From the report, it looks like the OS version is always Android 4.4

  3. Galaxy Note 2/3/4, HTC One, HTC Desire, Galaxy Tab 8, Galaxy S4, Xperia ZL, etc.

  4. My app uses following plugins:  Flurry, Google Play Services, Facebook, Chartboost.  They have been there and the crash wasn’t happening until recently.

#15

@ajaymccaled-work

@Corona

@Rob

Would the problem be related to “HttpClient is not threadsafe”? Some discussion here: https://geonet.esri.com/thread/94377  where the stack trace of the crash is exactly the same as what I’ve been seeing.

I just updated a new version of my app with latest CoronaSDK and the crash reports from Google still come in like hell. I really hope Corona can look further into this issue to see any possible workaround. The rate of the crash is never seen before & it has caused my installation number started to decline (# of uninstallation > # of installation after this crash issue is observed).

#16

Found an online article having similar issue: http://stackoverflow.com/questions/16950492/abstractgoogleclient-fatal-signal-11-sigsegv

Is it the same problem & possible fix for CoronaSDK?

Corona, please help. I am bombarded with user complaints and 1-star reviews.

#17

Here others discussion with same problem: https://github.com/square/okhttp/issues/1338

#18

If you haven’t changed anything, then there is nothing magical that we do behind the scene that would fix this.  It sounds like Google broke something.  I’ll alert Engineering.

Rob

#19

There are two possibilities:

(1) I did update my app with newer-version CoronaSDK and CoronaSDK might have changed something

(2) Google broke something

Even for case (2) that Google broke something, the crash has been discussed online with some possible fix, I hope Corona can try to work around it even it’s Google’s fault.

#20

You said in the top post that you did not make any changes.  Did you update the app?  What version of Corona SDK did you use to update your app with?

Rob