Realm-java: Investigate connection between BadVersionException and calling load() on async query results

[Zhuinden]

Ported hypothesis from http://stackoverflow.com/questions/41804318/realmchangelistener-does-not-execute-when-added-shortly-after-asynchronous-write?noredirect=1#comment70833516_41804318

Realm version: 2.3.0

[ppamorim]

It is a know issue. Should be fixed on: https://github.com/realm/realm-java/pull/3834

[kneth]

The SO user is claiming he got it to work.

[benj56]

@kneth
I think this is a misunderstanding, I did not get it to work, it is just working more reliable depending on delays between the write and the queries, which is not really a solution.

I believe there might be a problem involving the "caller thread is behind the worker thread" case in the HandlerController.java.

Anyway, maybe the root cause of the problem that leads to my ChangeListener not triggering is what causes the BadVersionException (and sometimes SEGFAULT instead) when calling load() on these Results.

This is the SIGSEGV error message:

01-26 15:30:26.626 29367-29367/com.appname A/libc: Fatal signal 11 (SIGSEGV), code 2, fault addr 0x72fd8bcd20 in tid 29367 (com.appname)

                                                   [ 01-26 15:30:26.627   375:  375 W/         ]
                                                   debuggerd: handling request: pid=29367 uid=10117 gid=10117 tid=29367

01-26 15:30:26.731 29931-29931/? A/DEBUG: *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** ***
01-26 15:30:26.731 29931-29931/? A/DEBUG: Build fingerprint: 'google/angler/angler:7.1.1/N4F26I/3532671:user/release-keys'
01-26 15:30:26.732 29931-29931/? A/DEBUG: Revision: '0'
01-26 15:30:26.732 29931-29931/? A/DEBUG: ABI: 'arm64'
01-26 15:30:26.732 29931-29931/? A/DEBUG: pid: 29367, tid: 29367, name: com.appname >>> com.appname <<<
01-26 15:30:26.732 29931-29931/? A/DEBUG: signal 11 (SIGSEGV), code 2 (SEGV_ACCERR), fault addr 0x72fd8bcd20
01-26 15:30:26.732 29931-29931/? A/DEBUG:     x0   00000072fd8bcd20  x1   00000072fd8bcd20  x2   00000072faace93c  x3   0000000000000001
01-26 15:30:26.732 29931-29931/? A/DEBUG:     x4   0000000000000030  x5   4000000000000000  x6   0000007fff2e6b40  x7   0000000000000000
01-26 15:30:26.732 29931-29931/? A/DEBUG:     x8   0000000000000001  x9   0000000000000000  x10  0000000000000000  x11  0000000000000000
01-26 15:30:26.732 29931-29931/? A/DEBUG:     x12  0000000000000070  x13  000000000000001d  x14  0000000000000000  x15  0000000000000002
01-26 15:30:26.732 29931-29931/? A/DEBUG:     x16  0000007fff2e6bf0  x17  0000007317c23908  x18  0000000c00000000  x19  00000072fd8cd3a0
01-26 15:30:26.732 29931-29931/? A/DEBUG:     x20  00000072fd8cd3a0  x21  000000731753c080  x22  00000072fd84a400  x23  000000731743e000
01-26 15:30:26.732 29931-29931/? A/DEBUG:     x24  000000730d0eff70  x25  0000000012e0fca0  x26  0000000014fecee0  x27  00000072fd8cd3a0
01-26 15:30:26.732 29931-29931/? A/DEBUG:     x28  0000000000000000  x29  0000007fff2e79f0  x30  00000072faa9f27c
01-26 15:30:26.732 29931-29931/? A/DEBUG:     sp   0000007fff2e79f0  pc   00000072fd8bcd20  pstate 0000000060000000
01-26 15:30:26.736 29931-29931/? A/DEBUG: backtrace:
01-26 15:30:26.736 29931-29931/? A/DEBUG:     #00 pc 00000000000bcd20  [anon:libc_malloc:00000072fd800000]
01-26 15:30:26.736 29931-29931/? A/DEBUG:     #01 pc 0000000000033278  /data/app/com.appname-1/lib/arm64/librealm-jni.so
01-26 15:30:26.736 29931-29931/? A/DEBUG:     #02 pc 0000000000072360  /data/app/com.appname-1/lib/arm64/librealm-jni.so (Java_io_realm_internal_TableQuery_nativeImportHandoverTableViewIntoSharedGroup+792)
01-26 15:30:26.736 29931-29931/? A/DEBUG:     #03 pc 0000000000065340  /data/data/com.appname/cache/slice-io.realm-realm-android-library-2.3.0_0b3de97409478aa637365bab43eb294ef79e31b9-classes.dex (offset 0x70000)

[benj56]

@cmelchior @kneth
Would it make sense to open this again or could I create a new issue?

[cmelchior]

@benj56 I re-opened this issue

[kneth]

@benj56 I hope that we have a snapshot release including #3834 soon.

[benj56]

@kneth Would very much appreciate it. I've been looking at #3834 every few days, but can't really tell the progress. Do you think a snapshot is ready within one-two weeks? Thanks!

[kneth]

@benj56 #3834 is a big PR so reviewing it to a state where we can merge it, is taking longer than expected. I do hope that it is merged within two weeks from now!

[cmelchior]

@benj56 Can you reproduce this with Realm Java 3.0.0, that should contain a fix for this?

[benj56]

@cmelchior It is fixed, thanks.