752772 - org/mozilla/gecko/sqlite/SQLiteBridge;.sqliteCallWithDb() crashes with "received null jstring" error

Status: RESOLVED WORKSFORME

(Firefox for Android Graveyard :: Data Providers, defect)

Description

[Chris Peterson [:cpeterson]]

Attachment: sqliteCallWithDb-null-jstring-crash.log

Nightly crashes on my Galaxy Nexus fairly often when I launch it. Here is an excerpt from my logcat. See the attached log for more info, including some Sync error messages and StrictMode violations.


D/GeckoPasswordsProvider( 7944): Successfully created database helper for profile: default
D/GeckoPasswordsProvider( 7944): No profile provided, using default
D/GeckoPasswordsProvider( 7944): Using path: /data/data/org.mozilla.fennec/files/mozilla/bx9mu8ke.default
D/GeckoPasswordsProvider( 7944): Successfully created database helper for profile: default
D/GeckoPasswordsProvider( 7944): No profile provided, using default
D/GeckoPasswordsProvider( 7944): Using path: /data/data/org.mozilla.fennec/files/mozilla/bx9mu8ke.default
D/GeckoPasswordsProvider( 7944): Successfully created database helper for profile: default
W/dalvikvm( 7944): JNI WARNING: received null jstring
W/dalvikvm( 7944):              in Lorg/mozilla/gecko/sqlite/SQLiteBridge;.sqliteCallWithDb:(JLjava/lang/String;[Ljava/lang/String;[J)Lorg/mozilla/gecko/sqlite/MatrixBlobCursor; (GetStringUTFChars)
E/dalvikvm( 7944): VM aborting
F/libc    ( 7944): Fatal signal 11 (SIGSEGV) at 0xdeadd00d (code=1)
I/DEBUG   (  116): *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** ***
I/DEBUG   (  116): Build fingerprint: 'google/yakju/maguro:4.0.4/IMM76I/330937:user/release-keys'
I/DEBUG   (  116): pid: 7944, tid: 7959  >>> org.mozilla.f3nn3c.PasswordsProvider <<<
I/DEBUG   (  116): signal 11 (SIGSEGV), code 1 (SEGV_MAPERR), fault addr deadd00d

Comment 1

[Chris Peterson [:cpeterson]]

I hit this crash (with Nightly 15) on my Galaxy Nexus about every 10-60 seconds. Neither Aurora 14 nor my private builds of m-c 15 crash, so something about my Nightly profile data must be hosed.

Comment 2

[Kartikaya Gupta (email:[e-mail])]

Cc'ing :wesj who wrote the relevant code.

Comment 3

[Chris Peterson [:cpeterson]]

wesj, this PasswordsProvider exception consistently crashes Nightly on my Galaxy Nexus every 1-3 minutes. It is very annoying. I suspect my profile has some bogus password data stored in my PasswordsProvider, but Fennec should not be crashing.

Comment 4

[Mogens Isager]

I just updated my phone from Gingerbread to Jellybean+ and started hitting this crash. It appears to happen when you have password entries without a user name. Looking through my list of saved passwords this seem to happen on a number of sites. I have created a small test page [1] that will allow you to easily create an entry that will trigger the crash. I have tested this using beta, aurora, and nightly and they all crash.

[1] http://isager.net/synctest.html

Comment 5

[Wesley Johnston (:wesj)]

I suspect this needs the patch in bug 745384. Unfortunately, it makes our testing machines very angry. I'll try to dig into it again. Thanks.

Comment 6

[Richard Newman [:rnewman]]

Wes, any movement on this? Apparently this is still very much a thing.

Comment 7

[Lukas Blakk [:lsblakk] use ?needinfo]

This may still be a 'thing' but it's been around since 15 and unless it's in the topcrash list for Fennec, it's not a candidate for release tracking in FF23 - please re-nom if the volume (in FF23) changes for the worse.

Comment 8

[Richard Newman [:rnewman]]

CCing margaret and sola: might run into this when using the SQLiteBridge for content providers.

Comment 9

[flying sheep]

Attachment: logcat.log

i have a similar crash. pleas help, i really want to use firefox…

Comment 10

[Richard Newman [:rnewman]]

(In reply to flying sheep from comment #9)

> i have a similar crash. pleas help, i really want to use firefox…

You're seeing Bug 966143, not this bug.

If you're using Master Password, consider turning it off, see if it fixes your issue.

Comment 12

[Martyn Haigh (:mhaigh)]

I'm getting this issue: https://etherpad.mozilla.org/1sIojsYzVa

Running dev build on Android M - N6

Comment 13

[Gian-Carlo Pascutto [:gcp]]

If you can reproduce it reliably and trace where the nulls are actually coming from that would be great. We've been unable to fix this in the past.

Comment 14

[Richard Newman [:rnewman]]

Perhaps fixed by Bug 1135281.

Comment 15

[Martyn Haigh (:mhaigh)]

I can no longer reproduce

Comment 16

[Richard Newman [:rnewman]]

Let's call this done.