1235257 - crash around nsAppShell::ProcessNextNativeEvent (@0xa49d336b | @0x7198ffff | nsAppShell::ProcessNextNativeEvent(bool)) (if Rapport is installed into the environment)

Status: RESOLVED FIXED

(Core :: General, defect)

Description

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)(Away until 6/30 except 24)]

This bug was filed from the Socorro interface and is 
report bp-2aa7ced4-8828-481b-9b23-dc94f2151227.
=============================================================

The victim said that this bug suddenly started to reproduce in this a couple of hours. And even she does nothing, this is reproduced.

Looks like that some other users reproduce this crash (crashed at @0xa49d336b after @0x7198ffff after nsAppShell::ProcessNextNativeEvent(bool)) a lot.
https://crash-stats.mozilla.com/report/list?product=Firefox&signature=nsAppShell%3A%3AProcessNextNativeEvent

Looks like a new regression of 43 or 43.0.1?

Comment 1

[Tooru Fujisawa [:arai]]

according to Correlations, it's somehow related to Trusteer Rapport.
> 100% (21/21) RapportTanzanEx430.DLL
> 100% (21/21) RapportGP.DLL
> 100% (21/21) rooksdol.dll
> 100% (21/21) TRF.dll
> 100% (21/21) RapportUtil.dll
> 100% (21/21) rooksbas.dll

similar issue reported before in bug 786153

Comment 2

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)(Away until 6/30 except 24)]

arai-san, thank you for your comment! You're right. The victim said, the latest Rapport seems to fix their bug.

Can we do something for this bug?

Comment 3

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)(Away until 6/30 except 24)]

FYI: Major bank in Japan, the bank of Mitsubishi-Tokyo UFJ, recommend to install Rapport for their net banking users. I guess that a lot of Japanese Firefox users meet this bug.

Comment 4

[Benjamin Smedberg]

You could add the broken versions of the Rapport DLLs to our DLL blocklist at mozglue/build/WindowsDllBlocklist.cpp if you know what versions those are. Trusteer Rapport has been a longstanding source of stability issues within Firefox.

Comment 5

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)(Away until 6/30 except 24)]

Thank you.

I'm not sure which dll causes this crash, though, looks like one of following dll is the cause because their symbols are not available and the stack just recorded the address:

* RapportGP.DLL 	3.5.1507.99
* RapportUtil.dll 	3.5.1507.99
* rooksbas.dll 	2.143.0.0

https://crash-stats.mozilla.com/report/list?product=Firefox&signature=nsAppShell%3A%3AProcessNextNativeEvent#tab-reports

Do you have any idea, Makoto-san?

Comment 6

[Makoto Kato [:m_kato]]

Humm, maybe, although a module loaded around 0x7149ffff, it was unloaded.  Or any module allocates procedure around it.  So I cannot find root cause except to Rapport.

When I check Rapport version, all signature seems to be the latest version.  So we should contact to IBM at first.  Because we have no information whether this issue is fixed by them.

Comment 7

[Makoto Kato [:m_kato]]

I will send technical support of Rapport for this.

Comment 8

[[:philipp]]

[Tracking Requested - why for this release]:
in early crash data for 44rc2 this crash is on #5, making up 1.66% of all crashers.

Comment 9

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)(Away until 6/30 except 24)]

According to Makoto-san, this reproducible with the latest Rapport.

Comment 10

[Sylvestre Ledru [:Sylvestre]]

I contacted them.

Comment 11

[Makoto Kato [:m_kato]]

Firefox 44 crashes are that version 3.5.1507.105 (the latest version?) is installed.

Comment 12

[Makoto Kato [:m_kato]]

(In reply to Sylvestre Ledru [:sylvestre] from comment #10)
> I contacted them.

Good news.  They confirm this issue, then they will release new fixed version soon.

Comment 13

[[:philipp]]

due to the recent development in crash stats i'm going to mark this issue as fixed by trusteer's update:
https://crash-stats.mozilla.com/signature/?date=%3E2016-01-15&signature=nsAppShell%3A%3AProcessNextNativeEvent#graphs