1305001 - Crash in uipfull.dll@0x6b0e2 (BaiduPinyin IME)

Status: RESOLVED WORKSFORME

(External Software Affecting Firefox :: Other, defect, P3)

Description

[[:philipp]]

This bug was filed from the Socorro interface and is 
report bp-46997b0c-df46-4781-a904-3010a2160923.
=============================================================
Crashing Thread (67)
Frame 	Module 	Signature 	Source
0 		@0x37ec743a 	
Ø 1 	uipfull.dll 	uipfull.dll@0x6b0e2 	
Ø 2 	uipfull.dll 	uipfull.dll@0x6b0e2 	
Ø 3 	uipfull.dll 	uipfull.dll@0x6ac81 	
Ø 4 	uipfull.dll 	uipfull.dll@0x6abd7 	
Ø 5 	uipfull.dll 	uipfull.dll@0x1c4f52 	
Ø 6 	uipfull.dll 	uipfull.dll@0x1c507a 	
7 	kernel32.dll 	BaseThreadStart

crashes with these signature first appeared on 2016-09-03 and got more frequent in the past days and weeks. uipfull.dll appears to be part of a chinese input method editor provided by baidu and mainly chinese builds are affected.

on a manual sport check of a couple of crash reports they all seemed to have version 4.1.3051.0 of UIPFull.dll present, whereas when you search the internet for that module name you mostly end up with results for v2 & v3 of this library. so the recent rise in crashiness might be caused by a program update by baidu and blocklisting the new dll version might be effective in reducing this crash (though maybe this will cause an issue with functionality once this IME is used with firefox).

Comment 1

[Marco Castelluccio [:marco]]

322 crash reports have been analyzed and the following versions have been found:
 - uipfull.dll
   - 3.4.2483.11: 1
   - 4.0.2955.0: 12
   - 4.0.2923.0: 2
   - 4.1.3051.0: 301
   - 4.0.2926.0: 2
   - 3.3.2.1102: 1
   - 3.4.2717.9: 3

Comment 2

[Benjamin Smedberg]

This appears to be a bug in their most recent update. It's possible (likely?) that this affects many apps and not just Firefox. Also, if people are using this input method, blocking the DLL would likely make Firefox much less usable.

Unless we have evidence that this causes a crash all the time for these users, I don't think blocklisting is a good choice here. Based on the current data, it seems very unlikely that this is causing consistent crashes.

Marco, if you think it's worthwhile, please reach out to Baidu to inform them of the problem. Otherwise we should close this INCOMPLETE.

Comment 3

[BugBot [:suhaib / :marco/ :calixte]]

Crash volume for signature 'uipfull.dll@0x6b0e2':
 - nightly (version 52): 0 crashes from 2016-09-19.
 - aurora  (version 51): 16 crashes from 2016-09-19.
 - beta    (version 50): 31 crashes from 2016-09-20.
 - release (version 49): 300 crashes from 2016-09-05.
 - esr     (version 45): 158 crashes from 2016-06-01.

Crash volume on the last weeks (Week N is from 10-03 to 10-09):
            W. N-1  W. N-2
 - nightly       0       0
 - aurora       15       1
 - beta         21      10
 - release     231      69
 - esr          62      51

Affected platform: Windows

Crash rank on the last 7 days:
           Browser     Content   Plugin
 - nightly
 - aurora  #77
 - beta    #656
 - release #217
 - esr     #184

Comment 4

[Marco Castelluccio [:marco]]

Dees, do you have contacts at Baidu?

Comment 5

[Desigan Chinniah [:cyberdees] [:dees] [London - GMT]]

(In reply to Marco Castelluccio [:marco] from comment #4)
> Dees, do you have contacts at Baidu?

Not off-hand - but let me investigate a little.

D'

Comment 6

[xunxun]

See also: https://bugzilla.mozilla.org/show_bug.cgi?id=1306563

Comment 8

[Yanfang Liu]

Our QA could not reproduce this crash, our BD team have contacted Baidu for more information, but not got any feedback yet. We will continue to follow up this problem.

Comment 9

[Jack Guo]

Hi, We got Baidu's response below. They need crash dump and other crash information to reproduce this crash. Can someone provide the crash dump file asap please? Thanks. 


========================
您好，jack：

我们已经收到您的反馈并已经着手跟进解决，但从邮件中提供的信息，我们无法确认具体原因。

现在需要你帮忙提供一下crash dump以及出现crash的场景，可联系我的QQ：654804087  或者微信：654804087

谢谢！

Comment 10

[Benjamin Smedberg]

Per our privacy policy, we cannot provide crash dump files to a 3rd party without explicit permission from a user (or unless the crash was caught by an employee). I don't think we have either of those currently.

If baidu can provide us with symbols (PDB files) we can provide them with detailed stack trace data.

Comment 11

[Hector Zhao [:hectorz]]

(In reply to Benjamin Smedberg [:bsmedberg] from comment #10)
> Per our privacy policy, we cannot provide crash dump files to a 3rd party
> without explicit permission from a user (or unless the crash was caught by
> an employee). I don't think we have either of those currently.

We'll try to reach out to users who left a email for STRs, now that I have access to emails again.

> 
> If baidu can provide us with symbols (PDB files) we can provide them with
> detailed stack trace data.

I'll let Baidu know about this.

Comment 12

[Desigan Chinniah [:cyberdees] [:dees] [London - GMT]]

Adding :hectorz to the Baidu Pinyin IME email thread...

Comment 13

[Jim Mathies [:jimm]]

Hey Dees, any update on this from your email thread?

Comment 14

[Marco Castelluccio [:marco]]

Long standing bug with low number of crash reports, we are not going to fix it by 57.

Comment 15

[Desigan Chinniah [:cyberdees] [:dees] [London - GMT]]

(In reply to Jim Mathies [:jimm] from comment #13)
> Hey Dees, any update on this from your email thread?

Non of them were fruitful, I'm afraid. Local language communication would likely get us further - perhaps Jack has a route in for us?

Comment 16

[Chris Peterson [:cpeterson]]

Very low volume and no reports from Nightly 58 yet.

Comment 17

[Hector Zhao [:hectorz]]

(In reply to Desigan Chinniah [:cyberdees] [:dees] [London - GMT] from comment #15)
> 
> Non of them were fruitful, I'm afraid. Local language communication would
> likely get us further - perhaps Jack has a route in for us?

Sorry but we didn't make any real progress, either.

(In reply to Hector Zhao [:hectorz] from comment #11)
> (In reply to Benjamin Smedberg [:bsmedberg] from comment #10)
> > Per our privacy policy, we cannot provide crash dump files to a 3rd party
> > without explicit permission from a user (or unless the crash was caught by
> > an employee). I don't think we have either of those currently.
> 
> We'll try to reach out to users who left a email for STRs, now that I have
> access to emails again.

We didn't hear back from any affected users on sharing their crash dump, or succeed in reproducing this crash.

> 
> > 
> > If baidu can provide us with symbols (PDB files) we can provide them with
> > detailed stack trace data.
> 
> I'll let Baidu know about this.

Nor did Baidu ever get back to us about uploading their debug symbols.

Comment 18

[BugBot [:suhaib / :marco/ :calixte]]

Closing because no crashes reported for 12 weeks.