Bug 691259 - Midori sends a SIGSEGV with the IcedTea NP Plugin
Summary: Midori sends a SIGSEGV with the IcedTea NP Plugin
Keywords:
Status: CLOSED ERRATA
Alias: None
Product: Fedora
Classification: Fedora
Component: icedtea-web
Version: 15
Hardware: i686
OS: Unspecified
unspecified
unspecified
Target Milestone: ---
Assignee: Deepak Bhole
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard: abrt_hash:(null)
: 709967 (view as bug list)
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2011-03-28 00:20 UTC by John Watzke
Modified: 2011-06-15 18:32 UTC (History)
13 users (show)

Fixed In Version: icedtea-web-1.0.3-1.fc15
Clone Of:
Environment:
Last Closed: 2011-06-15 18:32:21 UTC
Type: ---
Embargoed:


Attachments (Terms of Use)
File: event_log (5.21 KB, text/plain)
2011-03-28 00:20 UTC, John Watzke
no flags Details
File: smaps (187.79 KB, text/plain)
2011-03-28 00:20 UTC, John Watzke
no flags Details
File: dsos (28.71 KB, text/plain)
2011-03-28 00:20 UTC, John Watzke
no flags Details
File: maps (32.45 KB, text/plain)
2011-03-28 00:20 UTC, John Watzke
no flags Details
F14 Backtrace (50.85 KB, text/plain)
2011-03-29 03:13 UTC, John Watzke
no flags Details

Description John Watzke 2011-03-28 00:20:23 UTC
abrt version: 2.0.0
comment: I was retesting bug 632795 which originated in F13 but hasn't been resolved.  Essentially if you hit http://download.oracle.com/javase/tutorial/deployment/applet/getStarted.html with Midori, it crashes evertime.
package: midori-0.3.3-1.fc15
cmdline: midori
component: midori
executable: /usr/bin/midori
kernel: 2.6.38.1-6.fc15.i686.PAE
reason: Process /usr/bin/midori was killed by signal 11 (SIGSEGV)
architecture: i686
uid: 500
username: watzkej
os_release: Fedora release 15 (Lovelock)
time: 1301268911

Text file: event_log, 5332 bytes
Text file: smaps, 192293 bytes
Binary file: coredump, 194654208 bytes
Text file: dsos, 29398 bytes
Text file: maps, 33225 bytes

Comment 1 John Watzke 2011-03-28 00:20:25 UTC
Created attachment 488058 [details]
File: event_log

Comment 2 John Watzke 2011-03-28 00:20:27 UTC
Created attachment 488059 [details]
File: smaps

Comment 3 John Watzke 2011-03-28 00:20:30 UTC
Created attachment 488060 [details]
File: dsos

Comment 4 John Watzke 2011-03-28 00:20:32 UTC
Created attachment 488061 [details]
File: maps

Comment 5 Kevin Fenzi 2011-03-28 01:11:47 UTC
I hope you aren't uploading the 194MB coredump. ;) 

can you provide:

rpm -q webkitgtk java-1.6.0-openjdk-plugin

also, are you using any other plugins? sun java?

Comment 6 Kevin Fenzi 2011-03-28 01:23:42 UTC
I can duplicate this here with openjdk. 
The page loads and works fine without it... 

So, it seems like a java-1.6.0-openjdk-plugin-1.6.0.0-53.1.9.6.fc15.x86_64 issue. ;) 

Trying to get a backtrace out of the retrace server and can see if we can move over to that component for them to look at.

Comment 7 John Watzke 2011-03-28 01:41:07 UTC
Okay, see if you can get a backtrace.  I was going to replicate it on F14 so I could get a proper backtrace.  Unfortunately, ABRT 2.0 is hosed up a little in F15 (which explains the above uploads) and the retrace server seems to be hosed up too.

Comment 8 Kevin Fenzi 2011-03-28 03:06:53 UTC
Yeah, retrace failed here too. ;( 

Unexpected HTTP response from server: 413
HTTP/1.1 413 Request Entity Too Large

:( 

If you can get a f14 trace that would be great.

Comment 9 Jiri Moskovcak 2011-03-28 11:05:08 UTC
ABRT should't allow user to create a bug report without backtrace, so this is a bug.

Unexpected HTTP response from server: 413
HTTP/1.1 413 Request Entity Too Large

- we will look into this and try tweak the limits if possible..

Comment 10 Michal Toman 2011-03-28 14:28:22 UTC
Retrace server's upload size limit has been increased from 50 to 100 MB. You can retry - I guess it should be enough.

Comment 11 John Watzke 2011-03-29 03:03:58 UTC
Michael, the coredump file was reported to be 194654208 bytes so the limit needs to be a lot larger than that.  If we can get a large enough upload limit then I can try to do a F15 trace through the retrace server.

Comment 12 John Watzke 2011-03-29 03:13:46 UTC
Created attachment 488320 [details]
F14 Backtrace

This is a F14 backtrace.  Hopefully it is the same condition that's affecting F15 but we won't know until we somehow get a F15 backtrace.

Comment 13 Michal Toman 2011-03-29 07:17:49 UTC
The coredump is compressed before uploading (together with other required files). The limit is now 100M for uploaded archive and 1G for unpacked archive content. If 100M is not enough for you, I can increase it, but I don't know if it's really useful. Most users probably don't want to upload more than 100M - I'd say even 50M is enough.

Comment 14 Kevin Fenzi 2011-03-29 15:46:25 UTC
ok. I'm not sure if this is a webkit issue, or a openjdk one. 

Moving to openjdk first to see if they can track anything down with it...

Comment 15 Joachim Frieben 2011-04-13 12:49:11 UTC
(In reply to comment #13)
When trying to upload debugging information to retrace01.fedoraproject.org, the operation gets closed before completion because the upload limit seems to have been exceeded. However,

% du -hs ccpp-2011-04-12-23:53:57-2027/
35M	ccpp-2011-04-12-23:53:57-2027/

shows that only 35M would have been uploaded. The message is

Certificate is signed by an untrusted issuer: 'E=mtoman,CN=retrace01.fedoraproject.org,OU=BaseOS,O=Red Hat,L=Brno,C=CZ'.
Unexpected HTTP response from server: 413
HTTP/1.1 413 Request Entity Too Large

Date: Wed, 13 Apr 2011 11:50:01 GMT

Server: Apache/2.2.15 (Red Hat)

Content-Length: 0

AppTime: D=215402110

AppServer: retrace01.fedoraproject.org

Connection: close

Content-Type: text/plain


I have not found a related bug report, but I can open one if needed.

Comment 16 Deepak Bhole 2011-04-15 21:51:27 UTC
(In reply to comment #15)
> (In reply to comment #13)
> When trying to upload debugging information to retrace01.fedoraproject.org, the
> operation gets closed before completion because the upload limit seems to have
> been exceeded. However,
> 

I think you have the wrong bug :)

Comment 17 Deepak Bhole 2011-04-15 23:09:08 UTC
(In reply to comment #14)
> ok. I'm not sure if this is a webkit issue, or a openjdk one. 
> 
> Moving to openjdk first to see if they can track anything down with it...

Thanks, this is indeed a Java plugin issue. The plugin allows the code in NP_Initialize to run only once. This works for other browsers, but Midori makes two NP_Initialize calls, and uses the function pointers set in the latter. Since NP_Initialize was returning immediately, all plugin functions pointed to 0x0, causing a SIGSEGV.

I have proposed a patch upstream. Once it is in there, it should make it into Fedora shortly.

http://mail.openjdk.java.net/pipermail/distro-pkg-dev/2011-April/013650.html

Comment 18 Deepak Bhole 2011-04-15 23:11:56 UTC
Er, and ignore my erroneous comment 16 please.

Comment 19 Deepak Bhole 2011-06-09 16:04:49 UTC
*** Bug 709967 has been marked as a duplicate of this bug. ***

Comment 20 Joachim Frieben 2011-06-09 18:51:58 UTC
(In reply to comment #19)
Is this a bug in abrt? If the bug is a duplicate, it should have simply added me to the CC list instead of creating a new bug.

Comment 21 Deepak Bhole 2011-06-09 19:05:41 UTC
Sorry, I am not familiar enough with abrt to know. I have however seen it file duplicate bugs on occasion.

Comment 22 Fedora Update System 2011-06-13 21:59:45 UTC
icedtea-web-1.0.3-1.fc15 has been submitted as an update for Fedora 15.
https://admin.fedoraproject.org/updates/icedtea-web-1.0.3-1.fc15

Comment 23 Deepak Bhole 2011-06-13 22:02:00 UTC
The build mentioned in comment #22 fixes this bug. Please give it a try and add karma as applicable. Thanks!

Comment 24 Joachim Frieben 2011-06-14 05:23:32 UTC
Update icedtea-web-1.0.3-1.fc15 fixes this issue. Accordingly, the component should be set to icedtea-web and not to java-1.6.0-openjdk.

Comment 25 Deepak Bhole 2011-06-14 13:12:23 UTC
Thanks for trying it! I have changed the component to icedtea-web. The bot will close this issue when the update is pushed to stable.

Comment 26 Fedora Update System 2011-06-15 05:31:24 UTC
Package icedtea-web-1.0.3-1.fc15:
* should fix your issue,
* was pushed to the Fedora 15 testing repository,
* should be available at your local mirror within two days.
Update it with:
# su -c 'yum update --enablerepo=updates-testing icedtea-web-1.0.3-1.fc15'
as soon as you are able to.
Please go to the following url:
https://admin.fedoraproject.org/updates/icedtea-web-1.0.3-1.fc15
then log in and leave karma (feedback).

Comment 27 Fedora Update System 2011-06-15 18:32:16 UTC
icedtea-web-1.0.3-1.fc15 has been pushed to the Fedora 15 stable repository.  If problems still persist, please make note of it in this bug report.


Note You need to log in before you can comment on or make changes to this bug.