setScreen(x,y) to causes long script to crash Chrome (???)

[GoogleCodeExporter]

# Issue Description
Chrome crashes while running complex selenium-webdriver script, if a call to 
"setScreen(x,y)" was made.
This is not an easy or straightforward scenario, but it's reproducible.
I'll try to provide as much detail as possible.
And yes, I do realise it's a ridiculous bug from the sound of it, but bare with 
me.

# Configuration
OS: Linux Ubuntu 14.04.2 running in Docker 1.5.0
Chrome version: reproducible with 42 (both 42.0.2311.135-1 and 42.0.2311.90-1)
ChromeDriver version: 2.15
Selenium Webdriver version: 2.45.1, Node module
Setup is an headless, Xvfb + Openbox (like the one I saw used by Chromium team 
to run headless tests).

# Notes
This setup is the one used in New Relic Synthetics: this issue was discovered 
by a fairly complex Customer script (that I provide).

Also, when Chrome is run we inject a "traffic filter" extension that allows our 
product to alter Headers.
This is just loaded via the command line args and we interact with it through a 
WebSocket, transparently affecting outgoing traffic generated by the browser.

# Setup, script and logs
Given the amount of details, I thought I'd create a Gist on GitHub with all the 
details I could gather.

- setup_headless: 
https://gist.github.com/detro/9a0ac1119238aa4c4259#file-setup_headless-sh
- setup_chromedriver_instance: 
https://gist.github.com/detro/9a0ac1119238aa4c4259#file-setup_chromedriver_insta
nce-js
- offending_script: 
https://gist.github.com/detro/9a0ac1119238aa4c4259#file-offending_script-js
- offending_script_output: 
https://gist.github.com/detro/9a0ac1119238aa4c4259#file-offending_script-out
- chrome_driver_log: 
https://gist.github.com/detro/9a0ac1119238aa4c4259#file-chromedriver-log
- Xvfb_log: https://gist.github.com/detro/9a0ac1119238aa4c4259#file-xvfb-log
- openbox_log: 
https://gist.github.com/detro/9a0ac1119238aa4c4259#file-openbox-log

If more is needed, please ask.

Original issue reported on code.google.com by detroniz...@gmail.com on 6 May 2015 at 11:56

[GoogleCodeExporter]

I meant "bear" with me. But I make this mistake all the time...

Original comment by detroniz...@gmail.com on 6 May 2015 at 11:58

[GoogleCodeExporter]

I just updated the chromedriver.log with the output of a run in verbose mode.

Original comment by detroniz...@gmail.com on 6 May 2015 at 1:10

[GoogleCodeExporter]

As per request of our customer, I have removed the "offending_script" and the 
"offending script output": will provide it upon request of a Chromium team 
member that might look into this.

Thanks
Ivan

Original comment by detroniz...@gmail.com on 6 May 2015 at 1:58

[GoogleCodeExporter]

I ran strated headless Xvfb in my Linux machine Ubuntu 14.04
Then launched Remote Webserver in Xvfb simply with :-

Xvfb :1 -ac -screen 0 1027x768x8 & export DISPLAY=:1 && java -Xmx2666m -jar 
/usr/local/path/to/selenium-server-standalone-2.45.0.jar - port 4444 
-Dwebdriver.chrome.driver=/usr/local/path/to/chromedriver

then ran a simple script which resize chrome window almost 1000 times with 
these methods:-
 driver.manage().window().setSize(new Dimension(length,width));       
 System.out.println(driver.manage().window().getSize());

Result:- No Chrome crash

Is this because of    offending_script ??
OR is this due to "traffic filter" extension ??
is it reproducible without "offending_script" ??

how can we get them to recreate the issue ?

Original comment by agau...@chromium.org on 7 May 2015 at 12:15

• Added labels: Needs-Feedback

[GoogleCodeExporter]

I can provide the script in private to help with this.
I was expecting that calling "setSize()" multiple times won't crash anything: I 
did test this locally and all the normal scenarios didn't cause the crash.

I think I really need to provide the specific script.

I'd be happy to email it to you, but your email is of course obscured for now.
I'll share mine here so we can connect: detronizator[AT]... (this is my 
personal email, not the New Relic one)

My name is Ivan De Marino (just in case it can help to find me)

Original comment by detroniz...@gmail.com on 7 May 2015 at 1:17

[GoogleCodeExporter]

detronizator@

I have sent a mail in your mail ID please reply.

Thanks for your help.

Original comment by agau...@chromium.org on 8 May 2015 at 10:02

[GoogleCodeExporter]

Hello,

I have sent back details and scripts via email.
Any update?

Original comment by detroniz...@gmail.com on 5 Jun 2015 at 2:44

[GoogleCodeExporter]

Unable to reproduce using latest chromedriver.

Original comment by ssudunag...@chromium.org on 12 Jun 2015 at 7:52

[GoogleCodeExporter]

Were you able to reproduce with the script I sent in private?

I will test it again on my side to see if 2.16 fixes things.

Original comment by detroniz...@gmail.com on 12 Jun 2015 at 10:05