Browser Crashes or Hangs When Trying to Print

[ascii_elite]

The Pale Moon browser crashes or hangs (typically depending on the printer driver) when trying to print a a webpage utilizing Flash Player.

An example to reproduce the issue, with a print driver already installed:
1. Install a version Flash Player if it is not already installed ( For Flash Player installers see http://www.adobe.com/go/tn_14266 )
2.Navigate to http://kb.adobe.com/selfservice/viewCon ... d=tn_14157
3.Take note of the Flash Player content on the page
4.Select menu item File --> Print Preview
5. Use the Print button or action to display the Print dialog box
6. Select a Printer and the "Print to file" option to perform a soft print (and save the paper)
7. Perform the Print dialog box "OK" action.
8. Observe a browser crash or hang.

I did my testing on Windows XP Professional with Service Pack 3 and Pale Moon 25.7.3.


EDIT (update given post post #6): Flash Player content must run not run in a separate process (such as with the plug-in container).

dom.ipc.plugins.enabled preference msut be set to false.

Code: Select all

// Disable using separate process for plug-ins (using plug-in container)
//   General handling (Firefox default is true)
user_pref("dom.ipc.plugins.enabled", false);

[Tomaso]

No such problem here.

(Pale Moon v25.7.3 (x64) / Adobe Flash Player v19.0.0.226 / HP LJ300-400 drivers v5.0.12200.706 / Windows 8.1 Pro)

[Moonchild]

So, is it a crash, or is it a hang? XD

[ascii_elite]

So, is it a crash, or is it a hang? XD

It seem to depend on the print driver.

When using printer drivers:
The PDF printer driver: PDF-XChange Printer 2012 5.0 build 272.1, a crash
hp officejet 5500 series 2.335.5.0, a permanent hang
HP LaserJet 1100, a crash

[Moonchild]

Tried verifying this in a VM with XP: no issues seen.

[ascii_elite]

Tried verifying this in a VM with XP: no issues seen.

I have been doing some testing and have found that the problem does not manifest if the Flash Player content is run in a separate process (such as with the plug-in container).

When run in a single process, the problem manifests.

Please retest after setting the dom.ipc.plugins.enabled preference to false.

Code: Select all

// Disable using separate process for plug-ins (using plug-in container)
//   General handling (Firefox default is true)
user_pref("dom.ipc.plugins.enabled", false);

[Moonchild]

The crash occurs in the flash player plugin in that case (I verified this with the latest v19 where this indeed leads to a crash), and because you disable OOPP, it takes the browser down with it. I can't do anything about this because the crash is caused by Adobe's software.

Solution: don't disable OOPP. It's there for a reason

[ascii_elite]

The crash occurs in the flash player plugin...I can't do anything about this because the crash is caused by Adobe's software.

It would be good to report the problem to Adobe to have it fixed; in the past (some years ago), I tried to do so regarding another bug, but could never get Adobe to accept the bug report.

Solution: don't disable OOPP. It's there for a reason

I also disable OOP for good reason; unfortunately, using the current Pale Moon implementation for OOPP leads to other major problems.

In the past, the FIrefox OOPP implementation has been rather poor; it often was the cause of stability and compatibility issues and so I had often preferred to disable the plugin-container use.

I should hope that the OOPP implementation would be much better by now.

I have not recently tested stability using the plugin-container, but I have recently tested the OOPP implementation compatibility with my main (multi-resolution, multi-monitor) workstation and have found it partially, but critically, incompatible. The OOPP implementation appears to conflict with, the multi-monitor management software, UltraMon. The result is that the Pale Moon windows may not be draw correctly: window part incorrectly appearing in two places at the same time, part of the window content as if missing or transparent, page with plugin content sizing issues, etc. Wherefore, at the moment, it is preferable to disable the Pale Moon plugin container, as the alternative is worse.

There, unfortunately, does not appear to be a good alternative to the UltraMon software. I have invested a considerable amount of time already trying to find one, but have not been successful.

Perhaps compatibility improvements may be made to the Pale Moon implementation of OOPP?

[Moonchild]

Perhaps compatibility improvements may be made to the Pale Moon implementation of OOPP?

Honestly? This isn't very likely. Especially not for compatibility issues that arise on Windows XP -- NT5 simply isn't the best when it comes to inter-process communication like a hardware-accelerated plugin passing back a surface. In the end it really isn't our software that crashes either, so as said, nothing can really be done on our end.

[ascii_elite]

Perhaps compatibility improvements may be made to the Pale Moon implementation of OOPP?

Honestly? This isn't very likely. Especially not for compatibility issues that arise on Windows XP

I would think the issue would also manifest on Windows Server 2003. There might be issues on NT6 series operating systems; I, however, have not done much testing on those platforms.

NT5 simply isn't the best when it comes to inter-process communication like a hardware-accelerated plugin passing back a surface.

I looked somewhat into this. It appears that a greater focus on inter-process communication for this like surfaces was a notable change for Windows Display Driver Model (WDDM) which was new (and optional) as of Windows Vista.

In the end it really isn't our software that crashes either, so as said, nothing can really be done on our end.

I am aware. Further, I recently tested Flash Player 18.0.0.252 in the Presto-based Opera 10.63 browser and found that a crash manifests while trying to print on a page that contains a Flash Player object. The unfortunate side-effect of the OOPP is that it masks (some would say mitigates) the bugs in the underlying plugin and such bugs never get fixed or are often not given attention.

I guess for now, I shall have to workaround the problem probably by finding a convenient way to determine if an offending (Flash Player) plugin is used on a page and be able to exclude it easily. I would think this would require an extension.