294231 - with fastback on, clicking back *on your mouse* while viewing a PDF causes FF to hang

Status: RESOLVED FIXED

(Core :: DOM: Navigation, defect)

Description

[u49640]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8b2) Gecko/20050513 Firefox/1.0+
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8b2) Gecko/20050513 Firefox/1.0+

(note: Fast Back enabled, Adobe Reader 7.0.1 installed with the view PDFs in
Browser enabled)

goto any page that has a PDF link (like the one on the URL)
click on a PDF link
click on your mouse the go back button.

--> FF uses 99% CPU time every time i tried it, but it worked fine in a older
nightly)

It works if i click the Back Button in the Toolbar

Reproducible: Always

Steps to Reproduce:

Comment 1

[Steven]

I had it happen with the 12/05/2005 build too.

* Open a PDF...click backward with the "back" button *on the mouse*
Result: Fx sits at 85% together with acroread.exe 15%

* When using the back button *in the browser* (not mouse) it does not freeze

Used: Logitech MX 310 Mouse
O/S: W2K SP4
UA: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.8b2) Gecko/20050512
Firefox/1.0+
Acrobat 6 *Professional*

Comment 2

[Steven]

Addition to my post:

Fastback was also enabled (like original poster said)

Comment 3

[Steven]

Attachment: Task manager when clicked backward on the mouse

Comment 4

[u49640]

it also happens if i press the "Back" button on my Keyboard (note: my Keyboard
is a MS Internet Keyboard Pro and has an extra back button)

it doesnt appear if fast back is disabled (i'll try to switch it to
Core/Session:History)

Comment 5

[Steven]

Since you've mentioned the keyboard... I've tried it too
(never tought on it before) My keyboard also has some back / forward buttons.

Exactly the same result as you (Fx freezes)

Comment 6

[Dimitrios]

no problems for me with 20050513 on Win2K and Acrobat Reader 7, while reading
various pdfs from http://www.nsc.com :
1. I clicked back button while the pdf wasn't completely loaded. No hang.
2. I clicked back button after the pdf was completely loaded. No hang.

Is it possible for your issue to vanish after upgrading to Acrobat 7? Perhaps
the new plugin fixes it. Btw, the test url provided doesn't work with anonymous
login. You should provide another one where you see consistend hangs.

Comment 7

[u49640]

(In reply to comment #6)
> no problems for me with 20050513 on Win2K and Acrobat Reader 7, while reading
> various pdfs from http://www.nsc.com :
> 1. I clicked back button while the pdf wasn't completely loaded. No hang.
> 2. I clicked back button after the pdf was completely loaded. No hang.

you have to pressback on your mouse or your keyboard that have an extra "Back"
Button.

Pressing the Back icon on the toolbar works fine.
 
> Is it possible for your issue to vanish after upgrading to Acrobat 7? 

no, i have Adobe Reader 7.0.1

> Btw, the test url provided doesn't work with anonymous
> login. You should provide another one where you see consistend hangs.

it happens on *every* site that has a PDF linked, but i have changed it to an
other one.

Comment 8

[u49640]

okay, tested it a little bit more and it doesnt happen with a clean profile, so
i guess its an extension that i have installed.

(dont know if i should close this bug, or leave it open)

Comment 9

[Dimitrios]

I enabled the backspace operation on a button of my mouse. Interestingly, it
works as a "back button" in all but a pdf page (where it doesn't work at all).
Anyway, since you don't see it with a new profile (and if no one else cc'ed here
doesn't see it as well), you can resolve this bug as worksforme or invalid.

Comment 10

[u49640]

Attachment: the pref.js that causes the problem

okay, tested it even more and found out that something in this pref.js file
causes the bug

(i used a clean profile and only my old pref.js)

Comment 11

[Steven]

(In reply to comment #10)
> Created an attachment (id=183807) [edit]
> the pref.js that causes the problem
> 
> okay, tested it even more and found out that something in this pref.js file
> causes the bug
> 
> (i used a clean profile and only my old pref.js)

Removed prefs.js, and it worked for me too!
(using acrobat 6 professional)

I guess this can be closed?

Comment 12

[u49640]

i'm confused:

new profile,  this as prefs.js (the only entries)
user_pref("browser.search.selectedEngine", "Google");
user_pref("browser.sessionhistory.max_viewers", 7);
user_pref("browser.startup.homepage_override.mstone", "rv:1.8b2");
user_pref("intl.charsetmenu.browser.cache", "UTF-8");
user_pref("network.cookie.prefsMigrated", true);
user_pref("security.warn_entering_secure", false);

and i see the bug.

(In reply to comment #11)
> Removed prefs.js, and it worked for me too!
> (using acrobat 6 professional)

if you remove the prefs.js the fastback setting is disabled again

i dont think that i should close this bug. i just enabled the fast back and the
bug is there again.

Comment 13

[Steven]

(In reply to comment #12)
> i'm confused:
> 
> new profile,  this as prefs.js (the only entries)
> user_pref("browser.search.selectedEngine", "Google");
> user_pref("browser.sessionhistory.max_viewers", 7);
> user_pref("browser.startup.homepage_override.mstone", "rv:1.8b2");
> user_pref("intl.charsetmenu.browser.cache", "UTF-8");
> user_pref("network.cookie.prefsMigrated", true);
> user_pref("security.warn_entering_secure", false);
> 
> and i see the bug.
> 
> (In reply to comment #11)
> > Removed prefs.js, and it worked for me too!
> > (using acrobat 6 professional)
> 
> if you remove the prefs.js the fastback setting is disabled again
> 
> i dont think that i should close this bug. i just enabled the fast back and the
> bug is there again.

Yeah, that's true.
Sorry for jumping too fast to conclusions.

S.

Comment 14

[David Koppelman]

I apologize if this is a me-too, but I see this bug with a clean profile, the
only non-default setting is ..max_viewers = 2.  Using Acrobat reader 7.0.1. 
Firefox hangs when the back button is pressed, control is restored if an
AcroRead process is killed.

Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8b2) Gecko/20050517
Firefox/1.0+

Comment 15

[Peter van der Woude [:Peter6]]

Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.8b2) Gecko/20050518
Firefox/1.0+ ID:2005051808

Adobe Acrobat Plug-In Version 7.00

WFM

I simply give the pdf document time to load and unload and there's no crash at all
It's not a blazingly fast method I suppose.

Comment 16

[David Koppelman]

(In reply to comment #15)

> I simply give the pdf document time to load and unload and there's no crash at all
> It's not a blazingly fast method I suppose.

For me it's not a matter of impatience, the hang has lasted for over
eight minutes (at which time I kill the AcroRd32.exe process).

Some other data points: the problem does not occur if I use Alt-Left
Arrow. I'm using a MS USB IntelliMouse Optical with driver version
5.1.2600.0.

Comment 17

[David Koppelman]

Here's the build ID that I most recently tested on (see previous comment):
Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8b2) Gecko/20050518
Firefox/1.0+

Comment 18

[Peter van der Woude [:Peter6]]

others crash using the mouse backbutton on images too.
same problem ?

Comment 19

[Oliver Saier]

This is not really a crash, FF merely hangs at 100% CPU until you kill
acroRd32.exe, then it switches back to the previous page normally.

Comment 20

[David Koppelman]

(In reply to comment #18)
> others crash using the mouse backbutton on images too.
> same problem ?

With bfcache off FF crashes after pressing the mouse back button while viewing
an image.

With bfcache on (..max_viewers 2) the mouse back (and forward) button works
normally when moving from and to an image.  However, it crashes if the tab is
showing an image and I click on the X.

I wonder if all of the problems are connected?

Comment 21

[Caleb]

(In reply to comment #20)
> (In reply to comment #18)
> > others crash using the mouse backbutton on images too.
> > same problem ?
> 
> With bfcache off FF crashes after pressing the mouse back button while viewing
> an image.
> 

No, this is not connected to this bug, there is a seperate bug filed for it: bug
294630

Comment 22

[Brian Ryner (not reading)]

Attachment: patch

The problem happens in this case because we're inside of the plugin's event
loop (it has forwarded us the mouse button press).  Acrobat (and probably some
other plugins) don't like being torn down while handling an event.  I think we
can fix this by just tearing down the old content viewer asynchronously on a
PLEvent.  I no longer see the hang with this patch.

Comment 23

[Johnny Stenback (:jst)]

Comment on attachment 185091 [details] [diff] [review]
patch

r+sr=jst

Comment 24

[chris hofmann]

Comment on attachment 185091 [details] [diff] [review]
patch

a=chofmann

Comment 25

[Brian Ryner (not reading)]

checked in.

Comment 26

[u49640]

looks fine with Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8b2)
Gecko/20050602 Firefox/1.0+

marking verified

Comment 27

[u49640]

happens again with
Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8b2) Gecko/20050608
Firefox/1.0+
and with the 20050607 version

Comment 28

[Stebs]

Confirming the regression with:
Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8b2) Gecko/20050609
Firefox/1.0+ ID:2005060915

Comment 29

[Stebs]

Some Talkback IDs after killing the Acrobat Process because of hang (killing the
Firefox Process does not invoke Talkback): TB6593026Q, TB6592883E, TB6592881M
All with: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8b2)
Gecko/20050610 Firefox/1.0+ ID:2005061023

Comment 30

[Stebs]

Jay! This is WFM now with:
Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8b2) Gecko/20050616
Firefox/1.0+ ID:2005061621
One of the latest checkins for bfcache must have fixed it.
Well done!

Comment 31

[u49640]

yep, works in Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8b2)
Gecko/20050616 Firefox/1.0+