Closed GoogleCodeExporter closed 9 years ago
Issue 126 has been merged into this issue.
Original comment by magreenb...@gmail.com
on 29 Sep 2010 at 1:04
Have you noticed stability improvements using the integrated flash plugin vs.
the standard NPAPI flash plugin?
Original comment by magreenb...@gmail.com
on 29 Sep 2010 at 2:20
Was testing to see if a crash that happens on the normal flash plugin occurs on
the integrated one and it still does. However this means we can use flash no
matter if flash is installed or not.
Original comment by lodle...@gmail.com
on 30 Sep 2010 at 9:41
It seems the cause of the crash is the way flash is embeded. Here is a test
page that will always cause a crash
http://www.bobbyvandersluis.com/swfobject/testsuite_2_2/test_dynamic.html
(using http://code.google.com/p/swfobject/wiki/documentation)
Original comment by lodle...@gmail.com
on 8 Oct 2010 at 4:16
[deleted comment]
This is the same bug thats been bothering me for 2 months (issue 115),
chromiumembedded crashes on swfobject pages. It also crashes in the Chromium
test_shell though.
Original comment by ad...@enesce.com
on 14 Oct 2010 at 10:45
Looking into a way to move flash into a separate process so a crash doesnt
bring down the app
Original comment by lodle...@gmail.com
on 16 Oct 2010 at 1:32
Just did some hard core debugging and seems that if a flash is served with a
keep-alive header it causes flash to crash.
ive made a patch that removes the headers and for what i can see flash doesnt
crash no more.
Original comment by lodle...@gmail.com
on 16 Oct 2010 at 5:34
Attachments:
Patch doesn't seem to be affecting the request header, crashes and packet dump
is showing "Connection: keep-alive" still set (using
http://www.bobbyvandersluis.com/swfobject/testsuite_2_2/test_dynamic.html as a
test page in CEF)
Original comment by ad...@enesce.com
on 16 Oct 2010 at 11:19
Hmm yeah it doesnt work for me now. Some thing i change did allow it though.
Strange :P
Original comment by lodle...@gmail.com
on 24 Oct 2010 at 8:49
[deleted comment]
[deleted comment]
Worked out the flash crash. Its CEF sending the flash plugin data after it has
been unloaded.
And the reason it does that is because the unloading of the plugin happens as a
task and if a new instance of the plugin is loaded before the task executes it
doesnt cancel the pending unload. Thus it unloads the plugin but chrome still
thinks its loaded.
Original comment by lodle...@gmail.com
on 6 Nov 2010 at 7:20
Nice work, hope you are right. I'd like to see a comment from magreenblatt
about this.
I've been using a very hackish temporary fix for this bug... appending "<object
type="application/x-shockwave-flash" style="visibility: hidden; "></object>"
into the bottom of every page <body> tag. Stops the crash, but sometimes
doesn't work if you view the page a second time.
Original comment by ad...@enesce.com
on 13 Nov 2010 at 9:57
i have a better fix. Wrapped the flash plugin to stop it being unloaded at the
wrong time.
http://pastebin.com/9frTLeZf
Compile it as a npapi dll and use the patch above to point cef at it
Original comment by lodle...@gmail.com
on 13 Nov 2010 at 1:30
In response to Comment #13 by lodle.05:
Good analysis, this is exactly what's happening. I've filed a bug with Chromium:
http://code.google.com/p/chromium/issues/detail?id=63213
Original comment by magreenb...@gmail.com
on 15 Nov 2010 at 4:40
The crash is fixed in Chromium revision 66269.
Original comment by magreenb...@gmail.com
on 16 Nov 2010 at 2:29
Woot. :)
Original comment by lodle...@gmail.com
on 16 Nov 2010 at 3:40
The crash is fixed in CEF revision 138.
Original comment by magreenb...@gmail.com
on 16 Nov 2010 at 4:02
What about loading chrome flash plugin?
Original comment by lodle...@gmail.com
on 17 Nov 2010 at 8:29
A more general implementation for calling PluginList::AddExtraPluginPath would
be better -- no reason it needs to be labeled as specific for Flash. I'll
likely expose this as part of the upcoming pre-initialization configuration
options (along with user agent, WebPreferences, etc).
Original comment by magreenb...@gmail.com
on 18 Nov 2010 at 2:43
Original comment by magreenb...@gmail.com
on 18 Nov 2010 at 3:17
Original issue reported on code.google.com by
lodle...@gmail.com
on 28 Sep 2010 at 7:42Attachments: