On the page above, there is a Flash movie that is supposed to slowly animate the text "Cats In Space". In Mozilla, however, the animation is flashing like crazy. It is also causing Mozilla to increase it memory usage by about 1 MB per second. If I leave Mozilla on this page, it will eventually use up all my memory and swap until it crashes or X crashes or something. I am guessing that the Flash plugin is being constantly destroyed and recreated. The reason I am thinking this is that if I right click on the Flash, I either get the Flash context menu (which disappears immediately) or I get Mozilla's context menu, and I am guessing I would only ever get Mozilla's context menu if the Flash plugin had been destroyed. This page works fine in NN4 with the same version of the Flash plugin. Tested with Mozilla trunk build 2001081106 on Linux.

WORKSFORME.

Certainly does not work for me. Works just as described including eventually consuming all available memory. Let me reiterate: The page works fine with NN4 but not with Mozilla using the same ShockWave plugins (I tried both versions 4 and 5). Most flash sites work ok in Mozilla, but not this one. I tried creating a new profile to see if that would help. It did not. Mozilla trunk build 2001100808 on Linux. Note: works fine on Win2000.

Marking NEW.

maybe it's not the same bug, but on two flash sites i visited, the animation was paused until i was moving the mouse cursor, if i stop the animation (+sound) stop, tested with 2001121821

I've just found that this happend if i change the quality (high-normal-low- of the animation, when it's done, the bug appeared, even, if you put it back to high(default).

--- Mass reassigning Unix bugs to serge ---

I'm working on it

Reviews wanted, but plussing based on ADT triage.

Comment on attachment 71043 [details] [diff] [review] patch v1 r=peterl

After applying this patch browser hangs. Also the patches for the following bugs have been applied: #85958 and #116924.

I've forgotten to say that browser hangs on http://www.moontv.fi

Hmm, it looks like removing gtk_widget_destroy(mXtBin) call before NPP_Destroy() causes the crash: +#if 0 #if defined(MOZ_WIDGET_GTK) if (mXtBin) gtk_widget_destroy(mXtBin); @@ -751,6 +752,7 @@ if (mXlibXtBin) mXlibXtBin->xtbin_destroy(); #endif +#endif investigating...

I removed gtk_widget_destroy() from ns4xPluginInstance::Stop() because we destroy that widget in 694 ns4xPluginInstance :: ~ns4xPluginInstance(void) 695 { 696 PLUGIN_LOG(PLUGIN_LOG_BASIC, ("ns4xPluginInstance dtor: this=%p\n",this)); 697 #if defined(MOZ_WIDGET_GTK) 698 if (mXtBin) 699 gtk_widget_destroy(mXtBin); 700 #elif defined(MOZ_WIDGET_XLIB) but w/o gtk_widget_destroy(mXtBin) call bedore NPP_Destroy() mozilla always exits on reload of http://www.moontv.fi page with Gdk-ERROR **: BadWindow #0 exit (status=1) at exit.c:40 #1 0x4045f533 in gdk_x_error () from /usr/lib/libgdk-1.2.so.0 #2 0x4050dd2e in _XError () from /usr/X11R6/lib/libX11.so.6 #3 0x4050b552 in _XReadEvents () from /usr/X11R6/lib/libX11.so.6 #4 0x404f7b7b in XIfEvent () from /usr/X11R6/lib/libX11.so.6 #5 0x42f07c0b in object.2 () from /u/serge/plugins/linux/libflashplayer5.so so I'm leaving gtk_widget_destroy ()where it was before in ns4xPluginInstance::Stop(void) and adding mXtBin = 0; after that, to disable gtk_widget_destroy() for the same mXtBin in ns4xPluginInstance :: ~ns4xPluginInstance()

please review this patch

The new patch works correctly.

Comment on attachment 72868 [details] [diff] [review] patch v1.1 r=peterl

ccing blizzard for possible sr/a= the patch does fix a memory leak which we allocate in ns4xPluginInstance::SetWindow() 1060 window->ws_info = (NPSetWindowCallbackStruct *)PR_MALLOC(sizeof(NPSeindowCallbackStruct)); but never free, Also it emproves overall ObjectFrame performance, because window->ws_info now has a lifetime equal lifetime of nsObjectFrame object, and we are not doing destroy/recreate gtkxtbin widget on any nsObjectFrame::Reflow() call.

adding ADT2 in status whiteboard as per discussion with beppe.

*** Bug 135540 has been marked as a duplicate of this bug. ***

Comment on attachment 72868 [details] [diff] [review] patch v1.1 sr=beard, one comment though, you changed an #if 1 to #if 0, why not just remove the code you are commenting out?

Thanks Patrick, >why not just remove the code you are commenting out I've decided to keep it as reference, until final clean up, when we'll split this file by platform or widget set.

seeking for check in approval.

adt1.0.0+ (on behalf of the adt)

fix on the trunk /cvsroot/mozilla/layout/html/base/src/nsObjectFrame.cpp,v <-- nsObjectFrame.cpp new revision: 1.336; previous revision: 1.335 /cvsroot/mozilla/modules/plugin/base/src/ns4xPluginInstance.cpp,v <-- ns4xPluginInstance.cpp new revision: 1.91; previous revision: 1.90

checked into MOZILLA_1_0_0_BRANCH /cvsroot/mozilla/layout/html/base/src/nsObjectFrame.cpp,v <-- nsObjectFrame.cpp new revision: 1.335.2.2; previous revision: 1.335.2.1 /cvsroot/mozilla/modules/plugin/base/src/ns4xPluginInstance.cpp,v <-- ns4xPluginInstance.cpp new revision: 1.90.2.2; previous revision: 1.90.2.1 --- resolved as FIXED

The site http://www.mycatsnotfat.com/space.html is now working fine. Mozilla truk 2002-04-13-07 on Linux.

verified that this is working ok on trunk 0416 and branch 0411 on linux. I see no slowdown or any performance degradation using this page.

adding verified1.0.0 keyword (branch verification) based on bonsai checkin and QA comments.