13584 - MLK: 256 bytes device context impls

Status: VERIFIED FIXED

(Core :: XUL, defect, P3)

Description

[Bruce Mitchener]

MLK: 256 bytes leaked at 0x2cd688
  * This memory was allocated from:
        malloc         [rtlib.o]
        __bUiLtIn_nEw  [libraptorgfx.so]
        __builtin_new  [rtlib.o]
        __bUiLtIn_vEc_nEw [libraptorgfx.so]
        __builtin_vec_new [rtlib.o]
        DeviceContextImpl::DeviceContextImpl() [nsDeviceContext.cpp:62]
        nsDeviceContextGTK::nsDeviceContextGTK() [nsDeviceContextGTK.cpp:45]
        nsGfxFactoryGTK::CreateInstance(nsISupports*,const nsID&,void**)
[nsGfxFactoryGTK.cpp:139]
        nsComponentManagerImpl::CreateInstance(const nsID&,nsISupports*,const
nsID&,void**) [nsComponentManager.cpp:1218]
        nsComponentManager::CreateInstance(const nsID&,nsISupports*,const
nsID&,void**) [nsRepository.cpp:77]
        nsWebShell::Init(void*,int,int,int,int,nsScrollPreference,int,int)
[nsWebShell.cpp:1151]
        nsBrowserWindow::Init(nsIAppShell*,nsIPref*,const nsRect&,unsigned
int,int) [nsBrowserWindow.cpp:1152]
        nsViewerApp::OpenWindow() [nsViewerApp.cpp:600]
        nsNativeViewerApp::Run() [nsGtkMain.cpp:41]
        main           [nsGtkMain.cpp:116]
        _start         [crt1.o]

Comment 1

[brendan-obsolete]

Pavlov, I'm reading this as a one-time leak, or nearly so.  Marking M14 for you
to get it out of chofmann's and my hair -- move it further out if you like, but
it can't be beta unless my analysis and the description of this bug are wrong or
incomplete.

/be

Comment 2

[Peter Trudelle]

Putting on beta1 radar.

Comment 3

[leger]

Putting on pdt- radar for beta1.

Comment 4

[Peter Trudelle]

Moving to m15. Pav, if you think this bug needs to be fixed for beta, as 
indicated by your previous tfv, you'll need to say why.

Comment 5

[Stuart Parmenter]

not beta1 at all

Comment 6

[Stuart Parmenter]

I believe this leak is because we were leaking a webshell.  What'd you think
bruce?

Comment 7

[Stuart Parmenter]

bruce, didn't you fix this?

Comment 8

[Bruce Mitchener]

nope... I see that this leaks as of last night in mozilla-bin... but now it is 
from a docshell.

Comment 9

[Bruce Mitchener]

bryner did this fix.  I tested it under Purify and it works for me.

I've reviewed it and would be happy to check it in or bryner can.

Index: nsFrameImageLoader.cpp
===================================================================
RCS file: /cvsroot/mozilla/layout/base/src/nsFrameImageLoader.cpp,v
retrieving revision 1.48
diff -u -r1.48 nsFrameImageLoader.cpp
--- nsFrameImageLoader.cpp 2000/03/15 04:24:21 1.48
+++ nsFrameImageLoader.cpp 2000/03/18 01:36:54
@@ -155,6 +155,8 @@
     aPresContext->GetTwipsToPixels(&t2p);
     aPresContext->GetDeviceContext(&theDC);
     theDC->GetCanonicalPixelScale(devScale);
+    NS_IF_RELEASE(theDC);
+    
     desiredWidth = NSToCoordRound((mDesiredSize.width * t2p)/devScale);
     desiredHeight = NSToCoordRound((mDesiredSize.height * t2p)/devScale);
   }

Comment 10

[Johnny Stenback (:jst)]

Bruce, the patch looks good to me, consider it reviewed (once more).

Comment 11

[Bruce Mitchener]

bryner checked this in.  You can now duly accept the bug and mark it resolved.

Comment 12

[Bruce Mitchener]

Resolving this for Pavlov to get it off my active memory leaks radar.

Comment 13

[Claudius Gayle]

marking verified.