[webkit-qt] Memory Leak in QtWebKit

[Balazs Kelemen]

>
>
> You have mentioned that it is faster to free leaking memory by OS , 
> but in my view a program is
> responsible for doing cleanup work by itself and never depends OS to 
> do it.
> So I want to figure out the real reason why webkit is designed like 
> this? Can you provide some
> information about this?

Basically the policy in WebKit is _not_ to free global objects ever 
(caches for example). One reason for this is to make shutdown faster.
However, as long as you actually use WebKit (i.e. creating new views) 
you don't need to worry about that. These resources are not tied to a 
particular view, so a new view can reuse them. In order to purge the 
data that has been cached for the old page you can use 
QWebSettings::clearMemoryCaches(), that should actually free everything 
that cannot be reused in the new page. Another thing you can do is 
forcing a JS garbage collection - but in this case you need to build 
your own WebKit since afaik there is no public API for this. Hope that 
helps.

cheers!
-kbalazs

>
> Best Regards
>
> Eric Chen
>
>
> On 2012年02月29日 19:17, Alexis Menard wrote:
>> On Wed, Feb 29, 2012 at 12:53 AM, 陳敏華<chenmh at cn.fujitsu.com> wrote:
>>> Dear all
>>>
>>> I'd like to ask you question about memory leak in Webkit. Currently, 
>>> I am using
>>> QtWebkit self contained in Qt 4.8.0, and my test application is like 
>>> follows:
>>>
>>> int main(int argc, char *argv[])
>>> {
>>> QApplication a(argc, argv);
>>> QWebView* view = new QWebView();
>>> view->load(QUrl("http://www.baidu.com"));
>>> view->show();
>>> a.exec();
>>> delete view;
>>> return 0;
>>> }
>>>
>>> As you know, Webkit provide RefCountedLeakCounter class to detect 
>>> memory leak. When I run this test application in debug mode, I have 
>>> noticed that many WebCoreNodes and CachedResource have leaked. This 
>>> leak could be a big problem because we want to make use of WebKit in 
>>> embedded device.
>> But when the program is closed aren't the memory allocated to him
>> freed by the OS?
>>
>>> So, I'd like to ask you all whether this memory leak real or not? I 
>>> just can't believe Webkit would leak memory in such a simple test, 
>>> or if Webkit provides some other mechanism to ensure memory will not 
>>> be leaked, please tell me.
>> It happens that sometimes leaking on exit is not very bad as the OS
>> will free the memory for you anyway. In some scenarios it is faster to
>> do so...Now I haven't investigate myself but I know it sometimes
>> happens. In modern OS it's not a problem, maybe on yours it is fine
>> too. You have to check.
>>
>>> Can anyone provide some information about this problem? I really 
>>> appreciate your help.
>>>
>>> Best Regards
>>>
>>> Eric Chen
>>> ---------------------------------------------------
>>> Chen Minhua
>>> Development Dept.III
>>> Nanjing Fujitsu Nanda Software Tech. Co., Ltd.(FNST)
>>> No. 6 Wenzhu Road, Nanjing, 210012, China
>>> TEL: 86630566-615
>>> COINS: 79955-615
>>> FAX: 83317685
>>> MAIL: chenmh at cn.fujitsu.com
>>> ---------------------------------------------------
>>>
>>>
>>>
>>> _______________________________________________
>>> webkit-qt mailing list
>>> webkit-qt at lists.webkit.org
>>> http://lists.webkit.org/mailman/listinfo.cgi/webkit-qt
>>
>>
>
> _______________________________________________
> webkit-qt mailing list
> webkit-qt at lists.webkit.org
> http://lists.webkit.org/mailman/listinfo.cgi/webkit-qt