Memory leak when using CefSharp.BindObjectAsync().

[HTD]

• What version of the product are you using?
  65.0.0-pre02 (NuGet)

• What architecture x86 or x64?
  Both.

• On what operating system?
  Win10.

• Are you using WinForms, WPF or OffScreen?
  OffScreen.

• What steps will reproduce the problem?
  A huge memory leak occurs on each page reload when CefSharp.BindObjectAsync() is used.

To reproduce run the test project: https://github.com/HTD/CefSharpTest

When run in probably less than a minute memory usage of CefSharp.BrowserSubprocess.exe will grow exceeding 100MB, soon exceeding 1GB and more. Switch IsFullTestEnabled field to false and rerun the test. The described memory leak will not occur.

When IsFullTestEnabled field is set to false the bound class is still cached during each reload, but it is unavailable (not bound) after reloading.

It is expected, that binding the same object on each page reload should not cause a memory leak.

[HTD]

I just tested it with CefSharpSettings.LegacyJavascriptBindingEnabled = true; option and legacy JSB, no memory leak, work as charm, so clearly new JSB is broken. Any clues? Is it possible the problem is in CefSharp, not CEF itself?

[amaitland]

To reproduce run the test project: https://github.com/HTD/CefSharpTest

Thanks for providing an example 👍

Any clues? Is it possible the problem is in CefSharp, not CEF itself?

No idea at this stage. The V8 Javascript Engine is constantly evolving, what worked in a previous version doesn't always work in a newer ones, #2390 is a recent example of that.

[HTD]

Looks like a duplicate, but in my case the leak is not present in legacy mode, only in the native one. BTW, it seems to be present since 63.x, I just haven't tested it well enough with that version.

[amaitland]

Running the WPF example and using BindingTest.html as a more complete example, calling Reload every 10 seconds and I'm seeing pretty much what I expect, memory usage rises, then the GC kicks in and reclaims it. There does appear to be a leak when you refresh the browser every second, which has me wondering if the GC doesn't have time to run.

Relevant commit 57d60e9

[amaitland]

Testing with b0af05a and the Javascript GC appears to be working again, the memory usage goes up and after a short period of time comes back down to a manageable level. I've not seen it rise past 150mb on my machine, this will likely vary depending on hardware/operating system.

Closing as unable to reproduce with newer versions.