Fatal Error when calling GetSourceAsync()

[huazai76]

• What version of the product are you using?
  94.4.2+g6a963ca+chromium-94.0.4606.61

• What architecture x86 or x64?
  x64 - note: I also tried x86, and a similar error occured

• What version of .Net?
  .Net 4.6

• On what operating system?
  Win10

• Are you using WinForms, WPF or OffScreen?
  WinForms

• What steps will reproduce the problem?
  When executing following codes (for scrawling webpages) on a particular URL, an error occurs:

CefBrowser.Load("https://www.epri.com/research/products/3002012302");
Application.DoEvents();
Thread.Sleep(8000);

try
{
  Task<string> t = CefBrowser.GetSourceAsync();
  t.Wait();   
  var h = t.Result;
}
catch (Exception ex)
{
    MessageBox.Show(ex.Message);
}

• What is the expected output? What do you see instead?
  Error Messages are copied and translated into English (marked with **) below:

System.AccessViolationException
HResult=0x80004003
Message= ** try to read or write protected memory,....
Source=mscorlib
StackTrace:
at System.String.wcslen(Char* ptr)
at System.String.CtorCharPtr(Char* ptr)
at CefSharp.Internals.CefStringVisitorAdapter.Visit(CefStringVisitorAdapter* , CefStringBase* string) in C:\projects\cefsharp\packages\cef.sdk.93.1.14\CEF\include\internal\cef_string_wrappers.h:line 447

** this exception is initiated at the following calling stack :
string.wcslen(char*)
string.CtorCharPtr(char*)
CefSharp::Internals::CefStringVisitorAdapter::Visit(CefStringBase&) (in cef_string_wrappers.h)

[amaitland]

Converting to discussion as it's not clear this is a bug. You shouldn't rely on Thread.Sleep for loading of a website.

You can use LoadUrlAsync to wait for the website to load.
http://cefsharp.github.io/api/94.4.x/html/M_CefSharp_IWebBrowser_LoadUrlAsync.htm

GetSourceAsync is an asynchronous operation, Continue With or await should be used rather than blocking.

You can add a breakpoint at https://github.com/cefsharp/CefSharp/blob/cefsharp/94/CefSharp.Core.Runtime/Internals/CefStringVisitorAdapter.h#L33 debug the underlying exception.

[huazai76]

Still don't work for another URL. Same exception come out. After testing this case, it seem the content of page ( string returned by GetSource ) keep changing before end of loading and script execution, and it changes many times, in some cases, it first contain the keyword, then not ,and then in the end contain the keyword. So the way using if (Contain( "someKeywords")) to determine the page is completely loaded does not work. That means in some particular pages, I get the wrong source from intermidiate page content.

Thanks to amaitland and ukandrewc. Possibly you are right, the program will only work completely in the way not using the SLEEP.

[ukandrewc]

Just tried your URL and got the source in FrameLoadEnd, every time.

[amaitland]

I have try the minimal winform example, I put a button on it, and execute same code I have provided, and it worked.

The minimum example has a CachePath set by default. Please focus on reproducing the problem with the minimum example. I need a fork that reliably reproduces the problem before I can debug this and see if we can handle the exception gracefully.

[huazai76]

The minimum example has a CachePath set by default. Please focus on reproducing the problem with the minimum example. I need a fork that reliably reproduces the problem before I can debug this and see if we can handle the exception gracefully.

I works hours and cannot exactly locate the exceptiontiggering code. Up to now, by removing code sections each time and test, I think possibly three factors together can triggering the exception:
(1) cachdir is set;
(2) A FormResize methord defined, and some controls properties is set and got, such as width, clientwidth,
(3) CefBrowser.Dock = DockStyle.None, -- this shoud have a connection with factor 2.

After a long time editing, I have messed up my codes, maybe later I can provide the exact example.

GetSourceAsync is an asynchronous operation, Continue With or await should be used rather than blocking.

Besides, the GetSourceAsync() cannot not be used with await as the sample code provide you above, because, C# require the await followed by a method modified by " Async" ,,, however, this method is not defined like that as shown by the method name. The compilor raise an error for await GetSourceAsync(). Could you pls provide me some more details about how to use GetSourceAsync with await?

[amaitland]

Besides, the GetSourceAsync() cannot not be used with await as the sample code provide you above, because, C# require the await followed by a method modified by " Async" ,,

I'm not sure what reference you are quoting here. You can absolutely use GetSourceAsync() with await as it returns Task<string>. For an event handler you simply need to add the async keyword to the method declaration.

private async void BtnStart_Click(object sender, EventArgs e)
{
	await browser.LoadUrlAsync("https://www.epri.com/research/products/1013030");
	string html = "";
	do
	{
		html = await browser.GetSourceAsync();
	}
	while (!html.Contains("<h2 class=\"header\">Keywords</h2>"));

	MessageBox.Show(html);
}

[huazai76]

I tried but fail to use the WinformMinimum to trigger the exception. By removing codes of my app, I have got a very simple code that will trigger the exeption. But I cannot upload a attachment by dragging & dropping here. So I copy all the code below.
---code deleted. more concise code provided below.

[amaitland]

Please add your form to the minimal example and make any of the modifications nessicary.

[huazai76]

Sorry, I have no experience in version control or forking.. I have delete some more unnecessay codes and make a most concise version, and re-test to make sure that the three conditions altogether will trigger the exception. If removing anyone, no exception..
I have make comments in the code, showing the details of the controls on the form. I believe it is easy task for you as to re-construct the project.

---

using System;
using System.Threading.Tasks;
using System.Windows.Forms;
using System.IO;
using CefSharp;
using CefSharp.WinForms;

namespace EPRIProductsCrawler
{
    public partial class FrmMain : Form
    {
        private readonly ChromiumWebBrowser browser;  

        public FrmMain()
        {
            InitializeComponent();

            Cef.EnableHighDPISupport();
            CefSettings setting = new CefSettings()
            {
                CachePath = Path.Combine(Environment.GetFolderPath(Environment.SpecialFolder.LocalApplicationData), "CefSharp\\Cache"),
            };
            Cef.Initialize(setting, performDependencyCheck: false);

            browser = new ChromiumWebBrowser("about:version")
            {
                Dock = DockStyle.None,
            };
            this.Controls.Add(browser);
        }
        
        private void FrmMain_Resize(object sender, EventArgs e)
        {
            // The controls in and the arragement of FrmMain Form are: 
            //upper  is a label (LblToolbar) used as toolbar, and a textbox for show URL;  a Panel (PnlMain) docked left; 
            // and a CEFBrowser whose location is calculated by FrmMain_Resize method
            if (browser != null)
            {
                const int MIN_SIZE = 50;
                this.SuspendLayout();
                browser.Left =  PnlMain.Left + PnlMain.Width ;
                browser.Top = LblToolbar.Top + LblToolbar.Height;
                browser.Width = Math.Max(ClientSize.Width - browser.Left, MIN_SIZE);
                browser.Height = Math.Max(ClientSize.Height - browser.Top - StatusStrip.Height, MIN_SIZE);
                TxtURL.Width = Math.Max(ClientSize.Width - TxtURL.Left - 10, MIN_SIZE);
                this.ResumeLayout(false);
                this.PerformLayout();
            }
        }

        private void FrmMain_FormClosing(object sender, FormClosingEventArgs e)
        {
            browser.Dispose();
            System.Environment.Exit(0);  
        }
        
        private void BtnStart_Click(object sender, EventArgs e)
        {
            //This will triggering an  System.AccessViolationException  that is not handled and cannot be catched, and thus is fatal.
            // URLs triggering exception include at least：  1013030   3002012302    3002014760  1011829  1021228  
            // If remove brower's CachePath setting or Dock.none setting, or remove FrmMain_Resize(), no exception will occur

            browser.Load("https://www.epri.com/research/products/1013030");  
            string h ="";
            do
            {
                System.Threading.Thread.Sleep(1000);
                Application.DoEvents();
                try
                {
                    Task<string> t = browser.GetSourceAsync();
                    t.Wait();     
                    h = t.Result;
                }
                catch (Exception ex)
                {
                    MessageBox.Show(ex.Message);
                }
            } while (!h.Contains("<h2 class=\"header\">Keywords</h2>"));
            MessageBox.Show(h);   // for making sure no exception occur before
        }     
    }
}

[amaitland]

Sorry, I have no experience in version control or forking

https://docs.github.com/en/get-started/quickstart/fork-a-repo

The controls in and the arragement of FrmMain Form are:
//upper is a label (LblToolbar) used as toolbar, and a textbox for show URL; a Panel (PnlMain) docked left;
// and a CEFBrowser whose location is calculated by FrmMain_Resize

This is the reason I ask for a fork of the minimum example. Trying to reconstruct an example takes time and is open to interpretation. I'll test without this code as I'm not going to try implemting this portion.

If this portion is relevant then you'll need to provide the designer code as well. (No zip files as I won't download them from an unknown source).

[huazai76]

further info:
In all the 3000 identical URLs crawled, three webpage trigger the exception (and always trigger). They may have some common attibutes
URLs are : 1013030 3002012302 3002014760 (plus the prefix: https://www.epri.com/research/products/ )

[huazai76]

More info:
In the three factors, pls focus on Dock = DockStyle.None if you debug later. Because later I find other two URLs (1011829 1021228) that cause exception even when I remove the Form_Resize method and the CachePath setting .. And after Dock = DockStyle.None is also removed, no exception occur for these two URL.
But there is a weird thing, these two URL only still cause exception after removing CachePath in my actual code ( with a loop to crawl many pages), but do not cause exception in the test code after removing CachePath. This is just FYI, and I will not further investigate the reason for this weird issue, it is too complicated for me. If you need further info, I can try my best. I will just remove all these three code sections in my apps to aviod crash and hope no more other exceptions come.

[amaitland]

In the three factors, pls focus on Dock = DockStyle.None if you debug later. Because later I find other two URLs (1011829 1021228) that cause exception even when I remove the Form_Resize method and the CachePath setting .. And after Dock = DockStyle.None is also removed, no exception occur for these two URL.

I don't believe these are actually related to the exception you are seeing, it's possible they somehow impact the timing in someway. It's very unlikely they are directly related.

I'd suggest you test with the async code I provided above.

Where appropriate, we now utilize the default UTF string encoding format and
shared memory to reduce copies and conversions for the cross-process
transfer of arbitrary-length strings. For example, CefFrame::GetSource/GetText
now involves zero UTF conversions and zero copies in the browser process for
the CefString delivered to CefStringVisitor::Visit().

As per chromiumembedded/cef@ebee847

Looking at the CEF history this change in behaviour to CefStringVisitor::Visit() seems the most likely cause.

Based on the exception my best guess is that the shared memory section used for the string somehow becomes invalid. Maybe the render process crashes and the memory is unavailable? Without an example I have no way to test a possible solution, the best I can do is take a guess.

[amaitland]

Commit 7722881 changes the way string are handled in an attempt to get a copy of the string rather than using a ptr to workaround the upstream issue.

You can test out a nightly build for 94.4.20-CI4254, Instructions for installing the package via the Nuget Console are included on the page.

As I have no way of testing this all that I know is that it passes the existing unit tests.

[huazai76]

I got the new CEF package. And use the my minimum code to test all the konwn problematic URL. And it is OK now. No exception.!!
So the three factors should only be contributing and indirect factors, not the root cause. You might have eliminate the root cause.

Excepetion URLs tested： 1013030 3002012302 3002014760 1011829 1021228 1013030

[pkudrel]

I can confirm that the problem exists.
But my application is complicated and it is not easy to isolate the problem. I will try to reproduce the issue using WinformsMinimum example.

[amaitland]

Test with the build linked in #3838 (reply in thread)

[pkudrel]

Thank You - I'll do it.
But I want to give an example that generates the problem.

[amaitland]

Commit 7722881 changes the way string are handled in an attempt to get a copy of the string rather than using a ptr to workaround the upstream issue.

Changes were included in the 94.4.50 release.