TL;DR:
application is built with WPF, running on .Net 3.5
newest and older versions of the app work properly on many different machines
on one specific machine it malfunctions in a strange way:
either does not start, displays black screen instead of "welcome" screen; can't do anything on that black screen
or does start and works properly, until a TextBox is presented and until the user starts entering the text into it; then application immediatelly hangs
Background:
I have built a WPF application on .Net 3.5. I've checked the application on various machines with different OS, ranging from Windows XP up to Windows 7 - and it runs fine, no bugs, etc. It also works fine on my client's several target machines, XP and W7 included again.
However, on one machine, it suddenly ceased to work. My client sent me the machine so I can investigate. From the files on its drives, I can see there were older versions, and they were used by the client for a few months. I also see that the newest version is not installed yet on this machine. So, I tried running the old versions first, to see what's the problem.
What user sees:
oldest versions (say, v1, v2) of the application does not work at all. When they start up, the only thing the application displays is ... absolutely completely black window. There should be a "welcome" screen, composed of few images and labels and a button. The app is not frozen it is "responsive" (Win7 can communicate with it, the window is not dimmed).
last (but not newest, let's name it v3) application runs fine, everything is clickable, features work OK, except one. When user is asked to provide a password, the user is presented with single TextBox in a popup. When the user enters at least one character in that textbox, the application freezes. It freezes even before having rendered the masked '*' of the password. The freezing is total, the application stops responding (the window is dimmed and the OS wants to close it), CPU goes to 100% (actually 50%, two-core system) and the only way to stop is kill it via task manager.
then I installed and tried running the newest version (v4) - it behaves just like v3 above
I've also tried reinstalling all the versions, just to be ensure the .EXE/.DLL files are not damaged - no changes (v1/2 - black, v3/4 - freeze on text)
Technical bits, just FYI:
The app is very simple, one executable with no shared libraries and no shared configuration. There is just no way the newest version could interact with old versions. Seriously, I intentionally made the app in a way it can be "just copied" and several instances can be safely kept in separate folders.
All versions of the app (v4, v3, v2, v1) work properly on all other machines, with various versions of Windows
The problem can't be related to user files or configuration, I tried copying them from problematic machine to healthy ones, and the problem does not repeat
Related to password box: There are absolutely no event handlers in the code that could produce a deadlock or freezing. The password textbox has no text-changed handlers. My code reads the text only after an "OK" button is pressed. The app freezes immediatelly when user enters first character, the
user does not even get a chance to press "OK"
Related to startup: The app does almost nothing on startup. It merely displays some form of a splash/welcome screen. There is no logic there except for a nice UI and one button. The user has no chance of pressing that button. When the app starts, the screen is black and has no contents, not even the splash/welcome image.
Tracing and observations:
I've focused on the TextBox part, since it's more granular.
I've got the app to freeze again, and checked the threads with ProcessHacker.
Except for typical threads, one thread that was sitting and busy-spinning at:
wpfgfx_v0300.dll MilGlyphRun SetGeometryAtRenderTime
wpfgfx_v0300.dll MilGlyphRun SetGeometryAtRenderTime
...
wpfgfx_v0300.dll MilContent DetachFromHWND
That's certainly problem with the WPF layer itself. The DetachFromHWND tells me close to nothing, but MilGlyphRun is surely about the text rendering. It freezes while trying to render the text. I've checked the black-screens on older versions of the applications, but I couldn't catch it, since the apps were not frozen, just black.
So..
I've applied literally all pending updates (like, 150 of them, ~800 Mb, most of them were to the OS and .Net framework)
I've tried reinstalling all video drivers
No changes, still black screen, still freezing.
So, focusing back to the problem. It failed to render a text. Sounds obvious, since it deadlocked when user entered a character into a textbox, but it's damn not obvious! The user was presented with tons of different TextBoxes until he finally got to the PasswordBox. All TextBoxes worked OK, only this one fails. And why old verions of the app display black screen and the new version displays the "welcome" screen properly? How can be those issues related?
The connection is: Windows Presentation Foundation Font Cache service.
Sources:
http://support.microsoft.com/kb/937135
http://social.msdn.microsoft.com/Forums/vstudio/en-US/7cc032c1-5f4d-4518-adc6-f53afd051e6b/presentationhostexe-running-at-50-cpu?forum=wpf
WPF uses a Font-Cache service, running as a separate process, for, well, caching fonts. Whenever your WPF app needs to draw some text, it might ask the cache service for fonts. Once in a while the font cache might get corrupted and it then will/may randomly impact any WPF application.
Unfortunatelly, restarting the service is not enough. When this happens, you must delete all FontCache files that this service keeps on the disk. Before deleting them, you must stop the "Windows Presentation Foundation Font Cache service (or whatever it's called in your languge, for example Usługa buforowania czcionek platformy Windows Presentation Foundation in Polish..). Then, go to
Win7: %systemdrive%\Windows\ServiceProfiles\LocalService\AppData\Local\
Vista: %systemdrive%\Windows\ServiceProfiles\LocalService\AppData\Local\
WinXP: %systemdrive%\Documents and Settings\LocalService\Local Settings\Application Data
and remove all files looking like FontCache.dat. The source articles mentioned only one file (FontCache3.0.0.0.dat, but in my case there were around 8 files with similar naming convention).
After deleing them and restarting the service, all versions of my application were running properly again. No black screens, no freezing on textbox.
Related
I am designing a Windows Form app. I have an MDIParent form that loads in a maximized state, and loads its child forms in a maximized state as well. However, when I open an OpenFileDialog, or any datareader object, the MDIParent shrinks to a smaller size with all its forms and controls.
This solution Opening child form is causing mdiform to change size and shrink does not apply/work in my situation.
Also this solution https://support.microsoft.com/en-nz/help/967173/restoring-a-maximized-or-minimized-mdi-parent-form-causes-its-height-t did not work for me.
Some background: I have seen this behaviour in almost all my WinForm applications but I have never been keen to sort it out. I was able to narrow down to the causes as highlighted above when I started investigate it. Some posts are describing it as a windows bug, but it has existed for as long as the screen resolutions started going above 1024 (VS 2010) for my case. I hoped it is not just a windows bug...
I hoped it is not just a windows bug...
Feature, not a bug, but it is not one that Winforms programmers like very much. Notable is that there have been several questions about mystifying window shrinkage in the past few months. I think it is associated with the release of Win10 Fall Creators edition. It has deep changes to the legacy Win32 api layer and they've caused plenty of upheaval.
In your specific case, the "feature" is enabled by a shell extension. They get injected into your process when you use OpenFileDialog. The one that does this is very, very evil and does something that a shell extension must absolutely never do. It calls SetProcessDPIAware(). Notable is that it might have been written in WPF, it has a very sneaky backdoor to declare itself dpiAware. Just loading the PresentationCore assembly is enough. But not otherwise limited to WPF code, any code can do this and that might have been undetected for a long time.
One way to chase down this evil extension is by using SysInternals' AutoRuns utility. It lets you selectively disable extensions. But there is also a programmer's way, you can debug this in VS.
Use Project > Properties > Debug tab > tick the "Enable native code debugging" checkbox. Named slightly different in old VS versions btw. Then Debug > New Breakpoint > Function Breakpoint. Function name = user32!SetProcessDPIAware, Language = C. You can exercise this in a do-nothing WPF app to ensure that everything is set correctly. For completeness you can also add the breakpoint for SetProcessDPIAwareness, the new flavor.
Press F5 to start debugging and trigger the OpenFileDialog.ShowDialog() call. The breakpoint should now hit, use Debug > Windows > Call Stack to look at the stack trace. You typically will not see anything very recognizable in your case since the evil code lives in a DLL that you don't have a PDB for. But the DLL name and location (visible in Debug > Windows > Modules) ought to be helpful to identify the person you need to file a bug with. Uninstall it if you can live without it.
Last but not least, it is getting pretty important to start creating Winforms apps that are dpiAware so such a bug can never byte. You kick this off by declaring your app to be dpiAware so DPI virtualization is disabled. Plus whatever you need to do in your code to ensure the UI design scales properly.
We are building a kiosk app and noticing some odd behavior with Google Chrome. To debug, we are monitoring all of the incoming events with monitorEvents(document); through the dev console and noticed 2 different kind of scenarios.
Eventually after a few minutes of heavy multi-finger/gesture use, we are stuck in a touchstart/touchend loop where touchend does not propagate down to a mouse/click event (breaking our app as we only handle onClick). It is continuously stuck in touchstart-touchend-touchstart-touchend for every finger click.
A continuation of scenario 1 with heavier use, we are getting absolutely no events in the console, not even a touchstart.
We are using Chrome 62 with Windows 10 with the following flags:
C:\Program Files (x86)\Google\Chrome\Application\chrome.exe --kiosk --incognito --disable-pinch --overscroll-history-navigation=0 URL
We have tried to add --touch-events, add --disable-gpu, remove our previous flags, but still facing the same issue. It seems to be an issue between Chrome and Windows as we are able to replicate it on google.com (with getting no events). When we plug in a mouse, we are able to click the buttons so the app is not frozen, we are just not getting any touch events (only mouse works at this point). Touch is still working on the Windows and Chrome App level, just not our app viewport.
The kiosk is a Dell All-in-One machine. Has anyone else experienced this kind of issue?
Thanks!
UPDATE: Using Chrome 68+ I no longer have this issue. I have not tried versions between 62 and 68 to see where it was fixed but I am glad that it is!
It is a little late now, but maybe this will bring some closure. I experienced the same problem and like you, only in Chrome. The best I could determine was that this glitch is dependent on a particular model of touchscreen in the computer.
Check the hardware IDs for the touchscreen in Device Manager. Only the Inspiron 20-3059 models with a touchscreen made by Advanced Silicon (VID_2149&PID_4C39) exhibit this behavior. Other 20-3059 models with a different touchscreen model (VID_0457&PID_1192) work fine.
Oddly enough, later Inspiron 20-3064 models also work fine, though they also had Advanced Silicon touchscreens. The PID was increased to 4C3E so perhaps they discovered this issue and fixed it for the newer model.
I wasn't able to find any combination of driver/firmware/Chrome versions that would fix the problem, but this is as close to a root cause as I could get.
First I'd like to head off the "wait a minute" questions.
Yes, I've read every thread with similar titles and subjects on SO and beyond.
Yes, I'm using the very latest version of SP1 and any potential hotfixes that might be related to this problem.
So, with specificity, here is my problem:
When I'm debugging Silverlight applications and it hits a breakpoint, everything works just fine as long as I use the mouse. I can continue, step, stop debugging, disable and enable breakpoints all just fine -- as long as I do so with the mouse.
But when I use the keyboard, Visual Studio freezes for a number of seconds, sometimes over a minute. This is any use of the keyboard. But after a short time, it starts to work just fine until that debugging session is over. But the next debugging session (regardless of whether I use the same instance of Casini or not) it will do exactly the same thing.
When I say any keyboard use, I mean it. F5 (Continue), Shift-F5 (Stop Debugging), F10 (Step Over), as well as typing in the Immediate window or even hitting the ALT key to highlight the menu.
For instance, if I try to type something in the Immediate window, it freezes for about 15 seconds and then shows the first key and repeats the pattern about 5 times until it releases and everything is just fine.
When I say "freezes" I'm not talking just Visual Studio but it hangs the entire operating system except for my ability to move the mouse. Even so, the mouse cursor remains what it was when it froze regardless of the program it's over, such as the text selector.
It only happens with Silverlight and it doesn't matter which version I use.
Here's what I'm running:
Windows 7 Professional 64-bit, all of the latest updates included, 12 GB RAM & 1.5 TB HD.
Visual Studio 2010 Ultimate with all of the latest updates.
All software is legit and I've already tried removing and/or disabling just about everything that can be removed or disabled, including extensions. I've used Process Explorer from Sysinternals to kill or stop everything I can. I've also tried it on a very simple Silverlight application with nothing more than one line of code that doesn't access anything.
EDIT: Forgot to mention that I also don't have it loading symbols beyond the "Just My Code" that it normally would and I've tried every debug setting that I can think of as well. And Intellitrace is also turned off (since I do have Ultimate), but that makes no difference either.
UPDATE (11/18): It gets more and more odd. I determined that the temporary freeze doesn't occur only when Visual Studio has focus. When VS (during the super-simple Silverlight app) goes into debug mode, any keyboard access AT ALL causes the same effect. SO, it appears that it has to (in some way) be related (at least in part) to my Windows setup, contrary to my assertion in the comments, but I cannot possibly fathom how. And if other programs are doing something when it freezes, the delay is longer (which makes a certain amount of sense with the symptoms). I'm still at a complete loss.
Anyone have any ideas or see this before? It's really frustrating.
I solved the problem.
It turned out to be the LastPass toolbar. It's a very handy thing and I've used it for years without issue. Yeah, yeah, toolbars are evil, I get it. :)
Disabling it in IE fixed the problem completely. I can still use it in FF (which is my main browser anyway except for working for which I use IE) and everything works fine.
Even so, it's bizarre to me that the toolbar would only have an effect when SL debugs. I didn't have an issue with ASP.Net debugging and I didn't any issues when the SL app was actually running, only when I hit a breakpoint AND hit a key within the first 15 seconds or so.
In any case, problem solved and thanks for your suggestions.
Try changing registry setting LowLevelHooksTimeout from 5000 to something low, like 300.
It can be at
HKEY_LOCAL_MACHINE\SOFTWARE\Wow6432Node\Policies\Microsoft\Windows\Control Panel\Desktop
or
HKEY_CURRENT_USER\Control Panel\Desktop
(first setting at Policy key seems to globally override user setting).
Edit: In addition to the bounty, we're willing to pay $250 to have this bug fixed in the Firefox/Gecko codebase. Here is a simple test project (Visual Studio 2008 C#) that reproduces the problem.
Edit #2 we're willing to pay $600 to have this bug fixed. See above for sample project that reproduces the problem.
We have a Firefox (Gecko) ActiveX control on our C# Windows Form to display HTML.
When this Firefox ActiveX control is on our form, about 2-3% of our key presses don't make it through. Or rather, a different Windows message is sent:
We hold down the TAB key to tab through 3 regular WinForms text boxes. It will behave correctly 97% of the time. Spy++ tells us WM_KEYDOWN message is sent properly:
normal behavior http://judahhimango.com/images/normaltab.jpg
But randomly, maybe 2-3% of the time, the tab key (or other key) isn't processed right. Spy++ tells us WM_CHAR is being sent instead:
odd behavior http://judahhimango.com/images/screwytab.png
When the odd behavior occurs, either the key is not processed at all, or is processed incorrectly (such as inserting a '\t' character into a textbox that doesn't support tab characters.
This only occurs if the Firefox ActiveX control is on our form.
Our question is: does Firefox/Gecko engine install some kind of keyboard hook that might cause these side effects? Or better yet, how do we fix this problem?
The WM_CHAR message is generated by TranslateMessage call, so a good place to start looking would be the TranslateMessage calls in the Gecko source code.
In the first example code you provided the function is imported only by two libraries - mozctl.dll and xul.dll. Since you claim that the same error happens also with GeckoFX we can take mozctl.dll out of the equation. That leaves us with xul.dll, so given the Gecko source code I would suggest to look into widget\src\windows\nsToolkit.cpp. I am not sure if the code is run if the engine is embedded, but if it is then the library starts a whole new message pump in different thread, which is bound to break.
Unfortunately I can't run or compile the code on my machine (Windows 7 x64 w/o the Mozilla ActiveX control installed), so I can't verify any of this with a debugger. Hope it helps someone to track it down further.
The root problem is that when Mozilla is embedded in another application, it incorrectly pumps Windows messages when it dispatches internal events. Mozilla uses an event system to coordinate across threads or to schedule deferred processing on a thread (see nsIThread, nsIEventTarget). If you embed a web page with a lot of active XMLHTTPRequests, for example, Mozilla will use its event dispatching interface to dispatch events back to javascript and it will pump windows messages as a side effect. Once Mozilla events are fully dispatched, it goes back to the main event loop.
When Mozilla pumps windows messages, it doesn't include the extra processing done by the application's event loop - IsDialogMessage(), TranslateMessage(), PreTranslateMessage(), or any other pre-processing are skipped when Mozilla gets into this state. Symptoms therefore include tab key presses getting inserted as a character instead of being used for dialog navigation, keyboard hotkeys being sporadically ignored, or custom message pre-processing being sporadically skipped. For example, the Outlook 2007/2010 "Compose" screen sporadically loses keystrokes because it relies on custom message pre-processing to handle keyboard input.
See https://bugzilla.mozilla.org/show_bug.cgi?id=582790 for a patch that addresses the problem.
I have Snoop Free and PSM Anti-Keylogger.
One of them detected firefox trying to install a Keyboard Hook.
Mozilla/Firefox file xul.dll attempt at installing at keyboard hook.
DENIED.
I noticed that you have implemented all of the interoperability yourself. Can you try this with the GeckoFX project and see if you get the same error? I use this project at work and haven't encountered any issues yet.
i have a WPF application that sits on top of all other windows. 99% of the time it works perfectly however if i dial into my computer from another location using remote desktop and then close the connection and go back to the original computer, my application is not visible. checking the running processes i can see that it is still running however its not visible at all. I assume it has something to do with the remoting but i was wondering if there is some code i can run say every 20 seconds that sits on its own thread that checks whether it is visible and if not will execute some code whether by pInvoke or .NET to become visible again. Any ideas?
Make sure to upgrade to .NET 3.5sp1. There were some changes with remoting which lead to WPF applications being sent as bitmaps in remote desktop in all situations.
In earlier versions of the framework, there were problems with WPF applications causing very odd behavior when combined with remote desktop.
Another idea - make sure you're using the latest and greatest versions of your graphics drivers from your graphics card vendor. WPF makes heavy use of DirectX, and this could potentially be a driver bug preventing the context from resetting appropriately.