I'm using Visual Studio 2008 to compile and debug a plugin for an application that runs under Windows.
The plugin has some problem and it crashes. I have been running VS in Debug mode but when the application crashes VS stack trace has nothing from my code. It is possibly the application itself that crashes. So I though I could try to figure out which part of my code does something the host application doesn't like by sprinkling a few print statements here and there in my code. That generally works as expected on *nix systems but in Windows I can't find where the output of those print goes. Can anybody point me in the right direction?
Also, if using prints do debug is too 1970's and somebody knows a better way to figure out where too look when the stack trace doesn't reveal anything interesting please let me know.
If i know correctly, debug.print is only executed when a debugger is attached, otherwise it is lost. (this applies to c# and vb.net).
You can use a logger to log your problems to some database or text file. A like NLOG to do the simple "log to a txt file" job. Additionally, you do not need to remove your logging code when shipping your addon -simply supress logmessages with low debug level.
Also, you could log to the windows event log, but is cluttered ny boatloads of junk, so you might have a hard time to differentiate that output.
Eventlog: http://support.microsoft.com/kb/301279
Related
I have a VB6 application that I provide support for. This application works on both Windows XP and Windows 7. Some users were migrated from Windows XP to Windows 7 using the User State Migration tool. These users now receive a generic "Application has crashed" Windows error message when they open certain screens (forms) in the application. My assumption is that there is a missing dll/ocx reference, but I'm having trouble tracking it down.
I've tried many/varied troubleshooting techniques:
Full uninstall and reinstall of my application
Manually re-registering all dll's and ocx's that I know are used
Running Process Monitor on a broken computer and a working computer to compare what dll's and ocx's are accessed. The answer might be here but even after filtering out most of the background noise the amount of data is overwhelming. At a minimum I reviewed all of the calls right before it crashes and all of the calls that were not successful. All of the non-successful calls match between working and non-working.
Installed the Windows Debugger Tools and captured a crash dump. Analyzed the crash dump with DebugDiag. DebugDiag says the exception is in msvbvm60.dll. I tried building a PDB file for my exe and loading it in DebugDiag to get more detail about where the exception is occuring but DebugDiag doesn't want to accept the PDB (might be doing something wrong here, but it just seems to ignore it. This same PDB file works fine when I do remote debugging, however.)
I recompiled my VB6 program without any optimizations in PCode. I've read online that sometimes building in PCode, while bad for performance, will tell you the real exception.
Used the above created PDB file to remote debug the VB6 application. The debugger says that the application crashes after the new window has been created, on a line that sets MousePointer = vbHourGlass... To me it seems unlikely that this is the real cause of the error. There are at least 20 other locations in the program where this same line is called and all work fine.
(Forgot about this one)
Used Dependency Walker and profiled the application on both a working and non-working computer. All errors found by dependency walker were the same between the two computers. There were no additional dependencies found on a working computer, and all missing dependencies on the non-working computer were also missing on the working one.
None of these actions changed my error message or showed me what the error is (unless it really is the mouse cursor issue)... There are no entries in the Windows Event Log related to the app crash.
The non-working and working computers all have the same base Windows 7 image, the only difference is whatever is being changed by USMT, which further convinces me that this is some kind of quirky configuration change or a missing dll/ocx or perhaps an unregistered dll/ocx.
Any ideas or thoughts on how I can track down the root cause of the issue would be greatly appreciated.
Update 1 - Response to questions
#MarkHall I have tried running it as admin, though not with UAC off. The application runs fine on a Windows 7 box as a non-admin with full UAC. Windows XP was 32-bit, Windows 7 is 64-bit, but again it works just fine on a like for like box where the user was not migrated from Windows XP.
#Beaner It's possible that it stores settings somewhere that have been corrupted, but the remote debugging leads me to think that it's more likely something else since it seems to die on a step related to the UI, which then makes me think it's probably a missing dll/ocx reference.
#Bob77 The application is installed into Program Files (x86). While many of the libraries do reside in the same folder, they are all registered.
Peter, often I've noticed that the debugger will indicate a line of code that is actually incorrect, depending on WHERE in the actual assembly language the fault occurs. You should look REAL close around your statement that sets the cursor to vbHourGlass. Your exception is PROBABLY happening BEFORE that line of code, but that line is what the debugger thinks is the actual faulted line of code.
Since you said it happens when a window OPENS, I'd look real close at any ocx's you may have referenced on the form, but perhaps NOT actually being used, or called. You might have one there that you don't intend to be there, that could be causing security issues, or something on Win7? Edit the .frm file by hand if you have to, and look at all the GUIDs the form references.
It is possible that one machine is using PER-USER registration, and the other is using PER-SYSTEM registration?? I don't know...
I would take a much closer look at the form that you are trying to open, and be VERY cautious of everything you are doing in the form load events, and so on. This sounds like it could be something as stupid as Windows Aero being enabled on one system, and not another, or some other sort of "Theme" setting that is throwing the VB Form Rendering routine into a hissyfit... Perhaps even something as stupid as a transparent color index in the icon you selected for that from?
If you are still developing this app, (or at least maintaining it), create an entirely NEW form, and re-create all the controls, etc, on the form (resist the temptation to copy/paste them from the old one...), and then see if THAT does the trick. Then, copy all the event code to the new form one event at a time, with at LEAST enough event code to make the form function, even if it's just a "dead form", that loads no data, or whatever the form is supposed to do. Check and debug after each change, and you WILL find it eventually. Of course, make sure you isolate one of the defunct systems to have a platform that you can duplicate the issue on, or then it's just guessing. I find that using something like Acronis w/ Universal Restore is a great option to then take the image file into a good HV, like VirtualBox, and then restore that image as a VM, so you can debug without interfering with your actual users. This sounds like a lot of work, but then again, so is re-writing an application that already exists, right? :)
Failing THAT... /* and */ are your friends!! (Well, we're dealing with VB, so ' would be your best friend! heh... But I'd start commenting out all the code on the form until that sucker opens. Then once it opens, start putting one line back at a time, and re-running it... That's called "VooDoo Debugging", but sometimes, you gotta do what you gotta do...
THANKS A LOT PETER! :) Now you got ME so involved in this, I feel like I'M the one debugging this sucker! Like if it was MY code I was trying to fix! :)
Let me know if any of this helps... I am actually quite interested in what you discover.
I'm trying to debug an application on an embedded device running an old version of Linux/Qtopia. I asked for help on QT forums but the people there don't know about old software and embedded systems. I'd really like some help with debug strategies.
My program will crash after the main window has been constructed, i.e. some time into the event loop. But depending on the order of functions in the constructor, sometimes it will run only from the console and sometimes it will only run from the icon. Despite my best efforts I can't narrow down what is causing the problem.
There is no seg fault or signal but my program does not continue and the destructor does not get called. It seems to me that one of the first things that would happen in the event loop is a resize event and when this is called could vary if you ran from the console or icon. Also, the various widgets in my GUI would be initialised and drawn so that is also a potential source of error, if I haven't set up something properly.
My debugging options are limited as the area where the crash actually occurs is not under my control. I tried logging to a file and printing to stderr but this was no help. When I got to the state where it runs from the icon but not console, I tried running in gdb and strace but it ran OK - the classic problem of debug software initialising differently.
My next thought is to try to force a core dump and then analyse that. How do I force a core dump ? Is there a better strategy ?
Logging to a file or to a communication port (serial port, etc.) is probably the simplest way to see what is happening and maintaining the normal runtime (i.e. not in a debugger).
You say that logging to a file and printing to stderr was no help. Why not? Are you printing relevant debugging information to the file? Are you using the Linux/Qtopia sources and adding debug logging?
Assuming you have sources for all of the code you are running, it should be just a matter of adding debug logging in the right places to pinpoint where the problem is occurring.
I'm using VB Express 2008. I know how to tell if code was built in debug more or release, but now I would like know how to tell if it is running in the Visual Studio IDE or not (so that I can add some debug MsgBox() and not worry about them showing if I ever accidentally ship a debug version).
Is System.Diagnostics.Debugger.IsAttached what you're looking for?
If you're building for Test and Prod, consider using a preprocessor directive in your code.
#If DEBUG Then
MsgBox("Foo")
#End If
This falls down, of course, if you ship a debug-built binary to a non-dev environment. I understand this is attacking the problem from another angle from where you asked the question (the IDE).
Try checking the IsAttached property of System.Diagnostics.Debugger
If you want to make sure you never show debug messages to users, you can use Debug.Write() and Debug.WriteLine(). These commands will output the text supplied to the debug output window. Note that you can attack a debug output window to a program running production code on a customer's machine without installing a development environment!
Testing for an attached debugger does not indicate that the debugger is also inside an IDE. It is quite common in many environments to attach debuggers to production code running on a customer system to identify what is going wrong on a particular customer's installation and usage. Testing for a debugger and then presuming you are in an IDE will foul out this usage of debuggers in a production environment.
I have a customer who is getting a 100% reproduceable crash that I can't replicate in my program compiled in Visual Studio 2005. I sent them a debug build of my program and kept all the PDB and DLL files handy. They sent me the minidump file, but when I open it I get:
"Unhandled exception at 0x00000000 in MiniDump.dmp: 0xC0000005: Access violation reading location 0x00000000."
Then the call stack shows only "0x00000000()" and the disassembly shows me a dump of the memory at 0x0. I've set up the symbol server, loaded my PDB symbols, etc. But I can't see any way of knowing which of the many DLLs actually caused the jump to null. This is a large project with many dependencies, and some of them are binaries that I don't have the source or PDBs for, as I am using an API as a 3rd party.
So how on earth is this minidump useful? How do I see which DLL caused the crash? I've never really used minidumps for debugging before, but all the tutorials I have read seem to at least display a function name or something else that gives you a clue in the call stack. I just get the one line pointing to null.
I also tried using "Depends" to see if there was some DLL dependency that was unresolved; however on my three test machines with various Windows OS's, I seem to get three different sets of OS DLL dependencies (and yet can't replicate the crash); so this doesn't seem a particularly reliable method either to diagnose the problem.
What other methods are available to determine the cause of this problem? Is there some way to step back one instruction to see which DLL jumped to null?
Well it looks like the answer in this instance was "Use WinDbg instead of Visual Studio for debugging minidumps". I couldn't get any useful info out of VS, but WinDbg gave me a wealth of info on the chain of function calls that led to the crash.
In this instance it still didn't help solve my problem, as all of the functions were in the 3rd party library I am using, so it looks like the only definitive answer to my specific problem is to use log files to trace the state of my application that leads to the crash.
I guess if anyone else sees a similar problem with an unhelpful call stack when debugging a minidump, the best practice is to open it with WinDgb rather than Visual Studio. Seems odd that the best tool for the job is the free Microsoft product, not the commerical one.
The other lesson here is probably "any program that uses a third party library needs to write a log file".
The whole idea behind all 'simple' ways of post mortem debugging is the capture of a stack trace. If your application overwrites the stack there is no way for such analysis. Only very sophisticated methods, that record the whole program execution in dedicated hardware could help.
The way to go in such a case are log files. Spread some log statements very wide around the area where the fault occurs and transmit that version to the customer. After the crash you'll see the last log statement in your log file. Add more log statements between that point and the next log statement that has not been recorded in the log file, ship that version again. Repeat until you found the line causing the problem.
I wrote a two part article about this at ddj.com:
About Log Files Part 1
About Log Files Part 2
Just an observation, but the the stack is getting truncated or over-written, might this be a simple case of using an uninitialised field, or perhaps a buffer overrun ?
That might be fairly easy to locate.
Have you tried to set WinDbg on a customer's computer and use it as a default debugger for any application that causes a crash? You just need to add pdb files to the folder where your application resides. When a crush happens WinDbg starts and you can try to get call stack.
Possibly you already know this, but here are some points about minidump debugging:
1. You need to have exactly the same executables and PDB files, as on the client computer where minidump was created, and they should be placed exactly in the same directories. Just rebuilding the same version doesn't help.
2. Debugger must be connected to MS Symbols server.
3. When debugger starts, it prints process loading log in the Output window. Generally, all libraries should be successfully loaded with debug information. Libraries without debug information are loaded as well, but "no debug info" is printed. Learn this log - it can give you some information.
If executable stack contains frames from a library without debug information, it may be not shown. This happens, for example, if your code is running as third-party library callback.
Try to create minidump on your own computer, by adding some code which creates unhandled exception, and debug it immediately. Does this work? Compare loading log in successful and unsuccessful debugging sessions.
You may have called null function pointer. Current executing function information is needed to show call stack information. Force set instruction pointer to start of any simple function, then you'll see call stack information again.
void SimpleFunc()
{ // <- set next statement here
}
I have a VB app that I need to monitor while it is running. I added some variables to the Watch window, but while the app is running the watch window is greyed out. The only way that I have found to see the variable values is to use Debug -> Break All, but this stops the program.
I have used other IDEs and they allow active variables to be monitored. Is this possible in VS?
Sorry if this is a noob question.
UPDATE: To be clear, my app is communicating with a piece of lab equipment and and as data is sent or received or errors are detected counters are incremented. I would like to watch these counters but I don't want to build a screen to do this since they are for debugging. I just assumed that this is basic functionality in any IDE
SHOCKED: It seems that Visual Studio does not offer this (what I would consider) basic functionality. For those that seem to think that this is not possible with an interpreted language, consider this thought experiment. If you pressed Break All quickly followed by a Continue then you would refresh the watch window - correct? Why then can't Visual Studio do this as a single Refresh Watch command or better yet allow this function to automatically run at a period specified by the user. No debug writes, no log files, no stopping your program mid-stream and creating timeouts. I am just shocked that you cannot do this. Its a little like not having breakpoints.
Which IDE or development environment shows - in real time - the values of variables in the Watch window, without having to hit any breakpoints, while the application is running?
Visual Studio doesn't provide this. In order to get updated values in the Watch window, or edit items there, the app needs to be at a breakpoint or debugging.
After you've done "break" to give control of the program to the debugger, then you can "step" through the code using function keys like F10 and F11. During each 'step', the program evaluates one or more statements; after each step it stops (until the next step), and while (only while) it's stopped you can 'watch' its current state.
There are other ways too to break into the debugger (to use the Watch window while the program is stopped): other ways like to set 'breakpoints', and use the 'run to cursor' feature.
Of course, but stopping a program that is actively receiving or sending data to a some other process, driver, etc, stops this communication and causes timeouts and other problems.
That's true. To watch values change in real-time, I use a log file:
Add statements to my code, such that when I change the value of a variable I emit a new line to a log file (showing the changed value)
Run the program
Watch new lines being appended to the log file using a utility like tail -f.
I've never see a debugger with the functionality you mention. The closest thing to the functionality you mentioned (and which isn't exactly the functionality you mentioned) is How to: Set a Data Breakpoint (Native Only).
What you're attempting to do is not possible in Visual Studio. All of the variable inspection windows (watch, locals, autos, etc ...) rely on the debugee process being in a break state in order to function.
This is true of essentially any debugger I've worked with in the past. At least those which use a compiled language.
I'm curious as to what IDE's you're referring to? Did they deal with interpreted languages?
Make sure you are in "Debug" build and Microsoft Debugger is running as a service and not blocked/disabled.
This should help you: How to trace and debug in Visual C++ .NET and in Visual C++ 2005
my 88 year old memory remembers an old version of visual studio allowing a watch window to function while debugging.
OK, just me.