What tips can you share to help locate and fix access violations when writing applications in Delphi?
I believe access violations are usually caused by trying to access something in memory that has not yet been created such as an Object etc?
I find it hard to identify what triggers the access violations and then where to make the required changes to try and stop/fix them.
A example is a personal project I am working on now. I am storing in TTreeView Node.Data property some data for each node. Nodes can be multiple selected and exported (the export iterates through each selected node and saves specific data to a text file - the information saved to the text file is what is stored in the nodes.data). Files can also be imported into the Treeview (saving the contents of the text files into the node.data).
The issue in that example is if I import files into the Treeview and then export them, it works perfect. However if I add a node at runtime and export them I get:
"Access Violation at address 00405772 in module 'Project1.exe'. Read of address 00000388."
My thoughts on that must be the way I am assigning the data to created nodes, maybe differently to the way I assign it when they are imported, but it all looks ok to me. The access violation only shows up when exporting, and this never happens with imported files.
I am NOT looking for a fix to the above example, but mainly advice/tips how to find and fix such type of errors. I don't often get access violations, but when I do they are really hard to track down and fix.
So advice and tips would be very useful.
It means your code is accessing some part of the memory it isn't allowed to. That usually means you have a pointer or object reference pointing to the wrong memory. Maybe because it is not initialized or is already released.
Use a debugger, like Delphi. It will tell you on what line of code the AV occurred. From there figure out your problem by looking at the callstack and local variables etc. Sometimes it also helps if you compile with Debug DCUs.
If you don't have a debugger because it only happens on a client side, you might want to use MadExcept or JclDebug to log the exception with callstack and have it send to you. It gives you less details but might point you in the right direction.
There are some tools that might be able to find these kind of problems earlier by checking more aggressively. The FastMM memory manager has such options.
EDIT
"Access Violation at address 00405772
in module 'Project1.exe'. Read of
address 00000388."
So your problem results in a AV at addresss 00405772 in module 'Project1.exe'. The Delphi debugger will bring you to the right line of code (or use Find Error).
It is trying to read memory at address 00000388. That is pretty close to 00000000 (nil), so that would probably mean accessing some pointer/reference to an array or dynamic array that is nil. If it was an array of bytes, it would be item 388. Or it could be a field of a rather large object or record with lots of fields. The object or record pointer/reference would be nil.
I find that the really hard-to-find access violations don't always occur while I'm running in a debugger. Worse yet, they happen to customers and not to me. The accepted answer mentions this, but I really think it should be given more detail: MadExcept provides a stack traceback which gives me valuable context information and helps me see where the code fails, or has unhandled exceptions (it's not just for access violations). It even provides a way for customers to email you the bug reports right from inside your program. That leads to more access violations found and fixed, reported by your beta testers, or your users.
Secondly, I have noticed that compiler hints and warnings are in fact detecting for you, some of the common problems. Clean up hints and warnings and you might find many access violations and other subtle problems. Forgetting to declare your destructors properly, for example, can lead to a compiler warning, but to serious problems at runtime.
Thirdly, I've found tools like Pascal Analyzer from Peganza, and the audits-and-metrics feature in some editions of Delphi, can help you find areas of your code that have problems. As a single concrete example, Pascal Analyzer has found places where I forgot to do something important, that lead to a crash or access violation.
Fourth, you can hardly beat the technique of having another developer critique your code. You might feel a bit sheepish afterwards, but you're going to learn something, hopefully, and get better at doing what you're doing. Chances are, there is more way than one to use a tree view, and more way than one to do the work you're doing, and a better architecture, and a clean way of doing things is going to result in more reliable code that doesn't break each time you touch it. THere is not a finite list of rules to produce clean code, it is rather, a lifetime effort, and a matter of degrees. You'd be surprised how innocent looking code can be a hotbed of potential crashes, access violations, race conditions, freezes and deadlocks.
I would like to mention one more tool, that I use when other tools fail to detect AV. It's SafeMM (newer version). Once it pointed me to the small 5 line procedure. And I had to look more than 10 minutes at it, in order to see the AV that happened there. Probably that day my programming skills wasn't at their maximum, but you know, bad thing tend to happen exactly at such days.
Just want to mention other debugging or "code guard" techniques that were not mentioned in previous answers:
"Local" tools:
* Use FastMM in DebugMode - have it write zeros each time it dealocates memory. This will make your program PAINFULLY slow but you have a HUGE chance to find errors like trying to access a freed object.
* Use FreeAndNil(Obj) instead of Obj.Free. Some, people were complaining about it as creating problems but without actually providing a clear example where this might happen. Additionally, Emarcadero recently added the recommendation to use FreeAndNil in their manual (finally!).
* ALWAYS compile the application in Release and Debug mode. Make sure the Project Options are correctly set for debug mode. The DEFAULT settings for Debug mode are NOT correct/complete - at last not in Delphi XE7 and Tokyo. Maybe one day they will set the correct options for Debug mode. So, enable things like:
"Stack frames"
"Map file generation (detailed)"
"Range checking",
"Symbol reference info"
"Debug information"
"Overflow checking"
"Assertions"
"Debug DCUs"
Deactivate the "Compiler optimizations"!
3rd Party Tools:
Use MadShi or EurekaLog (I would recommend MadShi over EurekaLog)
Use Microsoft's ApplicationVerfier
Related
What tips can you share to help locate and fix access violations when writing applications in Delphi?
I believe access violations are usually caused by trying to access something in memory that has not yet been created such as an Object etc?
I find it hard to identify what triggers the access violations and then where to make the required changes to try and stop/fix them.
A example is a personal project I am working on now. I am storing in TTreeView Node.Data property some data for each node. Nodes can be multiple selected and exported (the export iterates through each selected node and saves specific data to a text file - the information saved to the text file is what is stored in the nodes.data). Files can also be imported into the Treeview (saving the contents of the text files into the node.data).
The issue in that example is if I import files into the Treeview and then export them, it works perfect. However if I add a node at runtime and export them I get:
"Access Violation at address 00405772 in module 'Project1.exe'. Read of address 00000388."
My thoughts on that must be the way I am assigning the data to created nodes, maybe differently to the way I assign it when they are imported, but it all looks ok to me. The access violation only shows up when exporting, and this never happens with imported files.
I am NOT looking for a fix to the above example, but mainly advice/tips how to find and fix such type of errors. I don't often get access violations, but when I do they are really hard to track down and fix.
So advice and tips would be very useful.
It means your code is accessing some part of the memory it isn't allowed to. That usually means you have a pointer or object reference pointing to the wrong memory. Maybe because it is not initialized or is already released.
Use a debugger, like Delphi. It will tell you on what line of code the AV occurred. From there figure out your problem by looking at the callstack and local variables etc. Sometimes it also helps if you compile with Debug DCUs.
If you don't have a debugger because it only happens on a client side, you might want to use MadExcept or JclDebug to log the exception with callstack and have it send to you. It gives you less details but might point you in the right direction.
There are some tools that might be able to find these kind of problems earlier by checking more aggressively. The FastMM memory manager has such options.
EDIT
"Access Violation at address 00405772
in module 'Project1.exe'. Read of
address 00000388."
So your problem results in a AV at addresss 00405772 in module 'Project1.exe'. The Delphi debugger will bring you to the right line of code (or use Find Error).
It is trying to read memory at address 00000388. That is pretty close to 00000000 (nil), so that would probably mean accessing some pointer/reference to an array or dynamic array that is nil. If it was an array of bytes, it would be item 388. Or it could be a field of a rather large object or record with lots of fields. The object or record pointer/reference would be nil.
I find that the really hard-to-find access violations don't always occur while I'm running in a debugger. Worse yet, they happen to customers and not to me. The accepted answer mentions this, but I really think it should be given more detail: MadExcept provides a stack traceback which gives me valuable context information and helps me see where the code fails, or has unhandled exceptions (it's not just for access violations). It even provides a way for customers to email you the bug reports right from inside your program. That leads to more access violations found and fixed, reported by your beta testers, or your users.
Secondly, I have noticed that compiler hints and warnings are in fact detecting for you, some of the common problems. Clean up hints and warnings and you might find many access violations and other subtle problems. Forgetting to declare your destructors properly, for example, can lead to a compiler warning, but to serious problems at runtime.
Thirdly, I've found tools like Pascal Analyzer from Peganza, and the audits-and-metrics feature in some editions of Delphi, can help you find areas of your code that have problems. As a single concrete example, Pascal Analyzer has found places where I forgot to do something important, that lead to a crash or access violation.
Fourth, you can hardly beat the technique of having another developer critique your code. You might feel a bit sheepish afterwards, but you're going to learn something, hopefully, and get better at doing what you're doing. Chances are, there is more way than one to use a tree view, and more way than one to do the work you're doing, and a better architecture, and a clean way of doing things is going to result in more reliable code that doesn't break each time you touch it. THere is not a finite list of rules to produce clean code, it is rather, a lifetime effort, and a matter of degrees. You'd be surprised how innocent looking code can be a hotbed of potential crashes, access violations, race conditions, freezes and deadlocks.
I would like to mention one more tool, that I use when other tools fail to detect AV. It's SafeMM (newer version). Once it pointed me to the small 5 line procedure. And I had to look more than 10 minutes at it, in order to see the AV that happened there. Probably that day my programming skills wasn't at their maximum, but you know, bad thing tend to happen exactly at such days.
Just want to mention other debugging or "code guard" techniques that were not mentioned in previous answers:
"Local" tools:
* Use FastMM in DebugMode - have it write zeros each time it dealocates memory. This will make your program PAINFULLY slow but you have a HUGE chance to find errors like trying to access a freed object.
* Use FreeAndNil(Obj) instead of Obj.Free. Some, people were complaining about it as creating problems but without actually providing a clear example where this might happen. Additionally, Emarcadero recently added the recommendation to use FreeAndNil in their manual (finally!).
* ALWAYS compile the application in Release and Debug mode. Make sure the Project Options are correctly set for debug mode. The DEFAULT settings for Debug mode are NOT correct/complete - at last not in Delphi XE7 and Tokyo. Maybe one day they will set the correct options for Debug mode. So, enable things like:
"Stack frames"
"Map file generation (detailed)"
"Range checking",
"Symbol reference info"
"Debug information"
"Overflow checking"
"Assertions"
"Debug DCUs"
Deactivate the "Compiler optimizations"!
3rd Party Tools:
Use MadShi or EurekaLog (I would recommend MadShi over EurekaLog)
Use Microsoft's ApplicationVerfier
I am getting this AV message about 3 to 5 seconds after the applications close as expected:
Exception EAccessViolation in module rtl160.bpl at 00073225. Access violation at address 500A3225 in module 'rtl160.bpl'. Read of address 00000004.
These (20) applications are very similar in that they are IBX business applications. About half of them did not cause the AV to occur.
These applications were ported from Delphi-xe and they worked flawlessly for a long time. No changes were made to the projects in the port. Both 32 and 64 bit builds gave the same results.
Is this a bug in some library's finalization section freeing a resource or something?
I am using Delphi-XE2 Update 3.
Would appreciate the help.
Try using madExcept / EurekaLog etc. - they give you detailed stack trace on AV. This is not always a panacea, but can point you to the problem.
Access Violations are by their nature already very troublesome beasts since they deal with invalid pointers in memory. One that occurs a while after an application shuts down is even worse because that's when your app is in "cleanup" mode. You're could be dealing with something that went wrong much earlier in the application, but is only exposing itself at shutdown.
General Tips:
Try to always undo things in the reverse order you did them. E.g.
Create A, Create B ... Destroy B, Destroy A
Connect to Database, Open Dataset ... Close Dataset, Disconnect from Database
Even making sure you've done all the above before shutting down can help tremendously.
Any threads that are still running while your application is running can cause problems.
Preferably ensure all your child threads are properly terminated before final shutdown.
Refer back to Closing datasets above. Depending on what you're doing, some database components will create their own threads.
If you're using COM, try ensure ComObj is high up in the initialization sequence (I.e. place it as high as possible in your DPR).
Delphi finalizes units in the reverse order that they were initialized.
And you don't want ComObj to finalize before other things that are dependent on ComObj have also done so.
If you're using interface references, make sure you resolve circular reference issues.
Some of these problems can be tricky to find, but you can do the following:
Setup a source-code "sandbox" environment (you're going to chuck all your changes as soon as you've found the problem).
Figure out the simplest set of steps required to guarantee the error. (Start app and immediately shutdown would be ideal.)
Then you're going to comment-out delete wipe out chunks of code between tests and basically follow a divide and conquer approach to:
rip out code
test
if the problem persists, repeat. Else roll-back and rip out a different chunk of code.
eventually your code base will be small enough to pinpoint likely problems which can be tackled with targeted testing.
I've had this kind of access violation problem on occasion with old Delphi or C++Builder projects. Today I had it with C++Builder. At the time of the crash, by looking in the Debug -> Call Stack window, I can see that it's happening inside a call to fflush, called by __exit_streams and _exit.
I'm not sure what is causing it, since it's so deep in the Borland library code, but it seems to come and go at random when the code changes. And it seems to be more common with multi-form applications.
This time the error went away when I just added a new button on the main form. A button which is just there, has no event handlers and does not do anything. I think that any random change to the code, classes, variables etc rearranges the memory layout when you relink the application, and that either triggers or untriggers the error.
For now, I just leave the new button on the form, set it to "not visible" so that there's no visible change. As it seems to work, it's good enough solution for me at this time.
An application in the field is getting this message intermittently:
I am not able to reproduce this on my machine. I have also traced what I believe is the relevant code and can't find any access to uninitialized objects.
I've never had to deal with this kind of problem.
I did a build with madExcept and unfortunately the program does not crash once it is bundled.
Any opinions on madExcept vs EurekaLog for finding this kind of thing? I've never used FastMM. Would it be useful in his situation? (Delphi 2010) Any suggested flags to set in FastMM? Any other recommendations?
Note the very low address you are attempting to read. This sort of error almost certainly means you attempted to dereference a nil pointer even if you can't find one.
Given your description of the behavior I would suspect you've got a memory stomp going on--something is blasting a zero on top of the pointer to an object. When you change things you move things around and the stomp moves to someplace harmless.
Turn on both range checking and overflow checking.
Note the offending object must be at least 3C0 bytes in size--this should help narrow it down, most objects will be smaller than this.
What I have done in the past with such errors that only show in the field is put logging checkpoints in--a bunch of lines that display something in an out of the way place--a simple sequence of numbers is fine. Find out what number is showing when it crashes and you know which of you checkpoints was the last to execute. If that doesn't narrow it down enough you can repeat the process now that you've narrowed it down.
With a full map file you can identify the exact point in the code where this occurs. I hope you have a full map file for this image! Subtract $00401000 from the address at which the exception is raised ($007ADE8B in your case) and that corresponds to the values in the map file.
Having done that you know which object is nil and from there it is usually not too hard to work out what is going on.
One of the most common ways for this to occur is when a constructor raises an exception. When this occurs the destructor runs. If you access, in a destructor, a field that has not been initialised, and do anything other than call Free on it, then you will get an exception like this.
Looks like a memory overwrite where changing memory layout (your machine vs field machine or adding madExcept) makes the overwrite change something harmless.
FastMM is great at of making this kind of problems happen more consistently (and finding their source). Download the full version of FastMM, add it as the first unit of your project, and turn on FullDebugMode on its settings.
It might cause the problem to be reproduceable in your machine right away. If not, don't forget to deploy FastMM_FullDebugMode.dll with your application for testing. Keep madExcept on and let it embed the .map file for call stacks.
Closed. This question does not meet Stack Overflow guidelines. It is not currently accepting answers.
Questions concerning problems with code you've written must describe the specific problem — and include valid code to reproduce it — in the question itself. See SSCCE.org for guidance.
Closed 9 years ago.
Improve this question
I have a piece of code that is trying to write to disk many files in one second. However, it fails wince I have installed Kaspersky Anrivirus 2011.
Stream:= TFileStream.Create(sName, fmCreate);
The code totally worked with Kaspersky 2010 and also works with Kaspersky 2011 if I disable its scanners (it cannot be totally unloaded from memory - unless it is uninstalled). The code also works if (Kaspersky 2011 is running and) I write to disk slooooowly. So it obviously is not fast enough to handle my disk requests.
The error I get is EFCreateError ('Cannot create xxx file blablabla'). Error is random. Most of the files are written to disk. About 10% fail.
I have tried to get support but is impossible to find a real person at Kaspersky to speak with. Their so called 'support' is actually a FAQ data base. Of course it speaks about how to install the product and related stuff. There is nothing about programing-related issues. Any ideas?
PS: this has repercussions for the entire Delphi community! All our customers will fail to use Delphi software if they are using KIS 2011 as antivirus. For the moment I recommend to my users to disable their antivirus but I need a real solution.
It will be nice if a person with KIS 2011 can confirm the problem. Just create a tiny program that write 200 small files to disk using TFileStream.
UPDATE:
The problem appears ONLY when the file does not exist and it is created (created as opposed to overwritten).
Similar report: https://forums.embarcadero.com/thread.jspa?threadID=32751&tstart=15
Similar report: http://forum.kaspersky.com/index.php?showtopic=120561
A possible solution that popped in my mind is to detect if KIS is running and if it is, to put a delay after each writing to disk. Or at leat, let the user know there may be problems. Anybody knows how to detect if a service is running?
I added a delay of 650ms (after each file creation) and the bug is still there). So is not about how fast you write to disk but about how many files you write.
Just uninstalled KIS 2011. The problem does not appear anymore.
Just reinstalled the good old KIS 2010. The bug is still there but it appear rarely (about every 300 files instead of about 30 as in KIS 2011).
The problem was confirmed on a second computer.
NEWS: The crash appears in TFileStream.Create however it may be caused by a function called earlier: TestWriteAccess. If I disable this function, the TFileStream.Create doesn't fail anymore. Well, this doesn't change things too much. No matter which line of code generates the error, the program still fails (randomly) to write files to disk while Kaspersky is running.
Still waiting a response from a real person from Kaspersky...
More automated responses received from Kaspersky support (I sent emails to support in several countries). All pointing to a FAQ database.
I change my status from Kaspersky fan (and customer) to Kaspersky hater because I finally receive an answer from a real person from Kaspersky support and it was plain and simple obnoxious.
To test the code, try to use the code in a loop, to create 1000 files. The program creates a bunch of files (random number) then it fails at StreamFile:= TFileStream.Create.
Update: The issue can be fixed by entering a small delay after creating each file.
https://docs.google.com/forms/d/1H3_O1z1iEqfh9ZT9u3B0R1tGEj-Hc9o7rAE0LKPr33Y
2013 Update
Starting with this afternoon (after an update) KIS conflicts with Delphi.
Every time I compile a project KIS spikes to 100% CPU utilization. I will have to uninstall it.
2017 Update
All false positive alarms disappeared magically for all my Delphi programs starting with 2017. It seems that it was enough for a program like Kaspersky remove Delphi-generated executables from its virus list; all other smaller antivirus programs followed.
Delphi 7, Win 7 (32), KIS 2011
You need to instruct your users, i.e. Kaspersky's customers, that Kaspersky is interfering with the operation of your software, and that THEY should report it. Express your frustration that you, as a developer, don't have access to a real human being. This is the only way that the anti-malware companies will ever react - bad PR with their paying customers.
Kaspersky = pirate company? Maybe yes, maybe no. Maybe just yet another company with a bad product and nonexistent support. Their "support" consists in a FAQ database and an automatic email answering program. Phones are hooked to answering machines also. Their automated answer keep explaining me how to add my program in KIS "exception" database. I keep replying to those stupid emails that I cannot personally go to all my customers at home and put my program in the "exception" database and that it will be better if they will fix the bug.
When I finally got a non-automatic answer (the only one), the support guy fella is as rude as possible.
Possible solutions for Delphi programmers:
* Don't check if the user has write permission to a file (in order not to trigger Kaspersky bug)
* Check if the user has write permission. If the bug appears inform the user that Kaspersky creates problem and it should be temporary disable (while the program is running). Use a TRY EXCEPT block to do this.
Advice (based on my past experience):
Don't always blame your code if you ever received strange bug reports from your users when your program was trying to write to disk. Check also external factors (like existence of Kaspersky antivirus).
UPDATE:
I just applied for a refund. I will go for a chargeback if they won't refund the money (I strongly feel they won't).
Conclusion
When I posted this on StackOverflow I didn't realized the magnitude of the problem and I didn't realized it will deviate so much from initial course. Still I think it is well within the purpose of StackOverflow. We have all learned that sometimes the problems in out programs may not be caused by our faulty code and neither we can control the source of these problems (21 persons voted this question up - which means a lot of other people encountered issues with KIS).
We can just hope that poor designed programs that interacts with user's system at a very low level (such as KIS antivirus program) will be soon fixed so our sales won't suffer (much).
It is just frustrating when your program is labeled "buggy" and you can't do much about it!
Not an answer to solve your problem, but you should inform Kaspersky, probably they don't know there is a virus signature associated with a Delphi library.
And if your program isn't too complex, you might want to try Lazarus/FPC. It's not as good as Delphi, but I've been using it for several years now, and have got good results in Windows/MacOS/Linux.
i had similar problems with kaspersky 2011 when i was trying to add my prog to windows startup using d2010's new TFile.Copy() as well as raw api function:
CopyFile(PChar('C:\chellenger.exe'), PChar('C:\Documents and Settings\Omair\Start Menu\Programs\Startup\chellenger.exe'), False);
my solution was to put my delphi app in vb.net app as a resource, the vb.net app extracted it and put it to startup without false positives . Mixing two languages for your problem might solve your problem too(1 possible solution but a very ugly and nonprofessional solution i admit)
When you create file, any antivirus checks it. There is probably some kind of collision between your application and KAV. Have you tried to combine fmCreate with share modes. You can see in help for TFileStream.Create for available modes.
If the problem is just with kapersky, then just have your program detect if it is running. If so, scale back your file creation / writes to whatever passes their detection. Make sure you have some little status message somewhere that tells the user why things are slow. Incidentally, virus writers already know this which is why those heuristics simply don't work.
After doing that, contact Kapersky and work with them directly to get this resolved.
This gets past your immediate issue and will give you and kapersky time to figure out a long term solution.
Alternatively, you could simply shut kapersky down.. Just make sure you grab all of their watch dogs in the process.. But that tends to be a little more combative.
Creating a huge amount of files sounds like something that isn't necessarily A Good Thing, but you probably have your reasons :)
When you get the error code in Delphi, does KAV pop up any heuristic warnings, or is it completely silent? It wouldn't be weird to get a heuristic "omg, that app is doing something bad!" from creating a ton of new file, but if KAV is silent I'd say it's a bug.
Can you post a delphi executable with the tiniest amount of code that reproduces the bug? And a version that does the same step but only creates one file, it might be interesting to trace with SysInternals' ProcMon.
First, do you really need to test for write permissions by creating a file? Can't you just check the permission directly? I feel that creating a file for that purpose only is a lame way of doing it in any case.
Second, like noted above, it's likely that after you create and then delete a file, there is some intervention by Kaspersky's security mechanisms. Probably a driver tries to check the contents of the file you deleted, and keeps it alive for a while. Like this:
You create the file and open it, incrementing the refcount.
Kaspersky driver notices that and opens the file too. Even if you set share mode deny, as a driver it probably has the power to open it anyway (if Kaspersky could not circumvent sharing denials, any virus could have used the same trick to hide its data!).
You close the file and delete it. When you delete the file, the system just marks it "FILE_FLAG_DELETE_ON_CLOSE", but the file is still there until all the handles to it are closed.
Kaspersky continues to scan file, still haven't released the handle.
Therefore the file is still there.
You try to create a new file and the call fails because the old file is still not deleted.
The reason for all this mess is, of course, partly Kaspersky's checking mechanics, but they did nothing especially wrong here. Kaspersky needs to scan the file anyway, hardly anything can be done about that - it's antivirus, for crying out loud. On the other hand, checking permissions by creating and then deleting a file is (probably) very, very wrong. So I guess, you're the one at fault here.
I had the same problem. KIS made all kind of troubles. Until I reinstalled it. So, it was just a faulty installation.
I have some really complicated legacy code I've been working on that crashes when collecting big chunks of data. I've been unable to find the exact reason for the crashes and am trying different ways to solve it or at least recover nicely. The last thing I did was enclose the crashing code in a
try
...
except
cleanup();
end;
just to make it behave. But the cleanup never gets done. Under what circumstances does an exception not get caught? This might be due to some memory overflow or something since the app is collecting quite a bit of data.
Oh and the exception I got before adding the try was "Access violation" (what else?) and the CPU window points to very low addresses. Any ideas or pointers would be much appreciated!
"Very low address" probably means that somebody tried to call a virtual method on an object that was not really there (i.e. was 'nil'). For example:
TStringList(nil).Clear;
The first part is very mysterious, though. I have no idea how that can happen.
I think you should try to catch that exception with madExcept. It has never failed me yet. (Disclaimer: I am not using D7.)
A trashed stack or a stack overflow can both cause irreparable harm to the structures on the stack that structured exception handling (SEH) in Windows uses to find the actual exception handlers.
If you have a buffer overflow in a buffer on the stack (e.g. a static array as a local variable but written beyond its end), and overwrite an exception record, then you can overwrite the "next" pointer, which points at the next exception record on the stack. If the pointer gets clobbered, there's nothing the OS can do to find the next exception handler and eventually reach your catch-all one.
Stack overflows are different: they can prevent calling functions at all, since every function call requires at least one dword of stack space for the return address.
you have a number of good answers. the wildest problems i've had to chase come from stack corruption issues like barry mentioned. i've seen stuff happen with the project's "Memory sizes" section on the linker page. i might be superstitious but it seemed like larger wasn't necessarily better. you might consider using the enhanced memory manager FastMM4--it's free & very helpful.
http://sourceforge.net/projects/fastmm/
i've used it with d7 and found some access to stale pointers and other evil things.
you may also wish to create a way to track valid objects and or instrument the code in other ways to have the code checking itself as it works.
when i'm seeing access to addresses like 0x00001000 or less, i think of access to a nil pointer. myStringList:=nil; myStringList.Clear;
when i'm seeing access to other addresses with much larger numbers, i think of stale pointers.
when things are strangely unstable & stack traces are proving to be nonsense and/or wildly varying, i know i have stack issues. one time it's in Controls.pas; next time it's in mmsys.pas, etc.
using the wrong calling convention to a DLL can really mess up your stack as well. this is because of the parameter passing/releasing when calling/returning from the DLL.
MadExcept will be helpful in finding the source of this, even if it shows nonsense...you'll win either way because you'll know where the problem is occurring or you'll know you have a stack issue.
is there any testing framework you can put on it to exercise it? i've found that to be very powerful because it makes it entirely repeatable.
i've fixed some pretty ugly problems this way.
I'll leave the reasons why the except might not work to Barry...
But I strongly suggest a simple strategy to narrow down the area where it happens.
Cut the big chunk in smaller parts surrounded by
try
OutputDebugString('entering part abc');
... // part abc code here
except
OutputDebugString('horror in part abc');
raise;
end;
...
try
OutputDebugString('entering in part xyz');
... // part xyz code here
except
OutputDebugString('horror in part xyz');
raise;
end;
and run your code with DebugView on the side... (works for apps without GUI as well like services).
You'll see which part is executed and if the exceptions are caught there.
I used to get this strange behabiour when calling some COM object that used a safecall calling convention. This object/method may raise an EOleException, not trapped by the usual try/except on the client code.
You should trap an EOleException and the handle it properly.
try
...
except
on E: EOleException do
...
end;
I don't know if it is the problem you are facing. But if it is, i recommend you to take a look at Implement error handling correctly, a very clarifiyng post about exception handling in delphi.
You can also enable your IDE Debug Options to stop on delhi exceptions e monitor the stack trace.
Is this perhaps a DLL or a COM object? If so, it is possible that the FPUExcpetion mask is being set by the host application to something different than Delphi is used to. An overflow, by default in Delphi produces an exception, but the FPUExcpetionmask can be set so that it doesn't, and the value is set to NAN. See the math.pas unit for more information on FPUExceptionmask
I've gotten exceptions in the initialization and finalization blocks of my code which madExcept doesn't even seem to even catch. This might occur if you're referencing external DLL's inside of that try block. I'm not certain of the reason.
Actually (and thanks to #Gung for informing me of the worthlessness of my ancient answer), I read this recently in the ancient O'Reilly Delphi Tome. You should put SysUtils as the first (or second after your non-standard memory manager unit) in your main form's DPR so that it's resident in memory with all it's Exception Catching goodness. Otherwise, if it's loaded from some other unit, it will be unloaded with that unit too and you can kiss built in exception handling goodbye.