Program Suddenly Crashes - Event 8?

Good evening, everyone,

a program that's been running for the last couple of months suddenly crashes:

'Event 8' is printed to the console, the Silverfrost Exception Report contains

Access violation (c0000005) at address 7747fe50

Within file ntdll.dll In RtlUnicodeToMultiByteSize at address 50 In GetModuleFileNameA at address D6 Within file KERNELBASE.dll In _system_exception_handler(<ptr>structÄ_EXCEPTION_POINTERS) at address 12F Within file SALFLIBC64.DLL in __system_exception_handler(<ptr>structÄ_EXCEPTION_POINTERS) at address 12f Within file ntdll.dll In RtlUnwindEx at address 852 In KiUserExceptionDispatcher at address 2E In RtlUnicodeToMultiByteSize at address 50 In GetModuleFileNameA at address D6 Within file KERNELBASE.dll In _system_exception_handler(<ptr>structÄ_EXCEPTION_POINTERS) at address 12F Within file SALFLIBC64.DLL in __system_exception_handler(<ptr>structÄ_EXCEPTION_POINTERS) at address 12f

When running the debugger, nothing happens, except that the program simply stops at some point.

Could somebody please tell me, what to make of this?

Thank you!

If possible switch to 32bit and find the reason, then return to 64

Unfortunately, that's not possible.

Can you post some Fortran code that generates the failure. Make it a short as possible.

Hello, Paul,

unfortunately I wasn't able to narrow the problem down far enough to create a short example.

It occured again, in another program, with different symptoms:

When running the program, everything works fine.

When debugging the program step by step, everything works fine.

When stepping over larger parts, event 8 occurs, occasionally just printing a notification 'EVENT 8', occasionally causing the debugging process to get stuck.

Is there a line in your code that prints 'Event 8'?

Not at all.

Also, the program simply stops after the print.

But I got another symptom to share: When running the program (not in the debugger), the error suddenly is printed to the console again (as in my original post). The difference being that now there's a reference included in the error message, which points to a blank line in my code, just before the END line of a subroutine.

I checked and double checked: the line IS blank.

Further, I checked if all variables worked like they should. based on the values assigned to them at the moment of the error, everything looks fine in that regard.

I am afraid that nothing useful can be done if you can only provide a subset of the symptoms and circumstances.

If you cannot provide a complete cut-down example, consider zipping up the full source code, along with any data files and instructions as needed, and placing the zip file in a cloud host such as Dropbox. Provide a public-access link to that file.

It's been a while, but the Event returned again.

I still can't reproduce it at will, but today I noticed a pattern, as it occured in a situation where I was able to switch to 32-bit, as Dan suggested:

It is somehow connected to mismatched array sizes. Whereas SDBG provides the according notification at all times (=array subscripts out of bounds), SDBG64 doesn't always provide the correct notification.

In my case, the error displayed in the debugger was 'access violation at ...' while simultaneously 'Event 8' was printed to the console.