Is there anyway to make SEH works when stask is broken

[Tal Rashas Guardianship]

SEH is exactly a good solution for exception handling, but it depends on the stack to call the user-defined exeception handler function, that means, when exception occurs, the OS catch the exception immediately, and it will call the user-defined callback to give the control to the application. If the trouble which caused the exception makes the stack broken, the calling to the user-defined callback may not success, in this situation, the application will not be notified that an exception occurred. Is there anyway to catch the exception in all the cases? whether stack is OK or NG, the mechanism should work. Thanks.

[Stuart Dootson]

Exception handling in general needs to walk the call stack to find handlers. Maybe you can find some way of using C's longjmp to do your exception handling?

[Stephen Hewitt]

I think that if the stack is "broken" you've got more to worry about than SEH not working.

Steve

[Rolf Kristensen]

I have very good succes with __try / __except for catching unhandled exceptions (Instead of using SetUnhandledExceptionFilter or set_se_translator with /EHa compiler option). Though one still need to set the following exception handlers: _set_purecall_handler( ); _set_invalid_parameter_handler( ); Where one raises an exception inside the exception handlers to get the proper exception information to create a useful minidump. But yes sometimes the memory overwrite be so bad that the exception information cannot give a useful call stack.

[Tal Rashas Guardianship]

Thanks

[Tal Rashas Guardianship]

Yes I think so. But I don't want to recover from the error in this case, I just want to record some useful information when an exception happened. Such as error thread ID

[Tal Rashas Guardianship]

I think the workflow is, firstly I catch an exception, and then jump to where to handle this exception. I wonder how to catch such an exception when the stack was broken.