Differences running a program from C:\Windows vs. C:\Windows\system32

Question

Calling a program in windows prompt by absolute path from C:\Windows directory is working, but calling this program from C:\Windows\System32 gives me errors.

• C:\Windows>D:\pathtoexe\program.exe --> OK
• C:\Windows\System32>D:\pathtoexe\myprogram.exe --> Terminates without doing it's task

The program reads a configuration file from the current directory. The configuration file exists in both (Windows and System32) directories. If I delete the file from the Windows directory then calling the program from the Windows directory terminates with the same error as if I call it from the System32 directory. There should be no file access problems:

• C:\Windows>type config.ini--> OK
• C:\Windows\system32>type config.ini --> OK

The same program with same configuration file on Windows Server 2003 is working for both directories!

How do I debug why it won't run from system32 on 2008?

Answer 1

I assume you are on a 64-bit OS, and program.exe is a 32-bit program.

If that is true, put the configuration file into the windows\syswow64 directory. This is what appears to a 32-bit process as system32.

Due to the way Windows starts processes, they do not directly inherit their working directory, but perform the equivalent of a cd during startup. Hence the 32-bit process, started in the 64-bit system directory, actually ends up in the 32-bit system directory (called syswow64 in a 64-bit cmd or Explorer).

Answer 2

Your app is sensitive to which directory it is run in due to expecting certain files- configuration files, maybe dlls or third party exes- from the current directory. I would suggest using pstools procmon to determine the.exact files being accessed.

This may also be a 32 vs 64 bit problem. Procmon will show this as well.