1

For a while now I've had this reproducible error:

If I try to open any file (so regardless of its extension) that has a % in its name, regardless of the method of opening (click on it in Explorer, open it manually in the command prompt etc.), the program which is used to send the request (Explorer and cmd.exe, in the previous examples) would crash 100%.

Does anyone know the exact mechanism/sequence Windows uses to open a file? I could help out with debug information that way.

Oliver Salzburg
  • 87,539
  • 63
  • 263
  • 308
Buote Xu
  • 13

1 Answers1

0

It's a long road from the point where you double-click on an icon on screen to the point where an application is started.

There is a lot go wrong.

  1. You click on a file.
  2. Windows determines the file type.
  3. Can this file be executed on its own, or does it require another application to open it?
    Let's check what the registry says about the file extension.
  4. Registry says it's a txtfile.
  5. Registry says, to open a txtfile we should use a program called C:\Windows\system32\NOTEPAD.EXE
  6. Now we can finally create our process: Process Create Event captures in Process Monitor

There is so much stuff involved, there is no way of telling what exactly is the problem. You need to perform further investigations yourself.

Please have a look at:

Community
  • 1
Oliver Salzburg
  • 87,539
  • 63
  • 263
  • 308
  • But it also crashes if I, say, try notepad.exe blub%.txt, which skips a lot of your described steps, shouldn't that narrow it down? – Buote Xu Apr 28 '12 at 10:24
  • @BuoteXu: Yes, it narrows it down from 10 trillion possible reasons to 5 trillion possible reasons. To be more serious, yes, it does, but not by enough to make me able to know what's going on without having more detailed data. – Oliver Salzburg Apr 28 '12 at 10:27
  • 1
    Solved it, the crash was caused by Spybot's SDHook32.dll, which somehow couldn't deal with the filename. – Buote Xu Apr 28 '12 at 10:55
  • 1
    @BuoteXu: Please keep in mind that it is perfectly fine to answer your own question and accept that as the correct solution, if you're able to solve your problem yourself. This could help future users. – Oliver Salzburg Apr 28 '12 at 10:57