If you follow the Best Practice Guide for EFS and export this recovery key immediately after install, you can avoid this little problem.
As far as the key being stored on the same system, it is protected by 5 layers of encryption. Changing the password of the user, changing domain membership of the machine, or reinstalling the OS obliterates the original key.
EFS in Windows 2000 uses DESX for encryption. 3DES is available in Windows XP. You should define dedicated cryptanalysis. Given enough time and cycles, any cipher can be cracked. You're security guy is either a genius or vastly overpaid.
Problem is, we're talking about Windows XP, so Mr. Pfeil is wrong.
Assuming one can get Admin access to the installed OS (re-installing OS destroys access to EFS-protected files), resetting the password on WinXP in a Workgroup (as opposed to changing it) destroys access to DPAPI-protected keys, and hence access to EFS-protected files.
Win2000 EFS is vulnerable to this sort of attack, but not WinXP.
With WinXP, an attacker should endeavor to crack the user's password rather than change it to a known value. Even so, this attack can be mitigated by a) using strong passwords, and b) using SYSKEY to protect the SAM from offline attack.
Other notes:
1) EFS was principally designed to protect data when the hardware has been compromised, so the premise of this whole comment is wrong.
2) EFS is one layer of defense-in-depth. It should be combined with strong passwords, SYSKEY, and proper recovery key management.
Slashdot Mirror
If you follow the Best Practice Guide for EFS and export this recovery key immediately after install, you can avoid this little problem. As far as the key being stored on the same system, it is protected by 5 layers of encryption. Changing the password of the user, changing domain membership of the machine, or reinstalling the OS obliterates the original key. EFS in Windows 2000 uses DESX for encryption. 3DES is available in Windows XP. You should define dedicated cryptanalysis. Given enough time and cycles, any cipher can be cracked. You're security guy is either a genius or vastly overpaid.
Afraid not. On a non-domain workstation or server, the first admin user to log on to the box (usually Administrator) is the Recovery Agent.
EFS Best Practice Guide states quite clearly that this recovery key should be exported and archived immediately.
I agree with you, however, in that this isn't a hole.
This is flat wrong on Windows XP. I invite you to try it.
Here's another rule: Know what you're talking about before making "obvious" comments.
Problem is, we're talking about Windows XP, so Mr. Pfeil is wrong.
Assuming one can get Admin access to the installed OS (re-installing OS destroys access to EFS-protected files), resetting the password on WinXP in a Workgroup (as opposed to changing it) destroys access to DPAPI-protected keys, and hence access to EFS-protected files.
Win2000 EFS is vulnerable to this sort of attack, but not WinXP.
With WinXP, an attacker should endeavor to crack the user's password rather than change it to a known value. Even so, this attack can be mitigated by a) using strong passwords, and b) using SYSKEY to protect the SAM from offline attack.
Other notes:
1) EFS was principally designed to protect data when the hardware has been compromised, so the premise of this whole comment is wrong.
2) EFS is one layer of defense-in-depth. It should be combined with strong passwords, SYSKEY, and proper recovery key management.
3) Windows XP Key security is discussed here.
4) EFS does not support keys on removeable devices as of WinXP.