RE: Another Low Blow From Microsoft: MBSA Failure!

[Frank Knobbe <frank () knobbe us>]

On Tue, 2004-02-10 at 13:26, Joe DeMarco wrote:
> Maybe it's just me but, I wouldn't consider a patch successfully
> applied
> until the machine is rebooted. Registry changes usually require this
> process.

I would go even further and question the reliability of just checking
for the presence of Registry keys that claim a patch has been installed.
Anything short of verifying the MD5 hash of a given DLL, driver file or
executable just makes assumptions about a patched version being present
or not. Those assumptions tend you come back to haunt you, and I believe
there are enough people that had exactly that happening. I remember some
patch (a year or so ago) that overwrote a previously patched DLL with a
vulnerable version. Anything checking Registry keys, like Windows Update
I believe, made the assumption that the system was patched when in fact
the defective DLL rendered the system vulnerable.

Any tool, Windows Update, MBSA, or 3rd party should check the actual
files in question, not just logfiles or Registry keys (or anything that
makes historical statements rather than actual statements).

Regards,
Frank

Attachment: signature.asc
Description: This is a digitally signed message part