Integrity checking vs patching
Posted Jul 23, 2005 14:24 UTC (Sat) by ayeomans
Parent article: OLS: Linux and trusted computing
I have a sneaking suspicion that the software integrity checking will never be made to work with a general-purpose OS. It just about works with a closed games machine (not forgetting the MechWarrior Xbox hack), but as soon as users install software, or the OS and application vendors need to install patches, this breaks. Not because it has to technically, but because the processes around installation and patching, including updating checksums for all language versions, are simply not that controlled.
Don't believe me? Have a look at issues with Windows SFC. The Win98 version was unusable once you patched anything (so many changes to manually accept that you had to accept all updates were correct; WinXP only scans a few system files). I've also had experience with AIX Unix - it's file checks never updated properly after patches.
Sure, trusted integrity checking has potential, but only if manually driven by someone who understands the process. In which case they might as well use Tripwire or similar.
to post comments)