Todd H. wrote:
> writes:
>
>> In my office we are considering using virtual PC software for security
>> reasons. It is seems to us that by using a virtual PC for web browsing we
>> can protect the host system from malware and virus.
>>
>> Is this assumption correct?
>
> By and large yes. There is talk of some malware that can break
> outside of virtualized jails, but I don't believe it's come to
> fruition yet, at least not publicly.
Breaking out of various "jails" is pretty trivial, due to numerous lacks of
safe programming (like dropping rights, file descriptors and various other
resources on startup) as well as various methods of IPC. Jails are supposed
to limit attacking surface and protect against random errors.
If by "jails" you refer to various secure VMs like Java or various complete
PC emulators, I'd like like to see any method to breaking out of these
isolations. At least for VMware (and of course Java) such secure isolation
is a major design goal and therefore well-implemented.
The most common breakout method is user stupidity. "Oh, it seems to behaved
well inside the VM, so I decided to run the code outside it" is an often
heared result of misconception, since it's almost always (and in real life
always) trivial for malware to detect that it's running inside a VM and
behave accordingly.