[Cryptography] Own on install. How grave it is?

[Jerry Leichter]

> This is well known, haven't seen it discussed.
> 
> In short doing clean install (factory defaults) has a window of
> opportunity when the device is vulnerable to a known network attack.
It depends on what you install.  For example, the only version of iOS you can actually install is the most recent.  (Well, after a new update, the previous version may also be installable, typically for couple of weeks - though Apple tends to shorten the period when the new version fixes serious bugs.)

In MacOS, you can generally get something with most of the recent patches, though no always.  Again, I think Apple tend to cut a new master when serious security bugs have been fixed.

If the low-level code to download the updates had vulnerabilities, you'd be screwed - but I've seen no reports about that happening.

Apple is actually moving to (so far, optionally) a setup in which *only* the most recent version (or an earlier "approved" version) can be installed.  (So far only the new iMac Pro can actually enforce this, but it's clear that this is the future.)

In any case, the issue is not a significant one for Apple OS's.  (Let's not get into the question of whether it's *appropriate* for Apple to outright *forbid* downgrading.  It's what they do on iOS, for better or worse - and it does avoid the "known attack" problem, as well as downgrade attacks.)

> It used to be common sense to reinstall after compromise (probably
> doesn't apply to the windows world where the antivirus takes care).
These days, antivirus programs arguably introduce more vulnerabilities than they avoid!

> All versions of windoze are affected by the SMB bug to my knowledge.
> Debian jessie (old stable) is vulnerable to malicious mirror attack.
> 
> More of interest to me are devices where the installation media is
> fixed and can't be changed.
> 
> This includes smartphones and wireless routers.
I don't know of any smartphone where there are *any* "installation media" in the traditional sense.  Off hand, I can't think of any wireless routers either:  They come with some version of their software and all further updating is on line.

> Some smartphones might be vulnerable to wifi RCE (found by google?).
> Some wireless routers might be vulnerable to wifi RCE or
> default admin password attack over wifi.
> 
> Internet of Things will make things worse (some NAS devices are
> affected).
> 
> Shielding the device might not be solution since updates must be
> applied.
> 
> Are the above concerns real?
> 
> Have this been studied systematically?
If the problem is:  To re-initialize a device, I'm forced to install an older, insecure OS - which may well be attacked before I can upgrade it:  That *used to be* a severe problem with Windows (someone found, with some old version of Windows, that if you installed from CD and then connected to the internet to get updates, you'd be infected within half an hour at most - while the updates took several hours to install).  I doubt this is an issue any more, though I don't keep up with the Windows world.

The big problem with many devices - and certainly many IoT devices - is that they have *no* mechanism for updating, not that they may be compromised before you can updated them.
                                                        -- Jerry