So yeah, the baseband firmware is a huge vulnerability. But if you use a separate modem/router, and disable the onboard baseband, that's far less of an issue.
And with the PinePhone, it's easy: https://wiki.pine64.org/index.php/PinePhone#Killswitch_confi...
The Register reported the avaerge time to infection was 20 minutes.
https://www.theregister.co.uk/2004/08/19/infected_in20_minut...
That said - because of the sheer number of phones in existence, on IPv4 you’re guaranteed to be running behind a giant NAT operated by your network carrier - and on IPv6 the address space is too big to port-scan (at least) but while it’s no help if attackers know your address - I understand there’s still a mix of carrier-based and handset-based network lockdown going on.
Cellular baseband is poorly secured, and it's privileged over userland. And its firmware is a closed-source blob, so it's ~impossible to fully assess the risks.
And so it's arguable that adversaries can pwn smartphones through baseband.
That's the analogy to Windows XP machines. Windows Firewall was just a stopgap. What helped most was going from dial-up modems, which are no more secure than network interfaces, to modem/routers with NAT firewalls.
So smartphones ought to have discrete cellular modem/routers. And that's an easy option for the PinePhone, given the kill switch.
This is how Charlie Miller and Chris Valasek were able to remotely compromise vehicles with a vulnerable infotainment system via a pwned femtocell.
"To find vulnerable vehicles you just need to scan on port 6667 from a Sprint device on the IP addresses 21.0.0.0/8 and 25.0.0.0/8."[0]
[0] http://illmatics.com/Remote%20Car%20Hacking.pdf, pdf page 46.
> It turns out that any Sprint device anywhere in the country can communicate [as in telnet] with any other Sprint device anywhere in the country.
So remote devices do have network connectivity to cellular baseband. At least on Sprint.
Do other cellular networks work like that?
More like "wry observation".
Given that most traffic is encrypted (eg any major website via https, or mail provider with smtp-over-tls, or most apps via https), a baseband vulnerability isn't a great deal of help without an iommu bypass (AFAIK only one has ever been publicly disclosed for ios, and it took the researchers over 6 months).
IIRC correctly (from a few years back) the pixel phones also have one, but most Samsung phones did not.
Prior discussion from 2016: https://news.ycombinator.com/item?id=10905643