[anti-censorship-team] Azure domain fronting, meek ESNI

[David Fifield]

On Sat, Mar 27, 2021 at 10:33:46AM -0400, Cecylia Bocovich wrote:
> It looks like Azure is going to shutdown domain fronting:
> https://www.microsoft.com/security/blog/2021/03/26/securing-our-approach-to-domain-fronting-within-azure/
> 
> There isn't a time frame listed in the article, and I haven't gotten any
> notifications through my Azure account yet.

One possible alternative is ESNI with Cloudflare, using the mainline
meek code and its support for a headless (ESNI-supporting) Firefox.
However, this will require a lot of Tor Browser work to swap meek
implementations and re-wire the headless browser support files.

The meek source code (but not the obfs4proxy meek_lite now used in Tor
Browser) supports using a headless Firefox for TLS camouflage. As a side
benefit, if Firefox supports ESNI, then meek does as well. I used it
successfully in 2018 with Cloudflare server ESNI and the changes are
merged into the master branch, though I haven't looked at it much since
then and it hasn't been tried in deployment.
	https://bugs.torproject.org/tpo/anti-censorship/pluggable-transports/meek/28168#note_2602599

It would be unfortunate, because the headless Firefox setup was hard to
maintain, and for that reason the meek implementation was switched to
obfs4proxy once obfs4proxy gained support for uTLS-based camouflage.
	Use uTLS for meek TLS camouflage in Tor Browser
		https://bugs.torproject.org/tpo/applications/tor-browser/29430
	clean up the old meek http helper browser profiles
		https://bugs.torproject.org/tpo/applications/tor-launcher/31491
	https://blog.torproject.org/new-release-tor-browser-90a6
	https://blog.torproject.org/new-release-tor-browser-90
But restoring the browser-camouflage version of meek is something that
could be done, in order to keep it working in the absence of Azure
domain fronting support. Back then, I made a demonstration branch
showing how to integrate the meek changes into Tor Browser:
	https://bugs.torproject.org/tpo/anti-censorship/pluggable-transports/meek/29347#note_2602669
	https://gitweb.torproject.org/user/dcf/tor-browser-build.git/log/?h=meek-webextension_2&id=cf500b9c37a70ea42b494760c80bd6974944f9c0

One problem with the headless Firefox model is that the TLS fingerprint
of the ESR release used by Tor Browser would rapidly become uncommon
(because most people don't run ESRs). See Section V of
https://tlsfingerprint.io/static/frolov2019.pdf. But we currently have
that problem anyway, as the version of uTLS we are using is two years
old (Chrome 72, Firefox 65, and even the dev branch is 9 months old).
	https://gitweb.torproject.org/builders/tor-browser-build.git/tree/projects/goutls/config?h=tbb-10.5a13-build1#n3
	https://gitlab.com/yawning/utls/-/tree/v0.0.10-1

I've said "ESNI" and not "ECH" in this post because while Firefox 85 now
has client support for ECH, there's no server support for it yet. And
the ESR on which Tor Browser is based (currently Firefox 78) does not
yet have ECH support. As far as I know, ESNI with Firefox ESR and
Cloudflare works fine today.
	https://www.ghacks.net/2021/02/24/the-case-of-the-missing-esni-support-in-firefox-85/