-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

On Thu, Feb 13, 2025 at 10:09:21PM +0000, 'grandbronze111' via qubes-devel 
wrote:
> I'm using the mailing list as I can't use GitHub with Tor Browser.
> 
> This mailing list post is a succinct version of my forum thread 
> https://forum.qubes-os.org/t/mac-randomization-is-flawed-proposed-new-solution/32150
> 
> https://github.com/QubesOS/qubes-issues/issues/938 needs to be reopened.
> 
> We can’t rely solely on either udev (Tails approach) or NetworkManager (Qubes 
> approach) to ensure the true MAC address is never leaked. Some drivers 
> restore to the true/permanent MAC on sleep/suspend and udev doesn't handle 
> this. NM, in normal operation, brings the NIC 'up' with the true MAC address 
> which can result in leaks. NM also restores to the true MAC address in other 
> cases, such as when the user restarts the NM service to apply an updated 
> config. Even if NM did do everything properly, drivers can still leak the 
> true MAC address.

Device gets re-initialized after suspend, and NM does re-apply its
configuration on resume, including relevant MAC randomization setting.
_If_ that doesn't work (the above looks like just speculation, not actual
observed behavior), then relevant piece may need fixing. But none of
those pieces are qubes-specific. If setting MAC is broken either in some
specific driver or in Network Manager, then it needs to be reported in
appropriate place (the driver maintainers or NetworkManager bug
tracker).

> As originally proposed in /issues/938, patching drivers is the only way to do 
> this without the possibility of leaks occurring. Patching these drivers is 
> usually very simple, as most of them already support generating random MACs 
> for situations where the EEPROM MAC is invalid.

If driver maintainers would accept such patch, then maybe. We are not
going to carry such patch ourselves.

> As an initial proposal, I suggest Qubes hosts a contrib repository where 
> these patches can be submitted. Then, Qubes can build a separate kernel with 
> these modified drivers that users can optionally set for their NetVM. I can 
> do the necessary work, but I would want Qubes to confirm how they would like 
> this structured so that it will be accepted. I have built a modified kernel 
> locally with qubesbuilderv2, but I'm not familiar enough with Qubes' build 
> system to say for certain what the best way to do all of this is. It may be 
> that there's an entirely different way to supply the modified drivers that 
> makes more sense than building a separate kernel.

See https://docs.kernel.org/process/submitting-patches.html how to
submit Linux patches.

- -- 
Best Regards,
Marek Marczykowski-Górecki
Invisible Things Lab
-----BEGIN PGP SIGNATURE-----

iQEzBAEBCAAdFiEEhrpukzGPukRmQqkK24/THMrX1ywFAmevI8oACgkQ24/THMrX
1yyJawf+K7IGlmRzuulRDIiCVtqGjnvj8E3/BsYrqPYruchmFS5JF/Lbj4eqHNHR
CNLpRLrShWgATu9/4GABAUTBInIohP6d0CK4WiCtLxZyJyjpEr879k324TyiXPf5
c42GQ8HzKItagUYY7d+T7TO+J2PTVGVAoAnT8VQtDMUK+/9B3G6oQGsqtSCi3Wm3
imz5MvTPmnTeJErUCyvKITN/TX9WEum0+On9H1SvypjbLqpxBjTmBj2ybP65wr/d
+jyPg6LYznSd+4jciRv7AqhA63DT85EydqMSsfZVDPJTJOYwzJ9CCcMC8IpVs+0S
bns0MvlGCgfF7tESpPeaAxzZ7WCrxQ==
=8Rfy
-----END PGP SIGNATURE-----

-- 
You received this message because you are subscribed to the Google Groups 
"qubes-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion visit 
https://groups.google.com/d/msgid/qubes-devel/Z68jygcJfZbSekzN%40mail-itl.

Reply via email to