On 24.09.21 09:36, Johannes Berg wrote:
[dropping now dead linuxwifi@xxxxxxxxx]
I know it's been (almost exactly) a year, but I was wondering about this
scenario recently
(due to e.g. https://bugzilla.kernel.org/show_bug.cgi?id=213059, though
I'm not sure that report even makes sense).
I had a quick look in the ticket. It may just be the "other end" which
is not able to re-key correctly. After all the tests I did with mvm I
should have noticed rekey issues.
Maybe. But we see this also on Cisco gear, which I'd hope wasn't so bad
:)
I no longer have any expectations here. In the end it's probably just
luck if they test for it and/or the developers notice it.
After all even the authors of ieee892.11 did miss the implications of
what they put in the standard. A few warning words or better something
like Extended Key ID would have avoided most of the issues quite easily
back then.
Also, I wasn't sure what NIC you had actually tested - there are three
different scenarios depending on the NIC/encryption, as I outlined in my
previous email (1a, 1b, 2).
AC 9260 and AC 9560
But there is one thing sticking out in the problem description:
kernel: wlan0: disassociated from 4c:a6:4d:54:32:6b (Reason:
15=4WAY_HANDSHAKE_TIMEOUT)
--> with that it's can't be a PTK0 rekey bug as we defined it so far.
With the 4way handshake incomplete the new key has never been install on
the device the log is from.
Hmm. Good point. But we did also see the message saying
Rekeying PTK for STA ... but driver can't safely do that.
so it would appear that some rekeying happened successfully.
Oh. So maybe in this particular case some rekeying happened
successfully, but then another rekeying later got stuck?
So this is probably a case where the other end - should be the STA and
not the AP - activated the new key and then encrypts EAPOL#4 with the
new key... When the other end is not using control port and the driver
has no special workaround this can easily happen..
Problem here is, that when wpa_supplicant hands over EAPOL #4 to the
kernel this only guarantees that one frame is send out. When there are
others in the QDISC that won't be the EAPOL#4 frame. See
https://patchwork.ozlabs.org/project/hostap/patch/20190912192002.6105-1-alexander@xxxxxxxxxxxxxx/
I'll look at that closer. But only way I see is:
1) STA sends out EAPOL#4
2) AP is not reviving/able to decrypt it
3) And then disassociates the STA.
That seems to indicate that the mvm card did send out EAPOL#4 encrypted
with the new key. Assuming the STA is modern enough to use control port
that indicates some serious delay in mvm HW queues..
Now when the STA is not using control port the observation in the bug
report would be more or less the "expected" behavior. We probably should
ask for a wpa_supplicant debug log here.
Well, we don't even know that the other end is using mac80211/hostapd,
but I get your point.
Did you ever send another patch? I can't seem to find any.
No, I never send a patch. To be able to detect incorrect key usage I
ended up implementing key color for mac80211 only to discover that this
in not adding anything useful and never submitted it. I ended up with
the understanding that we do not need the flush but have now problems to
reproduce the reasoning for that. But let's see what I can together here.
OK.
But basically, in mvm, we support two scenarios these days:
1) PN assigned by the driver, in iwl_mvm_set_tx_cmd_crypto(), with two
sub-cases:
a) key material embedded into the TX command (CCMP, TKIP, WEP)
b) key material taken from firmware key offset (CCMP-256, GCMP)
2) PN assigned by the device per the station, via the "new TX AP > selected in iwl_mvm_set_tx_params().
Am I wrong in thinking that both scenarios 1a) and 2) are completely
acceptable for CAN_REPLACE_PTK0, since there's never any possibility of
sending a frame with a mismatch between the PN assignment and key
material?
CAN_REPLACE_PTK0 only has to confirms that the driver is handling all
the races PTK0 rekeying has. Which is possible in all of the cases above.
In a nutshell the driver and cards must never send out a frame where the
PN is not matching to the PTK key used for encryption AND never send out
a frame - when we are set up for encryption - with a deleted keys
Right. Mostly, anyway. In 1b, we do have a race, because the PN is
assigned by driver and the key material can be programmed while frames
(with old PN) are on the TX queue(s), which can be handled by flushing
all the (affected) queues (i.e. for this station).
However, the race is often benign, because what happens is that we have
32 (I think? maybe 16) key "slots" and reuse them as late as possible.
Thus, if we do a pretty standard rekeying, the reference to the old key,
when dereferenced in the firmware, will actually still be to the correct
old key. Only if we had enough new keys installed to override that
reference, would we have any issues.
You want to skip deleting the key in the firmware when mac80211 issues
the DISABLE_KEY set_key() call for it? That should indeed work.
As long as the firmware plays along. It was acting a bit odd when having
two STA keys installed when I tried Extended Key ID with it.
The bigger problem here happens if we do have enough key slots for this
(e.g. because we're the AP) and then reuse it ... potentially with a key
for a *different* station/connection. Then when the firmware
dereferences it, it'd be (theoretically) readable to another station
that it wasn't intended for, though the good thing is that the receiver
won't be able to decrypt it (anyway even if it's with the old key).
When the key slot is reused it will have the new key installed for the
current (new) station. It should be decrypted and be perfectly readable.
Worse, if the PN was generated not by the HW it will bump the PN and
make the connection unusable.
Only if we roll over enough keys that the new PTK0 ends up in the same
slots would we actually end up in the "stuck connection" case.
Either way though, if we roll over the key (regardless with which key),
we have a problem.
However, it seems that scenario 1b) is what this patch attempted to
handle, by flushing the TX queues when the new key is installed, and I'm
not sure why you said it wasn't necessary - if the driver installs new
key material in the device while there are frames that already have a PN
set, then the old PN _might_ be used with a new key, leading to
problems.
You are probably right but it depends how the HW is handling that in detail:
mac80211/iwlwifi/mwm together make sure that when a skb is send to the
HW the HW also has a valid key for it. At that very moment the frame can
be send. Of course chances are it will be queued there again...
Now when the key is only de-referenced on dequeue we may have an issue
and need a flush or another mitigation.
Yes, this is what happens in 1b.
Regardless how the HW is handling that here are some indirect mitigations.
mac80211 sets first KEY_FLAG_TAINTED and then waits for all rcu
sections. After that the HW can't get any skb depending on the outgoing
key. Then the old key is deleted from the HW and when it still has
frames queued needing the deleted one it may go out with a null key or
so. Even more time and waiting on running RCU sections will pass till
the new key is installed and we can freeze the connection instead "only"
having a cleartext leak.
(Note the above description about key slot rollover)
When A-MPDU is active on rekey stopping it will also flush the
problematic frames for my understanding.
Yeah. I think so.
Actually, I think this whole business of stopping aggregation sessions
is another potential issue - though particularly *RX* sessions, because
we don't know if the AP will always re-establish them.
Works fine for my HW but I get your point.
Unfortunately exactly the RX sessions are the problem: Noting in the
standard is clearly forbidding to mix old and new PTK key in one session
and for sure it happens in reality all the time.
So when one A-MPDU group includes any frame with a PN from the old key
it will bump the PN of the new key and freeze the connection.
To avoid that we have to identify and (probably) drop the frames after
the MPDU reorder buffer. (I did not look into that at all so far.)
But looks like that needs a mac80211 API extensions and very likely
firmware support from (many) cards. Or some really ugly heuristic
potentially impacting security.
There's a probabilistic defense against this, in that we attempt to
reuse key offsets (the hw_key ID that goes into the TX command) as
rarely as possible, so that if we put a frame with key offset 0 into the
queue, and then reinstall the PTK, it would go to key offset 2 (1 being
used by the GTK), and 0 would stay unless we did another rekeying or so.
The HW must then also be able to detect when a frame refers to a free
hw_key ID and drop those frames.
No, the hw_key ID isn't marked free, it just keeps the old key material,
unless we reuse that slot. If it's not overwritten, this isn't really a
big deal since the receiver won't be able to decrypt it any more, but in
any case it's still protected well.
The complication just comes in if we reuse that slot, as described
above.
Got it now.
Maybe keeping the hw_key_id but setting and checking a key color would
be better?
That would require firmware support.
Not sure what the HW is capable of here but adding a single bit to the
key information and compare it to the color of the frame can hopefully
be done in the firmware.
It could be, yes. And I could even implement it, but it wouldn't really
roll out well, and it's a really old device. I don't think it's worth
the effort.
We then would have to make sure the hw_key_id is not changing on rekey
to be sure concurrent rekeys are not swapping the hw_key_ids somehow.
Yeah we could ensure that.
I was never able to get my mvm cards to mix up the key or send cleartext
frames. (I think I first played around with CCMP-256 and later hacked
CCMP to also use the key mapping.)
CCMP-256 was never supported, I think, but GCMP/GCMP-256.
But yes, I'm not surprised you weren't able to, given that key slot
reuse would have to occur.
But I may have mixed up mac80211 queues with HW queues:
While mac80211 queues are safe the HW queues may still need the flush.
We don't have to flush the mac80211 queues. Cycling or draining all
frames handed off to the HW must be sufficient.
Right.
So we have two solutions: Flushing the queues when we rekey
GCMP/GCMP-256 or cycle the key slots. Flushing is quite simple but
cycling the key ids is in >>99% getting the job done without a flush.
One of my attempts to fix the PTK0 issues tried to use something like
that for the ath9k driver. It turned out that we still have to stop TX
for the outgoing key and can only resume once the new key is active in
the HW, making key slot cycling unnecessary.
That said it should work here, but only when the HW is able to drop
frames pointing to an cleared hw key.
As I see it mvm only can get and hand over frames to the HW when the
correct key is still installed in the HW.
It's not HW, it's firmware, but yeah. The thing though is that the
driver has assigned the PN - so key slot reuse ... (Both are after all using keyid 0.)
So we either flush, rotate the key slots or find something else we can
do in the driver and does not need firmware updates.
The other question I had was concerning the documented requirements for
NL80211_EXT_FEATURE_CAN_REPLACE_PTK0, aren't those too strict? For
reference, this is what it says now:
* Mac80211 drivers should set the @NL80211_EXT_FEATURE_CAN_REPLACE_PTK0 flag
* when they are able to replace in-use PTK keys according to the following
* requirements:
* 1) They do not hand over frames decrypted with the old key to
mac80211 once the call to set_key() with command %DISABLE_KEY has been
completed when also setting @IEEE80211_KEY_FLAG_GENERATE_IV for any key,
This is I think a bit misleading, how's the RX-side related to
GENERATE_IV? It seems to me that this requirement is to ensure we don't
get a bad replay counter on RX, but then that's unrelated to
GENERATE_IV?
Yes, looks like that ended up in the wrong sections with all the
rewording I've done here.
OK, thanks. Want to send a patch? Or I can fix it too.
No preference, I'm fine when you do it. But I've submitted a patch now
so you hopeful just can merge it.
The idea is to not bump the PN of the new key with a frame decrypted by
a now deleted key only.
Yeah. I'd phrase it TX-side and say "the idea is to never do PN
assignment and encryption from two different keys" :)
Ups, should have read that prior to submitting the patch. Just change
whatever you want or make your own patch.
I just moved the wrong statement to the correct block.
2) either drop or continue to use the old key for any outgoing frames queued
at the time of the key deletion (including re-transmits),
That's mostly true for iwlwifi, apart from the case 1b) key offset reuse
I was explaining above.
3) never send out a frame queued prior to the set_key() %SET_KEY command
encrypted with the new key and
This I don't really know why - I think maybe *this* was meant to have
the "when also setting GENERATE_IV"?
Now I'm sure you are right:
Any frame queued prior to %SET_KEY with @IEEE80211_KEY_FLAG_GENERATE_IV
can bump the new PN to a invalid value. But when the hw or driver is
handling that it's ok.
Right.
Thanks a lot!
johannes
