1 |
On 06/03/20 19:39, Rich Freeman wrote: |
2 |
> On Fri, Mar 6, 2020 at 2:07 PM Wols Lists <antlists@××××××××××××.uk> wrote: |
3 |
>> |
4 |
>> On 06/03/20 13:48, Rich Freeman wrote: |
5 |
>>> If you fall into this camp you need to still update your firmware to |
6 |
>>> address the non-TPM-user and to avoid making it trivial for software |
7 |
>>> to steal your keys/etc. However, you need to be aware that you are no |
8 |
>>> longer secure against physical theft of your device. Somebody who |
9 |
>>> steals your laptop with passwordless encryption might be able to break |
10 |
>>> the encryption on your device. |
11 |
>> |
12 |
>> It's worse that that, he's dead, Jim! |
13 |
>> |
14 |
>> The summary on LWN is an easy read. Somebody who steals your Intel |
15 |
>> laptop WILL be able to break the encryption on your device. |
16 |
>> |
17 |
>> tl;dr summary - the microcode that *boots* the cpu has been compromised. |
18 |
>> So even while it is setting up tpm and all that malarkey, malware can be |
19 |
>> stealing keys etc. |
20 |
> |
21 |
> They don't detail the effort required. If the firmware is patched it |
22 |
> sounds like it still requires tinkering with hardware. |
23 |
|
24 |
By then it's TOO LATE. The firmware is signed for security, AND LOADED |
25 |
AT BOOT. But if the boot process is compromised, the attacker simply |
26 |
doesn't load the patched firmware. |
27 |
|
28 |
> However, there |
29 |
> really isn't nothing you said that doesn't agree with what I said. |
30 |
> |
31 |
> Whether they "WILL" be able to break the encryption on your device |
32 |
> depends a lot on the details and the knowledge of the attacker. Hence |
33 |
> the reason I said "might." In any case, might is good enough to not |
34 |
> rely on a broken security feature. |
35 |
> |
36 |
>> Which means that Intel's master signing key will soon be cracked and |
37 |
>> compromised. |
38 |
> |
39 |
> Yes, but keep in mind the signing keys have nothing to do with disk |
40 |
> encryption. It is for remote attestation. Hence my Netflix comment. |
41 |
> |
42 |
Signing keys have EVERYTHING to do with whether you can trust the CPU. |
43 |
If you can't trust the CPU, then it can simply read the disk encryption |
44 |
credentials without any reference to whether it SHOULD read them. |
45 |
|
46 |
If the system doesn't *require* manual intervention at boot to provide |
47 |
"something the user knows", then it's game over - your hard drive is |
48 |
wide open to the attacker. |
49 |
|
50 |
And it only takes ONE person to crack that master key ONCE, and |
51 |
EVERYBODY is up shit creek. |
52 |
|
53 |
At the end of the day, it's a "tree of trust". And once the root key is |
54 |
compromised, you can NOT trust ANY key that was secured by said root |
55 |
key. And given that it's Intel's master key, anything that relies on the |
56 |
security of the CPU has just been shafted. |
57 |
|
58 |
The ONLY systems that are safe are those that have a separate TPM chip, |
59 |
because you no longer rely on the CPU's integrity. |
60 |
|
61 |
Cheers, |
62 |
Wol |