RE: Use Cases | ACTION-13 Revisited

I was going for a humbler objective - if an app chooses to encrypt a single value in a stable key-value store, it can do that. But as you correctly pointed out, that is indistinguishable at the API level from use case 2.2.

It would be nice if applications could assume some strong security properties about client key-value stores, but I'm not sure that is going to happen in the short term. This is all about allowing apps to build some additional assurance on top of an untrusted stable store (which, per the other thread, appears to be what Facebook is after as well).

-----Original Message-----
From: Ryan Sleevi [mailto:sleevi@google.com] 
Sent: Monday, August 27, 2012 11:57 AM
To: Arun Ranganathan
Cc: Vijay Bharadwaj; public-webcrypto@w3.org; estark@mit.edu
Subject: Re: Use Cases | ACTION-13 Revisited

Vijay,

Just to make sure I followed - is your proposed use case the same as what I suggested - where the IndexedDB/storage provider takes some key and uses it protect the entire key/value/indexed store?

The provenance of the Key doesn't matter to the security properties of the storage. Instead of using a pre-provisioned key, you could equally have the service give the user agent the Key after the user had authenticated. Under such a scenario, you prevent offline access from working, but it nominally protects against device theft (unless, you know, the user stores their password within the OS/user agent, and the attacker can compromise that...)

Still, it does seem like we're in agreement that encrypting single key/value pairs stored in stable storage isn't desirable, but what is desirable is to encrypt/protect the entire store (eg: all of IndexeDB/WebStore). While such features would be useful and have a number of interesting properties, it does seem like something that would require multi-WG coordination, and a strong understanding of the security properties and guarantees, which I'm not sure we have quite yet.

On Mon, Aug 27, 2012 at 9:37 AM, Arun Ranganathan <arun@mozilla.com> wrote:
> Vijay,
>
> There are two things in this use case:
>
> 1. The ability for the API container for the Crypto API to access keys from secure elements.  As discussed in the F2F, ideally this is done seamlessly; namely, when a secure element is present, the user agent detects it as a key repository.
>
> 2. The ability to have an encrypted local store.  But this is the same as the question rsleevi raised before, namely is this reinventing the IndexedDB/localStorage wheel, but enabled for cryptography?  Is this something this group should take on?
>
> I envision cross-group coordination at some point, but I'm wondering whether for now we should not include encrypted local storage as a use case.
>
> -- A*
>
> On Aug 27, 2012, at 5:55 AM, Vijay Bharadwaj wrote:
>
>> Perhaps there is a case for locally encrypted content when you combine it with a secure token.
>>
>> Take for example a web app that stores its local data encrypted to a smart card (provisioned out of band, like we have been assuming all trusted smart cards are). Then while the app is vulnerable if it is used after the user agent is compromised, at least it raises the bar by requiring the attacker to do a two-touch attack. An attacker who just compromises the user agent cannot decrypt the locally stored data, because the user agent itself cannot decrypt it without the token.
>>
>> To be more specific:
>>
>> Use case: encrypted local storage
>>
>> When caching sensitive data locally, an application may wish to ensure that this data cannot be compromised in an offline attack. In such a case, the application may leverage a key stored on a secure token distributed out of band (such as a smart card) to encrypt the local cache. Thus, the cache may only be decrypted by the application when the secure token is present; at other times (such as when an attacker has stolen the machine) the local cache is inaccessible and all operations will require online authentication to the application's web service.
>>
>> -----Original Message-----
>> From: Arun Ranganathan [mailto:arun@mozilla.com]
>> Sent: Friday, August 17, 2012 7:57 AM
>> To: Ryan Sleevi
>> Cc: public-webcrypto@w3.org; estark@mit.edu
>> Subject: Re: Use Cases | ACTION-13 Revisited
>>
>> Ryan,
>>
>>
>> On Aug 16, 2012, at 7:16 PM, Ryan Sleevi wrote:
>>
>>> On Thu, Aug 16, 2012 at 3:55 PM, Arun Ranganathan <arun@mozilla.com> wrote:
>>>> While working through the use cases (per [ACTION-13]) with Wan-Teh 
>>>> (wtc), we came up with the following:
>>>>
>>
>> <snip/>
>>
>>>> 1. The use cases rsleevi added to the draft [spec] are pretty 
>>>> solid; they are only missing a "local storage" scenario, first 
>>>> mentioned on the Wiki [cf. local].
>>>> [cf. local]
>>>> http://www.w3.org/community/webcryptoapi/wiki/Use_Cases#Storing_loc
>>>> al
>>>> _storage
>>
>>
>>> I'm a little concerned about the "local storage" case, and wondering 
>>> whether it's something that would necessarily be in scope for this 
>>> group.
>>>
>>
>>> Consider the example of IndexedDB, which uses "Keys" (IDB keys - 
>>> http://www.w3.org/TR/IndexedDB/#key-construct ) and returns "Values" 
>>> ( http://www.w3.org/TR/IndexedDB/#value-construct ), and can 
>>> alternatively be accessed via indices ( 
>>> http://www.w3.org/TR/IndexedDB/#index-concept ).
>>>
>>
>>> A naieve assumption would be that this API would only protect the 
>>> Values - not the keys, nor the indices. However, as practically 
>>> deployed today, that wouldn't offer much protection, since both Keys 
>>> and Indices often reveal quite a bit of information.
>>>
>>> Further, by ciphering contents, it's a tradeoff between efficiency 
>>> and privacy. Perfect privacy (storing no relationships about 
>>> keys/indices, everything randomly distributed) is the worst 
>>> efficiency, while perfect efficiency (which is what is afforded by 
>>> today's IndexedDB) has no privacy/cryptography.
>>>
>>> A refinement might be to have the IndexedDB actually take a Key 
>>> (Crypto API key), that it can use to protect however the IndexedDB 
>>> is stored - keys, indices, everything. Call it an "EncryptedIndexedDB".
>>> This is better, in that it allows the user agent to decrypt on the 
>>> fly (see caveat), and allows applications to use existing indices/keys.
>>> The caveat, however, is that encryption requires defining an 
>>> encryption algorithm, and the choice of encryption algorithm 
>>> directly affects the efficiency of the API. For example, under 
>>> today's IndexedDB, a user agent can load data on the fly (eg: from 
>>> disk), but under EncryptedIndexedDB with say, a block cipher alg 
>>> like AES, it might have to read the entire DB into memory, then 
>>> decrypt, in order to be able to offer this functionality.
>>>
>>> Even more fundamentally though, is the question about what attack 
>>> this is trying to defend against. The arguments I've heard for 
>>> encrypted local storage seem to be about a remote server, serving a 
>>> web application, distrusting the client platform. If that's the 
>>> case, it doesn't seem like any level of cryptography will save them. 
>>> As I noted in the existing security considerations, it SHOULD be 
>>> perfectly valid for a user agent to store a key in plaintext on 
>>> disk, so what actual protections are afforded by this?
>>
>>
>> You're right -- if the use case is primarily about an untrusted multi-user machine or virtual computing environment, we're only as safe as general user safety anyway.  This doesn't seem to be a use case we can salvage, nor one that should influence the API.  We should probably not include it.
>>
>> But:
>>
>>>
>>> If something like EncryptedIndexedDB is what is meant here, then 
>>> this seems like something that would likely live in the Web Apps WG 
>>> (since it's about extending IndexedDB).
>>>
>>
>> Maybe -- I doubt it's worth their while to solve for that use case either :).  Interestingly enough (and not to confuse matters, but) we've just heard from Facebook [FB-ScriptSigning] about localStorage (or IndexedDB) used as a script cache.  People are already using IndexedDB and localStorage in unsafe-ish ways.  Of course, we shouldn't confuse script signing with a general use case for protected/encrypted local storage, but perhaps if we jettison the "protected local storage" use case, we can bolster the "document signing" use case to explicitly refer to documents extracted from local storage for signature verification.
>>
>> This raises the sticky issue of types of documents.  We might naively say that a script is no ordinary document, and can be used by the relevant JSON primitive if it passes signature validity.
>>
>> In a nutshell, I'm saying: perhaps we cannot cater to an encrypted 
>> local store use case, but we may be able to flesh out the use case 
>> for signature verification, including extraction from local storage.  
>> Our use cases should encourage patterns of behavior that we think are 
>> desirable.  We can't control or solve for undesirable patterns of 
>> behavior :)
>>
>>
>>> I just want to make sure that we're carefully considering the use 
>>> case and the security implications before committing to them, as 
>>> well as to figure out what parts of the spec may need to change in 
>>> order to meaningfully implement them.
>>
>>
>> +1.
>>
>> -- A*
>>
>> [FB-ScriptSigning] 
>> http://lists.w3.org/Archives/Public/public-webcrypto/2012Aug/0121.htm
>> l
>>
>>
>>
>>
>

Received on Tuesday, 28 August 2012 14:13:12 UTC