admin管理员组

文章数量:1201026

EDIT: Per discussion in the comments, let me clarify that this will be happening server side, behind SSL. I do not intend to expose the hashed password or the hashing scheme to the client.

Assume we have an existing asp identity database with the default tables (aspnet_Users, aspnet_Roles, etc.). Based on my understanding, the password hashing algorithm uses sha256 and stores the salt + (hashed password) as a base64 encoded string. EDIT: This assumption is incorrect, see answer below.

I would like to replicate the function of the Microsoft.AspNet.Identity.Crypto class' VerifyHashedPassword function with a JavaScript version.

Let's say that a password is welcome1 and its asp hashed password is ADOEtXqGCnWCuuc5UOAVIvMVJWjANOA/LoVy0E4XCyUHIfJ7dfSY0Id+uJ20DTtG+A==

So far I have been able to reproduce the parts of the method that get the salt and the stored sub key.

Where the C# implementation does more or less this:

var salt = new byte[SaltSize];
Buffer.BlockCopy(hashedPasswordBytes, 1, salt, 0, SaltSize);
var storedSubkey = new byte[PBKDF2SubkeyLength];
Buffer.BlockCopy(hashedPasswordBytes, 1 + SaltSize, storedSubkey, 0, PBKDF2SubkeyLength);

I have the following in JavaScript (not elegant by any stretch):

var hashedPwd = "ADOEtXqGCnWCuuc5UOAVIvMVJWjANOA/LoVy0E4XCyUHIfJ7dfSY0Id+uJ20DTtG+A==";
var hashedPasswordBytes = new Buffer(hashedPwd, 'base64');
var saltbytes = [];
var storedSubKeyBytes = [];

for(var i=1;i<hashedPasswordBytes.length;i++)
{
  if(i > 0 && i <= 16)
  {
    saltbytes.push(hashedPasswordBytes[i]);
  }
  if(i > 0 && i >16) {
    storedSubKeyBytes.push(hashedPasswordBytes[i]);
  }
}

Again, it ain't pretty, but after running this snippet the saltbytes and storedSubKeyBytes match byte for byte what I see in the C# debugger for salt and storedSubkey.

Finally, in C#, an instance of Rfc2898DeriveBytes is used to generate a new subkey based on the salt and the password provided, like so:

byte[] generatedSubkey;
using (var deriveBytes = new Rfc2898DeriveBytes(password, salt, PBKDF2IterCount))
{
   generatedSubkey = deriveBytes.GetBytes(PBKDF2SubkeyLength);
}

This is where I'm stuck. I have tried others' solutions such as this one, I have used Google's and Node's CryptoJS and crypto libraries respectively, and my output never generates anything resembling the C# version.

(Example:

var output = crypto.pbkdf2Sync(new Buffer('welcome1', 'utf16le'), 
    new Buffer(parsedSaltString), 1000, 32, 'sha256');
console.log(output.toString('base64'))

generates "LSJvaDM9u7pXRfIS7QDFnmBPvsaN2z7FMXURGHIuqdY=")

Many of the pointers I've found online indicate problems involving encoding mismatches (NodeJS / UTF-8 vs. .NET / UTF-16LE), so I've tried encoding using the default .NET encoding format but to no avail.

Or I could be completely wrong about what I assume these libraries are doing. But any pointers in the right direction would be much appreciated.

EDIT: Per discussion in the comments, let me clarify that this will be happening server side, behind SSL. I do not intend to expose the hashed password or the hashing scheme to the client.

Assume we have an existing asp.net identity database with the default tables (aspnet_Users, aspnet_Roles, etc.). Based on my understanding, the password hashing algorithm uses sha256 and stores the salt + (hashed password) as a base64 encoded string. EDIT: This assumption is incorrect, see answer below.

I would like to replicate the function of the Microsoft.AspNet.Identity.Crypto class' VerifyHashedPassword function with a JavaScript version.

Let's say that a password is welcome1 and its asp.net hashed password is ADOEtXqGCnWCuuc5UOAVIvMVJWjANOA/LoVy0E4XCyUHIfJ7dfSY0Id+uJ20DTtG+A==

So far I have been able to reproduce the parts of the method that get the salt and the stored sub key.

Where the C# implementation does more or less this:

var salt = new byte[SaltSize];
Buffer.BlockCopy(hashedPasswordBytes, 1, salt, 0, SaltSize);
var storedSubkey = new byte[PBKDF2SubkeyLength];
Buffer.BlockCopy(hashedPasswordBytes, 1 + SaltSize, storedSubkey, 0, PBKDF2SubkeyLength);

I have the following in JavaScript (not elegant by any stretch):

var hashedPwd = "ADOEtXqGCnWCuuc5UOAVIvMVJWjANOA/LoVy0E4XCyUHIfJ7dfSY0Id+uJ20DTtG+A==";
var hashedPasswordBytes = new Buffer(hashedPwd, 'base64');
var saltbytes = [];
var storedSubKeyBytes = [];

for(var i=1;i<hashedPasswordBytes.length;i++)
{
  if(i > 0 && i <= 16)
  {
    saltbytes.push(hashedPasswordBytes[i]);
  }
  if(i > 0 && i >16) {
    storedSubKeyBytes.push(hashedPasswordBytes[i]);
  }
}

Again, it ain't pretty, but after running this snippet the saltbytes and storedSubKeyBytes match byte for byte what I see in the C# debugger for salt and storedSubkey.

Finally, in C#, an instance of Rfc2898DeriveBytes is used to generate a new subkey based on the salt and the password provided, like so:

byte[] generatedSubkey;
using (var deriveBytes = new Rfc2898DeriveBytes(password, salt, PBKDF2IterCount))
{
   generatedSubkey = deriveBytes.GetBytes(PBKDF2SubkeyLength);
}

This is where I'm stuck. I have tried others' solutions such as this one, I have used Google's and Node's CryptoJS and crypto libraries respectively, and my output never generates anything resembling the C# version.

(Example:

var output = crypto.pbkdf2Sync(new Buffer('welcome1', 'utf16le'), 
    new Buffer(parsedSaltString), 1000, 32, 'sha256');
console.log(output.toString('base64'))

generates "LSJvaDM9u7pXRfIS7QDFnmBPvsaN2z7FMXURGHIuqdY=")

Many of the pointers I've found online indicate problems involving encoding mismatches (NodeJS / UTF-8 vs. .NET / UTF-16LE), so I've tried encoding using the default .NET encoding format but to no avail.

Or I could be completely wrong about what I assume these libraries are doing. But any pointers in the right direction would be much appreciated.

Share Improve this question edited May 23, 2017 at 12:03 CommunityBot 11 silver badge asked Feb 24, 2015 at 21:18 GojiraGojira 3,0414 gold badges22 silver badges32 bronze badges 5
  • Are you trying to generate password hash on the client and pass the hash down to the server for validation? – trailmax Commented Feb 25, 2015 at 0:53
  • No, I'm trying to generate the hash server side in node.js. Essentially, keeping the database the same but swapping out the IIS / asp.net layer for node. I'm not a security expert but I would be wary of trying to do any password operations client side. – Gojira Commented Feb 25, 2015 at 16:28
  • Ah, that clarifies my concerns. I'd mention this in your question. Sorry, can't actually help with JS-side of things here( – trailmax Commented Feb 25, 2015 at 17:01
  • @trailmax Could you please explain your concerns with client-side hashing a little further? – bonh Commented Jun 17, 2015 at 19:12
  • 1 @bonh See this explanation security.stackexchange.com/a/53606 – trailmax Commented Jun 17, 2015 at 19:14
Add a comment  | 

4 Answers 4

Reset to default 16

Ok, I think this problem ended up being quite a bit simpler than I was making it (aren't they always). After performing a RTFM operation on the pbkdf2 spec, I ran some side-by-side tests with Node crypto and .NET crypto, and have made pretty good progress on a solution.

The following JavaScript code correctly parses the stored salt and subkey, then verifies the given password by hashing it with the stored salt. There are doubtless better / cleaner / more secure tweaks, so comments welcome.

// NodeJS implementation of crypto, I'm sure google's 
// cryptoJS would work equally well.
var crypto = require('crypto');

// The value stored in [dbo].[AspNetUsers].[PasswordHash]
var hashedPwd = "ADOEtXqGCnWCuuc5UOAVIvMVJWjANOA/LoVy0E4XCyUHIfJ7dfSY0Id+uJ20DTtG+A==";
var hashedPasswordBytes = new Buffer(hashedPwd, 'base64');

var hexChar = ["0", "1", "2", "3", "4", "5", "6", "7", "8", "9", "A", "B", "C", "D", "E", "F"];

var saltString = "";
var storedSubKeyString = "";

// build strings of octets for the salt and the stored key
for (var i = 1; i < hashedPasswordBytes.length; i++) {
    if (i > 0 && i <= 16) {
        saltString += hexChar[(hashedPasswordBytes[i] >> 4) & 0x0f] + hexChar[hashedPasswordBytes[i] & 0x0f]
    }
    if (i > 0 && i > 16) {
        storedSubKeyString += hexChar[(hashedPasswordBytes[i] >> 4) & 0x0f] + hexChar[hashedPasswordBytes[i] & 0x0f];
    }
}

// password provided by the user
var password = 'welcome1';

// TODO remove debug - logging passwords in prod is considered 
// tasteless for some odd reason
console.log('cleartext: ' + password);
console.log('saltString: ' + saltString);
console.log('storedSubKeyString: ' + storedSubKeyString);

// This is where the magic happens. 
// If you are doing your own hashing, you can (and maybe should)
// perform more iterations of applying the salt and perhaps
// use a stronger hash than sha1, but if you want it to work
// with the [as of 2015] Microsoft Identity framework, keep
// these settings.
var nodeCrypto = crypto.pbkdf2Sync(new Buffer(password), new Buffer(saltString, 'hex'), 1000, 256, 'sha1');

// get a hex string of the derived bytes
var derivedKeyOctets = nodeCrypto.toString('hex').toUpperCase();

console.log("hex of derived key octets: " + derivedKeyOctets);

// The first 64 bytes of the derived key should
// match the stored sub key
if (derivedKeyOctets.indexOf(storedSubKeyString) === 0) {
    console.info("passwords match!");
} else {
    console.warn("passwords DO NOT match!");
}

Here's another option which actually compares the bytes as opposed to converting to a string representation.

const crypto = require('crypto');

const password = 'Password123';
const storedHashString = 'J9IBFSw0U1EFsH/ysL+wak6wb8s=';
const storedSaltString = '2nX0MZPZlwiW8bYLlVrfjBYLBKM=';

const storedHashBytes = new Buffer.from(storedHashString, 'base64');
const storedSaltBytes = new Buffer.from(storedSaltString, 'base64');

crypto.pbkdf2(password, storedSaltBytes, 1000, 20, 'sha1',
  (err, calculatedHashBytes) => {
    const correct = calculatedHashBytes.equals(storedHashBytes);
    console.log('Password is ' + (correct ? 'correct 

本文标签: JavaScript How to generate Rfc2898DeriveBytes like CStack Overflow