@digibyte have u guys seen this and have u done anything for this fix yet to digishield?
Re: Regarding Auroracoin TW exploit
Today at 01:30:08 PM
Reply with quote #35
Quote from: Nite69 on April 01, 2014, 06:18:18 PM
Quote from: illodin on April 01, 2014, 04:10:21 PM
Every coin that uses KGW should apply the fix as well asap right?
Yes, I think so.
Quote from: illodin on April 01, 2014, 04:10:21 PM
Any idea if digishield or dgw (version2) are susceptible to the same exploit?
Have not yet dig too deeply on it, but:
Digishield: not the same, but possibly similar. KGW counts a lot of blocks to calculate adjustment, digishield only one, so it quite different. In KGW, attacker can 'jump over' the block in the future, which is not possible with only 1 block. Not yet sure, but there might be another TW exploit: Can you generate 2 blocks with dT less than 2xtarget so difficulty won't rise? If you generate block with dT 0, you get diff +25%, if I calculated correctly. next question: can you generate a block with -25% difficulty with less than 2*timespan? I think you get only -12.5% with 2 blocks time, so this hole is not open :-). But not sure if there are others.
Edit: this is dogecoin digishiel, don't know others..
DGW: possibly, quick peek reveals no fix, but it might be there, have to make deeper look at it to be sure.
I must make a re-estimate on digishield. It does allow back to median 11 blocks, ie one can go back in time.
1 Go to the future 12:00:00 -> get diff -50%
2 Go back 11:59:30 -> get diff +25%
Yes, there is attack vector.
Report to moderator
Bitcoin: 17gNvfoD2FDqTfESUxNEmTukGbGVAiJhXp
Litecoin: LhbDew4s9wbV8xeNkrdFcLK5u78APSGLrR
AuroraCoin: AXVoGgYtSVkPv96JLL7CiwcyVvPxXHXRK9
Math
Newbie
*
Online Online
Activity: 29
View Profile Personal Message (Online)
Trust: 0: -0 / +0(0)
Ignore
Re: Regarding Auroracoin TW exploit
Today at 02:24:59 PM
Reply with quote #36
Quote from: Nite69 on Today at 01:30:08 PM
Quote from: Nite69 on April 01, 2014, 06:18:18 PM
Quote from: illodin on April 01, 2014, 04:10:21 PM
Every coin that uses KGW should apply the fix as well asap right?
Yes, I think so.
Quote from: illodin on April 01, 2014, 04:10:21 PM
Any idea if digishield or dgw (version2) are susceptible to the same exploit?
Have not yet dig too deeply on it, but:
Digishield: not the same, but possibly similar. KGW counts a lot of blocks to calculate adjustment, digishield only one, so it quite different. In KGW, attacker can 'jump over' the block in the future, which is not possible with only 1 block. Not yet sure, but there might be another TW exploit: Can you generate 2 blocks with dT less than 2xtarget so difficulty won't rise? If you generate block with dT 0, you get diff +25%, if I calculated correctly. next question: can you generate a block with -25% difficulty with less than 2*timespan? I think you get only -12.5% with 2 blocks time, so this hole is not open :-). But not sure if there are others.
Edit: this is dogecoin digishiel, don't know others..
DGW: possibly, quick peek reveals no fix, but it might be there, have to make deeper look at it to be sure.
I must make a re-estimate on digishield. It does allow back to median 11 blocks, ie one can go back in time.
1 Go to the future 12:00:00 -> get diff -50%
2 Go back 11:59:30 -> get diff +25%
Yes, there is attack vector.
When the KGW exploit came to light I began to look over other one block readjustment algorithms. The DigiShield, as DigiByte currently has it implemented, does allow for an attack through time stamp manipulation. I have looked over Dogecoin's implementation, and it is more than likely vulnerable as well. It is hard to say however, because Dogecoin has a significantly higher hash rate than most other coins and has included a dampening on the adjustment. Again, this could be worked around through a determined attack on the largest pools and would probably just cause an attacker to have to work longer at building his chain. BCX stated that Auroracoin would have needed a hash rate of at least 25X his in order to prevent his attack. If that number holds true for the DigiShield, then any coin implementing the algorithm is at risk unless changes are made. The DigiShield algorithm allows for larger drops in difficulty than it does for larger increases, and this makes an attack easier.
I have yet to run through the numbers on the DGW. The first version was definitely vulnerable, but the second version, which allows for negatives to decrease the average time, may have plugged the hole.
Thank you for posting, we will look into this for sure.
Show Posts
