Hi Steffan, at firsts thanks for taking a review!
On 22.10.2013 16:28, Steffan Karger wrote: > Thank you for coding and sharing! I marked your original mail for later > processing, just had not gotten to it yet. This is a cool feature, and > it has been proposed before on this list (see > http://thread.gmane.org/gmane.network.openvpn.devel/7384/focus=7395). Yepp, I really wondern when I discovered that it is currently missing. > Returning to your patch, I did not fully review the code, but it at > least has the same problem: you are using a non-constant time memcmp to > compare HMAC's, which would introduce a possible side-channel > vulnerability. The solution is simple: use memcmp_constant_time() from > crypto.c. No problem, I switched to the standard function because I thought it is not relevant in this scenario where the time "is added". > While taking a peak at your patch I tried to apply it on my local > codebase, but got conflicts on both the master and release/2.3 branches. > On what commit did you base the patch? It would be nice if it would > apply clean to at least the master branch. It should, but I will post a fixed patch tonight. Do you want it for master or release2.3? Kind regards, André
