Closed singe closed 1 year ago
I don't understand why the first two chunks are safe to remove but the other changes look obviously beneficial to me.
I suspect the randomization has to do with our own key generation. But I'll echo what Zero has already mentioned. And I'm not quite sure on what savings we get from it anyways. I'm definitely willing to accept the changes modulo those changes to the randomization. There is a clear win with the other changes
Should I resubmit without the randomisation changes?
Yes, do that. Still not sold on the benefits of the randomization ATM and if that is something we should be really maintaining.
From @wytshadow "blunderbuss’s version was averaging about 17 words/sec. @singe your modified version does around 20 words/sec. this was tested from same internal laptop hardware against the same AP."
When I get some time, I'll set up a test harness and proper profiler to figure out what else can be ripped out and quantify it.
Hi
This patch will speed up the SAE and notify of success/failure earlier.
SAE tried to prevent side channel attacks by performing the hunting and pecking algorithm in a constant time, lately of 40 rounds. Reducing this to 1 saves time on the brute.
I started playing with other places like reducing randomisation initialisations, there are numerous other places speedups can be gained.
This will also notify of a success or failure sooner than when it would normally hit notify, I think.