xor eax,eax scans, to an x86 assembly programmer, as "eax := 0". It's idiomatic for clearing a register. There ought not be any expectation from an assembly programmer that such a frequent operation would take the same amount of time as xor eax,ebx (in particular, xor eax,eax doesn't depend on the value of eax, so it doesn't need to stall waiting for its value to be calculated). But if xor eax,eax took a different amount of time depending on whether eax was equal to 42 or not, that would be a different matter.
Why wouldn't that approach work?
There are much better ways to achieve this, by always computing both sides of a (logical) branch or using computation instead of branching or table lookups. You likely have to write this in assembly to be sure that the compiler doesn't "optimize" any of your tricks back into branches. Some of this is hard in older crypto algorithms (AES was designed to use table lookups in software), but newer crypto algorithms are much more amenable to safe implementation.
Also, even assuming you can always finish early, that plan won't prevent, for example, measuring the delay of other responses. (As your process sleeping will have an impact on how fast other responses are sent. Due to cache pressure, in particular. But also due to CPU throttling, scheduling, etc, etc.
You measure page faults in terms of milliseconds, I was not suggesting you even out the cryptographic timings to the tune of milliseconds, especially when you consider most cryptographic functions are setup specifically to take longer than that.
There is a delta that you can come up with such that things like page faults fall within it and it would still have acceptable performance. All but the most pounded of servers would be able to fall within it with no problem whatsoever.
And most of the cryptographic functions we're talking about here are things for webservers, where low performance leaves you open to easy (D)DOSing.
So why not take another crack at explaining why that wouldn't solve the problem? And don't complain to me about DDOS, people specifically code cryptographic functions to take longer than they absolutely have to. If they're not worried about DDOS, neither are we for the purposes of this conversation.
AIUI tfa's point is that you can't assume that
xor eax,eax
and
xor eax,ebx
take the same amount of time, because "x86 is a high level language". Similarly for his other examples. This, he claims, makes it difficult to write code that resists timing attacks, if you ever want to have a branch that does nothing. Thus the discussion of cmov.