When PVS suggests that something can be done it doesn't necessarily mean that it should be done.
There are not many allocations of these structures and since the library heavily relies on multythreading false sharing can be much bigger problem than a few wasted bytes. And IMHO even readability and ease of understanding structure logic are more important in this case.

Is there a known problem with memory consumption by these structures?
Did you measure how much memory this patch saves and how it impacts performance?

When PVS suggests that something can be done it doesn't necessarily mean that it should be done. There are not many allocations of these structures and since the library heavily relies on multythreading false sharing can be much bigger problem than a few wasted bytes. And IMHO even readability and ease of understanding structure logic are more important in this case.

I generally agree with the false sharing statement, but both of these structures are less than a cache line in size, so false sharing doesn't really go away either way, and this way we win a few bytes for free. Unless you have specific readability concerns with this, I would merge it.

Unless you have specific readability concerns with this, I would merge it.

Yeah, sure. I don't see much profit from this, but I also don't expect any harm.