Messages in this thread | | | Date | Thu, 10 Jul 2008 13:53:56 -0400 | From | "H. Peter Anvin" <> | Subject | Re: [RFC 00/15] x86_64: Optimize percpu accesses |
| |
Christoph Lameter wrote: > > There will be no additional overhead since the memory already mapped 1-1 using 2MB TLBs and we want to use the same for the percpu areas. This is similar to the vmemmap solution. >
THAT sounds strange. If you're using dedicated virtual maps (which is what you're responding to here) then you will *always* have additional TLB pressure. Furthermore, if you use 2 MB pages, you:
a) can only allocate full 2 MB pages, which is expensive for the static users and difficult for the dynamic users;
b) increase pressure in the relatively small 2 MB TLB.
-hpa
| |