>> To elaborate on this: part of the idea behind mmapfd() is that it
>> gives the kernel the flexibility to make global-scope decisions about
>> where and how a file should be mapped. [...]
>
> The problem is that (at least in the B72 installation on the Oct 2007
> Summit's T2000 machine) the auto-MPSS stuff performs (IMO) poorly or
> does nothing

That's an implementation issue which is the focus of ongoing work.

mmapfd() is a new interface which is another step toward making the
kernel's automatic decisions better.  You will still be able to use the
memcntl(MC_HAT_ADVISE) (and mpss.so.1) interfaces with mmapfd(), just as
you can with mmap() today.

Blake

Reply via email to