madvise() vs posix_fadvise()

[John Baldwin]

On Monday, March 24, 2014 8:03:04 am Dmitry Sivachenko wrote:
> 
> On 21 марта 2014 г., at 20:27, Trond Endrestøl 
<Trond.Endrestol at fagskolen.gjovik.no> wrote:
> 
> > On Fri, 21 Mar 2014 18:56+0400, Dmitry Sivachenko wrote:
> > 
> >> Hello!
> >> 
> >> I have a program which uses large data files (read-only, via mmap()).
> >> 
> >> These machines have a bit more RAM that these files occupy, so it is 
> >> possible to have all these data in memory.
> >> 
> >> What techniques should I use to promote this data not to be purged 
> >> from RAM:
> >> 
> >> -- madvise(MADV_WILLNEED)
> >> -- posix_fadvise(POSIX_FADV_WILLNEED)
> >> -- both?
> > 
> > Although a bit dangerous, mlock(2) might be your ticket. That system 
> > call prevents your memory region from being swapped/paged away from 
> > physical memory.
> > 
> 
> 
> I know about mlock(2), it is a bit overkill.
> Can someone please explain the difference between madvise(MADV_WILLNEED) and 
posix_fadvise(POSIX_FADV_WILLNEED)?

Right now FADV_WILLNEED is a nop.  (I have some patches to implement it for
UFS.)  I can't recall off the top of my head if MADV_WILLNEED is also a nop.
However, if both are fully implemented they should be similar in terms of
requesting async read-ahead.  MADV_WILLNEED might also conceivably
pre-create PTEs while FADV_WILLNEED can be used on a file that isn't
mapped but is accessed via read(2).

-- 
John Baldwin