Re: Propose to Use madvise API on Runtime Loader

[Siddhesh Poyarekar <siddhesh at redhat dot com>]

On Fri, May 23, 2014 at 01:32:29PM +0800, lin zuojian wrote:
> On Fri, May 23, 2014 at 01:26:35AM -0400, Rich Felker wrote:
> > They are the same physical pages, so I don't see how you can drop one
> > without dropping the other.
> > 
>     When the virtual memory area specified is huge, hopefully there are
>     many pmd pages will get dropped. Even some of the pud will get
>     dropped.
> > RSS is not "accountable memory". It's largely meaningless, because it
> > includes memory that can be dropped at any time (without a need to
> > save it to swap). Defining a process's memory usage in a meaningful
> > way is difficult, but dirty pages, commit charge, and total RW mapping
> > sizes are more likely to be useful metrics than RSS is.
>     But it's one the most important element that OOM killer considers.

The OOM killer considers commit charge and MADV_DONTNEED does not
affect commit charge, so the call would be largely useless.

Siddhesh

Attachment: pgpsNL8ujebgV.pgp
Description: PGP signature