linux-kernel - Consensus inquiry regarding POSIX_FADV

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [day] [month] [year] [list]

Message-ID: <BANLkTikMLvgtAJp8Vex2u-Fcv5YGMvWBgg@mail.gmail.com>
Date:	Thu, 9 Jun 2011 15:47:29 -0400
From:	"Jonah H. Harris" <jonah.harris@...il.com>
To:	linux-kernel@...r.kernel.org
Subject: Consensus inquiry regarding POSIX_FADV_NOREUSE

Hello everyone,

I'm thinking about implementing the POSIX_FADV_NOREUSE advice option
of fadvise (mm/fadvise.c). While POSIX_FADV_DONTNEED works for many
use cases, it would be nice to forcibly evict pages as quickly as
possible to avoid having a single application pollute the cache --
especially when cache is used more heavily by other applications.

I have some ideas in regard to implementation, but as of yet I don't
have a concrete proposal. Instead, I wanted to see whether anyone is
averse to the idea before I start researching it more completely. To
that end, I've reviewed a large number of previous discussions and am
seeking clarification in regard to the current consensus of the
following:

  For developers who don't want their pages cached, should they use
O_DIRECT or fadvise?

In the past, I've used O_DIRECT quite successfully. But as it requires
dealing with alignment, it's not as simple to use or as general-case
as fadvise.

Thoughts?

--
Jonah H. Harris, VP of Architecture
myYearbook.com
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/