From: Ray Bryant <raybry@engr.sgi.com>
To: Paul Jackson <pj@sgi.com>
Cc: hch@infradead.org, ak@suse.de, raybry@sgi.com,
taka@valinux.co.jp, marcelo.tosatti@cyclades.com,
haveblue@us.ibm.com, linux-mm@kvack.org, nathans@sgi.com,
raybry@austin.rr.com, lhms-devel@lists.sourceforge.net,
jes@wildopensource.com
Subject: Re: [Lhms-devel] Re: [PATCH 2.6.12-rc3 1/8] mm: manual page migration-rc2 -- xfs-extended-attributes-rc2.patch
Date: Wed, 18 May 2005 09:49:07 -0500 [thread overview]
Message-ID: <428B55E3.1000201@engr.sgi.com> (raw)
In-Reply-To: <20050517232007.018d52ab.pj@sgi.com>
Paul Jackson wrote:
> Ray wrote:
>
>>If one were to use madvise() or mbind() to apply the migration
>>policy flags ... then ... it's necessary to reach down to a common subobject,
>>(such as the file struct, address space struct, or inode) and mark
>>that.
>
>
> As I wrote Ray offline earlier today (before noticing this thread), I
> suggested considering adding another flag to fcntl(), not madvise/mbind,
> since fcntl() is sometimes used to make changes to the underlying
> in-core inode.
>
> My rough idea was to have a S_SKIPMIGRATE flag in the in-core
> inode->i_flag's, defaulting to off, which could be set and gotten via
> fcntl on any file descriptor open on that inode. The value set would
> persist so long as some file held that dentry->inode open, or until
> changed again. Just before invoking the call to migrate a task, user
> code could examine each named file mapped into the task, and by opening
> each said file and performing the appropriate fcntl(), mark whether or
> not pages from that mapped file should be migrated.
>
Actually, we need two flags: S_SKIPMIGRATE, and S_MIGRATE_NON_SHARED.
I personally prefer using mbind() to set these attributes in the
address space object that maps a particular virtual address range,
I guess I don't see the advantage of invoking file system operations
to describe how memory should be dealt with in a migration operation.
So I'd prefer to keep all of the operations in the memory space, hence
mbind().
This makes sense to me since the migration operation is a policy issue,
and mbind() is in the business of setting memory policy.
--
Best Regards,
Ray
-----------------------------------------------
Ray Bryant
512-453-9679 (work) 512-507-7807 (cell)
raybry@sgi.com raybry@austin.rr.com
The box said: "Requires Windows 98 or better",
so I installed Linux.
-----------------------------------------------
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"aart@kvack.org"> aart@kvack.org </a>
next prev parent reply other threads:[~2005-05-18 14:49 UTC|newest]
Thread overview: 46+ messages / expand[flat|nested] mbox.gz Atom feed top
2005-05-11 4:37 [PATCH 2.6.12-rc3 0/8] mm: manual page migration-rc2 -- overview Ray Bryant
2005-05-11 4:38 ` [PATCH 2.6.12-rc3 1/8] mm: manual page migration-rc2 -- xfs-extended-attributes-rc2.patch Ray Bryant
2005-05-11 7:15 ` Christoph Hellwig
2005-05-11 12:10 ` [Lhms-devel] " Ray Bryant
2005-05-11 12:59 ` Andi Kleen
2005-05-11 18:43 ` Ray Bryant
2005-05-11 19:32 ` Andi Kleen
2005-05-11 20:00 ` Christoph Hellwig
2005-05-11 22:04 ` Ray Bryant
2005-05-12 10:45 ` Christoph Hellwig
2005-05-17 4:22 ` Ray Bryant
2005-05-18 6:20 ` Paul Jackson
2005-05-18 14:49 ` Ray Bryant [this message]
2005-05-20 22:26 ` Ray Bryant
2005-05-23 17:50 ` Steve Longerbeam
2005-05-24 4:53 ` Ray Bryant
2005-05-24 20:59 ` Christoph Lameter
2005-05-24 21:04 ` Martin J. Bligh
2005-05-25 6:42 ` Ray Bryant
2005-05-28 8:40 ` Christoph Hellwig
2005-05-28 16:12 ` Ray Bryant
2005-05-11 19:50 ` Christoph Hellwig
2005-05-11 21:30 ` Ray Bryant
2005-05-12 9:55 ` Christoph Hellwig
2005-05-12 15:47 ` Ray Bryant
2005-05-11 4:38 ` [PATCH 2.6.12-rc3 2/8] mm: manual page migration-rc2 -- xfs-migrate-page-rc2.patch Ray Bryant
2005-05-11 4:38 ` [PATCH 2.6.12-rc3 3/8] mm: manual page migration-rc2 -- add-node_map-arg-to-try_to_migrate_pages-rc2.patch Ray Bryant
2005-05-11 4:38 ` [PATCH 2.6.12-rc3 4/8] mm: manual page migration-rc2 -- add-sys_migrate_pages-rc2.patch Ray Bryant
2005-05-11 8:24 ` Christoph Hellwig
2005-05-18 19:07 ` Ray Bryant
2005-05-28 9:14 ` Christoph Hellwig
2005-05-28 15:53 ` Ray Bryant
2005-05-11 13:23 ` Hirokazu Takahashi
2005-05-11 13:26 ` Hirokazu Takahashi
2005-05-11 14:06 ` Ray Bryant
2005-05-12 6:41 ` Hirokazu Takahashi
2005-05-12 16:41 ` Ray Bryant
2005-05-12 23:50 ` Hirokazu Takahashi
2005-05-13 9:59 ` [Lhms-devel] " Ray Bryant
2005-05-11 4:38 ` [PATCH 2.6.12-rc3 5/8] mm: manual page migration-rc2 -- sys_migrate_pages-xattr-support-rc2.patch Ray Bryant
2005-05-11 4:38 ` [PATCH 2.6.12-rc3 6/8] mm: manual page migration-rc2 -- sys_migrate_pages-mempolicy-migration-rc2.patch Ray Bryant
2005-05-11 4:38 ` [PATCH 2.6.12-rc3 7/8] mm: manual page migration-rc2 -- sys_migrate_pages-cpuset-support-rc2.patch Ray Bryant
2005-05-11 12:37 ` Paul Jackson
2005-05-11 14:20 ` Ray Bryant
2005-05-11 18:55 ` [Lhms-devel] " Paul Jackson
2005-05-11 4:38 ` [PATCH 2.6.12-rc3 8/8] mm: manual page migration-rc2 -- sys_migrate_pages-permissions-check-rc2.patch Ray Bryant
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=428B55E3.1000201@engr.sgi.com \
--to=raybry@engr.sgi.com \
--cc=ak@suse.de \
--cc=haveblue@us.ibm.com \
--cc=hch@infradead.org \
--cc=jes@wildopensource.com \
--cc=lhms-devel@lists.sourceforge.net \
--cc=linux-mm@kvack.org \
--cc=marcelo.tosatti@cyclades.com \
--cc=nathans@sgi.com \
--cc=pj@sgi.com \
--cc=raybry@austin.rr.com \
--cc=raybry@sgi.com \
--cc=taka@valinux.co.jp \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox