linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
From: Dave Kleikamp <shaggy@linux.vnet.ibm.com>
To: linux-fsdevel <linux-fsdevel@vger.kernel.org>,
	linux-mm <linux-mm@kvack.org>
Subject: [RFC:PATCH 00/09] VM File Tails
Date: Thu, 8 Nov 2007 12:47:11 -0700	[thread overview]
Message-ID: <20071108194709.17862.16713.sendpatchset@norville.austin.ibm.com> (raw)

This is the latest version of my "VM File Tails" work.  The idea is to
store tails of files that are smaller than the base page size in kmalloc'ed
memory, allowing more efficient use of memory.  This is especially important
when the base page size is large, such as 64 KB on powerpc.

So far, my testing hasn't resulted in any performance gains.  The workloads
prompting this work are more involved, so more testing is needed.  Right
now, I don't have a case for inclusion of these patches, but there was
interest in the community, so here they are.

These patches are built against 2.6.24-rc2.

I had posted some patches earlier that were much more complex, and
introduced dummy pages into the page cache to account for the tails.  I
have abandoned that approach, and have arrived at a much simpler patch set.

The idea is to attach a buffer to the address space (page->mapping) to hold
the tail.  Whenever the page corresponding to the tail is requested, a new
page is allocated and the tail is unpacked to that page.  At some point,
pages that are eligible to be packed are copied into kmalloced buffers and
attached to the address space.  The eligible pages must be up-to-date, clean,
unmapped, not waiting for I/O, etc.

Since the last time I posted:
- I optimized generic_file_aio_read to copy data directly from the tail,
  rather than unpacking the tail and copying from the page cache
- Luiz Capitulino contributed a patch to add statistics in
  /sys/kernel/debug/vm_tail/

My To-Do list includes:
- Investigate more aggressive places to pack tails.  It's currently only
  being done in shrink_active_list()
- benchmark!

Comments are appreciated.

The patches can also be downloaded from:
ftp://kernel.org/pub/linux/kernel/people/shaggy/vm_file_tails/

Thanks,
Shaggy

--
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:"dont@kvack.org"> email@kvack.org </a>

             reply	other threads:[~2007-11-08 19:47 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2007-11-08 19:47 Dave Kleikamp [this message]
2007-11-08 19:47 ` [RFC:PATCH 01/09] Add tail to address space Dave Kleikamp
2007-11-08 19:47 ` [RFC:PATCH 02/09] Core function for packing, unpacking, and freeing file tails Dave Kleikamp
2007-11-08 19:47 ` [RFC:PATCH 03/09] Release tail when inode is freed Dave Kleikamp
2007-11-08 19:47 ` [RFC:PATCH 04/09] Unpack or remove file tail when inode is resized Dave Kleikamp
2007-11-08 19:47 ` [RFC:PATCH 05/09] find_get_page() and find_lock_page() need to unpack the tail Dave Kleikamp
2007-11-08 19:47 ` [RFC:PATCH 06/09] For readahead, leave data in tail Dave Kleikamp
2007-11-08 19:47 ` [RFC:PATCH 07/09] shrink_active_list: pack file tails rather than move to inactive list Dave Kleikamp
2007-11-08 19:48 ` [RFC:PATCH 08/09] generic_file_aio_read can read directly from the tail. No need to unpack Dave Kleikamp
2007-11-08 19:48 ` [RFC:PATCH 09/09] VM tail statistics support Dave Kleikamp, Luiz Fernando N. Capitulino

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=20071108194709.17862.16713.sendpatchset@norville.austin.ibm.com \
    --to=shaggy@linux.vnet.ibm.com \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    /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