All of lore.kernel.org
 help / color / mirror / Atom feed
From: Dave Hansen <haveblue@us.ibm.com>
To: Yasunori Goto <ygoto@us.fujitsu.com>
Cc: Linux Kernel ML <linux-kernel@vger.kernel.org>,
	Linux Hotplug Memory Support  <lhms-devel@lists.sourceforge.net>,
	Linux-Node-Hotplug <lhns-devel@lists.sourceforge.net>,
	linux-mm <linux-mm@kvack.org>,
	"BRADLEY CHRISTIANSEN [imap]" <bradc1@us.ibm.com>
Subject: Re: Merging Nonlinear and Numa style memory hotplug
Date: Thu, 24 Jun 2004 15:37:02 -0700	[thread overview]
Message-ID: <1088116621.3918.1060.camel@nighthawk> (raw)
In-Reply-To: <20040624135838.F009.YGOTO@us.fujitsu.com>

On Thu, 2004-06-24 at 15:19, Yasunori Goto wrote:
> BTW, I have a question about nonlinear patch.
> It is about difference between phys_section[] and mem_section[]
> I suppose that phys_section[] looks like no-meaning now.
> If it isn't necessary, __va() and __pa() translation can be more simple.
> What is the purpose of phys_section[]. Is it for ppc64?

This is the fun (read: confusing) part of nonlinear.

The mem_section[] array is where the pointer to the mem_map for the
section is stored, obviously.  It's indexed virtually, so that something
at a virtual address is in section number (address >> SECTION_SHIFT). 
So, that makes it easy to go from a virtual address to a 'struct page'
inside of the mem_map[].

But, given a physical address (or a pfn for that matter), you sometimes
also need to get to a 'struct page'.  It is for that reason that we have
the phys_section[] array.  Each entry in the phys_section[] points back
to a mem_section[], which then contains the mem_map[].

pfn_to_page(unsigned long pfn)
{
       return
&mem_section[phys_section[pfn_to_section(pfn)]].mem_map[section_offset_pfn(pfn)];
}

pfn_to_section(pfn) does a (pfn >> (SECTION_SHIFT - PAGE_SHIFT)), then
uses that section number to index into the phys_section[] array, which
gives an index into the mem_section[] array, from which you can get the
'struct page'.  


-- Dave


WARNING: multiple messages have this Message-ID (diff)
From: Dave Hansen <haveblue@us.ibm.com>
To: Yasunori Goto <ygoto@us.fujitsu.com>
Cc: Linux Kernel ML <linux-kernel@vger.kernel.org>,
	Linux Hotplug Memory Support <lhms-devel@lists.sourceforge.net>,
	Linux-Node-Hotplug <lhns-devel@lists.sourceforge.net>,
	linux-mm <linux-mm@kvack.org>,
	"BRADLEY CHRISTIANSEN [imap]" <bradc1@us.ibm.com>
Subject: Re: Merging Nonlinear and Numa style memory hotplug
Date: Thu, 24 Jun 2004 15:37:02 -0700	[thread overview]
Message-ID: <1088116621.3918.1060.camel@nighthawk> (raw)
In-Reply-To: <20040624135838.F009.YGOTO@us.fujitsu.com>

On Thu, 2004-06-24 at 15:19, Yasunori Goto wrote:
> BTW, I have a question about nonlinear patch.
> It is about difference between phys_section[] and mem_section[]
> I suppose that phys_section[] looks like no-meaning now.
> If it isn't necessary, __va() and __pa() translation can be more simple.
> What is the purpose of phys_section[]. Is it for ppc64?

This is the fun (read: confusing) part of nonlinear.

The mem_section[] array is where the pointer to the mem_map for the
section is stored, obviously.  It's indexed virtually, so that something
at a virtual address is in section number (address >> SECTION_SHIFT). 
So, that makes it easy to go from a virtual address to a 'struct page'
inside of the mem_map[].

But, given a physical address (or a pfn for that matter), you sometimes
also need to get to a 'struct page'.  It is for that reason that we have
the phys_section[] array.  Each entry in the phys_section[] points back
to a mem_section[], which then contains the mem_map[].

pfn_to_page(unsigned long pfn)
{
       return
&mem_section[phys_section[pfn_to_section(pfn)]].mem_map[section_offset_pfn(pfn)];
}

pfn_to_section(pfn) does a (pfn >> (SECTION_SHIFT - PAGE_SHIFT)), then
uses that section number to index into the phys_section[] array, which
gives an index into the mem_section[] array, from which you can get the
'struct page'.  


-- Dave

--
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>

  reply	other threads:[~2004-06-24 22:40 UTC|newest]

Thread overview: 32+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2004-06-22 19:00 Merging Nonlinear and Numa style memory hotplug Yasunori Goto
2004-06-22 19:00 ` Yasunori Goto
2004-06-23 22:32 ` [Lhns-devel] " Dave Hansen
2004-06-23 22:32   ` Dave Hansen
2004-06-24  3:04   ` Yasunori Goto
2004-06-24  3:04     ` Yasunori Goto
2004-06-24  3:26     ` Dave Hansen
2004-06-24  3:26       ` Dave Hansen
2004-06-24 13:28     ` [Lhms-devel] " Dave Hansen
2004-06-24 13:28       ` Dave Hansen
2004-06-24 22:19       ` Yasunori Goto
2004-06-24 22:19         ` Yasunori Goto
2004-06-24 22:37         ` Dave Hansen [this message]
2004-06-24 22:37           ` Dave Hansen
2004-06-25  3:11           ` [Lhms-devel] " Yasunori Goto
2004-06-25  3:11             ` Yasunori Goto
2004-06-25  3:19             ` Dave Hansen
2004-06-25  3:19               ` Dave Hansen
2004-06-25 18:48               ` Yasunori Goto
2004-06-25 18:48                 ` Yasunori Goto
2004-06-25 18:59                 ` Dave Hansen
2004-06-25 18:59                   ` Dave Hansen
2004-06-25 20:45                   ` Yasunori Goto
2004-06-25 20:45                     ` Yasunori Goto
2004-06-25 20:49                     ` Dave Hansen
2004-06-25 20:49                       ` Dave Hansen
2004-06-25 20:54                     ` Dave Hansen
2004-06-25 20:54                       ` Dave Hansen
2004-06-25  4:49         ` [Lhms-devel] Re: [Lhns-devel] " Shai Fultheim
2004-06-25  4:49           ` Shai Fultheim
2004-06-25 15:16           ` Dave Hansen
2004-06-25 15:16             ` Dave Hansen

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=1088116621.3918.1060.camel@nighthawk \
    --to=haveblue@us.ibm.com \
    --cc=bradc1@us.ibm.com \
    --cc=lhms-devel@lists.sourceforge.net \
    --cc=lhns-devel@lists.sourceforge.net \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=ygoto@us.fujitsu.com \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.