From: Andy Whitcroft <apw@shadowen.org>
To: Michael Ellerman <michael@ellerman.id.au>
Cc: Andy Whitcroft <apw@shadowen.org>, Andrew Morton <akpm@osdl.org>,
linux-kernel@vger.kernel.org, haveblue@us.ibm.com,
kravetz@us.ibm.com
Subject: [PATCH] sparsemem record nid during memory present
Date: Wed, 10 May 2006 11:51:02 +0100 [thread overview]
Message-ID: <20060510105102.GA9533@shadowen.org> (raw)
In-Reply-To: 1147241173.8091.21.camel@localhost.localdomain
Ok, here is the updated version. Better commentry on what we are doing
and how long the data is kept (suggested by Dave Hansen). Also added a
SECTION_NID_SHIFT to better describe its use. This showed a small
buglett as we were shifting by 4 (1<<2) instead of 2; now fixed.
-apw
=== 8< ===
sparsemem record nid during memory present
Record the node id as we mark sections for instantiation. Use this
nid during instantiation to direct allocations.
Signed-off-by: Andy Whitcroft <apw@shadowen.org>
---
include/linux/mmzone.h | 5 +++++
mm/sparse.c | 22 ++++++++++++++++++++--
2 files changed, 25 insertions(+), 2 deletions(-)
diff -upN reference/include/linux/mmzone.h current/include/linux/mmzone.h
--- reference/include/linux/mmzone.h
+++ current/include/linux/mmzone.h
@@ -508,6 +508,10 @@ struct mem_section {
* pages. However, it is stored with some other magic.
* (see sparse.c::sparse_init_one_section())
*
+ * Additionally during early boot we encode node id of
+ * the location of the section here to guide allocation.
+ * (see sparse.c::memory_present())
+ *
* Making it a UL at least makes someone do a cast
* before using it wrong.
*/
@@ -547,6 +551,7 @@ extern int __section_nr(struct mem_secti
#define SECTION_HAS_MEM_MAP (1UL<<1)
#define SECTION_MAP_LAST_BIT (1UL<<2)
#define SECTION_MAP_MASK (~(SECTION_MAP_LAST_BIT-1))
+#define SECTION_NID_SHIFT 2
static inline struct page *__section_mem_map_addr(struct mem_section *section)
{
diff -upN reference/mm/sparse.c current/mm/sparse.c
--- reference/mm/sparse.c
+++ current/mm/sparse.c
@@ -102,6 +102,22 @@ int __section_nr(struct mem_section* ms)
return (root_nr * SECTIONS_PER_ROOT) + (ms - root);
}
+/*
+ * During early boot, before section_mem_map is used for an actual
+ * mem_map, we use section_mem_map to store the section's NUMA
+ * node. This keeps us from having to use another data structure. The
+ * node information is cleared just before we store the real mem_map.
+ */
+static inline unsigned long sparse_encode_early_nid(int nid)
+{
+ return (nid << SECTION_NID_SHIFT);
+}
+
+static inline int sparse_early_nid(struct mem_section *section)
+{
+ return (section->section_mem_map >> SECTION_NID_SHIFT);
+}
+
/* Record a memory area against a node. */
void memory_present(int nid, unsigned long start, unsigned long end)
{
@@ -116,7 +132,8 @@ void memory_present(int nid, unsigned lo
ms = __nr_to_section(section);
if (!ms->section_mem_map)
- ms->section_mem_map = SECTION_MARKED_PRESENT;
+ ms->section_mem_map = sparse_encode_early_nid(nid) |
+ SECTION_MARKED_PRESENT;
}
}
@@ -167,6 +184,7 @@ static int sparse_init_one_section(struc
if (!valid_section(ms))
return -EINVAL;
+ ms->section_mem_map &= ~SECTION_MAP_MASK;
ms->section_mem_map |= sparse_encode_mem_map(mem_map, pnum);
return 1;
@@ -175,8 +193,8 @@ static int sparse_init_one_section(struc
static struct page *sparse_early_mem_map_alloc(unsigned long pnum)
{
struct page *map;
- int nid = early_pfn_to_nid(section_nr_to_pfn(pnum));
struct mem_section *ms = __nr_to_section(pnum);
+ int nid = sparse_early_nid(ms);
map = alloc_remap(nid, sizeof(struct page) * PAGES_PER_SECTION);
if (map)
next prev parent reply other threads:[~2006-05-10 10:51 UTC|newest]
Thread overview: 12+ messages / expand[flat|nested] mbox.gz Atom feed top
2006-05-09 7:03 [PATCH] SPARSEMEM + NUMA can't handle unaligned memory regions? Michael Ellerman
2006-05-09 8:32 ` Andy Whitcroft
2006-05-09 9:11 ` Michael Ellerman
2006-05-09 10:24 ` Andy Whitcroft
2006-05-09 13:34 ` Andy Whitcroft
2006-05-09 14:28 ` Andy Whitcroft
2006-05-09 16:26 ` Dave Hansen
2006-05-09 16:31 ` Andy Whitcroft
2006-05-10 6:06 ` Michael Ellerman
2006-05-10 10:51 ` Andy Whitcroft [this message]
2006-05-09 16:05 ` mike kravetz
2006-05-09 16:14 ` Andy Whitcroft
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=20060510105102.GA9533@shadowen.org \
--to=apw@shadowen.org \
--cc=akpm@osdl.org \
--cc=haveblue@us.ibm.com \
--cc=kravetz@us.ibm.com \
--cc=linux-kernel@vger.kernel.org \
--cc=michael@ellerman.id.au \
/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.