From: Christoph Lameter <cl@linux-foundation.org>
To: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Mel Gorman <mel@csn.ul.ie>,
Linux Memory Management List <linux-mm@kvack.org>,
Pekka Enberg <penberg@cs.helsinki.fi>,
Rik van Riel <riel@redhat.com>,
KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>,
Johannes Weiner <hannes@cmpxchg.org>,
Nick Piggin <npiggin@suse.de>,
Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
Lin Ming <ming.m.lin@intel.com>,
Zhang Yanmin <yanmin_zhang@linux.intel.com>,
Peter Zijlstra <peterz@infradead.org>
Subject: Re: [PATCH 24/27] Convert gfp_zone() to use a table of precalculated values
Date: Thu, 19 Mar 2009 12:53:34 -0400 (EDT) [thread overview]
Message-ID: <alpine.DEB.1.10.0903191251310.24152@qirst.com> (raw)
In-Reply-To: <alpine.DEB.1.10.0903191105090.8100@qirst.com>
On Thu, 19 Mar 2009, Christoph Lameter wrote:
> It would work if we could check for BAD_ZONE with a VM_BUG_ON or a
> BUILD_BUG_ON. If I get some time I will look into this.
Here is such a patch. Boots on my machine and working with that kernel
now. There is a slight gcc problem in that the table is likely repeated
for each compilation unit. Anyone know how to fix that?
Subject: Use a table lookup for gfp_zone and check for errors in flags passed to the page allocator
Use a table to lookup the zone to use given gfp_flags using gfp_zone().
This simplifies the code in gfp_zone() and also keeps the ability of the compiler to
use constant folding to get rid of gfp_zone processing.
One problem with this patch is that we define a static const array in gfp.h. This results
in every compilation unit to reserve its own space for the array. There must be some
trick to get the compiler to allocate this only once. The contents of the array
must be described in the header file otherwise the compiler will not be able to
determine the value of a lookup in the table.
Signed-off-by: Christoph Lameter <cl@linux-foundation.org>
Index: linux-2.6/include/linux/gfp.h
===================================================================
--- linux-2.6.orig/include/linux/gfp.h 2009-03-19 11:43:32.000000000 -0500
+++ linux-2.6/include/linux/gfp.h 2009-03-19 11:48:38.000000000 -0500
@@ -19,7 +19,8 @@
#define __GFP_DMA ((__force gfp_t)0x01u)
#define __GFP_HIGHMEM ((__force gfp_t)0x02u)
#define __GFP_DMA32 ((__force gfp_t)0x04u)
-
+#define __GFP_MOVABLE ((__force gfp_t)0x08u) /* Page is movable */
+#define GFP_ZONEMASK ((__force gfp_t)0x0fu)
/*
* Action modifiers - doesn't change the zoning
*
@@ -49,7 +50,6 @@
#define __GFP_HARDWALL ((__force gfp_t)0x20000u) /* Enforce hardwall cpuset memory allocs */
#define __GFP_THISNODE ((__force gfp_t)0x40000u)/* No fallback, no policies */
#define __GFP_RECLAIMABLE ((__force gfp_t)0x80000u) /* Page is reclaimable */
-#define __GFP_MOVABLE ((__force gfp_t)0x100000u) /* Page is movable */
#define __GFP_BITS_SHIFT 21 /* Room for 21 __GFP_FOO bits */
#define __GFP_BITS_MASK ((__force gfp_t)((1 << __GFP_BITS_SHIFT) - 1))
@@ -111,24 +111,56 @@
((gfp_flags & __GFP_RECLAIMABLE) != 0);
}
-static inline enum zone_type gfp_zone(gfp_t flags)
-{
+#ifdef CONFIG_ZONE_HIGHMEM
+#define OPT_ZONE_HIGHMEM ZONE_HIGHMEM
+#else
+#define OPT_ZONE_HIGHMEM ZONE_NORMAL
+#endif
+
#ifdef CONFIG_ZONE_DMA
- if (flags & __GFP_DMA)
- return ZONE_DMA;
+#define OPT_ZONE_DMA ZONE_DMA
+#else
+#define OPT_ZONE_DMA ZONE_NORMAL
#endif
+
#ifdef CONFIG_ZONE_DMA32
- if (flags & __GFP_DMA32)
- return ZONE_DMA32;
+#define OPT_ZONE_DMA32 ZONE_DMA32
+#else
+#define OPT_ZONE_DMA32 OPT_ZONE_DMA
#endif
- if ((flags & (__GFP_HIGHMEM | __GFP_MOVABLE)) ==
- (__GFP_HIGHMEM | __GFP_MOVABLE))
- return ZONE_MOVABLE;
-#ifdef CONFIG_HIGHMEM
- if (flags & __GFP_HIGHMEM)
- return ZONE_HIGHMEM;
+
+#define BAD_ZONE MAX_NR_ZONES
+
+static const enum zone_type gfp_zone_table[GFP_ZONEMASK + 1] = {
+ ZONE_NORMAL, /* 00 No flags set */
+ OPT_ZONE_DMA, /* 01 GFP_DMA */
+ OPT_ZONE_HIGHMEM, /* 02 GFP_HIGHMEM */
+ BAD_ZONE, /* 03 GFP_HIGHMEM GFP_DMA */
+ OPT_ZONE_DMA32, /* 04 GFP_DMA32 */
+ BAD_ZONE, /* 05 GFP_DMA32 GFP_DMA */
+ BAD_ZONE, /* 06 GFP_DMA32 GFP_HIGHMEM */
+ BAD_ZONE, /* 07 GFP_DMA32 GFP_HIGHMEM GFP_DMA */
+ ZONE_NORMAL, /* 08 ZONE_MOVABLE */
+ OPT_ZONE_DMA, /* 09 MOVABLE + DMA */
+ ZONE_MOVABLE, /* 0A MOVABLE + HIGHMEM */
+ BAD_ZONE, /* 0B MOVABLE + HIGHMEM + DMA */
+ OPT_ZONE_DMA32, /* 0C MOVABLE + DMA32 */
+ BAD_ZONE, /* 0D MOVABLE + DMA32 + DMA */
+ BAD_ZONE, /* 0E MOVABLE + DMA32 + HIGHMEM */
+ BAD_ZONE /* 0F MOVABLE + DMA32 + HIGHMEM + DMA */
+};
+
+static inline enum zone_type gfp_zone(gfp_t flags)
+{
+ enum zone_type zone = gfp_zone_table[flags & 0xf];
+
+ if (__builtin_constant_p(zone))
+ BUILD_BUG_ON(zone == BAD_ZONE);
+#ifdef CONFIG_DEBUG_VM
+ else
+ BUG_ON(zone == BAD_ZONE);
#endif
- return ZONE_NORMAL;
+ return zone;
}
/*
Index: linux-2.6/include/linux/mmzone.h
===================================================================
--- linux-2.6.orig/include/linux/mmzone.h 2009-03-19 11:47:00.000000000 -0500
+++ linux-2.6/include/linux/mmzone.h 2009-03-19 11:47:54.000000000 -0500
@@ -240,7 +240,8 @@
ZONE_HIGHMEM,
#endif
ZONE_MOVABLE,
- __MAX_NR_ZONES
+ __MAX_NR_ZONES,
+ BAD_ZONE
};
#ifndef __GENERATING_BOUNDS_H
--
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>
next prev parent reply other threads:[~2009-03-19 16:56 UTC|newest]
Thread overview: 44+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-03-16 17:53 [PATCH 00/26] Cleanup and optimise the page allocator V4 Mel Gorman
2009-03-16 17:53 ` [PATCH 01/27] Replace __alloc_pages_internal() with __alloc_pages_nodemask() Mel Gorman
2009-03-16 17:53 ` [PATCH 02/27] Do not sanity check order in the fast path Mel Gorman
2009-03-16 17:53 ` [PATCH 03/27] Do not check NUMA node ID when the caller knows the node is valid Mel Gorman
2009-03-16 17:53 ` [PATCH 04/27] Check only once if the zonelist is suitable for the allocation Mel Gorman
2009-03-16 17:53 ` [PATCH 05/27] Break up the allocator entry point into fast and slow paths Mel Gorman
2009-03-16 19:30 ` Christoph Lameter
2009-03-16 17:53 ` [PATCH 06/27] Move check for disabled anti-fragmentation out of fastpath Mel Gorman
2009-03-16 17:53 ` [PATCH 07/27] Check in advance if the zonelist needs additional filtering Mel Gorman
2009-03-16 17:53 ` [PATCH 08/27] Calculate the preferred zone for allocation only once Mel Gorman
2009-03-16 17:53 ` [PATCH 09/27] Calculate the migratetype " Mel Gorman
2009-03-16 17:53 ` [PATCH 10/27] Calculate the alloc_flags " Mel Gorman
2009-03-16 17:53 ` [PATCH 11/27] Calculate the cold parameter " Mel Gorman
2009-03-16 17:53 ` [PATCH 12/27] Remove a branch by assuming __GFP_HIGH == ALLOC_HIGH Mel Gorman
2009-03-16 17:53 ` [PATCH 13/27] Inline __rmqueue_smallest() Mel Gorman
2009-03-16 18:55 ` Christoph Lameter
2009-03-16 17:53 ` [PATCH 14/27] Inline buffered_rmqueue() Mel Gorman
2009-03-16 17:53 ` [PATCH 15/27] Inline __rmqueue_fallback() Mel Gorman
2009-03-16 17:53 ` [PATCH 16/27] Save text by reducing call sites of __rmqueue() Mel Gorman
2009-03-16 17:53 ` [PATCH 17/27] Do not call get_pageblock_migratetype() more than necessary Mel Gorman
2009-03-16 17:53 ` [PATCH 18/27] Do not disable interrupts in free_page_mlock() Mel Gorman
2009-03-16 18:57 ` Christoph Lameter
2009-03-16 17:53 ` [PATCH 19/27] Do not setup zonelist cache when there is only one node Mel Gorman
2009-03-16 17:53 ` [PATCH 20/27] Use a pre-calculated value for num_online_nodes() Mel Gorman
2009-03-16 17:53 ` [PATCH 21/27] Do not check for compound pages during the page allocator sanity checks Mel Gorman
2009-03-16 17:53 ` [PATCH 22/27] Use allocation flags as an index to the zone watermark Mel Gorman
2009-03-16 17:53 ` [PATCH 23/27] Update NR_FREE_PAGES only as necessary Mel Gorman
2009-03-16 17:53 ` [PATCH 24/27] Convert gfp_zone() to use a table of precalculated values Mel Gorman
2009-03-16 19:12 ` Christoph Lameter
2009-03-18 13:52 ` Mel Gorman
2009-03-18 14:15 ` Christoph Lameter
2009-03-18 15:35 ` Mel Gorman
2009-03-18 17:21 ` Christoph Lameter
2009-03-18 18:17 ` Mel Gorman
2009-03-18 19:07 ` Christoph Lameter
2009-03-18 19:46 ` Mel Gorman
2009-03-19 0:04 ` KAMEZAWA Hiroyuki
2009-03-19 15:05 ` Christoph Lameter
2009-03-19 16:53 ` Christoph Lameter [this message]
2009-03-19 18:11 ` Mel Gorman
2009-03-19 18:15 ` Christoph Lameter
2009-03-19 18:37 ` Christoph Lameter
2009-03-16 17:53 ` [PATCH 25/27] Re-sort GFP flags and fix whitespace alignment for easier reading Mel Gorman
2009-03-16 17:53 ` [PATCH 26/27] Get the pageblock migratetype without disabling interrupts Mel Gorman
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=alpine.DEB.1.10.0903191251310.24152@qirst.com \
--to=cl@linux-foundation.org \
--cc=hannes@cmpxchg.org \
--cc=kamezawa.hiroyu@jp.fujitsu.com \
--cc=kosaki.motohiro@jp.fujitsu.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=mel@csn.ul.ie \
--cc=ming.m.lin@intel.com \
--cc=npiggin@suse.de \
--cc=penberg@cs.helsinki.fi \
--cc=peterz@infradead.org \
--cc=riel@redhat.com \
--cc=yanmin_zhang@linux.intel.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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox