2010-07-12 12:36:09 +08:00
|
|
|
#ifndef _LINUX_MEMBLOCK_H
|
|
|
|
#define _LINUX_MEMBLOCK_H
|
|
|
|
#ifdef __KERNEL__
|
|
|
|
|
2010-07-28 13:28:21 +08:00
|
|
|
#ifdef CONFIG_HAVE_MEMBLOCK
|
2010-07-12 12:36:09 +08:00
|
|
|
/*
|
|
|
|
* Logical memory blocks.
|
|
|
|
*
|
|
|
|
* Copyright (C) 2001 Peter Bergner, IBM Corp.
|
|
|
|
*
|
|
|
|
* This program is free software; you can redistribute it and/or
|
|
|
|
* modify it under the terms of the GNU General Public License
|
|
|
|
* as published by the Free Software Foundation; either version
|
|
|
|
* 2 of the License, or (at your option) any later version.
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <linux/init.h>
|
|
|
|
#include <linux/mm.h>
|
|
|
|
|
2010-07-28 13:20:58 +08:00
|
|
|
#define INIT_MEMBLOCK_REGIONS 128
|
2014-01-30 01:16:01 +08:00
|
|
|
#define INIT_PHYSMEM_REGIONS 4
|
2010-07-12 12:36:09 +08:00
|
|
|
|
2014-01-22 07:49:23 +08:00
|
|
|
/* Definition of memblock flags. */
|
2015-06-25 07:58:09 +08:00
|
|
|
enum {
|
|
|
|
MEMBLOCK_NONE = 0x0, /* No special request */
|
|
|
|
MEMBLOCK_HOTPLUG = 0x1, /* hotpluggable region */
|
2015-06-25 07:58:12 +08:00
|
|
|
MEMBLOCK_MIRROR = 0x2, /* mirrored region */
|
2015-11-30 20:28:15 +08:00
|
|
|
MEMBLOCK_NOMAP = 0x4, /* don't add to kernel direct mapping */
|
2015-06-25 07:58:09 +08:00
|
|
|
};
|
2014-01-22 07:49:23 +08:00
|
|
|
|
2010-08-04 12:06:41 +08:00
|
|
|
struct memblock_region {
|
2010-08-04 11:34:42 +08:00
|
|
|
phys_addr_t base;
|
|
|
|
phys_addr_t size;
|
2014-01-22 07:49:20 +08:00
|
|
|
unsigned long flags;
|
2011-07-14 17:43:42 +08:00
|
|
|
#ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP
|
|
|
|
int nid;
|
|
|
|
#endif
|
2010-07-12 12:36:09 +08:00
|
|
|
};
|
|
|
|
|
2010-08-04 12:06:41 +08:00
|
|
|
struct memblock_type {
|
2010-07-07 06:39:06 +08:00
|
|
|
unsigned long cnt; /* number of regions */
|
|
|
|
unsigned long max; /* size of the allocated array */
|
2011-12-09 02:22:08 +08:00
|
|
|
phys_addr_t total_size; /* size of all regions */
|
2010-07-07 06:39:06 +08:00
|
|
|
struct memblock_region *regions;
|
2010-07-12 12:36:09 +08:00
|
|
|
};
|
|
|
|
|
|
|
|
struct memblock {
|
2013-11-13 07:07:59 +08:00
|
|
|
bool bottom_up; /* is bottom up direction? */
|
2010-08-04 11:34:42 +08:00
|
|
|
phys_addr_t current_limit;
|
2010-08-04 12:06:41 +08:00
|
|
|
struct memblock_type memory;
|
|
|
|
struct memblock_type reserved;
|
2014-01-30 01:16:01 +08:00
|
|
|
#ifdef CONFIG_HAVE_MEMBLOCK_PHYS_MAP
|
|
|
|
struct memblock_type physmem;
|
|
|
|
#endif
|
2010-07-12 12:36:09 +08:00
|
|
|
};
|
|
|
|
|
|
|
|
extern struct memblock memblock;
|
2010-07-28 13:07:21 +08:00
|
|
|
extern int memblock_debug;
|
2014-01-22 07:49:35 +08:00
|
|
|
#ifdef CONFIG_MOVABLE_NODE
|
|
|
|
/* If movable_node boot option specified */
|
|
|
|
extern bool movable_node_enabled;
|
|
|
|
#endif /* CONFIG_MOVABLE_NODE */
|
2010-07-28 13:07:21 +08:00
|
|
|
|
2016-01-16 08:57:11 +08:00
|
|
|
#ifdef CONFIG_ARCH_DISCARD_MEMBLOCK
|
|
|
|
#define __init_memblock __meminit
|
|
|
|
#define __initdata_memblock __meminitdata
|
|
|
|
#else
|
|
|
|
#define __init_memblock
|
|
|
|
#define __initdata_memblock
|
|
|
|
#endif
|
|
|
|
|
2010-07-28 13:07:21 +08:00
|
|
|
#define memblock_dbg(fmt, ...) \
|
|
|
|
if (memblock_debug) printk(KERN_INFO pr_fmt(fmt), ##__VA_ARGS__)
|
2010-07-12 12:36:09 +08:00
|
|
|
|
2014-01-22 07:50:14 +08:00
|
|
|
phys_addr_t memblock_find_in_range_node(phys_addr_t size, phys_addr_t align,
|
2016-07-27 06:24:47 +08:00
|
|
|
phys_addr_t start, phys_addr_t end,
|
|
|
|
int nid, ulong flags);
|
2011-07-12 15:58:10 +08:00
|
|
|
phys_addr_t memblock_find_in_range(phys_addr_t start, phys_addr_t end,
|
|
|
|
phys_addr_t size, phys_addr_t align);
|
2012-07-12 05:02:56 +08:00
|
|
|
phys_addr_t get_allocated_memblock_reserved_regions_info(phys_addr_t *addr);
|
2014-01-24 07:53:11 +08:00
|
|
|
phys_addr_t get_allocated_memblock_memory_regions_info(phys_addr_t *addr);
|
2011-12-09 02:22:08 +08:00
|
|
|
void memblock_allow_resize(void);
|
2011-12-09 02:22:08 +08:00
|
|
|
int memblock_add_node(phys_addr_t base, phys_addr_t size, int nid);
|
2011-12-09 02:22:06 +08:00
|
|
|
int memblock_add(phys_addr_t base, phys_addr_t size);
|
|
|
|
int memblock_remove(phys_addr_t base, phys_addr_t size);
|
|
|
|
int memblock_free(phys_addr_t base, phys_addr_t size);
|
|
|
|
int memblock_reserve(phys_addr_t base, phys_addr_t size);
|
2012-10-23 07:35:18 +08:00
|
|
|
void memblock_trim_memory(phys_addr_t align);
|
mem-hotplug: handle node hole when initializing numa_meminfo.
When parsing SRAT, all memory ranges are added into numa_meminfo. In
numa_init(), before entering numa_cleanup_meminfo(), all possible memory
ranges are in numa_meminfo. And numa_cleanup_meminfo() removes all
ranges over max_pfn or empty.
But, this only works if the nodes are continuous. Let's have a look at
the following example:
We have an SRAT like this:
SRAT: Node 0 PXM 0 [mem 0x00000000-0x5fffffff]
SRAT: Node 0 PXM 0 [mem 0x100000000-0x1ffffffffff]
SRAT: Node 1 PXM 1 [mem 0x20000000000-0x3ffffffffff]
SRAT: Node 4 PXM 2 [mem 0x40000000000-0x5ffffffffff] hotplug
SRAT: Node 5 PXM 3 [mem 0x60000000000-0x7ffffffffff] hotplug
SRAT: Node 2 PXM 4 [mem 0x80000000000-0x9ffffffffff] hotplug
SRAT: Node 3 PXM 5 [mem 0xa0000000000-0xbffffffffff] hotplug
SRAT: Node 6 PXM 6 [mem 0xc0000000000-0xdffffffffff] hotplug
SRAT: Node 7 PXM 7 [mem 0xe0000000000-0xfffffffffff] hotplug
On boot, only node 0,1,2,3 exist.
And the numa_meminfo will look like this:
numa_meminfo.nr_blks = 9
1. on node 0: [0, 60000000]
2. on node 0: [100000000, 20000000000]
3. on node 1: [20000000000, 40000000000]
4. on node 4: [40000000000, 60000000000]
5. on node 5: [60000000000, 80000000000]
6. on node 2: [80000000000, a0000000000]
7. on node 3: [a0000000000, a0800000000]
8. on node 6: [c0000000000, a0800000000]
9. on node 7: [e0000000000, a0800000000]
And numa_cleanup_meminfo() will merge 1 and 2, and remove 8,9 because the
end address is over max_pfn, which is a0800000000. But 4 and 5 are not
removed because their end addresses are less then max_pfn. But in fact,
node 4 and 5 don't exist.
In a word, numa_cleanup_meminfo() is not able to handle holes between nodes.
Since memory ranges in node 4 and 5 are in numa_meminfo, in
numa_register_memblks(), node 4 and 5 will be mistakenly set to online.
If you run lscpu, it will show:
NUMA node0 CPU(s): 0-14,128-142
NUMA node1 CPU(s): 15-29,143-157
NUMA node2 CPU(s):
NUMA node3 CPU(s):
NUMA node4 CPU(s): 62-76,190-204
NUMA node5 CPU(s): 78-92,206-220
In this patch, we use memblock_overlaps_region() to check if ranges in
numa_meminfo overlap with ranges in memory_block. Since memory_block
contains all available memory at boot time, if they overlap, it means the
ranges exist. If not, then remove them from numa_meminfo.
After this patch, lscpu will show:
NUMA node0 CPU(s): 0-14,128-142
NUMA node1 CPU(s): 15-29,143-157
NUMA node4 CPU(s): 62-76,190-204
NUMA node5 CPU(s): 78-92,206-220
Signed-off-by: Tang Chen <tangchen@cn.fujitsu.com>
Reviewed-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Tejun Heo <tj@kernel.org>
Cc: Luiz Capitulino <lcapitulino@redhat.com>
Cc: Xishi Qiu <qiuxishi@huawei.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Vladimir Murzin <vladimir.murzin@arm.com>
Cc: Fabian Frederick <fabf@skynet.be>
Cc: Alexander Kuleshov <kuleshovmail@gmail.com>
Cc: Baoquan He <bhe@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-09-09 06:02:03 +08:00
|
|
|
bool memblock_overlaps_region(struct memblock_type *type,
|
|
|
|
phys_addr_t base, phys_addr_t size);
|
2014-01-22 07:49:23 +08:00
|
|
|
int memblock_mark_hotplug(phys_addr_t base, phys_addr_t size);
|
|
|
|
int memblock_clear_hotplug(phys_addr_t base, phys_addr_t size);
|
2015-06-25 07:58:12 +08:00
|
|
|
int memblock_mark_mirror(phys_addr_t base, phys_addr_t size);
|
2015-11-30 20:28:15 +08:00
|
|
|
int memblock_mark_nomap(phys_addr_t base, phys_addr_t size);
|
2015-06-25 07:58:12 +08:00
|
|
|
ulong choose_memblock_flags(void);
|
2014-01-30 01:16:01 +08:00
|
|
|
|
|
|
|
/* Low level functions */
|
|
|
|
int memblock_add_range(struct memblock_type *type,
|
|
|
|
phys_addr_t base, phys_addr_t size,
|
|
|
|
int nid, unsigned long flags);
|
|
|
|
|
2015-06-25 07:58:09 +08:00
|
|
|
void __next_mem_range(u64 *idx, int nid, ulong flags,
|
|
|
|
struct memblock_type *type_a,
|
2014-01-30 01:16:01 +08:00
|
|
|
struct memblock_type *type_b, phys_addr_t *out_start,
|
|
|
|
phys_addr_t *out_end, int *out_nid);
|
|
|
|
|
2015-06-25 07:58:09 +08:00
|
|
|
void __next_mem_range_rev(u64 *idx, int nid, ulong flags,
|
|
|
|
struct memblock_type *type_a,
|
2014-01-30 01:16:01 +08:00
|
|
|
struct memblock_type *type_b, phys_addr_t *out_start,
|
|
|
|
phys_addr_t *out_end, int *out_nid);
|
|
|
|
|
memblock: introduce a for_each_reserved_mem_region iterator
Struct page initialisation had been identified as one of the reasons why
large machines take a long time to boot. Patches were posted a long time ago
to defer initialisation until they were first used. This was rejected on
the grounds it should not be necessary to hurt the fast paths. This series
reuses much of the work from that time but defers the initialisation of
memory to kswapd so that one thread per node initialises memory local to
that node.
After applying the series and setting the appropriate Kconfig variable I
see this in the boot log on a 64G machine
[ 7.383764] kswapd 0 initialised deferred memory in 188ms
[ 7.404253] kswapd 1 initialised deferred memory in 208ms
[ 7.411044] kswapd 3 initialised deferred memory in 216ms
[ 7.411551] kswapd 2 initialised deferred memory in 216ms
On a 1TB machine, I see
[ 8.406511] kswapd 3 initialised deferred memory in 1116ms
[ 8.428518] kswapd 1 initialised deferred memory in 1140ms
[ 8.435977] kswapd 0 initialised deferred memory in 1148ms
[ 8.437416] kswapd 2 initialised deferred memory in 1148ms
Once booted the machine appears to work as normal. Boot times were measured
from the time shutdown was called until ssh was available again. In the
64G case, the boot time savings are negligible. On the 1TB machine, the
savings were 16 seconds.
Nate Zimmer said:
: On an older 8 TB box with lots and lots of cpus the boot time, as
: measure from grub to login prompt, the boot time improved from 1484
: seconds to exactly 1000 seconds.
Waiman Long said:
: I ran a bootup timing test on a 12-TB 16-socket IvyBridge-EX system. From
: grub menu to ssh login, the bootup time was 453s before the patch and 265s
: after the patch - a saving of 188s (42%).
Daniel Blueman said:
: On a 7TB, 1728-core NumaConnect system with 108 NUMA nodes, we're seeing
: stock 4.0 boot in 7136s. This drops to 2159s, or a 70% reduction with
: this patchset. Non-temporal PMD init (https://lkml.org/lkml/2015/4/23/350)
: drops this to 1045s.
This patch (of 13):
As part of initializing struct page's in 2MiB chunks, we noticed that at
the end of free_all_bootmem(), there was nothing which had forced the
reserved/allocated 4KiB pages to be initialized.
This helper function will be used for that expansion.
Signed-off-by: Robin Holt <holt@sgi.com>
Signed-off-by: Nate Zimmer <nzimmer@sgi.com>
Signed-off-by: Mel Gorman <mgorman@suse.de>
Tested-by: Nate Zimmer <nzimmer@sgi.com>
Tested-by: Waiman Long <waiman.long@hp.com>
Tested-by: Daniel J Blueman <daniel@numascale.com>
Acked-by: Pekka Enberg <penberg@kernel.org>
Cc: Robin Holt <robinmholt@gmail.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Waiman Long <waiman.long@hp.com>
Cc: Scott Norton <scott.norton@hp.com>
Cc: "Luck, Tony" <tony.luck@intel.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-07-01 05:56:41 +08:00
|
|
|
void __next_reserved_mem_region(u64 *idx, phys_addr_t *out_start,
|
2016-07-27 06:24:47 +08:00
|
|
|
phys_addr_t *out_end);
|
memblock: introduce a for_each_reserved_mem_region iterator
Struct page initialisation had been identified as one of the reasons why
large machines take a long time to boot. Patches were posted a long time ago
to defer initialisation until they were first used. This was rejected on
the grounds it should not be necessary to hurt the fast paths. This series
reuses much of the work from that time but defers the initialisation of
memory to kswapd so that one thread per node initialises memory local to
that node.
After applying the series and setting the appropriate Kconfig variable I
see this in the boot log on a 64G machine
[ 7.383764] kswapd 0 initialised deferred memory in 188ms
[ 7.404253] kswapd 1 initialised deferred memory in 208ms
[ 7.411044] kswapd 3 initialised deferred memory in 216ms
[ 7.411551] kswapd 2 initialised deferred memory in 216ms
On a 1TB machine, I see
[ 8.406511] kswapd 3 initialised deferred memory in 1116ms
[ 8.428518] kswapd 1 initialised deferred memory in 1140ms
[ 8.435977] kswapd 0 initialised deferred memory in 1148ms
[ 8.437416] kswapd 2 initialised deferred memory in 1148ms
Once booted the machine appears to work as normal. Boot times were measured
from the time shutdown was called until ssh was available again. In the
64G case, the boot time savings are negligible. On the 1TB machine, the
savings were 16 seconds.
Nate Zimmer said:
: On an older 8 TB box with lots and lots of cpus the boot time, as
: measure from grub to login prompt, the boot time improved from 1484
: seconds to exactly 1000 seconds.
Waiman Long said:
: I ran a bootup timing test on a 12-TB 16-socket IvyBridge-EX system. From
: grub menu to ssh login, the bootup time was 453s before the patch and 265s
: after the patch - a saving of 188s (42%).
Daniel Blueman said:
: On a 7TB, 1728-core NumaConnect system with 108 NUMA nodes, we're seeing
: stock 4.0 boot in 7136s. This drops to 2159s, or a 70% reduction with
: this patchset. Non-temporal PMD init (https://lkml.org/lkml/2015/4/23/350)
: drops this to 1045s.
This patch (of 13):
As part of initializing struct page's in 2MiB chunks, we noticed that at
the end of free_all_bootmem(), there was nothing which had forced the
reserved/allocated 4KiB pages to be initialized.
This helper function will be used for that expansion.
Signed-off-by: Robin Holt <holt@sgi.com>
Signed-off-by: Nate Zimmer <nzimmer@sgi.com>
Signed-off-by: Mel Gorman <mgorman@suse.de>
Tested-by: Nate Zimmer <nzimmer@sgi.com>
Tested-by: Waiman Long <waiman.long@hp.com>
Tested-by: Daniel J Blueman <daniel@numascale.com>
Acked-by: Pekka Enberg <penberg@kernel.org>
Cc: Robin Holt <robinmholt@gmail.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Waiman Long <waiman.long@hp.com>
Cc: Scott Norton <scott.norton@hp.com>
Cc: "Luck, Tony" <tony.luck@intel.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-07-01 05:56:41 +08:00
|
|
|
|
2014-01-30 01:16:01 +08:00
|
|
|
/**
|
|
|
|
* for_each_mem_range - iterate through memblock areas from type_a and not
|
|
|
|
* included in type_b. Or just type_a if type_b is NULL.
|
|
|
|
* @i: u64 used as loop variable
|
|
|
|
* @type_a: ptr to memblock_type to iterate
|
|
|
|
* @type_b: ptr to memblock_type which excludes from the iteration
|
|
|
|
* @nid: node selector, %NUMA_NO_NODE for all nodes
|
2015-06-25 07:58:09 +08:00
|
|
|
* @flags: pick from blocks based on memory attributes
|
2014-01-30 01:16:01 +08:00
|
|
|
* @p_start: ptr to phys_addr_t for start address of the range, can be %NULL
|
|
|
|
* @p_end: ptr to phys_addr_t for end address of the range, can be %NULL
|
|
|
|
* @p_nid: ptr to int for nid of the range, can be %NULL
|
|
|
|
*/
|
2015-06-25 07:58:09 +08:00
|
|
|
#define for_each_mem_range(i, type_a, type_b, nid, flags, \
|
2014-01-30 01:16:01 +08:00
|
|
|
p_start, p_end, p_nid) \
|
2015-06-25 07:58:09 +08:00
|
|
|
for (i = 0, __next_mem_range(&i, nid, flags, type_a, type_b, \
|
2014-01-30 01:16:01 +08:00
|
|
|
p_start, p_end, p_nid); \
|
|
|
|
i != (u64)ULLONG_MAX; \
|
2015-06-25 07:58:09 +08:00
|
|
|
__next_mem_range(&i, nid, flags, type_a, type_b, \
|
2014-01-30 01:16:01 +08:00
|
|
|
p_start, p_end, p_nid))
|
|
|
|
|
|
|
|
/**
|
|
|
|
* for_each_mem_range_rev - reverse iterate through memblock areas from
|
|
|
|
* type_a and not included in type_b. Or just type_a if type_b is NULL.
|
|
|
|
* @i: u64 used as loop variable
|
|
|
|
* @type_a: ptr to memblock_type to iterate
|
|
|
|
* @type_b: ptr to memblock_type which excludes from the iteration
|
|
|
|
* @nid: node selector, %NUMA_NO_NODE for all nodes
|
2015-06-25 07:58:09 +08:00
|
|
|
* @flags: pick from blocks based on memory attributes
|
2014-01-30 01:16:01 +08:00
|
|
|
* @p_start: ptr to phys_addr_t for start address of the range, can be %NULL
|
|
|
|
* @p_end: ptr to phys_addr_t for end address of the range, can be %NULL
|
|
|
|
* @p_nid: ptr to int for nid of the range, can be %NULL
|
|
|
|
*/
|
2015-06-25 07:58:09 +08:00
|
|
|
#define for_each_mem_range_rev(i, type_a, type_b, nid, flags, \
|
2014-01-30 01:16:01 +08:00
|
|
|
p_start, p_end, p_nid) \
|
|
|
|
for (i = (u64)ULLONG_MAX, \
|
2015-06-25 07:58:09 +08:00
|
|
|
__next_mem_range_rev(&i, nid, flags, type_a, type_b,\
|
2016-07-27 06:24:47 +08:00
|
|
|
p_start, p_end, p_nid); \
|
2014-01-30 01:16:01 +08:00
|
|
|
i != (u64)ULLONG_MAX; \
|
2015-06-25 07:58:09 +08:00
|
|
|
__next_mem_range_rev(&i, nid, flags, type_a, type_b, \
|
2014-01-30 01:16:01 +08:00
|
|
|
p_start, p_end, p_nid))
|
|
|
|
|
memblock: introduce a for_each_reserved_mem_region iterator
Struct page initialisation had been identified as one of the reasons why
large machines take a long time to boot. Patches were posted a long time ago
to defer initialisation until they were first used. This was rejected on
the grounds it should not be necessary to hurt the fast paths. This series
reuses much of the work from that time but defers the initialisation of
memory to kswapd so that one thread per node initialises memory local to
that node.
After applying the series and setting the appropriate Kconfig variable I
see this in the boot log on a 64G machine
[ 7.383764] kswapd 0 initialised deferred memory in 188ms
[ 7.404253] kswapd 1 initialised deferred memory in 208ms
[ 7.411044] kswapd 3 initialised deferred memory in 216ms
[ 7.411551] kswapd 2 initialised deferred memory in 216ms
On a 1TB machine, I see
[ 8.406511] kswapd 3 initialised deferred memory in 1116ms
[ 8.428518] kswapd 1 initialised deferred memory in 1140ms
[ 8.435977] kswapd 0 initialised deferred memory in 1148ms
[ 8.437416] kswapd 2 initialised deferred memory in 1148ms
Once booted the machine appears to work as normal. Boot times were measured
from the time shutdown was called until ssh was available again. In the
64G case, the boot time savings are negligible. On the 1TB machine, the
savings were 16 seconds.
Nate Zimmer said:
: On an older 8 TB box with lots and lots of cpus the boot time, as
: measure from grub to login prompt, the boot time improved from 1484
: seconds to exactly 1000 seconds.
Waiman Long said:
: I ran a bootup timing test on a 12-TB 16-socket IvyBridge-EX system. From
: grub menu to ssh login, the bootup time was 453s before the patch and 265s
: after the patch - a saving of 188s (42%).
Daniel Blueman said:
: On a 7TB, 1728-core NumaConnect system with 108 NUMA nodes, we're seeing
: stock 4.0 boot in 7136s. This drops to 2159s, or a 70% reduction with
: this patchset. Non-temporal PMD init (https://lkml.org/lkml/2015/4/23/350)
: drops this to 1045s.
This patch (of 13):
As part of initializing struct page's in 2MiB chunks, we noticed that at
the end of free_all_bootmem(), there was nothing which had forced the
reserved/allocated 4KiB pages to be initialized.
This helper function will be used for that expansion.
Signed-off-by: Robin Holt <holt@sgi.com>
Signed-off-by: Nate Zimmer <nzimmer@sgi.com>
Signed-off-by: Mel Gorman <mgorman@suse.de>
Tested-by: Nate Zimmer <nzimmer@sgi.com>
Tested-by: Waiman Long <waiman.long@hp.com>
Tested-by: Daniel J Blueman <daniel@numascale.com>
Acked-by: Pekka Enberg <penberg@kernel.org>
Cc: Robin Holt <robinmholt@gmail.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Waiman Long <waiman.long@hp.com>
Cc: Scott Norton <scott.norton@hp.com>
Cc: "Luck, Tony" <tony.luck@intel.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-07-01 05:56:41 +08:00
|
|
|
/**
|
|
|
|
* for_each_reserved_mem_region - iterate over all reserved memblock areas
|
|
|
|
* @i: u64 used as loop variable
|
|
|
|
* @p_start: ptr to phys_addr_t for start address of the range, can be %NULL
|
|
|
|
* @p_end: ptr to phys_addr_t for end address of the range, can be %NULL
|
|
|
|
*
|
|
|
|
* Walks over reserved areas of memblock. Available as soon as memblock
|
|
|
|
* is initialized.
|
|
|
|
*/
|
|
|
|
#define for_each_reserved_mem_region(i, p_start, p_end) \
|
2016-07-27 06:24:47 +08:00
|
|
|
for (i = 0UL, __next_reserved_mem_region(&i, p_start, p_end); \
|
memblock: introduce a for_each_reserved_mem_region iterator
Struct page initialisation had been identified as one of the reasons why
large machines take a long time to boot. Patches were posted a long time ago
to defer initialisation until they were first used. This was rejected on
the grounds it should not be necessary to hurt the fast paths. This series
reuses much of the work from that time but defers the initialisation of
memory to kswapd so that one thread per node initialises memory local to
that node.
After applying the series and setting the appropriate Kconfig variable I
see this in the boot log on a 64G machine
[ 7.383764] kswapd 0 initialised deferred memory in 188ms
[ 7.404253] kswapd 1 initialised deferred memory in 208ms
[ 7.411044] kswapd 3 initialised deferred memory in 216ms
[ 7.411551] kswapd 2 initialised deferred memory in 216ms
On a 1TB machine, I see
[ 8.406511] kswapd 3 initialised deferred memory in 1116ms
[ 8.428518] kswapd 1 initialised deferred memory in 1140ms
[ 8.435977] kswapd 0 initialised deferred memory in 1148ms
[ 8.437416] kswapd 2 initialised deferred memory in 1148ms
Once booted the machine appears to work as normal. Boot times were measured
from the time shutdown was called until ssh was available again. In the
64G case, the boot time savings are negligible. On the 1TB machine, the
savings were 16 seconds.
Nate Zimmer said:
: On an older 8 TB box with lots and lots of cpus the boot time, as
: measure from grub to login prompt, the boot time improved from 1484
: seconds to exactly 1000 seconds.
Waiman Long said:
: I ran a bootup timing test on a 12-TB 16-socket IvyBridge-EX system. From
: grub menu to ssh login, the bootup time was 453s before the patch and 265s
: after the patch - a saving of 188s (42%).
Daniel Blueman said:
: On a 7TB, 1728-core NumaConnect system with 108 NUMA nodes, we're seeing
: stock 4.0 boot in 7136s. This drops to 2159s, or a 70% reduction with
: this patchset. Non-temporal PMD init (https://lkml.org/lkml/2015/4/23/350)
: drops this to 1045s.
This patch (of 13):
As part of initializing struct page's in 2MiB chunks, we noticed that at
the end of free_all_bootmem(), there was nothing which had forced the
reserved/allocated 4KiB pages to be initialized.
This helper function will be used for that expansion.
Signed-off-by: Robin Holt <holt@sgi.com>
Signed-off-by: Nate Zimmer <nzimmer@sgi.com>
Signed-off-by: Mel Gorman <mgorman@suse.de>
Tested-by: Nate Zimmer <nzimmer@sgi.com>
Tested-by: Waiman Long <waiman.long@hp.com>
Tested-by: Daniel J Blueman <daniel@numascale.com>
Acked-by: Pekka Enberg <penberg@kernel.org>
Cc: Robin Holt <robinmholt@gmail.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Waiman Long <waiman.long@hp.com>
Cc: Scott Norton <scott.norton@hp.com>
Cc: "Luck, Tony" <tony.luck@intel.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-07-01 05:56:41 +08:00
|
|
|
i != (u64)ULLONG_MAX; \
|
|
|
|
__next_reserved_mem_region(&i, p_start, p_end))
|
|
|
|
|
2014-01-22 07:49:35 +08:00
|
|
|
#ifdef CONFIG_MOVABLE_NODE
|
|
|
|
static inline bool memblock_is_hotpluggable(struct memblock_region *m)
|
|
|
|
{
|
|
|
|
return m->flags & MEMBLOCK_HOTPLUG;
|
|
|
|
}
|
|
|
|
|
2016-01-16 08:57:11 +08:00
|
|
|
static inline bool __init_memblock movable_node_is_enabled(void)
|
2014-01-22 07:49:35 +08:00
|
|
|
{
|
|
|
|
return movable_node_enabled;
|
|
|
|
}
|
|
|
|
#else
|
|
|
|
static inline bool memblock_is_hotpluggable(struct memblock_region *m)
|
|
|
|
{
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
static inline bool movable_node_is_enabled(void)
|
|
|
|
{
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
#endif
|
2010-07-07 06:39:01 +08:00
|
|
|
|
2015-06-25 07:58:12 +08:00
|
|
|
static inline bool memblock_is_mirror(struct memblock_region *m)
|
|
|
|
{
|
|
|
|
return m->flags & MEMBLOCK_MIRROR;
|
|
|
|
}
|
|
|
|
|
2015-11-30 20:28:15 +08:00
|
|
|
static inline bool memblock_is_nomap(struct memblock_region *m)
|
|
|
|
{
|
|
|
|
return m->flags & MEMBLOCK_NOMAP;
|
|
|
|
}
|
|
|
|
|
2011-12-09 02:22:09 +08:00
|
|
|
#ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP
|
2013-09-12 05:22:17 +08:00
|
|
|
int memblock_search_pfn_nid(unsigned long pfn, unsigned long *start_pfn,
|
|
|
|
unsigned long *end_pfn);
|
2011-12-09 02:22:09 +08:00
|
|
|
void __next_mem_pfn_range(int *idx, int nid, unsigned long *out_start_pfn,
|
|
|
|
unsigned long *out_end_pfn, int *out_nid);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* for_each_mem_pfn_range - early memory pfn range iterator
|
|
|
|
* @i: an integer used as loop variable
|
|
|
|
* @nid: node selector, %MAX_NUMNODES for all nodes
|
|
|
|
* @p_start: ptr to ulong for start pfn of the range, can be %NULL
|
|
|
|
* @p_end: ptr to ulong for end pfn of the range, can be %NULL
|
|
|
|
* @p_nid: ptr to int for nid of the range, can be %NULL
|
|
|
|
*
|
2012-10-09 07:32:24 +08:00
|
|
|
* Walks over configured memory ranges.
|
2011-12-09 02:22:09 +08:00
|
|
|
*/
|
|
|
|
#define for_each_mem_pfn_range(i, nid, p_start, p_end, p_nid) \
|
|
|
|
for (i = -1, __next_mem_pfn_range(&i, nid, p_start, p_end, p_nid); \
|
|
|
|
i >= 0; __next_mem_pfn_range(&i, nid, p_start, p_end, p_nid))
|
|
|
|
#endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */
|
|
|
|
|
2011-07-12 17:15:59 +08:00
|
|
|
/**
|
|
|
|
* for_each_free_mem_range - iterate through free memblock areas
|
|
|
|
* @i: u64 used as loop variable
|
2014-01-22 07:50:16 +08:00
|
|
|
* @nid: node selector, %NUMA_NO_NODE for all nodes
|
2016-01-15 07:22:04 +08:00
|
|
|
* @flags: pick from blocks based on memory attributes
|
2011-07-12 17:15:59 +08:00
|
|
|
* @p_start: ptr to phys_addr_t for start address of the range, can be %NULL
|
|
|
|
* @p_end: ptr to phys_addr_t for end address of the range, can be %NULL
|
|
|
|
* @p_nid: ptr to int for nid of the range, can be %NULL
|
|
|
|
*
|
|
|
|
* Walks over free (memory && !reserved) areas of memblock. Available as
|
|
|
|
* soon as memblock is initialized.
|
|
|
|
*/
|
2015-06-25 07:58:09 +08:00
|
|
|
#define for_each_free_mem_range(i, nid, flags, p_start, p_end, p_nid) \
|
2014-01-30 01:16:01 +08:00
|
|
|
for_each_mem_range(i, &memblock.memory, &memblock.reserved, \
|
2015-06-25 07:58:09 +08:00
|
|
|
nid, flags, p_start, p_end, p_nid)
|
2011-12-09 02:22:09 +08:00
|
|
|
|
|
|
|
/**
|
|
|
|
* for_each_free_mem_range_reverse - rev-iterate through free memblock areas
|
|
|
|
* @i: u64 used as loop variable
|
2014-01-22 07:50:16 +08:00
|
|
|
* @nid: node selector, %NUMA_NO_NODE for all nodes
|
2016-01-15 07:22:04 +08:00
|
|
|
* @flags: pick from blocks based on memory attributes
|
2011-12-09 02:22:09 +08:00
|
|
|
* @p_start: ptr to phys_addr_t for start address of the range, can be %NULL
|
|
|
|
* @p_end: ptr to phys_addr_t for end address of the range, can be %NULL
|
|
|
|
* @p_nid: ptr to int for nid of the range, can be %NULL
|
|
|
|
*
|
|
|
|
* Walks over free (memory && !reserved) areas of memblock in reverse
|
|
|
|
* order. Available as soon as memblock is initialized.
|
|
|
|
*/
|
2015-06-25 07:58:09 +08:00
|
|
|
#define for_each_free_mem_range_reverse(i, nid, flags, p_start, p_end, \
|
|
|
|
p_nid) \
|
2014-01-30 01:16:01 +08:00
|
|
|
for_each_mem_range_rev(i, &memblock.memory, &memblock.reserved, \
|
2015-06-25 07:58:09 +08:00
|
|
|
nid, flags, p_start, p_end, p_nid)
|
2011-12-09 02:22:09 +08:00
|
|
|
|
2014-01-22 07:49:23 +08:00
|
|
|
static inline void memblock_set_region_flags(struct memblock_region *r,
|
|
|
|
unsigned long flags)
|
|
|
|
{
|
|
|
|
r->flags |= flags;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void memblock_clear_region_flags(struct memblock_region *r,
|
|
|
|
unsigned long flags)
|
|
|
|
{
|
|
|
|
r->flags &= ~flags;
|
|
|
|
}
|
|
|
|
|
2011-07-14 17:43:42 +08:00
|
|
|
#ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP
|
2014-01-22 07:49:26 +08:00
|
|
|
int memblock_set_node(phys_addr_t base, phys_addr_t size,
|
|
|
|
struct memblock_type *type, int nid);
|
2011-07-14 17:43:42 +08:00
|
|
|
|
|
|
|
static inline void memblock_set_region_node(struct memblock_region *r, int nid)
|
|
|
|
{
|
|
|
|
r->nid = nid;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline int memblock_get_region_node(const struct memblock_region *r)
|
|
|
|
{
|
|
|
|
return r->nid;
|
|
|
|
}
|
|
|
|
#else
|
|
|
|
static inline void memblock_set_region_node(struct memblock_region *r, int nid)
|
|
|
|
{
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline int memblock_get_region_node(const struct memblock_region *r)
|
|
|
|
{
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
#endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */
|
|
|
|
|
2011-12-09 02:22:06 +08:00
|
|
|
phys_addr_t memblock_alloc_nid(phys_addr_t size, phys_addr_t align, int nid);
|
|
|
|
phys_addr_t memblock_alloc_try_nid(phys_addr_t size, phys_addr_t align, int nid);
|
2010-07-07 06:39:17 +08:00
|
|
|
|
2011-12-09 02:22:06 +08:00
|
|
|
phys_addr_t memblock_alloc(phys_addr_t size, phys_addr_t align);
|
2010-07-07 06:39:01 +08:00
|
|
|
|
2013-11-13 07:07:59 +08:00
|
|
|
#ifdef CONFIG_MOVABLE_NODE
|
|
|
|
/*
|
|
|
|
* Set the allocation direction to bottom-up or top-down.
|
|
|
|
*/
|
2014-08-07 07:05:03 +08:00
|
|
|
static inline void __init memblock_set_bottom_up(bool enable)
|
2013-11-13 07:07:59 +08:00
|
|
|
{
|
|
|
|
memblock.bottom_up = enable;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Check if the allocation direction is bottom-up or not.
|
|
|
|
* if this is true, that said, memblock will allocate memory
|
|
|
|
* in bottom-up direction.
|
|
|
|
*/
|
|
|
|
static inline bool memblock_bottom_up(void)
|
|
|
|
{
|
|
|
|
return memblock.bottom_up;
|
|
|
|
}
|
|
|
|
#else
|
2014-08-07 07:05:03 +08:00
|
|
|
static inline void __init memblock_set_bottom_up(bool enable) {}
|
2013-11-13 07:07:59 +08:00
|
|
|
static inline bool memblock_bottom_up(void) { return false; }
|
|
|
|
#endif
|
|
|
|
|
2010-07-07 06:39:01 +08:00
|
|
|
/* Flags for memblock_alloc_base() amd __memblock_alloc_base() */
|
2010-08-04 11:34:42 +08:00
|
|
|
#define MEMBLOCK_ALLOC_ANYWHERE (~(phys_addr_t)0)
|
2010-07-07 06:39:01 +08:00
|
|
|
#define MEMBLOCK_ALLOC_ACCESSIBLE 0
|
|
|
|
|
2014-06-05 07:06:53 +08:00
|
|
|
phys_addr_t __init memblock_alloc_range(phys_addr_t size, phys_addr_t align,
|
2015-06-25 07:58:09 +08:00
|
|
|
phys_addr_t start, phys_addr_t end,
|
|
|
|
ulong flags);
|
2011-12-09 02:22:06 +08:00
|
|
|
phys_addr_t memblock_alloc_base(phys_addr_t size, phys_addr_t align,
|
|
|
|
phys_addr_t max_addr);
|
|
|
|
phys_addr_t __memblock_alloc_base(phys_addr_t size, phys_addr_t align,
|
|
|
|
phys_addr_t max_addr);
|
|
|
|
phys_addr_t memblock_phys_mem_size(void);
|
2016-10-08 07:59:18 +08:00
|
|
|
phys_addr_t memblock_reserved_size(void);
|
2013-01-25 04:20:09 +08:00
|
|
|
phys_addr_t memblock_mem_size(unsigned long limit_pfn);
|
2011-12-09 02:22:06 +08:00
|
|
|
phys_addr_t memblock_start_of_DRAM(void);
|
|
|
|
phys_addr_t memblock_end_of_DRAM(void);
|
|
|
|
void memblock_enforce_memory_limit(phys_addr_t memory_limit);
|
2016-07-29 06:48:26 +08:00
|
|
|
void memblock_mem_limit_remove_map(phys_addr_t limit);
|
2016-01-15 07:18:54 +08:00
|
|
|
bool memblock_is_memory(phys_addr_t addr);
|
2015-11-30 20:28:15 +08:00
|
|
|
int memblock_is_map_memory(phys_addr_t addr);
|
2011-12-09 02:22:06 +08:00
|
|
|
int memblock_is_region_memory(phys_addr_t base, phys_addr_t size);
|
2016-01-15 07:18:54 +08:00
|
|
|
bool memblock_is_reserved(phys_addr_t addr);
|
2015-09-09 06:02:00 +08:00
|
|
|
bool memblock_is_region_reserved(phys_addr_t base, phys_addr_t size);
|
2011-12-09 02:22:06 +08:00
|
|
|
|
2011-12-09 02:22:06 +08:00
|
|
|
extern void __memblock_dump_all(void);
|
|
|
|
|
|
|
|
static inline void memblock_dump_all(void)
|
|
|
|
{
|
|
|
|
if (memblock_debug)
|
|
|
|
__memblock_dump_all();
|
|
|
|
}
|
2010-07-12 12:36:09 +08:00
|
|
|
|
2010-07-07 06:39:01 +08:00
|
|
|
/**
|
|
|
|
* memblock_set_current_limit - Set the current allocation limit to allow
|
|
|
|
* limiting allocations to what is currently
|
|
|
|
* accessible during boot
|
|
|
|
* @limit: New limit value (physical address)
|
|
|
|
*/
|
2011-12-09 02:22:06 +08:00
|
|
|
void memblock_set_current_limit(phys_addr_t limit);
|
2010-07-07 06:39:01 +08:00
|
|
|
|
2010-07-07 06:38:58 +08:00
|
|
|
|
2014-02-27 08:23:43 +08:00
|
|
|
phys_addr_t memblock_get_current_limit(void);
|
|
|
|
|
2010-08-04 11:40:38 +08:00
|
|
|
/*
|
|
|
|
* pfn conversion functions
|
|
|
|
*
|
|
|
|
* While the memory MEMBLOCKs should always be page aligned, the reserved
|
|
|
|
* MEMBLOCKs may not be. This accessor attempt to provide a very clear
|
|
|
|
* idea of what they return for such non aligned MEMBLOCKs.
|
|
|
|
*/
|
|
|
|
|
|
|
|
/**
|
2010-10-13 05:07:09 +08:00
|
|
|
* memblock_region_memory_base_pfn - Return the lowest pfn intersecting with the memory region
|
2010-08-04 11:40:38 +08:00
|
|
|
* @reg: memblock_region structure
|
|
|
|
*/
|
2010-10-13 05:07:09 +08:00
|
|
|
static inline unsigned long memblock_region_memory_base_pfn(const struct memblock_region *reg)
|
2010-08-04 11:40:38 +08:00
|
|
|
{
|
2010-10-13 05:07:09 +08:00
|
|
|
return PFN_UP(reg->base);
|
2010-08-04 11:40:38 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
2010-10-13 05:07:09 +08:00
|
|
|
* memblock_region_memory_end_pfn - Return the end_pfn this region
|
2010-08-04 11:40:38 +08:00
|
|
|
* @reg: memblock_region structure
|
|
|
|
*/
|
2010-10-13 05:07:09 +08:00
|
|
|
static inline unsigned long memblock_region_memory_end_pfn(const struct memblock_region *reg)
|
2010-08-04 11:40:38 +08:00
|
|
|
{
|
2010-10-13 05:07:09 +08:00
|
|
|
return PFN_DOWN(reg->base + reg->size);
|
2010-08-04 11:40:38 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
2010-10-13 05:07:09 +08:00
|
|
|
* memblock_region_reserved_base_pfn - Return the lowest pfn intersecting with the reserved region
|
2010-08-04 11:40:38 +08:00
|
|
|
* @reg: memblock_region structure
|
|
|
|
*/
|
2010-10-13 05:07:09 +08:00
|
|
|
static inline unsigned long memblock_region_reserved_base_pfn(const struct memblock_region *reg)
|
2010-08-04 11:40:38 +08:00
|
|
|
{
|
2010-10-13 05:07:09 +08:00
|
|
|
return PFN_DOWN(reg->base);
|
2010-08-04 11:40:38 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
2010-10-13 05:07:09 +08:00
|
|
|
* memblock_region_reserved_end_pfn - Return the end_pfn this region
|
2010-08-04 11:40:38 +08:00
|
|
|
* @reg: memblock_region structure
|
|
|
|
*/
|
2010-10-13 05:07:09 +08:00
|
|
|
static inline unsigned long memblock_region_reserved_end_pfn(const struct memblock_region *reg)
|
2010-08-04 11:40:38 +08:00
|
|
|
{
|
2010-10-13 05:07:09 +08:00
|
|
|
return PFN_UP(reg->base + reg->size);
|
2010-08-04 11:40:38 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
#define for_each_memblock(memblock_type, region) \
|
2016-07-27 06:24:47 +08:00
|
|
|
for (region = memblock.memblock_type.regions; \
|
2010-08-04 11:40:38 +08:00
|
|
|
region < (memblock.memblock_type.regions + memblock.memblock_type.cnt); \
|
|
|
|
region++)
|
|
|
|
|
2016-01-15 07:20:42 +08:00
|
|
|
#define for_each_memblock_type(memblock_type, rgn) \
|
2016-07-27 06:24:47 +08:00
|
|
|
for (idx = 0, rgn = &memblock_type->regions[0]; \
|
|
|
|
idx < memblock_type->cnt; \
|
|
|
|
idx++, rgn = &memblock_type->regions[idx])
|
2010-08-04 11:40:38 +08:00
|
|
|
|
2015-04-15 06:48:27 +08:00
|
|
|
#ifdef CONFIG_MEMTEST
|
2015-04-15 06:48:30 +08:00
|
|
|
extern void early_memtest(phys_addr_t start, phys_addr_t end);
|
2015-04-15 06:48:27 +08:00
|
|
|
#else
|
2015-04-15 06:48:30 +08:00
|
|
|
static inline void early_memtest(phys_addr_t start, phys_addr_t end)
|
2015-04-15 06:48:27 +08:00
|
|
|
{
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
2011-05-25 08:13:19 +08:00
|
|
|
#else
|
|
|
|
static inline phys_addr_t memblock_alloc(phys_addr_t size, phys_addr_t align)
|
|
|
|
{
|
2011-07-12 15:58:09 +08:00
|
|
|
return 0;
|
2011-05-25 08:13:19 +08:00
|
|
|
}
|
|
|
|
|
2010-07-28 13:28:21 +08:00
|
|
|
#endif /* CONFIG_HAVE_MEMBLOCK */
|
|
|
|
|
2010-07-12 12:36:09 +08:00
|
|
|
#endif /* __KERNEL__ */
|
|
|
|
|
|
|
|
#endif /* _LINUX_MEMBLOCK_H */
|