2005-04-17 02:20:36 +04:00
|
|
|
#ifndef _LINUX_SWAP_H
|
|
|
|
|
#define _LINUX_SWAP_H
|
|
|
|
|
|
|
|
|
|
#include <linux/spinlock.h>
|
|
|
|
|
#include <linux/linkage.h>
|
|
|
|
|
#include <linux/mmzone.h>
|
|
|
|
|
#include <linux/list.h>
|
2008-02-07 11:13:56 +03:00
|
|
|
#include <linux/memcontrol.h>
|
2005-04-17 02:20:36 +04:00
|
|
|
#include <linux/sched.h>
|
2008-10-19 07:26:53 +04:00
|
|
|
#include <linux/node.h>
|
2013-02-23 04:34:37 +04:00
|
|
|
#include <linux/fs.h>
|
2011-07-27 03:09:06 +04:00
|
|
|
#include <linux/atomic.h>
|
2013-07-04 02:02:34 +04:00
|
|
|
#include <linux/page-flags.h>
|
2005-04-17 02:20:36 +04:00
|
|
|
#include <asm/page.h>
|
|
|
|
|
|
2006-09-26 10:31:20 +04:00
|
|
|
struct notifier_block;
|
|
|
|
|
|
2006-09-26 10:32:42 +04:00
|
|
|
struct bio;
|
|
|
|
|
|
2005-04-17 02:20:36 +04:00
|
|
|
#define SWAP_FLAG_PREFER 0x8000 /* set if swap priority specified */
|
|
|
|
|
#define SWAP_FLAG_PRIO_MASK 0x7fff
|
|
|
|
|
#define SWAP_FLAG_PRIO_SHIFT 0
|
swap: discard while swapping only if SWAP_FLAG_DISCARD_PAGES
Considering the use cases where the swap device supports discard:
a) and can do it quickly;
b) but it's slow to do in small granularities (or concurrent with other
I/O);
c) but the implementation is so horrendous that you don't even want to
send one down;
And assuming that the sysadmin considers it useful to send the discards down
at all, we would (probably) want the following solutions:
i. do the fine-grained discards for freed swap pages, if device is
capable of doing so optimally;
ii. do single-time (batched) swap area discards, either at swapon
or via something like fstrim (not implemented yet);
iii. allow doing both single-time and fine-grained discards; or
iv. turn it off completely (default behavior)
As implemented today, one can only enable/disable discards for swap, but
one cannot select, for instance, solution (ii) on a swap device like (b)
even though the single-time discard is regarded to be interesting, or
necessary to the workload because it would imply (1), and the device is
not capable of performing it optimally.
This patch addresses the scenario depicted above by introducing a way to
ensure the (probably) wanted solutions (i, ii, iii and iv) can be flexibly
flagged through swapon(8) to allow a sysadmin to select the best suitable
swap discard policy accordingly to system constraints.
This patch introduces SWAP_FLAG_DISCARD_PAGES and SWAP_FLAG_DISCARD_ONCE
new flags to allow more flexibe swap discard policies being flagged
through swapon(8). The default behavior is to keep both single-time, or
batched, area discards (SWAP_FLAG_DISCARD_ONCE) and fine-grained discards
for page-clusters (SWAP_FLAG_DISCARD_PAGES) enabled, in order to keep
consistentcy with older kernel behavior, as well as maintain compatibility
with older swapon(8). However, through the new introduced flags the best
suitable discard policy can be selected accordingly to any given swap
device constraint.
[akpm@linux-foundation.org: tweak comments]
Signed-off-by: Rafael Aquini <aquini@redhat.com>
Acked-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Shaohua Li <shli@kernel.org>
Cc: Karel Zak <kzak@redhat.com>
Cc: Jeff Moyer <jmoyer@redhat.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-07-04 02:02:46 +04:00
|
|
|
#define SWAP_FLAG_DISCARD 0x10000 /* enable discard for swap */
|
|
|
|
|
#define SWAP_FLAG_DISCARD_ONCE 0x20000 /* discard swap area at swapon-time */
|
|
|
|
|
#define SWAP_FLAG_DISCARD_PAGES 0x40000 /* discard page-clusters after use */
|
2005-04-17 02:20:36 +04:00
|
|
|
|
2012-03-29 01:42:42 +04:00
|
|
|
#define SWAP_FLAGS_VALID (SWAP_FLAG_PRIO_MASK | SWAP_FLAG_PREFER | \
|
swap: discard while swapping only if SWAP_FLAG_DISCARD_PAGES
Considering the use cases where the swap device supports discard:
a) and can do it quickly;
b) but it's slow to do in small granularities (or concurrent with other
I/O);
c) but the implementation is so horrendous that you don't even want to
send one down;
And assuming that the sysadmin considers it useful to send the discards down
at all, we would (probably) want the following solutions:
i. do the fine-grained discards for freed swap pages, if device is
capable of doing so optimally;
ii. do single-time (batched) swap area discards, either at swapon
or via something like fstrim (not implemented yet);
iii. allow doing both single-time and fine-grained discards; or
iv. turn it off completely (default behavior)
As implemented today, one can only enable/disable discards for swap, but
one cannot select, for instance, solution (ii) on a swap device like (b)
even though the single-time discard is regarded to be interesting, or
necessary to the workload because it would imply (1), and the device is
not capable of performing it optimally.
This patch addresses the scenario depicted above by introducing a way to
ensure the (probably) wanted solutions (i, ii, iii and iv) can be flexibly
flagged through swapon(8) to allow a sysadmin to select the best suitable
swap discard policy accordingly to system constraints.
This patch introduces SWAP_FLAG_DISCARD_PAGES and SWAP_FLAG_DISCARD_ONCE
new flags to allow more flexibe swap discard policies being flagged
through swapon(8). The default behavior is to keep both single-time, or
batched, area discards (SWAP_FLAG_DISCARD_ONCE) and fine-grained discards
for page-clusters (SWAP_FLAG_DISCARD_PAGES) enabled, in order to keep
consistentcy with older kernel behavior, as well as maintain compatibility
with older swapon(8). However, through the new introduced flags the best
suitable discard policy can be selected accordingly to any given swap
device constraint.
[akpm@linux-foundation.org: tweak comments]
Signed-off-by: Rafael Aquini <aquini@redhat.com>
Acked-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Shaohua Li <shli@kernel.org>
Cc: Karel Zak <kzak@redhat.com>
Cc: Jeff Moyer <jmoyer@redhat.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-07-04 02:02:46 +04:00
|
|
|
SWAP_FLAG_DISCARD | SWAP_FLAG_DISCARD_ONCE | \
|
|
|
|
|
SWAP_FLAG_DISCARD_PAGES)
|
2017-02-23 02:45:33 +03:00
|
|
|
#define SWAP_BATCH 64
|
2012-03-29 01:42:42 +04:00
|
|
|
|
2005-04-17 02:20:36 +04:00
|
|
|
static inline int current_is_kswapd(void)
|
|
|
|
|
{
|
|
|
|
|
return current->flags & PF_KSWAPD;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* MAX_SWAPFILES defines the maximum number of swaptypes: things which can
|
|
|
|
|
* be swapped to. The swap type and the offset into that swap type are
|
|
|
|
|
* encoded into pte's and into pgoff_t's in the swapcache. Using five bits
|
|
|
|
|
* for the type means that the maximum number of swapcache pages is 27 bits
|
|
|
|
|
* on 32-bit-pgoff_t architectures. And that assumes that the architecture packs
|
|
|
|
|
* the type/offset into the pte as 5/27 as well.
|
|
|
|
|
*/
|
|
|
|
|
#define MAX_SWAPFILES_SHIFT 5
|
2009-09-16 13:50:05 +04:00
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Use some of the swap files numbers for other purposes. This
|
|
|
|
|
* is a convenient way to hook into the VM to trigger special
|
|
|
|
|
* actions on faults.
|
|
|
|
|
*/
|
|
|
|
|
|
2017-09-09 02:11:43 +03:00
|
|
|
/*
|
|
|
|
|
* Unaddressable device memory support. See include/linux/hmm.h and
|
|
|
|
|
* Documentation/vm/hmm.txt. Short description is we need struct pages for
|
|
|
|
|
* device memory that is unaddressable (inaccessible) by CPU, so that we can
|
|
|
|
|
* migrate part of a process memory to device memory.
|
|
|
|
|
*
|
|
|
|
|
* When a page is migrated from CPU to device, we set the CPU page table entry
|
|
|
|
|
* to a special SWP_DEVICE_* entry.
|
|
|
|
|
*/
|
|
|
|
|
#ifdef CONFIG_DEVICE_PRIVATE
|
|
|
|
|
#define SWP_DEVICE_NUM 2
|
|
|
|
|
#define SWP_DEVICE_WRITE (MAX_SWAPFILES+SWP_HWPOISON_NUM+SWP_MIGRATION_NUM)
|
|
|
|
|
#define SWP_DEVICE_READ (MAX_SWAPFILES+SWP_HWPOISON_NUM+SWP_MIGRATION_NUM+1)
|
|
|
|
|
#else
|
|
|
|
|
#define SWP_DEVICE_NUM 0
|
|
|
|
|
#endif
|
|
|
|
|
|
2009-09-16 13:50:05 +04:00
|
|
|
/*
|
|
|
|
|
* NUMA node memory migration support
|
|
|
|
|
*/
|
|
|
|
|
#ifdef CONFIG_MIGRATION
|
|
|
|
|
#define SWP_MIGRATION_NUM 2
|
|
|
|
|
#define SWP_MIGRATION_READ (MAX_SWAPFILES + SWP_HWPOISON_NUM)
|
|
|
|
|
#define SWP_MIGRATION_WRITE (MAX_SWAPFILES + SWP_HWPOISON_NUM + 1)
|
[PATCH] Swapless page migration: add R/W migration entries
Implement read/write migration ptes
We take the upper two swapfiles for the two types of migration ptes and define
a series of macros in swapops.h.
The VM is modified to handle the migration entries. migration entries can
only be encountered when the page they are pointing to is locked. This limits
the number of places one has to fix. We also check in copy_pte_range and in
mprotect_pte_range() for migration ptes.
We check for migration ptes in do_swap_cache and call a function that will
then wait on the page lock. This allows us to effectively stop all accesses
to apge.
Migration entries are created by try_to_unmap if called for migration and
removed by local functions in migrate.c
From: Hugh Dickins <hugh@veritas.com>
Several times while testing swapless page migration (I've no NUMA, just
hacking it up to migrate recklessly while running load), I've hit the
BUG_ON(!PageLocked(p)) in migration_entry_to_page.
This comes from an orphaned migration entry, unrelated to the current
correctly locked migration, but hit by remove_anon_migration_ptes as it
checks an address in each vma of the anon_vma list.
Such an orphan may be left behind if an earlier migration raced with fork:
copy_one_pte can duplicate a migration entry from parent to child, after
remove_anon_migration_ptes has checked the child vma, but before it has
removed it from the parent vma. (If the process were later to fault on this
orphaned entry, it would hit the same BUG from migration_entry_wait.)
This could be fixed by locking anon_vma in copy_one_pte, but we'd rather
not. There's no such problem with file pages, because vma_prio_tree_add
adds child vma after parent vma, and the page table locking at each end is
enough to serialize. Follow that example with anon_vma: add new vmas to the
tail instead of the head.
(There's no corresponding problem when inserting migration entries,
because a missed pte will leave the page count and mapcount high, which is
allowed for. And there's no corresponding problem when migrating via swap,
because a leftover swap entry will be correctly faulted. But the swapless
method has no refcounting of its entries.)
From: Ingo Molnar <mingo@elte.hu>
pte_unmap_unlock() takes the pte pointer as an argument.
From: Hugh Dickins <hugh@veritas.com>
Several times while testing swapless page migration, gcc has tried to exec
a pointer instead of a string: smells like COW mappings are not being
properly write-protected on fork.
The protection in copy_one_pte looks very convincing, until at last you
realize that the second arg to make_migration_entry is a boolean "write",
and SWP_MIGRATION_READ is 30.
Anyway, it's better done like in change_pte_range, using
is_write_migration_entry and make_migration_entry_read.
From: Hugh Dickins <hugh@veritas.com>
Remove unnecessary obfuscation from sys_swapon's range check on swap type,
which blew up causing memory corruption once swapless migration made
MAX_SWAPFILES no longer 2 ^ MAX_SWAPFILES_SHIFT.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Acked-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Christoph Lameter <clameter@engr.sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
From: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-23 13:03:35 +04:00
|
|
|
#else
|
2009-09-16 13:50:05 +04:00
|
|
|
#define SWP_MIGRATION_NUM 0
|
[PATCH] Swapless page migration: add R/W migration entries
Implement read/write migration ptes
We take the upper two swapfiles for the two types of migration ptes and define
a series of macros in swapops.h.
The VM is modified to handle the migration entries. migration entries can
only be encountered when the page they are pointing to is locked. This limits
the number of places one has to fix. We also check in copy_pte_range and in
mprotect_pte_range() for migration ptes.
We check for migration ptes in do_swap_cache and call a function that will
then wait on the page lock. This allows us to effectively stop all accesses
to apge.
Migration entries are created by try_to_unmap if called for migration and
removed by local functions in migrate.c
From: Hugh Dickins <hugh@veritas.com>
Several times while testing swapless page migration (I've no NUMA, just
hacking it up to migrate recklessly while running load), I've hit the
BUG_ON(!PageLocked(p)) in migration_entry_to_page.
This comes from an orphaned migration entry, unrelated to the current
correctly locked migration, but hit by remove_anon_migration_ptes as it
checks an address in each vma of the anon_vma list.
Such an orphan may be left behind if an earlier migration raced with fork:
copy_one_pte can duplicate a migration entry from parent to child, after
remove_anon_migration_ptes has checked the child vma, but before it has
removed it from the parent vma. (If the process were later to fault on this
orphaned entry, it would hit the same BUG from migration_entry_wait.)
This could be fixed by locking anon_vma in copy_one_pte, but we'd rather
not. There's no such problem with file pages, because vma_prio_tree_add
adds child vma after parent vma, and the page table locking at each end is
enough to serialize. Follow that example with anon_vma: add new vmas to the
tail instead of the head.
(There's no corresponding problem when inserting migration entries,
because a missed pte will leave the page count and mapcount high, which is
allowed for. And there's no corresponding problem when migrating via swap,
because a leftover swap entry will be correctly faulted. But the swapless
method has no refcounting of its entries.)
From: Ingo Molnar <mingo@elte.hu>
pte_unmap_unlock() takes the pte pointer as an argument.
From: Hugh Dickins <hugh@veritas.com>
Several times while testing swapless page migration, gcc has tried to exec
a pointer instead of a string: smells like COW mappings are not being
properly write-protected on fork.
The protection in copy_one_pte looks very convincing, until at last you
realize that the second arg to make_migration_entry is a boolean "write",
and SWP_MIGRATION_READ is 30.
Anyway, it's better done like in change_pte_range, using
is_write_migration_entry and make_migration_entry_read.
From: Hugh Dickins <hugh@veritas.com>
Remove unnecessary obfuscation from sys_swapon's range check on swap type,
which blew up causing memory corruption once swapless migration made
MAX_SWAPFILES no longer 2 ^ MAX_SWAPFILES_SHIFT.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Acked-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Christoph Lameter <clameter@engr.sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
From: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-23 13:03:35 +04:00
|
|
|
#endif
|
2005-04-17 02:20:36 +04:00
|
|
|
|
2009-09-16 13:50:05 +04:00
|
|
|
/*
|
|
|
|
|
* Handling of hardware poisoned pages with memory corruption.
|
|
|
|
|
*/
|
|
|
|
|
#ifdef CONFIG_MEMORY_FAILURE
|
|
|
|
|
#define SWP_HWPOISON_NUM 1
|
|
|
|
|
#define SWP_HWPOISON MAX_SWAPFILES
|
|
|
|
|
#else
|
|
|
|
|
#define SWP_HWPOISON_NUM 0
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
#define MAX_SWAPFILES \
|
2017-09-09 02:11:43 +03:00
|
|
|
((1 << MAX_SWAPFILES_SHIFT) - SWP_DEVICE_NUM - \
|
|
|
|
|
SWP_MIGRATION_NUM - SWP_HWPOISON_NUM)
|
2009-09-16 13:50:05 +04:00
|
|
|
|
2005-04-17 02:20:36 +04:00
|
|
|
/*
|
|
|
|
|
* Magic header for a swap area. The first part of the union is
|
|
|
|
|
* what the swap magic looks like for the old (limited to 128MB)
|
|
|
|
|
* swap area format, the second part of the union adds - in the
|
|
|
|
|
* old reserved area - some extra information. Note that the first
|
|
|
|
|
* kilobyte is reserved for boot loader or disk label stuff...
|
|
|
|
|
*
|
|
|
|
|
* Having the magic at the end of the PAGE_SIZE makes detecting swap
|
|
|
|
|
* areas somewhat tricky on machines that support multiple page sizes.
|
|
|
|
|
* For 2.5 we'll probably want to move the magic to just beyond the
|
|
|
|
|
* bootbits...
|
|
|
|
|
*/
|
|
|
|
|
union swap_header {
|
|
|
|
|
struct {
|
|
|
|
|
char reserved[PAGE_SIZE - 10];
|
|
|
|
|
char magic[10]; /* SWAP-SPACE or SWAPSPACE2 */
|
|
|
|
|
} magic;
|
|
|
|
|
struct {
|
2006-06-23 13:03:14 +04:00
|
|
|
char bootbits[1024]; /* Space for disklabel etc. */
|
|
|
|
|
__u32 version;
|
|
|
|
|
__u32 last_page;
|
|
|
|
|
__u32 nr_badpages;
|
|
|
|
|
unsigned char sws_uuid[16];
|
|
|
|
|
unsigned char sws_volume[16];
|
|
|
|
|
__u32 padding[117];
|
|
|
|
|
__u32 badpages[1];
|
2005-04-17 02:20:36 +04:00
|
|
|
} info;
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* current->reclaim_state points to one of these when a task is running
|
|
|
|
|
* memory reclaim
|
|
|
|
|
*/
|
|
|
|
|
struct reclaim_state {
|
|
|
|
|
unsigned long reclaimed_slab;
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
#ifdef __KERNEL__
|
|
|
|
|
|
|
|
|
|
struct address_space;
|
|
|
|
|
struct sysinfo;
|
|
|
|
|
struct writeback_control;
|
|
|
|
|
struct zone;
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* A swap extent maps a range of a swapfile's PAGE_SIZE pages onto a range of
|
|
|
|
|
* disk blocks. A list of swap extents maps the entire swapfile. (Where the
|
|
|
|
|
* term `swapfile' refers to either a blockdevice or an IS_REG file. Apart
|
|
|
|
|
* from setup, they're handled identically.
|
|
|
|
|
*
|
|
|
|
|
* We always assume that blocks are of size PAGE_SIZE.
|
|
|
|
|
*/
|
|
|
|
|
struct swap_extent {
|
|
|
|
|
struct list_head list;
|
|
|
|
|
pgoff_t start_page;
|
|
|
|
|
pgoff_t nr_pages;
|
|
|
|
|
sector_t start_block;
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Max bad pages in the new format..
|
|
|
|
|
*/
|
|
|
|
|
#define __swapoffset(x) ((unsigned long)&((union swap_header *)0)->x)
|
|
|
|
|
#define MAX_SWAP_BADPAGES \
|
|
|
|
|
((__swapoffset(magic.magic) - __swapoffset(info.badpages)) / sizeof(int))
|
|
|
|
|
|
|
|
|
|
enum {
|
|
|
|
|
SWP_USED = (1 << 0), /* is slot in swap_info[] used? */
|
|
|
|
|
SWP_WRITEOK = (1 << 1), /* ok to write to this swap? */
|
swap: discard while swapping only if SWAP_FLAG_DISCARD_PAGES
Considering the use cases where the swap device supports discard:
a) and can do it quickly;
b) but it's slow to do in small granularities (or concurrent with other
I/O);
c) but the implementation is so horrendous that you don't even want to
send one down;
And assuming that the sysadmin considers it useful to send the discards down
at all, we would (probably) want the following solutions:
i. do the fine-grained discards for freed swap pages, if device is
capable of doing so optimally;
ii. do single-time (batched) swap area discards, either at swapon
or via something like fstrim (not implemented yet);
iii. allow doing both single-time and fine-grained discards; or
iv. turn it off completely (default behavior)
As implemented today, one can only enable/disable discards for swap, but
one cannot select, for instance, solution (ii) on a swap device like (b)
even though the single-time discard is regarded to be interesting, or
necessary to the workload because it would imply (1), and the device is
not capable of performing it optimally.
This patch addresses the scenario depicted above by introducing a way to
ensure the (probably) wanted solutions (i, ii, iii and iv) can be flexibly
flagged through swapon(8) to allow a sysadmin to select the best suitable
swap discard policy accordingly to system constraints.
This patch introduces SWAP_FLAG_DISCARD_PAGES and SWAP_FLAG_DISCARD_ONCE
new flags to allow more flexibe swap discard policies being flagged
through swapon(8). The default behavior is to keep both single-time, or
batched, area discards (SWAP_FLAG_DISCARD_ONCE) and fine-grained discards
for page-clusters (SWAP_FLAG_DISCARD_PAGES) enabled, in order to keep
consistentcy with older kernel behavior, as well as maintain compatibility
with older swapon(8). However, through the new introduced flags the best
suitable discard policy can be selected accordingly to any given swap
device constraint.
[akpm@linux-foundation.org: tweak comments]
Signed-off-by: Rafael Aquini <aquini@redhat.com>
Acked-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Shaohua Li <shli@kernel.org>
Cc: Karel Zak <kzak@redhat.com>
Cc: Jeff Moyer <jmoyer@redhat.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-07-04 02:02:46 +04:00
|
|
|
SWP_DISCARDABLE = (1 << 2), /* blkdev support discard */
|
2009-01-07 01:39:53 +03:00
|
|
|
SWP_DISCARDING = (1 << 3), /* now discarding a free cluster */
|
2009-01-07 01:39:54 +03:00
|
|
|
SWP_SOLIDSTATE = (1 << 4), /* blkdev seeks are cheap */
|
swap_info: swap count continuations
Swap is duplicated (reference count incremented by one) whenever the same
swap page is inserted into another mm (when forking finds a swap entry in
place of a pte, or when reclaim unmaps a pte to insert the swap entry).
swap_info_struct's vmalloc'ed swap_map is the array of these reference
counts: but what happens when the unsigned short (or unsigned char since
the preceding patch) is full? (and its high bit is kept for a cache flag)
We then lose track of it, never freeing, leaving it in use until swapoff:
at which point we _hope_ that a single pass will have found all instances,
assume there are no more, and will lose user data if we're wrong.
Swapping of KSM pages has not yet been enabled; but it is implemented,
and makes it very easy for a user to overflow the maximum swap count:
possible with ordinary process pages, but unlikely, even when pid_max
has been raised from PID_MAX_DEFAULT.
This patch implements swap count continuations: when the count overflows,
a continuation page is allocated and linked to the original vmalloc'ed
map page, and this used to hold the continuation counts for that entry
and its neighbours. These continuation pages are seldom referenced:
the common paths all work on the original swap_map, only referring to
a continuation page when the low "digit" of a count is incremented or
decremented through SWAP_MAP_MAX.
Signed-off-by: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-12-15 04:58:46 +03:00
|
|
|
SWP_CONTINUED = (1 << 5), /* swap_map has count continuation */
|
2010-05-17 09:32:42 +04:00
|
|
|
SWP_BLKDEV = (1 << 6), /* its a block device */
|
2012-08-01 03:44:55 +04:00
|
|
|
SWP_FILE = (1 << 7), /* set after swap_activate success */
|
swap: discard while swapping only if SWAP_FLAG_DISCARD_PAGES
Considering the use cases where the swap device supports discard:
a) and can do it quickly;
b) but it's slow to do in small granularities (or concurrent with other
I/O);
c) but the implementation is so horrendous that you don't even want to
send one down;
And assuming that the sysadmin considers it useful to send the discards down
at all, we would (probably) want the following solutions:
i. do the fine-grained discards for freed swap pages, if device is
capable of doing so optimally;
ii. do single-time (batched) swap area discards, either at swapon
or via something like fstrim (not implemented yet);
iii. allow doing both single-time and fine-grained discards; or
iv. turn it off completely (default behavior)
As implemented today, one can only enable/disable discards for swap, but
one cannot select, for instance, solution (ii) on a swap device like (b)
even though the single-time discard is regarded to be interesting, or
necessary to the workload because it would imply (1), and the device is
not capable of performing it optimally.
This patch addresses the scenario depicted above by introducing a way to
ensure the (probably) wanted solutions (i, ii, iii and iv) can be flexibly
flagged through swapon(8) to allow a sysadmin to select the best suitable
swap discard policy accordingly to system constraints.
This patch introduces SWAP_FLAG_DISCARD_PAGES and SWAP_FLAG_DISCARD_ONCE
new flags to allow more flexibe swap discard policies being flagged
through swapon(8). The default behavior is to keep both single-time, or
batched, area discards (SWAP_FLAG_DISCARD_ONCE) and fine-grained discards
for page-clusters (SWAP_FLAG_DISCARD_PAGES) enabled, in order to keep
consistentcy with older kernel behavior, as well as maintain compatibility
with older swapon(8). However, through the new introduced flags the best
suitable discard policy can be selected accordingly to any given swap
device constraint.
[akpm@linux-foundation.org: tweak comments]
Signed-off-by: Rafael Aquini <aquini@redhat.com>
Acked-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Shaohua Li <shli@kernel.org>
Cc: Karel Zak <kzak@redhat.com>
Cc: Jeff Moyer <jmoyer@redhat.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-07-04 02:02:46 +04:00
|
|
|
SWP_AREA_DISCARD = (1 << 8), /* single-time swap area discards */
|
|
|
|
|
SWP_PAGE_DISCARD = (1 << 9), /* freed swap page-cluster discards */
|
mm: support anonymous stable page
During developemnt for zram-swap asynchronous writeback, I found strange
corruption of compressed page, resulting in:
Modules linked in: zram(E)
CPU: 3 PID: 1520 Comm: zramd-1 Tainted: G E 4.8.0-mm1-00320-ge0d4894c9c38-dirty #3274
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014
task: ffff88007620b840 task.stack: ffff880078090000
RIP: set_freeobj.part.43+0x1c/0x1f
RSP: 0018:ffff880078093ca8 EFLAGS: 00010246
RAX: 0000000000000018 RBX: ffff880076798d88 RCX: ffffffff81c408c8
RDX: 0000000000000018 RSI: 0000000000000000 RDI: 0000000000000246
RBP: ffff880078093cb0 R08: 0000000000000000 R09: 0000000000000000
R10: ffff88005bc43030 R11: 0000000000001df3 R12: ffff880076798d88
R13: 000000000005bc43 R14: ffff88007819d1b8 R15: 0000000000000001
FS: 0000000000000000(0000) GS:ffff88007e380000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007fc934048f20 CR3: 0000000077b01000 CR4: 00000000000406e0
Call Trace:
obj_malloc+0x22b/0x260
zs_malloc+0x1e4/0x580
zram_bvec_rw+0x4cd/0x830 [zram]
page_requests_rw+0x9c/0x130 [zram]
zram_thread+0xe6/0x173 [zram]
kthread+0xca/0xe0
ret_from_fork+0x25/0x30
With investigation, it reveals currently stable page doesn't support
anonymous page. IOW, reuse_swap_page can reuse the page without waiting
writeback completion so it can overwrite page zram is compressing.
Unfortunately, zram has used per-cpu stream feature from v4.7.
It aims for increasing cache hit ratio of scratch buffer for
compressing. Downside of that approach is that zram should ask
memory space for compressed page in per-cpu context which requires
stricted gfp flag which could be failed. If so, it retries to
allocate memory space out of per-cpu context so it could get memory
this time and compress the data again, copies it to the memory space.
In this scenario, zram assumes the data should never be changed
but it is not true unless stable page supports. So, If the data is
changed under us, zram can make buffer overrun because second
compression size could be bigger than one we got in previous trial
and blindly, copy bigger size object to smaller buffer which is
buffer overrun. The overrun breaks zsmalloc free object chaining
so system goes crash like above.
I think below is same problem.
https://bugzilla.suse.com/show_bug.cgi?id=997574
Unfortunately, reuse_swap_page should be atomic so that we cannot wait on
writeback in there so the approach in this patch is simply return false if
we found it needs stable page. Although it increases memory footprint
temporarily, it happens rarely and it should be reclaimed easily althoug
it happened. Also, It would be better than waiting of IO completion,
which is critial path for application latency.
Fixes: da9556a2367c ("zram: user per-cpu compression streams")
Link: http://lkml.kernel.org/r/20161120233015.GA14113@bbox
Link: http://lkml.kernel.org/r/1482366980-3782-2-git-send-email-minchan@kernel.org
Signed-off-by: Minchan Kim <minchan@kernel.org>
Acked-by: Hugh Dickins <hughd@google.com>
Cc: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Cc: Darrick J. Wong <darrick.wong@oracle.com>
Cc: Takashi Iwai <tiwai@suse.de>
Cc: Hyeoncheol Lee <cheol.lee@lge.com>
Cc: <yjay.kim@lge.com>
Cc: Sangseok Lee <sangseok.lee@lge.com>
Cc: <stable@vger.kernel.org> [4.7+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-01-11 03:58:15 +03:00
|
|
|
SWP_STABLE_WRITES = (1 << 10), /* no overwrite PG_writeback pages */
|
[PATCH] swap: scan_swap_map drop swap_device_lock
get_swap_page has often shown up on latency traces, doing lengthy scans while
holding two spinlocks. swap_list_lock is already dropped, now scan_swap_map
drop swap_device_lock before scanning the swap_map.
While scanning for an empty cluster, don't worry that racing tasks may
allocate what was free and free what was allocated; but when allocating an
entry, check it's still free after retaking the lock. Avoid dropping the lock
in the expected common path. No barriers beyond the locks, just let the
cookie crumble; highest_bit limit is volatile, but benign.
Guard against swapoff: must check SWP_WRITEOK before allocating, must raise
SWP_SCANNING reference count while in scan_swap_map, swapoff wait for that to
fall - just use schedule_timeout, we don't want to burden scan_swap_map
itself, and it's very unlikely that anyone can really still be in
scan_swap_map once swapoff gets this far.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-04 02:54:39 +04:00
|
|
|
/* add others here before... */
|
mm: support anonymous stable page
During developemnt for zram-swap asynchronous writeback, I found strange
corruption of compressed page, resulting in:
Modules linked in: zram(E)
CPU: 3 PID: 1520 Comm: zramd-1 Tainted: G E 4.8.0-mm1-00320-ge0d4894c9c38-dirty #3274
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014
task: ffff88007620b840 task.stack: ffff880078090000
RIP: set_freeobj.part.43+0x1c/0x1f
RSP: 0018:ffff880078093ca8 EFLAGS: 00010246
RAX: 0000000000000018 RBX: ffff880076798d88 RCX: ffffffff81c408c8
RDX: 0000000000000018 RSI: 0000000000000000 RDI: 0000000000000246
RBP: ffff880078093cb0 R08: 0000000000000000 R09: 0000000000000000
R10: ffff88005bc43030 R11: 0000000000001df3 R12: ffff880076798d88
R13: 000000000005bc43 R14: ffff88007819d1b8 R15: 0000000000000001
FS: 0000000000000000(0000) GS:ffff88007e380000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007fc934048f20 CR3: 0000000077b01000 CR4: 00000000000406e0
Call Trace:
obj_malloc+0x22b/0x260
zs_malloc+0x1e4/0x580
zram_bvec_rw+0x4cd/0x830 [zram]
page_requests_rw+0x9c/0x130 [zram]
zram_thread+0xe6/0x173 [zram]
kthread+0xca/0xe0
ret_from_fork+0x25/0x30
With investigation, it reveals currently stable page doesn't support
anonymous page. IOW, reuse_swap_page can reuse the page without waiting
writeback completion so it can overwrite page zram is compressing.
Unfortunately, zram has used per-cpu stream feature from v4.7.
It aims for increasing cache hit ratio of scratch buffer for
compressing. Downside of that approach is that zram should ask
memory space for compressed page in per-cpu context which requires
stricted gfp flag which could be failed. If so, it retries to
allocate memory space out of per-cpu context so it could get memory
this time and compress the data again, copies it to the memory space.
In this scenario, zram assumes the data should never be changed
but it is not true unless stable page supports. So, If the data is
changed under us, zram can make buffer overrun because second
compression size could be bigger than one we got in previous trial
and blindly, copy bigger size object to smaller buffer which is
buffer overrun. The overrun breaks zsmalloc free object chaining
so system goes crash like above.
I think below is same problem.
https://bugzilla.suse.com/show_bug.cgi?id=997574
Unfortunately, reuse_swap_page should be atomic so that we cannot wait on
writeback in there so the approach in this patch is simply return false if
we found it needs stable page. Although it increases memory footprint
temporarily, it happens rarely and it should be reclaimed easily althoug
it happened. Also, It would be better than waiting of IO completion,
which is critial path for application latency.
Fixes: da9556a2367c ("zram: user per-cpu compression streams")
Link: http://lkml.kernel.org/r/20161120233015.GA14113@bbox
Link: http://lkml.kernel.org/r/1482366980-3782-2-git-send-email-minchan@kernel.org
Signed-off-by: Minchan Kim <minchan@kernel.org>
Acked-by: Hugh Dickins <hughd@google.com>
Cc: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Cc: Darrick J. Wong <darrick.wong@oracle.com>
Cc: Takashi Iwai <tiwai@suse.de>
Cc: Hyeoncheol Lee <cheol.lee@lge.com>
Cc: <yjay.kim@lge.com>
Cc: Sangseok Lee <sangseok.lee@lge.com>
Cc: <stable@vger.kernel.org> [4.7+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-01-11 03:58:15 +03:00
|
|
|
SWP_SCANNING = (1 << 11), /* refcount in scan_swap_map */
|
2005-04-17 02:20:36 +04:00
|
|
|
};
|
|
|
|
|
|
2013-02-23 04:32:12 +04:00
|
|
|
#define SWAP_CLUSTER_MAX 32UL
|
2010-05-25 01:32:27 +04:00
|
|
|
#define COMPACT_CLUSTER_MAX SWAP_CLUSTER_MAX
|
2005-04-17 02:20:36 +04:00
|
|
|
|
swap_info: swap count continuations
Swap is duplicated (reference count incremented by one) whenever the same
swap page is inserted into another mm (when forking finds a swap entry in
place of a pte, or when reclaim unmaps a pte to insert the swap entry).
swap_info_struct's vmalloc'ed swap_map is the array of these reference
counts: but what happens when the unsigned short (or unsigned char since
the preceding patch) is full? (and its high bit is kept for a cache flag)
We then lose track of it, never freeing, leaving it in use until swapoff:
at which point we _hope_ that a single pass will have found all instances,
assume there are no more, and will lose user data if we're wrong.
Swapping of KSM pages has not yet been enabled; but it is implemented,
and makes it very easy for a user to overflow the maximum swap count:
possible with ordinary process pages, but unlikely, even when pid_max
has been raised from PID_MAX_DEFAULT.
This patch implements swap count continuations: when the count overflows,
a continuation page is allocated and linked to the original vmalloc'ed
map page, and this used to hold the continuation counts for that entry
and its neighbours. These continuation pages are seldom referenced:
the common paths all work on the original swap_map, only referring to
a continuation page when the low "digit" of a count is incremented or
decremented through SWAP_MAP_MAX.
Signed-off-by: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-12-15 04:58:46 +03:00
|
|
|
#define SWAP_MAP_MAX 0x3e /* Max duplication count, in first swap_map */
|
|
|
|
|
#define SWAP_MAP_BAD 0x3f /* Note pageblock is bad, in first swap_map */
|
|
|
|
|
#define SWAP_HAS_CACHE 0x40 /* Flag page is cached, in first swap_map */
|
|
|
|
|
#define SWAP_CONT_MAX 0x7f /* Max count, in each swap_map continuation */
|
|
|
|
|
#define COUNT_CONTINUED 0x80 /* See swap_map continuation for full count */
|
2009-12-15 04:58:47 +03:00
|
|
|
#define SWAP_MAP_SHMEM 0xbf /* Owned by shmem/tmpfs, in first swap_map */
|
2009-12-15 04:58:44 +03:00
|
|
|
|
swap: change block allocation algorithm for SSD
I'm using a fast SSD to do swap. scan_swap_map() sometimes uses up to
20~30% CPU time (when cluster is hard to find, the CPU time can be up to
80%), which becomes a bottleneck. scan_swap_map() scans a byte array to
search a 256 page cluster, which is very slow.
Here I introduced a simple algorithm to search cluster. Since we only
care about 256 pages cluster, we can just use a counter to track if a
cluster is free. Every 256 pages use one int to store the counter. If
the counter of a cluster is 0, the cluster is free. All free clusters
will be added to a list, so searching cluster is very efficient. With
this, scap_swap_map() overhead disappears.
This might help low end SD card swap too. Because if the cluster is
aligned, SD firmware can do flash erase more efficiently.
We only enable the algorithm for SSD. Hard disk swap isn't fast enough
and has downside with the algorithm which might introduce regression (see
below).
The patch slightly changes which cluster is choosen. It always adds free
cluster to list tail. This can help wear leveling for low end SSD too.
And if no cluster found, the scan_swap_map() will do search from the end
of last cluster. So if no cluster found, the scan_swap_map() will do
search from the end of last free cluster, which is random. For SSD, this
isn't a problem at all.
Another downside is the cluster must be aligned to 256 pages, which will
reduce the chance to find a cluster. I would expect this isn't a big
problem for SSD because of the non-seek penality. (And this is the reason
I only enable the algorithm for SSD).
Signed-off-by: Shaohua Li <shli@fusionio.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Kyungmin Park <kmpark@infradead.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Rafael Aquini <aquini@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-09-12 01:20:28 +04:00
|
|
|
/*
|
|
|
|
|
* We use this to track usage of a cluster. A cluster is a block of swap disk
|
|
|
|
|
* space with SWAPFILE_CLUSTER pages long and naturally aligns in disk. All
|
|
|
|
|
* free clusters are organized into a list. We fetch an entry from the list to
|
|
|
|
|
* get a free cluster.
|
|
|
|
|
*
|
|
|
|
|
* The data field stores next cluster if the cluster is free or cluster usage
|
|
|
|
|
* counter otherwise. The flags field determines if a cluster is free. This is
|
|
|
|
|
* protected by swap_info_struct.lock.
|
|
|
|
|
*/
|
|
|
|
|
struct swap_cluster_info {
|
mm/swap: add cluster lock
This patch is to reduce the lock contention of swap_info_struct->lock
via using a more fine grained lock in swap_cluster_info for some swap
operations. swap_info_struct->lock is heavily contended if multiple
processes reclaim pages simultaneously. Because there is only one lock
for each swap device. While in common configuration, there is only one
or several swap devices in the system. The lock protects almost all
swap related operations.
In fact, many swap operations only access one element of
swap_info_struct->swap_map array. And there is no dependency between
different elements of swap_info_struct->swap_map. So a fine grained
lock can be used to allow parallel access to the different elements of
swap_info_struct->swap_map.
In this patch, a spinlock is added to swap_cluster_info to protect the
elements of swap_info_struct->swap_map in the swap cluster and the
fields of swap_cluster_info. This reduced locking contention for
swap_info_struct->swap_map access greatly.
Because of the added spinlock, the size of swap_cluster_info increases
from 4 bytes to 8 bytes on the 64 bit and 32 bit system. This will use
additional 4k RAM for every 1G swap space.
Because the size of swap_cluster_info is much smaller than the size of
the cache line (8 vs 64 on x86_64 architecture), there may be false
cache line sharing between spinlocks in swap_cluster_info. To avoid the
false sharing in the first round of the swap cluster allocation, the
order of the swap clusters in the free clusters list is changed. So
that, the swap_cluster_info sharing the same cache line will be placed
as far as possible. After the first round of allocation, the order of
the clusters in free clusters list is expected to be random. So the
false sharing should be not serious.
Compared with a previous implementation using bit_spin_lock, the
sequential swap out throughput improved about 3.2%. Test was done on a
Xeon E5 v3 system. The swap device used is a RAM simulated PMEM
(persistent memory) device. To test the sequential swapping out, the
test case created 32 processes, which sequentially allocate and write to
the anonymous pages until the RAM and part of the swap device is used.
[ying.huang@intel.com: v5]
Link: http://lkml.kernel.org/r/878tqeuuic.fsf_-_@yhuang-dev.intel.com
[minchan@kernel.org: initialize spinlock for swap_cluster_info]
Link: http://lkml.kernel.org/r/1486434945-29753-1-git-send-email-minchan@kernel.org
[hughd@google.com: annotate nested locking for cluster lock]
Link: http://lkml.kernel.org/r/alpine.LSU.2.11.1702161050540.21773@eggly.anvils
Link: http://lkml.kernel.org/r/dbb860bbd825b1aaba18988015e8963f263c3f0d.1484082593.git.tim.c.chen@linux.intel.com
Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
Signed-off-by: Tim Chen <tim.c.chen@linux.intel.com>
Signed-off-by: Minchan Kim <minchan@kernel.org>
Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: Aaron Lu <aaron.lu@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Christian Borntraeger <borntraeger@de.ibm.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: Huang Ying <ying.huang@intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Jonathan Corbet <corbet@lwn.net> escreveu:
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Shaohua Li <shli@kernel.org>
Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-02-23 02:45:22 +03:00
|
|
|
spinlock_t lock; /*
|
|
|
|
|
* Protect swap_cluster_info fields
|
|
|
|
|
* and swap_info_struct->swap_map
|
|
|
|
|
* elements correspond to the swap
|
|
|
|
|
* cluster
|
|
|
|
|
*/
|
swap: change block allocation algorithm for SSD
I'm using a fast SSD to do swap. scan_swap_map() sometimes uses up to
20~30% CPU time (when cluster is hard to find, the CPU time can be up to
80%), which becomes a bottleneck. scan_swap_map() scans a byte array to
search a 256 page cluster, which is very slow.
Here I introduced a simple algorithm to search cluster. Since we only
care about 256 pages cluster, we can just use a counter to track if a
cluster is free. Every 256 pages use one int to store the counter. If
the counter of a cluster is 0, the cluster is free. All free clusters
will be added to a list, so searching cluster is very efficient. With
this, scap_swap_map() overhead disappears.
This might help low end SD card swap too. Because if the cluster is
aligned, SD firmware can do flash erase more efficiently.
We only enable the algorithm for SSD. Hard disk swap isn't fast enough
and has downside with the algorithm which might introduce regression (see
below).
The patch slightly changes which cluster is choosen. It always adds free
cluster to list tail. This can help wear leveling for low end SSD too.
And if no cluster found, the scan_swap_map() will do search from the end
of last cluster. So if no cluster found, the scan_swap_map() will do
search from the end of last free cluster, which is random. For SSD, this
isn't a problem at all.
Another downside is the cluster must be aligned to 256 pages, which will
reduce the chance to find a cluster. I would expect this isn't a big
problem for SSD because of the non-seek penality. (And this is the reason
I only enable the algorithm for SSD).
Signed-off-by: Shaohua Li <shli@fusionio.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Kyungmin Park <kmpark@infradead.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Rafael Aquini <aquini@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-09-12 01:20:28 +04:00
|
|
|
unsigned int data:24;
|
|
|
|
|
unsigned int flags:8;
|
|
|
|
|
};
|
|
|
|
|
#define CLUSTER_FLAG_FREE 1 /* This cluster is free */
|
|
|
|
|
#define CLUSTER_FLAG_NEXT_NULL 2 /* This cluster has no next cluster */
|
mm, THP, swap: support to reclaim swap space for THP swapped out
The normal swap slot reclaiming can be done when the swap count reaches
SWAP_HAS_CACHE. But for the swap slot which is backing a THP, all swap
slots backing one THP must be reclaimed together, because the swap slot
may be used again when the THP is swapped out again later. So the swap
slots backing one THP can be reclaimed together when the swap count for
all swap slots for the THP reached SWAP_HAS_CACHE. In the patch, the
functions to check whether the swap count for all swap slots backing one
THP reached SWAP_HAS_CACHE are implemented and used when checking
whether a swap slot can be reclaimed.
To make it easier to determine whether a swap slot is backing a THP, a
new swap cluster flag named CLUSTER_FLAG_HUGE is added to mark a swap
cluster which is backing a THP (Transparent Huge Page). Because THP
swap in as a whole isn't supported now. After deleting the THP from the
swap cache (for example, swapping out finished), the CLUSTER_FLAG_HUGE
flag will be cleared. So that, the normal pages inside THP can be
swapped in individually.
[ying.huang@intel.com: fix swap_page_trans_huge_swapped on HDD]
Link: http://lkml.kernel.org/r/874ltsm0bi.fsf@yhuang-dev.intel.com
Link: http://lkml.kernel.org/r/20170724051840.2309-3-ying.huang@intel.com
Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
Acked-by: Rik van Riel <riel@redhat.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Shaohua Li <shli@kernel.org>
Cc: "Kirill A . Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Ross Zwisler <ross.zwisler@intel.com> [for brd.c, zram_drv.c, pmem.c]
Cc: Vishal L Verma <vishal.l.verma@intel.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-09-07 02:22:16 +03:00
|
|
|
#define CLUSTER_FLAG_HUGE 4 /* This cluster is backing a transparent huge page */
|
swap: change block allocation algorithm for SSD
I'm using a fast SSD to do swap. scan_swap_map() sometimes uses up to
20~30% CPU time (when cluster is hard to find, the CPU time can be up to
80%), which becomes a bottleneck. scan_swap_map() scans a byte array to
search a 256 page cluster, which is very slow.
Here I introduced a simple algorithm to search cluster. Since we only
care about 256 pages cluster, we can just use a counter to track if a
cluster is free. Every 256 pages use one int to store the counter. If
the counter of a cluster is 0, the cluster is free. All free clusters
will be added to a list, so searching cluster is very efficient. With
this, scap_swap_map() overhead disappears.
This might help low end SD card swap too. Because if the cluster is
aligned, SD firmware can do flash erase more efficiently.
We only enable the algorithm for SSD. Hard disk swap isn't fast enough
and has downside with the algorithm which might introduce regression (see
below).
The patch slightly changes which cluster is choosen. It always adds free
cluster to list tail. This can help wear leveling for low end SSD too.
And if no cluster found, the scan_swap_map() will do search from the end
of last cluster. So if no cluster found, the scan_swap_map() will do
search from the end of last free cluster, which is random. For SSD, this
isn't a problem at all.
Another downside is the cluster must be aligned to 256 pages, which will
reduce the chance to find a cluster. I would expect this isn't a big
problem for SSD because of the non-seek penality. (And this is the reason
I only enable the algorithm for SSD).
Signed-off-by: Shaohua Li <shli@fusionio.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Kyungmin Park <kmpark@infradead.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Rafael Aquini <aquini@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-09-12 01:20:28 +04:00
|
|
|
|
swap: make cluster allocation per-cpu
swap cluster allocation is to get better request merge to improve
performance. But the cluster is shared globally, if multiple tasks are
doing swap, this will cause interleave disk access. While multiple tasks
swap is quite common, for example, each numa node has a kswapd thread
doing swap and multiple threads/processes doing direct page reclaim.
ioscheduler can't help too much here, because tasks don't send swapout IO
down to block layer in the meantime. Block layer does merge some IOs, but
a lot not, depending on how many tasks are doing swapout concurrently. In
practice, I've seen a lot of small size IO in swapout workloads.
We makes the cluster allocation per-cpu here. The interleave disk access
issue goes away. All tasks swapout to their own cluster, so swapout will
become sequential, which can be easily merged to big size IO. If one CPU
can't get its per-cpu cluster (for example, there is no free cluster
anymore in the swap), it will fallback to scan swap_map. The CPU can
still continue swap. We don't need recycle free swap entries of other
CPUs.
In my test (swap to a 2-disk raid0 partition), this improves around 10%
swapout throughput, and request size is increased significantly.
How does this impact swap readahead is uncertain though. On one side,
page reclaim always isolates and swaps several adjancent pages, this will
make page reclaim write the pages sequentially and benefit readahead. On
the other side, several CPU write pages interleave means the pages don't
live _sequentially_ but relatively _near_. In the per-cpu allocation
case, if adjancent pages are written by different cpus, they will live
relatively _far_. So how this impacts swap readahead depends on how many
pages page reclaim isolates and swaps one time. If the number is big,
this patch will benefit swap readahead. Of course, this is about
sequential access pattern. The patch has no impact for random access
pattern, because the new cluster allocation algorithm is just for SSD.
Alternative solution is organizing swap layout to be per-mm instead of
this per-cpu approach. In the per-mm layout, we allocate a disk range for
each mm, so pages of one mm live in swap disk adjacently. per-mm layout
has potential issues of lock contention if multiple reclaimers are swap
pages from one mm. For a sequential workload, per-mm layout is better to
implement swap readahead, because pages from the mm are adjacent in disk.
But per-cpu layout isn't very bad in this workload, as page reclaim always
isolates and swaps several pages one time, such pages will still live in
disk sequentially and readahead can utilize this. For a random workload,
per-mm layout isn't beneficial of request merge, because it's quite
possible pages from different mm are swapout in the meantime and IO can't
be merged in per-mm layout. while with per-cpu layout we can merge
requests from any mm. Considering random workload is more popular in
workloads with swap (and per-cpu approach isn't too bad for sequential
workload too), I'm choosing per-cpu layout.
[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Shaohua Li <shli@fusionio.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Kyungmin Park <kmpark@infradead.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Rafael Aquini <aquini@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-09-12 01:20:32 +04:00
|
|
|
/*
|
|
|
|
|
* We assign a cluster to each CPU, so each CPU can allocate swap entry from
|
|
|
|
|
* its own cluster and swapout sequentially. The purpose is to optimize swapout
|
|
|
|
|
* throughput.
|
|
|
|
|
*/
|
|
|
|
|
struct percpu_cluster {
|
|
|
|
|
struct swap_cluster_info index; /* Current cluster index */
|
|
|
|
|
unsigned int next; /* Likely next allocation offset */
|
|
|
|
|
};
|
|
|
|
|
|
2016-10-08 02:58:42 +03:00
|
|
|
struct swap_cluster_list {
|
|
|
|
|
struct swap_cluster_info head;
|
|
|
|
|
struct swap_cluster_info tail;
|
|
|
|
|
};
|
|
|
|
|
|
2005-04-17 02:20:36 +04:00
|
|
|
/*
|
|
|
|
|
* The in-memory structure used to track swap areas.
|
|
|
|
|
*/
|
|
|
|
|
struct swap_info_struct {
|
2009-12-15 04:58:41 +03:00
|
|
|
unsigned long flags; /* SWP_USED etc: see above */
|
|
|
|
|
signed short prio; /* swap priority of this type */
|
swap: change swap_list_head to plist, add swap_avail_head
Originally get_swap_page() started iterating through the singly-linked
list of swap_info_structs using swap_list.next or highest_priority_index,
which both were intended to point to the highest priority active swap
target that was not full. The first patch in this series changed the
singly-linked list to a doubly-linked list, and removed the logic to start
at the highest priority non-full entry; it starts scanning at the highest
priority entry each time, even if the entry is full.
Replace the manually ordered swap_list_head with a plist, swap_active_head.
Add a new plist, swap_avail_head. The original swap_active_head plist
contains all active swap_info_structs, as before, while the new
swap_avail_head plist contains only swap_info_structs that are active and
available, i.e. not full. Add a new spinlock, swap_avail_lock, to protect
the swap_avail_head list.
Mel Gorman suggested using plists since they internally handle ordering
the list entries based on priority, which is exactly what swap was doing
manually. All the ordering code is now removed, and swap_info_struct
entries and simply added to their corresponding plist and automatically
ordered correctly.
Using a new plist for available swap_info_structs simplifies and
optimizes get_swap_page(), which no longer has to iterate over full
swap_info_structs. Using a new spinlock for swap_avail_head plist
allows each swap_info_struct to add or remove themselves from the
plist when they become full or not-full; previously they could not
do so because the swap_info_struct->lock is held when they change
from full<->not-full, and the swap_lock protecting the main
swap_active_head must be ordered before any swap_info_struct->lock.
Signed-off-by: Dan Streetman <ddstreet@ieee.org>
Acked-by: Mel Gorman <mgorman@suse.de>
Cc: Shaohua Li <shli@fusionio.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Dan Streetman <ddstreet@ieee.org>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Christian Ehrhardt <ehrhardt@linux.vnet.ibm.com>
Cc: Weijie Yang <weijieut@gmail.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Bob Liu <bob.liu@oracle.com>
Cc: Paul Gortmaker <paul.gortmaker@windriver.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>
2014-06-05 03:09:59 +04:00
|
|
|
struct plist_node list; /* entry in swap_active_head */
|
swap: choose swap device according to numa node
If the system has more than one swap device and swap device has the node
information, we can make use of this information to decide which swap
device to use in get_swap_pages() to get better performance.
The current code uses a priority based list, swap_avail_list, to decide
which swap device to use and if multiple swap devices share the same
priority, they are used round robin. This patch changes the previous
single global swap_avail_list into a per-numa-node list, i.e. for each
numa node, it sees its own priority based list of available swap
devices. Swap device's priority can be promoted on its matching node's
swap_avail_list.
The current swap device's priority is set as: user can set a >=0 value,
or the system will pick one starting from -1 then downwards. The
priority value in the swap_avail_list is the negated value of the swap
device's due to plist being sorted from low to high. The new policy
doesn't change the semantics for priority >=0 cases, the previous
starting from -1 then downwards now becomes starting from -2 then
downwards and -1 is reserved as the promoted value.
Take 4-node EX machine as an example, suppose 4 swap devices are
available, each sit on a different node:
swapA on node 0
swapB on node 1
swapC on node 2
swapD on node 3
After they are all swapped on in the sequence of ABCD.
Current behaviour:
their priorities will be:
swapA: -1
swapB: -2
swapC: -3
swapD: -4
And their position in the global swap_avail_list will be:
swapA -> swapB -> swapC -> swapD
prio:1 prio:2 prio:3 prio:4
New behaviour:
their priorities will be(note that -1 is skipped):
swapA: -2
swapB: -3
swapC: -4
swapD: -5
And their positions in the 4 swap_avail_lists[nid] will be:
swap_avail_lists[0]: /* node 0's available swap device list */
swapA -> swapB -> swapC -> swapD
prio:1 prio:3 prio:4 prio:5
swap_avali_lists[1]: /* node 1's available swap device list */
swapB -> swapA -> swapC -> swapD
prio:1 prio:2 prio:4 prio:5
swap_avail_lists[2]: /* node 2's available swap device list */
swapC -> swapA -> swapB -> swapD
prio:1 prio:2 prio:3 prio:5
swap_avail_lists[3]: /* node 3's available swap device list */
swapD -> swapA -> swapB -> swapC
prio:1 prio:2 prio:3 prio:4
To see the effect of the patch, a test that starts N process, each mmap
a region of anonymous memory and then continually write to it at random
position to trigger both swap in and out is used.
On a 2 node Skylake EP machine with 64GiB memory, two 170GB SSD drives
are used as swap devices with each attached to a different node, the
result is:
runtime=30m/processes=32/total test size=128G/each process mmap region=4G
kernel throughput
vanilla 13306
auto-binding 15169 +14%
runtime=30m/processes=64/total test size=128G/each process mmap region=2G
kernel throughput
vanilla 11885
auto-binding 14879 +25%
[aaron.lu@intel.com: v2]
Link: http://lkml.kernel.org/r/20170814053130.GD2369@aaronlu.sh.intel.com
Link: http://lkml.kernel.org/r/20170816024439.GA10925@aaronlu.sh.intel.com
[akpm@linux-foundation.org: use kmalloc_array()]
Link: http://lkml.kernel.org/r/20170814053130.GD2369@aaronlu.sh.intel.com
Link: http://lkml.kernel.org/r/20170816024439.GA10925@aaronlu.sh.intel.com
Signed-off-by: Aaron Lu <aaron.lu@intel.com>
Cc: "Chen, Tim C" <tim.c.chen@intel.com>
Cc: Huang Ying <ying.huang@intel.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Hugh Dickins <hughd@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-09-07 02:24:57 +03:00
|
|
|
struct plist_node avail_lists[MAX_NUMNODES];/* entry in swap_avail_heads */
|
2009-12-15 04:58:41 +03:00
|
|
|
signed char type; /* strange name for an index */
|
2009-12-15 04:58:48 +03:00
|
|
|
unsigned int max; /* extent of the swap_map */
|
|
|
|
|
unsigned char *swap_map; /* vmalloc'ed array of usage counts */
|
swap: change block allocation algorithm for SSD
I'm using a fast SSD to do swap. scan_swap_map() sometimes uses up to
20~30% CPU time (when cluster is hard to find, the CPU time can be up to
80%), which becomes a bottleneck. scan_swap_map() scans a byte array to
search a 256 page cluster, which is very slow.
Here I introduced a simple algorithm to search cluster. Since we only
care about 256 pages cluster, we can just use a counter to track if a
cluster is free. Every 256 pages use one int to store the counter. If
the counter of a cluster is 0, the cluster is free. All free clusters
will be added to a list, so searching cluster is very efficient. With
this, scap_swap_map() overhead disappears.
This might help low end SD card swap too. Because if the cluster is
aligned, SD firmware can do flash erase more efficiently.
We only enable the algorithm for SSD. Hard disk swap isn't fast enough
and has downside with the algorithm which might introduce regression (see
below).
The patch slightly changes which cluster is choosen. It always adds free
cluster to list tail. This can help wear leveling for low end SSD too.
And if no cluster found, the scan_swap_map() will do search from the end
of last cluster. So if no cluster found, the scan_swap_map() will do
search from the end of last free cluster, which is random. For SSD, this
isn't a problem at all.
Another downside is the cluster must be aligned to 256 pages, which will
reduce the chance to find a cluster. I would expect this isn't a big
problem for SSD because of the non-seek penality. (And this is the reason
I only enable the algorithm for SSD).
Signed-off-by: Shaohua Li <shli@fusionio.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Kyungmin Park <kmpark@infradead.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Rafael Aquini <aquini@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-09-12 01:20:28 +04:00
|
|
|
struct swap_cluster_info *cluster_info; /* cluster info. Only for SSD */
|
2016-10-08 02:58:42 +03:00
|
|
|
struct swap_cluster_list free_clusters; /* free clusters list */
|
2009-12-15 04:58:48 +03:00
|
|
|
unsigned int lowest_bit; /* index of first free in swap_map */
|
|
|
|
|
unsigned int highest_bit; /* index of last free in swap_map */
|
|
|
|
|
unsigned int pages; /* total of usable pages of swap */
|
|
|
|
|
unsigned int inuse_pages; /* number of those currently in use */
|
|
|
|
|
unsigned int cluster_next; /* likely index for next allocation */
|
|
|
|
|
unsigned int cluster_nr; /* countdown to next cluster search */
|
swap: make cluster allocation per-cpu
swap cluster allocation is to get better request merge to improve
performance. But the cluster is shared globally, if multiple tasks are
doing swap, this will cause interleave disk access. While multiple tasks
swap is quite common, for example, each numa node has a kswapd thread
doing swap and multiple threads/processes doing direct page reclaim.
ioscheduler can't help too much here, because tasks don't send swapout IO
down to block layer in the meantime. Block layer does merge some IOs, but
a lot not, depending on how many tasks are doing swapout concurrently. In
practice, I've seen a lot of small size IO in swapout workloads.
We makes the cluster allocation per-cpu here. The interleave disk access
issue goes away. All tasks swapout to their own cluster, so swapout will
become sequential, which can be easily merged to big size IO. If one CPU
can't get its per-cpu cluster (for example, there is no free cluster
anymore in the swap), it will fallback to scan swap_map. The CPU can
still continue swap. We don't need recycle free swap entries of other
CPUs.
In my test (swap to a 2-disk raid0 partition), this improves around 10%
swapout throughput, and request size is increased significantly.
How does this impact swap readahead is uncertain though. On one side,
page reclaim always isolates and swaps several adjancent pages, this will
make page reclaim write the pages sequentially and benefit readahead. On
the other side, several CPU write pages interleave means the pages don't
live _sequentially_ but relatively _near_. In the per-cpu allocation
case, if adjancent pages are written by different cpus, they will live
relatively _far_. So how this impacts swap readahead depends on how many
pages page reclaim isolates and swaps one time. If the number is big,
this patch will benefit swap readahead. Of course, this is about
sequential access pattern. The patch has no impact for random access
pattern, because the new cluster allocation algorithm is just for SSD.
Alternative solution is organizing swap layout to be per-mm instead of
this per-cpu approach. In the per-mm layout, we allocate a disk range for
each mm, so pages of one mm live in swap disk adjacently. per-mm layout
has potential issues of lock contention if multiple reclaimers are swap
pages from one mm. For a sequential workload, per-mm layout is better to
implement swap readahead, because pages from the mm are adjacent in disk.
But per-cpu layout isn't very bad in this workload, as page reclaim always
isolates and swaps several pages one time, such pages will still live in
disk sequentially and readahead can utilize this. For a random workload,
per-mm layout isn't beneficial of request merge, because it's quite
possible pages from different mm are swapout in the meantime and IO can't
be merged in per-mm layout. while with per-cpu layout we can merge
requests from any mm. Considering random workload is more popular in
workloads with swap (and per-cpu approach isn't too bad for sequential
workload too), I'm choosing per-cpu layout.
[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Shaohua Li <shli@fusionio.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Kyungmin Park <kmpark@infradead.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Rafael Aquini <aquini@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-09-12 01:20:32 +04:00
|
|
|
struct percpu_cluster __percpu *percpu_cluster; /* per cpu's swap location */
|
2009-12-15 04:58:48 +03:00
|
|
|
struct swap_extent *curr_swap_extent;
|
|
|
|
|
struct swap_extent first_swap_extent;
|
|
|
|
|
struct block_device *bdev; /* swap device or bdev of swap file */
|
|
|
|
|
struct file *swap_file; /* seldom referenced */
|
|
|
|
|
unsigned int old_block_size; /* seldom referenced */
|
2012-04-10 03:08:06 +04:00
|
|
|
#ifdef CONFIG_FRONTSWAP
|
|
|
|
|
unsigned long *frontswap_map; /* frontswap in-use, one bit per page */
|
|
|
|
|
atomic_t frontswap_pages; /* frontswap pages in-use counter */
|
|
|
|
|
#endif
|
swap: add per-partition lock for swapfile
swap_lock is heavily contended when I test swap to 3 fast SSD (even
slightly slower than swap to 2 such SSD). The main contention comes
from swap_info_get(). This patch tries to fix the gap with adding a new
per-partition lock.
Global data like nr_swapfiles, total_swap_pages, least_priority and
swap_list are still protected by swap_lock.
nr_swap_pages is an atomic now, it can be changed without swap_lock. In
theory, it's possible get_swap_page() finds no swap pages but actually
there are free swap pages. But sounds not a big problem.
Accessing partition specific data (like scan_swap_map and so on) is only
protected by swap_info_struct.lock.
Changing swap_info_struct.flags need hold swap_lock and
swap_info_struct.lock, because scan_scan_map() will check it. read the
flags is ok with either the locks hold.
If both swap_lock and swap_info_struct.lock must be hold, we always hold
the former first to avoid deadlock.
swap_entry_free() can change swap_list. To delete that code, we add a
new highest_priority_index. Whenever get_swap_page() is called, we
check it. If it's valid, we use it.
It's a pity get_swap_page() still holds swap_lock(). But in practice,
swap_lock() isn't heavily contended in my test with this patch (or I can
say there are other much more heavier bottlenecks like TLB flush). And
BTW, looks get_swap_page() doesn't really need the lock. We never free
swap_info[] and we check SWAP_WRITEOK flag. The only risk without the
lock is we could swapout to some low priority swap, but we can quickly
recover after several rounds of swap, so sounds not a big deal to me.
But I'd prefer to fix this if it's a real problem.
"swap: make each swap partition have one address_space" improved the
swapout speed from 1.7G/s to 2G/s. This patch further improves the
speed to 2.3G/s, so around 15% improvement. It's a multi-process test,
so TLB flush isn't the biggest bottleneck before the patches.
[arnd@arndb.de: fix it for nommu]
[hughd@google.com: add missing unlock]
[minchan@kernel.org: get rid of lockdep whinge on sys_swapon]
Signed-off-by: Shaohua Li <shli@fusionio.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Seth Jennings <sjenning@linux.vnet.ibm.com>
Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: Xiao Guangrong <xiaoguangrong@linux.vnet.ibm.com>
Cc: Dan Magenheimer <dan.magenheimer@oracle.com>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Hugh Dickins <hughd@google.com>
Signed-off-by: Minchan Kim <minchan@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-23 04:34:38 +04:00
|
|
|
spinlock_t lock; /*
|
|
|
|
|
* protect map scan related fields like
|
|
|
|
|
* swap_map, lowest_bit, highest_bit,
|
|
|
|
|
* inuse_pages, cluster_next,
|
swap: make swap discard async
swap can do cluster discard for SSD, which is good, but there are some
problems here:
1. swap do the discard just before page reclaim gets a swap entry and
writes the disk sectors. This is useless for high end SSD, because an
overwrite to a sector implies a discard to original sector too. A
discard + overwrite == overwrite.
2. the purpose of doing discard is to improve SSD firmware garbage
collection. Idealy we should send discard as early as possible, so
firmware can do something smart. Sending discard just after swap entry
is freed is considered early compared to sending discard before write.
Of course, if workload is already bound to gc speed, sending discard
earlier or later doesn't make
3. block discard is a sync API, which will delay scan_swap_map()
significantly.
4. Write and discard command can be executed parallel in PCIe SSD.
Making swap discard async can make execution more efficiently.
This patch makes swap discard async and moves discard to where swap entry
is freed. Discard and write have no dependence now, so above issues can
be avoided. Idealy we should do discard for any freed sectors, but some
SSD discard is very slow. This patch still does discard for a whole
cluster.
My test does a several round of 'mmap, write, unmap', which will trigger a
lot of swap discard. In a fusionio card, with this patch, the test
runtime is reduced to 18% of the time without it, so around 5.5x faster.
[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Shaohua Li <shli@fusionio.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Kyungmin Park <kmpark@infradead.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Rafael Aquini <aquini@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-09-12 01:20:30 +04:00
|
|
|
* cluster_nr, lowest_alloc,
|
|
|
|
|
* highest_alloc, free/discard cluster
|
|
|
|
|
* list. other fields are only changed
|
|
|
|
|
* at swapon/swapoff, so are protected
|
|
|
|
|
* by swap_lock. changing flags need
|
|
|
|
|
* hold this lock and swap_lock. If
|
|
|
|
|
* both locks need hold, hold swap_lock
|
|
|
|
|
* first.
|
swap: add per-partition lock for swapfile
swap_lock is heavily contended when I test swap to 3 fast SSD (even
slightly slower than swap to 2 such SSD). The main contention comes
from swap_info_get(). This patch tries to fix the gap with adding a new
per-partition lock.
Global data like nr_swapfiles, total_swap_pages, least_priority and
swap_list are still protected by swap_lock.
nr_swap_pages is an atomic now, it can be changed without swap_lock. In
theory, it's possible get_swap_page() finds no swap pages but actually
there are free swap pages. But sounds not a big problem.
Accessing partition specific data (like scan_swap_map and so on) is only
protected by swap_info_struct.lock.
Changing swap_info_struct.flags need hold swap_lock and
swap_info_struct.lock, because scan_scan_map() will check it. read the
flags is ok with either the locks hold.
If both swap_lock and swap_info_struct.lock must be hold, we always hold
the former first to avoid deadlock.
swap_entry_free() can change swap_list. To delete that code, we add a
new highest_priority_index. Whenever get_swap_page() is called, we
check it. If it's valid, we use it.
It's a pity get_swap_page() still holds swap_lock(). But in practice,
swap_lock() isn't heavily contended in my test with this patch (or I can
say there are other much more heavier bottlenecks like TLB flush). And
BTW, looks get_swap_page() doesn't really need the lock. We never free
swap_info[] and we check SWAP_WRITEOK flag. The only risk without the
lock is we could swapout to some low priority swap, but we can quickly
recover after several rounds of swap, so sounds not a big deal to me.
But I'd prefer to fix this if it's a real problem.
"swap: make each swap partition have one address_space" improved the
swapout speed from 1.7G/s to 2G/s. This patch further improves the
speed to 2.3G/s, so around 15% improvement. It's a multi-process test,
so TLB flush isn't the biggest bottleneck before the patches.
[arnd@arndb.de: fix it for nommu]
[hughd@google.com: add missing unlock]
[minchan@kernel.org: get rid of lockdep whinge on sys_swapon]
Signed-off-by: Shaohua Li <shli@fusionio.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Seth Jennings <sjenning@linux.vnet.ibm.com>
Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: Xiao Guangrong <xiaoguangrong@linux.vnet.ibm.com>
Cc: Dan Magenheimer <dan.magenheimer@oracle.com>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Hugh Dickins <hughd@google.com>
Signed-off-by: Minchan Kim <minchan@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-23 04:34:38 +04:00
|
|
|
*/
|
swap: make swap discard async
swap can do cluster discard for SSD, which is good, but there are some
problems here:
1. swap do the discard just before page reclaim gets a swap entry and
writes the disk sectors. This is useless for high end SSD, because an
overwrite to a sector implies a discard to original sector too. A
discard + overwrite == overwrite.
2. the purpose of doing discard is to improve SSD firmware garbage
collection. Idealy we should send discard as early as possible, so
firmware can do something smart. Sending discard just after swap entry
is freed is considered early compared to sending discard before write.
Of course, if workload is already bound to gc speed, sending discard
earlier or later doesn't make
3. block discard is a sync API, which will delay scan_swap_map()
significantly.
4. Write and discard command can be executed parallel in PCIe SSD.
Making swap discard async can make execution more efficiently.
This patch makes swap discard async and moves discard to where swap entry
is freed. Discard and write have no dependence now, so above issues can
be avoided. Idealy we should do discard for any freed sectors, but some
SSD discard is very slow. This patch still does discard for a whole
cluster.
My test does a several round of 'mmap, write, unmap', which will trigger a
lot of swap discard. In a fusionio card, with this patch, the test
runtime is reduced to 18% of the time without it, so around 5.5x faster.
[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Shaohua Li <shli@fusionio.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Kyungmin Park <kmpark@infradead.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Rafael Aquini <aquini@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-09-12 01:20:30 +04:00
|
|
|
struct work_struct discard_work; /* discard worker */
|
2016-10-08 02:58:42 +03:00
|
|
|
struct swap_cluster_list discard_clusters; /* discard clusters list */
|
2005-04-17 02:20:36 +04:00
|
|
|
};
|
|
|
|
|
|
mm, swap: VMA based swap readahead
The swap readahead is an important mechanism to reduce the swap in
latency. Although pure sequential memory access pattern isn't very
popular for anonymous memory, the space locality is still considered
valid.
In the original swap readahead implementation, the consecutive blocks in
swap device are readahead based on the global space locality estimation.
But the consecutive blocks in swap device just reflect the order of page
reclaiming, don't necessarily reflect the access pattern in virtual
memory. And the different tasks in the system may have different access
patterns, which makes the global space locality estimation incorrect.
In this patch, when page fault occurs, the virtual pages near the fault
address will be readahead instead of the swap slots near the fault swap
slot in swap device. This avoid to readahead the unrelated swap slots.
At the same time, the swap readahead is changed to work on per-VMA from
globally. So that the different access patterns of the different VMAs
could be distinguished, and the different readahead policy could be
applied accordingly. The original core readahead detection and scaling
algorithm is reused, because it is an effect algorithm to detect the
space locality.
The test and result is as follow,
Common test condition
=====================
Test Machine: Xeon E5 v3 (2 sockets, 72 threads, 32G RAM) Swap device:
NVMe disk
Micro-benchmark with combined access pattern
============================================
vm-scalability, sequential swap test case, 4 processes to eat 50G
virtual memory space, repeat the sequential memory writing until 300
seconds. The first round writing will trigger swap out, the following
rounds will trigger sequential swap in and out.
At the same time, run vm-scalability random swap test case in
background, 8 processes to eat 30G virtual memory space, repeat the
random memory write until 300 seconds. This will trigger random swap-in
in the background.
This is a combined workload with sequential and random memory accessing
at the same time. The result (for sequential workload) is as follow,
Base Optimized
---- ---------
throughput 345413 KB/s 414029 KB/s (+19.9%)
latency.average 97.14 us 61.06 us (-37.1%)
latency.50th 2 us 1 us
latency.60th 2 us 1 us
latency.70th 98 us 2 us
latency.80th 160 us 2 us
latency.90th 260 us 217 us
latency.95th 346 us 369 us
latency.99th 1.34 ms 1.09 ms
ra_hit% 52.69% 99.98%
The original swap readahead algorithm is confused by the background
random access workload, so readahead hit rate is lower. The VMA-base
readahead algorithm works much better.
Linpack
=======
The test memory size is bigger than RAM to trigger swapping.
Base Optimized
---- ---------
elapsed_time 393.49 s 329.88 s (-16.2%)
ra_hit% 86.21% 98.82%
The score of base and optimized kernel hasn't visible changes. But the
elapsed time reduced and readahead hit rate improved, so the optimized
kernel runs better for startup and tear down stages. And the absolute
value of readahead hit rate is high, shows that the space locality is
still valid in some practical workloads.
Link: http://lkml.kernel.org/r/20170807054038.1843-4-ying.huang@intel.com
Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Shaohua Li <shli@kernel.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Fengguang Wu <fengguang.wu@intel.com>
Cc: Tim Chen <tim.c.chen@intel.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-09-07 02:24:36 +03:00
|
|
|
#ifdef CONFIG_64BIT
|
|
|
|
|
#define SWAP_RA_ORDER_CEILING 5
|
|
|
|
|
#else
|
|
|
|
|
/* Avoid stack overflow, because we need to save part of page table */
|
|
|
|
|
#define SWAP_RA_ORDER_CEILING 3
|
|
|
|
|
#define SWAP_RA_PTE_CACHE_SIZE (1 << SWAP_RA_ORDER_CEILING)
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
struct vma_swap_readahead {
|
|
|
|
|
unsigned short win;
|
|
|
|
|
unsigned short offset;
|
|
|
|
|
unsigned short nr_pte;
|
|
|
|
|
#ifdef CONFIG_64BIT
|
|
|
|
|
pte_t *ptes;
|
|
|
|
|
#else
|
|
|
|
|
pte_t ptes[SWAP_RA_PTE_CACHE_SIZE];
|
|
|
|
|
#endif
|
|
|
|
|
};
|
|
|
|
|
|
2014-04-04 01:47:51 +04:00
|
|
|
/* linux/mm/workingset.c */
|
|
|
|
|
void *workingset_eviction(struct address_space *mapping, struct page *page);
|
|
|
|
|
bool workingset_refault(void *shadow);
|
|
|
|
|
void workingset_activation(struct page *page);
|
2016-12-13 03:43:52 +03:00
|
|
|
void workingset_update_node(struct radix_tree_node *node, void *private);
|
2014-04-04 01:47:51 +04:00
|
|
|
|
2005-04-17 02:20:36 +04:00
|
|
|
/* linux/mm/page_alloc.c */
|
|
|
|
|
extern unsigned long totalram_pages;
|
2006-04-11 09:52:59 +04:00
|
|
|
extern unsigned long totalreserve_pages;
|
2013-02-23 04:35:43 +04:00
|
|
|
extern unsigned long nr_free_buffer_pages(void);
|
|
|
|
|
extern unsigned long nr_free_pagecache_pages(void);
|
2005-04-17 02:20:36 +04:00
|
|
|
|
2017-09-07 02:23:36 +03:00
|
|
|
/* Definition of global_zone_page_state not available yet */
|
|
|
|
|
#define nr_free_pages() global_zone_page_state(NR_FREE_PAGES)
|
2007-02-10 12:43:03 +03:00
|
|
|
|
|
|
|
|
|
2005-04-17 02:20:36 +04:00
|
|
|
/* linux/mm/swap.c */
|
2013-07-04 02:02:34 +04:00
|
|
|
extern void lru_cache_add(struct page *);
|
2014-06-05 03:07:31 +04:00
|
|
|
extern void lru_cache_add_anon(struct page *page);
|
|
|
|
|
extern void lru_cache_add_file(struct page *page);
|
2012-05-30 02:07:09 +04:00
|
|
|
extern void lru_add_page_tail(struct page *page, struct page *page_tail,
|
2013-04-30 02:08:36 +04:00
|
|
|
struct lruvec *lruvec, struct list_head *head);
|
2008-02-14 02:03:15 +03:00
|
|
|
extern void activate_page(struct page *);
|
|
|
|
|
extern void mark_page_accessed(struct page *);
|
2005-04-17 02:20:36 +04:00
|
|
|
extern void lru_add_drain(void);
|
2012-03-22 03:34:06 +04:00
|
|
|
extern void lru_add_drain_cpu(int cpu);
|
2013-09-13 02:13:55 +04:00
|
|
|
extern void lru_add_drain_all(void);
|
2017-07-11 01:50:06 +03:00
|
|
|
extern void lru_add_drain_all_cpuslocked(void);
|
2008-04-28 13:12:38 +04:00
|
|
|
extern void rotate_reclaimable_page(struct page *page);
|
2015-04-16 02:13:26 +03:00
|
|
|
extern void deactivate_file_page(struct page *page);
|
2017-05-04 00:52:29 +03:00
|
|
|
extern void mark_page_lazyfree(struct page *page);
|
2005-04-17 02:20:36 +04:00
|
|
|
extern void swap_setup(void);
|
|
|
|
|
|
Unevictable LRU Infrastructure
When the system contains lots of mlocked or otherwise unevictable pages,
the pageout code (kswapd) can spend lots of time scanning over these
pages. Worse still, the presence of lots of unevictable pages can confuse
kswapd into thinking that more aggressive pageout modes are required,
resulting in all kinds of bad behaviour.
Infrastructure to manage pages excluded from reclaim--i.e., hidden from
vmscan. Based on a patch by Larry Woodman of Red Hat. Reworked to
maintain "unevictable" pages on a separate per-zone LRU list, to "hide"
them from vmscan.
Kosaki Motohiro added the support for the memory controller unevictable
lru list.
Pages on the unevictable list have both PG_unevictable and PG_lru set.
Thus, PG_unevictable is analogous to and mutually exclusive with
PG_active--it specifies which LRU list the page is on.
The unevictable infrastructure is enabled by a new mm Kconfig option
[CONFIG_]UNEVICTABLE_LRU.
A new function 'page_evictable(page, vma)' in vmscan.c tests whether or
not a page may be evictable. Subsequent patches will add the various
!evictable tests. We'll want to keep these tests light-weight for use in
shrink_active_list() and, possibly, the fault path.
To avoid races between tasks putting pages [back] onto an LRU list and
tasks that might be moving the page from non-evictable to evictable state,
the new function 'putback_lru_page()' -- inverse to 'isolate_lru_page()'
-- tests the "evictability" of a page after placing it on the LRU, before
dropping the reference. If the page has become unevictable,
putback_lru_page() will redo the 'putback', thus moving the page to the
unevictable list. This way, we avoid "stranding" evictable pages on the
unevictable list.
[akpm@linux-foundation.org: fix fallout from out-of-order merge]
[riel@redhat.com: fix UNEVICTABLE_LRU and !PROC_PAGE_MONITOR build]
[nishimura@mxp.nes.nec.co.jp: remove redundant mapping check]
[kosaki.motohiro@jp.fujitsu.com: unevictable-lru-infrastructure: putback_lru_page()/unevictable page handling rework]
[kosaki.motohiro@jp.fujitsu.com: kill unnecessary lock_page() in vmscan.c]
[kosaki.motohiro@jp.fujitsu.com: revert migration change of unevictable lru infrastructure]
[kosaki.motohiro@jp.fujitsu.com: revert to unevictable-lru-infrastructure-kconfig-fix.patch]
[kosaki.motohiro@jp.fujitsu.com: restore patch failure of vmstat-unevictable-and-mlocked-pages-vm-events.patch]
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Rik van Riel <riel@redhat.com>
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Debugged-by: Benjamin Kidwell <benjkidwell@yahoo.com>
Signed-off-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-10-19 07:26:39 +04:00
|
|
|
extern void add_page_to_unevictable_list(struct page *page);
|
|
|
|
|
|
mm: memcontrol: rewrite charge API
These patches rework memcg charge lifetime to integrate more naturally
with the lifetime of user pages. This drastically simplifies the code and
reduces charging and uncharging overhead. The most expensive part of
charging and uncharging is the page_cgroup bit spinlock, which is removed
entirely after this series.
Here are the top-10 profile entries of a stress test that reads a 128G
sparse file on a freshly booted box, without even a dedicated cgroup (i.e.
executing in the root memcg). Before:
15.36% cat [kernel.kallsyms] [k] copy_user_generic_string
13.31% cat [kernel.kallsyms] [k] memset
11.48% cat [kernel.kallsyms] [k] do_mpage_readpage
4.23% cat [kernel.kallsyms] [k] get_page_from_freelist
2.38% cat [kernel.kallsyms] [k] put_page
2.32% cat [kernel.kallsyms] [k] __mem_cgroup_commit_charge
2.18% kswapd0 [kernel.kallsyms] [k] __mem_cgroup_uncharge_common
1.92% kswapd0 [kernel.kallsyms] [k] shrink_page_list
1.86% cat [kernel.kallsyms] [k] __radix_tree_lookup
1.62% cat [kernel.kallsyms] [k] __pagevec_lru_add_fn
After:
15.67% cat [kernel.kallsyms] [k] copy_user_generic_string
13.48% cat [kernel.kallsyms] [k] memset
11.42% cat [kernel.kallsyms] [k] do_mpage_readpage
3.98% cat [kernel.kallsyms] [k] get_page_from_freelist
2.46% cat [kernel.kallsyms] [k] put_page
2.13% kswapd0 [kernel.kallsyms] [k] shrink_page_list
1.88% cat [kernel.kallsyms] [k] __radix_tree_lookup
1.67% cat [kernel.kallsyms] [k] __pagevec_lru_add_fn
1.39% kswapd0 [kernel.kallsyms] [k] free_pcppages_bulk
1.30% cat [kernel.kallsyms] [k] kfree
As you can see, the memcg footprint has shrunk quite a bit.
text data bss dec hex filename
37970 9892 400 48262 bc86 mm/memcontrol.o.old
35239 9892 400 45531 b1db mm/memcontrol.o
This patch (of 4):
The memcg charge API charges pages before they are rmapped - i.e. have an
actual "type" - and so every callsite needs its own set of charge and
uncharge functions to know what type is being operated on. Worse,
uncharge has to happen from a context that is still type-specific, rather
than at the end of the page's lifetime with exclusive access, and so
requires a lot of synchronization.
Rewrite the charge API to provide a generic set of try_charge(),
commit_charge() and cancel_charge() transaction operations, much like
what's currently done for swap-in:
mem_cgroup_try_charge() attempts to reserve a charge, reclaiming
pages from the memcg if necessary.
mem_cgroup_commit_charge() commits the page to the charge once it
has a valid page->mapping and PageAnon() reliably tells the type.
mem_cgroup_cancel_charge() aborts the transaction.
This reduces the charge API and enables subsequent patches to
drastically simplify uncharging.
As pages need to be committed after rmap is established but before they
are added to the LRU, page_add_new_anon_rmap() must stop doing LRU
additions again. Revive lru_cache_add_active_or_unevictable().
[hughd@google.com: fix shmem_unuse]
[hughd@google.com: Add comments on the private use of -EAGAIN]
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Michal Hocko <mhocko@suse.cz>
Cc: Tejun Heo <tj@kernel.org>
Cc: Vladimir Davydov <vdavydov@parallels.com>
Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-08-09 01:19:20 +04:00
|
|
|
extern void lru_cache_add_active_or_unevictable(struct page *page,
|
|
|
|
|
struct vm_area_struct *vma);
|
|
|
|
|
|
2005-04-17 02:20:36 +04:00
|
|
|
/* linux/mm/vmscan.c */
|
2016-07-29 01:47:31 +03:00
|
|
|
extern unsigned long zone_reclaimable_pages(struct zone *zone);
|
2016-07-29 01:45:31 +03:00
|
|
|
extern unsigned long pgdat_reclaimable_pages(struct pglist_data *pgdat);
|
2008-04-28 13:12:12 +04:00
|
|
|
extern unsigned long try_to_free_pages(struct zonelist *zonelist, int order,
|
2009-04-01 02:23:31 +04:00
|
|
|
gfp_t gfp_mask, nodemask_t *mask);
|
2012-05-30 02:06:54 +04:00
|
|
|
extern int __isolate_lru_page(struct page *page, isolate_mode_t mode);
|
2014-10-10 02:28:56 +04:00
|
|
|
extern unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *memcg,
|
|
|
|
|
unsigned long nr_pages,
|
|
|
|
|
gfp_t gfp_mask,
|
|
|
|
|
bool may_swap);
|
2016-07-29 01:46:02 +03:00
|
|
|
extern unsigned long mem_cgroup_shrink_node(struct mem_cgroup *mem,
|
2011-09-15 03:21:58 +04:00
|
|
|
gfp_t gfp_mask, bool noswap,
|
2016-07-29 01:46:05 +03:00
|
|
|
pg_data_t *pgdat,
|
2011-09-15 03:21:58 +04:00
|
|
|
unsigned long *nr_scanned);
|
2006-03-22 11:08:19 +03:00
|
|
|
extern unsigned long shrink_all_memory(unsigned long nr_pages);
|
2005-04-17 02:20:36 +04:00
|
|
|
extern int vm_swappiness;
|
2006-03-22 11:09:12 +03:00
|
|
|
extern int remove_mapping(struct address_space *mapping, struct page *page);
|
2013-02-23 04:35:48 +04:00
|
|
|
extern unsigned long vm_total_pages;
|
2006-03-22 11:09:12 +03:00
|
|
|
|
2006-01-19 04:42:31 +03:00
|
|
|
#ifdef CONFIG_NUMA
|
2016-07-29 01:46:32 +03:00
|
|
|
extern int node_reclaim_mode;
|
2006-07-03 11:24:13 +04:00
|
|
|
extern int sysctl_min_unmapped_ratio;
|
2006-09-26 10:31:52 +04:00
|
|
|
extern int sysctl_min_slab_ratio;
|
2016-07-29 01:46:32 +03:00
|
|
|
extern int node_reclaim(struct pglist_data *, gfp_t, unsigned int);
|
2006-01-19 04:42:31 +03:00
|
|
|
#else
|
2016-07-29 01:46:32 +03:00
|
|
|
#define node_reclaim_mode 0
|
|
|
|
|
static inline int node_reclaim(struct pglist_data *pgdat, gfp_t mask,
|
|
|
|
|
unsigned int order)
|
2006-01-19 04:42:31 +03:00
|
|
|
{
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
#endif
|
|
|
|
|
|
2012-10-09 03:33:18 +04:00
|
|
|
extern int page_evictable(struct page *page);
|
SHM_UNLOCK: fix Unevictable pages stranded after swap
Commit cc39c6a9bbde ("mm: account skipped entries to avoid looping in
find_get_pages") correctly fixed an infinite loop; but left a problem
that find_get_pages() on shmem would return 0 (appearing to callers to
mean end of tree) when it meets a run of nr_pages swap entries.
The only uses of find_get_pages() on shmem are via pagevec_lookup(),
called from invalidate_mapping_pages(), and from shmctl SHM_UNLOCK's
scan_mapping_unevictable_pages(). The first is already commented, and
not worth worrying about; but the second can leave pages on the
Unevictable list after an unusual sequence of swapping and locking.
Fix that by using shmem_find_get_pages_and_swap() (then ignoring the
swap) instead of pagevec_lookup().
But I don't want to contaminate vmscan.c with shmem internals, nor
shmem.c with LRU locking. So move scan_mapping_unevictable_pages() into
shmem.c, renaming it shmem_unlock_mapping(); and rename
check_move_unevictable_page() to check_move_unevictable_pages(), looping
down an array of pages, oftentimes under the same lock.
Leave out the "rotate unevictable list" block: that's a leftover from
when this was used for /proc/sys/vm/scan_unevictable_pages, whose flawed
handling involved looking at pages at tail of LRU.
Was there significance to the sequence first ClearPageUnevictable, then
test page_evictable, then SetPageUnevictable here? I think not, we're
under LRU lock, and have no barriers between those.
Signed-off-by: Hugh Dickins <hughd@google.com>
Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Shaohua Li <shaohua.li@intel.com>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michel Lespinasse <walken@google.com>
Cc: <stable@vger.kernel.org> [back to 3.1 but will need respins]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-01-21 02:34:21 +04:00
|
|
|
extern void check_move_unevictable_pages(struct page **, int nr_pages);
|
2008-10-19 07:26:53 +04:00
|
|
|
|
2006-06-27 13:53:33 +04:00
|
|
|
extern int kswapd_run(int nid);
|
2009-12-15 04:58:33 +03:00
|
|
|
extern void kswapd_stop(int nid);
|
2015-09-09 01:01:02 +03:00
|
|
|
|
2005-04-17 02:20:36 +04:00
|
|
|
#ifdef CONFIG_SWAP
|
2016-11-01 16:40:16 +03:00
|
|
|
|
|
|
|
|
#include <linux/blk_types.h> /* for bio_end_io_t */
|
|
|
|
|
|
2005-04-17 02:20:36 +04:00
|
|
|
/* linux/mm/page_io.c */
|
swap: add block io poll in swapin path
For fast flash disk, async IO could introduce overhead because of
context switch. block-mq now supports IO poll, which improves
performance and latency a lot. swapin is a good place to use this
technique, because the task is waiting for the swapin page to continue
execution.
In my virtual machine, directly read 4k data from a NVMe with iopoll is
about 60% better than that without poll. With iopoll support in swapin
patch, my microbenchmark (a task does random memory write) is about
10%~25% faster. CPU utilization increases a lot though, 2x and even 3x
CPU utilization. This will depend on disk speed.
While iopoll in swapin isn't intended for all usage cases, it's a win
for latency sensistive workloads with high speed swap disk. block layer
has knob to control poll in runtime. If poll isn't enabled in block
layer, there should be no noticeable change in swapin.
I got a chance to run the same test in a NVMe with DRAM as the media.
In simple fio IO test, blkpoll boosts 50% performance in single thread
test and ~20% in 8 threads test. So this is the base line. In above
swap test, blkpoll boosts ~27% performance in single thread test.
blkpoll uses 2x CPU time though.
If we enable hybid polling, the performance gain has very slight drop
but CPU time is only 50% worse than that without blkpoll. Also we can
adjust parameter of hybid poll, with it, the CPU time penality is
reduced further. In 8 threads test, blkpoll doesn't help though. The
performance is similar to that without blkpoll, but cpu utilization is
similar too. There is lock contention in swap path. The cpu time
spending on blkpoll isn't high. So overall, blkpoll swapin isn't worse
than that without it.
The swapin readahead might read several pages in in the same time and
form a big IO request. Since the IO will take longer time, it doesn't
make sense to do poll, so the patch only does iopoll for single page
swapin.
[akpm@linux-foundation.org: coding-style fixes]
Link: http://lkml.kernel.org/r/070c3c3e40b711e7b1390002c991e86a-b5408f0@7511894063d3764ff01ea8111f5a004d7dd700ed078797c204a24e620ddb965c
Signed-off-by: Shaohua Li <shli@fb.com>
Cc: Tim Chen <tim.c.chen@intel.com>
Cc: Huang Ying <ying.huang@intel.com>
Cc: Jens Axboe <axboe@fb.com>
Cc: Hugh Dickins <hughd@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-07-11 01:47:11 +03:00
|
|
|
extern int swap_readpage(struct page *page, bool do_poll);
|
2005-04-17 02:20:36 +04:00
|
|
|
extern int swap_writepage(struct page *page, struct writeback_control *wbc);
|
2015-07-20 16:29:37 +03:00
|
|
|
extern void end_swap_bio_write(struct bio *bio);
|
2013-04-30 02:08:35 +04:00
|
|
|
extern int __swap_writepage(struct page *page, struct writeback_control *wbc,
|
2015-07-20 16:29:37 +03:00
|
|
|
bio_end_io_t end_write_func);
|
2012-08-01 03:44:55 +04:00
|
|
|
extern int swap_set_page_dirty(struct page *page);
|
2005-04-17 02:20:36 +04:00
|
|
|
|
2012-08-01 03:44:57 +04:00
|
|
|
int add_swap_extent(struct swap_info_struct *sis, unsigned long start_page,
|
|
|
|
|
unsigned long nr_pages, sector_t start_block);
|
|
|
|
|
int generic_swapfile_activate(struct swap_info_struct *, struct file *,
|
|
|
|
|
sector_t *);
|
|
|
|
|
|
2005-04-17 02:20:36 +04:00
|
|
|
/* linux/mm/swap_state.c */
|
mm/swap: split swap cache into 64MB trunks
The patch is to improve the scalability of the swap out/in via using
fine grained locks for the swap cache. In current kernel, one address
space will be used for each swap device. And in the common
configuration, the number of the swap device is very small (one is
typical). This causes the heavy lock contention on the radix tree of
the address space if multiple tasks swap out/in concurrently.
But in fact, there is no dependency between pages in the swap cache. So
that, we can split the one shared address space for each swap device
into several address spaces to reduce the lock contention. In the
patch, the shared address space is split into 64MB trunks. 64MB is
chosen to balance the memory space usage and effect of lock contention
reduction.
The size of struct address_space on x86_64 architecture is 408B, so with
the patch, 6528B more memory will be used for every 1GB swap space on
x86_64 architecture.
One address space is still shared for the swap entries in the same 64M
trunks. To avoid lock contention for the first round of swap space
allocation, the order of the swap clusters in the initial free clusters
list is changed. The swap space distance between the consecutive swap
clusters in the free cluster list is at least 64M. After the first
round of allocation, the swap clusters are expected to be freed
randomly, so the lock contention should be reduced effectively.
Link: http://lkml.kernel.org/r/735bab895e64c930581ffb0a05b661e01da82bc5.1484082593.git.tim.c.chen@linux.intel.com
Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
Signed-off-by: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Aaron Lu <aaron.lu@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Christian Borntraeger <borntraeger@de.ibm.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: Huang Ying <ying.huang@intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Jonathan Corbet <corbet@lwn.net> escreveu:
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Shaohua Li <shli@kernel.org>
Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-02-23 02:45:26 +03:00
|
|
|
/* One swap address space for each 64M swap space */
|
|
|
|
|
#define SWAP_ADDRESS_SPACE_SHIFT 14
|
|
|
|
|
#define SWAP_ADDRESS_SPACE_PAGES (1 << SWAP_ADDRESS_SPACE_SHIFT)
|
|
|
|
|
extern struct address_space *swapper_spaces[];
|
mm, swap: VMA based swap readahead
The swap readahead is an important mechanism to reduce the swap in
latency. Although pure sequential memory access pattern isn't very
popular for anonymous memory, the space locality is still considered
valid.
In the original swap readahead implementation, the consecutive blocks in
swap device are readahead based on the global space locality estimation.
But the consecutive blocks in swap device just reflect the order of page
reclaiming, don't necessarily reflect the access pattern in virtual
memory. And the different tasks in the system may have different access
patterns, which makes the global space locality estimation incorrect.
In this patch, when page fault occurs, the virtual pages near the fault
address will be readahead instead of the swap slots near the fault swap
slot in swap device. This avoid to readahead the unrelated swap slots.
At the same time, the swap readahead is changed to work on per-VMA from
globally. So that the different access patterns of the different VMAs
could be distinguished, and the different readahead policy could be
applied accordingly. The original core readahead detection and scaling
algorithm is reused, because it is an effect algorithm to detect the
space locality.
The test and result is as follow,
Common test condition
=====================
Test Machine: Xeon E5 v3 (2 sockets, 72 threads, 32G RAM) Swap device:
NVMe disk
Micro-benchmark with combined access pattern
============================================
vm-scalability, sequential swap test case, 4 processes to eat 50G
virtual memory space, repeat the sequential memory writing until 300
seconds. The first round writing will trigger swap out, the following
rounds will trigger sequential swap in and out.
At the same time, run vm-scalability random swap test case in
background, 8 processes to eat 30G virtual memory space, repeat the
random memory write until 300 seconds. This will trigger random swap-in
in the background.
This is a combined workload with sequential and random memory accessing
at the same time. The result (for sequential workload) is as follow,
Base Optimized
---- ---------
throughput 345413 KB/s 414029 KB/s (+19.9%)
latency.average 97.14 us 61.06 us (-37.1%)
latency.50th 2 us 1 us
latency.60th 2 us 1 us
latency.70th 98 us 2 us
latency.80th 160 us 2 us
latency.90th 260 us 217 us
latency.95th 346 us 369 us
latency.99th 1.34 ms 1.09 ms
ra_hit% 52.69% 99.98%
The original swap readahead algorithm is confused by the background
random access workload, so readahead hit rate is lower. The VMA-base
readahead algorithm works much better.
Linpack
=======
The test memory size is bigger than RAM to trigger swapping.
Base Optimized
---- ---------
elapsed_time 393.49 s 329.88 s (-16.2%)
ra_hit% 86.21% 98.82%
The score of base and optimized kernel hasn't visible changes. But the
elapsed time reduced and readahead hit rate improved, so the optimized
kernel runs better for startup and tear down stages. And the absolute
value of readahead hit rate is high, shows that the space locality is
still valid in some practical workloads.
Link: http://lkml.kernel.org/r/20170807054038.1843-4-ying.huang@intel.com
Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Shaohua Li <shli@kernel.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Fengguang Wu <fengguang.wu@intel.com>
Cc: Tim Chen <tim.c.chen@intel.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-09-07 02:24:36 +03:00
|
|
|
extern bool swap_vma_readahead;
|
mm/swap: split swap cache into 64MB trunks
The patch is to improve the scalability of the swap out/in via using
fine grained locks for the swap cache. In current kernel, one address
space will be used for each swap device. And in the common
configuration, the number of the swap device is very small (one is
typical). This causes the heavy lock contention on the radix tree of
the address space if multiple tasks swap out/in concurrently.
But in fact, there is no dependency between pages in the swap cache. So
that, we can split the one shared address space for each swap device
into several address spaces to reduce the lock contention. In the
patch, the shared address space is split into 64MB trunks. 64MB is
chosen to balance the memory space usage and effect of lock contention
reduction.
The size of struct address_space on x86_64 architecture is 408B, so with
the patch, 6528B more memory will be used for every 1GB swap space on
x86_64 architecture.
One address space is still shared for the swap entries in the same 64M
trunks. To avoid lock contention for the first round of swap space
allocation, the order of the swap clusters in the initial free clusters
list is changed. The swap space distance between the consecutive swap
clusters in the free cluster list is at least 64M. After the first
round of allocation, the swap clusters are expected to be freed
randomly, so the lock contention should be reduced effectively.
Link: http://lkml.kernel.org/r/735bab895e64c930581ffb0a05b661e01da82bc5.1484082593.git.tim.c.chen@linux.intel.com
Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
Signed-off-by: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Aaron Lu <aaron.lu@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Christian Borntraeger <borntraeger@de.ibm.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: Huang Ying <ying.huang@intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Jonathan Corbet <corbet@lwn.net> escreveu:
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Shaohua Li <shli@kernel.org>
Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-02-23 02:45:26 +03:00
|
|
|
#define swap_address_space(entry) \
|
|
|
|
|
(&swapper_spaces[swp_type(entry)][swp_offset(entry) \
|
|
|
|
|
>> SWAP_ADDRESS_SPACE_SHIFT])
|
2013-02-23 04:34:37 +04:00
|
|
|
extern unsigned long total_swapcache_pages(void);
|
2005-04-17 02:20:36 +04:00
|
|
|
extern void show_swap_cache_info(void);
|
2017-07-07 01:37:24 +03:00
|
|
|
extern int add_to_swap(struct page *page);
|
2008-02-05 09:28:50 +03:00
|
|
|
extern int add_to_swap_cache(struct page *, swp_entry_t, gfp_t);
|
2013-04-30 02:08:34 +04:00
|
|
|
extern int __add_to_swap_cache(struct page *page, swp_entry_t entry);
|
2005-04-17 02:20:36 +04:00
|
|
|
extern void __delete_from_swap_cache(struct page *);
|
|
|
|
|
extern void delete_from_swap_cache(struct page *);
|
|
|
|
|
extern void free_page_and_swap_cache(struct page *);
|
|
|
|
|
extern void free_pages_and_swap_cache(struct page **, int);
|
mm, swap: VMA based swap readahead
The swap readahead is an important mechanism to reduce the swap in
latency. Although pure sequential memory access pattern isn't very
popular for anonymous memory, the space locality is still considered
valid.
In the original swap readahead implementation, the consecutive blocks in
swap device are readahead based on the global space locality estimation.
But the consecutive blocks in swap device just reflect the order of page
reclaiming, don't necessarily reflect the access pattern in virtual
memory. And the different tasks in the system may have different access
patterns, which makes the global space locality estimation incorrect.
In this patch, when page fault occurs, the virtual pages near the fault
address will be readahead instead of the swap slots near the fault swap
slot in swap device. This avoid to readahead the unrelated swap slots.
At the same time, the swap readahead is changed to work on per-VMA from
globally. So that the different access patterns of the different VMAs
could be distinguished, and the different readahead policy could be
applied accordingly. The original core readahead detection and scaling
algorithm is reused, because it is an effect algorithm to detect the
space locality.
The test and result is as follow,
Common test condition
=====================
Test Machine: Xeon E5 v3 (2 sockets, 72 threads, 32G RAM) Swap device:
NVMe disk
Micro-benchmark with combined access pattern
============================================
vm-scalability, sequential swap test case, 4 processes to eat 50G
virtual memory space, repeat the sequential memory writing until 300
seconds. The first round writing will trigger swap out, the following
rounds will trigger sequential swap in and out.
At the same time, run vm-scalability random swap test case in
background, 8 processes to eat 30G virtual memory space, repeat the
random memory write until 300 seconds. This will trigger random swap-in
in the background.
This is a combined workload with sequential and random memory accessing
at the same time. The result (for sequential workload) is as follow,
Base Optimized
---- ---------
throughput 345413 KB/s 414029 KB/s (+19.9%)
latency.average 97.14 us 61.06 us (-37.1%)
latency.50th 2 us 1 us
latency.60th 2 us 1 us
latency.70th 98 us 2 us
latency.80th 160 us 2 us
latency.90th 260 us 217 us
latency.95th 346 us 369 us
latency.99th 1.34 ms 1.09 ms
ra_hit% 52.69% 99.98%
The original swap readahead algorithm is confused by the background
random access workload, so readahead hit rate is lower. The VMA-base
readahead algorithm works much better.
Linpack
=======
The test memory size is bigger than RAM to trigger swapping.
Base Optimized
---- ---------
elapsed_time 393.49 s 329.88 s (-16.2%)
ra_hit% 86.21% 98.82%
The score of base and optimized kernel hasn't visible changes. But the
elapsed time reduced and readahead hit rate improved, so the optimized
kernel runs better for startup and tear down stages. And the absolute
value of readahead hit rate is high, shows that the space locality is
still valid in some practical workloads.
Link: http://lkml.kernel.org/r/20170807054038.1843-4-ying.huang@intel.com
Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Shaohua Li <shli@kernel.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Fengguang Wu <fengguang.wu@intel.com>
Cc: Tim Chen <tim.c.chen@intel.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-09-07 02:24:36 +03:00
|
|
|
extern struct page *lookup_swap_cache(swp_entry_t entry,
|
|
|
|
|
struct vm_area_struct *vma,
|
|
|
|
|
unsigned long addr);
|
swapin needs gfp_mask for loop on tmpfs
Building in a filesystem on a loop device on a tmpfs file can hang when
swapping, the loop thread caught in that infamous throttle_vm_writeout.
In theory this is a long standing problem, which I've either never seen in
practice, or long ago suppressed the recollection, after discounting my load
and my tmpfs size as unrealistically high. But now, with the new aops, it has
become easy to hang on one machine.
Loop used to grab_cache_page before the old prepare_write to tmpfs, which
seems to have been enough to free up some memory for any swapin needed; but
the new write_begin lets tmpfs find or allocate the page (much nicer, since
grab_cache_page missed tmpfs pages in swapcache).
When allocating a fresh page, tmpfs respects loop's mapping_gfp_mask, which
has __GFP_IO|__GFP_FS stripped off, and throttle_vm_writeout is designed to
break out when __GFP_IO or GFP_FS is unset; but when tmfps swaps in,
read_swap_cache_async allocates with GFP_HIGHUSER_MOVABLE regardless of the
mapping_gfp_mask - hence the hang.
So, pass gfp_mask down the line from shmem_getpage to shmem_swapin to
swapin_readahead to read_swap_cache_async to add_to_swap_cache.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Acked-by: Rik van Riel <riel@redhat.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 09:28:42 +03:00
|
|
|
extern struct page *read_swap_cache_async(swp_entry_t, gfp_t,
|
swap: add block io poll in swapin path
For fast flash disk, async IO could introduce overhead because of
context switch. block-mq now supports IO poll, which improves
performance and latency a lot. swapin is a good place to use this
technique, because the task is waiting for the swapin page to continue
execution.
In my virtual machine, directly read 4k data from a NVMe with iopoll is
about 60% better than that without poll. With iopoll support in swapin
patch, my microbenchmark (a task does random memory write) is about
10%~25% faster. CPU utilization increases a lot though, 2x and even 3x
CPU utilization. This will depend on disk speed.
While iopoll in swapin isn't intended for all usage cases, it's a win
for latency sensistive workloads with high speed swap disk. block layer
has knob to control poll in runtime. If poll isn't enabled in block
layer, there should be no noticeable change in swapin.
I got a chance to run the same test in a NVMe with DRAM as the media.
In simple fio IO test, blkpoll boosts 50% performance in single thread
test and ~20% in 8 threads test. So this is the base line. In above
swap test, blkpoll boosts ~27% performance in single thread test.
blkpoll uses 2x CPU time though.
If we enable hybid polling, the performance gain has very slight drop
but CPU time is only 50% worse than that without blkpoll. Also we can
adjust parameter of hybid poll, with it, the CPU time penality is
reduced further. In 8 threads test, blkpoll doesn't help though. The
performance is similar to that without blkpoll, but cpu utilization is
similar too. There is lock contention in swap path. The cpu time
spending on blkpoll isn't high. So overall, blkpoll swapin isn't worse
than that without it.
The swapin readahead might read several pages in in the same time and
form a big IO request. Since the IO will take longer time, it doesn't
make sense to do poll, so the patch only does iopoll for single page
swapin.
[akpm@linux-foundation.org: coding-style fixes]
Link: http://lkml.kernel.org/r/070c3c3e40b711e7b1390002c991e86a-b5408f0@7511894063d3764ff01ea8111f5a004d7dd700ed078797c204a24e620ddb965c
Signed-off-by: Shaohua Li <shli@fb.com>
Cc: Tim Chen <tim.c.chen@intel.com>
Cc: Huang Ying <ying.huang@intel.com>
Cc: Jens Axboe <axboe@fb.com>
Cc: Hugh Dickins <hughd@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-07-11 01:47:11 +03:00
|
|
|
struct vm_area_struct *vma, unsigned long addr,
|
|
|
|
|
bool do_poll);
|
2015-09-09 01:05:00 +03:00
|
|
|
extern struct page *__read_swap_cache_async(swp_entry_t, gfp_t,
|
|
|
|
|
struct vm_area_struct *vma, unsigned long addr,
|
|
|
|
|
bool *new_page_allocated);
|
swapin needs gfp_mask for loop on tmpfs
Building in a filesystem on a loop device on a tmpfs file can hang when
swapping, the loop thread caught in that infamous throttle_vm_writeout.
In theory this is a long standing problem, which I've either never seen in
practice, or long ago suppressed the recollection, after discounting my load
and my tmpfs size as unrealistically high. But now, with the new aops, it has
become easy to hang on one machine.
Loop used to grab_cache_page before the old prepare_write to tmpfs, which
seems to have been enough to free up some memory for any swapin needed; but
the new write_begin lets tmpfs find or allocate the page (much nicer, since
grab_cache_page missed tmpfs pages in swapcache).
When allocating a fresh page, tmpfs respects loop's mapping_gfp_mask, which
has __GFP_IO|__GFP_FS stripped off, and throttle_vm_writeout is designed to
break out when __GFP_IO or GFP_FS is unset; but when tmfps swaps in,
read_swap_cache_async allocates with GFP_HIGHUSER_MOVABLE regardless of the
mapping_gfp_mask - hence the hang.
So, pass gfp_mask down the line from shmem_getpage to shmem_swapin to
swapin_readahead to read_swap_cache_async to add_to_swap_cache.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Acked-by: Rik van Riel <riel@redhat.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 09:28:42 +03:00
|
|
|
extern struct page *swapin_readahead(swp_entry_t, gfp_t,
|
2008-02-05 09:28:41 +03:00
|
|
|
struct vm_area_struct *vma, unsigned long addr);
|
|
|
|
|
|
mm, swap: VMA based swap readahead
The swap readahead is an important mechanism to reduce the swap in
latency. Although pure sequential memory access pattern isn't very
popular for anonymous memory, the space locality is still considered
valid.
In the original swap readahead implementation, the consecutive blocks in
swap device are readahead based on the global space locality estimation.
But the consecutive blocks in swap device just reflect the order of page
reclaiming, don't necessarily reflect the access pattern in virtual
memory. And the different tasks in the system may have different access
patterns, which makes the global space locality estimation incorrect.
In this patch, when page fault occurs, the virtual pages near the fault
address will be readahead instead of the swap slots near the fault swap
slot in swap device. This avoid to readahead the unrelated swap slots.
At the same time, the swap readahead is changed to work on per-VMA from
globally. So that the different access patterns of the different VMAs
could be distinguished, and the different readahead policy could be
applied accordingly. The original core readahead detection and scaling
algorithm is reused, because it is an effect algorithm to detect the
space locality.
The test and result is as follow,
Common test condition
=====================
Test Machine: Xeon E5 v3 (2 sockets, 72 threads, 32G RAM) Swap device:
NVMe disk
Micro-benchmark with combined access pattern
============================================
vm-scalability, sequential swap test case, 4 processes to eat 50G
virtual memory space, repeat the sequential memory writing until 300
seconds. The first round writing will trigger swap out, the following
rounds will trigger sequential swap in and out.
At the same time, run vm-scalability random swap test case in
background, 8 processes to eat 30G virtual memory space, repeat the
random memory write until 300 seconds. This will trigger random swap-in
in the background.
This is a combined workload with sequential and random memory accessing
at the same time. The result (for sequential workload) is as follow,
Base Optimized
---- ---------
throughput 345413 KB/s 414029 KB/s (+19.9%)
latency.average 97.14 us 61.06 us (-37.1%)
latency.50th 2 us 1 us
latency.60th 2 us 1 us
latency.70th 98 us 2 us
latency.80th 160 us 2 us
latency.90th 260 us 217 us
latency.95th 346 us 369 us
latency.99th 1.34 ms 1.09 ms
ra_hit% 52.69% 99.98%
The original swap readahead algorithm is confused by the background
random access workload, so readahead hit rate is lower. The VMA-base
readahead algorithm works much better.
Linpack
=======
The test memory size is bigger than RAM to trigger swapping.
Base Optimized
---- ---------
elapsed_time 393.49 s 329.88 s (-16.2%)
ra_hit% 86.21% 98.82%
The score of base and optimized kernel hasn't visible changes. But the
elapsed time reduced and readahead hit rate improved, so the optimized
kernel runs better for startup and tear down stages. And the absolute
value of readahead hit rate is high, shows that the space locality is
still valid in some practical workloads.
Link: http://lkml.kernel.org/r/20170807054038.1843-4-ying.huang@intel.com
Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Shaohua Li <shli@kernel.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Fengguang Wu <fengguang.wu@intel.com>
Cc: Tim Chen <tim.c.chen@intel.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-09-07 02:24:36 +03:00
|
|
|
extern struct page *swap_readahead_detect(struct vm_fault *vmf,
|
|
|
|
|
struct vma_swap_readahead *swap_ra);
|
|
|
|
|
extern struct page *do_swap_page_readahead(swp_entry_t fentry, gfp_t gfp_mask,
|
|
|
|
|
struct vm_fault *vmf,
|
|
|
|
|
struct vma_swap_readahead *swap_ra);
|
|
|
|
|
|
2005-04-17 02:20:36 +04:00
|
|
|
/* linux/mm/swapfile.c */
|
swap: add per-partition lock for swapfile
swap_lock is heavily contended when I test swap to 3 fast SSD (even
slightly slower than swap to 2 such SSD). The main contention comes
from swap_info_get(). This patch tries to fix the gap with adding a new
per-partition lock.
Global data like nr_swapfiles, total_swap_pages, least_priority and
swap_list are still protected by swap_lock.
nr_swap_pages is an atomic now, it can be changed without swap_lock. In
theory, it's possible get_swap_page() finds no swap pages but actually
there are free swap pages. But sounds not a big problem.
Accessing partition specific data (like scan_swap_map and so on) is only
protected by swap_info_struct.lock.
Changing swap_info_struct.flags need hold swap_lock and
swap_info_struct.lock, because scan_scan_map() will check it. read the
flags is ok with either the locks hold.
If both swap_lock and swap_info_struct.lock must be hold, we always hold
the former first to avoid deadlock.
swap_entry_free() can change swap_list. To delete that code, we add a
new highest_priority_index. Whenever get_swap_page() is called, we
check it. If it's valid, we use it.
It's a pity get_swap_page() still holds swap_lock(). But in practice,
swap_lock() isn't heavily contended in my test with this patch (or I can
say there are other much more heavier bottlenecks like TLB flush). And
BTW, looks get_swap_page() doesn't really need the lock. We never free
swap_info[] and we check SWAP_WRITEOK flag. The only risk without the
lock is we could swapout to some low priority swap, but we can quickly
recover after several rounds of swap, so sounds not a big deal to me.
But I'd prefer to fix this if it's a real problem.
"swap: make each swap partition have one address_space" improved the
swapout speed from 1.7G/s to 2G/s. This patch further improves the
speed to 2.3G/s, so around 15% improvement. It's a multi-process test,
so TLB flush isn't the biggest bottleneck before the patches.
[arnd@arndb.de: fix it for nommu]
[hughd@google.com: add missing unlock]
[minchan@kernel.org: get rid of lockdep whinge on sys_swapon]
Signed-off-by: Shaohua Li <shli@fusionio.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Seth Jennings <sjenning@linux.vnet.ibm.com>
Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: Xiao Guangrong <xiaoguangrong@linux.vnet.ibm.com>
Cc: Dan Magenheimer <dan.magenheimer@oracle.com>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Hugh Dickins <hughd@google.com>
Signed-off-by: Minchan Kim <minchan@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-23 04:34:38 +04:00
|
|
|
extern atomic_long_t nr_swap_pages;
|
2005-04-17 02:20:36 +04:00
|
|
|
extern long total_swap_pages;
|
2017-09-07 02:24:43 +03:00
|
|
|
extern atomic_t nr_rotate_swap;
|
2017-02-23 02:45:39 +03:00
|
|
|
extern bool has_usable_swap(void);
|
swap: add per-partition lock for swapfile
swap_lock is heavily contended when I test swap to 3 fast SSD (even
slightly slower than swap to 2 such SSD). The main contention comes
from swap_info_get(). This patch tries to fix the gap with adding a new
per-partition lock.
Global data like nr_swapfiles, total_swap_pages, least_priority and
swap_list are still protected by swap_lock.
nr_swap_pages is an atomic now, it can be changed without swap_lock. In
theory, it's possible get_swap_page() finds no swap pages but actually
there are free swap pages. But sounds not a big problem.
Accessing partition specific data (like scan_swap_map and so on) is only
protected by swap_info_struct.lock.
Changing swap_info_struct.flags need hold swap_lock and
swap_info_struct.lock, because scan_scan_map() will check it. read the
flags is ok with either the locks hold.
If both swap_lock and swap_info_struct.lock must be hold, we always hold
the former first to avoid deadlock.
swap_entry_free() can change swap_list. To delete that code, we add a
new highest_priority_index. Whenever get_swap_page() is called, we
check it. If it's valid, we use it.
It's a pity get_swap_page() still holds swap_lock(). But in practice,
swap_lock() isn't heavily contended in my test with this patch (or I can
say there are other much more heavier bottlenecks like TLB flush). And
BTW, looks get_swap_page() doesn't really need the lock. We never free
swap_info[] and we check SWAP_WRITEOK flag. The only risk without the
lock is we could swapout to some low priority swap, but we can quickly
recover after several rounds of swap, so sounds not a big deal to me.
But I'd prefer to fix this if it's a real problem.
"swap: make each swap partition have one address_space" improved the
swapout speed from 1.7G/s to 2G/s. This patch further improves the
speed to 2.3G/s, so around 15% improvement. It's a multi-process test,
so TLB flush isn't the biggest bottleneck before the patches.
[arnd@arndb.de: fix it for nommu]
[hughd@google.com: add missing unlock]
[minchan@kernel.org: get rid of lockdep whinge on sys_swapon]
Signed-off-by: Shaohua Li <shli@fusionio.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Seth Jennings <sjenning@linux.vnet.ibm.com>
Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: Xiao Guangrong <xiaoguangrong@linux.vnet.ibm.com>
Cc: Dan Magenheimer <dan.magenheimer@oracle.com>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Hugh Dickins <hughd@google.com>
Signed-off-by: Minchan Kim <minchan@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-23 04:34:38 +04:00
|
|
|
|
2017-09-07 02:24:43 +03:00
|
|
|
static inline bool swap_use_vma_readahead(void)
|
|
|
|
|
{
|
|
|
|
|
return READ_ONCE(swap_vma_readahead) && !atomic_read(&nr_rotate_swap);
|
|
|
|
|
}
|
|
|
|
|
|
swap: add per-partition lock for swapfile
swap_lock is heavily contended when I test swap to 3 fast SSD (even
slightly slower than swap to 2 such SSD). The main contention comes
from swap_info_get(). This patch tries to fix the gap with adding a new
per-partition lock.
Global data like nr_swapfiles, total_swap_pages, least_priority and
swap_list are still protected by swap_lock.
nr_swap_pages is an atomic now, it can be changed without swap_lock. In
theory, it's possible get_swap_page() finds no swap pages but actually
there are free swap pages. But sounds not a big problem.
Accessing partition specific data (like scan_swap_map and so on) is only
protected by swap_info_struct.lock.
Changing swap_info_struct.flags need hold swap_lock and
swap_info_struct.lock, because scan_scan_map() will check it. read the
flags is ok with either the locks hold.
If both swap_lock and swap_info_struct.lock must be hold, we always hold
the former first to avoid deadlock.
swap_entry_free() can change swap_list. To delete that code, we add a
new highest_priority_index. Whenever get_swap_page() is called, we
check it. If it's valid, we use it.
It's a pity get_swap_page() still holds swap_lock(). But in practice,
swap_lock() isn't heavily contended in my test with this patch (or I can
say there are other much more heavier bottlenecks like TLB flush). And
BTW, looks get_swap_page() doesn't really need the lock. We never free
swap_info[] and we check SWAP_WRITEOK flag. The only risk without the
lock is we could swapout to some low priority swap, but we can quickly
recover after several rounds of swap, so sounds not a big deal to me.
But I'd prefer to fix this if it's a real problem.
"swap: make each swap partition have one address_space" improved the
swapout speed from 1.7G/s to 2G/s. This patch further improves the
speed to 2.3G/s, so around 15% improvement. It's a multi-process test,
so TLB flush isn't the biggest bottleneck before the patches.
[arnd@arndb.de: fix it for nommu]
[hughd@google.com: add missing unlock]
[minchan@kernel.org: get rid of lockdep whinge on sys_swapon]
Signed-off-by: Shaohua Li <shli@fusionio.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Seth Jennings <sjenning@linux.vnet.ibm.com>
Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: Xiao Guangrong <xiaoguangrong@linux.vnet.ibm.com>
Cc: Dan Magenheimer <dan.magenheimer@oracle.com>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Hugh Dickins <hughd@google.com>
Signed-off-by: Minchan Kim <minchan@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-23 04:34:38 +04:00
|
|
|
/* Swap 50% full? Release swapcache more aggressively.. */
|
|
|
|
|
static inline bool vm_swap_full(void)
|
|
|
|
|
{
|
|
|
|
|
return atomic_long_read(&nr_swap_pages) * 2 < total_swap_pages;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static inline long get_nr_swap_pages(void)
|
|
|
|
|
{
|
|
|
|
|
return atomic_long_read(&nr_swap_pages);
|
|
|
|
|
}
|
|
|
|
|
|
2005-04-17 02:20:36 +04:00
|
|
|
extern void si_swapinfo(struct sysinfo *);
|
mm, THP, swap: delay splitting THP during swap out
Patch series "THP swap: Delay splitting THP during swapping out", v11.
This patchset is to optimize the performance of Transparent Huge Page
(THP) swap.
Recently, the performance of the storage devices improved so fast that
we cannot saturate the disk bandwidth with single logical CPU when do
page swap out even on a high-end server machine. Because the
performance of the storage device improved faster than that of single
logical CPU. And it seems that the trend will not change in the near
future. On the other hand, the THP becomes more and more popular
because of increased memory size. So it becomes necessary to optimize
THP swap performance.
The advantages of the THP swap support include:
- Batch the swap operations for the THP to reduce lock
acquiring/releasing, including allocating/freeing the swap space,
adding/deleting to/from the swap cache, and writing/reading the swap
space, etc. This will help improve the performance of the THP swap.
- The THP swap space read/write will be 2M sequential IO. It is
particularly helpful for the swap read, which are usually 4k random
IO. This will improve the performance of the THP swap too.
- It will help the memory fragmentation, especially when the THP is
heavily used by the applications. The 2M continuous pages will be
free up after THP swapping out.
- It will improve the THP utilization on the system with the swap
turned on. Because the speed for khugepaged to collapse the normal
pages into the THP is quite slow. After the THP is split during the
swapping out, it will take quite long time for the normal pages to
collapse back into the THP after being swapped in. The high THP
utilization helps the efficiency of the page based memory management
too.
There are some concerns regarding THP swap in, mainly because possible
enlarged read/write IO size (for swap in/out) may put more overhead on
the storage device. To deal with that, the THP swap in should be turned
on only when necessary. For example, it can be selected via
"always/never/madvise" logic, to be turned on globally, turned off
globally, or turned on only for VMA with MADV_HUGEPAGE, etc.
This patchset is the first step for the THP swap support. The plan is
to delay splitting THP step by step, finally avoid splitting THP during
the THP swapping out and swap out/in the THP as a whole.
As the first step, in this patchset, the splitting huge page is delayed
from almost the first step of swapping out to after allocating the swap
space for the THP and adding the THP into the swap cache. This will
reduce lock acquiring/releasing for the locks used for the swap cache
management.
With the patchset, the swap out throughput improves 15.5% (from about
3.73GB/s to about 4.31GB/s) in the vm-scalability swap-w-seq test case
with 8 processes. The test is done on a Xeon E5 v3 system. The swap
device used is a RAM simulated PMEM (persistent memory) device. To test
the sequential swapping out, the test case creates 8 processes, which
sequentially allocate and write to the anonymous pages until the RAM and
part of the swap device is used up.
This patch (of 5):
In this patch, splitting huge page is delayed from almost the first step
of swapping out to after allocating the swap space for the THP
(Transparent Huge Page) and adding the THP into the swap cache. This
will batch the corresponding operation, thus improve THP swap out
throughput.
This is the first step for the THP swap optimization. The plan is to
delay splitting the THP step by step and avoid splitting the THP
finally.
In this patch, one swap cluster is used to hold the contents of each THP
swapped out. So, the size of the swap cluster is changed to that of the
THP (Transparent Huge Page) on x86_64 architecture (512). For other
architectures which want such THP swap optimization,
ARCH_USES_THP_SWAP_CLUSTER needs to be selected in the Kconfig file for
the architecture. In effect, this will enlarge swap cluster size by 2
times on x86_64. Which may make it harder to find a free cluster when
the swap space becomes fragmented. So that, this may reduce the
continuous swap space allocation and sequential write in theory. The
performance test in 0day shows no regressions caused by this.
In the future of THP swap optimization, some information of the swapped
out THP (such as compound map count) will be recorded in the
swap_cluster_info data structure.
The mem cgroup swap accounting functions are enhanced to support charge
or uncharge a swap cluster backing a THP as a whole.
The swap cluster allocate/free functions are added to allocate/free a
swap cluster for a THP. A fair simple algorithm is used for swap
cluster allocation, that is, only the first swap device in priority list
will be tried to allocate the swap cluster. The function will fail if
the trying is not successful, and the caller will fallback to allocate a
single swap slot instead. This works good enough for normal cases. If
the difference of the number of the free swap clusters among multiple
swap devices is significant, it is possible that some THPs are split
earlier than necessary. For example, this could be caused by big size
difference among multiple swap devices.
The swap cache functions is enhanced to support add/delete THP to/from
the swap cache as a set of (HPAGE_PMD_NR) sub-pages. This may be
enhanced in the future with multi-order radix tree. But because we will
split the THP soon during swapping out, that optimization doesn't make
much sense for this first step.
The THP splitting functions are enhanced to support to split THP in swap
cache during swapping out. The page lock will be held during allocating
the swap cluster, adding the THP into the swap cache and splitting the
THP. So in the code path other than swapping out, if the THP need to be
split, the PageSwapCache(THP) will be always false.
The swap cluster is only available for SSD, so the THP swap optimization
in this patchset has no effect for HDD.
[ying.huang@intel.com: fix two issues in THP optimize patch]
Link: http://lkml.kernel.org/r/87k25ed8zo.fsf@yhuang-dev.intel.com
[hannes@cmpxchg.org: extensive cleanups and simplifications, reduce code size]
Link: http://lkml.kernel.org/r/20170515112522.32457-2-ying.huang@intel.com
Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Suggested-by: Andrew Morton <akpm@linux-foundation.org> [for config option]
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> [for changes in huge_memory.c and huge_mm.h]
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Ebru Akagunduz <ebru.akagunduz@gmail.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Shaohua Li <shli@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-07-07 01:37:18 +03:00
|
|
|
extern swp_entry_t get_swap_page(struct page *page);
|
2017-07-07 01:37:21 +03:00
|
|
|
extern void put_swap_page(struct page *page, swp_entry_t entry);
|
2010-09-10 03:38:07 +04:00
|
|
|
extern swp_entry_t get_swap_page_of_type(int);
|
mm, THP, swap: delay splitting THP during swap out
Patch series "THP swap: Delay splitting THP during swapping out", v11.
This patchset is to optimize the performance of Transparent Huge Page
(THP) swap.
Recently, the performance of the storage devices improved so fast that
we cannot saturate the disk bandwidth with single logical CPU when do
page swap out even on a high-end server machine. Because the
performance of the storage device improved faster than that of single
logical CPU. And it seems that the trend will not change in the near
future. On the other hand, the THP becomes more and more popular
because of increased memory size. So it becomes necessary to optimize
THP swap performance.
The advantages of the THP swap support include:
- Batch the swap operations for the THP to reduce lock
acquiring/releasing, including allocating/freeing the swap space,
adding/deleting to/from the swap cache, and writing/reading the swap
space, etc. This will help improve the performance of the THP swap.
- The THP swap space read/write will be 2M sequential IO. It is
particularly helpful for the swap read, which are usually 4k random
IO. This will improve the performance of the THP swap too.
- It will help the memory fragmentation, especially when the THP is
heavily used by the applications. The 2M continuous pages will be
free up after THP swapping out.
- It will improve the THP utilization on the system with the swap
turned on. Because the speed for khugepaged to collapse the normal
pages into the THP is quite slow. After the THP is split during the
swapping out, it will take quite long time for the normal pages to
collapse back into the THP after being swapped in. The high THP
utilization helps the efficiency of the page based memory management
too.
There are some concerns regarding THP swap in, mainly because possible
enlarged read/write IO size (for swap in/out) may put more overhead on
the storage device. To deal with that, the THP swap in should be turned
on only when necessary. For example, it can be selected via
"always/never/madvise" logic, to be turned on globally, turned off
globally, or turned on only for VMA with MADV_HUGEPAGE, etc.
This patchset is the first step for the THP swap support. The plan is
to delay splitting THP step by step, finally avoid splitting THP during
the THP swapping out and swap out/in the THP as a whole.
As the first step, in this patchset, the splitting huge page is delayed
from almost the first step of swapping out to after allocating the swap
space for the THP and adding the THP into the swap cache. This will
reduce lock acquiring/releasing for the locks used for the swap cache
management.
With the patchset, the swap out throughput improves 15.5% (from about
3.73GB/s to about 4.31GB/s) in the vm-scalability swap-w-seq test case
with 8 processes. The test is done on a Xeon E5 v3 system. The swap
device used is a RAM simulated PMEM (persistent memory) device. To test
the sequential swapping out, the test case creates 8 processes, which
sequentially allocate and write to the anonymous pages until the RAM and
part of the swap device is used up.
This patch (of 5):
In this patch, splitting huge page is delayed from almost the first step
of swapping out to after allocating the swap space for the THP
(Transparent Huge Page) and adding the THP into the swap cache. This
will batch the corresponding operation, thus improve THP swap out
throughput.
This is the first step for the THP swap optimization. The plan is to
delay splitting the THP step by step and avoid splitting the THP
finally.
In this patch, one swap cluster is used to hold the contents of each THP
swapped out. So, the size of the swap cluster is changed to that of the
THP (Transparent Huge Page) on x86_64 architecture (512). For other
architectures which want such THP swap optimization,
ARCH_USES_THP_SWAP_CLUSTER needs to be selected in the Kconfig file for
the architecture. In effect, this will enlarge swap cluster size by 2
times on x86_64. Which may make it harder to find a free cluster when
the swap space becomes fragmented. So that, this may reduce the
continuous swap space allocation and sequential write in theory. The
performance test in 0day shows no regressions caused by this.
In the future of THP swap optimization, some information of the swapped
out THP (such as compound map count) will be recorded in the
swap_cluster_info data structure.
The mem cgroup swap accounting functions are enhanced to support charge
or uncharge a swap cluster backing a THP as a whole.
The swap cluster allocate/free functions are added to allocate/free a
swap cluster for a THP. A fair simple algorithm is used for swap
cluster allocation, that is, only the first swap device in priority list
will be tried to allocate the swap cluster. The function will fail if
the trying is not successful, and the caller will fallback to allocate a
single swap slot instead. This works good enough for normal cases. If
the difference of the number of the free swap clusters among multiple
swap devices is significant, it is possible that some THPs are split
earlier than necessary. For example, this could be caused by big size
difference among multiple swap devices.
The swap cache functions is enhanced to support add/delete THP to/from
the swap cache as a set of (HPAGE_PMD_NR) sub-pages. This may be
enhanced in the future with multi-order radix tree. But because we will
split the THP soon during swapping out, that optimization doesn't make
much sense for this first step.
The THP splitting functions are enhanced to support to split THP in swap
cache during swapping out. The page lock will be held during allocating
the swap cluster, adding the THP into the swap cache and splitting the
THP. So in the code path other than swapping out, if the THP need to be
split, the PageSwapCache(THP) will be always false.
The swap cluster is only available for SSD, so the THP swap optimization
in this patchset has no effect for HDD.
[ying.huang@intel.com: fix two issues in THP optimize patch]
Link: http://lkml.kernel.org/r/87k25ed8zo.fsf@yhuang-dev.intel.com
[hannes@cmpxchg.org: extensive cleanups and simplifications, reduce code size]
Link: http://lkml.kernel.org/r/20170515112522.32457-2-ying.huang@intel.com
Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Suggested-by: Andrew Morton <akpm@linux-foundation.org> [for config option]
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> [for changes in huge_memory.c and huge_mm.h]
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Ebru Akagunduz <ebru.akagunduz@gmail.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Shaohua Li <shli@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-07-07 01:37:18 +03:00
|
|
|
extern int get_swap_pages(int n, bool cluster, swp_entry_t swp_entries[]);
|
swap_info: swap count continuations
Swap is duplicated (reference count incremented by one) whenever the same
swap page is inserted into another mm (when forking finds a swap entry in
place of a pte, or when reclaim unmaps a pte to insert the swap entry).
swap_info_struct's vmalloc'ed swap_map is the array of these reference
counts: but what happens when the unsigned short (or unsigned char since
the preceding patch) is full? (and its high bit is kept for a cache flag)
We then lose track of it, never freeing, leaving it in use until swapoff:
at which point we _hope_ that a single pass will have found all instances,
assume there are no more, and will lose user data if we're wrong.
Swapping of KSM pages has not yet been enabled; but it is implemented,
and makes it very easy for a user to overflow the maximum swap count:
possible with ordinary process pages, but unlikely, even when pid_max
has been raised from PID_MAX_DEFAULT.
This patch implements swap count continuations: when the count overflows,
a continuation page is allocated and linked to the original vmalloc'ed
map page, and this used to hold the continuation counts for that entry
and its neighbours. These continuation pages are seldom referenced:
the common paths all work on the original swap_map, only referring to
a continuation page when the low "digit" of a count is incremented or
decremented through SWAP_MAP_MAX.
Signed-off-by: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-12-15 04:58:46 +03:00
|
|
|
extern int add_swap_count_continuation(swp_entry_t, gfp_t);
|
2009-12-15 04:58:47 +03:00
|
|
|
extern void swap_shmem_alloc(swp_entry_t);
|
swap_info: swap count continuations
Swap is duplicated (reference count incremented by one) whenever the same
swap page is inserted into another mm (when forking finds a swap entry in
place of a pte, or when reclaim unmaps a pte to insert the swap entry).
swap_info_struct's vmalloc'ed swap_map is the array of these reference
counts: but what happens when the unsigned short (or unsigned char since
the preceding patch) is full? (and its high bit is kept for a cache flag)
We then lose track of it, never freeing, leaving it in use until swapoff:
at which point we _hope_ that a single pass will have found all instances,
assume there are no more, and will lose user data if we're wrong.
Swapping of KSM pages has not yet been enabled; but it is implemented,
and makes it very easy for a user to overflow the maximum swap count:
possible with ordinary process pages, but unlikely, even when pid_max
has been raised from PID_MAX_DEFAULT.
This patch implements swap count continuations: when the count overflows,
a continuation page is allocated and linked to the original vmalloc'ed
map page, and this used to hold the continuation counts for that entry
and its neighbours. These continuation pages are seldom referenced:
the common paths all work on the original swap_map, only referring to
a continuation page when the low "digit" of a count is incremented or
decremented through SWAP_MAP_MAX.
Signed-off-by: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-12-15 04:58:46 +03:00
|
|
|
extern int swap_duplicate(swp_entry_t);
|
|
|
|
|
extern int swapcache_prepare(swp_entry_t);
|
2005-04-17 02:20:36 +04:00
|
|
|
extern void swap_free(swp_entry_t);
|
2017-02-23 02:45:36 +03:00
|
|
|
extern void swapcache_free_entries(swp_entry_t *entries, int n);
|
2009-01-07 01:40:10 +03:00
|
|
|
extern int free_swap_and_cache(swp_entry_t);
|
2007-01-06 03:36:28 +03:00
|
|
|
extern int swap_type_of(dev_t, sector_t, struct block_device **);
|
2006-03-23 13:59:59 +03:00
|
|
|
extern unsigned int count_swap_pages(int, int);
|
2009-12-15 04:58:49 +03:00
|
|
|
extern sector_t map_swap_page(struct page *, struct block_device **);
|
2006-12-07 07:34:10 +03:00
|
|
|
extern sector_t swapdev_block(int, pgoff_t);
|
shmem: replace page if mapping excludes its zone
The GMA500 GPU driver uses GEM shmem objects, but with a new twist: the
backing RAM has to be below 4GB. Not a problem while the boards
supported only 4GB: but now Intel's D2700MUD boards support 8GB, and
their GMA3600 is managed by the GMA500 driver.
shmem/tmpfs has never pretended to support hardware restrictions on the
backing memory, but it might have appeared to do so before v3.1, and
even now it works fine until a page is swapped out then back in. When
read_cache_page_gfp() supplied a freshly allocated page for copy, that
compensated for whatever choice might have been made by earlier swapin
readahead; but swapoff was likely to destroy the illusion.
We'd like to continue to support GMA500, so now add a new
shmem_should_replace_page() check on the zone when about to move a page
from swapcache to filecache (in swapin and swapoff cases), with
shmem_replace_page() to allocate and substitute a suitable page (given
gma500/gem.c's mapping_set_gfp_mask GFP_KERNEL | __GFP_DMA32).
This does involve a minor extension to mem_cgroup_replace_page_cache()
(the page may or may not have already been charged); and I've removed a
comment and call to mem_cgroup_uncharge_cache_page(), which in fact is
always a no-op while PageSwapCache.
Also removed optimization of an unlikely path in shmem_getpage_gfp(),
now that we need to check PageSwapCache more carefully (a racing caller
might already have made the copy). And at one point shmem_unuse_inode()
needs to use the hitherto private page_swapcount(), to guard against
racing with inode eviction.
It would make sense to extend shmem_should_replace_page(), to cover
cpuset and NUMA mempolicy restrictions too, but set that aside for now:
needs a cleanup of shmem mempolicy handling, and more testing, and ought
to handle swap faults in do_swap_page() as well as shmem.
Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: Christoph Hellwig <hch@infradead.org>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Cc: Stephane Marchesin <marcheu@chromium.org>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Dave Airlie <airlied@gmail.com>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: Rob Clark <rob.clark@linaro.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-05-30 02:06:38 +04:00
|
|
|
extern int page_swapcount(struct page *);
|
2017-02-23 02:45:29 +03:00
|
|
|
extern int __swp_swapcount(swp_entry_t entry);
|
2015-09-09 01:00:24 +03:00
|
|
|
extern int swp_swapcount(swp_entry_t entry);
|
2012-08-01 03:44:47 +04:00
|
|
|
extern struct swap_info_struct *page_swap_info(struct page *);
|
mm: thp: calculate the mapcount correctly for THP pages during WP faults
This will provide fully accuracy to the mapcount calculation in the
write protect faults, so page pinning will not get broken by false
positive copy-on-writes.
total_mapcount() isn't the right calculation needed in
reuse_swap_page(), so this introduces a page_trans_huge_mapcount()
that is effectively the full accurate return value for page_mapcount()
if dealing with Transparent Hugepages, however we only use the
page_trans_huge_mapcount() during COW faults where it strictly needed,
due to its higher runtime cost.
This also provide at practical zero cost the total_mapcount
information which is needed to know if we can still relocate the page
anon_vma to the local vma. If page_trans_huge_mapcount() returns 1 we
can reuse the page no matter if it's a pte or a pmd_trans_huge
triggering the fault, but we can only relocate the page anon_vma to
the local vma->anon_vma if we're sure it's only this "vma" mapping the
whole THP physical range.
Kirill A. Shutemov discovered the problem with moving the page
anon_vma to the local vma->anon_vma in a previous version of this
patch and another problem in the way page_move_anon_rmap() was called.
Andrew Morton discovered that CONFIG_SWAP=n wouldn't build in a
previous version, because reuse_swap_page must be a macro to call
page_trans_huge_mapcount from swap.h, so this uses a macro again
instead of an inline function. With this change at least it's a less
dangerous usage than it was before, because "page" is used only once
now, while with the previous code reuse_swap_page(page++) would have
called page_mapcount on page+1 and it would have increased page twice
instead of just once.
Dean Luick noticed an uninitialized variable that could result in a
rmap inefficiency for the non-THP case in a previous version.
Mike Marciniszyn said:
: Our RDMA tests are seeing an issue with memory locking that bisects to
: commit 61f5d698cc97 ("mm: re-enable THP")
:
: The test program registers two rather large MRs (512M) and RDMA
: writes data to a passive peer using the first and RDMA reads it back
: into the second MR and compares that data. The sizes are chosen randomly
: between 0 and 1024 bytes.
:
: The test will get through a few (<= 4 iterations) and then gets a
: compare error.
:
: Tracing indicates the kernel logical addresses associated with the individual
: pages at registration ARE correct , the data in the "RDMA read response only"
: packets ARE correct.
:
: The "corruption" occurs when the packet crosse two pages that are not physically
: contiguous. The second page reads back as zero in the program.
:
: It looks like the user VA at the point of the compare error no longer points to
: the same physical address as was registered.
:
: This patch totally resolves the issue!
Link: http://lkml.kernel.org/r/1462547040-1737-2-git-send-email-aarcange@redhat.com
Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
Reviewed-by: "Kirill A. Shutemov" <kirill@shutemov.name>
Reviewed-by: Dean Luick <dean.luick@intel.com>
Tested-by: Alex Williamson <alex.williamson@redhat.com>
Tested-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Tested-by: Josh Collier <josh.d.collier@intel.com>
Cc: Marc Haber <mh+linux-kernel@zugschlus.de>
Cc: <stable@vger.kernel.org> [4.5]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-05-13 01:42:25 +03:00
|
|
|
extern bool reuse_swap_page(struct page *, int *);
|
2009-01-07 01:39:36 +03:00
|
|
|
extern int try_to_free_swap(struct page *);
|
2005-04-17 02:20:36 +04:00
|
|
|
struct backing_dev_info;
|
mm/swap: split swap cache into 64MB trunks
The patch is to improve the scalability of the swap out/in via using
fine grained locks for the swap cache. In current kernel, one address
space will be used for each swap device. And in the common
configuration, the number of the swap device is very small (one is
typical). This causes the heavy lock contention on the radix tree of
the address space if multiple tasks swap out/in concurrently.
But in fact, there is no dependency between pages in the swap cache. So
that, we can split the one shared address space for each swap device
into several address spaces to reduce the lock contention. In the
patch, the shared address space is split into 64MB trunks. 64MB is
chosen to balance the memory space usage and effect of lock contention
reduction.
The size of struct address_space on x86_64 architecture is 408B, so with
the patch, 6528B more memory will be used for every 1GB swap space on
x86_64 architecture.
One address space is still shared for the swap entries in the same 64M
trunks. To avoid lock contention for the first round of swap space
allocation, the order of the swap clusters in the initial free clusters
list is changed. The swap space distance between the consecutive swap
clusters in the free cluster list is at least 64M. After the first
round of allocation, the swap clusters are expected to be freed
randomly, so the lock contention should be reduced effectively.
Link: http://lkml.kernel.org/r/735bab895e64c930581ffb0a05b661e01da82bc5.1484082593.git.tim.c.chen@linux.intel.com
Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
Signed-off-by: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Aaron Lu <aaron.lu@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Christian Borntraeger <borntraeger@de.ibm.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: Huang Ying <ying.huang@intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Jonathan Corbet <corbet@lwn.net> escreveu:
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Shaohua Li <shli@kernel.org>
Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-02-23 02:45:26 +03:00
|
|
|
extern int init_swap_address_space(unsigned int type, unsigned long nr_pages);
|
|
|
|
|
extern void exit_swap_address_space(unsigned int type);
|
2005-04-17 02:20:36 +04:00
|
|
|
|
|
|
|
|
#else /* CONFIG_SWAP */
|
|
|
|
|
|
2013-09-12 01:21:29 +04:00
|
|
|
#define swap_address_space(entry) (NULL)
|
swap: add per-partition lock for swapfile
swap_lock is heavily contended when I test swap to 3 fast SSD (even
slightly slower than swap to 2 such SSD). The main contention comes
from swap_info_get(). This patch tries to fix the gap with adding a new
per-partition lock.
Global data like nr_swapfiles, total_swap_pages, least_priority and
swap_list are still protected by swap_lock.
nr_swap_pages is an atomic now, it can be changed without swap_lock. In
theory, it's possible get_swap_page() finds no swap pages but actually
there are free swap pages. But sounds not a big problem.
Accessing partition specific data (like scan_swap_map and so on) is only
protected by swap_info_struct.lock.
Changing swap_info_struct.flags need hold swap_lock and
swap_info_struct.lock, because scan_scan_map() will check it. read the
flags is ok with either the locks hold.
If both swap_lock and swap_info_struct.lock must be hold, we always hold
the former first to avoid deadlock.
swap_entry_free() can change swap_list. To delete that code, we add a
new highest_priority_index. Whenever get_swap_page() is called, we
check it. If it's valid, we use it.
It's a pity get_swap_page() still holds swap_lock(). But in practice,
swap_lock() isn't heavily contended in my test with this patch (or I can
say there are other much more heavier bottlenecks like TLB flush). And
BTW, looks get_swap_page() doesn't really need the lock. We never free
swap_info[] and we check SWAP_WRITEOK flag. The only risk without the
lock is we could swapout to some low priority swap, but we can quickly
recover after several rounds of swap, so sounds not a big deal to me.
But I'd prefer to fix this if it's a real problem.
"swap: make each swap partition have one address_space" improved the
swapout speed from 1.7G/s to 2G/s. This patch further improves the
speed to 2.3G/s, so around 15% improvement. It's a multi-process test,
so TLB flush isn't the biggest bottleneck before the patches.
[arnd@arndb.de: fix it for nommu]
[hughd@google.com: add missing unlock]
[minchan@kernel.org: get rid of lockdep whinge on sys_swapon]
Signed-off-by: Shaohua Li <shli@fusionio.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Seth Jennings <sjenning@linux.vnet.ibm.com>
Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: Xiao Guangrong <xiaoguangrong@linux.vnet.ibm.com>
Cc: Dan Magenheimer <dan.magenheimer@oracle.com>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Hugh Dickins <hughd@google.com>
Signed-off-by: Minchan Kim <minchan@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-23 04:34:38 +04:00
|
|
|
#define get_nr_swap_pages() 0L
|
2009-01-07 01:39:41 +03:00
|
|
|
#define total_swap_pages 0L
|
2013-02-23 04:34:37 +04:00
|
|
|
#define total_swapcache_pages() 0UL
|
swap: add per-partition lock for swapfile
swap_lock is heavily contended when I test swap to 3 fast SSD (even
slightly slower than swap to 2 such SSD). The main contention comes
from swap_info_get(). This patch tries to fix the gap with adding a new
per-partition lock.
Global data like nr_swapfiles, total_swap_pages, least_priority and
swap_list are still protected by swap_lock.
nr_swap_pages is an atomic now, it can be changed without swap_lock. In
theory, it's possible get_swap_page() finds no swap pages but actually
there are free swap pages. But sounds not a big problem.
Accessing partition specific data (like scan_swap_map and so on) is only
protected by swap_info_struct.lock.
Changing swap_info_struct.flags need hold swap_lock and
swap_info_struct.lock, because scan_scan_map() will check it. read the
flags is ok with either the locks hold.
If both swap_lock and swap_info_struct.lock must be hold, we always hold
the former first to avoid deadlock.
swap_entry_free() can change swap_list. To delete that code, we add a
new highest_priority_index. Whenever get_swap_page() is called, we
check it. If it's valid, we use it.
It's a pity get_swap_page() still holds swap_lock(). But in practice,
swap_lock() isn't heavily contended in my test with this patch (or I can
say there are other much more heavier bottlenecks like TLB flush). And
BTW, looks get_swap_page() doesn't really need the lock. We never free
swap_info[] and we check SWAP_WRITEOK flag. The only risk without the
lock is we could swapout to some low priority swap, but we can quickly
recover after several rounds of swap, so sounds not a big deal to me.
But I'd prefer to fix this if it's a real problem.
"swap: make each swap partition have one address_space" improved the
swapout speed from 1.7G/s to 2G/s. This patch further improves the
speed to 2.3G/s, so around 15% improvement. It's a multi-process test,
so TLB flush isn't the biggest bottleneck before the patches.
[arnd@arndb.de: fix it for nommu]
[hughd@google.com: add missing unlock]
[minchan@kernel.org: get rid of lockdep whinge on sys_swapon]
Signed-off-by: Shaohua Li <shli@fusionio.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Seth Jennings <sjenning@linux.vnet.ibm.com>
Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: Xiao Guangrong <xiaoguangrong@linux.vnet.ibm.com>
Cc: Dan Magenheimer <dan.magenheimer@oracle.com>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Hugh Dickins <hughd@google.com>
Signed-off-by: Minchan Kim <minchan@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-23 04:34:38 +04:00
|
|
|
#define vm_swap_full() 0
|
2005-04-17 02:20:36 +04:00
|
|
|
|
|
|
|
|
#define si_swapinfo(val) \
|
|
|
|
|
do { (val)->freeswap = (val)->totalswap = 0; } while (0)
|
2005-08-07 20:42:24 +04:00
|
|
|
/* only sparc can not include linux/pagemap.h in this file
|
2016-04-01 15:29:48 +03:00
|
|
|
* so leave put_page and release_pages undeclared... */
|
2005-04-17 02:20:36 +04:00
|
|
|
#define free_page_and_swap_cache(page) \
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 15:29:47 +03:00
|
|
|
put_page(page)
|
2005-04-17 02:20:36 +04:00
|
|
|
#define free_pages_and_swap_cache(pages, nr) \
|
2014-06-05 03:10:22 +04:00
|
|
|
release_pages((pages), (nr), false);
|
2005-04-17 02:20:36 +04:00
|
|
|
|
2006-06-23 13:03:42 +04:00
|
|
|
static inline void show_swap_cache_info(void)
|
|
|
|
|
{
|
|
|
|
|
}
|
|
|
|
|
|
2017-09-09 02:11:43 +03:00
|
|
|
#define free_swap_and_cache(e) ({(is_migration_entry(e) || is_device_private_entry(e));})
|
|
|
|
|
#define swapcache_prepare(e) ({(is_migration_entry(e) || is_device_private_entry(e));})
|
2006-06-23 13:03:42 +04:00
|
|
|
|
swap_info: swap count continuations
Swap is duplicated (reference count incremented by one) whenever the same
swap page is inserted into another mm (when forking finds a swap entry in
place of a pte, or when reclaim unmaps a pte to insert the swap entry).
swap_info_struct's vmalloc'ed swap_map is the array of these reference
counts: but what happens when the unsigned short (or unsigned char since
the preceding patch) is full? (and its high bit is kept for a cache flag)
We then lose track of it, never freeing, leaving it in use until swapoff:
at which point we _hope_ that a single pass will have found all instances,
assume there are no more, and will lose user data if we're wrong.
Swapping of KSM pages has not yet been enabled; but it is implemented,
and makes it very easy for a user to overflow the maximum swap count:
possible with ordinary process pages, but unlikely, even when pid_max
has been raised from PID_MAX_DEFAULT.
This patch implements swap count continuations: when the count overflows,
a continuation page is allocated and linked to the original vmalloc'ed
map page, and this used to hold the continuation counts for that entry
and its neighbours. These continuation pages are seldom referenced:
the common paths all work on the original swap_map, only referring to
a continuation page when the low "digit" of a count is incremented or
decremented through SWAP_MAP_MAX.
Signed-off-by: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-12-15 04:58:46 +03:00
|
|
|
static inline int add_swap_count_continuation(swp_entry_t swp, gfp_t gfp_mask)
|
2009-06-17 02:32:53 +04:00
|
|
|
{
|
swap_info: swap count continuations
Swap is duplicated (reference count incremented by one) whenever the same
swap page is inserted into another mm (when forking finds a swap entry in
place of a pte, or when reclaim unmaps a pte to insert the swap entry).
swap_info_struct's vmalloc'ed swap_map is the array of these reference
counts: but what happens when the unsigned short (or unsigned char since
the preceding patch) is full? (and its high bit is kept for a cache flag)
We then lose track of it, never freeing, leaving it in use until swapoff:
at which point we _hope_ that a single pass will have found all instances,
assume there are no more, and will lose user data if we're wrong.
Swapping of KSM pages has not yet been enabled; but it is implemented,
and makes it very easy for a user to overflow the maximum swap count:
possible with ordinary process pages, but unlikely, even when pid_max
has been raised from PID_MAX_DEFAULT.
This patch implements swap count continuations: when the count overflows,
a continuation page is allocated and linked to the original vmalloc'ed
map page, and this used to hold the continuation counts for that entry
and its neighbours. These continuation pages are seldom referenced:
the common paths all work on the original swap_map, only referring to
a continuation page when the low "digit" of a count is incremented or
decremented through SWAP_MAP_MAX.
Signed-off-by: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-12-15 04:58:46 +03:00
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
2009-12-15 04:58:47 +03:00
|
|
|
static inline void swap_shmem_alloc(swp_entry_t swp)
|
|
|
|
|
{
|
|
|
|
|
}
|
|
|
|
|
|
swap_info: swap count continuations
Swap is duplicated (reference count incremented by one) whenever the same
swap page is inserted into another mm (when forking finds a swap entry in
place of a pte, or when reclaim unmaps a pte to insert the swap entry).
swap_info_struct's vmalloc'ed swap_map is the array of these reference
counts: but what happens when the unsigned short (or unsigned char since
the preceding patch) is full? (and its high bit is kept for a cache flag)
We then lose track of it, never freeing, leaving it in use until swapoff:
at which point we _hope_ that a single pass will have found all instances,
assume there are no more, and will lose user data if we're wrong.
Swapping of KSM pages has not yet been enabled; but it is implemented,
and makes it very easy for a user to overflow the maximum swap count:
possible with ordinary process pages, but unlikely, even when pid_max
has been raised from PID_MAX_DEFAULT.
This patch implements swap count continuations: when the count overflows,
a continuation page is allocated and linked to the original vmalloc'ed
map page, and this used to hold the continuation counts for that entry
and its neighbours. These continuation pages are seldom referenced:
the common paths all work on the original swap_map, only referring to
a continuation page when the low "digit" of a count is incremented or
decremented through SWAP_MAP_MAX.
Signed-off-by: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-12-15 04:58:46 +03:00
|
|
|
static inline int swap_duplicate(swp_entry_t swp)
|
|
|
|
|
{
|
|
|
|
|
return 0;
|
2009-06-17 02:32:53 +04:00
|
|
|
}
|
|
|
|
|
|
2006-06-23 13:03:42 +04:00
|
|
|
static inline void swap_free(swp_entry_t swp)
|
|
|
|
|
{
|
|
|
|
|
}
|
|
|
|
|
|
2017-07-07 01:37:21 +03:00
|
|
|
static inline void put_swap_page(struct page *page, swp_entry_t swp)
|
2009-06-17 02:32:52 +04:00
|
|
|
{
|
|
|
|
|
}
|
|
|
|
|
|
swapin needs gfp_mask for loop on tmpfs
Building in a filesystem on a loop device on a tmpfs file can hang when
swapping, the loop thread caught in that infamous throttle_vm_writeout.
In theory this is a long standing problem, which I've either never seen in
practice, or long ago suppressed the recollection, after discounting my load
and my tmpfs size as unrealistically high. But now, with the new aops, it has
become easy to hang on one machine.
Loop used to grab_cache_page before the old prepare_write to tmpfs, which
seems to have been enough to free up some memory for any swapin needed; but
the new write_begin lets tmpfs find or allocate the page (much nicer, since
grab_cache_page missed tmpfs pages in swapcache).
When allocating a fresh page, tmpfs respects loop's mapping_gfp_mask, which
has __GFP_IO|__GFP_FS stripped off, and throttle_vm_writeout is designed to
break out when __GFP_IO or GFP_FS is unset; but when tmfps swaps in,
read_swap_cache_async allocates with GFP_HIGHUSER_MOVABLE regardless of the
mapping_gfp_mask - hence the hang.
So, pass gfp_mask down the line from shmem_getpage to shmem_swapin to
swapin_readahead to read_swap_cache_async to add_to_swap_cache.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Acked-by: Rik van Riel <riel@redhat.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-05 09:28:42 +03:00
|
|
|
static inline struct page *swapin_readahead(swp_entry_t swp, gfp_t gfp_mask,
|
2006-06-23 13:03:42 +04:00
|
|
|
struct vm_area_struct *vma, unsigned long addr)
|
|
|
|
|
{
|
|
|
|
|
return NULL;
|
|
|
|
|
}
|
|
|
|
|
|
mm, swap: VMA based swap readahead
The swap readahead is an important mechanism to reduce the swap in
latency. Although pure sequential memory access pattern isn't very
popular for anonymous memory, the space locality is still considered
valid.
In the original swap readahead implementation, the consecutive blocks in
swap device are readahead based on the global space locality estimation.
But the consecutive blocks in swap device just reflect the order of page
reclaiming, don't necessarily reflect the access pattern in virtual
memory. And the different tasks in the system may have different access
patterns, which makes the global space locality estimation incorrect.
In this patch, when page fault occurs, the virtual pages near the fault
address will be readahead instead of the swap slots near the fault swap
slot in swap device. This avoid to readahead the unrelated swap slots.
At the same time, the swap readahead is changed to work on per-VMA from
globally. So that the different access patterns of the different VMAs
could be distinguished, and the different readahead policy could be
applied accordingly. The original core readahead detection and scaling
algorithm is reused, because it is an effect algorithm to detect the
space locality.
The test and result is as follow,
Common test condition
=====================
Test Machine: Xeon E5 v3 (2 sockets, 72 threads, 32G RAM) Swap device:
NVMe disk
Micro-benchmark with combined access pattern
============================================
vm-scalability, sequential swap test case, 4 processes to eat 50G
virtual memory space, repeat the sequential memory writing until 300
seconds. The first round writing will trigger swap out, the following
rounds will trigger sequential swap in and out.
At the same time, run vm-scalability random swap test case in
background, 8 processes to eat 30G virtual memory space, repeat the
random memory write until 300 seconds. This will trigger random swap-in
in the background.
This is a combined workload with sequential and random memory accessing
at the same time. The result (for sequential workload) is as follow,
Base Optimized
---- ---------
throughput 345413 KB/s 414029 KB/s (+19.9%)
latency.average 97.14 us 61.06 us (-37.1%)
latency.50th 2 us 1 us
latency.60th 2 us 1 us
latency.70th 98 us 2 us
latency.80th 160 us 2 us
latency.90th 260 us 217 us
latency.95th 346 us 369 us
latency.99th 1.34 ms 1.09 ms
ra_hit% 52.69% 99.98%
The original swap readahead algorithm is confused by the background
random access workload, so readahead hit rate is lower. The VMA-base
readahead algorithm works much better.
Linpack
=======
The test memory size is bigger than RAM to trigger swapping.
Base Optimized
---- ---------
elapsed_time 393.49 s 329.88 s (-16.2%)
ra_hit% 86.21% 98.82%
The score of base and optimized kernel hasn't visible changes. But the
elapsed time reduced and readahead hit rate improved, so the optimized
kernel runs better for startup and tear down stages. And the absolute
value of readahead hit rate is high, shows that the space locality is
still valid in some practical workloads.
Link: http://lkml.kernel.org/r/20170807054038.1843-4-ying.huang@intel.com
Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Shaohua Li <shli@kernel.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Fengguang Wu <fengguang.wu@intel.com>
Cc: Tim Chen <tim.c.chen@intel.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-09-07 02:24:36 +03:00
|
|
|
static inline bool swap_use_vma_readahead(void)
|
|
|
|
|
{
|
|
|
|
|
return false;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static inline struct page *swap_readahead_detect(
|
|
|
|
|
struct vm_fault *vmf, struct vma_swap_readahead *swap_ra)
|
|
|
|
|
{
|
|
|
|
|
return NULL;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static inline struct page *do_swap_page_readahead(
|
|
|
|
|
swp_entry_t fentry, gfp_t gfp_mask,
|
|
|
|
|
struct vm_fault *vmf, struct vma_swap_readahead *swap_ra)
|
|
|
|
|
{
|
|
|
|
|
return NULL;
|
|
|
|
|
}
|
|
|
|
|
|
shmem: writepage directly to swap
Synopsis: if shmem_writepage calls swap_writepage directly, most shmem
swap loads benefit, and a catastrophic interaction between SLUB and some
flash storage is avoided.
shmem_writepage() has always been peculiar in making no attempt to write:
it has just transferred a shmem page from file cache to swap cache, then
let that page make its way around the LRU again before being written and
freed.
The idea was that people use tmpfs because they want those pages to stay
in RAM; so although we give it an overflow to swap, we should resist
writing too soon, giving those pages a second chance before they can be
reclaimed.
That was always questionable, and I've toyed with this patch for years;
but never had a clear justification to depart from the original design.
It became more questionable in 2.6.28, when the split LRU patches classed
shmem and tmpfs pages as SwapBacked rather than as file_cache: that in
itself gives them more resistance to reclaim than normal file pages. I
prepared this patch for 2.6.29, but the merge window arrived before I'd
completed gathering statistics to justify sending it in.
Then while comparing SLQB against SLUB, running SLUB on a laptop I'd
habitually used with SLAB, I found SLUB to run my tmpfs kbuild swapping
tests five times slower than SLAB or SLQB - other machines slower too, but
nowhere near so bad. Simpler "cp -a" swapping tests showed the same.
slub_max_order=0 brings sanity to all, but heavy swapping is too far from
normal to justify such a tuning. The crucial factor on that laptop turns
out to be that I'm using an SD card for swap. What happens is this:
By default, SLUB uses order-2 pages for shmem_inode_cache (and many other
fs inodes), so creating tmpfs files under memory pressure brings lumpy
reclaim into play. One subpage of the order is chosen from the bottom of
the LRU as usual, then the other three picked out from their random
positions on the LRUs.
In a tmpfs load, many of these pages will be ones which already passed
through shmem_writepage, so already have swap allocated. And though their
offsets on swap were probably allocated sequentially, now that the pages
are picked off at random, their swap offsets are scattered.
But the flash storage on the SD card is very sensitive to having its
writes merged: once swap is written at scattered offsets, performance
falls apart. Rotating disk seeks increase too, but less disastrously.
So: stop giving shmem/tmpfs pages a second pass around the LRU, write them
out to swap as soon as their swap has been allocated.
It's surely possible to devise an artificial load which runs faster the
old way, one whose sizing is such that the tmpfs pages on their second
pass are the ones that are wanted again, and other pages not.
But I've not yet found such a load: on all machines, under the loads I've
tried, immediate swap_writepage speeds up shmem swapping: especially when
using the SLUB allocator (and more effectively than slub_max_order=0), but
also with the others; and it also reduces the variance between runs. How
much faster varies widely: a factor of five is rare, 5% is common.
One load which might have suffered: imagine a swapping shmem load in a
limited mem_cgroup on a machine with plenty of memory. Before 2.6.29 the
swapcache was not charged, and such a load would have run quickest with
the shmem swapcache never written to swap. But now swapcache is charged,
so even this load benefits from shmem_writepage directly to swap.
Apologies for the #ifndef CONFIG_SWAP swap_writepage() stub in swap.h:
it's silly because that will never get called; but refactoring shmem.c
sensibly according to CONFIG_SWAP will be a separate task.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Acked-by: Pekka Enberg <penberg@cs.helsinki.fi>
Acked-by: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-04-01 02:23:33 +04:00
|
|
|
static inline int swap_writepage(struct page *p, struct writeback_control *wbc)
|
|
|
|
|
{
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
mm, swap: VMA based swap readahead
The swap readahead is an important mechanism to reduce the swap in
latency. Although pure sequential memory access pattern isn't very
popular for anonymous memory, the space locality is still considered
valid.
In the original swap readahead implementation, the consecutive blocks in
swap device are readahead based on the global space locality estimation.
But the consecutive blocks in swap device just reflect the order of page
reclaiming, don't necessarily reflect the access pattern in virtual
memory. And the different tasks in the system may have different access
patterns, which makes the global space locality estimation incorrect.
In this patch, when page fault occurs, the virtual pages near the fault
address will be readahead instead of the swap slots near the fault swap
slot in swap device. This avoid to readahead the unrelated swap slots.
At the same time, the swap readahead is changed to work on per-VMA from
globally. So that the different access patterns of the different VMAs
could be distinguished, and the different readahead policy could be
applied accordingly. The original core readahead detection and scaling
algorithm is reused, because it is an effect algorithm to detect the
space locality.
The test and result is as follow,
Common test condition
=====================
Test Machine: Xeon E5 v3 (2 sockets, 72 threads, 32G RAM) Swap device:
NVMe disk
Micro-benchmark with combined access pattern
============================================
vm-scalability, sequential swap test case, 4 processes to eat 50G
virtual memory space, repeat the sequential memory writing until 300
seconds. The first round writing will trigger swap out, the following
rounds will trigger sequential swap in and out.
At the same time, run vm-scalability random swap test case in
background, 8 processes to eat 30G virtual memory space, repeat the
random memory write until 300 seconds. This will trigger random swap-in
in the background.
This is a combined workload with sequential and random memory accessing
at the same time. The result (for sequential workload) is as follow,
Base Optimized
---- ---------
throughput 345413 KB/s 414029 KB/s (+19.9%)
latency.average 97.14 us 61.06 us (-37.1%)
latency.50th 2 us 1 us
latency.60th 2 us 1 us
latency.70th 98 us 2 us
latency.80th 160 us 2 us
latency.90th 260 us 217 us
latency.95th 346 us 369 us
latency.99th 1.34 ms 1.09 ms
ra_hit% 52.69% 99.98%
The original swap readahead algorithm is confused by the background
random access workload, so readahead hit rate is lower. The VMA-base
readahead algorithm works much better.
Linpack
=======
The test memory size is bigger than RAM to trigger swapping.
Base Optimized
---- ---------
elapsed_time 393.49 s 329.88 s (-16.2%)
ra_hit% 86.21% 98.82%
The score of base and optimized kernel hasn't visible changes. But the
elapsed time reduced and readahead hit rate improved, so the optimized
kernel runs better for startup and tear down stages. And the absolute
value of readahead hit rate is high, shows that the space locality is
still valid in some practical workloads.
Link: http://lkml.kernel.org/r/20170807054038.1843-4-ying.huang@intel.com
Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Shaohua Li <shli@kernel.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Fengguang Wu <fengguang.wu@intel.com>
Cc: Tim Chen <tim.c.chen@intel.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-09-07 02:24:36 +03:00
|
|
|
static inline struct page *lookup_swap_cache(swp_entry_t swp,
|
|
|
|
|
struct vm_area_struct *vma,
|
|
|
|
|
unsigned long addr)
|
2006-06-23 13:03:42 +04:00
|
|
|
{
|
|
|
|
|
return NULL;
|
|
|
|
|
}
|
|
|
|
|
|
2017-07-07 01:37:24 +03:00
|
|
|
static inline int add_to_swap(struct page *page)
|
2009-01-07 01:39:40 +03:00
|
|
|
{
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
2008-02-05 09:28:50 +03:00
|
|
|
static inline int add_to_swap_cache(struct page *page, swp_entry_t entry,
|
|
|
|
|
gfp_t gfp_mask)
|
2006-06-23 13:03:42 +04:00
|
|
|
{
|
2008-02-05 09:28:50 +03:00
|
|
|
return -1;
|
2006-06-23 13:03:42 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static inline void __delete_from_swap_cache(struct page *page)
|
|
|
|
|
{
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static inline void delete_from_swap_cache(struct page *page)
|
|
|
|
|
{
|
|
|
|
|
}
|
|
|
|
|
|
shmem: replace page if mapping excludes its zone
The GMA500 GPU driver uses GEM shmem objects, but with a new twist: the
backing RAM has to be below 4GB. Not a problem while the boards
supported only 4GB: but now Intel's D2700MUD boards support 8GB, and
their GMA3600 is managed by the GMA500 driver.
shmem/tmpfs has never pretended to support hardware restrictions on the
backing memory, but it might have appeared to do so before v3.1, and
even now it works fine until a page is swapped out then back in. When
read_cache_page_gfp() supplied a freshly allocated page for copy, that
compensated for whatever choice might have been made by earlier swapin
readahead; but swapoff was likely to destroy the illusion.
We'd like to continue to support GMA500, so now add a new
shmem_should_replace_page() check on the zone when about to move a page
from swapcache to filecache (in swapin and swapoff cases), with
shmem_replace_page() to allocate and substitute a suitable page (given
gma500/gem.c's mapping_set_gfp_mask GFP_KERNEL | __GFP_DMA32).
This does involve a minor extension to mem_cgroup_replace_page_cache()
(the page may or may not have already been charged); and I've removed a
comment and call to mem_cgroup_uncharge_cache_page(), which in fact is
always a no-op while PageSwapCache.
Also removed optimization of an unlikely path in shmem_getpage_gfp(),
now that we need to check PageSwapCache more carefully (a racing caller
might already have made the copy). And at one point shmem_unuse_inode()
needs to use the hitherto private page_swapcount(), to guard against
racing with inode eviction.
It would make sense to extend shmem_should_replace_page(), to cover
cpuset and NUMA mempolicy restrictions too, but set that aside for now:
needs a cleanup of shmem mempolicy handling, and more testing, and ought
to handle swap faults in do_swap_page() as well as shmem.
Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: Christoph Hellwig <hch@infradead.org>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Cc: Stephane Marchesin <marcheu@chromium.org>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Dave Airlie <airlied@gmail.com>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: Rob Clark <rob.clark@linaro.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-05-30 02:06:38 +04:00
|
|
|
static inline int page_swapcount(struct page *page)
|
|
|
|
|
{
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
2017-02-23 02:45:29 +03:00
|
|
|
static inline int __swp_swapcount(swp_entry_t entry)
|
|
|
|
|
{
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
2015-09-09 01:00:24 +03:00
|
|
|
static inline int swp_swapcount(swp_entry_t entry)
|
|
|
|
|
{
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
2017-09-07 02:22:19 +03:00
|
|
|
#define reuse_swap_page(page, total_map_swapcount) \
|
|
|
|
|
(page_trans_huge_mapcount(page, total_map_swapcount) == 1)
|
2005-04-17 02:20:36 +04:00
|
|
|
|
2009-01-07 01:39:36 +03:00
|
|
|
static inline int try_to_free_swap(struct page *page)
|
2008-10-19 07:26:23 +04:00
|
|
|
{
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
mm, THP, swap: delay splitting THP during swap out
Patch series "THP swap: Delay splitting THP during swapping out", v11.
This patchset is to optimize the performance of Transparent Huge Page
(THP) swap.
Recently, the performance of the storage devices improved so fast that
we cannot saturate the disk bandwidth with single logical CPU when do
page swap out even on a high-end server machine. Because the
performance of the storage device improved faster than that of single
logical CPU. And it seems that the trend will not change in the near
future. On the other hand, the THP becomes more and more popular
because of increased memory size. So it becomes necessary to optimize
THP swap performance.
The advantages of the THP swap support include:
- Batch the swap operations for the THP to reduce lock
acquiring/releasing, including allocating/freeing the swap space,
adding/deleting to/from the swap cache, and writing/reading the swap
space, etc. This will help improve the performance of the THP swap.
- The THP swap space read/write will be 2M sequential IO. It is
particularly helpful for the swap read, which are usually 4k random
IO. This will improve the performance of the THP swap too.
- It will help the memory fragmentation, especially when the THP is
heavily used by the applications. The 2M continuous pages will be
free up after THP swapping out.
- It will improve the THP utilization on the system with the swap
turned on. Because the speed for khugepaged to collapse the normal
pages into the THP is quite slow. After the THP is split during the
swapping out, it will take quite long time for the normal pages to
collapse back into the THP after being swapped in. The high THP
utilization helps the efficiency of the page based memory management
too.
There are some concerns regarding THP swap in, mainly because possible
enlarged read/write IO size (for swap in/out) may put more overhead on
the storage device. To deal with that, the THP swap in should be turned
on only when necessary. For example, it can be selected via
"always/never/madvise" logic, to be turned on globally, turned off
globally, or turned on only for VMA with MADV_HUGEPAGE, etc.
This patchset is the first step for the THP swap support. The plan is
to delay splitting THP step by step, finally avoid splitting THP during
the THP swapping out and swap out/in the THP as a whole.
As the first step, in this patchset, the splitting huge page is delayed
from almost the first step of swapping out to after allocating the swap
space for the THP and adding the THP into the swap cache. This will
reduce lock acquiring/releasing for the locks used for the swap cache
management.
With the patchset, the swap out throughput improves 15.5% (from about
3.73GB/s to about 4.31GB/s) in the vm-scalability swap-w-seq test case
with 8 processes. The test is done on a Xeon E5 v3 system. The swap
device used is a RAM simulated PMEM (persistent memory) device. To test
the sequential swapping out, the test case creates 8 processes, which
sequentially allocate and write to the anonymous pages until the RAM and
part of the swap device is used up.
This patch (of 5):
In this patch, splitting huge page is delayed from almost the first step
of swapping out to after allocating the swap space for the THP
(Transparent Huge Page) and adding the THP into the swap cache. This
will batch the corresponding operation, thus improve THP swap out
throughput.
This is the first step for the THP swap optimization. The plan is to
delay splitting the THP step by step and avoid splitting the THP
finally.
In this patch, one swap cluster is used to hold the contents of each THP
swapped out. So, the size of the swap cluster is changed to that of the
THP (Transparent Huge Page) on x86_64 architecture (512). For other
architectures which want such THP swap optimization,
ARCH_USES_THP_SWAP_CLUSTER needs to be selected in the Kconfig file for
the architecture. In effect, this will enlarge swap cluster size by 2
times on x86_64. Which may make it harder to find a free cluster when
the swap space becomes fragmented. So that, this may reduce the
continuous swap space allocation and sequential write in theory. The
performance test in 0day shows no regressions caused by this.
In the future of THP swap optimization, some information of the swapped
out THP (such as compound map count) will be recorded in the
swap_cluster_info data structure.
The mem cgroup swap accounting functions are enhanced to support charge
or uncharge a swap cluster backing a THP as a whole.
The swap cluster allocate/free functions are added to allocate/free a
swap cluster for a THP. A fair simple algorithm is used for swap
cluster allocation, that is, only the first swap device in priority list
will be tried to allocate the swap cluster. The function will fail if
the trying is not successful, and the caller will fallback to allocate a
single swap slot instead. This works good enough for normal cases. If
the difference of the number of the free swap clusters among multiple
swap devices is significant, it is possible that some THPs are split
earlier than necessary. For example, this could be caused by big size
difference among multiple swap devices.
The swap cache functions is enhanced to support add/delete THP to/from
the swap cache as a set of (HPAGE_PMD_NR) sub-pages. This may be
enhanced in the future with multi-order radix tree. But because we will
split the THP soon during swapping out, that optimization doesn't make
much sense for this first step.
The THP splitting functions are enhanced to support to split THP in swap
cache during swapping out. The page lock will be held during allocating
the swap cluster, adding the THP into the swap cache and splitting the
THP. So in the code path other than swapping out, if the THP need to be
split, the PageSwapCache(THP) will be always false.
The swap cluster is only available for SSD, so the THP swap optimization
in this patchset has no effect for HDD.
[ying.huang@intel.com: fix two issues in THP optimize patch]
Link: http://lkml.kernel.org/r/87k25ed8zo.fsf@yhuang-dev.intel.com
[hannes@cmpxchg.org: extensive cleanups and simplifications, reduce code size]
Link: http://lkml.kernel.org/r/20170515112522.32457-2-ying.huang@intel.com
Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Suggested-by: Andrew Morton <akpm@linux-foundation.org> [for config option]
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> [for changes in huge_memory.c and huge_mm.h]
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Ebru Akagunduz <ebru.akagunduz@gmail.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Shaohua Li <shli@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-07-07 01:37:18 +03:00
|
|
|
static inline swp_entry_t get_swap_page(struct page *page)
|
2005-04-17 02:20:36 +04:00
|
|
|
{
|
|
|
|
|
swp_entry_t entry;
|
|
|
|
|
entry.val = 0;
|
|
|
|
|
return entry;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
#endif /* CONFIG_SWAP */
|
2016-01-21 02:03:05 +03:00
|
|
|
|
2017-09-07 02:22:34 +03:00
|
|
|
#ifdef CONFIG_THP_SWAP
|
|
|
|
|
extern int split_swap_cluster(swp_entry_t entry);
|
|
|
|
|
#else
|
|
|
|
|
static inline int split_swap_cluster(swp_entry_t entry)
|
|
|
|
|
{
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
#endif
|
|
|
|
|
|
2016-01-21 02:03:05 +03:00
|
|
|
#ifdef CONFIG_MEMCG
|
|
|
|
|
static inline int mem_cgroup_swappiness(struct mem_cgroup *memcg)
|
|
|
|
|
{
|
2016-05-06 02:22:03 +03:00
|
|
|
/* Cgroup2 doesn't have per-cgroup swappiness */
|
|
|
|
|
if (cgroup_subsys_on_dfl(memory_cgrp_subsys))
|
|
|
|
|
return vm_swappiness;
|
|
|
|
|
|
2016-01-21 02:03:05 +03:00
|
|
|
/* root ? */
|
|
|
|
|
if (mem_cgroup_disabled() || !memcg->css.parent)
|
|
|
|
|
return vm_swappiness;
|
|
|
|
|
|
|
|
|
|
return memcg->swappiness;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
#else
|
|
|
|
|
static inline int mem_cgroup_swappiness(struct mem_cgroup *mem)
|
|
|
|
|
{
|
|
|
|
|
return vm_swappiness;
|
|
|
|
|
}
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
#ifdef CONFIG_MEMCG_SWAP
|
|
|
|
|
extern void mem_cgroup_swapout(struct page *page, swp_entry_t entry);
|
|
|
|
|
extern int mem_cgroup_try_charge_swap(struct page *page, swp_entry_t entry);
|
mm, THP, swap: delay splitting THP during swap out
Patch series "THP swap: Delay splitting THP during swapping out", v11.
This patchset is to optimize the performance of Transparent Huge Page
(THP) swap.
Recently, the performance of the storage devices improved so fast that
we cannot saturate the disk bandwidth with single logical CPU when do
page swap out even on a high-end server machine. Because the
performance of the storage device improved faster than that of single
logical CPU. And it seems that the trend will not change in the near
future. On the other hand, the THP becomes more and more popular
because of increased memory size. So it becomes necessary to optimize
THP swap performance.
The advantages of the THP swap support include:
- Batch the swap operations for the THP to reduce lock
acquiring/releasing, including allocating/freeing the swap space,
adding/deleting to/from the swap cache, and writing/reading the swap
space, etc. This will help improve the performance of the THP swap.
- The THP swap space read/write will be 2M sequential IO. It is
particularly helpful for the swap read, which are usually 4k random
IO. This will improve the performance of the THP swap too.
- It will help the memory fragmentation, especially when the THP is
heavily used by the applications. The 2M continuous pages will be
free up after THP swapping out.
- It will improve the THP utilization on the system with the swap
turned on. Because the speed for khugepaged to collapse the normal
pages into the THP is quite slow. After the THP is split during the
swapping out, it will take quite long time for the normal pages to
collapse back into the THP after being swapped in. The high THP
utilization helps the efficiency of the page based memory management
too.
There are some concerns regarding THP swap in, mainly because possible
enlarged read/write IO size (for swap in/out) may put more overhead on
the storage device. To deal with that, the THP swap in should be turned
on only when necessary. For example, it can be selected via
"always/never/madvise" logic, to be turned on globally, turned off
globally, or turned on only for VMA with MADV_HUGEPAGE, etc.
This patchset is the first step for the THP swap support. The plan is
to delay splitting THP step by step, finally avoid splitting THP during
the THP swapping out and swap out/in the THP as a whole.
As the first step, in this patchset, the splitting huge page is delayed
from almost the first step of swapping out to after allocating the swap
space for the THP and adding the THP into the swap cache. This will
reduce lock acquiring/releasing for the locks used for the swap cache
management.
With the patchset, the swap out throughput improves 15.5% (from about
3.73GB/s to about 4.31GB/s) in the vm-scalability swap-w-seq test case
with 8 processes. The test is done on a Xeon E5 v3 system. The swap
device used is a RAM simulated PMEM (persistent memory) device. To test
the sequential swapping out, the test case creates 8 processes, which
sequentially allocate and write to the anonymous pages until the RAM and
part of the swap device is used up.
This patch (of 5):
In this patch, splitting huge page is delayed from almost the first step
of swapping out to after allocating the swap space for the THP
(Transparent Huge Page) and adding the THP into the swap cache. This
will batch the corresponding operation, thus improve THP swap out
throughput.
This is the first step for the THP swap optimization. The plan is to
delay splitting the THP step by step and avoid splitting the THP
finally.
In this patch, one swap cluster is used to hold the contents of each THP
swapped out. So, the size of the swap cluster is changed to that of the
THP (Transparent Huge Page) on x86_64 architecture (512). For other
architectures which want such THP swap optimization,
ARCH_USES_THP_SWAP_CLUSTER needs to be selected in the Kconfig file for
the architecture. In effect, this will enlarge swap cluster size by 2
times on x86_64. Which may make it harder to find a free cluster when
the swap space becomes fragmented. So that, this may reduce the
continuous swap space allocation and sequential write in theory. The
performance test in 0day shows no regressions caused by this.
In the future of THP swap optimization, some information of the swapped
out THP (such as compound map count) will be recorded in the
swap_cluster_info data structure.
The mem cgroup swap accounting functions are enhanced to support charge
or uncharge a swap cluster backing a THP as a whole.
The swap cluster allocate/free functions are added to allocate/free a
swap cluster for a THP. A fair simple algorithm is used for swap
cluster allocation, that is, only the first swap device in priority list
will be tried to allocate the swap cluster. The function will fail if
the trying is not successful, and the caller will fallback to allocate a
single swap slot instead. This works good enough for normal cases. If
the difference of the number of the free swap clusters among multiple
swap devices is significant, it is possible that some THPs are split
earlier than necessary. For example, this could be caused by big size
difference among multiple swap devices.
The swap cache functions is enhanced to support add/delete THP to/from
the swap cache as a set of (HPAGE_PMD_NR) sub-pages. This may be
enhanced in the future with multi-order radix tree. But because we will
split the THP soon during swapping out, that optimization doesn't make
much sense for this first step.
The THP splitting functions are enhanced to support to split THP in swap
cache during swapping out. The page lock will be held during allocating
the swap cluster, adding the THP into the swap cache and splitting the
THP. So in the code path other than swapping out, if the THP need to be
split, the PageSwapCache(THP) will be always false.
The swap cluster is only available for SSD, so the THP swap optimization
in this patchset has no effect for HDD.
[ying.huang@intel.com: fix two issues in THP optimize patch]
Link: http://lkml.kernel.org/r/87k25ed8zo.fsf@yhuang-dev.intel.com
[hannes@cmpxchg.org: extensive cleanups and simplifications, reduce code size]
Link: http://lkml.kernel.org/r/20170515112522.32457-2-ying.huang@intel.com
Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Suggested-by: Andrew Morton <akpm@linux-foundation.org> [for config option]
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> [for changes in huge_memory.c and huge_mm.h]
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Ebru Akagunduz <ebru.akagunduz@gmail.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Shaohua Li <shli@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-07-07 01:37:18 +03:00
|
|
|
extern void mem_cgroup_uncharge_swap(swp_entry_t entry, unsigned int nr_pages);
|
2016-01-21 02:03:07 +03:00
|
|
|
extern long mem_cgroup_get_nr_swap_pages(struct mem_cgroup *memcg);
|
2016-01-21 02:03:10 +03:00
|
|
|
extern bool mem_cgroup_swap_full(struct page *page);
|
2016-01-21 02:03:05 +03:00
|
|
|
#else
|
|
|
|
|
static inline void mem_cgroup_swapout(struct page *page, swp_entry_t entry)
|
|
|
|
|
{
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static inline int mem_cgroup_try_charge_swap(struct page *page,
|
|
|
|
|
swp_entry_t entry)
|
|
|
|
|
{
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
mm, THP, swap: delay splitting THP during swap out
Patch series "THP swap: Delay splitting THP during swapping out", v11.
This patchset is to optimize the performance of Transparent Huge Page
(THP) swap.
Recently, the performance of the storage devices improved so fast that
we cannot saturate the disk bandwidth with single logical CPU when do
page swap out even on a high-end server machine. Because the
performance of the storage device improved faster than that of single
logical CPU. And it seems that the trend will not change in the near
future. On the other hand, the THP becomes more and more popular
because of increased memory size. So it becomes necessary to optimize
THP swap performance.
The advantages of the THP swap support include:
- Batch the swap operations for the THP to reduce lock
acquiring/releasing, including allocating/freeing the swap space,
adding/deleting to/from the swap cache, and writing/reading the swap
space, etc. This will help improve the performance of the THP swap.
- The THP swap space read/write will be 2M sequential IO. It is
particularly helpful for the swap read, which are usually 4k random
IO. This will improve the performance of the THP swap too.
- It will help the memory fragmentation, especially when the THP is
heavily used by the applications. The 2M continuous pages will be
free up after THP swapping out.
- It will improve the THP utilization on the system with the swap
turned on. Because the speed for khugepaged to collapse the normal
pages into the THP is quite slow. After the THP is split during the
swapping out, it will take quite long time for the normal pages to
collapse back into the THP after being swapped in. The high THP
utilization helps the efficiency of the page based memory management
too.
There are some concerns regarding THP swap in, mainly because possible
enlarged read/write IO size (for swap in/out) may put more overhead on
the storage device. To deal with that, the THP swap in should be turned
on only when necessary. For example, it can be selected via
"always/never/madvise" logic, to be turned on globally, turned off
globally, or turned on only for VMA with MADV_HUGEPAGE, etc.
This patchset is the first step for the THP swap support. The plan is
to delay splitting THP step by step, finally avoid splitting THP during
the THP swapping out and swap out/in the THP as a whole.
As the first step, in this patchset, the splitting huge page is delayed
from almost the first step of swapping out to after allocating the swap
space for the THP and adding the THP into the swap cache. This will
reduce lock acquiring/releasing for the locks used for the swap cache
management.
With the patchset, the swap out throughput improves 15.5% (from about
3.73GB/s to about 4.31GB/s) in the vm-scalability swap-w-seq test case
with 8 processes. The test is done on a Xeon E5 v3 system. The swap
device used is a RAM simulated PMEM (persistent memory) device. To test
the sequential swapping out, the test case creates 8 processes, which
sequentially allocate and write to the anonymous pages until the RAM and
part of the swap device is used up.
This patch (of 5):
In this patch, splitting huge page is delayed from almost the first step
of swapping out to after allocating the swap space for the THP
(Transparent Huge Page) and adding the THP into the swap cache. This
will batch the corresponding operation, thus improve THP swap out
throughput.
This is the first step for the THP swap optimization. The plan is to
delay splitting the THP step by step and avoid splitting the THP
finally.
In this patch, one swap cluster is used to hold the contents of each THP
swapped out. So, the size of the swap cluster is changed to that of the
THP (Transparent Huge Page) on x86_64 architecture (512). For other
architectures which want such THP swap optimization,
ARCH_USES_THP_SWAP_CLUSTER needs to be selected in the Kconfig file for
the architecture. In effect, this will enlarge swap cluster size by 2
times on x86_64. Which may make it harder to find a free cluster when
the swap space becomes fragmented. So that, this may reduce the
continuous swap space allocation and sequential write in theory. The
performance test in 0day shows no regressions caused by this.
In the future of THP swap optimization, some information of the swapped
out THP (such as compound map count) will be recorded in the
swap_cluster_info data structure.
The mem cgroup swap accounting functions are enhanced to support charge
or uncharge a swap cluster backing a THP as a whole.
The swap cluster allocate/free functions are added to allocate/free a
swap cluster for a THP. A fair simple algorithm is used for swap
cluster allocation, that is, only the first swap device in priority list
will be tried to allocate the swap cluster. The function will fail if
the trying is not successful, and the caller will fallback to allocate a
single swap slot instead. This works good enough for normal cases. If
the difference of the number of the free swap clusters among multiple
swap devices is significant, it is possible that some THPs are split
earlier than necessary. For example, this could be caused by big size
difference among multiple swap devices.
The swap cache functions is enhanced to support add/delete THP to/from
the swap cache as a set of (HPAGE_PMD_NR) sub-pages. This may be
enhanced in the future with multi-order radix tree. But because we will
split the THP soon during swapping out, that optimization doesn't make
much sense for this first step.
The THP splitting functions are enhanced to support to split THP in swap
cache during swapping out. The page lock will be held during allocating
the swap cluster, adding the THP into the swap cache and splitting the
THP. So in the code path other than swapping out, if the THP need to be
split, the PageSwapCache(THP) will be always false.
The swap cluster is only available for SSD, so the THP swap optimization
in this patchset has no effect for HDD.
[ying.huang@intel.com: fix two issues in THP optimize patch]
Link: http://lkml.kernel.org/r/87k25ed8zo.fsf@yhuang-dev.intel.com
[hannes@cmpxchg.org: extensive cleanups and simplifications, reduce code size]
Link: http://lkml.kernel.org/r/20170515112522.32457-2-ying.huang@intel.com
Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Suggested-by: Andrew Morton <akpm@linux-foundation.org> [for config option]
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> [for changes in huge_memory.c and huge_mm.h]
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Ebru Akagunduz <ebru.akagunduz@gmail.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Shaohua Li <shli@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-07-07 01:37:18 +03:00
|
|
|
static inline void mem_cgroup_uncharge_swap(swp_entry_t entry,
|
|
|
|
|
unsigned int nr_pages)
|
2016-01-21 02:03:05 +03:00
|
|
|
{
|
|
|
|
|
}
|
2016-01-21 02:03:07 +03:00
|
|
|
|
|
|
|
|
static inline long mem_cgroup_get_nr_swap_pages(struct mem_cgroup *memcg)
|
|
|
|
|
{
|
|
|
|
|
return get_nr_swap_pages();
|
|
|
|
|
}
|
2016-01-21 02:03:10 +03:00
|
|
|
|
|
|
|
|
static inline bool mem_cgroup_swap_full(struct page *page)
|
|
|
|
|
{
|
|
|
|
|
return vm_swap_full();
|
|
|
|
|
}
|
2016-01-21 02:03:05 +03:00
|
|
|
#endif
|
|
|
|
|
|
2005-04-17 02:20:36 +04:00
|
|
|
#endif /* __KERNEL__*/
|
|
|
|
|
#endif /* _LINUX_SWAP_H */
|
