From: Christoph Lameter <cl@linux.com>
To: Matthew Wilcox <willy@infradead.org>
Cc: linux-mm@kvack.org, Pekka Enberg <penberg@cs.helsinki.fi>,
akpm@linux-foundation.org, Mel Gorman <mel@skynet.ie>,
andi@firstfloor.org, Rik van Riel <riel@redhat.com>,
Dave Chinner <dchinner@redhat.com>,
Christoph Hellwig <hch@lst.de>
Subject: [RFC 3/8] slub: Add isolate() and migrate() methods
Date: Wed, 27 Dec 2017 16:06:39 -0600 [thread overview]
Message-ID: <20171227220652.402842142@linux.com> (raw)
In-Reply-To: 20171227220636.361857279@linux.com
[-- Attachment #1: isolate_and_migrate_methods --]
[-- Type: text/plain, Size: 6693 bytes --]
Add the two methods needed for moving objects and enable the
display of the callbacks via the /sys/kernel/slab interface.
Add documentation explaining the use of these methods and the prototypes
for slab.h. Add functions to setup the callbacks method for a slab cache.
Add empty functions for SLAB/SLOB. The API is generic so it
could be theoretically implemented for these allocators as well.
Signed-off-by: Christoph Lameter <cl@linux.com>
---
include/linux/slab.h | 50 +++++++++++++++++++++++++++++++++++++++++++++++
include/linux/slub_def.h | 3 ++
mm/slub.c | 29 ++++++++++++++++++++++++++-
3 files changed, 81 insertions(+), 1 deletion(-)
Index: linux/include/linux/slub_def.h
===================================================================
--- linux.orig/include/linux/slub_def.h
+++ linux/include/linux/slub_def.h
@@ -98,6 +98,9 @@ struct kmem_cache {
gfp_t allocflags; /* gfp flags to use on each alloc */
int refcount; /* Refcount for slab cache destroy */
void (*ctor)(void *);
+ kmem_isolate_func *isolate;
+ kmem_migrate_func *migrate;
+
int inuse; /* Offset to metadata */
int align; /* Alignment */
int reserved; /* Reserved bytes at the end of slabs */
Index: linux/mm/slub.c
===================================================================
--- linux.orig/mm/slub.c
+++ linux/mm/slub.c
@@ -3479,7 +3479,6 @@ static int calculate_sizes(struct kmem_c
else
s->flags &= ~__OBJECT_POISON;
-
/*
* If we are Redzoning then check if there is some space between the
* end of the object and the free pointer. If not then add an
@@ -4275,6 +4274,25 @@ int __kmem_cache_create(struct kmem_cach
return err;
}
+void kmem_cache_setup_mobility(struct kmem_cache *s,
+ kmem_isolate_func isolate, kmem_migrate_func migrate)
+{
+ /*
+ * Defragmentable slabs must have a ctor otherwise objects may be
+ * in an undetermined state after they are allocated.
+ */
+ BUG_ON(!s->ctor);
+ s->isolate = isolate;
+ s->migrate = migrate;
+ /*
+ * Sadly serialization requirements currently mean that we have
+ * to disable fast cmpxchg based processing.
+ */
+ s->flags &= ~__CMPXCHG_DOUBLE;
+
+}
+EXPORT_SYMBOL(kmem_cache_setup_mobility);
+
void *__kmalloc_track_caller(size_t size, gfp_t gfpflags, unsigned long caller)
{
struct kmem_cache *s;
@@ -4969,6 +4987,20 @@ static ssize_t ops_show(struct kmem_cach
if (s->ctor)
x += sprintf(buf + x, "ctor : %pS\n", s->ctor);
+
+ if (s->isolate) {
+ x += sprintf(buf + x, "isolate : ");
+ x += sprint_symbol(buf + x,
+ (unsigned long)s->isolate);
+ x += sprintf(buf + x, "\n");
+ }
+
+ if (s->migrate) {
+ x += sprintf(buf + x, "migrate : ");
+ x += sprint_symbol(buf + x,
+ (unsigned long)s->migrate);
+ x += sprintf(buf + x, "\n");
+ }
return x;
}
SLAB_ATTR_RO(ops);
Index: linux/include/linux/slab.h
===================================================================
--- linux.orig/include/linux/slab.h
+++ linux/include/linux/slab.h
@@ -146,6 +146,68 @@ void memcg_deactivate_kmem_caches(struct
void memcg_destroy_kmem_caches(struct mem_cgroup *);
/*
+ * Function prototypes passed to kmem_cache_setup_mobility() to enable mobile
+ * objects and targeted reclaim in slab caches.
+ */
+
+/*
+ * kmem_cache_isolate_func() is called with locks held so that the slab
+ * objects cannot be freed. We are in an atomic context and no slab
+ * operations may be performed. The purpose of kmem_cache_isolate_func()
+ * is to pin the object so that it cannot be freed until
+ * kmem_cache_migrate_func() has processed them. This may be accomplished
+ * by increasing the refcount or setting a flag.
+ *
+ * Parameters passed are the number of objects to process and an array of
+ * pointers to objects which are intended to be moved.
+ *
+ * Returns a pointer that is passed to the migrate function. If any objects
+ * cannot be touched at this point then the pointer may indicate a
+ * failure and then the migration function can simply remove the references
+ * that were already obtained. The private data could be used to track
+ * the objects that were already pinned.
+ *
+ * The object pointer array passed is also passed to kmem_cache_migrate().
+ * The function may remove objects from the array by setting pointers to
+ * NULL. This is useful if we can determine that an object is being freed
+ * because kmem_cache_isolate_func() was called when the subsystem
+ * was calling kmem_cache_free().
+ * In that case it is not necessary to increase the refcount or
+ * specially mark the object because the release of the slab lock
+ * will lead to the immediate freeing of the object.
+ */
+typedef void *kmem_isolate_func(struct kmem_cache *, void **, int);
+
+/*
+ * kmem_cache_move_migrate_func is called with no locks held and interrupts
+ * enabled. Sleeping is possible. Any operation may be performed in
+ * migrate(). kmem_cache_migrate_func should allocate new objects and
+ * free all the objects.
+ **
+ * Parameters passed are the number of objects in the array, the array of
+ * pointers to the objects, the NUMA node where the object should be
+ * allocated and the pointer returned by kmem_cache_isolate_func().
+ *
+ * Success is checked by examining the number of remaining objects in
+ * the slab. If the number is zero then the objects will be freed.
+ */
+typedef void kmem_migrate_func(struct kmem_cache *, void **, int nr, int node, void *private);
+
+/*
+ * kmem_cache_setup_mobility() is used to setup callbacks for a slab cache.
+ */
+#ifdef CONFIG_SLUB
+void kmem_cache_setup_mobility(struct kmem_cache *, kmem_isolate_func,
+ kmem_migrate_func);
+#else
+static inline void kmem_cache_setup_mobility(struct kmem_cache *s,
+ kmem_isolate_func isolate, kmem_migrate_func migrate) {}
+#endif
+
+/*
+ * Allocator specific definitions. These are mainly used to establish optimized
+ * ways to convert kmalloc() calls to kmem_cache_alloc() invocations by
+ * selecting the appropriate general cache at compile time.
* Please use this macro to create slab caches. Simply specify the
* name of the structure and maybe some flags that are listed above.
*
Index: linux/mm/slab_common.c
===================================================================
--- linux.orig/mm/slab_common.c
+++ linux/mm/slab_common.c
@@ -278,7 +278,7 @@ int slab_unmergeable(struct kmem_cache *
if (!is_root_cache(s))
return 1;
- if (s->ctor)
+ if (s->ctor || s->isolate || s->migrate)
return 1;
/*
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
next prev parent reply other threads:[~2017-12-27 22:09 UTC|newest]
Thread overview: 24+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-12-27 22:06 [RFC 0/8] Xarray object migration V1 Christoph Lameter
2017-12-27 22:06 ` [RFC 2/8] slub: Add defrag_ratio field and sysfs support Christoph Lameter
2017-12-30 6:20 ` Matthew Wilcox
2018-01-02 14:53 ` Christopher Lameter
2017-12-27 22:06 ` Christoph Lameter [this message]
2017-12-30 6:42 ` [RFC 3/8] slub: Add isolate() and migrate() methods Matthew Wilcox
2018-01-01 21:20 ` Matthew Wilcox
2018-01-02 14:56 ` Christopher Lameter
2018-01-02 14:55 ` Christopher Lameter
2017-12-27 22:06 ` [RFC 4/8] slub: Sort slab cache list and establish maximum objects for defrag slabs Christoph Lameter
2017-12-27 22:06 ` [RFC 5/8] slub: Slab defrag core Christoph Lameter
2017-12-27 22:06 ` [RFC 6/8] slub: Extend slabinfo to support -D and -F options Christoph Lameter
2017-12-27 22:06 ` [RFC 7/8] xarray: Implement migration function for objects Christoph Lameter
2017-12-27 22:06 ` [RFC 8/8] Add debugging output Christoph Lameter
2017-12-28 5:19 ` [RFC 0/8] Xarray object migration V1 Randy Dunlap
2017-12-28 14:57 ` Christopher Lameter
2017-12-28 17:18 ` James Bottomley
2017-12-28 17:33 ` Benjamin LaHaise
2017-12-28 17:40 ` James Bottomley
2017-12-28 19:17 ` Benjamin LaHaise
2017-12-28 20:00 ` James Bottomley
2017-12-28 20:33 ` Benjamin LaHaise
2017-12-28 22:24 ` Dave Chinner
2017-12-29 0:19 ` Christopher Lameter
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20171227220652.402842142@linux.com \
--to=cl@linux.com \
--cc=akpm@linux-foundation.org \
--cc=andi@firstfloor.org \
--cc=dchinner@redhat.com \
--cc=hch@lst.de \
--cc=linux-mm@kvack.org \
--cc=mel@skynet.ie \
--cc=penberg@cs.helsinki.fi \
--cc=riel@redhat.com \
--cc=willy@infradead.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.