Implement only __kmem_cache_alloc_node() in slab allocators and make
kmem_cache_alloc{,node,lru} wrapper of it.
Now that kmem_cache_alloc{,node,lru} is inline function, we should
use _THIS_IP_ instead of _RET_IP_ for consistency.
Signed-off-by: Hyeonggon Yoo <[email protected]>
---
include/linux/slab.h | 52 ++++++++++++++++++++++++++++++++-----
mm/slab.c | 61 +++++---------------------------------------
mm/slob.c | 27 ++++++--------------
mm/slub.c | 35 +++++--------------------
4 files changed, 67 insertions(+), 108 deletions(-)
diff --git a/include/linux/slab.h b/include/linux/slab.h
index 143830f57a7f..1b5bdcb0fd31 100644
--- a/include/linux/slab.h
+++ b/include/linux/slab.h
@@ -429,9 +429,52 @@ void *__kmalloc(size_t size, gfp_t flags)
return __kmalloc_node(size, flags, NUMA_NO_NODE);
}
-void *kmem_cache_alloc(struct kmem_cache *s, gfp_t flags) __assume_slab_alignment __malloc;
-void *kmem_cache_alloc_lru(struct kmem_cache *s, struct list_lru *lru,
- gfp_t gfpflags) __assume_slab_alignment __malloc;
+
+void *__kmem_cache_alloc_node(struct kmem_cache *s, struct list_lru *lru,
+ gfp_t gfpflags, int node, unsigned long caller __maybe_unused)
+ __assume_slab_alignment __malloc;
+
+/**
+ * kmem_cache_alloc - Allocate an object
+ * @cachep: The cache to allocate from.
+ * @flags: See kmalloc().
+ *
+ * Allocate an object from this cache. The flags are only relevant
+ * if the cache has no available objects.
+ *
+ * Return: pointer to the new object or %NULL in case of error
+ */
+static __always_inline __malloc
+void *kmem_cache_alloc(struct kmem_cache *s, gfp_t flags)
+{
+ return __kmem_cache_alloc_node(s, NULL, flags, NUMA_NO_NODE, _THIS_IP_);
+}
+
+/**
+ * kmem_cache_alloc_node - Allocate an object on the specified node
+ * @s: The cache to allocate from.
+ * @flags: See kmalloc().
+ * @node: node number of the target node.
+ *
+ * Identical to kmem_cache_alloc but it will allocate memory on the given
+ * node, which can improve the performance for cpu bound structures.
+ *
+ * Fallback to other node is possible if __GFP_THISNODE is not set.
+ *
+ * Return: pointer to the new object or %NULL in case of error
+ */
+static __always_inline __malloc
+void *kmem_cache_alloc_node(struct kmem_cache *s, gfp_t flags, int node)
+{
+ return __kmem_cache_alloc_node(s, NULL, flags, node, _THIS_IP_);
+}
+
+static __always_inline __malloc
+void *kmem_cache_alloc_lru(struct kmem_cache *s, struct list_lru *lru, gfp_t gfpflags)
+{
+ return __kmem_cache_alloc_node(s, lru, gfpflags, NUMA_NO_NODE, _THIS_IP_);
+}
+
void kmem_cache_free(struct kmem_cache *s, void *objp);
/*
@@ -453,9 +496,6 @@ static __always_inline void kfree_bulk(size_t size, void **p)
kmem_cache_free_bulk(NULL, size, p);
}
-void *kmem_cache_alloc_node(struct kmem_cache *s, gfp_t flags, int node) __assume_slab_alignment
- __malloc;
-
#ifdef CONFIG_TRACING
extern void *kmem_cache_alloc_trace(struct kmem_cache *s, gfp_t flags, size_t size)
__assume_slab_alignment __alloc_size(3);
diff --git a/mm/slab.c b/mm/slab.c
index db7eab9e2e9f..c5ffe54c207a 100644
--- a/mm/slab.c
+++ b/mm/slab.c
@@ -3442,40 +3442,18 @@ void ___cache_free(struct kmem_cache *cachep, void *objp,
__free_one(ac, objp);
}
-static __always_inline
-void *__kmem_cache_alloc_lru(struct kmem_cache *cachep, struct list_lru *lru,
- gfp_t flags)
+void *__kmem_cache_alloc_node(struct kmem_cache *cachep, struct list_lru *lru,
+ gfp_t flags, int nodeid, unsigned long caller)
{
- void *ret = slab_alloc(cachep, lru, flags, cachep->object_size, _RET_IP_);
+ void *ret = slab_alloc_node(cachep, lru, flags, nodeid,
+ cachep->object_size, caller);
- trace_kmem_cache_alloc(_RET_IP_, ret,
- cachep->object_size, cachep->size, flags);
+ trace_kmem_cache_alloc_node(caller, ret, cachep->object_size,
+ cachep->size, flags, nodeid);
return ret;
}
-
-/**
- * kmem_cache_alloc - Allocate an object
- * @cachep: The cache to allocate from.
- * @flags: See kmalloc().
- *
- * Allocate an object from this cache. The flags are only relevant
- * if the cache has no available objects.
- *
- * Return: pointer to the new object or %NULL in case of error
- */
-void *kmem_cache_alloc(struct kmem_cache *cachep, gfp_t flags)
-{
- return __kmem_cache_alloc_lru(cachep, NULL, flags);
-}
-EXPORT_SYMBOL(kmem_cache_alloc);
-
-void *kmem_cache_alloc_lru(struct kmem_cache *cachep, struct list_lru *lru,
- gfp_t flags)
-{
- return __kmem_cache_alloc_lru(cachep, lru, flags);
-}
-EXPORT_SYMBOL(kmem_cache_alloc_lru);
+EXPORT_SYMBOL(__kmem_cache_alloc_node);
static __always_inline void
cache_alloc_debugcheck_after_bulk(struct kmem_cache *s, gfp_t flags,
@@ -3545,31 +3523,6 @@ kmem_cache_alloc_trace(struct kmem_cache *cachep, gfp_t flags, size_t size)
EXPORT_SYMBOL(kmem_cache_alloc_trace);
#endif
-/**
- * kmem_cache_alloc_node - Allocate an object on the specified node
- * @cachep: The cache to allocate from.
- * @flags: See kmalloc().
- * @nodeid: node number of the target node.
- *
- * Identical to kmem_cache_alloc but it will allocate memory on the given
- * node, which can improve the performance for cpu bound structures.
- *
- * Fallback to other node is possible if __GFP_THISNODE is not set.
- *
- * Return: pointer to the new object or %NULL in case of error
- */
-void *kmem_cache_alloc_node(struct kmem_cache *cachep, gfp_t flags, int nodeid)
-{
- void *ret = slab_alloc_node(cachep, NULL, flags, nodeid, cachep->object_size, _RET_IP_);
-
- trace_kmem_cache_alloc_node(_RET_IP_, ret,
- cachep->object_size, cachep->size,
- flags, nodeid);
-
- return ret;
-}
-EXPORT_SYMBOL(kmem_cache_alloc_node);
-
#ifdef CONFIG_TRACING
void *kmem_cache_alloc_node_trace(struct kmem_cache *cachep,
gfp_t flags,
diff --git a/mm/slob.c b/mm/slob.c
index ab67c8219e8d..6c7c30845056 100644
--- a/mm/slob.c
+++ b/mm/slob.c
@@ -586,7 +586,8 @@ int __kmem_cache_create(struct kmem_cache *c, slab_flags_t flags)
return 0;
}
-static void *slob_alloc_node(struct kmem_cache *c, gfp_t flags, int node)
+static void *slob_alloc_node(struct kmem_cache *c, gfp_t flags, int node,
+ unsigned long caller)
{
void *b;
@@ -596,12 +597,12 @@ static void *slob_alloc_node(struct kmem_cache *c, gfp_t flags, int node)
if (c->size < PAGE_SIZE) {
b = slob_alloc(c->size, flags, c->align, node, 0);
- trace_kmem_cache_alloc_node(_RET_IP_, b, c->object_size,
+ trace_kmem_cache_alloc_node(caller, b, c->object_size,
SLOB_UNITS(c->size) * SLOB_UNIT,
flags, node);
} else {
b = slob_new_pages(flags, get_order(c->size), node);
- trace_kmem_cache_alloc_node(_RET_IP_, b, c->object_size,
+ trace_kmem_cache_alloc_node(caller, b, c->object_size,
PAGE_SIZE << get_order(c->size),
flags, node);
}
@@ -615,30 +616,18 @@ static void *slob_alloc_node(struct kmem_cache *c, gfp_t flags, int node)
return b;
}
-void *kmem_cache_alloc(struct kmem_cache *cachep, gfp_t flags)
-{
- return slob_alloc_node(cachep, flags, NUMA_NO_NODE);
-}
-EXPORT_SYMBOL(kmem_cache_alloc);
-
-
-void *kmem_cache_alloc_lru(struct kmem_cache *cachep, struct list_lru *lru, gfp_t flags)
-{
- return slob_alloc_node(cachep, flags, NUMA_NO_NODE);
-}
-EXPORT_SYMBOL(kmem_cache_alloc_lru);
-
void *__kmalloc_node(size_t size, gfp_t gfp, int node)
{
return __do_kmalloc_node(size, gfp, node, _RET_IP_);
}
EXPORT_SYMBOL(__kmalloc_node);
-void *kmem_cache_alloc_node(struct kmem_cache *cachep, gfp_t gfp, int node)
+void *__kmem_cache_alloc_node(struct kmem_cache *cachep, struct list_lru *lru __maybe_unused,
+ gfp_t gfp, int node, unsigned long caller __maybe_unused)
{
- return slob_alloc_node(cachep, gfp, node);
+ return slob_alloc_node(cachep, gfp, node, caller);
}
-EXPORT_SYMBOL(kmem_cache_alloc_node);
+EXPORT_SYMBOL(__kmem_cache_alloc_node);
static void __kmem_cache_free(void *b, int size)
{
diff --git a/mm/slub.c b/mm/slub.c
index f10a892f1772..2a2be2a8a5d0 100644
--- a/mm/slub.c
+++ b/mm/slub.c
@@ -3216,30 +3216,6 @@ static __always_inline void *slab_alloc(struct kmem_cache *s, struct list_lru *l
return slab_alloc_node(s, lru, gfpflags, NUMA_NO_NODE, addr, orig_size);
}
-static __always_inline
-void *__kmem_cache_alloc_lru(struct kmem_cache *s, struct list_lru *lru,
- gfp_t gfpflags)
-{
- void *ret = slab_alloc(s, lru, gfpflags, _RET_IP_, s->object_size);
-
- trace_kmem_cache_alloc(_RET_IP_, ret, s->object_size,
- s->size, gfpflags);
-
- return ret;
-}
-
-void *kmem_cache_alloc(struct kmem_cache *s, gfp_t gfpflags)
-{
- return __kmem_cache_alloc_lru(s, NULL, gfpflags);
-}
-EXPORT_SYMBOL(kmem_cache_alloc);
-
-void *kmem_cache_alloc_lru(struct kmem_cache *s, struct list_lru *lru,
- gfp_t gfpflags)
-{
- return __kmem_cache_alloc_lru(s, lru, gfpflags);
-}
-EXPORT_SYMBOL(kmem_cache_alloc_lru);
#ifdef CONFIG_TRACING
void *kmem_cache_alloc_trace(struct kmem_cache *s, gfp_t gfpflags, size_t size)
@@ -3252,16 +3228,17 @@ void *kmem_cache_alloc_trace(struct kmem_cache *s, gfp_t gfpflags, size_t size)
EXPORT_SYMBOL(kmem_cache_alloc_trace);
#endif
-void *kmem_cache_alloc_node(struct kmem_cache *s, gfp_t gfpflags, int node)
+void *__kmem_cache_alloc_node(struct kmem_cache *s, struct list_lru *lru, gfp_t gfpflags,
+ int node, unsigned long caller __maybe_unused)
{
- void *ret = slab_alloc_node(s, NULL, gfpflags, node, _RET_IP_, s->object_size);
+ void *ret = slab_alloc_node(s, lru, gfpflags, node, caller, s->object_size);
- trace_kmem_cache_alloc_node(_RET_IP_, ret,
- s->object_size, s->size, gfpflags, node);
+ trace_kmem_cache_alloc_node(caller, ret, s->object_size,
+ s->size, gfpflags, node);
return ret;
}
-EXPORT_SYMBOL(kmem_cache_alloc_node);
+EXPORT_SYMBOL(__kmem_cache_alloc_node);
#ifdef CONFIG_TRACING
void *kmem_cache_alloc_node_trace(struct kmem_cache *s,
--
2.32.0
On 4/14/22 10:57, Hyeonggon Yoo wrote:
> Implement only __kmem_cache_alloc_node() in slab allocators and make
> kmem_cache_alloc{,node,lru} wrapper of it.
>
> Now that kmem_cache_alloc{,node,lru} is inline function, we should
> use _THIS_IP_ instead of _RET_IP_ for consistency.
Hm yeah looks like this actually fixes some damage of obscured actual
__RET_IP_ by the recent addition and wrapping of __kmem_cache_alloc_lru().
> Signed-off-by: Hyeonggon Yoo <[email protected]>
Reviewed-by: Vlastimil Babka <[email protected]>
Some nits:
> ---
> include/linux/slab.h | 52 ++++++++++++++++++++++++++++++++-----
> mm/slab.c | 61 +++++---------------------------------------
> mm/slob.c | 27 ++++++--------------
> mm/slub.c | 35 +++++--------------------
> 4 files changed, 67 insertions(+), 108 deletions(-)
>
> diff --git a/include/linux/slab.h b/include/linux/slab.h
> index 143830f57a7f..1b5bdcb0fd31 100644
> --- a/include/linux/slab.h
> +++ b/include/linux/slab.h
> @@ -429,9 +429,52 @@ void *__kmalloc(size_t size, gfp_t flags)
> return __kmalloc_node(size, flags, NUMA_NO_NODE);
> }
>
> -void *kmem_cache_alloc(struct kmem_cache *s, gfp_t flags) __assume_slab_alignment __malloc;
> -void *kmem_cache_alloc_lru(struct kmem_cache *s, struct list_lru *lru,
> - gfp_t gfpflags) __assume_slab_alignment __malloc;
> +
> +void *__kmem_cache_alloc_node(struct kmem_cache *s, struct list_lru *lru,
> + gfp_t gfpflags, int node, unsigned long caller __maybe_unused)
> + __assume_slab_alignment __malloc;
I don't think caller needs to be __maybe_unused in the declaration nor any
of the implementations of __kmem_cache_alloc_node(), all actually pass it on?
On Tue, Apr 26, 2022 at 08:01:27PM +0200, Vlastimil Babka wrote:
> On 4/14/22 10:57, Hyeonggon Yoo wrote:
> > Implement only __kmem_cache_alloc_node() in slab allocators and make
> > kmem_cache_alloc{,node,lru} wrapper of it.
> >
> > Now that kmem_cache_alloc{,node,lru} is inline function, we should
> > use _THIS_IP_ instead of _RET_IP_ for consistency.
>
> Hm yeah looks like this actually fixes some damage of obscured actual
> __RET_IP_ by the recent addition and wrapping of __kmem_cache_alloc_lru().
>
> > Signed-off-by: Hyeonggon Yoo <[email protected]>
>
> Reviewed-by: Vlastimil Babka <[email protected]>
>
> Some nits:
>
> > ---
> > include/linux/slab.h | 52 ++++++++++++++++++++++++++++++++-----
> > mm/slab.c | 61 +++++---------------------------------------
> > mm/slob.c | 27 ++++++--------------
> > mm/slub.c | 35 +++++--------------------
> > 4 files changed, 67 insertions(+), 108 deletions(-)
> >
> > diff --git a/include/linux/slab.h b/include/linux/slab.h
> > index 143830f57a7f..1b5bdcb0fd31 100644
> > --- a/include/linux/slab.h
> > +++ b/include/linux/slab.h
> > @@ -429,9 +429,52 @@ void *__kmalloc(size_t size, gfp_t flags)
> > return __kmalloc_node(size, flags, NUMA_NO_NODE);
> > }
> >
> > -void *kmem_cache_alloc(struct kmem_cache *s, gfp_t flags) __assume_slab_alignment __malloc;
> > -void *kmem_cache_alloc_lru(struct kmem_cache *s, struct list_lru *lru,
> > - gfp_t gfpflags) __assume_slab_alignment __malloc;
> > +
> > +void *__kmem_cache_alloc_node(struct kmem_cache *s, struct list_lru *lru,
> > + gfp_t gfpflags, int node, unsigned long caller __maybe_unused)
> > + __assume_slab_alignment __malloc;
>
> I don't think caller needs to be __maybe_unused in the declaration nor any
> of the implementations of __kmem_cache_alloc_node(), all actually pass it on?
My intention was to give hints to compilers when CONFIG_TRACING=n.
I'll check if the compiler just optimizes them without __maybe_unused.
Thanks!