All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH] slub: do not merge cache if slub_debug contains a never-merge flag
@ 2016-12-22 23:59 Grygorii Maistrenko
  2016-12-23 18:30 ` Christoph Lameter
  0 siblings, 1 reply; 9+ messages in thread
From: Grygorii Maistrenko @ 2016-12-22 23:59 UTC (permalink / raw)
  To: linux-mm
  Cc: Christoph Lameter, Pekka Enberg, David Rientjes, Joonsoo Kim,
	Andrew Morton

In case CONFIG_SLUB_DEBUG_ON=n find_mergeable() gets debug features
from commandline but never checks if there are features from the
SLAB_NEVER_MERGE set.
As a result selected by slub_debug caches are always mergeable if they
have been created without a custom constructor set or without one of the
SLAB_* debug features on.

This adds the necessary check and makes selected slab caches unmergeable
if one of the SLAB_NEVER_MERGE features is set from commandline.

Signed-off-by: Grygorii Maistrenko <grygoriimkd@gmail.com>
---
 mm/slab_common.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/mm/slab_common.c b/mm/slab_common.c
index 329b03843863..7341cba8c58b 100644
--- a/mm/slab_common.c
+++ b/mm/slab_common.c
@@ -266,6 +266,9 @@ struct kmem_cache *find_mergeable(size_t size, size_t align,
 	size = ALIGN(size, align);
 	flags = kmem_cache_flags(size, flags, name, NULL);
 
+	if (flags & SLAB_NEVER_MERGE)
+		return NULL;
+
 	list_for_each_entry_reverse(s, &slab_caches, list) {
 		if (slab_unmergeable(s))
 			continue;
-- 
2.11.0

--
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>

^ permalink raw reply related	[flat|nested] 9+ messages in thread

* Re: [PATCH] slub: do not merge cache if slub_debug contains a never-merge flag
  2016-12-22 23:59 [PATCH] slub: do not merge cache if slub_debug contains a never-merge flag Grygorii Maistrenko
@ 2016-12-23 18:30 ` Christoph Lameter
  2016-12-23 19:00   ` Grygorii Maistrenko
  0 siblings, 1 reply; 9+ messages in thread
From: Christoph Lameter @ 2016-12-23 18:30 UTC (permalink / raw)
  To: Grygorii Maistrenko
  Cc: linux-mm, Pekka Enberg, David Rientjes, Joonsoo Kim, Andrew Morton


On Fri, 23 Dec 2016, Grygorii Maistrenko wrote:

> In case CONFIG_SLUB_DEBUG_ON=n find_mergeable() gets debug features
> from commandline but never checks if there are features from the
> SLAB_NEVER_MERGE set.
> As a result selected by slub_debug caches are always mergeable if they
> have been created without a custom constructor set or without one of the
> SLAB_* debug features on.

WTF is this nonsense? That check is done a few lines earlier!

struct kmem_cache *ind_mergeable(size_t size, size_t align,
                unsigned long flags, const char *name, void (*ctor)(void *))
{
        struct kmem_cache *s;

        if (slab_nomerge || (flags & SLAB_NEVER_MERGE))    <----- !!!!!!
                return NULL;

        if (ctor)
                return NULL;

        size = ALIGN(size, sizeof(void *));
        align = calculate_alignment(flags,


>
> This adds the necessary check and makes selected slab caches unmergeable
> if one of the SLAB_NEVER_MERGE features is set from commandline.
>
> Signed-off-by: Grygorii Maistrenko <grygoriimkd@gmail.com>
> ---
>  mm/slab_common.c | 3 +++
>  1 file changed, 3 insertions(+)
>
> diff --git a/mm/slab_common.c b/mm/slab_common.c
> index 329b03843863..7341cba8c58b 100644
> --- a/mm/slab_common.c
> +++ b/mm/slab_common.c
> @@ -266,6 +266,9 @@ struct kmem_cache *find_mergeable(size_t size, size_t align,
>  	size = ALIGN(size, align);
>  	flags = kmem_cache_flags(size, flags, name, NULL);
>
> +	if (flags & SLAB_NEVER_MERGE)
> +		return NULL;
> +
>  	list_for_each_entry_reverse(s, &slab_caches, list) {
>  		if (slab_unmergeable(s))
>  			continue;
>

--
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>

^ permalink raw reply	[flat|nested] 9+ messages in thread

* Re: [PATCH] slub: do not merge cache if slub_debug contains a never-merge flag
  2016-12-23 18:30 ` Christoph Lameter
@ 2016-12-23 19:00   ` Grygorii Maistrenko
  2016-12-24 23:09     ` Christoph Lameter
  0 siblings, 1 reply; 9+ messages in thread
From: Grygorii Maistrenko @ 2016-12-23 19:00 UTC (permalink / raw)
  To: Christoph Lameter
  Cc: linux-mm, Pekka Enberg, David Rientjes, Joonsoo Kim, Andrew Morton

On Fri, Dec 23, 2016 at 12:30:02PM -0600, Christoph Lameter wrote:
> 
> On Fri, 23 Dec 2016, Grygorii Maistrenko wrote:
> 
> > In case CONFIG_SLUB_DEBUG_ON=n find_mergeable() gets debug features
> > from commandline but never checks if there are features from the
> > SLAB_NEVER_MERGE set.
> > As a result selected by slub_debug caches are always mergeable if they
> > have been created without a custom constructor set or without one of the
> > SLAB_* debug features on.
> 
> WTF is this nonsense? That check is done a few lines earlier!
> 
> struct kmem_cache *ind_mergeable(size_t size, size_t align,
>                 unsigned long flags, const char *name, void (*ctor)(void *))
> {
>         struct kmem_cache *s;
> 
>         if (slab_nomerge || (flags & SLAB_NEVER_MERGE))    <----- !!!!!!
>                 return NULL;

This one check is done on flags passed to kmem_cache_create().

> 
>         if (ctor)
>                 return NULL;
> 
>         size = ALIGN(size, sizeof(void *));
>         align = calculate_alignment(flags,
	flags = kmem_cache_flags(size, flags, name, NULL);

I added here the missing line. This updates flags from commandline and
after this we do not check it.

> 
> 
> >
> > This adds the necessary check and makes selected slab caches unmergeable
> > if one of the SLAB_NEVER_MERGE features is set from commandline.
> >
> > Signed-off-by: Grygorii Maistrenko <grygoriimkd@gmail.com>
> > ---
> >  mm/slab_common.c | 3 +++
> >  1 file changed, 3 insertions(+)
> >
> > diff --git a/mm/slab_common.c b/mm/slab_common.c
> > index 329b03843863..7341cba8c58b 100644
> > --- a/mm/slab_common.c
> > +++ b/mm/slab_common.c
> > @@ -266,6 +266,9 @@ struct kmem_cache *find_mergeable(size_t size, size_t align,
> >  	size = ALIGN(size, align);
> >  	flags = kmem_cache_flags(size, flags, name, NULL);
> >
> > +	if (flags & SLAB_NEVER_MERGE)
> > +		return NULL;
> > +
> >  	list_for_each_entry_reverse(s, &slab_caches, list) {
> >  		if (slab_unmergeable(s))
> >  			continue;
> >

--
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>

^ permalink raw reply	[flat|nested] 9+ messages in thread

* Re: [PATCH] slub: do not merge cache if slub_debug contains a never-merge flag
  2016-12-23 19:00   ` Grygorii Maistrenko
@ 2016-12-24 23:09     ` Christoph Lameter
  2016-12-26 19:08       ` [PATCH v2] " Grygorii Maistrenko
  2017-01-01 12:44       ` Grygorii Maistrenko
  0 siblings, 2 replies; 9+ messages in thread
From: Christoph Lameter @ 2016-12-24 23:09 UTC (permalink / raw)
  To: Grygorii Maistrenko
  Cc: linux-mm, Pekka Enberg, David Rientjes, Joonsoo Kim, Andrew Morton

On Fri, 23 Dec 2016, Grygorii Maistrenko wrote:

> > struct kmem_cache *ind_mergeable(size_t size, size_t align,
> >                 unsigned long flags, const char *name, void (*ctor)(void *))
> > {
> >         struct kmem_cache *s;
> >
> >         if (slab_nomerge || (flags & SLAB_NEVER_MERGE))    <----- !!!!!!
> >                 return NULL;
>
> This one check is done on flags passed to kmem_cache_create().
>
> >
> >         if (ctor)
> >                 return NULL;
> >
> >         size = ALIGN(size, sizeof(void *));
> >         align = calculate_alignment(flags,
> 	flags = kmem_cache_flags(size, flags, name, NULL);
>
> I added here the missing line. This updates flags from commandline and
> after this we do not check it.

Then please move the check down below the flags update.

--
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>

^ permalink raw reply	[flat|nested] 9+ messages in thread

* [PATCH v2] slub: do not merge cache if slub_debug contains a never-merge flag
  2016-12-24 23:09     ` Christoph Lameter
@ 2016-12-26 19:08       ` Grygorii Maistrenko
  2016-12-27  9:50         ` Pekka Enberg
  2016-12-28  0:17         ` David Rientjes
  2017-01-01 12:44       ` Grygorii Maistrenko
  1 sibling, 2 replies; 9+ messages in thread
From: Grygorii Maistrenko @ 2016-12-26 19:08 UTC (permalink / raw)
  To: Christoph Lameter
  Cc: linux-mm, Pekka Enberg, David Rientjes, Joonsoo Kim, Andrew Morton

In case CONFIG_SLUB_DEBUG_ON=n, find_mergeable() gets debug features
from commandline but never checks if there are features from the
SLAB_NEVER_MERGE set.
As a result selected by slub_debug caches are always mergeable if they
have been created without a custom constructor set or without one of the
SLAB_* debug features on.

This moves the SLAB_NEVER_MERGE check below the flags update from
commandline to make sure it won't merge the slab cache if one of the
debug features is on.

Signed-off-by: Grygorii Maistrenko <grygoriimkd@gmail.com>
---
 mm/slab_common.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

New in v2:
	- (flags & SLAB_NEVER_MERGE) check is moved down below the flags update
	  as suggested by Christoph Lameter

diff --git a/mm/slab_common.c b/mm/slab_common.c
index 329b03843863..a85a01439490 100644
--- a/mm/slab_common.c
+++ b/mm/slab_common.c
@@ -255,7 +255,7 @@ struct kmem_cache *find_mergeable(size_t size, size_t align,
 {
 	struct kmem_cache *s;
 
-	if (slab_nomerge || (flags & SLAB_NEVER_MERGE))
+	if (slab_nomerge)
 		return NULL;
 
 	if (ctor)
@@ -266,6 +266,9 @@ struct kmem_cache *find_mergeable(size_t size, size_t align,
 	size = ALIGN(size, align);
 	flags = kmem_cache_flags(size, flags, name, NULL);
 
+	if (flags & SLAB_NEVER_MERGE)
+		return NULL;
+
 	list_for_each_entry_reverse(s, &slab_caches, list) {
 		if (slab_unmergeable(s))
 			continue;
-- 
2.11.0

--
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>

^ permalink raw reply related	[flat|nested] 9+ messages in thread

* Re: [PATCH v2] slub: do not merge cache if slub_debug contains a never-merge flag
  2016-12-26 19:08       ` [PATCH v2] " Grygorii Maistrenko
@ 2016-12-27  9:50         ` Pekka Enberg
  2016-12-28  0:17         ` David Rientjes
  1 sibling, 0 replies; 9+ messages in thread
From: Pekka Enberg @ 2016-12-27  9:50 UTC (permalink / raw)
  To: Grygorii Maistrenko, Christoph Lameter
  Cc: linux-mm, Pekka Enberg, David Rientjes, Joonsoo Kim, Andrew Morton

On 12/26/2016 09:08 PM, Grygorii Maistrenko wrote:
> In case CONFIG_SLUB_DEBUG_ON=n, find_mergeable() gets debug features
> from commandline but never checks if there are features from the
> SLAB_NEVER_MERGE set.
> As a result selected by slub_debug caches are always mergeable if they
> have been created without a custom constructor set or without one of the
> SLAB_* debug features on.
>
> This moves the SLAB_NEVER_MERGE check below the flags update from
> commandline to make sure it won't merge the slab cache if one of the
> debug features is on.
>
> Signed-off-by: Grygorii Maistrenko <grygoriimkd@gmail.com>

Reviewed-by: Pekka Enberg <penberg@kernel.org>

> ---
>   mm/slab_common.c | 5 ++++-
>   1 file changed, 4 insertions(+), 1 deletion(-)
>
> New in v2:
> 	- (flags & SLAB_NEVER_MERGE) check is moved down below the flags update
> 	  as suggested by Christoph Lameter
>
> diff --git a/mm/slab_common.c b/mm/slab_common.c
> index 329b03843863..a85a01439490 100644
> --- a/mm/slab_common.c
> +++ b/mm/slab_common.c
> @@ -255,7 +255,7 @@ struct kmem_cache *find_mergeable(size_t size, size_t align,
>   {
>   	struct kmem_cache *s;
>   
> -	if (slab_nomerge || (flags & SLAB_NEVER_MERGE))
> +	if (slab_nomerge)
>   		return NULL;
>   
>   	if (ctor)
> @@ -266,6 +266,9 @@ struct kmem_cache *find_mergeable(size_t size, size_t align,
>   	size = ALIGN(size, align);
>   	flags = kmem_cache_flags(size, flags, name, NULL);
>   
> +	if (flags & SLAB_NEVER_MERGE)
> +		return NULL;
> +
>   	list_for_each_entry_reverse(s, &slab_caches, list) {
>   		if (slab_unmergeable(s))
>   			continue;

--
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>

^ permalink raw reply	[flat|nested] 9+ messages in thread

* Re: [PATCH v2] slub: do not merge cache if slub_debug contains a never-merge flag
  2016-12-26 19:08       ` [PATCH v2] " Grygorii Maistrenko
  2016-12-27  9:50         ` Pekka Enberg
@ 2016-12-28  0:17         ` David Rientjes
  1 sibling, 0 replies; 9+ messages in thread
From: David Rientjes @ 2016-12-28  0:17 UTC (permalink / raw)
  To: Grygorii Maistrenko
  Cc: Christoph Lameter, linux-mm, Pekka Enberg, Joonsoo Kim, Andrew Morton

On Mon, 26 Dec 2016, Grygorii Maistrenko wrote:

> In case CONFIG_SLUB_DEBUG_ON=n, find_mergeable() gets debug features
> from commandline but never checks if there are features from the
> SLAB_NEVER_MERGE set.
> As a result selected by slub_debug caches are always mergeable if they
> have been created without a custom constructor set or without one of the
> SLAB_* debug features on.
> 
> This moves the SLAB_NEVER_MERGE check below the flags update from
> commandline to make sure it won't merge the slab cache if one of the
> debug features is on.
> 
> Signed-off-by: Grygorii Maistrenko <grygoriimkd@gmail.com>

Acked-by: David Rientjes <rientjes@google.com>

--
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>

^ permalink raw reply	[flat|nested] 9+ messages in thread

* [PATCH v2] slub: do not merge cache if slub_debug contains a never-merge flag
  2016-12-24 23:09     ` Christoph Lameter
  2016-12-26 19:08       ` [PATCH v2] " Grygorii Maistrenko
@ 2017-01-01 12:44       ` Grygorii Maistrenko
  2017-01-04 15:55         ` Christoph Lameter
  1 sibling, 1 reply; 9+ messages in thread
From: Grygorii Maistrenko @ 2017-01-01 12:44 UTC (permalink / raw)
  To: Christoph Lameter
  Cc: linux-mm, Pekka Enberg, David Rientjes, Joonsoo Kim, Andrew Morton

In case CONFIG_SLUB_DEBUG_ON=n, find_mergeable() gets debug features
from commandline but never checks if there are features from the
SLAB_NEVER_MERGE set.
As a result selected by slub_debug caches are always mergeable if they
have been created without a custom constructor set or without one of the
SLAB_* debug features on.

This moves the SLAB_NEVER_MERGE check below the flags update from
commandline to make sure it won't merge the slab cache if one of the
debug features is on.

Signed-off-by: Grygorii Maistrenko <grygoriimkd@gmail.com>
Reviewed-by: Pekka Enberg <penberg@kernel.org>
Acked-by: David Rientjes <rientjes@google.com>
---
 mm/slab_common.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

New in v2:
	- (flags & SLAB_NEVER_MERGE) check is moved down below the flags update
	  as suggested by Christoph Lameter

diff --git a/mm/slab_common.c b/mm/slab_common.c
index 329b03843863..a85a01439490 100644
--- a/mm/slab_common.c
+++ b/mm/slab_common.c
@@ -255,7 +255,7 @@ struct kmem_cache *find_mergeable(size_t size, size_t align,
 {
 	struct kmem_cache *s;
 
-	if (slab_nomerge || (flags & SLAB_NEVER_MERGE))
+	if (slab_nomerge)
 		return NULL;
 
 	if (ctor)
@@ -266,6 +266,9 @@ struct kmem_cache *find_mergeable(size_t size, size_t align,
 	size = ALIGN(size, align);
 	flags = kmem_cache_flags(size, flags, name, NULL);
 
+	if (flags & SLAB_NEVER_MERGE)
+		return NULL;
+
 	list_for_each_entry_reverse(s, &slab_caches, list) {
 		if (slab_unmergeable(s))
 			continue;
-- 
2.11.0

--
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>

^ permalink raw reply related	[flat|nested] 9+ messages in thread

* Re: [PATCH v2] slub: do not merge cache if slub_debug contains a never-merge flag
  2017-01-01 12:44       ` Grygorii Maistrenko
@ 2017-01-04 15:55         ` Christoph Lameter
  0 siblings, 0 replies; 9+ messages in thread
From: Christoph Lameter @ 2017-01-04 15:55 UTC (permalink / raw)
  To: Grygorii Maistrenko
  Cc: linux-mm, Pekka Enberg, David Rientjes, Joonsoo Kim, Andrew Morton


Acked-by: Christoph Lameter <cl@linux.com

--
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>

^ permalink raw reply	[flat|nested] 9+ messages in thread

end of thread, other threads:[~2017-01-04 15:55 UTC | newest]

Thread overview: 9+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2016-12-22 23:59 [PATCH] slub: do not merge cache if slub_debug contains a never-merge flag Grygorii Maistrenko
2016-12-23 18:30 ` Christoph Lameter
2016-12-23 19:00   ` Grygorii Maistrenko
2016-12-24 23:09     ` Christoph Lameter
2016-12-26 19:08       ` [PATCH v2] " Grygorii Maistrenko
2016-12-27  9:50         ` Pekka Enberg
2016-12-28  0:17         ` David Rientjes
2017-01-01 12:44       ` Grygorii Maistrenko
2017-01-04 15:55         ` Christoph Lameter

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.