* [PATCH v2] mm, slob: Drop usage of page->private for storing page-sized allocations
@ 2012-09-04 10:13 Ezequiel Garcia
2012-09-04 10:15 ` Glauber Costa
0 siblings, 1 reply; 8+ messages in thread
From: Ezequiel Garcia @ 2012-09-04 10:13 UTC (permalink / raw)
To: linux-mm; +Cc: Ezequiel Garcia, Pekka Enberg, Christoph Lameter, Glauber Costa
This field was being used to store size allocation so it could be
retrieved by ksize(). However, it is a bad practice to not mark a page
as a slab page and then use fields for special purposes.
There is no need to store the allocated size and
ksize() can simply return PAGE_SIZE << compound_order(page).
Cc: Pekka Enberg <penberg@kernel.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: Glauber Costa <glommer@parallels.com>
Signed-off-by: Ezequiel Garcia <elezegarcia@gmail.com>
---
Changes from v1:
* Rebased on top of slab/next
mm/slob.c | 24 ++++++++++--------------
1 files changed, 10 insertions(+), 14 deletions(-)
diff --git a/mm/slob.c b/mm/slob.c
index 45d4ca7..ae46edc 100644
--- a/mm/slob.c
+++ b/mm/slob.c
@@ -28,9 +28,8 @@
* from kmalloc are prepended with a 4-byte header with the kmalloc size.
* If kmalloc is asked for objects of PAGE_SIZE or larger, it calls
* alloc_pages() directly, allocating compound pages so the page order
- * does not have to be separately tracked, and also stores the exact
- * allocation size in page->private so that it can be used to accurately
- * provide ksize(). These objects are detected in kfree() because slob_page()
+ * does not have to be separately tracked.
+ * These objects are detected in kfree() because PageSlab()
* is false for them.
*
* SLAB is emulated on top of SLOB by simply calling constructors and
@@ -454,11 +453,6 @@ void *__kmalloc_node(size_t size, gfp_t gfp, int node)
if (likely(order))
gfp |= __GFP_COMP;
ret = slob_new_pages(gfp, order, node);
- if (ret) {
- struct page *page;
- page = virt_to_page(ret);
- page->private = size;
- }
trace_kmalloc_node(_RET_IP_, ret,
size, PAGE_SIZE << order, gfp, node);
@@ -493,18 +487,20 @@ EXPORT_SYMBOL(kfree);
size_t ksize(const void *block)
{
struct page *sp;
+ int align;
+ unsigned int *m;
BUG_ON(!block);
if (unlikely(block == ZERO_SIZE_PTR))
return 0;
sp = virt_to_page(block);
- if (PageSlab(sp)) {
- int align = max(ARCH_KMALLOC_MINALIGN, ARCH_SLAB_MINALIGN);
- unsigned int *m = (unsigned int *)(block - align);
- return SLOB_UNITS(*m) * SLOB_UNIT;
- } else
- return sp->private;
+ if (unlikely(!PageSlab(sp)))
+ return PAGE_SIZE << compound_order(sp);
+
+ align = max(ARCH_KMALLOC_MINALIGN, ARCH_SLAB_MINALIGN);
+ m = (unsigned int *)(block - align);
+ return SLOB_UNITS(*m) * SLOB_UNIT;
}
EXPORT_SYMBOL(ksize);
--
1.7.8.6
--
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] 8+ messages in thread
* Re: [PATCH v2] mm, slob: Drop usage of page->private for storing page-sized allocations
2012-09-04 10:13 [PATCH v2] mm, slob: Drop usage of page->private for storing page-sized allocations Ezequiel Garcia
@ 2012-09-04 10:15 ` Glauber Costa
2012-09-04 10:34 ` Ezequiel Garcia
0 siblings, 1 reply; 8+ messages in thread
From: Glauber Costa @ 2012-09-04 10:15 UTC (permalink / raw)
To: Ezequiel Garcia; +Cc: linux-mm, Pekka Enberg, Christoph Lameter
On 09/04/2012 02:13 PM, Ezequiel Garcia wrote:
> This field was being used to store size allocation so it could be
> retrieved by ksize(). However, it is a bad practice to not mark a page
> as a slab page and then use fields for special purposes.
> There is no need to store the allocated size and
> ksize() can simply return PAGE_SIZE << compound_order(page).
What happens for allocations smaller than a page?
It seems you are breaking ksize for those.
--
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] 8+ messages in thread
* Re: [PATCH v2] mm, slob: Drop usage of page->private for storing page-sized allocations
2012-09-04 10:15 ` Glauber Costa
@ 2012-09-04 10:34 ` Ezequiel Garcia
2012-09-04 11:07 ` Glauber Costa
0 siblings, 1 reply; 8+ messages in thread
From: Ezequiel Garcia @ 2012-09-04 10:34 UTC (permalink / raw)
To: Glauber Costa; +Cc: linux-mm, Pekka Enberg, Christoph Lameter
Hi Glauber,
On Tue, Sep 4, 2012 at 7:15 AM, Glauber Costa <glommer@parallels.com> wrote:
> On 09/04/2012 02:13 PM, Ezequiel Garcia wrote:
>> This field was being used to store size allocation so it could be
>> retrieved by ksize(). However, it is a bad practice to not mark a page
>> as a slab page and then use fields for special purposes.
>> There is no need to store the allocated size and
>> ksize() can simply return PAGE_SIZE << compound_order(page).
>
> What happens for allocations smaller than a page?
> It seems you are breaking ksize for those.
>
Allocations smaller than a page save its own size on a header
located at each returned pointer. This is documented at the beginning of slob:
"Above this is an implementation of kmalloc/kfree. Blocks returned
from kmalloc are prepended with a 4-byte header with the kmalloc size."
For this objects (smaller than a page) ksize works fine:
size_t ksize(const void *block) {
[...]
unsigned int *m = (unsigned int *)(block - align);
return SLOB_UNITS(*m) * SLOB_UNIT;
(see how ksize substract 'align' from block to find the header?)
Of course, it's possible I've overlooked something, but I think this
should work.
Thanks!
Ezequiel.
--
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] 8+ messages in thread
* Re: [PATCH v2] mm, slob: Drop usage of page->private for storing page-sized allocations
2012-09-04 10:34 ` Ezequiel Garcia
@ 2012-09-04 11:07 ` Glauber Costa
2012-09-04 18:00 ` Christoph Lameter
0 siblings, 1 reply; 8+ messages in thread
From: Glauber Costa @ 2012-09-04 11:07 UTC (permalink / raw)
To: Ezequiel Garcia; +Cc: linux-mm, Pekka Enberg, Christoph Lameter
On 09/04/2012 02:34 PM, Ezequiel Garcia wrote:
> Hi Glauber,
>
> On Tue, Sep 4, 2012 at 7:15 AM, Glauber Costa <glommer@parallels.com> wrote:
>> On 09/04/2012 02:13 PM, Ezequiel Garcia wrote:
>>> This field was being used to store size allocation so it could be
>>> retrieved by ksize(). However, it is a bad practice to not mark a page
>>> as a slab page and then use fields for special purposes.
>>> There is no need to store the allocated size and
>>> ksize() can simply return PAGE_SIZE << compound_order(page).
>>
>> What happens for allocations smaller than a page?
>> It seems you are breaking ksize for those.
>>
>
> Allocations smaller than a page save its own size on a header
> located at each returned pointer. This is documented at the beginning of slob:
>
> "Above this is an implementation of kmalloc/kfree. Blocks returned
> from kmalloc are prepended with a 4-byte header with the kmalloc size."
>
> For this objects (smaller than a page) ksize works fine:
>
> size_t ksize(const void *block) {
> [...]
> unsigned int *m = (unsigned int *)(block - align);
> return SLOB_UNITS(*m) * SLOB_UNIT;
>
> (see how ksize substract 'align' from block to find the header?)
>
> Of course, it's possible I've overlooked something, but I think this
> should work.
>
Ok. Slob also goes to the page allocator for higher order kmallocs,
so it should work.
--
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] 8+ messages in thread
* Re: [PATCH v2] mm, slob: Drop usage of page->private for storing page-sized allocations
2012-09-04 11:07 ` Glauber Costa
@ 2012-09-04 18:00 ` Christoph Lameter
2012-09-05 14:11 ` Ezequiel Garcia
0 siblings, 1 reply; 8+ messages in thread
From: Christoph Lameter @ 2012-09-04 18:00 UTC (permalink / raw)
To: Glauber Costa; +Cc: Ezequiel Garcia, linux-mm, Pekka Enberg
I thought I acked it before?
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] 8+ messages in thread
* Re: [PATCH v2] mm, slob: Drop usage of page->private for storing page-sized allocations
2012-09-04 18:00 ` Christoph Lameter
@ 2012-09-05 14:11 ` Ezequiel Garcia
2012-09-24 17:11 ` Ezequiel Garcia
0 siblings, 1 reply; 8+ messages in thread
From: Ezequiel Garcia @ 2012-09-05 14:11 UTC (permalink / raw)
To: Christoph Lameter; +Cc: Glauber Costa, linux-mm, Pekka Enberg
On Tue, Sep 4, 2012 at 3:00 PM, Christoph Lameter <cl@linux.com> wrote:
> I thought I acked it before?
>
> Acked-by: Christoph Lameter <cl@linux.com>
>
Yes you did. This is a v2, that I rebased on top of slab/next for Pekka.
Thanks!
Ezequiel.
--
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] 8+ messages in thread
* Re: [PATCH v2] mm, slob: Drop usage of page->private for storing page-sized allocations
2012-09-05 14:11 ` Ezequiel Garcia
@ 2012-09-24 17:11 ` Ezequiel Garcia
2012-10-09 14:51 ` Ezequiel Garcia
0 siblings, 1 reply; 8+ messages in thread
From: Ezequiel Garcia @ 2012-09-24 17:11 UTC (permalink / raw)
To: Pekka Enberg; +Cc: Glauber Costa, linux-mm, Christoph Lameter, Matt Mackall
Pekka,
On Wed, Sep 5, 2012 at 11:11 AM, Ezequiel Garcia <elezegarcia@gmail.com> wrote:
> On Tue, Sep 4, 2012 at 3:00 PM, Christoph Lameter <cl@linux.com> wrote:
>> I thought I acked it before?
>>
>> Acked-by: Christoph Lameter <cl@linux.com>
>>
>
Will you pick this for v3.7 pull request?
Or is there anything wrong with it?
--
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] 8+ messages in thread
* Re: [PATCH v2] mm, slob: Drop usage of page->private for storing page-sized allocations
2012-09-24 17:11 ` Ezequiel Garcia
@ 2012-10-09 14:51 ` Ezequiel Garcia
0 siblings, 0 replies; 8+ messages in thread
From: Ezequiel Garcia @ 2012-10-09 14:51 UTC (permalink / raw)
To: Pekka Enberg; +Cc: Glauber Costa, linux-mm, Christoph Lameter, Matt Mackall
Pekka,
On Mon, Sep 24, 2012 at 2:11 PM, Ezequiel Garcia <elezegarcia@gmail.com> wrote:
> Pekka,
>
> On Wed, Sep 5, 2012 at 11:11 AM, Ezequiel Garcia <elezegarcia@gmail.com> wrote:
>> On Tue, Sep 4, 2012 at 3:00 PM, Christoph Lameter <cl@linux.com> wrote:
>>> I thought I acked it before?
>>>
>>> Acked-by: Christoph Lameter <cl@linux.com>
>>>
>>
>
> Will you pick this for v3.7 pull request?
> Or is there anything wrong with it?
I think you've missed this for your pull request.
And I'm a bit puzzled, since it has been acked by Christoph,
or are we still waiting to matt's ack?
Thanks!
Ezequiel
--
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] 8+ messages in thread
end of thread, other threads:[~2012-10-09 14:51 UTC | newest]
Thread overview: 8+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2012-09-04 10:13 [PATCH v2] mm, slob: Drop usage of page->private for storing page-sized allocations Ezequiel Garcia
2012-09-04 10:15 ` Glauber Costa
2012-09-04 10:34 ` Ezequiel Garcia
2012-09-04 11:07 ` Glauber Costa
2012-09-04 18:00 ` Christoph Lameter
2012-09-05 14:11 ` Ezequiel Garcia
2012-09-24 17:11 ` Ezequiel Garcia
2012-10-09 14:51 ` Ezequiel Garcia
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.