On 08/28/14 at 09:47am, Christoph Lameter wrote:
> On Thu, 28 Aug 2014, WANG Chao wrote:
>
> > What about failslab_store()? SLAB_FAILSLAB is also a nomerge flag.
>
>
> Subject: slub: Disable tracing and failslab for merged slabs
>
> Tracing of mergeable slabs as well as uses of failslab are
> c
On Thu, 28 Aug 2014, WANG Chao wrote:
> What about failslab_store()? SLAB_FAILSLAB is also a nomerge flag.
Subject: slub: Disable tracing and failslab for merged slabs
Tracing of mergeable slabs as well as uses of failslab are
confusing since the objects of multiple slab caches will be
affected
On 08/27/14 at 10:32am, Christoph Lameter wrote:
> Maybe something like this may be a proper fix:
>
> Subject: slub: Disable tracing of mergeable slabs
>
> Tracing of mergeable slabs is confusing since the objects
> of multiple slab caches will be traced. Moreover this creates
> a situation where
On 08/27/14 at 10:25am, Christoph Lameter wrote:
> On Wed, 27 Aug 2014, WANG Chao wrote:
>
> > Mergeable slab can be changed to unmergeable after tuning its sysfs
> > interface, for example echo 1 > trace. But the sysfs kobject with the unique
> > name will be still there.
>
> Hmmm... Merging sho
Maybe something like this may be a proper fix:
Subject: slub: Disable tracing of mergeable slabs
Tracing of mergeable slabs is confusing since the objects
of multiple slab caches will be traced. Moreover this creates
a situation where a mergeable slab will become unmergeable.
If tracing is desir
On Wed, 27 Aug 2014, WANG Chao wrote:
> Mergeable slab can be changed to unmergeable after tuning its sysfs
> interface, for example echo 1 > trace. But the sysfs kobject with the unique
> name will be still there.
Hmmm... Merging should be switched off if any debugging features are
enabled. Mayb
6 matches
Mail list logo
