On Wed, 22 Jul 2015 06:26:42 +0000
"Koehrer Mathias (ETAS/ESW5)" <mathias.koehrer@xxxxxxxx> wrote:
> Hi all,
> > > >similar situation here with 3.18.11-rt7. The problem occurred once so far.
> > >
> > > Is this _with_ or _without_ the patch in question?
> >
> > With the patch. slub_delay_ctor_on_rt.patch is applied. It is 3.18.11-rt7 with
> > additional drivers
> > for our hardware.
>
> Since I have removed the slub_delay_ctor_on_rt.patch from 3.18.13-rt10 I have never seen this issue anymore on our systems.
>From my files, that is this patch:
----
From: Sebastian Andrzej Siewior <bigeasy@xxxxxxxxxxxxx>
Subject: slub: delay ctor until the object is requested
It seems that allocation of plenty objects causes latency on ARM since that code can not be preempted
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@xxxxxxxxxxxxx>
---
mm/slub.c | 6 ++++++
1 file changed, 6 insertions(+)
--- a/mm/slub.c
+++ b/mm/slub.c
@@ -1376,8 +1376,10 @@ static void setup_object(struct kmem_cac
void *object)
{
setup_object_debug(s, page, object);
+#ifndef CONFIG_PREEMPT_RT_FULL
if (unlikely(s->ctor))
s->ctor(object);
+#endif
}
static struct page *new_slab(struct kmem_cache *s, gfp_t flags, int node)
@@ -2499,6 +2501,10 @@ static __always_inline void *slab_alloc_
if (unlikely(gfpflags & __GFP_ZERO) && object)
memset(object, 0, s->object_size);
+#ifdef CONFIG_PREEMPT_RT_FULL
+ if (unlikely(s->ctor) && object)
+ s->ctor(object);
+#endif
slab_post_alloc_hook(s, gfpflags, object);
----
>
> Now I had a look at the latest 3.18.17-rt14: Here I do not see that patch anymore.
It's still there. Note, that when I pull -rt into stable, I do a
git quiltimport, which renames the patch set. It's now called:
0089-slub-delay-ctor-until-the-object-is-requested.patch
-- Steve
> Has it been removed from the patchset?
> Is there still a need to ignore one of the patches?
> Neither in the announcement to 3.18.16-rt13 nor to 3.18.17-rt14 I read any comment on that.
>
> Regards
>
> Mathias
--
To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
