Re: [PATCH v2 1/2] mm, slab: move memcg charging to post-alloc hook

[Roman Gushchin <roman.gushchin@linux.dev>]

On Wed, Apr 03, 2024 at 05:48:24PM +0200, Vlastimil Babka wrote:
> On 4/3/24 1:39 PM, Aishwarya TCV wrote:
> > 
> > 
> > On 25/03/2024 08:20, Vlastimil Babka wrote:
> >> The MEMCG_KMEM integration with slab currently relies on two hooks
> >> during allocation. memcg_slab_pre_alloc_hook() determines the objcg and
> >> charges it, and memcg_slab_post_alloc_hook() assigns the objcg pointer
> >> to the allocated object(s).
> >> 
> >> As Linus pointed out, this is unnecessarily complex. Failing to charge
> >> due to memcg limits should be rare, so we can optimistically allocate
> >> the object(s) and do the charging together with assigning the objcg
> >> pointer in a single post_alloc hook. In the rare case the charging
> >> fails, we can free the object(s) back.
> >> 
> >> This simplifies the code (no need to pass around the objcg pointer) and
> >> potentially allows to separate charging from allocation in cases where
> >> it's common that the allocation would be immediately freed, and the
> >> memcg handling overhead could be saved.
> >> 
> >> Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
> >> Link: https://lore.kernel.org/all/CAHk-=whYOOdM7jWy5jdrAm8LxcgCMFyk2bt8fYYvZzM4U-zAQA@mail.gmail.com/
> >> Reviewed-by: Roman Gushchin <roman.gushchin@linux.dev>
> >> Reviewed-by: Chengming Zhou <chengming.zhou@linux.dev>
> >> Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
> >> ---
> >>  mm/slub.c | 180 +++++++++++++++++++++++++++-----------------------------------
> >>  1 file changed, 77 insertions(+), 103 deletions(-)
> > 
> > Hi Vlastimil,
> > 
> > When running the LTP test "memcg_limit_in_bytes" against next-master
> > (next-20240402) kernel with Arm64 on JUNO, oops is observed in our CI. I
> > can send the full logs if required. It is observed to work fine on
> > softiron-overdrive-3000.
> > 
> > A bisect identified 11bb2d9d91627935c63ea3e6a031fd238c846e1 as the first
> > bad commit. Bisected it on the tag "next-20240402" at repo
> > "https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git".
> > 
> > This works fine on  Linux version v6.9-rc2
> 
> Oops, sorry, can you verify that this fixes it?
> Thanks.
> 
> ----8<----
> From b0597c220624fef4f10e26079a3ff1c86f02a12b Mon Sep 17 00:00:00 2001
> From: Vlastimil Babka <vbabka@suse.cz>
> Date: Wed, 3 Apr 2024 17:45:15 +0200
> Subject: [PATCH] fixup! mm, slab: move memcg charging to post-alloc hook
> 
> The call to memcg_alloc_abort_single() is wrong, it expects a pointer to
> single object, not an array.
> 
> Reported-by: Aishwarya TCV <aishwarya.tcv@arm.com>
> Signed-off-by: Vlastimil Babka <vbabka@suse.cz>

Oh, indeed.
Reviewed-by: Roman Gushchin <roman.gushchin@linux.dev>

Vlastimil, here is another small comments fixup for the same original patch:

diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 0745a28782de..9bd0ffd4c547 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -353,7 +353,7 @@ static void memcg_reparent_objcgs(struct mem_cgroup *memcg,

 /*
  * A lot of the calls to the cache allocation functions are expected to be
- * inlined by the compiler. Since the calls to memcg_slab_pre_alloc_hook() are
+ * inlined by the compiler. Since the calls to memcg_slab_post_alloc_hook() are
  * conditional to this static branch, we'll have to allow modules that does
  * kmem_cache_alloc and the such to see this symbol as well
  */



Thanks!