From: Roman Gushchin <roman.gushchin@linux.dev>
To: Christoph Lameter <cl@linux.com>
Cc: Matthew Wilcox <willy@infradead.org>,
linux-mm@kvack.org, cgroups@vger.kernel.org,
Shakeel Butt <shakeelb@google.com>,
Michal Hocko <mhocko@suse.com>
Subject: Re: cgroups: warning for metadata allocation with GFP_NOFAIL (was Re: folio_alloc_buffers() doing allocations > order 1 with GFP_NOFAIL)
Date: Tue, 7 Nov 2023 10:05:24 -0800 [thread overview]
Message-ID: <ZUp8ZFGxwmCx4ZFr@P9FQF9L96D.corp.robot.car> (raw)
In-Reply-To: <8f6d3d89-3632-01a8-80b8-6a788a4ba7a8@linux.com>
On Mon, Nov 06, 2023 at 06:57:05PM -0800, Christoph Lameter wrote:
> Right.. Well lets add the cgoup folks to this.
Hello!
I think it's the best thing we can do now. Thoughts?
From 5ed3e88f4f052b6ce8dbec0545dfc80eb7534a1a Mon Sep 17 00:00:00 2001
From: Roman Gushchin <roman.gushchin@linux.dev>
Date: Tue, 7 Nov 2023 09:18:02 -0800
Subject: [PATCH] mm: kmem: drop __GFP_NOFAIL when allocating objcg vectors
Objcg vectors attached to slab pages to store slab object ownership
information are allocated using gfp flags for the original slab
allocation. Depending on slab page order and the size of slab objects,
objcg vector can take several pages.
If the original allocation was done with the __GFP_NOFAIL flag, it
triggered a warning in the page allocation code. Indeed, order > 1
pages should not been allocated with the __GFP_NOFAIL flag.
Fix this by simple dropping the __GFP_NOFAIL flag when allocating
the objcg vector. It effectively allows to skip the accounting of a
single slab object under a heavy memory pressure.
An alternative would be to implement the mechanism to fallback to
order-0 allocations for accounting metadata, which is also not perfect
because it will increase performance penalty and memory footprint
of the kernel memory accounting under memory pressure.
Reported-by: Christoph Lameter <cl@linux.com>
Signed-off-by: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Matthew Wilcox <willy@infradead.org>
---
mm/memcontrol.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 774bd6e21e27..1c1061df9cd1 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -2936,7 +2936,8 @@ void mem_cgroup_commit_charge(struct folio *folio, struct mem_cgroup *memcg)
* Moreover, it should not come from DMA buffer and is not readily
* reclaimable. So those GFP bits should be masked off.
*/
-#define OBJCGS_CLEAR_MASK (__GFP_DMA | __GFP_RECLAIMABLE | __GFP_ACCOUNT)
+#define OBJCGS_CLEAR_MASK (__GFP_DMA | __GFP_RECLAIMABLE | \
+ __GFP_ACCOUNT | __GFP_NOFAIL)
/*
* mod_objcg_mlstate() may be called with irq enabled, so
--
2.42.0
next prev parent reply other threads:[~2023-11-07 18:05 UTC|newest]
Thread overview: 15+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-11-01 0:13 folio_alloc_buffers() doing allocations > order 1 with GFP_NOFAIL Christoph Lameter (Ampere)
2023-11-01 8:08 ` Matthew Wilcox
2023-11-07 2:57 ` cgroups: warning for metadata allocation with GFP_NOFAIL (was Re: folio_alloc_buffers() doing allocations > order 1 with GFP_NOFAIL) Christoph Lameter
2023-11-07 18:05 ` Roman Gushchin [this message]
2023-11-07 18:18 ` Shakeel Butt
2023-11-08 10:33 ` Michal Hocko
2023-11-09 6:37 ` Shakeel Butt
2023-11-09 17:36 ` Roman Gushchin
2023-11-07 19:24 ` Matthew Wilcox
2023-11-07 21:33 ` Roman Gushchin
2023-11-07 21:37 ` Matthew Wilcox
2023-11-10 13:38 ` Matthew Wilcox
2023-11-13 19:48 ` Christoph Lameter
2023-11-13 22:48 ` Matthew Wilcox
2023-11-14 17:29 ` Roman Gushchin
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=ZUp8ZFGxwmCx4ZFr@P9FQF9L96D.corp.robot.car \
--to=roman.gushchin@linux.dev \
--cc=cgroups@vger.kernel.org \
--cc=cl@linux.com \
--cc=linux-mm@kvack.org \
--cc=mhocko@suse.com \
--cc=shakeelb@google.com \
--cc=willy@infradead.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox