linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
From: Nico Pache <npache@redhat.com>
To: Michal Hocko <mhocko@suse.com>,
	Kirill Tkhai <ktkhai@virtuozzo.com>,
	David Hildenbrand <david@redhat.com>
Cc: linux-kernel@vger.kernel.org, linux-mm@kvack.org,
	akpm@linux-foundation.org, shakeelb@google.com,
	shy828301@gmail.com, guro@fb.com, vbabka@suse.cz,
	vdavydov.dev@gmail.com, raquini@redhat.com
Subject: Re: [RFC PATCH 2/2] mm/vmscan.c: Prevent allocating shrinker_info on offlined nodes
Date: Wed, 8 Dec 2021 14:00:01 -0500	[thread overview]
Message-ID: <aa8a8deb-0fdb-9408-48d4-adadb5602d72@redhat.com> (raw)
In-Reply-To: <Ya4PGJZL8tSb/Prj@dhcp22.suse.cz>



On 12/6/21 08:24, Michal Hocko wrote:
> On Mon 06-12-21 16:19:12, Kirill Tkhai wrote:
>> On 06.12.2021 13:45, David Hildenbrand wrote:
>>>> This doesn't seen complete. Slab shrinkers are used in the reclaim
>>>> context. Previously offline nodes could be onlined later and this would
>>>> lead to NULL ptr because there is no hook to allocate new shrinker
>>>> infos. This would be also really impractical because this would have to
>>>> update all existing memcgs...
>>>
>>> Instead of going through the trouble of updating...
>>>
>>> ...  maybe just keep for_each_node() and check if the target node is
>>> offline. If it's offline, just allocate from the first online node.
>>> After all, we're not using __GFP_THISNODE, so there are no guarantees
>>> either way ...
>>
>> Hm, can't we add shrinker maps allocation to __try_online_node() in addition
>> to this patch? 
> 
> Either that or through hotplug notifier (which would be a better
> solution). But allocating a new shrinker map for each memcg would have
> to be done as has been mentioned earlier.

I took a stab at this approach. It may be incomplete but please let me know what
you think. This would go on top of this series.

diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h
index 0c5c403f4be6..6c842382fa73 100644
--- a/include/linux/memcontrol.h
+++ b/include/linux/memcontrol.h
@@ -520,6 +520,7 @@ static inline struct mem_cgroup *page_memcg_check(struct
page *page)
 	return (struct mem_cgroup *)(memcg_data & ~MEMCG_DATA_FLAGS_MASK);
 }

+int alloc_mem_cgroup_per_node_info(struct mem_cgroup *memcg, int node);
 #ifdef CONFIG_MEMCG_KMEM
 /*
  * folio_memcg_kmem - Check if the folio has the memcg_kmem flag set.
diff --git a/include/linux/node.h b/include/linux/node.h
index bb21fd631b16..5e8c737ea751 100644
--- a/include/linux/node.h
+++ b/include/linux/node.h
@@ -19,7 +19,7 @@
 #include <linux/cpumask.h>
 #include <linux/list.h>
 #include <linux/workqueue.h>
-
+#include <linux/memcontrol.h>
 /**
  * struct node_hmem_attrs - heterogeneous memory performance attributes
  *
@@ -118,6 +118,7 @@ extern int __register_one_node(int nid);
 /* Registers an online node */
 static inline int register_one_node(int nid)
 {
+	struct mem_cgroup *memcg;
 	int error = 0;

 	if (node_online(nid)) {
@@ -130,6 +131,14 @@ static inline int register_one_node(int nid)
 			return error;
 		/* link memory sections under this node */
 		link_mem_sections(nid, start_pfn, end_pfn, MEMINIT_EARLY);
+		/* Iterate over memcgs and update nodeinfo  */
+		memcg = mem_cgroup_iter(NULL, NULL, NULL);
+		do {
+			if (alloc_mem_cgroup_per_node_info(memcg,nid)) {
+				mem_cgroup_iter_break(NULL, memcg);
+				return error;
+			}
+		} while ((memcg = mem_cgroup_iter(NULL, memcg, NULL)) != NULL);
 	}

 	return error;
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 6863a834ed42..2d55fad3229b 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -5041,18 +5041,11 @@ struct mem_cgroup *mem_cgroup_from_id(unsigned short id)
 	return idr_find(&mem_cgroup_idr, id);
 }

-static int alloc_mem_cgroup_per_node_info(struct mem_cgroup *memcg, int node)
+int alloc_mem_cgroup_per_node_info(struct mem_cgroup *memcg, int node)
 {
 	struct mem_cgroup_per_node *pn;
 	int tmp = node;
-	/*
-	 * This routine is called against possible nodes.
-	 * But it's BUG to call kmalloc() against offline node.
-	 *
-	 * TODO: this routine can waste much memory for nodes which will
-	 *       never be onlined. It's better to use memory hotplug callback
-	 *       function.
-	 */
+
 	if (!node_state(node, N_NORMAL_MEMORY))
 		tmp = -1;
 	pn = kzalloc_node(sizeof(*pn), GFP_KERNEL, tmp);
@@ -5130,7 +5123,7 @@ static struct mem_cgroup *mem_cgroup_alloc(void)
 	if (!memcg->vmstats_percpu)
 		goto fail;

-	for_each_node(node)
+	for_each_online_node(node)
 		if (alloc_mem_cgroup_per_node_info(memcg, node))
 			goto fail;



  reply	other threads:[~2021-12-08 19:00 UTC|newest]

Thread overview: 28+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-12-06  3:33 [RFC PATCH 0/2] mm: Dont allocate pages on a offline node Nico Pache
2021-12-06  3:33 ` [RFC PATCH 1/2] include/linux/gfp.h: Do not allocate pages on a offlined node Nico Pache
2021-12-06  3:37   ` Matthew Wilcox
2021-12-06  9:22   ` Michal Hocko
2021-12-07 21:24     ` Nico Pache
2021-12-06  3:33 ` [RFC PATCH 2/2] mm/vmscan.c: Prevent allocating shrinker_info on offlined nodes Nico Pache
2021-12-06  9:22   ` Michal Hocko
2021-12-06  9:24     ` Michal Hocko
2021-12-07 21:34     ` Nico Pache
     [not found]     ` <d9d14beb-ee20-7ebb-e007-fbf58fb28535@redhat.com>
2021-12-06 10:54       ` Michal Hocko
     [not found]         ` <840cb3d0-61fe-b6cb-9918-69146ba06cf7@redhat.com>
2021-12-06 11:22           ` Michal Hocko
     [not found]             ` <51c65635-1dae-6ba4-daf9-db9df0ec35d8@redhat.com>
2021-12-06 13:06               ` Michal Hocko
     [not found]                 ` <05157de4-e5df-11fc-fc46-8a9f79d0ddb4@redhat.com>
2021-12-06 14:06                   ` Michal Hocko
     [not found]                     ` <d4f281e6-1999-a3de-b879-c6ca6a25ae67@redhat.com>
2021-12-06 14:21                       ` Michal Hocko
2021-12-06 14:30                         ` Vlastimil Babka
2021-12-06 14:53                           ` Michal Hocko
2021-12-06 18:26                             ` Yang Shi
2021-12-07 10:15                               ` Michal Hocko
2021-12-06 14:15                   ` Michal Hocko
2021-12-07 21:40       ` Nico Pache
     [not found]       ` <24b4455c-aff9-ca9f-e29f-350833e7a0d1@virtuozzo.com>
2021-12-06 13:24         ` Michal Hocko
2021-12-08 19:00           ` Nico Pache [this message]
2021-12-06 18:42         ` Yang Shi
     [not found]           ` <a48c16d6-07df-ff44-67e6-f0942672ec28@redhat.com>
2021-12-06 21:28             ` Yang Shi
2021-12-07 10:15               ` David Hildenbrand
2021-12-07 10:55             ` Michal Hocko
2021-12-07 21:45         ` Nico Pache
2021-12-06 18:45   ` Yang Shi

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=aa8a8deb-0fdb-9408-48d4-adadb5602d72@redhat.com \
    --to=npache@redhat.com \
    --cc=akpm@linux-foundation.org \
    --cc=david@redhat.com \
    --cc=guro@fb.com \
    --cc=ktkhai@virtuozzo.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=mhocko@suse.com \
    --cc=raquini@redhat.com \
    --cc=shakeelb@google.com \
    --cc=shy828301@gmail.com \
    --cc=vbabka@suse.cz \
    --cc=vdavydov.dev@gmail.com \
    /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