From: Michal Hocko <mhocko@suse.com>
To: Zach O'Keefe <zokeefe@google.com>
Cc: Yang Shi <shy828301@gmail.com>,
akpm@linux-foundation.org, linux-mm@kvack.org,
linux-kernel@vger.kernel.org
Subject: Re: [PATCH] mm: don't warn if the node is offlined
Date: Wed, 2 Nov 2022 08:39:24 +0100 [thread overview]
Message-ID: <Y2IerOXJ+ZoRTHcs@dhcp22.suse.cz> (raw)
In-Reply-To: <CAAa6QmRe1zMp8P-gZjR63Fg6KhOw+fP-v7SQWLNKuc2Y9ZxvyA@mail.gmail.com>
On Tue 01-11-22 12:13:35, Zach O'Keefe wrote:
[...]
> This is slightly tangential - but I don't want to send a new mail
> about it -- but I wonder if we should be doing __GFP_THISNODE +
> explicit node vs having hpage_collapse_find_target_node() set a
> nodemask. We could then provide fallback nodes for ties, or if some
> node contained > some threshold number of pages.
I would simply go with something like this (not even compile tested):
diff --git a/mm/khugepaged.c b/mm/khugepaged.c
index 4734315f7940..947a5158fe11 100644
--- a/mm/khugepaged.c
+++ b/mm/khugepaged.c
@@ -96,9 +96,6 @@ struct collapse_control {
/* Num pages scanned per node */
u32 node_load[MAX_NUMNODES];
-
- /* Last target selected in hpage_collapse_find_target_node() */
- int last_target_node;
};
/**
@@ -734,7 +731,6 @@ static void khugepaged_alloc_sleep(void)
struct collapse_control khugepaged_collapse_control = {
.is_khugepaged = true,
- .last_target_node = NUMA_NO_NODE,
};
static bool hpage_collapse_scan_abort(int nid, struct collapse_control *cc)
@@ -772,7 +768,7 @@ static inline gfp_t alloc_hugepage_khugepaged_gfpmask(void)
}
#ifdef CONFIG_NUMA
-static int hpage_collapse_find_target_node(struct collapse_control *cc)
+static int hpage_collapse_find_target_node(struct collapse_control *cc, nodemask_t *alloc_mask)
{
int nid, target_node = 0, max_value = 0;
@@ -783,28 +779,25 @@ static int hpage_collapse_find_target_node(struct collapse_control *cc)
target_node = nid;
}
+ nodes_clear(&alloc_mask);
/* do some balance if several nodes have the same hit record */
- if (target_node <= cc->last_target_node)
- for (nid = cc->last_target_node + 1; nid < MAX_NUMNODES;
- nid++)
- if (max_value == cc->node_load[nid]) {
- target_node = nid;
- break;
- }
+ for_each_online_node(nid) {_
+ if (max_value == cc->node_load[nid])
+ node_set(nid, &alloc_mask)
+ }
- cc->last_target_node = target_node;
return target_node;
}
#else
-static int hpage_collapse_find_target_node(struct collapse_control *cc)
+static int hpage_collapse_find_target_node(struct collapse_control *cc, nodemask_t *alloc_mask)
{
return 0;
}
#endif
-static bool hpage_collapse_alloc_page(struct page **hpage, gfp_t gfp, int node)
+static bool hpage_collapse_alloc_page(struct page **hpage, gfp_t gfp, int node, nodemask_t *nmask)
{
- *hpage = __alloc_pages_node(node, gfp, HPAGE_PMD_ORDER);
+ *hpage = __alloc_pages(gfp, HPAGE_PMD_ORDER, node, nmask);
if (unlikely(!*hpage)) {
count_vm_event(THP_COLLAPSE_ALLOC_FAILED);
return false;
@@ -958,9 +951,18 @@ static int alloc_charge_hpage(struct page **hpage, struct mm_struct *mm,
/* Only allocate from the target node */
gfp_t gfp = (cc->is_khugepaged ? alloc_hugepage_khugepaged_gfpmask() :
GFP_TRANSHUGE) | __GFP_THISNODE;
- int node = hpage_collapse_find_target_node(cc);
+ NODEMASK_ALLOC(nodemask_t, nmask, GFP_KERNEL);
+ int node;
+ int ret;
+
+ if (!nmaks)
+ return SCAN_ALLOC_HUGE_PAGE_FAIL;
+
+ node = hpage_collapse_find_target_node(cc, nmask);
+ ret = hpage_collapse_alloc_page(hpage, gfp, node, nmask);
+ NODEMASK_FREE(nmask);
- if (!hpage_collapse_alloc_page(hpage, gfp, node))
+ if (!ret)
return SCAN_ALLOC_HUGE_PAGE_FAIL;
if (unlikely(mem_cgroup_charge(page_folio(*hpage), mm, gfp)))
return SCAN_CGROUP_CHARGE_FAIL;
@@ -2576,7 +2578,6 @@ int madvise_collapse(struct vm_area_struct *vma, struct vm_area_struct **prev,
if (!cc)
return -ENOMEM;
cc->is_khugepaged = false;
- cc->last_target_node = NUMA_NO_NODE;
mmgrab(mm);
lru_add_drain_all();
--
Michal Hocko
SUSE Labs
next prev parent reply other threads:[~2022-11-02 7:39 UTC|newest]
Thread overview: 25+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-10-31 18:31 [PATCH] mm: don't warn if the node is offlined Yang Shi
2022-10-31 21:16 ` Zach O'Keefe
2022-10-31 22:08 ` Michal Hocko
2022-11-01 0:05 ` Zach O'Keefe
2022-11-01 7:54 ` Michal Hocko
2022-11-01 17:12 ` Yang Shi
2022-11-01 19:13 ` Zach O'Keefe
2022-11-01 20:09 ` Yang Shi
2022-11-01 22:05 ` Zach O'Keefe
2022-11-02 7:39 ` Michal Hocko [this message]
2022-11-02 7:49 ` Michal Hocko
2022-11-02 16:03 ` Yang Shi
2022-11-02 16:15 ` Michal Hocko
2022-11-02 17:36 ` Yang Shi
2022-11-02 17:47 ` Michal Hocko
2022-11-02 18:18 ` Yang Shi
2022-11-02 18:58 ` Zach O'Keefe
2022-11-02 20:08 ` Yang Shi
2022-11-02 20:21 ` Zach O'Keefe
2022-11-03 7:54 ` Michal Hocko
2022-11-03 17:13 ` Yang Shi
2022-11-03 7:51 ` Michal Hocko
2022-11-02 7:14 ` Michal Hocko
2022-11-02 15:58 ` Yang Shi
2022-11-02 16:11 ` Michal Hocko
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=Y2IerOXJ+ZoRTHcs@dhcp22.suse.cz \
--to=mhocko@suse.com \
--cc=akpm@linux-foundation.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=shy828301@gmail.com \
--cc=zokeefe@google.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;
as well as URLs for NNTP newsgroup(s).