From: Oscar Salvador <osalvador@suse.de>
To: Muchun Song <songmuchun@bytedance.com>
Cc: corbet@lwn.net, mike.kravetz@oracle.com, tglx@linutronix.de,
mingo@redhat.com, bp@alien8.de, x86@kernel.org, hpa@zytor.com,
dave.hansen@linux.intel.com, luto@kernel.org,
peterz@infradead.org, viro@zeniv.linux.org.uk,
akpm@linux-foundation.org, paulmck@kernel.org,
mchehab+huawei@kernel.org, pawan.kumar.gupta@linux.intel.com,
rdunlap@infradead.org, oneukum@suse.com,
anshuman.khandual@arm.com, jroedel@suse.de,
almasrymina@google.com, rientjes@google.com, willy@infradead.org,
mhocko@suse.com, song.bao.hua@hisilicon.com, david@redhat.com,
naoya.horiguchi@nec.com, duanxiongchun@bytedance.com,
linux-doc@vger.kernel.org, linux-kernel@vger.kernel.org,
linux-mm@kvack.org, linux-fsdevel@vger.kernel.org
Subject: Re: [PATCH v14 4/8] mm: hugetlb: alloc the vmemmap pages associated with each HugeTLB page
Date: Fri, 5 Feb 2021 12:54:03 +0100 [thread overview]
Message-ID: <20210205115351.GA16428@linux> (raw)
In-Reply-To: <20210204035043.36609-5-songmuchun@bytedance.com>
On Thu, Feb 04, 2021 at 11:50:39AM +0800, Muchun Song wrote:
> When we free a HugeTLB page to the buddy allocator, we should allocate the
> vmemmap pages associated with it. But we may cannot allocate vmemmap pages
> when the system is under memory pressure, in this case, we just refuse to
> free the HugeTLB page instead of looping forever trying to allocate the
> pages.
>
> Signed-off-by: Muchun Song <songmuchun@bytedance.com>
[...]
> diff --git a/mm/hugetlb.c b/mm/hugetlb.c
> index 4cfca27c6d32..5518283aa667 100644
> --- a/mm/hugetlb.c
> +++ b/mm/hugetlb.c
> @@ -1397,16 +1397,26 @@ static void __free_huge_page(struct page *page)
> h->resv_huge_pages++;
>
> if (HPageTemporary(page)) {
> - list_del(&page->lru);
> ClearHPageTemporary(page);
> +
> + if (alloc_huge_page_vmemmap(h, page, GFP_ATOMIC)) {
> + h->surplus_huge_pages++;
> + h->surplus_huge_pages_node[nid]++;
> + goto enqueue;
> + }
> + list_del(&page->lru);
> update_and_free_page(h, page);
> } else if (h->surplus_huge_pages_node[nid]) {
> + if (alloc_huge_page_vmemmap(h, page, GFP_ATOMIC))
> + goto enqueue;
> +
> /* remove the page from active list */
> list_del(&page->lru);
> update_and_free_page(h, page);
> h->surplus_huge_pages--;
> h->surplus_huge_pages_node[nid]--;
> } else {
> +enqueue:
> arch_clear_hugepage_flags(page);
> enqueue_huge_page(h, page);
Ok, we just keep them in the pool in case we fail to allocate.
> diff --git a/mm/hugetlb_vmemmap.c b/mm/hugetlb_vmemmap.c
> index ddd872ab6180..0bd6b8d7282d 100644
> --- a/mm/hugetlb_vmemmap.c
> +++ b/mm/hugetlb_vmemmap.c
> @@ -169,6 +169,8 @@
> * (last) level. So this type of HugeTLB page can be optimized only when its
> * size of the struct page structs is greater than 2 pages.
[...]
> +int alloc_huge_page_vmemmap(struct hstate *h, struct page *head, gfp_t gfp_mask)
> +{
> + int ret;
> + unsigned long vmemmap_addr = (unsigned long)head;
> + unsigned long vmemmap_end, vmemmap_reuse;
> +
> + if (!free_vmemmap_pages_per_hpage(h))
> + return 0;
> +
> + vmemmap_addr += RESERVE_VMEMMAP_SIZE;
> + vmemmap_end = vmemmap_addr + free_vmemmap_pages_size_per_hpage(h);
> + vmemmap_reuse = vmemmap_addr - PAGE_SIZE;
> +
> + /*
> + * The pages which the vmemmap virtual address range [@vmemmap_addr,
> + * @vmemmap_end) are mapped to are freed to the buddy allocator, and
> + * the range is mapped to the page which @vmemmap_reuse is mapped to.
> + * When a HugeTLB page is freed to the buddy allocator, previously
> + * discarded vmemmap pages must be allocated and remapping.
> + */
> + ret = vmemmap_remap_alloc(vmemmap_addr, vmemmap_end, vmemmap_reuse,
> + gfp_mask | __GFP_NOWARN | __GFP_THISNODE);
Why don't you set all the GFP flags here?
vmemmap_remap_alloc(vmemmap_addr, vmemmap_end, vmemmap_reuse, GFP_ATOMIC|
__GFP_NOWARN | __GFP_THISNODE) ?
> diff --git a/mm/sparse-vmemmap.c b/mm/sparse-vmemmap.c
> index 50c1dc00b686..277eb43aebd5 100644
> --- a/mm/sparse-vmemmap.c
> +++ b/mm/sparse-vmemmap.c
[...]
> +static int alloc_vmemmap_page_list(unsigned long start, unsigned long end,
> + gfp_t gfp_mask, struct list_head *list)
I think it would make more sense for this function to get the nid and the
nr_pages to allocate directly.
> +{
> + unsigned long addr;
> + int nid = page_to_nid((const void *)start);
Uh, that void is a bit ugly. page_to_nid(struct page *)start).
Do not need the const either.
> + struct page *page, *next;
> +
> + for (addr = start; addr < end; addr += PAGE_SIZE) {
> + page = alloc_pages_node(nid, gfp_mask, 0);
> + if (!page)
> + goto out;
> + list_add_tail(&page->lru, list);
> + }
and replace this by while(--nr_pages) etc.
I did not really go in depth, but looks good to me, and much more simply
overall.
The only thing I am not sure about is the use of GFP_ATOMIC.
It has been raised before than when we are close to OOM, the user might want
to try to free up some memory by dissolving free_huge_pages, and so we might
want to dip in the reserves.
Given the fact that we are prepared to fail, and that we do not retry, I would
rather use GFP_KERNEL than to have X pages atomically allocated and then realize
we need to drop them on the ground because we cannot go further at some point.
I think those reserves would be better off used by someone else in that
situation.
But this is just my thoughs, and given the fact that there seems to be a consensus
of susing GFP_ATOMIC.
--
Oscar Salvador
SUSE L3
next prev parent reply other threads:[~2021-02-05 11:58 UTC|newest]
Thread overview: 29+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-02-04 3:50 [PATCH v14 0/8] Free some vmemmap pages of HugeTLB page Muchun Song
2021-02-04 3:50 ` [PATCH v14 1/8] mm: memory_hotplug: factor out bootmem core functions to bootmem_info.c Muchun Song
2021-02-04 3:50 ` [PATCH v14 2/8] mm: hugetlb: introduce a new config HUGETLB_PAGE_FREE_VMEMMAP Muchun Song
2021-02-04 11:44 ` Miaohe Lin
2021-02-04 3:50 ` [PATCH v14 3/8] mm: hugetlb: free the vmemmap pages associated with each HugeTLB page Muchun Song
2021-02-05 8:54 ` Oscar Salvador
2021-02-05 16:01 ` [External] " Muchun Song
2021-02-04 3:50 ` [PATCH v14 4/8] mm: hugetlb: alloc " Muchun Song
2021-02-05 9:29 ` Muchun Song
2021-02-05 11:54 ` Oscar Salvador [this message]
2021-02-06 8:01 ` [External] " Muchun Song
2021-02-04 3:50 ` [PATCH v14 5/8] mm: hugetlb: add a kernel parameter hugetlb_free_vmemmap Muchun Song
2021-02-05 7:25 ` Miaohe Lin
2021-02-04 3:50 ` [PATCH v14 6/8] mm: hugetlb: introduce nr_free_vmemmap_pages in the struct hstate Muchun Song
2021-02-05 7:29 ` Miaohe Lin
2021-02-05 8:22 ` Oscar Salvador
2021-02-05 8:39 ` Miaohe Lin
2021-02-05 8:56 ` Oscar Salvador
2021-02-05 9:12 ` Miaohe Lin
2021-02-04 3:50 ` [PATCH v14 7/8] mm: hugetlb: gather discrete indexes of tail page Muchun Song
2021-02-05 7:30 ` Miaohe Lin
2021-02-04 3:50 ` [PATCH v14 8/8] mm: hugetlb: optimize the code with the help of the compiler Muchun Song
2021-02-04 6:33 ` Miaohe Lin
2021-02-05 9:09 ` Oscar Salvador
2021-02-05 9:16 ` [External] " Muchun Song
2021-02-05 8:59 ` [PATCH v14 0/8] Free some vmemmap pages of HugeTLB page Oscar Salvador
2021-02-05 9:30 ` [External] " Muchun Song
2021-02-05 16:00 ` Joao Martins
2021-02-05 16:13 ` [External] " Muchun Song
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=20210205115351.GA16428@linux \
--to=osalvador@suse.de \
--cc=akpm@linux-foundation.org \
--cc=almasrymina@google.com \
--cc=anshuman.khandual@arm.com \
--cc=bp@alien8.de \
--cc=corbet@lwn.net \
--cc=dave.hansen@linux.intel.com \
--cc=david@redhat.com \
--cc=duanxiongchun@bytedance.com \
--cc=hpa@zytor.com \
--cc=jroedel@suse.de \
--cc=linux-doc@vger.kernel.org \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=luto@kernel.org \
--cc=mchehab+huawei@kernel.org \
--cc=mhocko@suse.com \
--cc=mike.kravetz@oracle.com \
--cc=mingo@redhat.com \
--cc=naoya.horiguchi@nec.com \
--cc=oneukum@suse.com \
--cc=paulmck@kernel.org \
--cc=pawan.kumar.gupta@linux.intel.com \
--cc=peterz@infradead.org \
--cc=rdunlap@infradead.org \
--cc=rientjes@google.com \
--cc=song.bao.hua@hisilicon.com \
--cc=songmuchun@bytedance.com \
--cc=tglx@linutronix.de \
--cc=viro@zeniv.linux.org.uk \
--cc=willy@infradead.org \
--cc=x86@kernel.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;
as well as URLs for NNTP newsgroup(s).