From: Muchun Song <songmuchun@bytedance.com>
To: Mike Kravetz <mike.kravetz@oracle.com>,
Oscar Salvador <osalvador@suse.de>
Cc: "Jonathan Corbet" <corbet@lwn.net>,
"Thomas Gleixner" <tglx@linutronix.de>,
mingo@redhat.com, bp@alien8.de, x86@kernel.org, hpa@zytor.com,
dave.hansen@linux.intel.com, luto@kernel.org,
"Peter Zijlstra" <peterz@infradead.org>,
viro@zeniv.linux.org.uk,
"Andrew Morton" <akpm@linux-foundation.org>,
paulmck@kernel.org, mchehab+huawei@kernel.org,
pawan.kumar.gupta@linux.intel.com,
"Randy Dunlap" <rdunlap@infradead.org>,
oneukum@suse.com, anshuman.khandual@arm.com, jroedel@suse.de,
"Mina Almasry" <almasrymina@google.com>,
"David Rientjes" <rientjes@google.com>,
"Matthew Wilcox" <willy@infradead.org>,
"Michal Hocko" <mhocko@suse.com>,
"Song Bao Hua (Barry Song)" <song.bao.hua@hisilicon.com>,
"David Hildenbrand" <david@redhat.com>,
"HORIGUCHI NAOYA(堀口 直也)" <naoya.horiguchi@nec.com>,
"Xiongchun duan" <duanxiongchun@bytedance.com>,
linux-doc@vger.kernel.org, LKML <linux-kernel@vger.kernel.org>,
"Linux Memory Management List" <linux-mm@kvack.org>,
linux-fsdevel <linux-fsdevel@vger.kernel.org>
Subject: Re: [External] Re: [PATCH v12 04/13] mm/hugetlb: Free the vmemmap pages associated with each HugeTLB page
Date: Thu, 14 Jan 2021 18:54:30 +0800 [thread overview]
Message-ID: <CAMZfGtUhhMDCaZKeayS1+w0MvBijDZC2AiUV4z5rUFrfbXBefw@mail.gmail.com> (raw)
In-Reply-To: <a9baf18c-22c7-4946-9778-678f6bc808dc@oracle.com>
On Thu, Jan 14, 2021 at 7:27 AM Mike Kravetz <mike.kravetz@oracle.com> wrote:
>
> On 1/13/21 1:20 AM, Oscar Salvador wrote:
> > On Tue, Jan 12, 2021 at 07:33:33PM +0800, Muchun Song wrote:
> >>> It seems a bit odd to only pass "start" for the BUG_ON.
> >>> Also, I kind of dislike the "addr += PAGE_SIZE" in vmemmap_pte_range.
> >>>
> >>> I wonder if adding a ".remap_start_addr" would make more sense.
> >>> And adding it here with the vmemmap_remap_walk init.
> >>
> >> How about introducing a new function which aims to get the reuse
> >> page? In this case, we can drop the BUG_ON() and "addr += PAGE_SIZE"
> >> which is in vmemmap_pte_range. The vmemmap_remap_range only
> >> does the remapping.
> >
> > How would that look?
> > It might be good, dunno, but the point is, we should try to make the rules as
> > simple as possible, dropping weird assumptions.
> >
> > Callers of vmemmap_remap_free should know three things:
> >
> > - Range to be remapped
> > - Addr to remap to
> > - Current implemantion needs addr to be remap to to be part of the complete
> > range
> >
> > right?
>
> And, current implementation needs must have remap addr be the first in the
> complete range. This is just because of the way the page tables are walked
> for remapping. The remap/reuse page must be found first so that the following
> pages can be remapped to it.
You are right.
>
> That implementation seems to be the 'most efficient' for hugetlb pages where
> we want vmemmap pages n+3 and beyond mapped to n+2.
>
> In a more general purpose vmemmap_remap_free implementation, the reuse/remap
> address would not necessarily need to be related to the range. However, this
> would require a separate page table walk/validation for the reuse address
> independent of the range. This may be what Muchun was proposing for 'a new
> function which aims to get the reuse page'.
Agree.
>
> IMO, the decision on how to implement depends on the intended use case.
> - If this is going to be hugetlb only (or perhaps generic huge page only)
> functionality, then I am OK with an efficient implementation that has
> some restrictions.
> - If we see this being used for more general purpose remapping, then we
> should go with a more general purpose implementation.
I think this approach may be only suitable for generic huge page only.
So we can implement it only for huge page.
Hi Oscar,
What's your opinion about this?
>
> Again, just my opinion.
> --
> Mike Kravetz
next prev parent reply other threads:[~2021-01-14 10:56 UTC|newest]
Thread overview: 26+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-01-06 14:19 [PATCH v12 00/13] Free some vmemmap pages of HugeTLB page Muchun Song
2021-01-06 14:19 ` [PATCH v12 01/13] mm/memory_hotplug: Factor out bootmem core functions to bootmem_info.c Muchun Song
2021-01-06 14:19 ` [PATCH v12 02/13] mm/hugetlb: Introduce a new config HUGETLB_PAGE_FREE_VMEMMAP Muchun Song
2021-01-06 14:19 ` [PATCH v12 03/13] mm: Introduce VM_WARN_ON_PAGE macro Muchun Song
2021-01-13 22:30 ` Mike Kravetz
2021-01-14 2:50 ` [External] " Muchun Song
2021-01-06 14:19 ` [PATCH v12 04/13] mm/hugetlb: Free the vmemmap pages associated with each HugeTLB page Muchun Song
2021-01-12 8:04 ` Oscar Salvador
2021-01-12 11:33 ` [External] " Muchun Song
2021-01-13 9:20 ` Oscar Salvador
2021-01-13 23:27 ` Mike Kravetz
2021-01-14 10:54 ` Muchun Song [this message]
2021-01-14 11:52 ` Oscar Salvador
2021-01-14 13:05 ` Muchun Song
2021-01-14 10:57 ` Muchun Song
2021-01-06 14:19 ` [PATCH v12 05/13] mm/hugetlb: Defer freeing of HugeTLB pages Muchun Song
2021-01-06 14:19 ` [PATCH v12 06/13] mm/hugetlb: Allocate the vmemmap pages associated with each HugeTLB page Muchun Song
2021-01-06 14:19 ` [PATCH v12 07/13] mm/hugetlb: Set the PageHWPoison to the raw error page Muchun Song
2021-01-06 14:19 ` [PATCH v12 08/13] mm/hugetlb: Flush work when dissolving a HugeTLB page Muchun Song
2021-01-06 14:19 ` [PATCH v12 09/13] mm/hugetlb: Introduce PageHugeInflight Muchun Song
2021-01-06 14:19 ` [PATCH v12 10/13] mm/hugetlb: Add a kernel parameter hugetlb_free_vmemmap Muchun Song
2021-01-06 14:19 ` [PATCH v12 11/13] mm/hugetlb: Introduce nr_free_vmemmap_pages in the struct hstate Muchun Song
2021-01-06 14:19 ` [PATCH v12 12/13] mm/hugetlb: Gather discrete indexes of tail page Muchun Song
2021-01-14 0:18 ` Mike Kravetz
2021-01-14 2:47 ` [External] " Muchun Song
2021-01-06 14:19 ` [PATCH v12 13/13] mm/hugetlb: Optimize the code with the help of the compiler 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=CAMZfGtUhhMDCaZKeayS1+w0MvBijDZC2AiUV4z5rUFrfbXBefw@mail.gmail.com \
--to=songmuchun@bytedance.com \
--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=osalvador@suse.de \
--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=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).