From: Yang Shi <yang.shi@linux.alibaba.com>
To: "Kirill A. Shutemov" <kirill@shutemov.name>
Cc: mhocko@kernel.org, willy@infradead.org,
ldufour@linux.vnet.ibm.com, vbabka@suse.cz,
akpm@linux-foundation.org, linux-mm@kvack.org,
linux-kernel@vger.kernel.org
Subject: Re: [v2 PATCH 2/2 -mm] mm: brk: dwongrade mmap_sem to read when shrinking
Date: Thu, 27 Sep 2018 09:06:29 -0700 [thread overview]
Message-ID: <d5443d11-7422-6b5b-f6a9-db09311bc827@linux.alibaba.com> (raw)
In-Reply-To: <20180927125025.xnvoh2btdq5kjmai@kshutemo-mobl1>
On 9/27/18 5:50 AM, Kirill A. Shutemov wrote:
> On Thu, Sep 27, 2018 at 02:10:34AM +0800, Yang Shi wrote:
>> brk might be used to shinrk memory mapping too other than munmap().
> s/shinrk/shrink/
>
>> So, it may hold write mmap_sem for long time when shrinking large
>> mapping, as what commit ("mm: mmap: zap pages with read mmap_sem in
>> munmap") described.
>>
>> The brk() will not manipulate vmas anymore after __do_munmap() call for
>> the mapping shrink use case. But, it may set mm->brk after
>> __do_munmap(), which needs hold write mmap_sem.
>>
>> However, a simple trick can workaround this by setting mm->brk before
>> __do_munmap(). Then restore the original value if __do_munmap() fails.
>> With this trick, it is safe to downgrade to read mmap_sem.
>>
>> So, the same optimization, which downgrades mmap_sem to read for
>> zapping pages, is also feasible and reasonable to this case.
>>
>> The period of holding exclusive mmap_sem for shrinking large mapping
>> would be reduced significantly with this optimization.
>>
>> Cc: Michal Hocko <mhocko@kernel.org>
>> Cc: Kirill A. Shutemov <kirill@shutemov.name>
>> Cc: Matthew Wilcox <willy@infradead.org>
>> Cc: Laurent Dufour <ldufour@linux.vnet.ibm.com>
>> Cc: Vlastimil Babka <vbabka@suse.cz>
>> Cc: Andrew Morton <akpm@linux-foundation.org>
>> Signed-off-by: Yang Shi <yang.shi@linux.alibaba.com>
>> ---
>> v2: Rephrase the commit per Michal
>>
>> mm/mmap.c | 40 ++++++++++++++++++++++++++++++----------
>> 1 file changed, 30 insertions(+), 10 deletions(-)
>>
>> diff --git a/mm/mmap.c b/mm/mmap.c
>> index 017bcfa..0d2fae1 100644
>> --- a/mm/mmap.c
>> +++ b/mm/mmap.c
>> @@ -193,9 +193,11 @@ static int do_brk_flags(unsigned long addr, unsigned long request, unsigned long
>> unsigned long retval;
>> unsigned long newbrk, oldbrk;
>> struct mm_struct *mm = current->mm;
>> + unsigned long origbrk = mm->brk;
> Is it safe to read mm->brk outside the lock?
Aha, thanks for catching this. It can be moved inside down_write().
Will solve in the next version.
Thanks,
Yang
>
>> struct vm_area_struct *next;
>> unsigned long min_brk;
>> bool populate;
>> + bool downgrade = false;
> Again,
>
> s/downgrade/downgraded/ ?
>
>> LIST_HEAD(uf);
>>
>> if (down_write_killable(&mm->mmap_sem))
>> @@ -229,14 +231,29 @@ static int do_brk_flags(unsigned long addr, unsigned long request, unsigned long
>>
>> newbrk = PAGE_ALIGN(brk);
>> oldbrk = PAGE_ALIGN(mm->brk);
>> - if (oldbrk == newbrk)
>> - goto set_brk;
>> + if (oldbrk == newbrk) {
>> + mm->brk = brk;
>> + goto success;
>> + }
>>
>> - /* Always allow shrinking brk. */
>> + /*
>> + * Always allow shrinking brk.
>> + * __do_munmap() may downgrade mmap_sem to read.
>> + */
>> if (brk <= mm->brk) {
>> - if (!do_munmap(mm, newbrk, oldbrk-newbrk, &uf))
>> - goto set_brk;
>> - goto out;
>> + /*
>> + * mm->brk need to be protected by write mmap_sem, update it
>> + * before downgrading mmap_sem.
>> + * When __do_munmap fail, it will be restored from origbrk.
>> + */
>> + mm->brk = brk;
>> + retval = __do_munmap(mm, newbrk, oldbrk-newbrk, &uf, true);
>> + if (retval < 0) {
>> + mm->brk = origbrk;
>> + goto out;
>> + } else if (retval == 1)
>> + downgrade = true;
>> + goto success;
>> }
>>
>> /* Check against existing mmap mappings. */
>> @@ -247,18 +264,21 @@ static int do_brk_flags(unsigned long addr, unsigned long request, unsigned long
>> /* Ok, looks good - let it rip. */
>> if (do_brk_flags(oldbrk, newbrk-oldbrk, 0, &uf) < 0)
>> goto out;
>> -
>> -set_brk:
>> mm->brk = brk;
>> +
>> +success:
>> populate = newbrk > oldbrk && (mm->def_flags & VM_LOCKED) != 0;
>> - up_write(&mm->mmap_sem);
>> + if (downgrade)
>> + up_read(&mm->mmap_sem);
>> + else
>> + up_write(&mm->mmap_sem);
>> userfaultfd_unmap_complete(mm, &uf);
>> if (populate)
>> mm_populate(oldbrk, newbrk - oldbrk);
>> return brk;
>>
>> out:
>> - retval = mm->brk;
>> + retval = origbrk;
>> up_write(&mm->mmap_sem);
>> return retval;
>> }
>> --
>> 1.8.3.1
>>
next prev parent reply other threads:[~2018-09-27 16:06 UTC|newest]
Thread overview: 10+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-09-26 18:10 [v2 PATCH 1/2 -mm] mm: mremap: dwongrade mmap_sem to read when shrinking Yang Shi
2018-09-26 18:10 ` [v2 PATCH 2/2 -mm] mm: brk: " Yang Shi
2018-09-27 12:14 ` Vlastimil Babka
2018-09-27 16:05 ` Yang Shi
2018-09-27 12:50 ` Kirill A. Shutemov
2018-09-27 13:21 ` Vlastimil Babka
2018-09-27 16:06 ` Yang Shi [this message]
2018-09-27 11:50 ` [v2 PATCH 1/2 -mm] mm: mremap: " Vlastimil Babka
2018-09-27 16:04 ` Yang Shi
2018-09-27 12:46 ` Kirill A. Shutemov
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=d5443d11-7422-6b5b-f6a9-db09311bc827@linux.alibaba.com \
--to=yang.shi@linux.alibaba.com \
--cc=akpm@linux-foundation.org \
--cc=kirill@shutemov.name \
--cc=ldufour@linux.vnet.ibm.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=mhocko@kernel.org \
--cc=vbabka@suse.cz \
--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;
as well as URLs for NNTP newsgroup(s).