From: Hugh Dickins <hughd@google.com>
To: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Hugh Dickins <hughd@google.com>, Dave Jones <davej@redhat.com>,
Al Viro <viro@zeniv.linux.org.uk>,
Linux Kernel <linux-kernel@vger.kernel.org>,
Rik van Riel <riel@redhat.com>, Ingo Molnar <mingo@redhat.com>,
Michel Lespinasse <walken@google.com>,
"Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>,
Mel Gorman <mgorman@suse.de>,
Sasha Levin <sasha.levin@oracle.com>
Subject: Re: pipe/page fault oddness.
Date: Thu, 2 Oct 2014 01:47:47 -0700 (PDT) [thread overview]
Message-ID: <alpine.LSU.2.11.1410020142230.6383@eggly.anvils> (raw)
In-Reply-To: <CA+55aFyJ09+iv2HX1nfAaDi0-7=L3KxKi11CACbM8K_Coo6kzg@mail.gmail.com>
On Wed, 1 Oct 2014, Linus Torvalds wrote:
> On Wed, Oct 1, 2014 at 1:19 AM, Hugh Dickins <hughd@google.com> wrote:
>
> Can we please just get rid of _PAGE_NUMA. There is no excuse for it.
I'm no lover of _PAGE_NUMA, and hope that it can be simplified away
as you outline. What we have in 3.16+3.17 is already an attempt to
improve on what you hated before, but not obviously an improvement.
Mel is the one who knows these issues inside out: maybe he's been
blinkered, but I wouldn't dare to pull it apart without his input.
Myself, I'm not looking beyond fixing whatever is the bug for 3.17.
> > However, that would still not explain Dave's endless refaulting;
>
> Why not? You start out with a PROTNONE, trigger shrink_page_list() on
> a hugepage,.which calls add_to_swap(), which does
> split_huge_page_to_list(), which in turn calls __split_huge_page(),
> and that turns (_PAGE_PROTNONE) into (_PAGE_PROTNONE|_PAGE_NUMA),
> which you will then fault on forever, because the kernel thinks the
> page is present, but not a NUMA page.
I hesitate to admit, I still don't see it: please illuminate further.
We're talking about the loop in __split_huge_page_map(), where it does
entry = mk_pte(page + i, vma->vm_page_prot);
entry = maybe_mkwrite(pte_mkdirty(entry), vma);
if (!pmd_write(*pmd))
entry = pte_wrprotect(entry);
if (!pmd_young(*pmd))
entry = pte_mkold(entry);
if (pmd_numa(*pmd))
entry = pte_mknuma(entry);
, right? I only see that adding _PAGE_NUMA to _PAGE_PROTNONE if
pmd_numa(*pmd): but that would mean we had already gone wrong, setting
pmd_numa in a PROT_NONE vma, which task_numa_work takes care not to do;
or have mprotected an area to PROT_NONE without doing the pmd_mknonnuma.
Ah, we won't have mmap_sem in the add_to_swap case; so we could be
racing with an mprotect which already updated vm_flags and vm_page_prot,
but has not yet reached this pmd: is that a part of how you see it?
Or are you noticing a deficiency in the pmd locking? I have not
worked my way through that, so cannot guarantee it, but please
point me to the weakness where you see it.
But when you convince me on that, then I still don't see how we get to
doing the repeated write fault, instead of hitting access_error(), as
you pointed out originally. That still seems to require a PROTNONE
pte to be left behind in a VM_WRITE vma, which I do not see happening
here. pte_modify leaves _PAGE_NUMA alone, but updates _PAGE_PROTNONE.
The pte_numa<->pte_special confusion is messy, but I don't yet get
how it would manifest in the manner observed.
But certainly, a bug in the THP splitting feels just right to
match the frequency of the sightings: I hope you've got it.
Hugh
next prev parent reply other threads:[~2014-10-02 8:49 UTC|newest]
Thread overview: 43+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-09-30 3:33 pipe/page fault oddness Dave Jones
2014-09-30 4:27 ` Linus Torvalds
2014-09-30 4:33 ` Dave Jones
[not found] ` <CA+55aFwxdOBKHwwp7Zq1k19mHCyHYmYqigCVt59AtB-P7Zva1w@mail.gmail.com>
2014-09-30 15:52 ` Linus Torvalds
2014-09-30 16:03 ` Rik van Riel
2014-09-30 16:07 ` Dave Jones
2014-09-30 16:26 ` Linus Torvalds
2014-09-30 16:05 ` Dave Jones
2014-09-30 16:10 ` Linus Torvalds
2014-09-30 16:22 ` Dave Jones
2014-09-30 16:40 ` Dave Jones
2014-09-30 16:46 ` Linus Torvalds
2014-09-30 18:20 ` Dave Jones
2014-09-30 18:58 ` Linus Torvalds
2014-10-01 8:19 ` Hugh Dickins
2014-10-01 16:01 ` Linus Torvalds
2014-10-01 16:18 ` Linus Torvalds
2014-10-01 17:29 ` Rik van Riel
2014-10-02 8:28 ` Peter Zijlstra
2014-10-01 20:20 ` Linus Torvalds
2014-10-01 21:09 ` Rik van Riel
2014-10-01 22:08 ` Sasha Levin
2014-10-01 22:28 ` Chuck Ebbert
2014-10-02 3:32 ` Sasha Levin
2014-10-02 8:03 ` Chuck Ebbert
2014-10-02 14:49 ` Sasha Levin
2014-10-01 22:42 ` Linus Torvalds
2014-10-02 14:25 ` Kirill A. Shutemov
2014-10-02 16:01 ` Linus Torvalds
2014-10-02 16:35 ` Kirill A. Shutemov
2014-10-02 15:04 ` Sasha Levin
2014-10-02 16:10 ` Linus Torvalds
2014-10-03 5:00 ` Sasha Levin
2014-10-03 15:43 ` Linus Torvalds
2014-10-03 15:58 ` Dave Jones
2014-10-03 16:02 ` Sasha Levin
2014-10-02 12:45 ` Mel Gorman
2014-10-06 19:18 ` Aneesh Kumar K.V
2014-10-07 12:45 ` Linus Torvalds
2014-10-08 10:37 ` Aneesh Kumar K.V
2014-10-02 8:47 ` Hugh Dickins [this message]
2014-10-02 15:57 ` Linus Torvalds
2014-09-30 4:35 ` Al Viro
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=alpine.LSU.2.11.1410020142230.6383@eggly.anvils \
--to=hughd@google.com \
--cc=davej@redhat.com \
--cc=kirill.shutemov@linux.intel.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mgorman@suse.de \
--cc=mingo@redhat.com \
--cc=riel@redhat.com \
--cc=sasha.levin@oracle.com \
--cc=torvalds@linux-foundation.org \
--cc=viro@zeniv.linux.org.uk \
--cc=walken@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).