All of lore.kernel.org
 help / color / mirror / Atom feed
From: Paolo Bonzini <pbonzini@redhat.com>
To: Andres Lagar-Cavilla <andreslc@google.com>
Cc: Gleb Natapov <gleb@redhat.com>, Rik van Riel <riel@redhat.com>,
	Peter Zijlstra <peterz@infradead.org>,
	Mel Gorman <mgorman@suse.de>,
	Andy Lutomirski <luto@amacapital.net>,
	Andrew Morton <akpm@linux-foundation.org>,
	Andrea Arcangeli <aarcange@redhat.com>,
	Sasha Levin <sasha.levin@oracle.com>,
	Jianyu Zhan <nasa4836@gmail.com>,
	Paul Cassella <cassella@cray.com>,
	Hugh Dickins <hughd@google.com>,
	Peter Feiner <pfeiner@google.com>,
	kvm@vger.kernel.org, linux-kernel@vger.kernel.org,
	linux-mm@kvack.org
Subject: Re: [PATCH] kvm: Faults which trigger IO release the mmap_sem
Date: Wed, 17 Sep 2014 09:43:46 +0200	[thread overview]
Message-ID: <54193BB2.8010500@redhat.com> (raw)
In-Reply-To: <CAJu=L58z-=_KkZXpEiPjDUup8GpH7079HH39csmvgUxGkvXy0A@mail.gmail.com>

Il 16/09/2014 20:42, Andres Lagar-Cavilla ha scritto:
> On Tue, Sep 16, 2014 at 11:29 AM, Paolo Bonzini <pbonzini@redhat.com> wrote:
>>>     I think a first patch should introduce kvm_get_user_page_retry ("Retry a
>>>     fault after a gup with FOLL_NOWAIT.") and the second would add
>>>     FOLL_TRIED ("This properly relinquishes mmap semaphore if the
>>>     filemap/swap has to wait on page lock (and retries the gup to completion
>>>     after that").
>>>
>>> That's not what FOLL_TRIED does. The relinquishing of mmap semaphore is
>>> done by this patch minus the FOLL_TRIED bits. FOLL_TRIED will let the
>>> fault handler (e.g. filemap) know that we've been there and waited on
>>> the IO already, so in the common case we won't need to redo the IO.
>>
>> Yes, that's not what FOLL_TRIED does.  But it's the difference between
>> get_user_pages and kvm_get_user_page_retry, right?
> 
> Unfortunately get_user_pages does not expose the param (int
> *nonblocking) that __gup will use to set FAULT_FLAG_ALLOW_RETRY. So
> that's one difference. The second difference is that kvm_gup_retry
> will call two times if necessary (the second without _RETRY but with
> _TRIED).

Yeah, that's how it is in your patch.  I can see that.

What I'm saying is that your patch is two changes in one:

1) do not use gup_fast in hva_to_pfn_slow, instead use gup as in
async_pf_execute.  This change can already introduce a function called
kvm_get_user_page_retry, and can already use it in async_pf_execute and
hva_to_pfn_slow

2) introduce the two-phase RETRY + TRIED mechanism in
kvm_get_user_page_retry, so that the mmap semaphore is relinquished
properly if the filemap or swap has to wait on the page lock.

I would prefer to split it in two patches.  Is it clearer now?

Paolo

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

WARNING: multiple messages have this Message-ID (diff)
From: Paolo Bonzini <pbonzini@redhat.com>
To: Andres Lagar-Cavilla <andreslc@google.com>
Cc: Gleb Natapov <gleb@redhat.com>, Rik van Riel <riel@redhat.com>,
	Peter Zijlstra <peterz@infradead.org>,
	Mel Gorman <mgorman@suse.de>,
	Andy Lutomirski <luto@amacapital.net>,
	Andrew Morton <akpm@linux-foundation.org>,
	Andrea Arcangeli <aarcange@redhat.com>,
	Sasha Levin <sasha.levin@oracle.com>,
	Jianyu Zhan <nasa4836@gmail.com>,
	Paul Cassella <cassella@cray.com>,
	Hugh Dickins <hughd@google.com>,
	Peter Feiner <pfeiner@google.com>,
	kvm@vger.kernel.org, linux-kernel@vger.kernel.org,
	linux-mm@kvack.org
Subject: Re: [PATCH] kvm: Faults which trigger IO release the mmap_sem
Date: Wed, 17 Sep 2014 09:43:46 +0200	[thread overview]
Message-ID: <54193BB2.8010500@redhat.com> (raw)
In-Reply-To: <CAJu=L58z-=_KkZXpEiPjDUup8GpH7079HH39csmvgUxGkvXy0A@mail.gmail.com>

Il 16/09/2014 20:42, Andres Lagar-Cavilla ha scritto:
> On Tue, Sep 16, 2014 at 11:29 AM, Paolo Bonzini <pbonzini@redhat.com> wrote:
>>>     I think a first patch should introduce kvm_get_user_page_retry ("Retry a
>>>     fault after a gup with FOLL_NOWAIT.") and the second would add
>>>     FOLL_TRIED ("This properly relinquishes mmap semaphore if the
>>>     filemap/swap has to wait on page lock (and retries the gup to completion
>>>     after that").
>>>
>>> That's not what FOLL_TRIED does. The relinquishing of mmap semaphore is
>>> done by this patch minus the FOLL_TRIED bits. FOLL_TRIED will let the
>>> fault handler (e.g. filemap) know that we've been there and waited on
>>> the IO already, so in the common case we won't need to redo the IO.
>>
>> Yes, that's not what FOLL_TRIED does.  But it's the difference between
>> get_user_pages and kvm_get_user_page_retry, right?
> 
> Unfortunately get_user_pages does not expose the param (int
> *nonblocking) that __gup will use to set FAULT_FLAG_ALLOW_RETRY. So
> that's one difference. The second difference is that kvm_gup_retry
> will call two times if necessary (the second without _RETRY but with
> _TRIED).

Yeah, that's how it is in your patch.  I can see that.

What I'm saying is that your patch is two changes in one:

1) do not use gup_fast in hva_to_pfn_slow, instead use gup as in
async_pf_execute.  This change can already introduce a function called
kvm_get_user_page_retry, and can already use it in async_pf_execute and
hva_to_pfn_slow

2) introduce the two-phase RETRY + TRIED mechanism in
kvm_get_user_page_retry, so that the mmap semaphore is relinquished
properly if the filemap or swap has to wait on the page lock.

I would prefer to split it in two patches.  Is it clearer now?

Paolo

  reply	other threads:[~2014-09-17  7:43 UTC|newest]

Thread overview: 70+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-09-15 20:11 [PATCH] kvm: Faults which trigger IO release the mmap_sem Andres Lagar-Cavilla
2014-09-15 20:11 ` Andres Lagar-Cavilla
2014-09-16 13:51 ` Paolo Bonzini
2014-09-16 13:51   ` Paolo Bonzini
2014-09-16 16:52   ` Andres Lagar-Cavilla
2014-09-16 16:55     ` Andres Lagar-Cavilla
2014-09-16 16:55       ` Andres Lagar-Cavilla
2014-09-16 18:29     ` Paolo Bonzini
2014-09-16 18:29       ` Paolo Bonzini
2014-09-16 18:42       ` Andres Lagar-Cavilla
2014-09-16 18:42         ` Andres Lagar-Cavilla
2014-09-17  7:43         ` Paolo Bonzini [this message]
2014-09-17  7:43           ` Paolo Bonzini
2014-09-17 16:58           ` Andres Lagar-Cavilla
2014-09-17 16:58             ` Andres Lagar-Cavilla
2014-09-17 20:01             ` Paolo Bonzini
2014-09-17 20:01               ` Paolo Bonzini
2014-09-16 20:51   ` Radim Krčmář
2014-09-16 20:51     ` Radim Krčmář
2014-09-16 20:51     ` Radim Krčmář
2014-09-16 21:01     ` Andres Lagar-Cavilla
2014-09-16 21:01       ` Andres Lagar-Cavilla
2014-09-16 22:34       ` Radim Krčmář
2014-09-16 22:34         ` Radim Krčmář
2014-09-16 22:34         ` Radim Krčmář
2014-09-17  4:15         ` Andres Lagar-Cavilla
2014-09-17  4:15           ` Andres Lagar-Cavilla
2014-09-17 11:35       ` Radim Krčmář
2014-09-17 11:35         ` Radim Krčmář
2014-09-17 11:35         ` Radim Krčmář
2014-09-17 10:26 ` Gleb Natapov
2014-09-17 10:26   ` Gleb Natapov
2014-09-17 11:27   ` Radim Krčmář
2014-09-17 11:27     ` Radim Krčmář
2014-09-17 11:42     ` Gleb Natapov
2014-09-17 11:42       ` Gleb Natapov
2014-09-17 17:00       ` Andres Lagar-Cavilla
2014-09-17 17:00         ` Andres Lagar-Cavilla
2014-09-17 17:08         ` Gleb Natapov
2014-09-17 17:08           ` Gleb Natapov
2014-09-17 17:13           ` Andres Lagar-Cavilla
2014-09-17 17:13             ` Andres Lagar-Cavilla
2014-09-17 17:21             ` Gleb Natapov
2014-09-17 17:21               ` Gleb Natapov
2014-09-17 17:41               ` Andres Lagar-Cavilla
2014-09-17 17:41                 ` Andres Lagar-Cavilla
2014-09-17 17:51 ` [PATCH v2] " Andres Lagar-Cavilla
2014-09-17 17:51   ` Andres Lagar-Cavilla
2014-09-18  0:29   ` Wanpeng Li
2014-09-18  0:29     ` Wanpeng Li
2014-09-18  6:13     ` Gleb Natapov
2014-09-18  6:13       ` Gleb Natapov
2014-09-19  0:32       ` Wanpeng Li
2014-09-19  0:32         ` Wanpeng Li
2014-09-19  3:58         ` Andres Lagar-Cavilla
2014-09-19  3:58           ` Andres Lagar-Cavilla
2014-09-19  6:08           ` Paolo Bonzini
2014-09-19  6:08             ` Paolo Bonzini
2014-09-22 20:49             ` Andres Lagar-Cavilla
2014-09-22 20:49               ` Andres Lagar-Cavilla
2014-09-22 21:32               ` Paolo Bonzini
2014-09-22 21:32                 ` Paolo Bonzini
2014-09-22 21:53                 ` Andrew Morton
2014-09-22 21:53                   ` Andrew Morton
2014-09-18  6:15   ` Gleb Natapov
2014-09-18  6:15     ` Gleb Natapov
2014-09-25 21:16   ` Andrea Arcangeli
2014-09-25 21:16     ` Andrea Arcangeli
2014-09-25 21:50     ` Andres Lagar-Cavilla
2014-09-25 21:50       ` Andres Lagar-Cavilla

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=54193BB2.8010500@redhat.com \
    --to=pbonzini@redhat.com \
    --cc=aarcange@redhat.com \
    --cc=akpm@linux-foundation.org \
    --cc=andreslc@google.com \
    --cc=cassella@cray.com \
    --cc=gleb@redhat.com \
    --cc=hughd@google.com \
    --cc=kvm@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=luto@amacapital.net \
    --cc=mgorman@suse.de \
    --cc=nasa4836@gmail.com \
    --cc=peterz@infradead.org \
    --cc=pfeiner@google.com \
    --cc=riel@redhat.com \
    --cc=sasha.levin@oracle.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.