Linux-Next Archive on lore.kernel.org
 help / color / Atom feed
From: Peter Zijlstra <peterz@infradead.org>
To: Ingo Molnar <mingo@kernel.org>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>,
	Andrew Morton <akpm@linux-foundation.org>,
	Thomas Gleixner <tglx@linutronix.de>, Ingo Molnar <mingo@elte.hu>,
	"H. Peter Anvin" <hpa@zytor.com>,
	Linux-Next Mailing List <linux-next@vger.kernel.org>,
	Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
	Nadav Amit <namit@vmware.com>,
	Linus <torvalds@linux-foundation.org>,
	minchan@kernel.org
Subject: Re: linux-next: manual merge of the akpm-current tree with the tip tree
Date: Fri, 11 Aug 2017 16:04:50 +0200
Message-ID: <20170811140450.irhxa2bhdpmmhhpv@hirez.programming.kicks-ass.net> (raw)
In-Reply-To: <20170811115607.p2vgqcp7w3wurhvw@gmail.com>


Ok, so I have the below to still go on-top.

Ideally someone would clarify the situation around
mm_tlb_flush_nested(), because ideally we'd remove the
smp_mb__after_atomic() and go back to relying on PTL alone.

This also removes the pointless smp_mb__before_atomic()

---
Subject: mm: Fix barriers for the tlb_flush_pending thing
From: Peter Zijlstra <peterz@infradead.org>
Date: Fri Aug 11 12:43:33 CEST 2017

I'm not 100% sure we always care about the same PTL and when we have
SPLIT_PTE_PTLOCKS and have RCpc locks (PPC) the UNLOCK of one does not
in fact order against the LOCK of another lock. Therefore the
documented scheme does not work if we care about multiple PTLs

mm_tlb_flush_pending() appears to only care about a single PTL:

 - arch pte_accessible() (x86, arm64) only cares about that one PTE.
 - do_huge_pmd_numa_page() also only cares about a single (huge) page.
 - ksm write_protect_page() also only cares about a single page.

however mm_tlb_flush_nested() is a mystery, it appears to care about
anything inside the range. For now rely on it doing at least _a_ PTL
lock instead of taking  _the_ PTL lock.

Therefore add an explicit smp_mb__after_atomic() to cure things.

Also remove the smp_mb__before_atomic() on the dec side, as its
completely pointless. We must rely on flush_tlb_range() to DTRT.

Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
---
 include/linux/mm_types.h |   38 ++++++++++++++++++++++----------------
 1 file changed, 22 insertions(+), 16 deletions(-)

--- a/include/linux/mm_types.h
+++ b/include/linux/mm_types.h
@@ -537,13 +537,13 @@ static inline bool mm_tlb_flush_pending(
 {
 	/*
 	 * Must be called with PTL held; such that our PTL acquire will have
-	 * observed the store from set_tlb_flush_pending().
+	 * observed the increment from inc_tlb_flush_pending().
 	 */
-	return atomic_read(&mm->tlb_flush_pending) > 0;
+	return atomic_read(&mm->tlb_flush_pending);
 }
 
 /*
- * Returns true if there are two above TLB batching threads in parallel.
+ * Returns true if there are two or more TLB batching threads in parallel.
  */
 static inline bool mm_tlb_flush_nested(struct mm_struct *mm)
 {
@@ -558,15 +558,12 @@ static inline void init_tlb_flush_pendin
 static inline void inc_tlb_flush_pending(struct mm_struct *mm)
 {
 	atomic_inc(&mm->tlb_flush_pending);
-
 	/*
-	 * The only time this value is relevant is when there are indeed pages
-	 * to flush. And we'll only flush pages after changing them, which
-	 * requires the PTL.
-	 *
 	 * So the ordering here is:
 	 *
 	 *	atomic_inc(&mm->tlb_flush_pending);
+	 *	smp_mb__after_atomic();
+	 *
 	 *	spin_lock(&ptl);
 	 *	...
 	 *	set_pte_at();
@@ -580,21 +577,30 @@ static inline void inc_tlb_flush_pending
 	 *	flush_tlb_range();
 	 *	atomic_dec(&mm->tlb_flush_pending);
 	 *
-	 * So the =true store is constrained by the PTL unlock, and the =false
-	 * store is constrained by the TLB invalidate.
+	 * Where we order the increment against the PTE modification with the
+	 * smp_mb__after_atomic(). It would appear that the spin_unlock(&ptl)
+	 * is sufficient to constrain the inc, because we only care about the
+	 * value if there is indeed a pending PTE modification. However with
+	 * SPLIT_PTE_PTLOCKS and RCpc locks (PPC) the UNLOCK of one lock does
+	 * not order against the LOCK of another lock.
+	 *
+	 * The decrement is ordered by the flush_tlb_range(), such that
+	 * mm_tlb_flush_pending() will not return false unless all flushes have
+	 * completed.
 	 */
+	smp_mb__after_atomic();
 }
 
-/* Clearing is done after a TLB flush, which also provides a barrier. */
 static inline void dec_tlb_flush_pending(struct mm_struct *mm)
 {
 	/*
-	 * Guarantee that the tlb_flush_pending does not not leak into the
-	 * critical section, since we must order the PTE change and changes to
-	 * the pending TLB flush indication. We could have relied on TLB flush
-	 * as a memory barrier, but this behavior is not clearly documented.
+	 * See inc_tlb_flush_pending().
+	 *
+	 * This cannot be smp_mb__before_atomic() because smp_mb() simply does
+	 * not order against TLB invalidate completion, which is what we need.
+	 *
+	 * Therefore we must rely on tlb_flush_*() to guarantee order.
 	 */
-	smp_mb__before_atomic();
 	atomic_dec(&mm->tlb_flush_pending);
 }
 

  parent reply index

Thread overview: 87+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-08-11  7:53 Stephen Rothwell
2017-08-11  9:34 ` Peter Zijlstra
2017-08-11 10:48   ` Peter Zijlstra
2017-08-11 11:45   ` Stephen Rothwell
2017-08-11 11:56     ` Ingo Molnar
2017-08-11 12:17       ` Peter Zijlstra
2017-08-11 12:44         ` Ingo Molnar
2017-08-11 13:49           ` Stephen Rothwell
2017-08-11 14:04       ` Peter Zijlstra [this message]
2017-08-13  6:06         ` Nadav Amit
2017-08-13 12:50           ` Peter Zijlstra
2017-08-14  3:16             ` Minchan Kim
2017-08-14  5:07               ` Nadav Amit
2017-08-14  5:23                 ` Minchan Kim
2017-08-14  8:38                 ` Minchan Kim
2017-08-14 19:57                   ` Peter Zijlstra
2017-08-16  4:14                     ` Minchan Kim
2017-08-14 19:38                 ` Peter Zijlstra
2017-08-15  7:51                   ` Nadav Amit
2017-08-14  3:09         ` Minchan Kim
2017-08-14 18:54           ` Peter Zijlstra
  -- strict thread matches above, loose matches on Subject: below --
2019-10-31  5:43 Stephen Rothwell
2019-06-24 10:24 Stephen Rothwell
2019-05-01 11:10 Stephen Rothwell
2019-01-31  4:31 Stephen Rothwell
2018-08-20  4:32 Stephen Rothwell
2018-08-20 19:52 ` Andrew Morton
2018-03-23  5:59 Stephen Rothwell
2017-12-18  5:04 Stephen Rothwell
2017-11-10  4:33 Stephen Rothwell
2017-11-02  7:19 Stephen Rothwell
2017-08-22  6:57 Stephen Rothwell
2017-08-23  6:39 ` Vlastimil Babka
2017-04-12  6:46 Stephen Rothwell
2017-04-12 20:53 ` Vlastimil Babka
2017-04-20  2:17   ` NeilBrown
2017-03-24  5:25 Stephen Rothwell
2017-02-17  4:40 Stephen Rothwell
2016-11-14  6:08 Stephen Rothwell
2016-07-29  4:14 Stephen Rothwell
2016-06-15  5:23 Stephen Rothwell
2016-06-18 19:39 ` Manfred Spraul
2016-04-29  6:12 Stephen Rothwell
2016-04-29  6:26 ` Ingo Molnar
2016-03-02  5:40 Stephen Rothwell
2016-02-26  5:07 Stephen Rothwell
2016-02-26 21:35 ` Andrew Morton
2016-02-19  4:09 Stephen Rothwell
2016-02-19 15:26 ` Ard Biesheuvel
2015-12-07  8:06 Stephen Rothwell
2015-10-02  4:21 Stephen Rothwell
2015-07-28  6:00 Stephen Rothwell
2015-07-29 17:12 ` Andrea Arcangeli
2015-07-29 17:47   ` Andy Lutomirski
2015-07-29 18:46     ` Thomas Gleixner
2015-07-30 15:38       ` Andrea Arcangeli
2015-07-29 23:06   ` Stephen Rothwell
2015-07-29 23:07     ` Thomas Gleixner
2015-09-07 23:35   ` Stephen Rothwell
2015-09-08 18:11     ` Linus Torvalds
2015-09-08 22:56       ` Stephen Rothwell
2015-09-08 23:03         ` Linus Torvalds
2015-09-08 23:21           ` Andrew Morton
2015-09-16  6:58             ` Geert Uytterhoeven
2015-06-04 12:07 Stephen Rothwell
2015-04-08  8:28 Stephen Rothwell
2015-04-08  8:25 Stephen Rothwell
2014-03-17  9:31 Stephen Rothwell
2014-03-17  9:36 ` Peter Zijlstra
2014-03-19 23:27   ` Andrew Morton
2014-01-14  4:53 Stephen Rothwell
2014-01-14  5:04 ` Davidlohr Bueso
2014-01-14 12:51 ` Peter Zijlstra
2014-01-14 13:17   ` Geert Uytterhoeven
2014-01-14 13:33     ` Peter Zijlstra
2014-01-14 16:19     ` H. Peter Anvin
2014-01-14 15:15   ` H. Peter Anvin
2014-01-14 15:20     ` Geert Uytterhoeven
2014-01-14 15:41       ` Peter Zijlstra
2014-01-14 15:48         ` H. Peter Anvin
2014-01-07  6:00 Stephen Rothwell
2014-01-07  6:34 ` Tang Chen
2013-11-08  7:48 Stephen Rothwell
2013-11-08 18:58 ` Josh Triplett
2013-11-08 23:20   ` Stephen Rothwell
2013-11-09  0:19     ` Josh Triplett
2013-10-30  6:40 Stephen Rothwell

Reply instructions:

You may reply publically 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=20170811140450.irhxa2bhdpmmhhpv@hirez.programming.kicks-ass.net \
    --to=peterz@infradead.org \
    --cc=akpm@linux-foundation.org \
    --cc=hpa@zytor.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-next@vger.kernel.org \
    --cc=minchan@kernel.org \
    --cc=mingo@elte.hu \
    --cc=mingo@kernel.org \
    --cc=namit@vmware.com \
    --cc=sfr@canb.auug.org.au \
    --cc=tglx@linutronix.de \
    --cc=torvalds@linux-foundation.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

Linux-Next Archive on lore.kernel.org

Archives are clonable:
	git clone --mirror https://lore.kernel.org/linux-next/0 linux-next/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 linux-next linux-next/ https://lore.kernel.org/linux-next \
		linux-next@vger.kernel.org
	public-inbox-index linux-next

Example config snippet for mirrors

Newsgroup available over NNTP:
	nntp://nntp.lore.kernel.org/org.kernel.vger.linux-next


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git