linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
From: Nadav Amit <nadav.amit@gmail.com>
To: linux-mm@kvack.org
Cc: linux-kernel@vger.kernel.org,
	Andrew Morton <akpm@linux-foundation.org>,
	Mike Rapoport <rppt@linux.ibm.com>,
	Axel Rasmussen <axelrasmussen@google.com>,
	Nadav Amit <namit@vmware.com>,
	Andrea Arcangeli <aarcange@redhat.com>,
	Andrew Cooper <andrew.cooper3@citrix.com>,
	Andy Lutomirski <luto@kernel.org>,
	Dave Hansen <dave.hansen@linux.intel.com>,
	David Hildenbrand <david@redhat.com>,
	Peter Xu <peterx@redhat.com>,
	Peter Zijlstra <peterz@infradead.org>,
	Thomas Gleixner <tglx@linutronix.de>,
	Will Deacon <will@kernel.org>, Yu Zhao <yuzhao@google.com>,
	Nick Piggin <npiggin@gmail.com>
Subject: [RFC PATCH 14/14] mm: conditional check of pfn in pte_flush_type
Date: Mon, 18 Jul 2022 05:02:12 -0700	[thread overview]
Message-ID: <20220718120212.3180-15-namit@vmware.com> (raw)
In-Reply-To: <20220718120212.3180-1-namit@vmware.com>

From: Nadav Amit <namit@vmware.com>

Checking whether PFNs in two PTEs are the same takes surprisingly large
number of instructions. Yet in fact, in most cases the caller to
pte_flush_type() already knows if the PFN was changed. For instance,
mprotect() does not change the PFN, but only modifies the protection
flags.

Add argument to pte_flush_type() to indicate whether the PFN should be
checked. Keep checking it in mm-debug to see if some caller was wrong to
assume the PFN is the same.

Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Andrew Cooper <andrew.cooper3@citrix.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Will Deacon <will@kernel.org>
Cc: Yu Zhao <yuzhao@google.com>
Cc: Nick Piggin <npiggin@gmail.com>
Signed-off-by: Nadav Amit <namit@vmware.com>
---
 arch/x86/include/asm/tlbflush.h | 14 ++++++++++----
 include/asm-generic/tlb.h       |  6 ++++--
 mm/huge_memory.c                |  2 +-
 mm/mprotect.c                   |  2 +-
 mm/rmap.c                       |  2 +-
 5 files changed, 17 insertions(+), 9 deletions(-)

diff --git a/arch/x86/include/asm/tlbflush.h b/arch/x86/include/asm/tlbflush.h
index 58c95e36b098..50349861fdc9 100644
--- a/arch/x86/include/asm/tlbflush.h
+++ b/arch/x86/include/asm/tlbflush.h
@@ -340,14 +340,17 @@ static inline enum pte_flush_type pte_flags_flush_type(unsigned long oldflags,
  * whether a strict or relaxed TLB flush is need. It should only be used on
  * userspace PTEs.
  */
-static inline enum pte_flush_type pte_flush_type(pte_t oldpte, pte_t newpte)
+static inline enum pte_flush_type pte_flush_type(pte_t oldpte, pte_t newpte,
+						   bool check_pfn)
 {
 	/* !PRESENT -> * ; no need for flush */
 	if (!(pte_flags(oldpte) & _PAGE_PRESENT))
 		return PTE_FLUSH_NONE;
 
 	/* PFN changed ; needs flush */
-	if (pte_pfn(oldpte) != pte_pfn(newpte))
+	if (!check_pfn)
+		VM_BUG_ON(pte_pfn(oldpte) != pte_pfn(newpte));
+	else if (pte_pfn(oldpte) != pte_pfn(newpte))
 		return PTE_FLUSH_STRICT;
 
 	/*
@@ -363,14 +366,17 @@ static inline enum pte_flush_type pte_flush_type(pte_t oldpte, pte_t newpte)
  * huge_pmd_flush_type() checks whether permissions were demoted and require a
  * flush. It should only be used for userspace huge PMDs.
  */
-static inline enum pte_flush_type huge_pmd_flush_type(pmd_t oldpmd, pmd_t newpmd)
+static inline enum pte_flush_type huge_pmd_flush_type(pmd_t oldpmd, pmd_t newpmd,
+						      bool check_pfn)
 {
 	/* !PRESENT -> * ; no need for flush */
 	if (!(pmd_flags(oldpmd) & _PAGE_PRESENT))
 		return PTE_FLUSH_NONE;
 
 	/* PFN changed ; needs flush */
-	if (pmd_pfn(oldpmd) != pmd_pfn(newpmd))
+	if (!check_pfn)
+		VM_BUG_ON(pmd_pfn(oldpmd) != pmd_pfn(newpmd));
+	else if (pmd_pfn(oldpmd) != pmd_pfn(newpmd))
 		return PTE_FLUSH_STRICT;
 
 	/*
diff --git a/include/asm-generic/tlb.h b/include/asm-generic/tlb.h
index 07b3eb8caf63..aee9da6cc5d5 100644
--- a/include/asm-generic/tlb.h
+++ b/include/asm-generic/tlb.h
@@ -677,14 +677,16 @@ static inline void tlb_flush_p4d_range(struct mmu_gather *tlb,
 #endif
 
 #ifndef pte_flush_type
-static inline struct pte_flush_type pte_flush_type(pte_t oldpte, pte_t newpte)
+static inline struct pte_flush_type pte_flush_type(pte_t oldpte, pte_t newpte,
+						   bool check_pfn)
 {
 	return PTE_FLUSH_STRICT;
 }
 #endif
 
 #ifndef huge_pmd_flush_type
-static inline bool huge_pmd_flush_type(pmd_t oldpmd, pmd_t newpmd)
+static inline bool huge_pmd_flush_type(pmd_t oldpmd, pmd_t newpmd,
+				       bool check_pfn)
 {
 	return PTE_FLUSH_STRICT;
 }
diff --git a/mm/huge_memory.c b/mm/huge_memory.c
index b32b7da0f6f7..92a7b3ca317f 100644
--- a/mm/huge_memory.c
+++ b/mm/huge_memory.c
@@ -1818,7 +1818,7 @@ int change_huge_pmd(struct mmu_gather *tlb, struct vm_area_struct *vma,
 
 	flush_type = PTE_FLUSH_STRICT;
 	if (!tlb->strict)
-		flush_type = huge_pmd_flush_type(oldpmd, entry);
+		flush_type = huge_pmd_flush_type(oldpmd, entry, false);
 	if (flush_type != PTE_FLUSH_NONE)
 		tlb_flush_pmd_range(tlb, addr, HPAGE_PMD_SIZE,
 				    flush_type == PTE_FLUSH_STRICT);
diff --git a/mm/mprotect.c b/mm/mprotect.c
index cf775f6c8c08..78081d7f4edf 100644
--- a/mm/mprotect.c
+++ b/mm/mprotect.c
@@ -204,7 +204,7 @@ static unsigned long change_pte_range(struct mmu_gather *tlb,
 
 			flush_type = PTE_FLUSH_STRICT;
 			if (!tlb->strict)
-				flush_type = pte_flush_type(oldpte, ptent);
+				flush_type = pte_flush_type(oldpte, ptent, false);
 			if (flush_type != PTE_FLUSH_NONE)
 				tlb_flush_pte_range(tlb, addr, PAGE_SIZE,
 						flush_type == PTE_FLUSH_STRICT);
diff --git a/mm/rmap.c b/mm/rmap.c
index 62f4b2a4f067..63261619b607 100644
--- a/mm/rmap.c
+++ b/mm/rmap.c
@@ -974,7 +974,7 @@ static int page_vma_mkclean_one(struct page_vma_mapped_walk *pvmw)
 			entry = pte_wrprotect(oldpte);
 			entry = pte_mkclean(entry);
 
-			if (pte_flush_type(oldpte, entry) != PTE_FLUSH_NONE ||
+			if (pte_flush_type(oldpte, entry, false) != PTE_FLUSH_NONE ||
 			    mm_tlb_flush_pending(vma->vm_mm))
 				flush_tlb_page(vma, address);
 
-- 
2.25.1



      parent reply	other threads:[~2022-07-18 19:37 UTC|newest]

Thread overview: 33+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
     [not found] <20220718120212.3180-1-namit@vmware.com>
2022-07-18 12:01 ` [RFC PATCH 01/14] userfaultfd: set dirty and young on writeprotect Nadav Amit
2022-07-19 20:47   ` Peter Xu
2022-07-20  9:39     ` David Hildenbrand
2022-07-20 13:10       ` Peter Xu
2022-07-20 15:10         ` David Hildenbrand
2022-07-20 19:15           ` Peter Xu
2022-07-20 19:33             ` David Hildenbrand
2022-07-20 19:48               ` Peter Xu
2022-07-20 19:55                 ` David Hildenbrand
2022-07-20 20:22                   ` Nadav Amit
2022-07-20 20:38                     ` David Hildenbrand
2022-07-20 20:56                       ` Nadav Amit
2022-07-21  7:52                         ` David Hildenbrand
2022-07-21 14:10                           ` David Hildenbrand
2022-07-20  9:42   ` David Hildenbrand
2022-07-20 17:36     ` Nadav Amit
2022-07-20 18:00       ` David Hildenbrand
2022-07-20 18:09         ` Nadav Amit
2022-07-20 18:11           ` David Hildenbrand
2022-07-18 12:02 ` [RFC PATCH 02/14] userfaultfd: try to map write-unprotected pages Nadav Amit
2022-07-19 20:49   ` Peter Xu
2022-07-18 12:02 ` [RFC PATCH 03/14] mm/mprotect: allow exclusive anon pages to be writable Nadav Amit
2022-07-20 15:19   ` David Hildenbrand
2022-07-20 17:25     ` Nadav Amit
2022-07-21  7:45       ` David Hildenbrand
2022-07-18 12:02 ` [RFC PATCH 04/14] mm/mprotect: preserve write with MM_CP_TRY_CHANGE_WRITABLE Nadav Amit
2022-07-18 12:02 ` [RFC PATCH 06/14] mm/rmap: avoid flushing on page_vma_mkclean_one() when possible Nadav Amit
2022-07-18 12:02 ` [RFC PATCH 07/14] mm: do fix spurious page-faults for instruction faults Nadav Amit
2022-07-18 12:02 ` [RFC PATCH 08/14] x86/mm: introduce flush_tlb_fix_spurious_fault Nadav Amit
2022-07-18 12:02 ` [RFC PATCH 10/14] x86/mm: introduce relaxed TLB flushes Nadav Amit
2022-07-18 12:02 ` [RFC PATCH 11/14] x86/mm: use relaxed TLB flushes when protection is removed Nadav Amit
2022-07-18 12:02 ` [RFC PATCH 12/14] x86/tlb: no flush on PTE change from RW->RO when PTE is clean Nadav Amit
2022-07-18 12:02 ` Nadav Amit [this message]

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=20220718120212.3180-15-namit@vmware.com \
    --to=nadav.amit@gmail.com \
    --cc=aarcange@redhat.com \
    --cc=akpm@linux-foundation.org \
    --cc=andrew.cooper3@citrix.com \
    --cc=axelrasmussen@google.com \
    --cc=dave.hansen@linux.intel.com \
    --cc=david@redhat.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=luto@kernel.org \
    --cc=namit@vmware.com \
    --cc=npiggin@gmail.com \
    --cc=peterx@redhat.com \
    --cc=peterz@infradead.org \
    --cc=rppt@linux.ibm.com \
    --cc=tglx@linutronix.de \
    --cc=will@kernel.org \
    --cc=yuzhao@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