All of lore.kernel.org
 help / color / mirror / Atom feed
From: Dan Williams <dan.j.williams@intel.com>
To: akpm@linux-foundation.org
Cc: Andrea Arcangeli <aarcange@redhat.com>,
	Dave Hansen <dave@sr71.net>,
	linux-nvdimm@lists.01.org, Peter Zijlstra <peterz@infradead.org>,
	linux-mm@kvack.org, Mel Gorman <mgorman@suse.de>,
	Sasha Levin <sasha.levin@oracle.com>,
	Matthew Wilcox <willy@linux.intel.com>,
	"Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Subject: [-mm PATCH v5 15/18] mm, dax: dax-pmd vs thp-pmd vs hugetlbfs-pmd
Date: Thu, 24 Dec 2015 16:59:28 -0800	[thread overview]
Message-ID: <20151225005623.19962.1972.stgit@dwillia2-desk3.jf.intel.com> (raw)
In-Reply-To: <20151221054526.34542.25205.stgit@dwillia2-desk3.jf.intel.com>

A dax-huge-page mapping while it uses some thp helpers is ultimately not a
transparent huge page.  The distinction is especially important in the
get_user_pages() path.  pmd_devmap() is used to distinguish dax-pmds from
pmd_huge() and pmd_trans_huge() which have slightly different semantics.

Explicitly mark the pmd_trans_huge() helpers that dax needs by adding
pmd_devmap() checks.

Cc: Dave Hansen <dave@sr71.net>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Reported-by: Matthew Wilcox <willy@linux.intel.com>
Reported-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Reported-by: Sasha Levin <sasha.levin@oracle.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
---

Changes since v4:
1/ Fold in a __split_huge_pmd() regression fix (Kirill)

2/ Add a missing pmd_devmap() check to pmd_trans_huge_lock() (Willy)

 arch/x86/include/asm/pgtable.h |    9 ++++++++-
 include/linux/huge_mm.h        |    5 +++--
 include/linux/mm.h             |    7 +++++++
 mm/huge_memory.c               |   38 +++++++++++++++++++++-----------------
 mm/memory.c                    |    8 ++++----
 mm/mprotect.c                  |    5 +++--
 mm/pgtable-generic.c           |    2 +-
 7 files changed, 47 insertions(+), 27 deletions(-)

diff --git a/arch/x86/include/asm/pgtable.h b/arch/x86/include/asm/pgtable.h
index dc962ae41597..993ce3c84ff4 100644
--- a/arch/x86/include/asm/pgtable.h
+++ b/arch/x86/include/asm/pgtable.h
@@ -167,13 +167,20 @@ static inline int pmd_large(pmd_t pte)
 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
 static inline int pmd_trans_huge(pmd_t pmd)
 {
-	return pmd_val(pmd) & _PAGE_PSE;
+	return (pmd_val(pmd) & (_PAGE_PSE|_PAGE_DEVMAP)) == _PAGE_PSE;
 }
 
 static inline int has_transparent_hugepage(void)
 {
 	return cpu_has_pse;
 }
+
+#ifdef __HAVE_ARCH_PTE_DEVMAP
+static inline int pmd_devmap(pmd_t pmd)
+{
+	return !!(pmd_val(pmd) & _PAGE_DEVMAP);
+}
+#endif
 #endif /* CONFIG_TRANSPARENT_HUGEPAGE */
 
 static inline pte_t pte_set_flags(pte_t pte, pteval_t set)
diff --git a/include/linux/huge_mm.h b/include/linux/huge_mm.h
index 8ca35a131904..d39fa60bd6bf 100644
--- a/include/linux/huge_mm.h
+++ b/include/linux/huge_mm.h
@@ -104,7 +104,8 @@ void __split_huge_pmd(struct vm_area_struct *vma, pmd_t *pmd,
 #define split_huge_pmd(__vma, __pmd, __address)				\
 	do {								\
 		pmd_t *____pmd = (__pmd);				\
-		if (pmd_trans_huge(*____pmd))				\
+		if (pmd_trans_huge(*____pmd)				\
+					|| pmd_devmap(*____pmd))	\
 			__split_huge_pmd(__vma, __pmd, __address);	\
 	}  while (0)
 
@@ -124,7 +125,7 @@ static inline bool pmd_trans_huge_lock(pmd_t *pmd, struct vm_area_struct *vma,
 		spinlock_t **ptl)
 {
 	VM_BUG_ON_VMA(!rwsem_is_locked(&vma->vm_mm->mmap_sem), vma);
-	if (pmd_trans_huge(*pmd))
+	if (pmd_trans_huge(*pmd) || pmd_devmap(*pmd))
 		return __pmd_trans_huge_lock(pmd, vma, ptl);
 	else
 		return false;
diff --git a/include/linux/mm.h b/include/linux/mm.h
index 957afd1b10a5..96f396bbcc9f 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -1471,6 +1471,13 @@ static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd,
 int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
 #endif
 
+#if !defined(__HAVE_ARCH_PTE_DEVMAP) || !defined(CONFIG_TRANSPARENT_HUGEPAGE)
+static inline int pmd_devmap(pmd_t pmd)
+{
+	return 0;
+}
+#endif
+
 #if defined(__PAGETABLE_PMD_FOLDED) || !defined(CONFIG_MMU)
 static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
 						unsigned long address)
diff --git a/mm/huge_memory.c b/mm/huge_memory.c
index 7356857d7356..4521bec67364 100644
--- a/mm/huge_memory.c
+++ b/mm/huge_memory.c
@@ -1024,7 +1024,7 @@ int copy_huge_pmd(struct mm_struct *dst_mm, struct mm_struct *src_mm,
 
 	ret = -EAGAIN;
 	pmd = *src_pmd;
-	if (unlikely(!pmd_trans_huge(pmd))) {
+	if (unlikely(!pmd_trans_huge(pmd) && !pmd_devmap(pmd))) {
 		pte_free(dst_mm, pgtable);
 		goto out_unlock;
 	}
@@ -1047,17 +1047,20 @@ int copy_huge_pmd(struct mm_struct *dst_mm, struct mm_struct *src_mm,
 		goto out_unlock;
 	}
 
-	src_page = pmd_page(pmd);
-	VM_BUG_ON_PAGE(!PageHead(src_page), src_page);
-	get_page(src_page);
-	page_dup_rmap(src_page, true);
-	add_mm_counter(dst_mm, MM_ANONPAGES, HPAGE_PMD_NR);
+	if (pmd_trans_huge(pmd)) {
+		/* thp accounting separate from pmd_devmap accounting */
+		src_page = pmd_page(pmd);
+		VM_BUG_ON_PAGE(!PageHead(src_page), src_page);
+		get_page(src_page);
+		page_dup_rmap(src_page, true);
+		add_mm_counter(dst_mm, MM_ANONPAGES, HPAGE_PMD_NR);
+		atomic_long_inc(&dst_mm->nr_ptes);
+		pgtable_trans_huge_deposit(dst_mm, dst_pmd, pgtable);
+	}
 
 	pmdp_set_wrprotect(src_mm, addr, src_pmd);
 	pmd = pmd_mkold(pmd_wrprotect(pmd));
-	pgtable_trans_huge_deposit(dst_mm, dst_pmd, pgtable);
 	set_pmd_at(dst_mm, addr, dst_pmd, pmd);
-	atomic_long_inc(&dst_mm->nr_ptes);
 
 	ret = 0;
 out_unlock:
@@ -1745,7 +1748,7 @@ bool __pmd_trans_huge_lock(pmd_t *pmd, struct vm_area_struct *vma,
 		spinlock_t **ptl)
 {
 	*ptl = pmd_lock(vma->vm_mm, pmd);
-	if (likely(pmd_trans_huge(*pmd)))
+	if (likely(pmd_trans_huge(*pmd) || pmd_devmap(*pmd)))
 		return true;
 	spin_unlock(*ptl);
 	return false;
@@ -2862,7 +2865,7 @@ static void __split_huge_pmd_locked(struct vm_area_struct *vma, pmd_t *pmd,
 	VM_BUG_ON(haddr & ~HPAGE_PMD_MASK);
 	VM_BUG_ON_VMA(vma->vm_start > haddr, vma);
 	VM_BUG_ON_VMA(vma->vm_end < haddr + HPAGE_PMD_SIZE, vma);
-	VM_BUG_ON(!pmd_trans_huge(*pmd));
+	VM_BUG_ON(!pmd_trans_huge(*pmd) && !pmd_devmap(*pmd));
 
 	count_vm_event(THP_SPLIT_PMD);
 
@@ -2975,14 +2978,15 @@ void __split_huge_pmd(struct vm_area_struct *vma, pmd_t *pmd,
 
 	mmu_notifier_invalidate_range_start(mm, haddr, haddr + HPAGE_PMD_SIZE);
 	ptl = pmd_lock(mm, pmd);
-	if (unlikely(!pmd_trans_huge(*pmd)))
+	if (pmd_trans_huge(*pmd)) {
+		page = pmd_page(*pmd);
+		if (PageMlocked(page))
+			get_page(page);
+		else
+			page = NULL;
+	} else if (!pmd_devmap(*pmd))
 		goto out;
-	page = pmd_page(*pmd);
 	__split_huge_pmd_locked(vma, pmd, haddr, false);
-	if (PageMlocked(page))
-		get_page(page);
-	else
-		page = NULL;
 out:
 	spin_unlock(ptl);
 	mmu_notifier_invalidate_range_end(mm, haddr, haddr + HPAGE_PMD_SIZE);
@@ -3012,7 +3016,7 @@ static void split_huge_pmd_address(struct vm_area_struct *vma,
 		return;
 
 	pmd = pmd_offset(pud, address);
-	if (!pmd_present(*pmd) || !pmd_trans_huge(*pmd))
+	if (!pmd_present(*pmd) || (!pmd_trans_huge(*pmd) && !pmd_devmap(*pmd)))
 		return;
 	/*
 	 * Caller holds the mmap_sem write mode, so a huge pmd cannot
diff --git a/mm/memory.c b/mm/memory.c
index 9483d2b1dd3b..03b6fa406a28 100644
--- a/mm/memory.c
+++ b/mm/memory.c
@@ -950,7 +950,7 @@ static inline int copy_pmd_range(struct mm_struct *dst_mm, struct mm_struct *src
 	src_pmd = pmd_offset(src_pud, addr);
 	do {
 		next = pmd_addr_end(addr, end);
-		if (pmd_trans_huge(*src_pmd)) {
+		if (pmd_trans_huge(*src_pmd) || pmd_devmap(*src_pmd)) {
 			int err;
 			VM_BUG_ON(next-addr != HPAGE_PMD_SIZE);
 			err = copy_huge_pmd(dst_mm, src_mm,
@@ -1177,7 +1177,7 @@ static inline unsigned long zap_pmd_range(struct mmu_gather *tlb,
 	pmd = pmd_offset(pud, addr);
 	do {
 		next = pmd_addr_end(addr, end);
-		if (pmd_trans_huge(*pmd)) {
+		if (pmd_trans_huge(*pmd) || pmd_devmap(*pmd)) {
 			if (next - addr != HPAGE_PMD_SIZE) {
 #ifdef CONFIG_DEBUG_VM
 				if (!rwsem_is_locked(&tlb->mm->mmap_sem)) {
@@ -3375,7 +3375,7 @@ static int __handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
 		int ret;
 
 		barrier();
-		if (pmd_trans_huge(orig_pmd)) {
+		if (pmd_trans_huge(orig_pmd) || pmd_devmap(orig_pmd)) {
 			unsigned int dirty = flags & FAULT_FLAG_WRITE;
 
 			if (pmd_protnone(orig_pmd))
@@ -3404,7 +3404,7 @@ static int __handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
 	    unlikely(__pte_alloc(mm, vma, pmd, address)))
 		return VM_FAULT_OOM;
 	/* if an huge pmd materialized from under us just retry later */
-	if (unlikely(pmd_trans_huge(*pmd)))
+	if (unlikely(pmd_trans_huge(*pmd) || pmd_devmap(*pmd)))
 		return 0;
 	/*
 	 * A regular pmd is established and it can't morph into a huge pmd
diff --git a/mm/mprotect.c b/mm/mprotect.c
index 9c1445dc8a4c..732e07baf76c 100644
--- a/mm/mprotect.c
+++ b/mm/mprotect.c
@@ -149,7 +149,8 @@ static inline unsigned long change_pmd_range(struct vm_area_struct *vma,
 		unsigned long this_pages;
 
 		next = pmd_addr_end(addr, end);
-		if (!pmd_trans_huge(*pmd) && pmd_none_or_clear_bad(pmd))
+		if (!pmd_trans_huge(*pmd) && !pmd_devmap(*pmd)
+				&& pmd_none_or_clear_bad(pmd))
 			continue;
 
 		/* invoke the mmu notifier if the pmd is populated */
@@ -158,7 +159,7 @@ static inline unsigned long change_pmd_range(struct vm_area_struct *vma,
 			mmu_notifier_invalidate_range_start(mm, mni_start, end);
 		}
 
-		if (pmd_trans_huge(*pmd)) {
+		if (pmd_trans_huge(*pmd) || pmd_devmap(*pmd)) {
 			if (next - addr != HPAGE_PMD_SIZE)
 				split_huge_pmd(vma, pmd, addr);
 			else {
diff --git a/mm/pgtable-generic.c b/mm/pgtable-generic.c
index c311a2ec6fea..9d4767698a1c 100644
--- a/mm/pgtable-generic.c
+++ b/mm/pgtable-generic.c
@@ -132,7 +132,7 @@ pmd_t pmdp_huge_clear_flush(struct vm_area_struct *vma, unsigned long address,
 {
 	pmd_t pmd;
 	VM_BUG_ON(address & ~HPAGE_PMD_MASK);
-	VM_BUG_ON(!pmd_trans_huge(*pmdp));
+	VM_BUG_ON(!pmd_trans_huge(*pmdp) && !pmd_devmap(*pmdp));
 	pmd = pmdp_huge_get_and_clear(vma->vm_mm, address, pmdp);
 	flush_pmd_tlb_range(vma, address, address + HPAGE_PMD_SIZE);
 	return pmd;

--
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>

  reply	other threads:[~2015-12-25  0:59 UTC|newest]

Thread overview: 29+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-12-21  5:44 [-mm PATCH v4 00/18] get_user_pages() for dax pte and pmd mappings Dan Williams
2015-12-21  5:44 ` [-mm PATCH v4 01/18] kvm: rename pfn_t to kvm_pfn_t Dan Williams
2015-12-21  5:44 ` [-mm PATCH v4 02/18] mm, dax, pmem: introduce pfn_t Dan Williams
2015-12-21  5:44 ` [-mm PATCH v4 03/18] mm: skip memory block registration for ZONE_DEVICE Dan Williams
2015-12-21  5:44 ` [-mm PATCH v4 04/18] mm: introduce find_dev_pagemap() Dan Williams
2015-12-21  5:44 ` [-mm PATCH v4 05/18] x86, mm: introduce vmem_altmap to augment vmemmap_populate() Dan Williams
2015-12-27  8:40   ` Bob Liu
2015-12-21  5:44 ` [-mm PATCH v4 06/18] libnvdimm, pfn, pmem: allocate memmap array in persistent memory Dan Williams
2015-12-21  5:44 ` [-mm PATCH v4 07/18] avr32: convert to asm-generic/memory_model.h Dan Williams
2015-12-21  5:44 ` [-mm PATCH v4 08/18] hugetlb: fix compile error on tile Dan Williams
2015-12-21  5:44 ` [-mm PATCH v4 09/18] frv: fix compiler warning from definition of __pmd() Dan Williams
2015-12-21  5:45 ` [-mm PATCH v4 10/18] x86, mm: introduce _PAGE_DEVMAP Dan Williams
2015-12-21  5:45 ` [-mm PATCH v4 11/18] mm, dax, gpu: convert vm_insert_mixed to pfn_t Dan Williams
2015-12-21  5:45 ` [-mm PATCH v4 12/18] mm, dax: convert vmf_insert_pfn_pmd() " Dan Williams
2015-12-21  5:45 ` [-mm PATCH v4 13/18] libnvdimm, pmem: move request_queue allocation earlier in probe Dan Williams
2015-12-21  5:45 ` [-mm PATCH v4 14/18] mm, dax, pmem: introduce {get|put}_dev_pagemap() for dax-gup Dan Williams
2015-12-27  8:46   ` Bob Liu
2015-12-27 19:02     ` Dan Williams
2015-12-21  5:45 ` [-mm PATCH v4 15/18] mm, dax: dax-pmd vs thp-pmd vs hugetlbfs-pmd Dan Williams
2015-12-25  0:59   ` Dan Williams [this message]
2015-12-25  1:11     ` [-mm PATCH v5 " Sasha Levin
2015-12-30  5:32   ` [-mm PATCH v4 " Williams, Dan J
2015-12-21  5:45 ` [-mm PATCH v4 16/18] mm, x86: get_user_pages() for dax mappings Dan Williams
2015-12-25  1:03   ` [-mm PATCH v5 " Dan Williams
2015-12-21  5:45 ` [-mm PATCH v4 17/18] dax: provide diagnostics for pmd mapping failures Dan Williams
2015-12-21  5:45 ` [-mm PATCH v4 18/18] dax: re-enable dax pmd mappings Dan Williams
2015-12-27  8:33 ` [-mm PATCH v4 00/18] get_user_pages() for dax pte and " Bob Liu
2015-12-27 18:55   ` Dan Williams
2015-12-29  3:23     ` Bob Liu

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=20151225005623.19962.1972.stgit@dwillia2-desk3.jf.intel.com \
    --to=dan.j.williams@intel.com \
    --cc=aarcange@redhat.com \
    --cc=akpm@linux-foundation.org \
    --cc=dave@sr71.net \
    --cc=kirill.shutemov@linux.intel.com \
    --cc=linux-mm@kvack.org \
    --cc=linux-nvdimm@lists.01.org \
    --cc=mgorman@suse.de \
    --cc=peterz@infradead.org \
    --cc=sasha.levin@oracle.com \
    --cc=willy@linux.intel.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.