Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1413230 > unrolled thread

[PATCH v2 1/3] mm/hugetlb: Simplify hugetlb unmap

Back to article view | Back to linux.kernel

Contents

  [PATCH v2 1/3] mm/hugetlb: Simplify hugetlb unmap "Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com> - 2016-06-03 15:20 +0200
    Re: [PATCH v2 1/3] mm/hugetlb: Simplify hugetlb unmap "Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com> - 2016-06-03 15:40 +0200

#1413230 — [PATCH v2 1/3] mm/hugetlb: Simplify hugetlb unmap

For hugetlb like THP (and unlike regular page), we do tlb flush after
dropping ptl. Because of the above, we don't need to track force_flush
like we do now. Instead we can simply call tlb_remove_page() which
will do the flush if needed.

No functionality change in this patch.

Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
---
 mm/hugetlb.c | 54 +++++++++++++++++++++---------------------------------
 1 file changed, 21 insertions(+), 33 deletions(-)

diff --git a/mm/hugetlb.c b/mm/hugetlb.c
index d26162e81fea..741429d01668 100644
--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c
@@ -3138,7 +3138,6 @@ void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma,
 			    unsigned long start, unsigned long end,
 			    struct page *ref_page)
 {
-	int force_flush = 0;
 	struct mm_struct *mm = vma->vm_mm;
 	unsigned long address;
 	pte_t *ptep;
@@ -3157,19 +3156,22 @@ void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma,
 	tlb_start_vma(tlb, vma);
 	mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end);
 	address = start;
-again:
 	for (; address < end; address += sz) {
 		ptep = huge_pte_offset(mm, address);
 		if (!ptep)
 			continue;
 
 		ptl = huge_pte_lock(h, mm, ptep);
-		if (huge_pmd_unshare(mm, &address, ptep))
-			goto unlock;
+		if (huge_pmd_unshare(mm, &address, ptep)) {
+			spin_unlock(ptl);
+			continue;
+		}
 
 		pte = huge_ptep_get(ptep);
-		if (huge_pte_none(pte))
-			goto unlock;
+		if (huge_pte_none(pte)) {
+			spin_unlock(ptl);
+			continue;
+		}
 
 		/*
 		 * Migrating hugepage or HWPoisoned hugepage is already
@@ -3177,7 +3179,8 @@ again:
 		 */
 		if (unlikely(!pte_present(pte))) {
 			huge_pte_clear(mm, address, ptep);
-			goto unlock;
+			spin_unlock(ptl);
+			continue;
 		}
 
 		page = pte_page(pte);
@@ -3187,9 +3190,10 @@ again:
 		 * are about to unmap is the actual page of interest.
 		 */
 		if (ref_page) {
-			if (page != ref_page)
-				goto unlock;
-
+			if (page != ref_page) {
+				spin_unlock(ptl);
+				continue;
+			}
 			/*
 			 * Mark the VMA as having unmapped its page so that
 			 * future faults in this VMA will fail rather than
@@ -3205,30 +3209,14 @@ again:
 
 		hugetlb_count_sub(pages_per_huge_page(h), mm);
 		page_remove_rmap(page, true);
-		force_flush = !__tlb_remove_page(tlb, page);
-		if (force_flush) {
-			address += sz;
-			spin_unlock(ptl);
-			break;
-		}
-		/* Bail out after unmapping reference page if supplied */
-		if (ref_page) {
-			spin_unlock(ptl);
-			break;
-		}
-unlock:
+
 		spin_unlock(ptl);
-	}
-	/*
-	 * mmu_gather ran out of room to batch pages, we break out of
-	 * the PTE lock to avoid doing the potential expensive TLB invalidate
-	 * and page-free while holding it.
-	 */
-	if (force_flush) {
-		force_flush = 0;
-		tlb_flush_mmu(tlb);
-		if (address < end && !ref_page)
-			goto again;
+		tlb_remove_page(tlb, page);
+		/*
+		 * Bail out after unmapping reference page if supplied
+		 */
+		if (ref_page)
+			break;
 	}
 	mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end);
 	tlb_end_vma(tlb, vma);
-- 
2.7.4

[toc] | [next] | [standalone]

#1413246

Hi Andrew,

The updated version includes a build fix for [PATCH 2/3. I also dropped the
powerpc related changes from the series because that have dependencies
against other patches not yet merged upstream. I am adding the same
below for reference.

From 1f0975adfd3138f3709b2dec8771065ddc38de40 Mon Sep 17 00:00:00 2001
From: "Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com>
Date: Tue, 10 May 2016 12:24:34 +0530
Subject: [PATCH 4/4] powerpc/mm/radix: Implement tlb mmu gather flush
 efficiently

Now that we track page size in mmu_gather, we can use address based
tlbie format when doing a tlb_flush(). We don't do this if we are
invalidating the full address space.

Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
---
 .../powerpc/include/asm/book3s/64/tlbflush-radix.h |  2 +
 arch/powerpc/mm/tlb-radix.c                        | 73 +++++++++++++++++++++-
 2 files changed, 74 insertions(+), 1 deletion(-)

diff --git a/arch/powerpc/include/asm/book3s/64/tlbflush-radix.h b/arch/powerpc/include/asm/book3s/64/tlbflush-radix.h
index 3fa94fcac628..862c8fa50268 100644
--- a/arch/powerpc/include/asm/book3s/64/tlbflush-radix.h
+++ b/arch/powerpc/include/asm/book3s/64/tlbflush-radix.h
@@ -10,6 +10,8 @@ static inline int mmu_get_ap(int psize)
 	return mmu_psize_defs[psize].ap;
 }
 
+extern void radix__flush_tlb_range_psize(struct mm_struct *mm, unsigned long start,
+					 unsigned long end, int psize);
 extern void radix__flush_tlb_range(struct vm_area_struct *vma, unsigned long start,
 			    unsigned long end);
 extern void radix__flush_tlb_kernel_range(unsigned long start, unsigned long end);
diff --git a/arch/powerpc/mm/tlb-radix.c b/arch/powerpc/mm/tlb-radix.c
index 231e3ed2e684..03e719ee6747 100644
--- a/arch/powerpc/mm/tlb-radix.c
+++ b/arch/powerpc/mm/tlb-radix.c
@@ -279,9 +279,80 @@ void radix__flush_tlb_range(struct vm_area_struct *vma, unsigned long start,
 }
 EXPORT_SYMBOL(radix__flush_tlb_range);
 
+static int radix_get_mmu_psize(int page_size)
+{
+	int psize;
+
+	if (page_size == (1UL << mmu_psize_defs[mmu_virtual_psize].shift))
+		psize = mmu_virtual_psize;
+	else if (page_size == (1UL << mmu_psize_defs[MMU_PAGE_2M].shift))
+		psize = MMU_PAGE_2M;
+	else if (page_size == (1UL << mmu_psize_defs[MMU_PAGE_1G].shift))
+		psize = MMU_PAGE_1G;
+	else
+		return -1;
+	return psize;
+}
 
 void radix__tlb_flush(struct mmu_gather *tlb)
 {
+	int psize = 0;
 	struct mm_struct *mm = tlb->mm;
-	radix__flush_tlb_mm(mm);
+	int page_size = tlb->page_size;
+
+	psize = radix_get_mmu_psize(page_size);
+	/*
+	 * if page size is not something we understand, do a full mm flush
+	 */
+	if (psize != -1 && !tlb->fullmm && !tlb->need_flush_all)
+		radix__flush_tlb_range_psize(mm, tlb->start, tlb->end, psize);
+	else
+		radix__flush_tlb_mm(mm);
+}
+
+#define TLB_FLUSH_ALL -1UL
+/*
+ * Number of pages above which we will do a bcast tlbie. Just a
+ * number at this point copied from x86
+ */
+static unsigned long tlb_single_page_flush_ceiling __read_mostly = 33;
+
+void radix__flush_tlb_range_psize(struct mm_struct *mm, unsigned long start,
+				  unsigned long end, int psize)
+{
+	unsigned long pid;
+	unsigned long addr;
+	int local = mm_is_core_local(mm);
+	unsigned long ap = mmu_get_ap(psize);
+	int lock_tlbie = !mmu_has_feature(MMU_FTR_LOCKLESS_TLBIE);
+	unsigned long page_size = 1UL << mmu_psize_defs[psize].shift;
+
+
+	preempt_disable();
+	pid = mm ? mm->context.id : 0;
+	if (unlikely(pid == MMU_NO_CONTEXT))
+		goto err_out;
+
+	if (end == TLB_FLUSH_ALL ||
+	    (end - start) > tlb_single_page_flush_ceiling * page_size) {
+		if (local)
+			_tlbiel_pid(pid, RIC_FLUSH_TLB);
+		else
+			_tlbie_pid(pid, RIC_FLUSH_TLB);
+		goto err_out;
+	}
+	for (addr = start; addr < end; addr += page_size) {
+
+		if (local)
+			_tlbiel_va(addr, pid, ap, RIC_FLUSH_TLB);
+		else {
+			if (lock_tlbie)
+				raw_spin_lock(&native_tlbie_lock);
+			_tlbie_va(addr, pid, ap, RIC_FLUSH_TLB);
+			if (lock_tlbie)
+				raw_spin_unlock(&native_tlbie_lock);
+		}
+	}
+err_out:
+	preempt_enable();
 }
-- 
2.7.4

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web