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

Groups > linux.kernel > #1475067 > unrolled thread

[RFC PATCH v2 0/3] Add support for eXclusive Page Frame Ownership (XPFO)

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  [RFC PATCH v2 0/3] Add support for eXclusive Page Frame Ownership (XPFO) Juerg Haefliger <juerg.haefliger@hpe.com> - 2016-09-02 13:40 +0200
    [RFC PATCH v2 1/3] Add support for eXclusive Page Frame Ownership (XPFO) Juerg Haefliger <juerg.haefliger@hpe.com> - 2016-09-02 13:40 +0200
    [RFC PATCH v2 3/3] block: Always use a bounce buffer when XPFO is enabled Juerg Haefliger <juerg.haefliger@hpe.com> - 2016-09-02 13:50 +0200
    [RFC PATCH v2 2/3] xpfo: Only put previous userspace pages into the hot cache Juerg Haefliger <juerg.haefliger@hpe.com> - 2016-09-02 13:50 +0200
      Re: [RFC PATCH v2 2/3] xpfo: Only put previous userspace pages into  the hot cache Dave Hansen <dave.hansen@intel.com> - 2016-09-02 22:40 +0200
        Re: [RFC PATCH v2 2/3] xpfo: Only put previous userspace pages into  the hot cache Juerg Haefliger <juerg.haefliger@hpe.com> - 2016-09-05 14:00 +0200

#1475067 — [RFC PATCH v2 0/3] Add support for eXclusive Page Frame Ownership (XPFO)

Changes from:
  v1 -> v2:
    - Moved the code from arch/x86/mm/ to mm/ since it's (mostly)
      arch-agnostic.
    - Moved the config to the generic layer and added ARCH_SUPPORTS_XPFO
      for x86.
    - Use page_ext for the additional per-page data.
    - Removed the clearing of pages. This can be accomplished by using
      PAGE_POISONING.
    - Split up the patch into multiple patches.
    - Fixed additional issues identified by reviewers.

This patch series adds support for XPFO which protects against 'ret2dir'
kernel attacks. The basic idea is to enforce exclusive ownership of page
frames by either the kernel or userspace, unless explicitly requested by
the kernel. Whenever a page destined for userspace is allocated, it is
unmapped from physmap (the kernel's page table). When such a page is
reclaimed from userspace, it is mapped back to physmap.

Additional fields in the page_ext struct are used for XPFO housekeeping.
Specifically two flags to distinguish user vs. kernel pages and to tag
unmapped pages and a reference counter to balance kmap/kunmap operations
and a lock to serialize access to the XPFO fields.

Known issues/limitations:
  - Only supports x86-64 (for now)
  - Only supports 4k pages (for now)
  - There are most likely some legitimate uses cases where the kernel needs
    to access userspace which need to be made XPFO-aware
  - Performance penalty

Reference paper by the original patch authors:
  http://www.cs.columbia.edu/~vpk/papers/ret2dir.sec14.pdf

Juerg Haefliger (3):
  Add support for eXclusive Page Frame Ownership (XPFO)
  xpfo: Only put previous userspace pages into the hot cache
  block: Always use a bounce buffer when XPFO is enabled

 arch/x86/Kconfig         |   3 +-
 arch/x86/mm/init.c       |   2 +-
 block/blk-map.c          |   2 +-
 include/linux/highmem.h  |  15 +++-
 include/linux/page_ext.h |   7 ++
 include/linux/xpfo.h     |  41 +++++++++
 lib/swiotlb.c            |   3 +-
 mm/Makefile              |   1 +
 mm/page_alloc.c          |  10 ++-
 mm/page_ext.c            |   4 +
 mm/xpfo.c                | 213 +++++++++++++++++++++++++++++++++++++++++++++++
 security/Kconfig         |  20 +++++
 12 files changed, 314 insertions(+), 7 deletions(-)
 create mode 100644 include/linux/xpfo.h
 create mode 100644 mm/xpfo.c

-- 
2.9.3

[toc] | [next] | [standalone]

#1475068 — [RFC PATCH v2 1/3] Add support for eXclusive Page Frame Ownership (XPFO)

This patch adds support for XPFO which protects against 'ret2dir' kernel
attacks. The basic idea is to enforce exclusive ownership of page frames
by either the kernel or userspace, unless explicitly requested by the
kernel. Whenever a page destined for userspace is allocated, it is
unmapped from physmap (the kernel's page table). When such a page is
reclaimed from userspace, it is mapped back to physmap.

Additional fields in the page_ext struct are used for XPFO housekeeping.
Specifically two flags to distinguish user vs. kernel pages and to tag
unmapped pages and a reference counter to balance kmap/kunmap operations
and a lock to serialize access to the XPFO fields.

Known issues/limitations:
  - Only supports x86-64 (for now)
  - Only supports 4k pages (for now)
  - There are most likely some legitimate uses cases where the kernel needs
    to access userspace which need to be made XPFO-aware
  - Performance penalty

Reference paper by the original patch authors:
  http://www.cs.columbia.edu/~vpk/papers/ret2dir.sec14.pdf

Suggested-by: Vasileios P. Kemerlis <vpk@cs.columbia.edu>
Signed-off-by: Juerg Haefliger <juerg.haefliger@hpe.com>
---
 arch/x86/Kconfig         |   3 +-
 arch/x86/mm/init.c       |   2 +-
 include/linux/highmem.h  |  15 +++-
 include/linux/page_ext.h |   7 ++
 include/linux/xpfo.h     |  39 +++++++++
 lib/swiotlb.c            |   3 +-
 mm/Makefile              |   1 +
 mm/page_alloc.c          |   2 +
 mm/page_ext.c            |   4 +
 mm/xpfo.c                | 205 +++++++++++++++++++++++++++++++++++++++++++++++
 security/Kconfig         |  20 +++++
 11 files changed, 296 insertions(+), 5 deletions(-)
 create mode 100644 include/linux/xpfo.h
 create mode 100644 mm/xpfo.c

diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
index c580d8c33562..dc5604a710c6 100644
--- a/arch/x86/Kconfig
+++ b/arch/x86/Kconfig
@@ -165,6 +165,7 @@ config X86
 	select HAVE_STACK_VALIDATION		if X86_64
 	select ARCH_USES_HIGH_VMA_FLAGS		if X86_INTEL_MEMORY_PROTECTION_KEYS
 	select ARCH_HAS_PKEYS			if X86_INTEL_MEMORY_PROTECTION_KEYS
+	select ARCH_SUPPORTS_XPFO		if X86_64
 
 config INSTRUCTION_DECODER
 	def_bool y
@@ -1350,7 +1351,7 @@ config ARCH_DMA_ADDR_T_64BIT
 
 config X86_DIRECT_GBPAGES
 	def_bool y
-	depends on X86_64 && !DEBUG_PAGEALLOC && !KMEMCHECK
+	depends on X86_64 && !DEBUG_PAGEALLOC && !KMEMCHECK && !XPFO
 	---help---
 	  Certain kernel features effectively disable kernel
 	  linear 1 GB mappings (even if the CPU otherwise
diff --git a/arch/x86/mm/init.c b/arch/x86/mm/init.c
index d28a2d741f9e..426427b54639 100644
--- a/arch/x86/mm/init.c
+++ b/arch/x86/mm/init.c
@@ -161,7 +161,7 @@ static int page_size_mask;
 
 static void __init probe_page_size_mask(void)
 {
-#if !defined(CONFIG_KMEMCHECK)
+#if !defined(CONFIG_KMEMCHECK) && !defined(CONFIG_XPFO)
 	/*
 	 * For CONFIG_KMEMCHECK or pagealloc debugging, identity mapping will
 	 * use small pages.
diff --git a/include/linux/highmem.h b/include/linux/highmem.h
index bb3f3297062a..7a17c166532f 100644
--- a/include/linux/highmem.h
+++ b/include/linux/highmem.h
@@ -7,6 +7,7 @@
 #include <linux/mm.h>
 #include <linux/uaccess.h>
 #include <linux/hardirq.h>
+#include <linux/xpfo.h>
 
 #include <asm/cacheflush.h>
 
@@ -55,24 +56,34 @@ static inline struct page *kmap_to_page(void *addr)
 #ifndef ARCH_HAS_KMAP
 static inline void *kmap(struct page *page)
 {
+	void *kaddr;
+
 	might_sleep();
-	return page_address(page);
+	kaddr = page_address(page);
+	xpfo_kmap(kaddr, page);
+	return kaddr;
 }
 
 static inline void kunmap(struct page *page)
 {
+	xpfo_kunmap(page_address(page), page);
 }
 
 static inline void *kmap_atomic(struct page *page)
 {
+	void *kaddr;
+
 	preempt_disable();
 	pagefault_disable();
-	return page_address(page);
+	kaddr = page_address(page);
+	xpfo_kmap(kaddr, page);
+	return kaddr;
 }
 #define kmap_atomic_prot(page, prot)	kmap_atomic(page)
 
 static inline void __kunmap_atomic(void *addr)
 {
+	xpfo_kunmap(addr, virt_to_page(addr));
 	pagefault_enable();
 	preempt_enable();
 }
diff --git a/include/linux/page_ext.h b/include/linux/page_ext.h
index 03f2a3e7d76d..fdf63dcc399e 100644
--- a/include/linux/page_ext.h
+++ b/include/linux/page_ext.h
@@ -27,6 +27,8 @@ enum page_ext_flags {
 	PAGE_EXT_DEBUG_POISON,		/* Page is poisoned */
 	PAGE_EXT_DEBUG_GUARD,
 	PAGE_EXT_OWNER,
+	PAGE_EXT_XPFO_KERNEL,		/* Page is a kernel page */
+	PAGE_EXT_XPFO_UNMAPPED,		/* Page is unmapped */
 #if defined(CONFIG_IDLE_PAGE_TRACKING) && !defined(CONFIG_64BIT)
 	PAGE_EXT_YOUNG,
 	PAGE_EXT_IDLE,
@@ -48,6 +50,11 @@ struct page_ext {
 	int last_migrate_reason;
 	depot_stack_handle_t handle;
 #endif
+#ifdef CONFIG_XPFO
+	int inited;		/* Map counter and lock initialized */
+	atomic_t mapcount;	/* Counter for balancing map/unmap requests */
+	spinlock_t maplock;	/* Lock to serialize map/unmap requests */
+#endif
 };
 
 extern void pgdat_page_ext_init(struct pglist_data *pgdat);
diff --git a/include/linux/xpfo.h b/include/linux/xpfo.h
new file mode 100644
index 000000000000..77187578ca33
--- /dev/null
+++ b/include/linux/xpfo.h
@@ -0,0 +1,39 @@
+/*
+ * Copyright (C) 2016 Hewlett Packard Enterprise Development, L.P.
+ * Copyright (C) 2016 Brown University. All rights reserved.
+ *
+ * Authors:
+ *   Juerg Haefliger <juerg.haefliger@hpe.com>
+ *   Vasileios P. Kemerlis <vpk@cs.brown.edu>
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 as published by
+ * the Free Software Foundation.
+ */
+
+#ifndef _LINUX_XPFO_H
+#define _LINUX_XPFO_H
+
+#ifdef CONFIG_XPFO
+
+extern struct page_ext_operations page_xpfo_ops;
+
+extern void xpfo_kmap(void *kaddr, struct page *page);
+extern void xpfo_kunmap(void *kaddr, struct page *page);
+extern void xpfo_alloc_page(struct page *page, int order, gfp_t gfp);
+extern void xpfo_free_page(struct page *page, int order);
+
+extern bool xpfo_page_is_unmapped(struct page *page);
+
+#else /* !CONFIG_XPFO */
+
+static inline void xpfo_kmap(void *kaddr, struct page *page) { }
+static inline void xpfo_kunmap(void *kaddr, struct page *page) { }
+static inline void xpfo_alloc_page(struct page *page, int order, gfp_t gfp) { }
+static inline void xpfo_free_page(struct page *page, int order) { }
+
+static inline bool xpfo_page_is_unmapped(struct page *page) { return false; }
+
+#endif /* CONFIG_XPFO */
+
+#endif /* _LINUX_XPFO_H */
diff --git a/lib/swiotlb.c b/lib/swiotlb.c
index 22e13a0e19d7..455eff44604e 100644
--- a/lib/swiotlb.c
+++ b/lib/swiotlb.c
@@ -390,8 +390,9 @@ static void swiotlb_bounce(phys_addr_t orig_addr, phys_addr_t tlb_addr,
 {
 	unsigned long pfn = PFN_DOWN(orig_addr);
 	unsigned char *vaddr = phys_to_virt(tlb_addr);
+	struct page *page = pfn_to_page(pfn);
 
-	if (PageHighMem(pfn_to_page(pfn))) {
+	if (PageHighMem(page) || xpfo_page_is_unmapped(page)) {
 		/* The buffer does not have a mapping.  Map it in and copy */
 		unsigned int offset = orig_addr & ~PAGE_MASK;
 		char *buffer;
diff --git a/mm/Makefile b/mm/Makefile
index 2ca1faf3fa09..e6f8894423da 100644
--- a/mm/Makefile
+++ b/mm/Makefile
@@ -103,3 +103,4 @@ obj-$(CONFIG_IDLE_PAGE_TRACKING) += page_idle.o
 obj-$(CONFIG_FRAME_VECTOR) += frame_vector.o
 obj-$(CONFIG_DEBUG_PAGE_REF) += debug_page_ref.o
 obj-$(CONFIG_HARDENED_USERCOPY) += usercopy.o
+obj-$(CONFIG_XPFO) += xpfo.o
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 3fbe73a6fe4b..0241c8a7e72a 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -1029,6 +1029,7 @@ static __always_inline bool free_pages_prepare(struct page *page,
 	kernel_poison_pages(page, 1 << order, 0);
 	kernel_map_pages(page, 1 << order, 0);
 	kasan_free_pages(page, order);
+	xpfo_free_page(page, order);
 
 	return true;
 }
@@ -1726,6 +1727,7 @@ inline void post_alloc_hook(struct page *page, unsigned int order,
 	kernel_map_pages(page, 1 << order, 1);
 	kernel_poison_pages(page, 1 << order, 1);
 	kasan_alloc_pages(page, order);
+	xpfo_alloc_page(page, order, gfp_flags);
 	set_page_owner(page, order, gfp_flags);
 }
 
diff --git a/mm/page_ext.c b/mm/page_ext.c
index 44a4c029c8e7..1cd7d7f460cc 100644
--- a/mm/page_ext.c
+++ b/mm/page_ext.c
@@ -7,6 +7,7 @@
 #include <linux/kmemleak.h>
 #include <linux/page_owner.h>
 #include <linux/page_idle.h>
+#include <linux/xpfo.h>
 
 /*
  * struct page extension
@@ -63,6 +64,9 @@ static struct page_ext_operations *page_ext_ops[] = {
 #if defined(CONFIG_IDLE_PAGE_TRACKING) && !defined(CONFIG_64BIT)
 	&page_idle_ops,
 #endif
+#ifdef CONFIG_XPFO
+	&page_xpfo_ops,
+#endif
 };
 
 static unsigned long total_usage;
diff --git a/mm/xpfo.c b/mm/xpfo.c
new file mode 100644
index 000000000000..ddb1be05485d
--- /dev/null
+++ b/mm/xpfo.c
@@ -0,0 +1,205 @@
+/*
+ * Copyright (C) 2016 Hewlett Packard Enterprise Development, L.P.
+ * Copyright (C) 2016 Brown University. All rights reserved.
+ *
+ * Authors:
+ *   Juerg Haefliger <juerg.haefliger@hpe.com>
+ *   Vasileios P. Kemerlis <vpk@cs.brown.edu>
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 as published by
+ * the Free Software Foundation.
+ */
+
+#include <linux/mm.h>
+#include <linux/module.h>
+#include <linux/page_ext.h>
+#include <linux/xpfo.h>
+
+#include <asm/tlbflush.h>
+
+DEFINE_STATIC_KEY_FALSE(xpfo_inited);
+
+static bool need_xpfo(void)
+{
+	return true;
+}
+
+static void init_xpfo(void)
+{
+	printk(KERN_INFO "XPFO enabled\n");
+	static_branch_enable(&xpfo_inited);
+}
+
+struct page_ext_operations page_xpfo_ops = {
+	.need = need_xpfo,
+	.init = init_xpfo,
+};
+
+/*
+ * Update a single kernel page table entry
+ */
+static inline void set_kpte(struct page *page, unsigned long kaddr,
+			    pgprot_t prot) {
+	unsigned int level;
+	pte_t *kpte = lookup_address(kaddr, &level);
+
+	/* We only support 4k pages for now */
+	BUG_ON(!kpte || level != PG_LEVEL_4K);
+
+	set_pte_atomic(kpte, pfn_pte(page_to_pfn(page), canon_pgprot(prot)));
+}
+
+void xpfo_alloc_page(struct page *page, int order, gfp_t gfp)
+{
+	int i, flush_tlb = 0;
+	struct page_ext *page_ext;
+	unsigned long kaddr;
+
+	if (!static_branch_unlikely(&xpfo_inited))
+		return;
+
+	for (i = 0; i < (1 << order); i++)  {
+		page_ext = lookup_page_ext(page + i);
+
+		BUG_ON(test_bit(PAGE_EXT_XPFO_UNMAPPED, &page_ext->flags));
+
+		/* Initialize the map lock and map counter */
+		if (!page_ext->inited) {
+			spin_lock_init(&page_ext->maplock);
+			atomic_set(&page_ext->mapcount, 0);
+			page_ext->inited = 1;
+		}
+		BUG_ON(atomic_read(&page_ext->mapcount));
+
+		if ((gfp & GFP_HIGHUSER) == GFP_HIGHUSER) {
+			/*
+			 * Flush the TLB if the page was previously allocated
+			 * to the kernel.
+			 */
+			if (test_and_clear_bit(PAGE_EXT_XPFO_KERNEL,
+					       &page_ext->flags))
+				flush_tlb = 1;
+		} else {
+			/* Tag the page as a kernel page */
+			set_bit(PAGE_EXT_XPFO_KERNEL, &page_ext->flags);
+		}
+	}
+
+	if (flush_tlb) {
+		kaddr = (unsigned long)page_address(page);
+		flush_tlb_kernel_range(kaddr, kaddr + (1 << order) *
+				       PAGE_SIZE);
+	}
+}
+
+void xpfo_free_page(struct page *page, int order)
+{
+	int i;
+	struct page_ext *page_ext;
+	unsigned long kaddr;
+
+	if (!static_branch_unlikely(&xpfo_inited))
+		return;
+
+	for (i = 0; i < (1 << order); i++) {
+		page_ext = lookup_page_ext(page + i);
+
+		if (!page_ext->inited) {
+			/*
+			 * The page was allocated before page_ext was
+			 * initialized, so it is a kernel page and it needs to
+			 * be tagged accordingly.
+			 */
+			set_bit(PAGE_EXT_XPFO_KERNEL, &page_ext->flags);
+			continue;
+		}
+
+		/*
+		 * Map the page back into the kernel if it was previously
+		 * allocated to user space.
+		 */
+		if (test_and_clear_bit(PAGE_EXT_XPFO_UNMAPPED,
+				       &page_ext->flags)) {
+			kaddr = (unsigned long)page_address(page + i);
+			set_kpte(page + i,  kaddr, __pgprot(__PAGE_KERNEL));
+		}
+	}
+}
+
+void xpfo_kmap(void *kaddr, struct page *page)
+{
+	struct page_ext *page_ext;
+	unsigned long flags;
+
+	if (!static_branch_unlikely(&xpfo_inited))
+		return;
+
+	page_ext = lookup_page_ext(page);
+
+	/*
+	 * The page was allocated before page_ext was initialized (which means
+	 * it's a kernel page) or it's allocated to the kernel, so nothing to
+	 * do.
+	 */
+	if (!page_ext->inited ||
+	    test_bit(PAGE_EXT_XPFO_KERNEL, &page_ext->flags))
+		return;
+
+	spin_lock_irqsave(&page_ext->maplock, flags);
+
+	/*
+	 * The page was previously allocated to user space, so map it back
+	 * into the kernel. No TLB flush required.
+	 */
+	if ((atomic_inc_return(&page_ext->mapcount) == 1) &&
+	    test_and_clear_bit(PAGE_EXT_XPFO_UNMAPPED, &page_ext->flags))
+		set_kpte(page, (unsigned long)kaddr, __pgprot(__PAGE_KERNEL));
+
+	spin_unlock_irqrestore(&page_ext->maplock, flags);
+}
+EXPORT_SYMBOL(xpfo_kmap);
+
+void xpfo_kunmap(void *kaddr, struct page *page)
+{
+	struct page_ext *page_ext;
+	unsigned long flags;
+
+	if (!static_branch_unlikely(&xpfo_inited))
+		return;
+
+	page_ext = lookup_page_ext(page);
+
+	/*
+	 * The page was allocated before page_ext was initialized (which means
+	 * it's a kernel page) or it's allocated to the kernel, so nothing to
+	 * do.
+	 */
+	if (!page_ext->inited ||
+	    test_bit(PAGE_EXT_XPFO_KERNEL, &page_ext->flags))
+		return;
+
+	spin_lock_irqsave(&page_ext->maplock, flags);
+
+	/*
+	 * The page is to be allocated back to user space, so unmap it from the
+	 * kernel, flush the TLB and tag it as a user page.
+	 */
+	if (atomic_dec_return(&page_ext->mapcount) == 0) {
+		BUG_ON(test_bit(PAGE_EXT_XPFO_UNMAPPED, &page_ext->flags));
+		set_bit(PAGE_EXT_XPFO_UNMAPPED, &page_ext->flags);
+		set_kpte(page, (unsigned long)kaddr, __pgprot(0));
+		__flush_tlb_one((unsigned long)kaddr);
+	}
+
+	spin_unlock_irqrestore(&page_ext->maplock, flags);
+}
+EXPORT_SYMBOL(xpfo_kunmap);
+
+inline bool xpfo_page_is_unmapped(struct page *page)
+{
+	if (!static_branch_unlikely(&xpfo_inited))
+		return false;
+
+	return test_bit(PAGE_EXT_XPFO_UNMAPPED, &lookup_page_ext(page)->flags);
+}
diff --git a/security/Kconfig b/security/Kconfig
index da10d9b573a4..1eac37a9bec2 100644
--- a/security/Kconfig
+++ b/security/Kconfig
@@ -6,6 +6,26 @@ menu "Security options"
 
 source security/keys/Kconfig
 
+config ARCH_SUPPORTS_XPFO
+	bool
+
+config XPFO
+	bool "Enable eXclusive Page Frame Ownership (XPFO)"
+	default n
+	depends on DEBUG_KERNEL && ARCH_SUPPORTS_XPFO
+	select DEBUG_TLBFLUSH
+	select PAGE_EXTENSION
+	help
+	  This option offers protection against 'ret2dir' kernel attacks.
+	  When enabled, every time a page frame is allocated to user space, it
+	  is unmapped from the direct mapped RAM region in kernel space
+	  (physmap). Similarly, when a page frame is freed/reclaimed, it is
+	  mapped back to physmap.
+
+	  There is a slight performance impact when this option is enabled.
+
+	  If in doubt, say "N".
+
 config SECURITY_DMESG_RESTRICT
 	bool "Restrict unprivileged access to the kernel syslog"
 	default n
-- 
2.9.3

[toc] | [prev] | [next] | [standalone]

#1475071 — [RFC PATCH v2 3/3] block: Always use a bounce buffer when XPFO is enabled

This is a temporary hack to prevent the use of bio_map_user_iov()
which causes XPFO page faults.

Signed-off-by: Juerg Haefliger <juerg.haefliger@hpe.com>
---
 block/blk-map.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/block/blk-map.c b/block/blk-map.c
index b8657fa8dc9a..e889dbfee6fb 100644
--- a/block/blk-map.c
+++ b/block/blk-map.c
@@ -52,7 +52,7 @@ static int __blk_rq_map_user_iov(struct request *rq,
 	struct bio *bio, *orig_bio;
 	int ret;
 
-	if (copy)
+	if (copy || IS_ENABLED(CONFIG_XPFO))
 		bio = bio_copy_user_iov(q, map_data, iter, gfp_mask);
 	else
 		bio = bio_map_user_iov(q, iter, gfp_mask);
-- 
2.9.3

[toc] | [prev] | [next] | [standalone]

#1475072 — [RFC PATCH v2 2/3] xpfo: Only put previous userspace pages into the hot cache

Allocating a page to userspace that was previously allocated to the
kernel requires an expensive TLB shootdown. To minimize this, we only
put non-kernel pages into the hot cache to favor their allocation.

Signed-off-by: Juerg Haefliger <juerg.haefliger@hpe.com>
---
 include/linux/xpfo.h | 2 ++
 mm/page_alloc.c      | 8 +++++++-
 mm/xpfo.c            | 8 ++++++++
 3 files changed, 17 insertions(+), 1 deletion(-)

diff --git a/include/linux/xpfo.h b/include/linux/xpfo.h
index 77187578ca33..077d1cfadfa2 100644
--- a/include/linux/xpfo.h
+++ b/include/linux/xpfo.h
@@ -24,6 +24,7 @@ extern void xpfo_alloc_page(struct page *page, int order, gfp_t gfp);
 extern void xpfo_free_page(struct page *page, int order);
 
 extern bool xpfo_page_is_unmapped(struct page *page);
+extern bool xpfo_page_is_kernel(struct page *page);
 
 #else /* !CONFIG_XPFO */
 
@@ -33,6 +34,7 @@ static inline void xpfo_alloc_page(struct page *page, int order, gfp_t gfp) { }
 static inline void xpfo_free_page(struct page *page, int order) { }
 
 static inline bool xpfo_page_is_unmapped(struct page *page) { return false; }
+static inline bool xpfo_page_is_kernel(struct page *page) { return false; }
 
 #endif /* CONFIG_XPFO */
 
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 0241c8a7e72a..83404b41e52d 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -2421,7 +2421,13 @@ void free_hot_cold_page(struct page *page, bool cold)
 	}
 
 	pcp = &this_cpu_ptr(zone->pageset)->pcp;
-	if (!cold)
+	/*
+	 * XPFO: Allocating a page to userspace that was previously allocated
+	 * to the kernel requires an expensive TLB shootdown. To minimize this,
+	 * we only put non-kernel pages into the hot cache to favor their
+	 * allocation.
+	 */
+	if (!cold && !xpfo_page_is_kernel(page))
 		list_add(&page->lru, &pcp->lists[migratetype]);
 	else
 		list_add_tail(&page->lru, &pcp->lists[migratetype]);
diff --git a/mm/xpfo.c b/mm/xpfo.c
index ddb1be05485d..f8dffda0c961 100644
--- a/mm/xpfo.c
+++ b/mm/xpfo.c
@@ -203,3 +203,11 @@ inline bool xpfo_page_is_unmapped(struct page *page)
 
 	return test_bit(PAGE_EXT_XPFO_UNMAPPED, &lookup_page_ext(page)->flags);
 }
+
+inline bool xpfo_page_is_kernel(struct page *page)
+{
+	if (!static_branch_unlikely(&xpfo_inited))
+		return false;
+
+	return test_bit(PAGE_EXT_XPFO_KERNEL, &lookup_page_ext(page)->flags);
+}
-- 
2.9.3

[toc] | [prev] | [next] | [standalone]

#1475437 — Re: [RFC PATCH v2 2/3] xpfo: Only put previous userspace pages into the hot cache

On 09/02/2016 04:39 AM, Juerg Haefliger wrote:
> Allocating a page to userspace that was previously allocated to the
> kernel requires an expensive TLB shootdown. To minimize this, we only
> put non-kernel pages into the hot cache to favor their allocation.

But kernel allocations do allocate from these pools, right?  Does this
just mean that kernel allocations usually have to pay the penalty to
convert a page?

So, what's the logic here?  You're assuming that order-0 kernel
allocations are more rare than allocations for userspace?

[toc] | [prev] | [next] | [standalone]

#1476335 — Re: [RFC PATCH v2 2/3] xpfo: Only put previous userspace pages into the hot cache

[Multipart message — attachments visible in raw view] — view raw

On 09/02/2016 10:39 PM, Dave Hansen wrote:
> On 09/02/2016 04:39 AM, Juerg Haefliger wrote:
>> Allocating a page to userspace that was previously allocated to the
>> kernel requires an expensive TLB shootdown. To minimize this, we only
>> put non-kernel pages into the hot cache to favor their allocation.
> 
> But kernel allocations do allocate from these pools, right?

Yes.


> Does this
> just mean that kernel allocations usually have to pay the penalty to
> convert a page?

Only pages that are allocated for userspace (gfp & GFP_HIGHUSER == GFP_HIGHUSER) which were
previously allocated for the kernel (gfp & GFP_HIGHUSER != GFP_HIGHUSER) have to pay the penalty.


> So, what's the logic here?  You're assuming that order-0 kernel
> allocations are more rare than allocations for userspace?

The logic is to put reclaimed kernel pages into the cold cache to postpone their allocation as long
as possible to minimize (potential) TLB flushes.

...Juerg

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web