generic debug pagealloc
CONFIG_DEBUG_PAGEALLOC is now supported by x86, powerpc, sparc64, and
s390. This patch implements it for the rest of the architectures by
filling the pages with poison byte patterns after free_pages() and
verifying the poison patterns before alloc_pages().
This generic one cannot detect invalid page accesses immediately but
invalid read access may cause invalid dereference by poisoned memory and
invalid write access can be detected after a long delay.
Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Cc: <linux-arch@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
diff --git a/arch/avr32/mm/fault.c b/arch/avr32/mm/fault.c
index ce4e429..62d4abb 100644
--- a/arch/avr32/mm/fault.c
+++ b/arch/avr32/mm/fault.c
@@ -250,21 +250,3 @@
dump_dtlb();
die("Bus Error", regs, SIGKILL);
}
-
-/*
- * This functionality is currently not possible to implement because
- * we're using segmentation to ensure a fixed mapping of the kernel
- * virtual address space.
- *
- * It would be possible to implement this, but it would require us to
- * disable segmentation at startup and load the kernel mappings into
- * the TLB like any other pages. There will be lots of trickery to
- * avoid recursive invocation of the TLB miss handler, though...
- */
-#ifdef CONFIG_DEBUG_PAGEALLOC
-void kernel_map_pages(struct page *page, int numpages, int enable)
-{
-
-}
-EXPORT_SYMBOL(kernel_map_pages);
-#endif
diff --git a/arch/powerpc/Kconfig b/arch/powerpc/Kconfig
index ad6b1c0..45192dc 100644
--- a/arch/powerpc/Kconfig
+++ b/arch/powerpc/Kconfig
@@ -228,6 +228,9 @@
depends on PPC64 # not supported on 32 bits yet
default n
+config ARCH_SUPPORTS_DEBUG_PAGEALLOC
+ def_bool y
+
source "init/Kconfig"
source "kernel/Kconfig.freezer"
diff --git a/arch/powerpc/Kconfig.debug b/arch/powerpc/Kconfig.debug
index 22091bb..6aa0b5e 100644
--- a/arch/powerpc/Kconfig.debug
+++ b/arch/powerpc/Kconfig.debug
@@ -30,6 +30,7 @@
config DEBUG_PAGEALLOC
bool "Debug page memory allocations"
depends on DEBUG_KERNEL && !HIBERNATION
+ depends on ARCH_SUPPORTS_DEBUG_PAGEALLOC
help
Unmap pages from the kernel linear mapping after free_pages().
This results in a large slowdown, but helps to find certain types
diff --git a/arch/s390/Kconfig b/arch/s390/Kconfig
index 2a8af5e..dcb667c 100644
--- a/arch/s390/Kconfig
+++ b/arch/s390/Kconfig
@@ -72,6 +72,9 @@
config VIRT_CPU_ACCOUNTING
def_bool y
+config ARCH_SUPPORTS_DEBUG_PAGEALLOC
+ def_bool y
+
mainmenu "Linux Kernel Configuration"
config S390
diff --git a/arch/s390/Kconfig.debug b/arch/s390/Kconfig.debug
index 4599fa0..7e297a3 100644
--- a/arch/s390/Kconfig.debug
+++ b/arch/s390/Kconfig.debug
@@ -9,6 +9,7 @@
config DEBUG_PAGEALLOC
bool "Debug page memory allocations"
depends on DEBUG_KERNEL
+ depends on ARCH_SUPPORTS_DEBUG_PAGEALLOC
help
Unmap pages from the kernel linear mapping after free_pages().
This results in a slowdown, but helps to find certain types of
diff --git a/arch/sparc/Kconfig b/arch/sparc/Kconfig
index c3ea215..cc12cd4 100644
--- a/arch/sparc/Kconfig
+++ b/arch/sparc/Kconfig
@@ -124,6 +124,9 @@
config OF
def_bool y
+config ARCH_SUPPORTS_DEBUG_PAGEALLOC
+ def_bool y if SPARC64
+
source "init/Kconfig"
source "kernel/Kconfig.freezer"
diff --git a/arch/sparc/Kconfig.debug b/arch/sparc/Kconfig.debug
index b8a15e2..d001b42 100644
--- a/arch/sparc/Kconfig.debug
+++ b/arch/sparc/Kconfig.debug
@@ -24,7 +24,8 @@
config DEBUG_PAGEALLOC
bool "Debug page memory allocations"
- depends on SPARC64 && DEBUG_KERNEL && !HIBERNATION
+ depends on DEBUG_KERNEL && !HIBERNATION
+ depends on ARCH_SUPPORTS_DEBUG_PAGEALLOC
help
Unmap pages from the kernel linear mapping after free_pages().
This results in a large slowdown, but helps to find certain types
diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
index 45161b8..748e50a 100644
--- a/arch/x86/Kconfig
+++ b/arch/x86/Kconfig
@@ -165,6 +165,9 @@
config ARCH_SUPPORTS_OPTIMIZED_INLINING
def_bool y
+config ARCH_SUPPORTS_DEBUG_PAGEALLOC
+ def_bool y
+
# Use the generic interrupt handling code in kernel/irq/:
config GENERIC_HARDIRQS
bool
diff --git a/arch/x86/Kconfig.debug b/arch/x86/Kconfig.debug
index fdb45df..a345cb5 100644
--- a/arch/x86/Kconfig.debug
+++ b/arch/x86/Kconfig.debug
@@ -75,6 +75,7 @@
config DEBUG_PAGEALLOC
bool "Debug page memory allocations"
depends on DEBUG_KERNEL
+ depends on ARCH_SUPPORTS_DEBUG_PAGEALLOC
---help---
Unmap pages from the kernel linear mapping after free_pages().
This results in a large slowdown, but helps to find certain types
diff --git a/include/linux/mm_types.h b/include/linux/mm_types.h
index d84feb7..ddadb4d 100644
--- a/include/linux/mm_types.h
+++ b/include/linux/mm_types.h
@@ -11,6 +11,7 @@
#include <linux/rwsem.h>
#include <linux/completion.h>
#include <linux/cpumask.h>
+#include <linux/page-debug-flags.h>
#include <asm/page.h>
#include <asm/mmu.h>
@@ -174,6 +175,9 @@
#ifdef CONFIG_NUMA
struct mempolicy *vm_policy; /* NUMA policy for the VMA */
#endif
+#ifdef CONFIG_WANT_PAGE_DEBUG_FLAGS
+ unsigned long debug_flags; /* Use atomic bitops on this */
+#endif
};
struct core_thread {
diff --git a/include/linux/page-debug-flags.h b/include/linux/page-debug-flags.h
new file mode 100644
index 0000000..b0638fd9
--- /dev/null
+++ b/include/linux/page-debug-flags.h
@@ -0,0 +1,30 @@
+#ifndef LINUX_PAGE_DEBUG_FLAGS_H
+#define LINUX_PAGE_DEBUG_FLAGS_H
+
+/*
+ * page->debug_flags bits:
+ *
+ * PAGE_DEBUG_FLAG_POISON is set for poisoned pages. This is used to
+ * implement generic debug pagealloc feature. The pages are filled with
+ * poison patterns and set this flag after free_pages(). The poisoned
+ * pages are verified whether the patterns are not corrupted and clear
+ * the flag before alloc_pages().
+ */
+
+enum page_debug_flags {
+ PAGE_DEBUG_FLAG_POISON, /* Page is poisoned */
+};
+
+/*
+ * Ensure that CONFIG_WANT_PAGE_DEBUG_FLAGS reliably
+ * gets turned off when no debug features are enabling it!
+ */
+
+#ifdef CONFIG_WANT_PAGE_DEBUG_FLAGS
+#if !defined(CONFIG_PAGE_POISONING) \
+/* && !defined(CONFIG_PAGE_DEBUG_SOMETHING_ELSE) && ... */
+#error WANT_PAGE_DEBUG_FLAGS is turned on with no debug features!
+#endif
+#endif /* CONFIG_WANT_PAGE_DEBUG_FLAGS */
+
+#endif /* LINUX_PAGE_DEBUG_FLAGS_H */
diff --git a/include/linux/poison.h b/include/linux/poison.h
index 9f31683..6729f7d 100644
--- a/include/linux/poison.h
+++ b/include/linux/poison.h
@@ -17,6 +17,9 @@
*/
#define TIMER_ENTRY_STATIC ((void *) 0x74737461)
+/********** mm/debug-pagealloc.c **********/
+#define PAGE_POISON 0xaa
+
/********** mm/slab.c **********/
/*
* Magic nums for obj red zoning.
diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug
index 58bfe7e..9638d99 100644
--- a/lib/Kconfig.debug
+++ b/lib/Kconfig.debug
@@ -796,6 +796,7 @@
to properly maintain and use. This enables checks that help
you to keep things correct.
+source mm/Kconfig.debug
source kernel/trace/Kconfig
config PROVIDE_OHCI1394_DMA_INIT
diff --git a/mm/Kconfig.debug b/mm/Kconfig.debug
new file mode 100644
index 0000000..c8d62d4
--- /dev/null
+++ b/mm/Kconfig.debug
@@ -0,0 +1,17 @@
+config WANT_PAGE_DEBUG_FLAGS
+ bool
+
+config PAGE_POISONING
+ bool "Debug page memory allocations"
+ depends on DEBUG_KERNEL && !ARCH_SUPPORTS_DEBUG_PAGEALLOC
+ depends on !HIBERNATION
+ select DEBUG_PAGEALLOC
+ select WANT_PAGE_DEBUG_FLAGS
+ help
+ Fill the pages with poison patterns after free_pages() and verify
+ the patterns before alloc_pages(). This results in a large slowdown,
+ but helps to find certain types of memory corruptions.
+
+ This option cannot enalbe with hibernation. Otherwise, it will get
+ wrong messages for memory corruption because the free pages are not
+ saved to the suspend image.
diff --git a/mm/Makefile b/mm/Makefile
index 818569b6..ec73c68 100644
--- a/mm/Makefile
+++ b/mm/Makefile
@@ -24,6 +24,7 @@
obj-$(CONFIG_TMPFS_POSIX_ACL) += shmem_acl.o
obj-$(CONFIG_SLOB) += slob.o
obj-$(CONFIG_MMU_NOTIFIER) += mmu_notifier.o
+obj-$(CONFIG_PAGE_POISONING) += debug-pagealloc.o
obj-$(CONFIG_SLAB) += slab.o
obj-$(CONFIG_SLUB) += slub.o
obj-$(CONFIG_FAILSLAB) += failslab.o
diff --git a/mm/debug-pagealloc.c b/mm/debug-pagealloc.c
new file mode 100644
index 0000000..a1e3324
--- /dev/null
+++ b/mm/debug-pagealloc.c
@@ -0,0 +1,129 @@
+#include <linux/kernel.h>
+#include <linux/mm.h>
+#include <linux/page-debug-flags.h>
+#include <linux/poison.h>
+
+static inline void set_page_poison(struct page *page)
+{
+ __set_bit(PAGE_DEBUG_FLAG_POISON, &page->debug_flags);
+}
+
+static inline void clear_page_poison(struct page *page)
+{
+ __clear_bit(PAGE_DEBUG_FLAG_POISON, &page->debug_flags);
+}
+
+static inline bool page_poison(struct page *page)
+{
+ return test_bit(PAGE_DEBUG_FLAG_POISON, &page->debug_flags);
+}
+
+static void poison_highpage(struct page *page)
+{
+ /*
+ * Page poisoning for highmem pages is not implemented.
+ *
+ * This can be called from interrupt contexts.
+ * So we need to create a new kmap_atomic slot for this
+ * application and it will need interrupt protection.
+ */
+}
+
+static void poison_page(struct page *page)
+{
+ void *addr;
+
+ if (PageHighMem(page)) {
+ poison_highpage(page);
+ return;
+ }
+ set_page_poison(page);
+ addr = page_address(page);
+ memset(addr, PAGE_POISON, PAGE_SIZE);
+}
+
+static void poison_pages(struct page *page, int n)
+{
+ int i;
+
+ for (i = 0; i < n; i++)
+ poison_page(page + i);
+}
+
+static bool single_bit_flip(unsigned char a, unsigned char b)
+{
+ unsigned char error = a ^ b;
+
+ return error && !(error & (error - 1));
+}
+
+static void check_poison_mem(unsigned char *mem, size_t bytes)
+{
+ unsigned char *start;
+ unsigned char *end;
+
+ for (start = mem; start < mem + bytes; start++) {
+ if (*start != PAGE_POISON)
+ break;
+ }
+ if (start == mem + bytes)
+ return;
+
+ for (end = mem + bytes - 1; end > start; end--) {
+ if (*end != PAGE_POISON)
+ break;
+ }
+
+ if (!printk_ratelimit())
+ return;
+ else if (start == end && single_bit_flip(*start, PAGE_POISON))
+ printk(KERN_ERR "pagealloc: single bit error\n");
+ else
+ printk(KERN_ERR "pagealloc: memory corruption\n");
+
+ print_hex_dump(KERN_ERR, "", DUMP_PREFIX_ADDRESS, 16, 1, start,
+ end - start + 1, 1);
+ dump_stack();
+}
+
+static void unpoison_highpage(struct page *page)
+{
+ /*
+ * See comment in poison_highpage().
+ * Highmem pages should not be poisoned for now
+ */
+ BUG_ON(page_poison(page));
+}
+
+static void unpoison_page(struct page *page)
+{
+ if (PageHighMem(page)) {
+ unpoison_highpage(page);
+ return;
+ }
+ if (page_poison(page)) {
+ void *addr = page_address(page);
+
+ check_poison_mem(addr, PAGE_SIZE);
+ clear_page_poison(page);
+ }
+}
+
+static void unpoison_pages(struct page *page, int n)
+{
+ int i;
+
+ for (i = 0; i < n; i++)
+ unpoison_page(page + i);
+}
+
+void kernel_map_pages(struct page *page, int numpages, int enable)
+{
+ if (!debug_pagealloc_enabled)
+ return;
+
+ if (enable)
+ unpoison_pages(page, numpages);
+ else
+ poison_pages(page, numpages);
+}