x86/kasan, mm: Introduce generic kasan_populate_zero_shadow()
authorAndrey Ryabinin <ryabinin.a.a@gmail.com>
Thu, 13 Aug 2015 05:37:24 +0000 (08:37 +0300)
committerIngo Molnar <mingo@kernel.org>
Sat, 22 Aug 2015 12:54:55 +0000 (14:54 +0200)
Introduce generic kasan_populate_zero_shadow(shadow_start,
shadow_end). This function maps kasan_zero_page to the
[shadow_start, shadow_end] addresses.

This replaces x86_64 specific populate_zero_shadow() and will
be used for ARM64 in follow on patches.

The main changes from original version are:

 * Use p?d_populate*() instead of set_p?d()
 * Use memblock allocator directly instead of vmemmap_alloc_block()
 * __pa() instead of __pa_nodebug(). __pa() causes troubles
   iff we use it before kasan_early_init(). kasan_populate_zero_shadow()
   will be used later, so we ok with __pa() here.

Signed-off-by: Andrey Ryabinin <ryabinin.a.a@gmail.com>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Alexey Klimov <klimov.linux@gmail.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: David Keitel <dkeitel@codeaurora.org>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Linus Walleij <linus.walleij@linaro.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Yury <yury.norov@gmail.com>
Cc: linux-arm-kernel@lists.infradead.org
Cc: linux-mm@kvack.org
Link: http://lkml.kernel.org/r/1439444244-26057-3-git-send-email-ryabinin.a.a@gmail.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
arch/x86/mm/kasan_init_64.c
include/linux/kasan.h
mm/kasan/Makefile
mm/kasan/kasan_init.c [new file with mode: 0644]

index e1840f3db5b583984e53bcc7d11787706dbbed26..9ce5da27b136beec916e5f1ad19c88ba493bb5f5 100644 (file)
 extern pgd_t early_level4_pgt[PTRS_PER_PGD];
 extern struct range pfn_mapped[E820_X_MAX];
 
-static pud_t kasan_zero_pud[PTRS_PER_PUD] __page_aligned_bss;
-static pmd_t kasan_zero_pmd[PTRS_PER_PMD] __page_aligned_bss;
-static pte_t kasan_zero_pte[PTRS_PER_PTE] __page_aligned_bss;
-
-/*
- * This page used as early shadow. We don't use empty_zero_page
- * at early stages, stack instrumentation could write some garbage
- * to this page.
- * Latter we reuse it as zero shadow for large ranges of memory
- * that allowed to access, but not instrumented by kasan
- * (vmalloc/vmemmap ...).
- */
-static unsigned char kasan_zero_page[PAGE_SIZE] __page_aligned_bss;
-
 static int __init map_range(struct range *range)
 {
        unsigned long start;
@@ -62,106 +48,6 @@ static void __init kasan_map_early_shadow(pgd_t *pgd)
        }
 }
 
-static int __init zero_pte_populate(pmd_t *pmd, unsigned long addr,
-                               unsigned long end)
-{
-       pte_t *pte = pte_offset_kernel(pmd, addr);
-
-       while (addr + PAGE_SIZE <= end) {
-               WARN_ON(!pte_none(*pte));
-               set_pte(pte, __pte(__pa_nodebug(kasan_zero_page)
-                                       | __PAGE_KERNEL_RO));
-               addr += PAGE_SIZE;
-               pte = pte_offset_kernel(pmd, addr);
-       }
-       return 0;
-}
-
-static int __init zero_pmd_populate(pud_t *pud, unsigned long addr,
-                               unsigned long end)
-{
-       int ret = 0;
-       pmd_t *pmd = pmd_offset(pud, addr);
-
-       while (IS_ALIGNED(addr, PMD_SIZE) && addr + PMD_SIZE <= end) {
-               WARN_ON(!pmd_none(*pmd));
-               set_pmd(pmd, __pmd(__pa_nodebug(kasan_zero_pte)
-                                       | _KERNPG_TABLE));
-               addr += PMD_SIZE;
-               pmd = pmd_offset(pud, addr);
-       }
-       if (addr < end) {
-               if (pmd_none(*pmd)) {
-                       void *p = vmemmap_alloc_block(PAGE_SIZE, NUMA_NO_NODE);
-                       if (!p)
-                               return -ENOMEM;
-                       set_pmd(pmd, __pmd(__pa_nodebug(p) | _KERNPG_TABLE));
-               }
-               ret = zero_pte_populate(pmd, addr, end);
-       }
-       return ret;
-}
-
-
-static int __init zero_pud_populate(pgd_t *pgd, unsigned long addr,
-                               unsigned long end)
-{
-       int ret = 0;
-       pud_t *pud = pud_offset(pgd, addr);
-
-       while (IS_ALIGNED(addr, PUD_SIZE) && addr + PUD_SIZE <= end) {
-               WARN_ON(!pud_none(*pud));
-               set_pud(pud, __pud(__pa_nodebug(kasan_zero_pmd)
-                                       | _KERNPG_TABLE));
-               addr += PUD_SIZE;
-               pud = pud_offset(pgd, addr);
-       }
-
-       if (addr < end) {
-               if (pud_none(*pud)) {
-                       void *p = vmemmap_alloc_block(PAGE_SIZE, NUMA_NO_NODE);
-                       if (!p)
-                               return -ENOMEM;
-                       set_pud(pud, __pud(__pa_nodebug(p) | _KERNPG_TABLE));
-               }
-               ret = zero_pmd_populate(pud, addr, end);
-       }
-       return ret;
-}
-
-static int __init zero_pgd_populate(unsigned long addr, unsigned long end)
-{
-       int ret = 0;
-       pgd_t *pgd = pgd_offset_k(addr);
-
-       while (IS_ALIGNED(addr, PGDIR_SIZE) && addr + PGDIR_SIZE <= end) {
-               WARN_ON(!pgd_none(*pgd));
-               set_pgd(pgd, __pgd(__pa_nodebug(kasan_zero_pud)
-                                       | _KERNPG_TABLE));
-               addr += PGDIR_SIZE;
-               pgd = pgd_offset_k(addr);
-       }
-
-       if (addr < end) {
-               if (pgd_none(*pgd)) {
-                       void *p = vmemmap_alloc_block(PAGE_SIZE, NUMA_NO_NODE);
-                       if (!p)
-                               return -ENOMEM;
-                       set_pgd(pgd, __pgd(__pa_nodebug(p) | _KERNPG_TABLE));
-               }
-               ret = zero_pud_populate(pgd, addr, end);
-       }
-       return ret;
-}
-
-
-static void __init populate_zero_shadow(const void *start, const void *end)
-{
-       if (zero_pgd_populate((unsigned long)start, (unsigned long)end))
-               panic("kasan: unable to map zero shadow!");
-}
-
-
 #ifdef CONFIG_KASAN_INLINE
 static int kasan_die_handler(struct notifier_block *self,
                             unsigned long val,
@@ -213,7 +99,7 @@ void __init kasan_init(void)
 
        clear_pgds(KASAN_SHADOW_START, KASAN_SHADOW_END);
 
-       populate_zero_shadow((void *)KASAN_SHADOW_START,
+       kasan_populate_zero_shadow((void *)KASAN_SHADOW_START,
                        kasan_mem_to_shadow((void *)PAGE_OFFSET));
 
        for (i = 0; i < E820_X_MAX; i++) {
@@ -223,14 +109,15 @@ void __init kasan_init(void)
                if (map_range(&pfn_mapped[i]))
                        panic("kasan: unable to allocate shadow!");
        }
-       populate_zero_shadow(kasan_mem_to_shadow((void *)PAGE_OFFSET + MAXMEM),
-                       kasan_mem_to_shadow((void *)__START_KERNEL_map));
+       kasan_populate_zero_shadow(
+               kasan_mem_to_shadow((void *)PAGE_OFFSET + MAXMEM),
+               kasan_mem_to_shadow((void *)__START_KERNEL_map));
 
        vmemmap_populate((unsigned long)kasan_mem_to_shadow(_stext),
                        (unsigned long)kasan_mem_to_shadow(_end),
                        NUMA_NO_NODE);
 
-       populate_zero_shadow(kasan_mem_to_shadow((void *)MODULES_END),
+       kasan_populate_zero_shadow(kasan_mem_to_shadow((void *)MODULES_END),
                        (void *)KASAN_SHADOW_END);
 
        memset(kasan_zero_page, 0, PAGE_SIZE);
index 6fb1c7d4292c0bbbe8ef528f5d958a9659e20e3b..4b9f85c963d0738e20ee7c89e2dcb28ba3cc1aaa 100644 (file)
@@ -12,8 +12,17 @@ struct vm_struct;
 #define KASAN_SHADOW_SCALE_SHIFT 3
 
 #include <asm/kasan.h>
+#include <asm/pgtable.h>
 #include <linux/sched.h>
 
+extern unsigned char kasan_zero_page[PAGE_SIZE];
+extern pte_t kasan_zero_pte[PTRS_PER_PTE];
+extern pmd_t kasan_zero_pmd[PTRS_PER_PMD];
+extern pud_t kasan_zero_pud[PTRS_PER_PUD];
+
+void kasan_populate_zero_shadow(const void *shadow_start,
+                               const void *shadow_end);
+
 static inline void *kasan_mem_to_shadow(const void *addr)
 {
        return (void *)((unsigned long)addr >> KASAN_SHADOW_SCALE_SHIFT)
index bd837b8c2f4126ec1911f43c39edcc8831ad8f50..64710148941ed33d7467192797711c5c98e9e4a3 100644 (file)
@@ -5,4 +5,4 @@ CFLAGS_REMOVE_kasan.o = -pg
 # see: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63533
 CFLAGS_kasan.o := $(call cc-option, -fno-conserve-stack -fno-stack-protector)
 
-obj-y := kasan.o report.o
+obj-y := kasan.o report.o kasan_init.o
diff --git a/mm/kasan/kasan_init.c b/mm/kasan/kasan_init.c
new file mode 100644 (file)
index 0000000..3f9a41c
--- /dev/null
@@ -0,0 +1,152 @@
+/*
+ * This file contains some kasan initialization code.
+ *
+ * Copyright (c) 2015 Samsung Electronics Co., Ltd.
+ * Author: Andrey Ryabinin <ryabinin.a.a@gmail.com>
+ *
+ * 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/bootmem.h>
+#include <linux/init.h>
+#include <linux/kasan.h>
+#include <linux/kernel.h>
+#include <linux/memblock.h>
+#include <linux/pfn.h>
+
+#include <asm/page.h>
+#include <asm/pgalloc.h>
+
+/*
+ * This page serves two purposes:
+ *   - It used as early shadow memory. The entire shadow region populated
+ *     with this page, before we will be able to setup normal shadow memory.
+ *   - Latter it reused it as zero shadow to cover large ranges of memory
+ *     that allowed to access, but not handled by kasan (vmalloc/vmemmap ...).
+ */
+unsigned char kasan_zero_page[PAGE_SIZE] __page_aligned_bss;
+
+#if CONFIG_PGTABLE_LEVELS > 3
+pud_t kasan_zero_pud[PTRS_PER_PUD] __page_aligned_bss;
+#endif
+#if CONFIG_PGTABLE_LEVELS > 2
+pmd_t kasan_zero_pmd[PTRS_PER_PMD] __page_aligned_bss;
+#endif
+pte_t kasan_zero_pte[PTRS_PER_PTE] __page_aligned_bss;
+
+static __init void *early_alloc(size_t size, int node)
+{
+       return memblock_virt_alloc_try_nid(size, size, __pa(MAX_DMA_ADDRESS),
+                                       BOOTMEM_ALLOC_ACCESSIBLE, node);
+}
+
+static void __init zero_pte_populate(pmd_t *pmd, unsigned long addr,
+                               unsigned long end)
+{
+       pte_t *pte = pte_offset_kernel(pmd, addr);
+       pte_t zero_pte;
+
+       zero_pte = pfn_pte(PFN_DOWN(__pa(kasan_zero_page)), PAGE_KERNEL);
+       zero_pte = pte_wrprotect(zero_pte);
+
+       while (addr + PAGE_SIZE <= end) {
+               set_pte_at(&init_mm, addr, pte, zero_pte);
+               addr += PAGE_SIZE;
+               pte = pte_offset_kernel(pmd, addr);
+       }
+}
+
+static void __init zero_pmd_populate(pud_t *pud, unsigned long addr,
+                               unsigned long end)
+{
+       pmd_t *pmd = pmd_offset(pud, addr);
+       unsigned long next;
+
+       do {
+               next = pmd_addr_end(addr, end);
+
+               if (IS_ALIGNED(addr, PMD_SIZE) && end - addr >= PMD_SIZE) {
+                       pmd_populate_kernel(&init_mm, pmd, kasan_zero_pte);
+                       continue;
+               }
+
+               if (pmd_none(*pmd)) {
+                       pmd_populate_kernel(&init_mm, pmd,
+                                       early_alloc(PAGE_SIZE, NUMA_NO_NODE));
+               }
+               zero_pte_populate(pmd, addr, next);
+       } while (pmd++, addr = next, addr != end);
+}
+
+static void __init zero_pud_populate(pgd_t *pgd, unsigned long addr,
+                               unsigned long end)
+{
+       pud_t *pud = pud_offset(pgd, addr);
+       unsigned long next;
+
+       do {
+               next = pud_addr_end(addr, end);
+               if (IS_ALIGNED(addr, PUD_SIZE) && end - addr >= PUD_SIZE) {
+                       pmd_t *pmd;
+
+                       pud_populate(&init_mm, pud, kasan_zero_pmd);
+                       pmd = pmd_offset(pud, addr);
+                       pmd_populate_kernel(&init_mm, pmd, kasan_zero_pte);
+                       continue;
+               }
+
+               if (pud_none(*pud)) {
+                       pud_populate(&init_mm, pud,
+                               early_alloc(PAGE_SIZE, NUMA_NO_NODE));
+               }
+               zero_pmd_populate(pud, addr, next);
+       } while (pud++, addr = next, addr != end);
+}
+
+/**
+ * kasan_populate_zero_shadow - populate shadow memory region with
+ *                               kasan_zero_page
+ * @shadow_start - start of the memory range to populate
+ * @shadow_end   - end of the memory range to populate
+ */
+void __init kasan_populate_zero_shadow(const void *shadow_start,
+                               const void *shadow_end)
+{
+       unsigned long addr = (unsigned long)shadow_start;
+       unsigned long end = (unsigned long)shadow_end;
+       pgd_t *pgd = pgd_offset_k(addr);
+       unsigned long next;
+
+       do {
+               next = pgd_addr_end(addr, end);
+
+               if (IS_ALIGNED(addr, PGDIR_SIZE) && end - addr >= PGDIR_SIZE) {
+                       pud_t *pud;
+                       pmd_t *pmd;
+
+                       /*
+                        * kasan_zero_pud should be populated with pmds
+                        * at this moment.
+                        * [pud,pmd]_populate*() below needed only for
+                        * 3,2 - level page tables where we don't have
+                        * puds,pmds, so pgd_populate(), pud_populate()
+                        * is noops.
+                        */
+                       pgd_populate(&init_mm, pgd, kasan_zero_pud);
+                       pud = pud_offset(pgd, addr);
+                       pud_populate(&init_mm, pud, kasan_zero_pmd);
+                       pmd = pmd_offset(pud, addr);
+                       pmd_populate_kernel(&init_mm, pmd, kasan_zero_pte);
+                       continue;
+               }
+
+               if (pgd_none(*pgd)) {
+                       pgd_populate(&init_mm, pgd,
+                               early_alloc(PAGE_SIZE, NUMA_NO_NODE));
+               }
+               zero_pud_populate(pgd, addr, next);
+       } while (pgd++, addr = next, addr != end);
+}