Add 3 new tests for tag-based KASAN modes:

1. Check that match-all pointer tag is not assigned randomly.
2. Check that 0xff works as a match-all pointer tag.
3. Check that there are no match-all memory tags.

Note, that test #3 causes a significant number (255) of KASAN reports
to be printed during execution for the SW_TAGS mode.

Signed-off-by: Andrey Konovalov <andre...@google.com>
Link: https://linux-review.googlesource.com/id/I78f1375efafa162b37f3abcb2c5bc2f3955dfd8e
---
lib/test_kasan.c | 93 ++++++++++++++++++++++++++++++++++++++++++++++++
mm/kasan/kasan.h | 6 ++++
2 files changed, 99 insertions(+)

diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index 46e578c8e842..f1eda0bcc780 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c
@@ -13,6 +13,7 @@
#include <linux/mman.h>
#include <linux/module.h>
#include <linux/printk.h>
+#include <linux/random.h>
#include <linux/slab.h>
#include <linux/string.h>
#include <linux/uaccess.h>
@@ -790,6 +791,95 @@ static void vmalloc_oob(struct kunit *test)
vfree(area);
}

+/*
+ * Check that match-all pointer tag is not assigned randomly for
+ * tag-based modes.
+ */
+static void match_all_not_assigned(struct kunit *test)
+{
+ char *ptr;
+ struct page *pages;
+ int i, size, order;
+
+ for (i = 0; i < 256; i++) {
+ size = get_random_int() % KMALLOC_MAX_SIZE;
+ ptr = kmalloc(128, GFP_KERNEL);
+ KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);
+ KUNIT_EXPECT_NE(test, (u8)get_tag(ptr), (u8)KASAN_TAG_KERNEL);
+ kfree(ptr);
+ }
+
+ for (i = 0; i < 256; i++) {
+ order = get_random_int() % 4;
+ pages = alloc_pages(GFP_KERNEL, order);
+ ptr = page_address(pages);
+ KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);
+ KUNIT_EXPECT_NE(test, (u8)get_tag(ptr), (u8)KASAN_TAG_KERNEL);
+ free_pages((unsigned long)ptr, order);
+ }
+}
+
+/* Check that 0xff works as a match-all pointer tag for tag-based modes. */
+static void match_all_ptr_tag(struct kunit *test)
+{
+ char *ptr;
+ u8 tag;
+
+ if (IS_ENABLED(CONFIG_KASAN_GENERIC)) {
+ kunit_info(test, "skipping, CONFIG_KASAN_SW/HW_TAGS required");
+ return;
+ }
+
+ ptr = kmalloc(128, GFP_KERNEL);
+ KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);
+
+ /* Backup the assigned tag. */
+ tag = get_tag(ptr);
+ KUNIT_EXPECT_NE(test, tag, (u8)KASAN_TAG_KERNEL);
+
+ /* Reset the tag to 0xff.*/
+ ptr = set_tag(ptr, KASAN_TAG_KERNEL);

+
+ /* This access shouldn't trigger a KASAN report. */

+ *ptr = 0;
+
+ /* Recover the pointer tag and free. */
+ ptr = set_tag(ptr, tag);
+ kfree(ptr);
+}
+
+/* Check that there are no match-all memory tags for tag-based modes. */
+static void match_all_mem_tag(struct kunit *test)
+{
+ char *ptr;
+ int tag;
+
+ if (IS_ENABLED(CONFIG_KASAN_GENERIC)) {
+ kunit_info(test, "skipping, CONFIG_KASAN_SW/HW_TAGS required");
+ return;
+ }
+
+ ptr = kmalloc(128, GFP_KERNEL);
+ KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);
+ KUNIT_EXPECT_NE(test, (u8)get_tag(ptr), (u8)KASAN_TAG_KERNEL);
+
+ /* For each possible tag value not matching the pointer tag. */
+ for (tag = KASAN_TAG_MIN; tag <= KASAN_TAG_KERNEL; tag++) {
+ if (tag == get_tag(ptr))
+ continue;
+
+ /* Mark the first memory granule with the chosen memory tag. */
+ kasan_poison(ptr, KASAN_GRANULE_SIZE, (u8)tag);
+
+ /* This access must cause a KASAN report. */
+ KUNIT_EXPECT_KASAN_FAIL(test, *ptr = 0);
+ }
+
+ /* Recover the memory tag and free. */
+ kasan_poison(ptr, KASAN_GRANULE_SIZE, get_tag(ptr));
+ kfree(ptr);
+}
+
static struct kunit_case kasan_kunit_test_cases[] = {
KUNIT_CASE(kmalloc_oob_right),
KUNIT_CASE(kmalloc_oob_left),
@@ -829,6 +919,9 @@ static struct kunit_case kasan_kunit_test_cases[] = {
KUNIT_CASE(kasan_bitops_tags),
KUNIT_CASE(kmalloc_double_kzfree),
KUNIT_CASE(vmalloc_oob),
+ KUNIT_CASE(match_all_not_assigned),
+ KUNIT_CASE(match_all_ptr_tag),
+ KUNIT_CASE(match_all_mem_tag),
{}
};

diff --git a/mm/kasan/kasan.h b/mm/kasan/kasan.h
index 3b38baddec47..c3fb9bf241d3 100644
--- a/mm/kasan/kasan.h
+++ b/mm/kasan/kasan.h
@@ -36,6 +36,12 @@ extern bool kasan_flag_panic __ro_after_init;
#define KASAN_TAG_INVALID 0xFE /* inaccessible memory tag */
#define KASAN_TAG_MAX 0xFD /* maximum value for random tags */

+#ifdef CONFIG_KASAN_HW_TAGS
+#define KASAN_TAG_MIN 0xF0 /* mimimum value for random tags */
+#else
+#define KASAN_TAG_MIN 0x00 /* mimimum value for random tags */
+#endif
+
#ifdef CONFIG_KASAN_GENERIC
#define KASAN_FREE_PAGE 0xFF /* page was freed */
#define KASAN_PAGE_REDZONE 0xFE /* redzone for kmalloc_large allocations */
--
2.29.2.729.g45daf8777d-goog

Andrey Konovalov

unread,

Jan 5, 2021, 1:28:16 PM1/5/21

to Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

On a high level, this patch allows running KUnit KASAN tests with the
hardware tag-based KASAN mode.

Internally, this change reenables tag checking at the end of each KASAN
test that triggers a tag fault and leads to tag checking being disabled.

With this patch KASAN tests are still failing for the hardware tag-based
mode; fixes come in the next few patches.

Signed-off-by: Andrey Konovalov <andre...@google.com>
Link: https://linux-review.googlesource.com/id/Id94dc9eccd33b23cda4950be408c27f879e474c8
---
arch/arm64/include/asm/memory.h | 1 +
arch/arm64/include/asm/mte-kasan.h | 12 +++++++++
arch/arm64/kernel/mte.c | 12 +++++++++
arch/arm64/mm/fault.c | 16 +++++++-----
lib/Kconfig.kasan | 4 +--
lib/test_kasan.c | 42 +++++++++++++++++++++---------
mm/kasan/kasan.h | 9 +++++++
7 files changed, 75 insertions(+), 21 deletions(-)

diff --git a/arch/arm64/include/asm/memory.h b/arch/arm64/include/asm/memory.h
index 18fce223b67b..cedfc9e97bcc 100644
--- a/arch/arm64/include/asm/memory.h
+++ b/arch/arm64/include/asm/memory.h
@@ -232,6 +232,7 @@ static inline const void *__tag_set(const void *addr, u8 tag)

#ifdef CONFIG_KASAN_HW_TAGS
#define arch_enable_tagging() mte_enable_kernel()
+#define arch_set_tagging_report_once(state) mte_set_report_once(state)
#define arch_init_tags(max_tag) mte_init_tags(max_tag)
#define arch_get_random_tag() mte_get_random_tag()
#define arch_get_mem_tag(addr) mte_get_mem_tag(addr)
diff --git a/arch/arm64/include/asm/mte-kasan.h b/arch/arm64/include/asm/mte-kasan.h
index 26349a4b5e2e..3748d5bb88c0 100644
--- a/arch/arm64/include/asm/mte-kasan.h
+++ b/arch/arm64/include/asm/mte-kasan.h
@@ -32,6 +32,9 @@ void *mte_set_mem_tag_range(void *addr, size_t size, u8 tag);
void mte_enable_kernel(void);
void mte_init_tags(u64 max_tag);

+void mte_set_report_once(bool state);
+bool mte_report_once(void);
+
#else /* CONFIG_ARM64_MTE */

static inline u8 mte_get_ptr_tag(void *ptr)
@@ -60,6 +63,15 @@ static inline void mte_init_tags(u64 max_tag)
{
}

+static inline void mte_set_report_once(bool state)
+{
+}
+
+static inline bool mte_report_once(void)
+{
+ return false;
+}
+
#endif /* CONFIG_ARM64_MTE */

#endif /* __ASSEMBLY__ */
diff --git a/arch/arm64/kernel/mte.c b/arch/arm64/kernel/mte.c
index dc9ada64feed..c63b3d7a3cd9 100644
--- a/arch/arm64/kernel/mte.c
+++ b/arch/arm64/kernel/mte.c
@@ -25,6 +25,8 @@

u64 gcr_kernel_excl __ro_after_init;

+static bool report_fault_once = true;
+
static void mte_sync_page_tags(struct page *page, pte_t *ptep, bool check_swap)
{
pte_t old_pte = READ_ONCE(*ptep);
@@ -158,6 +160,16 @@ void mte_enable_kernel(void)
isb();
}

+void mte_set_report_once(bool state)
+{
+ WRITE_ONCE(report_fault_once, state);
+}
+
+bool mte_report_once(void)
+{
+ return READ_ONCE(report_fault_once);
+}
+
static void update_sctlr_el1_tcf0(u64 tcf0)
{
/* ISB required for the kernel uaccess routines */
diff --git a/arch/arm64/mm/fault.c b/arch/arm64/mm/fault.c
index 3c40da479899..57d3f165d907 100644
--- a/arch/arm64/mm/fault.c
+++ b/arch/arm64/mm/fault.c
@@ -302,12 +302,20 @@ static void die_kernel_fault(const char *msg, unsigned long addr,
static void report_tag_fault(unsigned long addr, unsigned int esr,
struct pt_regs *regs)
{
- bool is_write = ((esr & ESR_ELx_WNR) >> ESR_ELx_WNR_SHIFT) != 0;
+ static bool reported;
+ bool is_write;
+
+ if (READ_ONCE(reported))
+ return;
+
+ if (mte_report_once())
+ WRITE_ONCE(reported, true);

/*
* SAS bits aren't set for all faults reported in EL1, so we can't
* find out access size.
*/
+ is_write = ((esr & ESR_ELx_WNR) >> ESR_ELx_WNR_SHIFT) != 0;
kasan_report(addr, 0, is_write, regs->pc);
}
#else
@@ -319,12 +327,8 @@ static inline void report_tag_fault(unsigned long addr, unsigned int esr,
static void do_tag_recovery(unsigned long addr, unsigned int esr,
struct pt_regs *regs)
{
- static bool reported;

- if (!READ_ONCE(reported)) {
- report_tag_fault(addr, esr, regs);
- WRITE_ONCE(reported, true);
- }
+ report_tag_fault(addr, esr, regs);

/*
* Disable MTE Tag Checking on the local CPU for the current EL.
diff --git a/lib/Kconfig.kasan b/lib/Kconfig.kasan
index f5fa4ba126bf..3091432acb0a 100644
--- a/lib/Kconfig.kasan
+++ b/lib/Kconfig.kasan
@@ -190,11 +190,11 @@ config KASAN_KUNIT_TEST
kernel debugging features like KASAN.

For more information on KUnit and unit tests in general, please refer
- to the KUnit documentation in Documentation/dev-tools/kunit
+ to the KUnit documentation in Documentation/dev-tools/kunit.

config TEST_KASAN_MODULE
tristate "KUnit-incompatible tests of KASAN bug detection capabilities"
- depends on m && KASAN
+ depends on m && KASAN && !KASAN_HW_TAGS
help
This is a part of the KASAN test suite that is incompatible with
KUnit. Currently includes tests that do bad copy_from/to_user
diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index f1eda0bcc780..dd3d2f95c24e 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c
@@ -41,16 +41,20 @@ static bool multishot;

/*

* Temporarily enable multi-shot mode. Otherwise, KASAN would only report the

- * first detected bug and panic the kernel if panic_on_warn is enabled.
+ * first detected bug and panic the kernel if panic_on_warn is enabled. For
+ * hardware tag-based KASAN also allow tag checking to be reenabled for each
+ * test, see the comment for KUNIT_EXPECT_KASAN_FAIL().

*/
static int kasan_test_init(struct kunit *test)
{

multishot = kasan_save_enable_multi_shot();
+ hw_set_tagging_report_once(false);
return 0;

}

static void kasan_test_exit(struct kunit *test)

{
+ hw_set_tagging_report_once(true);
kasan_restore_multi_shot(multishot);
}

@@ -59,19 +63,31 @@ static void kasan_test_exit(struct kunit *test)

* KASAN report; causes a test failure otherwise. This relies on a KUnit

* resource named "kasan_data". Do not use this name for KUnit resources

* outside of KASAN tests.

+ *
+ * For hardware tag-based KASAN, when a tag fault happens, tag checking is
+ * normally auto-disabled. When this happens, this test handler reenables
+ * tag checking. As tag checking can be only disabled or enabled per CPU, this
+ * handler disables migration (preemption).
*/
-#define KUNIT_EXPECT_KASAN_FAIL(test, expression) do { \
- fail_data.report_expected = true; \
- fail_data.report_found = false; \
- kunit_add_named_resource(test, \
- NULL, \
- NULL, \
- &resource, \
- "kasan_data", &fail_data); \
- expression; \
- KUNIT_EXPECT_EQ(test, \
- fail_data.report_expected, \
- fail_data.report_found); \
+#define KUNIT_EXPECT_KASAN_FAIL(test, expression) do { \
+ if (IS_ENABLED(CONFIG_KASAN_HW_TAGS)) \
+ migrate_disable(); \
+ fail_data.report_expected = true; \
+ fail_data.report_found = false; \
+ kunit_add_named_resource(test, \
+ NULL, \
+ NULL, \
+ &resource, \
+ "kasan_data", &fail_data); \
+ expression; \
+ KUNIT_EXPECT_EQ(test, \
+ fail_data.report_expected, \
+ fail_data.report_found); \
+ if (IS_ENABLED(CONFIG_KASAN_HW_TAGS)) { \
+ if (fail_data.report_found) \
+ hw_enable_tagging(); \
+ migrate_enable(); \
+ } \
} while (0)

static void kmalloc_oob_right(struct kunit *test)
diff --git a/mm/kasan/kasan.h b/mm/kasan/kasan.h
index c3fb9bf241d3..292dfbc37deb 100644
--- a/mm/kasan/kasan.h
+++ b/mm/kasan/kasan.h
@@ -280,6 +280,9 @@ static inline const void *arch_kasan_set_tag(const void *addr, u8 tag)
#ifndef arch_init_tags
#define arch_init_tags(max_tag)
#endif
+#ifndef arch_set_tagging_report_once
+#define arch_set_tagging_report_once(state)
+#endif
#ifndef arch_get_random_tag
#define arch_get_random_tag() (0xFF)
#endif
@@ -292,10 +295,16 @@ static inline const void *arch_kasan_set_tag(const void *addr, u8 tag)

#define hw_enable_tagging() arch_enable_tagging()
#define hw_init_tags(max_tag) arch_init_tags(max_tag)
+#define hw_set_tagging_report_once(state) arch_set_tagging_report_once(state)
#define hw_get_random_tag() arch_get_random_tag()
#define hw_get_mem_tag(addr) arch_get_mem_tag(addr)
#define hw_set_mem_tag_range(addr, size, tag) arch_set_mem_tag_range((addr), (size), (tag))

+#else /* CONFIG_KASAN_HW_TAGS */
+
+#define hw_enable_tagging()
+#define hw_set_tagging_report_once(state)
+

#endif /* CONFIG_KASAN_HW_TAGS */

#ifdef CONFIG_KASAN_SW_TAGS

--
2.29.2.729.g45daf8777d-goog

Andrey Konovalov

unread,

Jan 5, 2021, 1:28:19 PM1/5/21

to Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

Rename CONFIG_TEST_KASAN_MODULE to CONFIG_KASAN_MODULE_TEST.

This naming is more consistent with the existing CONFIG_KASAN_KUNIT_TEST.

Signed-off-by: Andrey Konovalov <andre...@google.com>
Link: https://linux-review.googlesource.com/id/Id347dfa5fe8788b7a1a189863e039f409da0ae5f
---
Documentation/dev-tools/kasan.rst | 6 +++---
lib/Kconfig.kasan | 2 +-
lib/Makefile | 2 +-
3 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/Documentation/dev-tools/kasan.rst b/Documentation/dev-tools/kasan.rst
index 26c99852a852..72535816145d 100644
--- a/Documentation/dev-tools/kasan.rst
+++ b/Documentation/dev-tools/kasan.rst
@@ -374,8 +374,8 @@ unmapped. This will require changes in arch-specific code.
This allows ``VMAP_STACK`` support on x86, and can simplify support of
architectures that do not have a fixed module region.

-CONFIG_KASAN_KUNIT_TEST & CONFIG_TEST_KASAN_MODULE
---------------------------------------------------
+CONFIG_KASAN_KUNIT_TEST and CONFIG_KASAN_MODULE_TEST
+----------------------------------------------------

KASAN tests consist on two parts:

@@ -384,7 +384,7 @@ KASAN tests consist on two parts:
automatically in a few different ways, see the instructions below.

2. Tests that are currently incompatible with KUnit. Enabled with
-``CONFIG_TEST_KASAN_MODULE`` and can only be run as a module. These tests can
+``CONFIG_KASAN_MODULE_TEST`` and can only be run as a module. These tests can
only be verified manually, by loading the kernel module and inspecting the
kernel log for KASAN reports.

diff --git a/lib/Kconfig.kasan b/lib/Kconfig.kasan
index 3091432acb0a..624ae1df7984 100644
--- a/lib/Kconfig.kasan
+++ b/lib/Kconfig.kasan
@@ -192,7 +192,7 @@ config KASAN_KUNIT_TEST

For more information on KUnit and unit tests in general, please refer

to the KUnit documentation in Documentation/dev-tools/kunit.

-config TEST_KASAN_MODULE
+config KASAN_MODULE_TEST

tristate "KUnit-incompatible tests of KASAN bug detection capabilities"

depends on m && KASAN && !KASAN_HW_TAGS
help

diff --git a/lib/Makefile b/lib/Makefile
index afeff05fa8c5..122f25d6407e 100644
--- a/lib/Makefile
+++ b/lib/Makefile
@@ -68,7 +68,7 @@ obj-$(CONFIG_TEST_IDA) += test_ida.o
obj-$(CONFIG_KASAN_KUNIT_TEST) += test_kasan.o
CFLAGS_test_kasan.o += -fno-builtin
CFLAGS_test_kasan.o += $(call cc-disable-warning, vla)
-obj-$(CONFIG_TEST_KASAN_MODULE) += test_kasan_module.o
+obj-$(CONFIG_KASAN_MODULE_TEST) += test_kasan_module.o
CFLAGS_test_kasan_module.o += -fno-builtin
obj-$(CONFIG_TEST_UBSAN) += test_ubsan.o
CFLAGS_test_ubsan.o += $(call cc-disable-warning, vla)
--
2.29.2.729.g45daf8777d-goog

Andrey Konovalov

unread,

Jan 5, 2021, 1:28:22 PM1/5/21

to Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

It might not be obvious to the compiler that the expression must be
executed between writing and reading to fail_data. In this case, the
compiler might reorder or optimize away some of the accesses, and
the tests will fail.

Add compiler barriers around the expression in KUNIT_EXPECT_KASAN_FAIL.

Signed-off-by: Andrey Konovalov <andre...@google.com>
Link: https://linux-review.googlesource.com/id/I046079f48641a1d36fe627fc8827a9249102fd50
---
lib/test_kasan.c | 2 ++
1 file changed, 2 insertions(+)

diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index dd3d2f95c24e..b5077a47b95a 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c
@@ -79,7 +79,9 @@ static void kasan_test_exit(struct kunit *test)

NULL, \
&resource, \
"kasan_data", &fail_data); \

+ barrier(); \
expression; \
+ barrier(); \
KUNIT_EXPECT_EQ(test, \
fail_data.report_expected, \
fail_data.report_found); \
--
2.29.2.729.g45daf8777d-goog

Andrey Konovalov

unread,

Jan 5, 2021, 1:28:24 PM1/5/21

to Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

In the kmalloc_uaf2() test, the pointers to the two allocated memory
blocks might be the same, and the test will fail. With the software
tag-based mode, the probability of the that happening is 1/254, so it's
hard to observe the failure. For the hardware tag-based mode though,
the probablity is 1/14, which is quite noticable.

Allow up to 4 attempts at generating different tags for the tag-based
modes.

Signed-off-by: Andrey Konovalov <andre...@google.com>
Link: https://linux-review.googlesource.com/id/Ibfa458ef2804ff465d8eb07434a300bf36388d55
---
lib/test_kasan.c | 9 +++++++++
1 file changed, 9 insertions(+)

diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index b5077a47b95a..b67da7f6e17f 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c
@@ -375,7 +375,9 @@ static void kmalloc_uaf2(struct kunit *test)
{
char *ptr1, *ptr2;
size_t size = 43;
+ int counter = 0;

+again:
ptr1 = kmalloc(size, GFP_KERNEL);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr1);

@@ -384,6 +386,13 @@ static void kmalloc_uaf2(struct kunit *test)
ptr2 = kmalloc(size, GFP_KERNEL);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr2);

+ /*
+ * For tag-based KASAN ptr1 and ptr2 tags might happen to be the same.
+ * Allow up to 4 attempts at generating different tags.
+ */
+ if (!IS_ENABLED(CONFIG_KASAN_GENERIC) && ptr1 == ptr2 && counter++ < 4)
+ goto again;
+
KUNIT_EXPECT_KASAN_FAIL(test, ptr1[40] = 'x');
KUNIT_EXPECT_PTR_NE(test, ptr1, ptr2);

--
2.29.2.729.g45daf8777d-goog

Andrey Konovalov

unread,

Jan 5, 2021, 1:28:27 PM1/5/21

to Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

Since the hardware tag-based KASAN mode might not have a redzone that
comes after an allocated object (when kasan.mode=prod is enabled), the
kasan_bitops_tags() test ends up corrupting the next object in memory.

Change the test so it always accesses the redzone that lies within the
allocated object's boundaries.

Signed-off-by: Andrey Konovalov <andre...@google.com>
Link: https://linux-review.googlesource.com/id/I67f51d1ee48f0a8d0fe2658c2a39e4879fe0832a
---
lib/test_kasan.c | 12 ++++++------
1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index b67da7f6e17f..3ea52da52714 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c
@@ -771,17 +771,17 @@ static void kasan_bitops_tags(struct kunit *test)

/* This test is specifically crafted for the tag-based mode. */
if (IS_ENABLED(CONFIG_KASAN_GENERIC)) {

- kunit_info(test, "skipping, CONFIG_KASAN_SW_TAGS required");

+ kunit_info(test, "skipping, CONFIG_KASAN_SW/HW_TAGS required");

return;
}

- /* Allocation size will be rounded to up granule size, which is 16. */
- bits = kzalloc(sizeof(*bits), GFP_KERNEL);
+ /* kmalloc-64 cache will be used and the last 16 bytes will be the redzone. */
+ bits = kzalloc(48, GFP_KERNEL);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, bits);

- /* Do the accesses past the 16 allocated bytes. */
- kasan_bitops_modify(test, BITS_PER_LONG, &bits[1]);
- kasan_bitops_test_and_modify(test, BITS_PER_LONG + BITS_PER_BYTE, &bits[1]);
+ /* Do the accesses past the 48 allocated bytes, but within the redone. */
+ kasan_bitops_modify(test, BITS_PER_LONG, (void *)bits + 48);
+ kasan_bitops_test_and_modify(test, BITS_PER_LONG + BITS_PER_BYTE, (void *)bits + 48);

kfree(bits);
}
--
2.29.2.729.g45daf8777d-goog

Andrey Konovalov

unread,

Jan 5, 2021, 1:28:30 PM1/5/21

to Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

The currently existing kasan_check_read/write() annotations are intended
to be used for kernel modules that have KASAN compiler instrumentation
disabled. Thus, they are only relevant for the software KASAN modes that
rely on compiler instrumentation.

However there's another use case for these annotations: ksize() checks
that the object passed to it is indeed accessible before unpoisoning the
whole object. This is currently done via __kasan_check_read(), which is
compiled away for the hardware tag-based mode that doesn't rely on
compiler instrumentation. This leads to KASAN missing detecting some
memory corruptions.

Provide another annotation called kasan_check_byte() that is available
for all KASAN modes. As the implementation rename and reuse
kasan_check_invalid_free(). Use this new annotation in ksize().

Also add a new ksize_uaf() test that checks that a use-after-free is
detected via ksize() itself, and via plain accesses that happen later.

Signed-off-by: Andrey Konovalov <andre...@google.com>
Link: https://linux-review.googlesource.com/id/Iaabf771881d0f9ce1b969f2a62938e99d3308ec5
---
include/linux/kasan-checks.h | 6 ++++++
include/linux/kasan.h | 13 +++++++++++++
lib/test_kasan.c | 20 ++++++++++++++++++++
mm/kasan/common.c | 11 ++++++++++-
mm/kasan/generic.c | 4 ++--
mm/kasan/kasan.h | 10 +++++-----
mm/kasan/sw_tags.c | 6 +++---
mm/slab_common.c | 15 +++++++++------
8 files changed, 68 insertions(+), 17 deletions(-)

diff --git a/include/linux/kasan-checks.h b/include/linux/kasan-checks.h
index ca5e89fb10d3..3d6d22a25bdc 100644
--- a/include/linux/kasan-checks.h
+++ b/include/linux/kasan-checks.h
@@ -4,6 +4,12 @@

#include <linux/types.h>

+/*
+ * The annotations present in this file are only relevant for the software
+ * KASAN modes that rely on compiler instrumentation, and will be optimized
+ * away for the hardware tag-based KASAN mode. Use kasan_check_byte() instead.
+ */
+
/*
* __kasan_check_*: Always available when KASAN is enabled. This may be used
* even in compilation units that selectively disable KASAN, but must use KASAN
diff --git a/include/linux/kasan.h b/include/linux/kasan.h
index 5e0655fb2a6f..992ba5c653a3 100644
--- a/include/linux/kasan.h
+++ b/include/linux/kasan.h
@@ -243,6 +243,18 @@ static __always_inline void kasan_kfree_large(void *ptr, unsigned long ip)
__kasan_kfree_large(ptr, ip);
}

+/*
+ * Unlike kasan_check_read/write(), kasan_check_byte() is performed even for
+ * the hardware tag-based mode that doesn't rely on compiler instrumentation.
+ */
+bool __kasan_check_byte(const void *addr, unsigned long ip);
+static __always_inline bool kasan_check_byte(const void *addr, unsigned long ip)
+{
+ if (kasan_enabled())
+ return __kasan_check_byte(addr, ip);
+ return true;
+}
+
bool kasan_save_enable_multi_shot(void);
void kasan_restore_multi_shot(bool enabled);

@@ -299,6 +311,7 @@ static inline void *kasan_krealloc(const void *object, size_t new_size,
return (void *)object;
}
static inline void kasan_kfree_large(void *ptr, unsigned long ip) {}
+static inline bool kasan_check_byte(const void *address, unsigned long ip) {}

#endif /* CONFIG_KASAN */

diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index 3ea52da52714..6261521e57ad 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c
@@ -490,6 +490,7 @@ static void kasan_global_oob(struct kunit *test)
KUNIT_EXPECT_KASAN_FAIL(test, *(volatile char *)p);
}

+/* Check that ksize() makes the whole object accessible. */

static void ksize_unpoisons_memory(struct kunit *test)

{
char *ptr;
@@ -508,6 +509,24 @@ static void ksize_unpoisons_memory(struct kunit *test)
kfree(ptr);
}

+/*
+ * Check that a use-after-free is detected by ksize() and via normal accesses
+ * after it.
+ */
+static void ksize_uaf(struct kunit *test)
+{
+ char *ptr;
+ int size = 128 - KASAN_GRANULE_SIZE;
+
+ ptr = kmalloc(size, GFP_KERNEL);
+ KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);
+ kfree(ptr);
+
+ KUNIT_EXPECT_KASAN_FAIL(test, ksize(ptr));
+ KUNIT_EXPECT_KASAN_FAIL(test, kasan_int_result = *ptr);
+ KUNIT_EXPECT_KASAN_FAIL(test, kasan_int_result = *(ptr + size));
+}
+

static void kasan_stack_oob(struct kunit *test)
{

char stack_array[10];
@@ -937,6 +956,7 @@ static struct kunit_case kasan_kunit_test_cases[] = {
KUNIT_CASE(kasan_alloca_oob_left),
KUNIT_CASE(kasan_alloca_oob_right),
KUNIT_CASE(ksize_unpoisons_memory),
+ KUNIT_CASE(ksize_uaf),
KUNIT_CASE(kmem_cache_double_free),
KUNIT_CASE(kmem_cache_invalid_free),
KUNIT_CASE(kasan_memchr),
diff --git a/mm/kasan/common.c b/mm/kasan/common.c
index eedc3e0fe365..45ab2c7073a8 100644
--- a/mm/kasan/common.c
+++ b/mm/kasan/common.c
@@ -345,7 +345,7 @@ static bool ____kasan_slab_free(struct kmem_cache *cache, void *object,

if (unlikely(cache->flags & SLAB_TYPESAFE_BY_RCU))
return false;

- if (kasan_check_invalid_free(tagged_object)) {
+ if (!kasan_check(tagged_object)) {
kasan_report_invalid_free(tagged_object, ip);
return true;
}
@@ -490,3 +490,12 @@ void __kasan_kfree_large(void *ptr, unsigned long ip)
kasan_report_invalid_free(ptr, ip);
/* The object will be poisoned by kasan_free_pages(). */
}
+
+bool __kasan_check_byte(const void *address, unsigned long ip)
+{
+ if (!kasan_check(address)) {
+ kasan_report_invalid_free((void *)address, ip);
+ return false;
+ }
+ return true;
+}
diff --git a/mm/kasan/generic.c b/mm/kasan/generic.c
index acab8862dc67..b3631ad9a8ef 100644
--- a/mm/kasan/generic.c
+++ b/mm/kasan/generic.c
@@ -185,11 +185,11 @@ bool kasan_check_range(unsigned long addr, size_t size, bool write,

return check_region_inline(addr, size, write, ret_ip);
}

-bool kasan_check_invalid_free(void *addr)
+bool kasan_check(const void *addr)

{
s8 shadow_byte = READ_ONCE(*(s8 *)kasan_mem_to_shadow(addr));

- return shadow_byte < 0 || shadow_byte >= KASAN_GRANULE_SIZE;
+ return shadow_byte >= 0 && shadow_byte < KASAN_GRANULE_SIZE;
}

void kasan_cache_shrink(struct kmem_cache *cache)
diff --git a/mm/kasan/kasan.h b/mm/kasan/kasan.h
index 292dfbc37deb..f17591545279 100644
--- a/mm/kasan/kasan.h
+++ b/mm/kasan/kasan.h
@@ -329,20 +329,20 @@ static inline void kasan_unpoison(const void *address, size_t size)
round_up(size, KASAN_GRANULE_SIZE), get_tag(address));
}

-static inline bool kasan_check_invalid_free(void *addr)
+static inline bool kasan_check(const void *addr)
{
u8 ptr_tag = get_tag(addr);
- u8 mem_tag = hw_get_mem_tag(addr);
+ u8 mem_tag = hw_get_mem_tag((void *)addr);

- return (mem_tag == KASAN_TAG_INVALID) ||
- (ptr_tag != KASAN_TAG_KERNEL && ptr_tag != mem_tag);
+ return (mem_tag != KASAN_TAG_INVALID) &&
+ (ptr_tag == KASAN_TAG_KERNEL || ptr_tag == mem_tag);
}

#else /* CONFIG_KASAN_HW_TAGS */

void kasan_poison(const void *address, size_t size, u8 value);
void kasan_unpoison(const void *address, size_t size);
-bool kasan_check_invalid_free(void *addr);
+bool kasan_check(const void *addr);

#endif /* CONFIG_KASAN_HW_TAGS */

diff --git a/mm/kasan/sw_tags.c b/mm/kasan/sw_tags.c
index cc271fceb5d5..e326caaaaca3 100644
--- a/mm/kasan/sw_tags.c
+++ b/mm/kasan/sw_tags.c
@@ -118,13 +118,13 @@ bool kasan_check_range(unsigned long addr, size_t size, bool write,
return true;
}

-bool kasan_check_invalid_free(void *addr)
+bool kasan_check(const void *addr)

{
u8 tag = get_tag(addr);
u8 shadow_byte = READ_ONCE(*(u8 *)kasan_mem_to_shadow(kasan_reset_tag(addr)));

- return (shadow_byte == KASAN_TAG_INVALID) ||
- (tag != KASAN_TAG_KERNEL && tag != shadow_byte);
+ return (shadow_byte != KASAN_TAG_INVALID) &&
+ (tag == KASAN_TAG_KERNEL || tag == shadow_byte);
}

#define DEFINE_HWASAN_LOAD_STORE(size) \
diff --git a/mm/slab_common.c b/mm/slab_common.c
index e981c80d216c..a3bb44516623 100644
--- a/mm/slab_common.c
+++ b/mm/slab_common.c
@@ -1157,11 +1157,13 @@ size_t ksize(const void *objp)
size_t size;

/*
- * We need to check that the pointed to object is valid, and only then
- * unpoison the shadow memory below. We use __kasan_check_read(), to
- * generate a more useful report at the time ksize() is called (rather
- * than later where behaviour is undefined due to potential
- * use-after-free or double-free).
+ * We need to first check that the pointer to the object is valid, and
+ * only then unpoison the memory. The report printed from ksize() is
+ * more useful, then when it's printed later when the behaviour could
+ * be undefined due to a potential use-after-free or double-free.
+ *
+ * We use kasan_check_byte(), which is supported for hardware tag-based
+ * KASAN mode, unlike kasan_check_read/write().
*
* If the pointed to memory is invalid we return 0, to avoid users of
* ksize() writing to and potentially corrupting the memory region.
@@ -1169,7 +1171,8 @@ size_t ksize(const void *objp)
* We want to perform the check before __ksize(), to avoid potentially
* crashing in __ksize() due to accessing invalid metadata.
*/
- if (unlikely(ZERO_OR_NULL_PTR(objp)) || !__kasan_check_read(objp, 1))
+ if (unlikely(ZERO_OR_NULL_PTR(objp)) ||
+ !kasan_check_byte(objp, _RET_IP_))
return 0;

size = __ksize(objp);
--
2.29.2.729.g45daf8777d-goog

Andrey Konovalov

unread,

Jan 5, 2021, 1:28:32 PM1/5/21

to Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

The currently existing page allocator tests rely on kmalloc fallback
with large sizes that is only present for SLUB. Add proper tests that
use alloc/free_pages().

Signed-off-by: Andrey Konovalov <andre...@google.com>
Link: https://linux-review.googlesource.com/id/Ia173d5a1b215fe6b2548d814ef0f4433cf983570
---
lib/test_kasan.c | 54 +++++++++++++++++++++++++++++++++++++++++++-----
1 file changed, 49 insertions(+), 5 deletions(-)

diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index 6261521e57ad..24798c034d05 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c
@@ -128,6 +128,12 @@ static void kmalloc_node_oob_right(struct kunit *test)
kfree(ptr);
}

+/*
+ * These kmalloc_pagealloc_* tests try allocating a memory chunk that doesn't
+ * fit into a slab cache and therefore is allocated via the page allocator
+ * fallback. Since this kind of fallback is only implemented for SLUB, these
+ * tests are limited to that allocator.
+ */
static void kmalloc_pagealloc_oob_right(struct kunit *test)
{
char *ptr;
@@ -138,14 +144,11 @@ static void kmalloc_pagealloc_oob_right(struct kunit *test)
return;
}

- /*
- * Allocate a chunk that does not fit into a SLUB cache to trigger
- * the page allocator fallback.
- */
ptr = kmalloc(size, GFP_KERNEL);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);

KUNIT_EXPECT_KASAN_FAIL(test, ptr[size + OOB_TAG_OFF] = 0);
+
kfree(ptr);
}

@@ -161,8 +164,8 @@ static void kmalloc_pagealloc_uaf(struct kunit *test)

ptr = kmalloc(size, GFP_KERNEL);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);

-
kfree(ptr);
+
KUNIT_EXPECT_KASAN_FAIL(test, ptr[0] = 0);
}

@@ -182,6 +185,45 @@ static void kmalloc_pagealloc_invalid_free(struct kunit *test)
KUNIT_EXPECT_KASAN_FAIL(test, kfree(ptr + 1));
}

+static void pagealloc_oob_right(struct kunit *test)
+{
+ char *ptr;
+ struct page *pages;
+ size_t order = 4;
+ size_t size = (1UL << (PAGE_SHIFT + order));
+
+ /*
+ * With generic KASAN page allocations have no redzones, thus
+ * out-of-bounds detection is not guaranteed.
+ * See https://bugzilla.kernel.org/show_bug.cgi?id=210503.
+ */
+ if (IS_ENABLED(CONFIG_KASAN_GENERIC)) {
+ kunit_info(test, "skipping, CONFIG_KASAN_GENERIC enabled");
+ return;
+ }
+

+ pages = alloc_pages(GFP_KERNEL, order);
+ ptr = page_address(pages);
+ KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);
+

+ KUNIT_EXPECT_KASAN_FAIL(test, ptr[size] = 0);

+ free_pages((unsigned long)ptr, order);
+}
+

+static void pagealloc_uaf(struct kunit *test)
+{
+ char *ptr;
+ struct page *pages;
+ size_t order = 4;
+

+ pages = alloc_pages(GFP_KERNEL, order);
+ ptr = page_address(pages);
+ KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);

+ free_pages((unsigned long)ptr, order);
+

+ KUNIT_EXPECT_KASAN_FAIL(test, ptr[0] = 0);
+}
+
static void kmalloc_large_oob_right(struct kunit *test)
{
char *ptr;
@@ -933,6 +975,8 @@ static struct kunit_case kasan_kunit_test_cases[] = {
KUNIT_CASE(kmalloc_pagealloc_oob_right),
KUNIT_CASE(kmalloc_pagealloc_uaf),
KUNIT_CASE(kmalloc_pagealloc_invalid_free),
+ KUNIT_CASE(pagealloc_oob_right),
+ KUNIT_CASE(pagealloc_uaf),
KUNIT_CASE(kmalloc_large_oob_right),
KUNIT_CASE(kmalloc_oob_krealloc_more),
KUNIT_CASE(kmalloc_oob_krealloc_less),
--
2.29.2.729.g45daf8777d-goog

Alexander Potapenko

unread,

Jan 12, 2021, 2:39:09 AM1/12/21

to Andrey Konovalov, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

On Tue, Jan 5, 2021 at 7:28 PM Andrey Konovalov <andre...@google.com> wrote:
>
> There's a number of internal KASAN functions that are used across multiple
> source code files and therefore aren't marked as static inline. To avoid
> littering the kernel function names list with generic functions, prefix
> all such KASAN functions with kasan_.
>
> As a part of this change:
>
> - Rename internal (un)poison_range() to kasan_(un)poison() (no _range)
> to avoid name collision with a public kasan_unpoison_range().
>
> - Rename check_memory_region() to kasan_check_range(), as it seems to be
> a more fitting name.
>
> Suggested-by: Marco Elver <el...@google.com>
> Signed-off-by: Andrey Konovalov <andre...@google.com>
> Link: https://linux-review.googlesource.com/id/I719cc93483d4ba288a634dba80ee6b7f2809cd26

Reviewed-by: Alexander Potapenko <gli...@google.com>

--
Alexander Potapenko
Software Engineer

Google Germany GmbH
Erika-Mann-Straße, 33
80636 München

Geschäftsführer: Paul Manicle, Halimah DeLaine Prado
Registergericht und -nummer: Hamburg, HRB 86891
Sitz der Gesellschaft: Hamburg

Alexander Potapenko

unread,

Jan 12, 2021, 2:40:55 AM1/12/21

to Andrey Konovalov, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

On Tue, Jan 5, 2021 at 7:28 PM Andrey Konovalov <andre...@google.com> wrote:
>

> Mention in the documentation that enabling CONFIG_KASAN_HW_TAGS
> always results in in-kernel TBI (Top Byte Ignore) being enabled.
>
> Also do a few minor documentation cleanups.
>
> Signed-off-by: Andrey Konovalov <andre...@google.com>
> Link: https://linux-review.googlesource.com/id/Iba2a6697e3c6304cb53f89ec61dedc77fa29e3ae

Reviewed-by: Alexander Potapenko <gli...@google.com>

Alexander Potapenko

unread,

Jan 12, 2021, 2:53:36 AM1/12/21

to Andrey Konovalov, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

On Tue, Jan 5, 2021 at 7:28 PM Andrey Konovalov <andre...@google.com> wrote:
>

> Clarify and update comments and info messages in KASAN tests.
>
> Signed-off-by: Andrey Konovalov <andre...@google.com>
> Link: https://linux-review.googlesource.com/id/I6c816c51fa1e0eb7aa3dead6bda1f339d2af46c8

> void *kasan_ptr_result;
> int kasan_int_result;
Shouldn't these two variables be static, by the way?

>
> @@ -39,14 +38,13 @@ static struct kunit_resource resource;
> static struct kunit_kasan_expectation fail_data;
> static bool multishot;
>
> +/*
> + * Temporarily enable multi-shot mode. Otherwise, KASAN would only report the
> + * first detected bug and panic the kernel if panic_on_warn is enabled.
> + */

YMMV, but I think this comment was at its place already.

> static int kasan_test_init(struct kunit *test)
> {
> - /*
> - * Temporarily enable multi-shot mode and set panic_on_warn=0.
> - * Otherwise, we'd only get a report for the first case.
> - */
> multishot = kasan_save_enable_multi_shot();

Unrelated to this change, but have you considered storing
test-specific data in test->priv instead of globals?

unread,

Jan 12, 2021, 3:26:05 AM1/12/21

to Andrey Konovalov, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

Nit: s/adopt/adapt in the title.

> +again:
> ptr1 = kmalloc(size, GFP_KERNEL);
> KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr1);
>
> @@ -384,6 +386,13 @@ static void kmalloc_uaf2(struct kunit *test)
> ptr2 = kmalloc(size, GFP_KERNEL);
> KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr2);
>
> + /*
> + * For tag-based KASAN ptr1 and ptr2 tags might happen to be the same.
> + * Allow up to 4 attempts at generating different tags.
> + */
> + if (!IS_ENABLED(CONFIG_KASAN_GENERIC) && ptr1 == ptr2 && counter++ < 4)
> + goto again;
> +

Looks like we are leaking memory allocated for ptr2 here?

Alexander Potapenko

unread,

Jan 12, 2021, 3:30:29 AM1/12/21

to Andrey Konovalov, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

On Tue, Jan 5, 2021 at 7:28 PM Andrey Konovalov <andre...@google.com> wrote:
>

I think it might make sense to call ksize() here to ensure we have
these spare bytes.

Alexander Potapenko

unread,

Jan 12, 2021, 3:57:39 AM1/12/21

to Andrey Konovalov, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

On Tue, Jan 5, 2021 at 7:28 PM Andrey Konovalov <andre...@google.com> wrote:
>

> The currently existing page allocator tests rely on kmalloc fallback
> with large sizes that is only present for SLUB. Add proper tests that
> use alloc/free_pages().
>
> Signed-off-by: Andrey Konovalov <andre...@google.com>
> Link: https://linux-review.googlesource.com/id/Ia173d5a1b215fe6b2548d814ef0f4433cf983570

Reviewed-by: Alexander Potapenko <gli...@google.com>

Marco Elver

unread,

Jan 12, 2021, 8:39:11 AM1/12/21

to Andrey Konovalov, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Andrew Morton, Will Deacon, Andrey Ryabinin, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org

On Tue, Jan 05, 2021 at 07:27PM +0100, Andrey Konovalov wrote:

Why do we even need a limit? Why not retry until ptr1 != ptr2?

Marco Elver

unread,

Jan 12, 2021, 8:55:26 AM1/12/21

to Andrey Konovalov, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Andrew Morton, Will Deacon, Andrey Ryabinin, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org

On Tue, Jan 05, 2021 at 07:27PM +0100, Andrey Konovalov wrote:

> Since the hardware tag-based KASAN mode might not have a redzone that
> comes after an allocated object (when kasan.mode=prod is enabled), the
> kasan_bitops_tags() test ends up corrupting the next object in memory.
>
> Change the test so it always accesses the redzone that lies within the
> allocated object's boundaries.
>
> Signed-off-by: Andrey Konovalov <andre...@google.com>
> Link: https://linux-review.googlesource.com/id/I67f51d1ee48f0a8d0fe2658c2a39e4879fe0832a

Reviewed-by: Marco Elver <el...@google.com>

Marco Elver

unread,

Jan 12, 2021, 9:32:47 AM1/12/21

to Andrey Konovalov, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Andrew Morton, Will Deacon, Andrey Ryabinin, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org

On Tue, Jan 05, 2021 at 07:27PM +0100, Andrey Konovalov wrote:

We have too many check-functions, and the name needs to be more precise.
Intuitively, I would have thought this should have access-type, i.e.
read or write, effectively mirroring a normal access.

Would kasan_check_byte_read() be better (and just not have a 'write'
variant because we do not need it)? This would restore ksize() closest
to what it was before (assuming reporting behaviour is fixed, too).

This is strange: why does it report an invalid free? Should this be a
use-after-free? I think this could just call kasan_report(....) for 1
byte, and we'd get the right report.

Definitely prefer shorted names, but we're in the unfortunate situation
of having numerous kasan_check-functions, so we probably need to be more
precise.

kasan_check() makes me think this also does reporting, but it does not
(it seems to only check the metadata for validity).

The internal function could therefore be kasan_check_allocated() (it's
now the inverse of kasan_check_invalid_free()).

Marco Elver

unread,

Jan 12, 2021, 9:34:23 AM1/12/21

to Andrey Konovalov, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Andrew Morton, Will Deacon, Andrey Ryabinin, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org

On Tue, Jan 05, 2021 at 07:27PM +0100, Andrey Konovalov wrote:

> The currently existing page allocator tests rely on kmalloc fallback
> with large sizes that is only present for SLUB. Add proper tests that
> use alloc/free_pages().
>
> Signed-off-by: Andrey Konovalov <andre...@google.com>
> Link: https://linux-review.googlesource.com/id/Ia173d5a1b215fe6b2548d814ef0f4433cf983570

Reviewed-by: Marco Elver <el...@google.com>

Andrey Konovalov

unread,

Jan 12, 2021, 12:56:05 PM1/12/21

to Alexander Potapenko, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

On Tue, Jan 12, 2021 at 8:53 AM Alexander Potapenko <gli...@google.com> wrote:
>
> On Tue, Jan 5, 2021 at 7:28 PM Andrey Konovalov <andre...@google.com> wrote:
> >
> > Clarify and update comments and info messages in KASAN tests.
> >
> > Signed-off-by: Andrey Konovalov <andre...@google.com>
> > Link: https://linux-review.googlesource.com/id/I6c816c51fa1e0eb7aa3dead6bda1f339d2af46c8
>
> > void *kasan_ptr_result;
> > int kasan_int_result;
> Shouldn't these two variables be static, by the way?

No, then the compiler starts eliminating accesses.

> > @@ -39,14 +38,13 @@ static struct kunit_resource resource;
> > static struct kunit_kasan_expectation fail_data;
> > static bool multishot;
> >
> > +/*
> > + * Temporarily enable multi-shot mode. Otherwise, KASAN would only report the
> > + * first detected bug and panic the kernel if panic_on_warn is enabled.
> > + */
>
> YMMV, but I think this comment was at its place already.

It gets updated by one of the subsequent patches.

> > static int kasan_test_init(struct kunit *test)
> > {
> > - /*
> > - * Temporarily enable multi-shot mode and set panic_on_warn=0.
> > - * Otherwise, we'd only get a report for the first case.
> > - */
> > multishot = kasan_save_enable_multi_shot();
>
> Unrelated to this change, but have you considered storing
> test-specific data in test->priv instead of globals?

I'd say that test->priv is for some per-test data that's used in the
tests, and multishot is not a part of that.

> > if (!IS_ENABLED(CONFIG_SLUB)) {
> > - kunit_info(test, "CONFIG_SLUB is not enabled.");
> > + kunit_info(test, "skipping, CONFIG_SLUB required");
> > return;
> > }
>
> You may want to introduce a macro that takes a config name and prints
> the warning/returns if it's not enabled.

Good idea, will do in v2.

Thanks!

Andrey Konovalov

unread,

Jan 12, 2021, 1:10:20 PM1/12/21

to Alexander Potapenko, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

Will fix in v2, thanks!

Andrey Konovalov

unread,

Jan 12, 2021, 1:11:23 PM1/12/21

to Marco Elver, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Andrew Morton, Will Deacon, Andrey Ryabinin, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

Indeed, will fix in v2, thanks!

Andrey Konovalov

unread,

Jan 12, 2021, 1:26:39 PM1/12/21

to Alexander Potapenko, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

Will do in v2, thanks!

Andrey Konovalov

unread,

Jan 12, 2021, 1:28:36 PM1/12/21

to Marco Elver, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Andrew Morton, Will Deacon, Andrey Ryabinin, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

Most tests in lib/ start with test_, so not using that pattern for
KASAN tests could be confusing. Maybe we can move them to mm/kasan.
Anyway, I won't look into this right now.

Thanks!

Catalin Marinas

unread,

Jan 12, 2021, 2:01:38 PM1/12/21

to Andrey Konovalov, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org

On Tue, Jan 05, 2021 at 07:27:49PM +0100, Andrey Konovalov wrote:
> diff --git a/arch/arm64/mm/fault.c b/arch/arm64/mm/fault.c
> index 3c40da479899..57d3f165d907 100644
> --- a/arch/arm64/mm/fault.c
> +++ b/arch/arm64/mm/fault.c
> @@ -302,12 +302,20 @@ static void die_kernel_fault(const char *msg, unsigned long addr,
> static void report_tag_fault(unsigned long addr, unsigned int esr,
> struct pt_regs *regs)
> {
> - bool is_write = ((esr & ESR_ELx_WNR) >> ESR_ELx_WNR_SHIFT) != 0;
> + static bool reported;
> + bool is_write;
> +
> + if (READ_ONCE(reported))
> + return;
> +
> + if (mte_report_once())
> + WRITE_ONCE(reported, true);

I guess the assumption here is that you don't get any report before the
tests start and temporarily set report_once to false. It's probably
fine, if we get a tag check failure we'd notice in the logs anyway.

> /*
> * SAS bits aren't set for all faults reported in EL1, so we can't
> * find out access size.
> */
> + is_write = ((esr & ESR_ELx_WNR) >> ESR_ELx_WNR_SHIFT) != 0;

I now noticed, you could write this in a shorter way:

is_write = !!(esr & ESR_ELx_WNR);

> kasan_report(addr, 0, is_write, regs->pc);
> }

The patch looks fine to me.

Reviewed-by: Catalin Marinas <catalin...@arm.com>

Andrey Konovalov

unread,

Jan 12, 2021, 2:51:06 PM1/12/21

to Alexander Potapenko, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

On Tue, Jan 12, 2021 at 9:18 AM Alexander Potapenko <gli...@google.com> wrote:
>
> On Tue, Jan 5, 2021 at 7:28 PM Andrey Konovalov <andre...@google.com> wrote:
> >
> > It might not be obvious to the compiler that the expression must be
> > executed between writing and reading to fail_data. In this case, the
> > compiler might reorder or optimize away some of the accesses, and
> > the tests will fail.
>
> Have you seen this happen in practice?

Yes.

> Are these accesses to fail_data that are optimized (in which case we

> could make it volatile)?

Yes. AFAIU compiler doesn't expect expression to change fail_data
fields, no those accesses and checks are optimized away.

> Note that compiler barriers won't probably help against removing
> memory accesses, they only prevent reordering.
>
> > + barrier(); \
> > expression; \
> > + barrier(); \
>
> The need for barriers is not obvious to the reader, so a comment in
> the code clarifying that would be nice.

Will add a comment in v2, thanks!

Andrey Konovalov

unread,

Jan 12, 2021, 2:57:16 PM1/12/21

to Alexander Potapenko, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

On Tue, Jan 12, 2021 at 8:50 PM Andrey Konovalov <andre...@google.com> wrote:
>
> On Tue, Jan 12, 2021 at 9:18 AM Alexander Potapenko <gli...@google.com> wrote:
> >
> > On Tue, Jan 5, 2021 at 7:28 PM Andrey Konovalov <andre...@google.com> wrote:
> > >
> > > It might not be obvious to the compiler that the expression must be
> > > executed between writing and reading to fail_data. In this case, the
> > > compiler might reorder or optimize away some of the accesses, and
> > > the tests will fail.
> >
> > Have you seen this happen in practice?
>
> Yes.
>
> > Are these accesses to fail_data that are optimized (in which case we
> > could make it volatile)?
>
> Yes. AFAIU compiler doesn't expect expression to change fail_data
> fields, no those accesses and checks are optimized away.

Ah, actually no, it reorders the expression and puts it after
fail_data fields checks. That's why I put the barriers.

> > Note that compiler barriers won't probably help against removing
> > memory accesses, they only prevent reordering.

But using WRITE/READ_ONCE() might also be a good idea, as technically
the compiler can optimize away the accesses.

unread,

Jan 12, 2021, 4:16:50 PM1/12/21

to Marco Elver, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Andrew Morton, Will Deacon, Andrey Ryabinin, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

On Tue, Jan 12, 2021 at 3:32 PM Marco Elver <el...@google.com> wrote:
>
> > +/*
> > + * Unlike kasan_check_read/write(), kasan_check_byte() is performed even for
> > + * the hardware tag-based mode that doesn't rely on compiler instrumentation.
> > + */
>
> We have too many check-functions, and the name needs to be more precise.
> Intuitively, I would have thought this should have access-type, i.e.
> read or write, effectively mirroring a normal access.
>
> Would kasan_check_byte_read() be better (and just not have a 'write'
> variant because we do not need it)? This would restore ksize() closest
> to what it was before (assuming reporting behaviour is fixed, too).

> > void kasan_poison(const void *address, size_t size, u8 value);
> > void kasan_unpoison(const void *address, size_t size);
> > -bool kasan_check_invalid_free(void *addr);
> > +bool kasan_check(const void *addr);
>
> Definitely prefer shorted names, but we're in the unfortunate situation
> of having numerous kasan_check-functions, so we probably need to be more
> precise.
>
> kasan_check() makes me think this also does reporting, but it does not
> (it seems to only check the metadata for validity).
>
> The internal function could therefore be kasan_check_allocated() (it's
> now the inverse of kasan_check_invalid_free()).

Re: kasan_check_byte():

I think the _read suffix is only making the name longer. ksize() isn't
checking that the memory is readable (or writable), it's checking that
it's addressable. At least that's the intention of the annotation, so
it makes sense to name it correspondingly despite the implementation.

Having all kasan_check_*() functions both checking and reporting makes
sense, so let's keep the kasan_check_ prefix.

What isn't obvious from the name is that this function is present for
every kasan mode. Maybe kasan_check_byte_always()? Although it also
seems too long.

But I'm OK with keeping kasan_check_byte().

Re kasan_check():

Here we can use Andrew's suggestion about the name being related to
what the function returns. And also drop the kasan_check_ prefix as
this function only does the checking.

Let's use kasan_byte_accessible() instead of kasan_check().

> > +bool __kasan_check_byte(const void *address, unsigned long ip)
> > +{
> > + if (!kasan_check(address)) {
> > + kasan_report_invalid_free((void *)address, ip);
>
> This is strange: why does it report an invalid free? Should this be a
> use-after-free? I think this could just call kasan_report(....) for 1
> byte, and we'd get the right report.

Will fix in v2.

Thanks!

Marco Elver

unread,

Jan 12, 2021, 5:54:55 PM1/12/21

to Andrey Konovalov, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Andrew Morton, Will Deacon, Andrey Ryabinin, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

This is fine.

> Re kasan_check():
>
> Here we can use Andrew's suggestion about the name being related to
> what the function returns. And also drop the kasan_check_ prefix as
> this function only does the checking.
>
> Let's use kasan_byte_accessible() instead of kasan_check().

Sounds reasonable to me.

Thanks,
-- Marco

Alexander Potapenko

unread,

Jan 13, 2021, 7:30:34 AM1/13/21

to Andrey Konovalov, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

On Tue, Jan 12, 2021 at 9:07 PM 'Andrey Konovalov' via kasan-dev
<kasa...@googlegroups.com> wrote:
>
> On Tue, Jan 12, 2021 at 9:30 AM Alexander Potapenko <gli...@google.com> wrote:
> >
> > On Tue, Jan 5, 2021 at 7:28 PM Andrey Konovalov <andre...@google.com> wrote:
> > >
> > > Since the hardware tag-based KASAN mode might not have a redzone that
> > > comes after an allocated object (when kasan.mode=prod is enabled), the
> > > kasan_bitops_tags() test ends up corrupting the next object in memory.
> > >
> > > Change the test so it always accesses the redzone that lies within the
> > > allocated object's boundaries.
> > >
> > > Signed-off-by: Andrey Konovalov <andre...@google.com>
> > > Link: https://linux-review.googlesource.com/id/I67f51d1ee48f0a8d0fe2658c2a39e4879fe0832a

Reviewed-by: Alexander Potapenko <gli...@google.com>

> > > ---
> > > lib/test_kasan.c | 12 ++++++------
> > > 1 file changed, 6 insertions(+), 6 deletions(-)
> > >
> > > diff --git a/lib/test_kasan.c b/lib/test_kasan.c
> > > index b67da7f6e17f..3ea52da52714 100644
> > > --- a/lib/test_kasan.c
> > > +++ b/lib/test_kasan.c
> > > @@ -771,17 +771,17 @@ static void kasan_bitops_tags(struct kunit *test)
> > >
> > > /* This test is specifically crafted for the tag-based mode. */
> > > if (IS_ENABLED(CONFIG_KASAN_GENERIC)) {
> > > - kunit_info(test, "skipping, CONFIG_KASAN_SW_TAGS required");
> > > + kunit_info(test, "skipping, CONFIG_KASAN_SW/HW_TAGS required");
> > > return;
> > > }
> > >
> > > - /* Allocation size will be rounded to up granule size, which is 16. */
> > > - bits = kzalloc(sizeof(*bits), GFP_KERNEL);
> > > + /* kmalloc-64 cache will be used and the last 16 bytes will be the redzone. */
> > > + bits = kzalloc(48, GFP_KERNEL);
> >
> > I think it might make sense to call ksize() here to ensure we have
> > these spare bytes.
>
> Calling ksize() will unpoison the whole object.

Ah, that's right.

> I think it's OK to make assumptions about KASAN internals in tests. I
> would actually say that we need more tests that check such internal
> properties.

Agreed.

Andrey Konovalov

unread,

Jan 13, 2021, 11:21:46 AM1/13/21

to Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

This patchset adds support for running KASAN-KUnit tests with the
hardware tag-based mode and also contains a few fixes.

Changes in v1->v2:
- Fix return values of kasan_check_byte().
- Use KASAN_TEST_NEEDS_*() macros for checking configs required for tests.
- Fix unused size in match_all_not_assigned().
- Fix typo in KASAN docs: "consist on" => "consist of".
- Use READ/WRITE_ONCE() for accessing fail_data fields.
- Doesn't leak memory in kmalloc_uaf2().
- Do up to 16 attempts in kmalloc_uaf2().
- Use kasan_report() for reporting from ksize() check.
- Rename kasan_check() to kasan_byte_accessible().
- Add a test for kmem_cache_bulk_alloc().
- Checks that pointer tags are assigned from [KASAN_TAG_MIN, KASAN_TAG_KERNEL).
- Don't run tests with kasan.mode=off.

Andrey Konovalov (14):
kasan: prefix global functions with kasan_
kasan: clarify HW_TAGS impact on TBI
kasan: clean up comments in tests
kasan: add macros to simplify checking test constraints
kasan: add match-all tag tests
kasan, arm64: allow using KUnit tests with HW_TAGS mode
kasan: rename CONFIG_TEST_KASAN_MODULE
kasan: add compiler barriers to KUNIT_EXPECT_KASAN_FAIL
kasan: adapt kmalloc_uaf2 test to HW_TAGS mode
kasan: fix memory corruption in kasan_bitops_tags test
kasan: fix bug detection via ksize for HW_TAGS mode
kasan: add proper page allocator tests
kasan: add a test for kmem_cache_alloc/free_bulk
kasan: don't run tests when KASAN is not enabled

Documentation/dev-tools/kasan.rst | 24 +-
arch/arm64/include/asm/memory.h | 1 +
arch/arm64/include/asm/mte-kasan.h | 12 +
arch/arm64/kernel/mte.c | 12 +
arch/arm64/mm/fault.c | 16 +-
include/linux/kasan-checks.h | 6 +
include/linux/kasan.h | 16 ++
lib/Kconfig.kasan | 6 +-
lib/Makefile | 2 +-
lib/test_kasan.c | 423 +++++++++++++++++++++--------
lib/test_kasan_module.c | 5 +-
mm/kasan/common.c | 56 ++--
mm/kasan/generic.c | 38 +--
mm/kasan/kasan.h | 69 +++--
mm/kasan/quarantine.c | 22 +-
mm/kasan/report.c | 15 +-
mm/kasan/report_generic.c | 8 +-
mm/kasan/report_hw_tags.c | 8 +-
mm/kasan/report_sw_tags.c | 8 +-
mm/kasan/shadow.c | 26 +-
mm/kasan/sw_tags.c | 20 +-
mm/slab_common.c | 15 +-
tools/objtool/check.c | 2 +-
23 files changed, 544 insertions(+), 266 deletions(-)

--
2.30.0.284.gd98b1dd5eaa7-goog

Andrey Konovalov

unread,

Jan 13, 2021, 11:21:50 AM1/13/21

to Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

There's a number of internal KASAN functions that are used across multiple
source code files and therefore aren't marked as static inline. To avoid

littering the kernel function names list with generic function names,

prefix all such KASAN functions with kasan_.

As a part of this change:

- Rename internal (un)poison_range() to kasan_(un)poison() (no _range)
to avoid name collision with a public kasan_unpoison_range().

- Rename check_memory_region() to kasan_check_range(), as it's a more
fitting name.

Link: https://linux-review.googlesource.com/id/I719cc93483d4ba288a634dba80ee6b7f2809cd26
Suggested-by: Marco Elver <el...@google.com>
Reviewed-by: Marco Elver <el...@google.com>
Reviewed-by: Alexander Potapenko <gli...@google.com>
Signed-off-by: Andrey Konovalov <andre...@google.com>

---
mm/kasan/common.c | 47 +++++++++++++++++++-------------------
mm/kasan/generic.c | 36 ++++++++++++++---------------
mm/kasan/kasan.h | 48 +++++++++++++++++++--------------------
mm/kasan/quarantine.c | 22 +++++++++---------
mm/kasan/report.c | 13 ++++++-----
mm/kasan/report_generic.c | 8 +++----
mm/kasan/report_hw_tags.c | 8 +++----
mm/kasan/report_sw_tags.c | 8 +++----
mm/kasan/shadow.c | 26 ++++++++++-----------
mm/kasan/sw_tags.c | 16 ++++++-------
tools/objtool/check.c | 2 +-
11 files changed, 117 insertions(+), 117 deletions(-)

diff --git a/mm/kasan/common.c b/mm/kasan/common.c
index b25167664ead..eedc3e0fe365 100644
--- a/mm/kasan/common.c
+++ b/mm/kasan/common.c

@@ -346,12 +345,12 @@ static bool ____kasan_slab_free(struct kmem_cache *cache, void *object,

if (unlikely(cache->flags & SLAB_TYPESAFE_BY_RCU))
return false;

- if (check_invalid_free(tagged_object)) {
+ if (kasan_check_invalid_free(tagged_object)) {
kasan_report_invalid_free(tagged_object, ip);
return true;
}

- poison_range(object, cache->object_size, KASAN_KMALLOC_FREE);
+ kasan_poison(object, cache->object_size, KASAN_KMALLOC_FREE);

if (!kasan_stack_collection_enabled())
return false;
@@ -361,7 +360,7 @@ static bool ____kasan_slab_free(struct kmem_cache *cache, void *object,

kasan_set_free_info(cache, object, tag);

- return quarantine_put(cache, object);
+ return kasan_quarantine_put(cache, object);
}

bool __kasan_slab_free(struct kmem_cache *cache, void *object, unsigned long ip)

@@ -386,7 +385,7 @@ void __kasan_slab_free_mempool(void *ptr, unsigned long ip)
kasan_report_invalid_free(ptr, ip);

diff --git a/mm/kasan/generic.c b/mm/kasan/generic.c
index 5106b84b07d4..acab8862dc67 100644
--- a/mm/kasan/generic.c
+++ b/mm/kasan/generic.c

@@ -158,7 +158,7 @@ static __always_inline bool memory_is_poisoned(unsigned long addr, size_t size)
return memory_is_poisoned_n(addr, size);
}

-static __always_inline bool check_memory_region_inline(unsigned long addr,

+static __always_inline bool check_region_inline(unsigned long addr,
size_t size, bool write,

unsigned long ret_ip)
{
@@ -179,13 +179,13 @@ static __always_inline bool check_memory_region_inline(unsigned long addr,
return !kasan_report(addr, size, write, ret_ip);
}

-bool check_memory_region(unsigned long addr, size_t size, bool write,
- unsigned long ret_ip)
+bool kasan_check_range(unsigned long addr, size_t size, bool write,
+ unsigned long ret_ip)
{
- return check_memory_region_inline(addr, size, write, ret_ip);

+ return check_region_inline(addr, size, write, ret_ip);
}

-bool check_invalid_free(void *addr)
+bool kasan_check_invalid_free(void *addr)

{
s8 shadow_byte = READ_ONCE(*(s8 *)kasan_mem_to_shadow(addr));

diff --git a/mm/kasan/kasan.h b/mm/kasan/kasan.h
index cc4d9e1d49b1..3b38baddec47 100644
--- a/mm/kasan/kasan.h
+++ b/mm/kasan/kasan.h

+static inline void kasan_unpoison(const void *address, size_t size)
{
hw_set_mem_tag_range(kasan_reset_tag(address),
round_up(size, KASAN_GRANULE_SIZE), get_tag(address));
}

-static inline bool check_invalid_free(void *addr)

+static inline bool kasan_check_invalid_free(void *addr)
{
u8 ptr_tag = get_tag(addr);

u8 mem_tag = hw_get_mem_tag(addr);
@@ -325,9 +325,9 @@ static inline bool check_invalid_free(void *addr)

#else /* CONFIG_KASAN_HW_TAGS */

-void poison_range(const void *address, size_t size, u8 value);
-void unpoison_range(const void *address, size_t size);
-bool check_invalid_free(void *addr);
+void kasan_poison(const void *address, size_t size, u8 value);
+void kasan_unpoison(const void *address, size_t size);
+bool kasan_check_invalid_free(void *addr);

#endif /* CONFIG_KASAN_HW_TAGS */

diff --git a/mm/kasan/sw_tags.c b/mm/kasan/sw_tags.c
index 5dcd830805b2..cc271fceb5d5 100644
--- a/mm/kasan/sw_tags.c
+++ b/mm/kasan/sw_tags.c

@@ -57,7 +57,7 @@ void __init kasan_init_sw_tags(void)
* sequence has in fact positive effect, since interrupts that randomly skew
* PRNG at unpredictable points do only good.
*/
-u8 random_tag(void)
+u8 kasan_random_tag(void)
{
u32 state = this_cpu_read(prng_state);

@@ -67,7 +67,7 @@ u8 random_tag(void)
return (u8)(state % (KASAN_TAG_MAX + 1));
}

-bool check_memory_region(unsigned long addr, size_t size, bool write,
+bool kasan_check_range(unsigned long addr, size_t size, bool write,
unsigned long ret_ip)
{
u8 tag;

@@ -118,7 +118,7 @@ bool check_memory_region(unsigned long addr, size_t size, bool write,
return true;
}

-bool check_invalid_free(void *addr)
+bool kasan_check_invalid_free(void *addr)

{
u8 tag = get_tag(addr);
u8 shadow_byte = READ_ONCE(*(u8 *)kasan_mem_to_shadow(kasan_reset_tag(addr)));

2.30.0.284.gd98b1dd5eaa7-goog

Andrey Konovalov

unread,

Jan 13, 2021, 11:21:51 AM1/13/21

to Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

Mention in the documentation that enabling CONFIG_KASAN_HW_TAGS
always results in in-kernel TBI (Top Byte Ignore) being enabled.

Also do a few minor documentation cleanups.

Link: https://linux-review.googlesource.com/id/Iba2a6697e3c6304cb53f89ec61dedc77fa29e3ae

Reviewed-by: Marco Elver <el...@google.com>
Reviewed-by: Alexander Potapenko <gli...@google.com>
Signed-off-by: Andrey Konovalov <andre...@google.com>
---

Documentation/dev-tools/kasan.rst | 16 +++++++++++-----
1 file changed, 11 insertions(+), 5 deletions(-)

diff --git a/Documentation/dev-tools/kasan.rst b/Documentation/dev-tools/kasan.rst
index 0fc3fb1860c4..26c99852a852 100644
--- a/Documentation/dev-tools/kasan.rst
+++ b/Documentation/dev-tools/kasan.rst

2.30.0.284.gd98b1dd5eaa7-goog

Andrey Konovalov

unread,

Jan 13, 2021, 11:21:54 AM1/13/21

to Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

Clarify and update comments in KASAN tests.

Link: https://linux-review.googlesource.com/id/I6c816c51fa1e0eb7aa3dead6bda1f339d2af46c8

Signed-off-by: Andrey Konovalov <andre...@google.com>
---

lib/test_kasan.c | 59 +++++++++++++++++++++++++----------------
lib/test_kasan_module.c | 5 ++--
2 files changed, 39 insertions(+), 25 deletions(-)

diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index 2947274cc2d3..6f46e27c2af7 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c

@@ -28,10 +28,9 @@
#define OOB_TAG_OFF (IS_ENABLED(CONFIG_KASAN_GENERIC) ? 0 : KASAN_GRANULE_SIZE)

/*
- * We assign some test results to these globals to make sure the tests
- * are not eliminated as dead code.
+ * Some tests use these global variables to store return values from function
+ * calls that could otherwise be eliminated by the compiler as dead code.
*/
-
void *kasan_ptr_result;
int kasan_int_result;

@@ -39,14 +38,13 @@ static struct kunit_resource resource;
static struct kunit_kasan_expectation fail_data;
static bool multishot;

+/*
+ * Temporarily enable multi-shot mode. Otherwise, KASAN would only report the
+ * first detected bug and panic the kernel if panic_on_warn is enabled.
+ */

static int kasan_test_init(struct kunit *test)
{
- /*
- * Temporarily enable multi-shot mode and set panic_on_warn=0.
- * Otherwise, we'd only get a report for the first case.
- */
multishot = kasan_save_enable_multi_shot();

-
return 0;
}

@@ -56,12 +54,12 @@ static void kasan_test_exit(struct kunit *test)
}

/**
- * KUNIT_EXPECT_KASAN_FAIL() - Causes a test failure when the expression does
- * not cause a KASAN error. This uses a KUnit resource named "kasan_data." Do
- * Do not use this name for a KUnit resource outside here.
- *
+ * KUNIT_EXPECT_KASAN_FAIL() - check that the executed expression produces a
+ * KASAN report; causes a test failure otherwise. This relies on a KUnit
+ * resource named "kasan_data". Do not use this name for KUnit resources
+ * outside of KASAN tests.
*/
-#define KUNIT_EXPECT_KASAN_FAIL(test, condition) do { \
+#define KUNIT_EXPECT_KASAN_FAIL(test, expression) do { \
fail_data.report_expected = true; \
fail_data.report_found = false; \
kunit_add_named_resource(test, \

@@ -69,7 +67,7 @@ static void kasan_test_exit(struct kunit *test)

NULL, \
&resource, \
"kasan_data", &fail_data); \

- condition; \
+ expression; \
KUNIT_EXPECT_EQ(test, \
fail_data.report_expected, \
fail_data.report_found); \

@@ -121,7 +119,8 @@ static void kmalloc_pagealloc_oob_right(struct kunit *test)
return;
}

- /* Allocate a chunk that does not fit into a SLUB cache to trigger
+ /*
+ * Allocate a chunk that does not fit into a SLUB cache to trigger

* the page allocator fallback.

*/
ptr = kmalloc(size, GFP_KERNEL);

@@ -168,7 +167,9 @@ static void kmalloc_large_oob_right(struct kunit *test)
{
char *ptr;

size_t size = KMALLOC_MAX_CACHE_SIZE - 256;
- /* Allocate a chunk that is large enough, but still fits into a slab

+
+ /*

+ * Allocate a chunk that is large enough, but still fits into a slab
* and does not trigger the page allocator fallback in SLUB.

*/
ptr = kmalloc(size, GFP_KERNEL);

@@ -469,10 +470,13 @@ static void ksize_unpoisons_memory(struct kunit *test)

ptr = kmalloc(size, GFP_KERNEL);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);

real_size = ksize(ptr);
- /* This access doesn't trigger an error. */

+
+ /* This access shouldn't trigger a KASAN report. */

ptr[size] = 'x';
- /* This one does. */
+
+ /* This one must. */
KUNIT_EXPECT_KASAN_FAIL(test, ptr[real_size] = 'y');
+
kfree(ptr);
}

@@ -568,7 +572,7 @@ static void kmem_cache_invalid_free(struct kunit *test)
return;
}

- /* Trigger invalid free, the object doesn't get freed */
+ /* Trigger invalid free, the object doesn't get freed. */
KUNIT_EXPECT_KASAN_FAIL(test, kmem_cache_free(cache, p + 1));

/*

@@ -585,7 +589,10 @@ static void kasan_memchr(struct kunit *test)

char *ptr;
size_t size = 24;

- /* See https://bugzilla.kernel.org/show_bug.cgi?id=206337 */
+ /*
+ * str* functions are not instrumented with CONFIG_AMD_MEM_ENCRYPT.
+ * See https://bugzilla.kernel.org/show_bug.cgi?id=206337 for details.
+ */
if (IS_ENABLED(CONFIG_AMD_MEM_ENCRYPT)) {

kunit_info(test,

"str* functions are not instrumented with CONFIG_AMD_MEM_ENCRYPT");

@@ -610,7 +617,10 @@ static void kasan_memcmp(struct kunit *test)

size_t size = 24;
int arr[9];

- /* See https://bugzilla.kernel.org/show_bug.cgi?id=206337 */
+ /*
+ * str* functions are not instrumented with CONFIG_AMD_MEM_ENCRYPT.
+ * See https://bugzilla.kernel.org/show_bug.cgi?id=206337 for details.
+ */
if (IS_ENABLED(CONFIG_AMD_MEM_ENCRYPT)) {

kunit_info(test,

"str* functions are not instrumented with CONFIG_AMD_MEM_ENCRYPT");

@@ -634,7 +644,10 @@ static void kasan_strings(struct kunit *test)

char *ptr;
size_t size = 24;

- /* See https://bugzilla.kernel.org/show_bug.cgi?id=206337 */
+ /*
+ * str* functions are not instrumented with CONFIG_AMD_MEM_ENCRYPT.
+ * See https://bugzilla.kernel.org/show_bug.cgi?id=206337 for details.
+ */
if (IS_ENABLED(CONFIG_AMD_MEM_ENCRYPT)) {

kunit_info(test,

"str* functions are not instrumented with CONFIG_AMD_MEM_ENCRYPT");

@@ -706,7 +719,7 @@ static void kasan_bitops_generic(struct kunit *test)
}

/*

- * Allocate 1 more byte, which causes kzalloc to round up to 16-bytes;
+ * Allocate 1 more byte, which causes kzalloc to round up to 16 bytes;
* this way we do not actually corrupt other memory.
*/
bits = kzalloc(sizeof(*bits) + 1, GFP_KERNEL);

diff --git a/lib/test_kasan_module.c b/lib/test_kasan_module.c
index 3b4cc77992d2..eee017ff8980 100644
--- a/lib/test_kasan_module.c
+++ b/lib/test_kasan_module.c
@@ -123,8 +123,9 @@ static noinline void __init kasan_workqueue_uaf(void)
static int __init test_kasan_module_init(void)
{

/*
- * Temporarily enable multi-shot mode. Otherwise, we'd only get a
- * report for the first case.

+ * Temporarily enable multi-shot mode. Otherwise, KASAN would only

+ * report the first detected bug and panic the kernel if panic_on_warn
+ * is enabled.
*/
bool multishot = kasan_save_enable_multi_shot();

--

2.30.0.284.gd98b1dd5eaa7-goog

Andrey Konovalov

unread,

Jan 13, 2021, 11:21:56 AM1/13/21

to Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

Some KASAN tests require specific kernel configs to be enabled.
Instead of copy-pasting the checks for these configs add a few helper
macros and use them.

Link: https://linux-review.googlesource.com/id/I237484a7fddfedf4a4aae9cc61ecbcdbe85a0a63
Suggested-by: Alexander Potapenko <gli...@google.com>

Signed-off-by: Andrey Konovalov <andre...@google.com>
---

lib/test_kasan.c | 101 +++++++++++++++--------------------------------
1 file changed, 31 insertions(+), 70 deletions(-)

diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index 6f46e27c2af7..714ea27fcc3e 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c
@@ -73,6 +73,20 @@ static void kasan_test_exit(struct kunit *test)
fail_data.report_found); \
} while (0)

+#define KASAN_TEST_NEEDS_CONFIG_ON(test, config) do { \
+ if (!IS_ENABLED(config)) { \
+ kunit_info((test), "skipping, " #config " required"); \
+ return; \
+ } \
+} while (0)
+
+#define KASAN_TEST_NEEDS_CONFIG_OFF(test, config) do { \
+ if (IS_ENABLED(config)) { \
+ kunit_info((test), "skipping, " #config " enabled"); \
+ return; \
+ } \
+} while (0)
+
static void kmalloc_oob_right(struct kunit *test)
{
char *ptr;
@@ -114,10 +128,7 @@ static void kmalloc_pagealloc_oob_right(struct kunit *test)
char *ptr;
size_t size = KMALLOC_MAX_CACHE_SIZE + 10;

- if (!IS_ENABLED(CONFIG_SLUB)) {

- kunit_info(test, "CONFIG_SLUB is not enabled.");

- return;
- }
+ KASAN_TEST_NEEDS_CONFIG_ON(test, CONFIG_SLUB);

/*

* Allocate a chunk that does not fit into a SLUB cache to trigger

@@ -135,10 +146,7 @@ static void kmalloc_pagealloc_uaf(struct kunit *test)
char *ptr;
size_t size = KMALLOC_MAX_CACHE_SIZE + 10;

- if (!IS_ENABLED(CONFIG_SLUB)) {

- kunit_info(test, "CONFIG_SLUB is not enabled.");

- return;
- }
+ KASAN_TEST_NEEDS_CONFIG_ON(test, CONFIG_SLUB);

ptr = kmalloc(size, GFP_KERNEL);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);

@@ -152,10 +160,7 @@ static void kmalloc_pagealloc_invalid_free(struct kunit *test)
char *ptr;
size_t size = KMALLOC_MAX_CACHE_SIZE + 10;

- if (!IS_ENABLED(CONFIG_SLUB)) {

- kunit_info(test, "CONFIG_SLUB is not enabled.");

- return;
- }
+ KASAN_TEST_NEEDS_CONFIG_ON(test, CONFIG_SLUB);

ptr = kmalloc(size, GFP_KERNEL);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);

@@ -218,10 +223,7 @@ static void kmalloc_oob_16(struct kunit *test)
} *ptr1, *ptr2;

/* This test is specifically crafted for the generic mode. */
- if (!IS_ENABLED(CONFIG_KASAN_GENERIC)) {
- kunit_info(test, "CONFIG_KASAN_GENERIC required\n");
- return;
- }
+ KASAN_TEST_NEEDS_CONFIG_ON(test, CONFIG_KASAN_GENERIC);

ptr1 = kmalloc(sizeof(*ptr1) - 3, GFP_KERNEL);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr1);
@@ -454,10 +456,7 @@ static void kasan_global_oob(struct kunit *test)
char *p = &global_array[ARRAY_SIZE(global_array) + i];

/* Only generic mode instruments globals. */
- if (!IS_ENABLED(CONFIG_KASAN_GENERIC)) {
- kunit_info(test, "CONFIG_KASAN_GENERIC required");
- return;
- }
+ KASAN_TEST_NEEDS_CONFIG_ON(test, CONFIG_KASAN_GENERIC);

KUNIT_EXPECT_KASAN_FAIL(test, *(volatile char *)p);
}
@@ -486,10 +485,7 @@ static void kasan_stack_oob(struct kunit *test)
volatile int i = OOB_TAG_OFF;
char *p = &stack_array[ARRAY_SIZE(stack_array) + i];

- if (!IS_ENABLED(CONFIG_KASAN_STACK)) {
- kunit_info(test, "CONFIG_KASAN_STACK is not enabled");
- return;
- }
+ KASAN_TEST_NEEDS_CONFIG_ON(test, CONFIG_KASAN_STACK);

KUNIT_EXPECT_KASAN_FAIL(test, *(volatile char *)p);
}
@@ -501,15 +497,8 @@ static void kasan_alloca_oob_left(struct kunit *test)
char *p = alloca_array - 1;

/* Only generic mode instruments dynamic allocas. */
- if (!IS_ENABLED(CONFIG_KASAN_GENERIC)) {
- kunit_info(test, "CONFIG_KASAN_GENERIC required");
- return;
- }
-
- if (!IS_ENABLED(CONFIG_KASAN_STACK)) {
- kunit_info(test, "CONFIG_KASAN_STACK is not enabled");
- return;
- }
+ KASAN_TEST_NEEDS_CONFIG_ON(test, CONFIG_KASAN_GENERIC);
+ KASAN_TEST_NEEDS_CONFIG_ON(test, CONFIG_KASAN_STACK);

KUNIT_EXPECT_KASAN_FAIL(test, *(volatile char *)p);
}
@@ -521,15 +510,8 @@ static void kasan_alloca_oob_right(struct kunit *test)
char *p = alloca_array + i;

/* Only generic mode instruments dynamic allocas. */
- if (!IS_ENABLED(CONFIG_KASAN_GENERIC)) {
- kunit_info(test, "CONFIG_KASAN_GENERIC required");
- return;
- }
-
- if (!IS_ENABLED(CONFIG_KASAN_STACK)) {
- kunit_info(test, "CONFIG_KASAN_STACK is not enabled");
- return;
- }
+ KASAN_TEST_NEEDS_CONFIG_ON(test, CONFIG_KASAN_GENERIC);
+ KASAN_TEST_NEEDS_CONFIG_ON(test, CONFIG_KASAN_STACK);

KUNIT_EXPECT_KASAN_FAIL(test, *(volatile char *)p);
}
@@ -593,11 +575,7 @@ static void kasan_memchr(struct kunit *test)

* str* functions are not instrumented with CONFIG_AMD_MEM_ENCRYPT.

* See https://bugzilla.kernel.org/show_bug.cgi?id=206337 for details.

*/
- if (IS_ENABLED(CONFIG_AMD_MEM_ENCRYPT)) {
- kunit_info(test,
- "str* functions are not instrumented with CONFIG_AMD_MEM_ENCRYPT");
- return;
- }
+ KASAN_TEST_NEEDS_CONFIG_OFF(test, CONFIG_AMD_MEM_ENCRYPT);

if (OOB_TAG_OFF)
size = round_up(size, OOB_TAG_OFF);
@@ -621,11 +599,7 @@ static void kasan_memcmp(struct kunit *test)

* str* functions are not instrumented with CONFIG_AMD_MEM_ENCRYPT.

* See https://bugzilla.kernel.org/show_bug.cgi?id=206337 for details.

*/
- if (IS_ENABLED(CONFIG_AMD_MEM_ENCRYPT)) {
- kunit_info(test,
- "str* functions are not instrumented with CONFIG_AMD_MEM_ENCRYPT");
- return;
- }
+ KASAN_TEST_NEEDS_CONFIG_OFF(test, CONFIG_AMD_MEM_ENCRYPT);

if (OOB_TAG_OFF)
size = round_up(size, OOB_TAG_OFF);
@@ -648,11 +622,7 @@ static void kasan_strings(struct kunit *test)

* str* functions are not instrumented with CONFIG_AMD_MEM_ENCRYPT.

* See https://bugzilla.kernel.org/show_bug.cgi?id=206337 for details.

*/
- if (IS_ENABLED(CONFIG_AMD_MEM_ENCRYPT)) {
- kunit_info(test,
- "str* functions are not instrumented with CONFIG_AMD_MEM_ENCRYPT");
- return;
- }
+ KASAN_TEST_NEEDS_CONFIG_OFF(test, CONFIG_AMD_MEM_ENCRYPT);

ptr = kmalloc(size, GFP_KERNEL | __GFP_ZERO);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);
@@ -713,10 +683,7 @@ static void kasan_bitops_generic(struct kunit *test)
long *bits;

/* This test is specifically crafted for the generic mode. */
- if (!IS_ENABLED(CONFIG_KASAN_GENERIC)) {
- kunit_info(test, "CONFIG_KASAN_GENERIC required\n");
- return;
- }
+ KASAN_TEST_NEEDS_CONFIG_ON(test, CONFIG_KASAN_GENERIC);

/*

* Allocate 1 more byte, which causes kzalloc to round up to 16 bytes;

@@ -744,11 +711,8 @@ static void kasan_bitops_tags(struct kunit *test)
{
long *bits;

- /* This test is specifically crafted for the tag-based mode. */
- if (IS_ENABLED(CONFIG_KASAN_GENERIC)) {
- kunit_info(test, "CONFIG_KASAN_SW_TAGS required\n");
- return;
- }
+ /* This test is specifically crafted for tag-based modes. */
+ KASAN_TEST_NEEDS_CONFIG_OFF(test, CONFIG_KASAN_GENERIC);

/* Allocation size will be rounded to up granule size, which is 16. */

bits = kzalloc(sizeof(*bits), GFP_KERNEL);
@@ -777,10 +741,7 @@ static void vmalloc_oob(struct kunit *test)
{
void *area;

- if (!IS_ENABLED(CONFIG_KASAN_VMALLOC)) {
- kunit_info(test, "CONFIG_KASAN_VMALLOC is not enabled.");
- return;
- }
+ KASAN_TEST_NEEDS_CONFIG_ON(test, CONFIG_KASAN_VMALLOC);

/*
* We have to be careful not to hit the guard page.
--
2.30.0.284.gd98b1dd5eaa7-goog

Andrey Konovalov

unread,

Jan 13, 2021, 11:21:59 AM1/13/21

to Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

Add 3 new tests for tag-based KASAN modes:

1. Check that match-all pointer tag is not assigned randomly.
2. Check that 0xff works as a match-all pointer tag.
3. Check that there are no match-all memory tags.

Note, that test #3 causes a significant number (255) of KASAN reports
to be printed during execution for the SW_TAGS mode.

Link: https://linux-review.googlesource.com/id/I78f1375efafa162b37f3abcb2c5bc2f3955dfd8e

Signed-off-by: Andrey Konovalov <andre...@google.com>
---

lib/test_kasan.c | 92 ++++++++++++++++++++++++++++++++++++++++++++++++
mm/kasan/kasan.h | 6 ++++
2 files changed, 98 insertions(+)

diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index 714ea27fcc3e..f5470bed50b6 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c

@@ -13,6 +13,7 @@
#include <linux/mman.h>
#include <linux/module.h>
#include <linux/printk.h>
+#include <linux/random.h>
#include <linux/slab.h>
#include <linux/string.h>
#include <linux/uaccess.h>

@@ -754,6 +755,94 @@ static void vmalloc_oob(struct kunit *test)
vfree(area);
}

+/*
+ * Check that the assigned pointer tag falls within the [KASAN_TAG_MIN,
+ * KASAN_TAG_KERNEL) range (note: excluding the match-all tag) for tag-based
+ * modes.

+ */
+static void match_all_not_assigned(struct kunit *test)
+{
+ char *ptr;
+ struct page *pages;
+ int i, size, order;
+

+ KASAN_TEST_NEEDS_CONFIG_OFF(test, CONFIG_KASAN_GENERIC);

+
+ for (i = 0; i < 256; i++) {

+ size = get_random_int() % 1024;
+ ptr = kmalloc(size, GFP_KERNEL);
+ KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);
+ KUNIT_EXPECT_GE(test, (u8)get_tag(ptr), (u8)KASAN_TAG_MIN);
+ KUNIT_EXPECT_LT(test, (u8)get_tag(ptr), (u8)KASAN_TAG_KERNEL);

+ kfree(ptr);
+ }
+
+ for (i = 0; i < 256; i++) {
+ order = get_random_int() % 4;
+ pages = alloc_pages(GFP_KERNEL, order);
+ ptr = page_address(pages);
+ KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);

+ KUNIT_EXPECT_GE(test, (u8)get_tag(ptr), (u8)KASAN_TAG_MIN);
+ KUNIT_EXPECT_LT(test, (u8)get_tag(ptr), (u8)KASAN_TAG_KERNEL);

+ free_pages((unsigned long)ptr, order);
+ }
+}
+
+/* Check that 0xff works as a match-all pointer tag for tag-based modes. */
+static void match_all_ptr_tag(struct kunit *test)
+{
+ char *ptr;
+ u8 tag;
+

+ KASAN_TEST_NEEDS_CONFIG_OFF(test, CONFIG_KASAN_GENERIC);

+
+ ptr = kmalloc(128, GFP_KERNEL);
+ KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);
+
+ /* Backup the assigned tag. */
+ tag = get_tag(ptr);
+ KUNIT_EXPECT_NE(test, tag, (u8)KASAN_TAG_KERNEL);
+
+ /* Reset the tag to 0xff.*/
+ ptr = set_tag(ptr, KASAN_TAG_KERNEL);

+
+ /* This access shouldn't trigger a KASAN report. */

+ *ptr = 0;
+
+ /* Recover the pointer tag and free. */
+ ptr = set_tag(ptr, tag);
+ kfree(ptr);
+}
+
+/* Check that there are no match-all memory tags for tag-based modes. */
+static void match_all_mem_tag(struct kunit *test)
+{
+ char *ptr;
+ int tag;
+

+ KASAN_TEST_NEEDS_CONFIG_OFF(test, CONFIG_KASAN_GENERIC);

+
+ ptr = kmalloc(128, GFP_KERNEL);
+ KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);
+ KUNIT_EXPECT_NE(test, (u8)get_tag(ptr), (u8)KASAN_TAG_KERNEL);
+
+ /* For each possible tag value not matching the pointer tag. */
+ for (tag = KASAN_TAG_MIN; tag <= KASAN_TAG_KERNEL; tag++) {
+ if (tag == get_tag(ptr))
+ continue;
+
+ /* Mark the first memory granule with the chosen memory tag. */
+ kasan_poison(ptr, KASAN_GRANULE_SIZE, (u8)tag);

+

+ /* This access must cause a KASAN report. */
+ KUNIT_EXPECT_KASAN_FAIL(test, *ptr = 0);
+ }
+
+ /* Recover the memory tag and free. */
+ kasan_poison(ptr, KASAN_GRANULE_SIZE, get_tag(ptr));
+ kfree(ptr);
+}
+
static struct kunit_case kasan_kunit_test_cases[] = {
KUNIT_CASE(kmalloc_oob_right),
KUNIT_CASE(kmalloc_oob_left),

@@ -793,6 +882,9 @@ static struct kunit_case kasan_kunit_test_cases[] = {

KUNIT_CASE(kasan_bitops_tags),
KUNIT_CASE(kmalloc_double_kzfree),
KUNIT_CASE(vmalloc_oob),
+ KUNIT_CASE(match_all_not_assigned),
+ KUNIT_CASE(match_all_ptr_tag),
+ KUNIT_CASE(match_all_mem_tag),
{}
};

diff --git a/mm/kasan/kasan.h b/mm/kasan/kasan.h
index 3b38baddec47..c3fb9bf241d3 100644
--- a/mm/kasan/kasan.h
+++ b/mm/kasan/kasan.h

@@ -36,6 +36,12 @@ extern bool kasan_flag_panic __ro_after_init;
#define KASAN_TAG_INVALID 0xFE /* inaccessible memory tag */
#define KASAN_TAG_MAX 0xFD /* maximum value for random tags */

+#ifdef CONFIG_KASAN_HW_TAGS
+#define KASAN_TAG_MIN 0xF0 /* mimimum value for random tags */
+#else
+#define KASAN_TAG_MIN 0x00 /* mimimum value for random tags */
+#endif
+
#ifdef CONFIG_KASAN_GENERIC
#define KASAN_FREE_PAGE 0xFF /* page was freed */
#define KASAN_PAGE_REDZONE 0xFE /* redzone for kmalloc_large allocations */
--

2.30.0.284.gd98b1dd5eaa7-goog

Andrey Konovalov

unread,

Jan 13, 2021, 11:22:01 AM1/13/21

to Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

On a high level, this patch allows running KUnit KASAN tests with the
hardware tag-based KASAN mode.

Internally, this change reenables tag checking at the end of each KASAN
test that triggers a tag fault and leads to tag checking being disabled.

With this patch KASAN tests are still failing for the hardware tag-based
mode; fixes come in the next few patches.

Link: https://linux-review.googlesource.com/id/Id94dc9eccd33b23cda4950be408c27f879e474c8

Signed-off-by: Andrey Konovalov <andre...@google.com>
---

arch/arm64/include/asm/memory.h | 1 +
arch/arm64/include/asm/mte-kasan.h | 12 +++++++++
arch/arm64/kernel/mte.c | 12 +++++++++
arch/arm64/mm/fault.c | 16 +++++++-----
lib/Kconfig.kasan | 4 +--
lib/test_kasan.c | 42 +++++++++++++++++++++---------
mm/kasan/kasan.h | 9 +++++++
7 files changed, 75 insertions(+), 21 deletions(-)

diff --git a/arch/arm64/include/asm/memory.h b/arch/arm64/include/asm/memory.h
index 18fce223b67b..cedfc9e97bcc 100644
--- a/arch/arm64/include/asm/memory.h
+++ b/arch/arm64/include/asm/memory.h
@@ -232,6 +232,7 @@ static inline const void *__tag_set(const void *addr, u8 tag)

#ifdef CONFIG_KASAN_HW_TAGS
#define arch_enable_tagging() mte_enable_kernel()
+#define arch_set_tagging_report_once(state) mte_set_report_once(state)
#define arch_init_tags(max_tag) mte_init_tags(max_tag)
#define arch_get_random_tag() mte_get_random_tag()
#define arch_get_mem_tag(addr) mte_get_mem_tag(addr)
diff --git a/arch/arm64/include/asm/mte-kasan.h b/arch/arm64/include/asm/mte-kasan.h
index 26349a4b5e2e..3748d5bb88c0 100644
--- a/arch/arm64/include/asm/mte-kasan.h
+++ b/arch/arm64/include/asm/mte-kasan.h
@@ -32,6 +32,9 @@ void *mte_set_mem_tag_range(void *addr, size_t size, u8 tag);
void mte_enable_kernel(void);
void mte_init_tags(u64 max_tag);

+void mte_set_report_once(bool state);
+bool mte_report_once(void);
+
#else /* CONFIG_ARM64_MTE */

static inline u8 mte_get_ptr_tag(void *ptr)
@@ -60,6 +63,15 @@ static inline void mte_init_tags(u64 max_tag)
{
}

+static inline void mte_set_report_once(bool state)
+{
+}
+
+static inline bool mte_report_once(void)
+{

+ return false;
+}
+

#endif /* CONFIG_ARM64_MTE */

#endif /* __ASSEMBLY__ */
diff --git a/arch/arm64/kernel/mte.c b/arch/arm64/kernel/mte.c
index dc9ada64feed..c63b3d7a3cd9 100644
--- a/arch/arm64/kernel/mte.c
+++ b/arch/arm64/kernel/mte.c
@@ -25,6 +25,8 @@

u64 gcr_kernel_excl __ro_after_init;

+static bool report_fault_once = true;
+
static void mte_sync_page_tags(struct page *page, pte_t *ptep, bool check_swap)
{
pte_t old_pte = READ_ONCE(*ptep);
@@ -158,6 +160,16 @@ void mte_enable_kernel(void)
isb();
}

+void mte_set_report_once(bool state)
+{
+ WRITE_ONCE(report_fault_once, state);
+}
+
+bool mte_report_once(void)
+{
+ return READ_ONCE(report_fault_once);
+}
+
static void update_sctlr_el1_tcf0(u64 tcf0)
{
/* ISB required for the kernel uaccess routines */
diff --git a/arch/arm64/mm/fault.c b/arch/arm64/mm/fault.c
index a218f6f2fdc8..f1b77dc79948 100644
--- a/arch/arm64/mm/fault.c
+++ b/arch/arm64/mm/fault.c
@@ -302,7 +302,14 @@ static void die_kernel_fault(const char *msg, unsigned long addr,

static void report_tag_fault(unsigned long addr, unsigned int esr,
struct pt_regs *regs)
{
- bool is_write = ((esr & ESR_ELx_WNR) >> ESR_ELx_WNR_SHIFT) != 0;
+ static bool reported;
+ bool is_write;
+
+ if (READ_ONCE(reported))
+ return;
+
+ if (mte_report_once())
+ WRITE_ONCE(reported, true);

/* The format of KASAN tags is 0xF<x>. */
addr |= (0xF0UL << MTE_TAG_SHIFT);
@@ -310,6 +317,7 @@ static void report_tag_fault(unsigned long addr, unsigned int esr,

* SAS bits aren't set for all faults reported in EL1, so we can't
* find out access size.
*/
+ is_write = ((esr & ESR_ELx_WNR) >> ESR_ELx_WNR_SHIFT) != 0;

kasan_report(addr, 0, is_write, regs->pc);
}

#else
@@ -321,12 +329,8 @@ static inline void report_tag_fault(unsigned long addr, unsigned int esr,
static void do_tag_recovery(unsigned long addr, unsigned int esr,
struct pt_regs *regs)
{
- static bool reported;

- if (!READ_ONCE(reported)) {
- report_tag_fault(addr, esr, regs);
- WRITE_ONCE(reported, true);
- }
+ report_tag_fault(addr, esr, regs);

/*
* Disable MTE Tag Checking on the local CPU for the current EL.
diff --git a/lib/Kconfig.kasan b/lib/Kconfig.kasan
index f5fa4ba126bf..3091432acb0a 100644
--- a/lib/Kconfig.kasan
+++ b/lib/Kconfig.kasan
@@ -190,11 +190,11 @@ config KASAN_KUNIT_TEST
kernel debugging features like KASAN.

For more information on KUnit and unit tests in general, please refer

- to the KUnit documentation in Documentation/dev-tools/kunit
+ to the KUnit documentation in Documentation/dev-tools/kunit.

config TEST_KASAN_MODULE

tristate "KUnit-incompatible tests of KASAN bug detection capabilities"

- depends on m && KASAN
+ depends on m && KASAN && !KASAN_HW_TAGS
help
This is a part of the KASAN test suite that is incompatible with
KUnit. Currently includes tests that do bad copy_from/to_user
diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index f5470bed50b6..5c8aa3a5ce93 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c
@@ -41,16 +41,20 @@ static bool multishot;

/*

* Temporarily enable multi-shot mode. Otherwise, KASAN would only report the

- * first detected bug and panic the kernel if panic_on_warn is enabled.
+ * first detected bug and panic the kernel if panic_on_warn is enabled. For
+ * hardware tag-based KASAN also allow tag checking to be reenabled for each
+ * test, see the comment for KUNIT_EXPECT_KASAN_FAIL().

*/
static int kasan_test_init(struct kunit *test)
{

multishot = kasan_save_enable_multi_shot();
+ hw_set_tagging_report_once(false);
return 0;

}

static void kasan_test_exit(struct kunit *test)

{
+ hw_set_tagging_report_once(true);
kasan_restore_multi_shot(multishot);
}

@@ -59,19 +63,31 @@ static void kasan_test_exit(struct kunit *test)

* KASAN report; causes a test failure otherwise. This relies on a KUnit

* resource named "kasan_data". Do not use this name for KUnit resources

* outside of KASAN tests.

+ *
+ * For hardware tag-based KASAN, when a tag fault happens, tag checking is
+ * normally auto-disabled. When this happens, this test handler reenables
+ * tag checking. As tag checking can be only disabled or enabled per CPU, this
+ * handler disables migration (preemption).
*/
-#define KUNIT_EXPECT_KASAN_FAIL(test, expression) do { \
- fail_data.report_expected = true; \
- fail_data.report_found = false; \
- kunit_add_named_resource(test, \
- NULL, \
- NULL, \
- &resource, \
- "kasan_data", &fail_data); \
- expression; \
- KUNIT_EXPECT_EQ(test, \
- fail_data.report_expected, \
- fail_data.report_found); \
+#define KUNIT_EXPECT_KASAN_FAIL(test, expression) do { \
+ if (IS_ENABLED(CONFIG_KASAN_HW_TAGS)) \
+ migrate_disable(); \
+ fail_data.report_expected = true; \
+ fail_data.report_found = false; \
+ kunit_add_named_resource(test, \
+ NULL, \
+ NULL, \
+ &resource, \
+ "kasan_data", &fail_data); \
+ expression; \
+ KUNIT_EXPECT_EQ(test, \
+ fail_data.report_expected, \
+ fail_data.report_found); \
+ if (IS_ENABLED(CONFIG_KASAN_HW_TAGS)) { \
+ if (fail_data.report_found) \
+ hw_enable_tagging(); \
+ migrate_enable(); \
+ } \
} while (0)

#define KASAN_TEST_NEEDS_CONFIG_ON(test, config) do { \
diff --git a/mm/kasan/kasan.h b/mm/kasan/kasan.h
index c3fb9bf241d3..292dfbc37deb 100644
--- a/mm/kasan/kasan.h
+++ b/mm/kasan/kasan.h
@@ -280,6 +280,9 @@ static inline const void *arch_kasan_set_tag(const void *addr, u8 tag)
#ifndef arch_init_tags
#define arch_init_tags(max_tag)
#endif
+#ifndef arch_set_tagging_report_once
+#define arch_set_tagging_report_once(state)
+#endif
#ifndef arch_get_random_tag
#define arch_get_random_tag() (0xFF)
#endif
@@ -292,10 +295,16 @@ static inline const void *arch_kasan_set_tag(const void *addr, u8 tag)

#define hw_enable_tagging() arch_enable_tagging()
#define hw_init_tags(max_tag) arch_init_tags(max_tag)
+#define hw_set_tagging_report_once(state) arch_set_tagging_report_once(state)
#define hw_get_random_tag() arch_get_random_tag()
#define hw_get_mem_tag(addr) arch_get_mem_tag(addr)
#define hw_set_mem_tag_range(addr, size, tag) arch_set_mem_tag_range((addr), (size), (tag))

+#else /* CONFIG_KASAN_HW_TAGS */
+
+#define hw_enable_tagging()
+#define hw_set_tagging_report_once(state)
+

#endif /* CONFIG_KASAN_HW_TAGS */

#ifdef CONFIG_KASAN_SW_TAGS

--
2.30.0.284.gd98b1dd5eaa7-goog

Andrey Konovalov

unread,

Jan 13, 2021, 11:22:04 AM1/13/21

to Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

Rename CONFIG_TEST_KASAN_MODULE to CONFIG_KASAN_MODULE_TEST.

This naming is more consistent with the existing CONFIG_KASAN_KUNIT_TEST.

Link: https://linux-review.googlesource.com/id/Id347dfa5fe8788b7a1a189863e039f409da0ae5f

Reviewed-by: Marco Elver <el...@google.com>
Reviewed-by: Alexander Potapenko <gli...@google.com>

Signed-off-by: Andrey Konovalov <andre...@google.com>
---

Documentation/dev-tools/kasan.rst | 8 ++++----

lib/Kconfig.kasan | 2 +-
lib/Makefile | 2 +-

3 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/Documentation/dev-tools/kasan.rst b/Documentation/dev-tools/kasan.rst
index 26c99852a852..b25ae43d683e 100644
--- a/Documentation/dev-tools/kasan.rst
+++ b/Documentation/dev-tools/kasan.rst
@@ -374,17 +374,17 @@ unmapped. This will require changes in arch-specific code.

This allows ``VMAP_STACK`` support on x86, and can simplify support of
architectures that do not have a fixed module region.

-CONFIG_KASAN_KUNIT_TEST & CONFIG_TEST_KASAN_MODULE
---------------------------------------------------
+CONFIG_KASAN_KUNIT_TEST and CONFIG_KASAN_MODULE_TEST
+----------------------------------------------------

-KASAN tests consist on two parts:
+KASAN tests consist of two parts:

1. Tests that are integrated with the KUnit Test Framework. Enabled with
``CONFIG_KASAN_KUNIT_TEST``. These tests can be run and partially verified

automatically in a few different ways, see the instructions below.

2. Tests that are currently incompatible with KUnit. Enabled with
-``CONFIG_TEST_KASAN_MODULE`` and can only be run as a module. These tests can
+``CONFIG_KASAN_MODULE_TEST`` and can only be run as a module. These tests can
only be verified manually, by loading the kernel module and inspecting the
kernel log for KASAN reports.

diff --git a/lib/Kconfig.kasan b/lib/Kconfig.kasan
index 3091432acb0a..624ae1df7984 100644
--- a/lib/Kconfig.kasan
+++ b/lib/Kconfig.kasan

@@ -192,7 +192,7 @@ config KASAN_KUNIT_TEST

For more information on KUnit and unit tests in general, please refer

to the KUnit documentation in Documentation/dev-tools/kunit.

-config TEST_KASAN_MODULE
+config KASAN_MODULE_TEST

tristate "KUnit-incompatible tests of KASAN bug detection capabilities"

depends on m && KASAN && !KASAN_HW_TAGS
help

diff --git a/lib/Makefile b/lib/Makefile
index afeff05fa8c5..122f25d6407e 100644
--- a/lib/Makefile
+++ b/lib/Makefile
@@ -68,7 +68,7 @@ obj-$(CONFIG_TEST_IDA) += test_ida.o
obj-$(CONFIG_KASAN_KUNIT_TEST) += test_kasan.o
CFLAGS_test_kasan.o += -fno-builtin
CFLAGS_test_kasan.o += $(call cc-disable-warning, vla)
-obj-$(CONFIG_TEST_KASAN_MODULE) += test_kasan_module.o
+obj-$(CONFIG_KASAN_MODULE_TEST) += test_kasan_module.o
CFLAGS_test_kasan_module.o += -fno-builtin

obj-$(CONFIG_TEST_UBSAN) += test_ubsan.o
CFLAGS_test_ubsan.o += $(call cc-disable-warning, vla)
--
2.30.0.284.gd98b1dd5eaa7-goog

Andrey Konovalov

unread,

Jan 13, 2021, 11:22:06 AM1/13/21

to Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

It might not be obvious to the compiler that the expression must be
executed between writing and reading to fail_data. In this case, the
compiler might reorder or optimize away some of the accesses, and
the tests will fail.

Add compiler barriers around the expression in KUNIT_EXPECT_KASAN_FAIL
and use READ/WRITE_ONCE() for accessing fail_data fields.

Link: https://linux-review.googlesource.com/id/I046079f48641a1d36fe627fc8827a9249102fd50

Signed-off-by: Andrey Konovalov <andre...@google.com>
---

lib/test_kasan.c | 17 ++++++++++++-----
mm/kasan/report.c | 2 +-
2 files changed, 13 insertions(+), 6 deletions(-)

diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index 5c8aa3a5ce93..283feda9882a 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c
@@ -68,23 +68,30 @@ static void kasan_test_exit(struct kunit *test)

* normally auto-disabled. When this happens, this test handler reenables

* tag checking. As tag checking can be only disabled or enabled per CPU, this

* handler disables migration (preemption).

+ *
+ * Since the compiler doesn't see that the expression can change the fail_data
+ * fields, it can reorder or optimize away the accesses to those fields.
+ * Use READ/WRITE_ONCE() for the accesses and compiler barriers around the
+ * expression to prevent that.
*/
#define KUNIT_EXPECT_KASAN_FAIL(test, expression) do { \
if (IS_ENABLED(CONFIG_KASAN_HW_TAGS)) \
migrate_disable(); \

- fail_data.report_expected = true; \
- fail_data.report_found = false; \

+ WRITE_ONCE(fail_data.report_expected, true); \
+ WRITE_ONCE(fail_data.report_found, false); \
kunit_add_named_resource(test, \
NULL, \

NULL, \
&resource, \
"kasan_data", &fail_data); \

+ barrier(); \
expression; \
+ barrier(); \

KUNIT_EXPECT_EQ(test, \
- fail_data.report_expected, \
- fail_data.report_found); \

+ READ_ONCE(fail_data.report_expected), \
+ READ_ONCE(fail_data.report_found)); \
if (IS_ENABLED(CONFIG_KASAN_HW_TAGS)) { \
- if (fail_data.report_found) \
+ if (READ_ONCE(fail_data.report_found)) \
hw_enable_tagging(); \
migrate_enable(); \
} \
diff --git a/mm/kasan/report.c b/mm/kasan/report.c
index e93d7973792e..234f35a84f19 100644
--- a/mm/kasan/report.c
+++ b/mm/kasan/report.c
@@ -331,7 +331,7 @@ static void kasan_update_kunit_status(struct kunit *cur_test)
}

kasan_data = (struct kunit_kasan_expectation *)resource->data;
- kasan_data->report_found = true;
+ WRITE_ONCE(kasan_data->report_found, true);
kunit_put_resource(resource);
}
#endif /* IS_ENABLED(CONFIG_KUNIT) */
--
2.30.0.284.gd98b1dd5eaa7-goog

Andrey Konovalov

unread,

Jan 13, 2021, 11:22:08 AM1/13/21

to Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

In the kmalloc_uaf2() test, the pointers to the two allocated memory

blocks might happen to be the same, and the test will fail. With the
software tag-based mode, the probability of the that is 1/254, so it's

hard to observe the failure. For the hardware tag-based mode though,
the probablity is 1/14, which is quite noticable.

Allow up to 16 attempts at generating different tags for the tag-based
modes.

Link: https://linux-review.googlesource.com/id/Ibfa458ef2804ff465d8eb07434a300bf36388d55

Signed-off-by: Andrey Konovalov <andre...@google.com>
---

lib/test_kasan.c | 11 +++++++++++
1 file changed, 11 insertions(+)

diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index 283feda9882a..a1a35d75ee1e 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c
@@ -382,7 +382,9 @@ static void kmalloc_uaf2(struct kunit *test)

{
char *ptr1, *ptr2;
size_t size = 43;
+ int counter = 0;

+again:
ptr1 = kmalloc(size, GFP_KERNEL);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr1);

@@ -391,6 +393,15 @@ static void kmalloc_uaf2(struct kunit *test)

ptr2 = kmalloc(size, GFP_KERNEL);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr2);

+ /*
+ * For tag-based KASAN ptr1 and ptr2 tags might happen to be the same.

+ * Allow up to 16 attempts at generating different tags.
+ */
+ if (!IS_ENABLED(CONFIG_KASAN_GENERIC) && ptr1 == ptr2 && counter++ < 16) {
+ kfree(ptr2);
+ goto again;
+ }
+
KUNIT_EXPECT_KASAN_FAIL(test, ptr1[40] = 'x');
KUNIT_EXPECT_PTR_NE(test, ptr1, ptr2);

--
2.30.0.284.gd98b1dd5eaa7-goog

Andrey Konovalov

unread,

Jan 13, 2021, 11:22:11 AM1/13/21

to Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

Since the hardware tag-based KASAN mode might not have a redzone that
comes after an allocated object (when kasan.mode=prod is enabled), the
kasan_bitops_tags() test ends up corrupting the next object in memory.

Change the test so it always accesses the redzone that lies within the
allocated object's boundaries.

Link: https://linux-review.googlesource.com/id/I67f51d1ee48f0a8d0fe2658c2a39e4879fe0832a

Reviewed-by: Marco Elver <el...@google.com>
Reviewed-by: Alexander Potapenko <gli...@google.com>

Signed-off-by: Andrey Konovalov <andre...@google.com>
---

lib/test_kasan.c | 10 +++++-----
1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index a1a35d75ee1e..63252d1fd58c 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c
@@ -749,13 +749,13 @@ static void kasan_bitops_tags(struct kunit *test)

/* This test is specifically crafted for tag-based modes. */

KASAN_TEST_NEEDS_CONFIG_OFF(test, CONFIG_KASAN_GENERIC);

- /* Allocation size will be rounded to up granule size, which is 16. */
- bits = kzalloc(sizeof(*bits), GFP_KERNEL);
+ /* kmalloc-64 cache will be used and the last 16 bytes will be the redzone. */
+ bits = kzalloc(48, GFP_KERNEL);

KUNIT_ASSERT_NOT_ERR_OR_NULL(test, bits);

- /* Do the accesses past the 16 allocated bytes. */
- kasan_bitops_modify(test, BITS_PER_LONG, &bits[1]);
- kasan_bitops_test_and_modify(test, BITS_PER_LONG + BITS_PER_BYTE, &bits[1]);
+ /* Do the accesses past the 48 allocated bytes, but within the redone. */
+ kasan_bitops_modify(test, BITS_PER_LONG, (void *)bits + 48);
+ kasan_bitops_test_and_modify(test, BITS_PER_LONG + BITS_PER_BYTE, (void *)bits + 48);

kfree(bits);
}
--
2.30.0.284.gd98b1dd5eaa7-goog

Andrey Konovalov

unread,

Jan 13, 2021, 11:22:14 AM1/13/21

to Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

The currently existing kasan_check_read/write() annotations are intended
to be used for kernel modules that have KASAN compiler instrumentation
disabled. Thus, they are only relevant for the software KASAN modes that
rely on compiler instrumentation.

However there's another use case for these annotations: ksize() checks
that the object passed to it is indeed accessible before unpoisoning the
whole object. This is currently done via __kasan_check_read(), which is
compiled away for the hardware tag-based mode that doesn't rely on
compiler instrumentation. This leads to KASAN missing detecting some
memory corruptions.

Provide another annotation called kasan_check_byte() that is available
for all KASAN modes. As the implementation rename and reuse
kasan_check_invalid_free(). Use this new annotation in ksize().

Also add a new ksize_uaf() test that checks that a use-after-free is
detected via ksize() itself, and via plain accesses that happen later.

Link: https://linux-review.googlesource.com/id/Iaabf771881d0f9ce1b969f2a62938e99d3308ec5

Signed-off-by: Andrey Konovalov <andre...@google.com>
---

include/linux/kasan-checks.h | 6 ++++++
include/linux/kasan.h | 16 ++++++++++++++++
lib/test_kasan.c | 20 ++++++++++++++++++++
mm/kasan/common.c | 11 ++++++++++-
mm/kasan/generic.c | 4 ++--
mm/kasan/kasan.h | 10 +++++-----
mm/kasan/sw_tags.c | 6 +++---
mm/slab_common.c | 15 +++++++++------
8 files changed, 71 insertions(+), 17 deletions(-)

diff --git a/include/linux/kasan-checks.h b/include/linux/kasan-checks.h
index ca5e89fb10d3..3d6d22a25bdc 100644
--- a/include/linux/kasan-checks.h
+++ b/include/linux/kasan-checks.h
@@ -4,6 +4,12 @@

#include <linux/types.h>

+/*
+ * The annotations present in this file are only relevant for the software
+ * KASAN modes that rely on compiler instrumentation, and will be optimized
+ * away for the hardware tag-based KASAN mode. Use kasan_check_byte() instead.
+ */
+
/*
* __kasan_check_*: Always available when KASAN is enabled. This may be used
* even in compilation units that selectively disable KASAN, but must use KASAN
diff --git a/include/linux/kasan.h b/include/linux/kasan.h
index 5e0655fb2a6f..b723895b157c 100644
--- a/include/linux/kasan.h
+++ b/include/linux/kasan.h
@@ -243,6 +243,18 @@ static __always_inline void kasan_kfree_large(void *ptr, unsigned long ip)
__kasan_kfree_large(ptr, ip);

}

+/*
+ * Unlike kasan_check_read/write(), kasan_check_byte() is performed even for
+ * the hardware tag-based mode that doesn't rely on compiler instrumentation.
+ */

+bool __kasan_check_byte(const void *addr, unsigned long ip);
+static __always_inline bool kasan_check_byte(const void *addr, unsigned long ip)
+{
+ if (kasan_enabled())
+ return __kasan_check_byte(addr, ip);
+ return true;
+}
+
bool kasan_save_enable_multi_shot(void);
void kasan_restore_multi_shot(bool enabled);

@@ -299,6 +311,10 @@ static inline void *kasan_krealloc(const void *object, size_t new_size,
return (void *)object;
}
static inline void kasan_kfree_large(void *ptr, unsigned long ip) {}
+static inline bool kasan_check_byte(const void *address, unsigned long ip)
+{
+ return true;
+}

#endif /* CONFIG_KASAN */

diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index 63252d1fd58c..710e714dc0cb 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c
@@ -496,6 +496,7 @@ static void kasan_global_oob(struct kunit *test)
KUNIT_EXPECT_KASAN_FAIL(test, *(volatile char *)p);
}

+/* Check that ksize() makes the whole object accessible. */
static void ksize_unpoisons_memory(struct kunit *test)
{
char *ptr;
@@ -514,6 +515,24 @@ static void ksize_unpoisons_memory(struct kunit *test)
kfree(ptr);
}

+/*
+ * Check that a use-after-free is detected by ksize() and via normal accesses
+ * after it.
+ */
+static void ksize_uaf(struct kunit *test)
+{
+ char *ptr;
+ int size = 128 - KASAN_GRANULE_SIZE;
+

+ ptr = kmalloc(size, GFP_KERNEL);
+ KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);

+ kfree(ptr);
+
+ KUNIT_EXPECT_KASAN_FAIL(test, ksize(ptr));
+ KUNIT_EXPECT_KASAN_FAIL(test, kasan_int_result = *ptr);
+ KUNIT_EXPECT_KASAN_FAIL(test, kasan_int_result = *(ptr + size));
+}
+

static void kasan_stack_oob(struct kunit *test)

{
char stack_array[10];
@@ -907,6 +926,7 @@ static struct kunit_case kasan_kunit_test_cases[] = {
KUNIT_CASE(kasan_alloca_oob_left),
KUNIT_CASE(kasan_alloca_oob_right),
KUNIT_CASE(ksize_unpoisons_memory),
+ KUNIT_CASE(ksize_uaf),
KUNIT_CASE(kmem_cache_double_free),
KUNIT_CASE(kmem_cache_invalid_free),
KUNIT_CASE(kasan_memchr),
diff --git a/mm/kasan/common.c b/mm/kasan/common.c
index eedc3e0fe365..b18189ef3a92 100644
--- a/mm/kasan/common.c
+++ b/mm/kasan/common.c
@@ -345,7 +345,7 @@ static bool ____kasan_slab_free(struct kmem_cache *cache, void *object,

if (unlikely(cache->flags & SLAB_TYPESAFE_BY_RCU))
return false;

- if (kasan_check_invalid_free(tagged_object)) {
+ if (!kasan_byte_accessible(tagged_object)) {
kasan_report_invalid_free(tagged_object, ip);
return true;
}
@@ -490,3 +490,12 @@ void __kasan_kfree_large(void *ptr, unsigned long ip)
kasan_report_invalid_free(ptr, ip);
/* The object will be poisoned by kasan_free_pages(). */
}
+

+bool __kasan_check_byte(const void *address, unsigned long ip)
+{

+ if (!kasan_byte_accessible(address)) {
+ kasan_report((unsigned long)address, 1, false, ip);
+ return false;
+ }
+ return true;
+}
diff --git a/mm/kasan/generic.c b/mm/kasan/generic.c
index acab8862dc67..3f17a1218055 100644
--- a/mm/kasan/generic.c
+++ b/mm/kasan/generic.c
@@ -185,11 +185,11 @@ bool kasan_check_range(unsigned long addr, size_t size, bool write,

return check_region_inline(addr, size, write, ret_ip);
}

-bool kasan_check_invalid_free(void *addr)
+bool kasan_byte_accessible(const void *addr)

{
s8 shadow_byte = READ_ONCE(*(s8 *)kasan_mem_to_shadow(addr));

- return shadow_byte < 0 || shadow_byte >= KASAN_GRANULE_SIZE;
+ return shadow_byte >= 0 && shadow_byte < KASAN_GRANULE_SIZE;
}

void kasan_cache_shrink(struct kmem_cache *cache)
diff --git a/mm/kasan/kasan.h b/mm/kasan/kasan.h
index 292dfbc37deb..bd4ee6fab648 100644
--- a/mm/kasan/kasan.h
+++ b/mm/kasan/kasan.h
@@ -329,20 +329,20 @@ static inline void kasan_unpoison(const void *address, size_t size)
round_up(size, KASAN_GRANULE_SIZE), get_tag(address));
}

-static inline bool kasan_check_invalid_free(void *addr)
+static inline bool kasan_byte_accessible(const void *addr)
{
u8 ptr_tag = get_tag(addr);
- u8 mem_tag = hw_get_mem_tag(addr);
+ u8 mem_tag = hw_get_mem_tag((void *)addr);

- return (mem_tag == KASAN_TAG_INVALID) ||
- (ptr_tag != KASAN_TAG_KERNEL && ptr_tag != mem_tag);
+ return (mem_tag != KASAN_TAG_INVALID) &&
+ (ptr_tag == KASAN_TAG_KERNEL || ptr_tag == mem_tag);
}

#else /* CONFIG_KASAN_HW_TAGS */

void kasan_poison(const void *address, size_t size, u8 value);
void kasan_unpoison(const void *address, size_t size);
-bool kasan_check_invalid_free(void *addr);
+bool kasan_byte_accessible(const void *addr);

#endif /* CONFIG_KASAN_HW_TAGS */

diff --git a/mm/kasan/sw_tags.c b/mm/kasan/sw_tags.c
index cc271fceb5d5..94c2d33be333 100644
--- a/mm/kasan/sw_tags.c
+++ b/mm/kasan/sw_tags.c
@@ -118,13 +118,13 @@ bool kasan_check_range(unsigned long addr, size_t size, bool write,
return true;
}

-bool kasan_check_invalid_free(void *addr)
+bool kasan_byte_accessible(const void *addr)

{
u8 tag = get_tag(addr);
u8 shadow_byte = READ_ONCE(*(u8 *)kasan_mem_to_shadow(kasan_reset_tag(addr)));

- return (shadow_byte == KASAN_TAG_INVALID) ||
- (tag != KASAN_TAG_KERNEL && tag != shadow_byte);
+ return (shadow_byte != KASAN_TAG_INVALID) &&
+ (tag == KASAN_TAG_KERNEL || tag == shadow_byte);
}

#define DEFINE_HWASAN_LOAD_STORE(size) \
diff --git a/mm/slab_common.c b/mm/slab_common.c
index e981c80d216c..a3bb44516623 100644
--- a/mm/slab_common.c
+++ b/mm/slab_common.c
@@ -1157,11 +1157,13 @@ size_t ksize(const void *objp)
size_t size;

/*
- * We need to check that the pointed to object is valid, and only then
- * unpoison the shadow memory below. We use __kasan_check_read(), to
- * generate a more useful report at the time ksize() is called (rather
- * than later where behaviour is undefined due to potential
- * use-after-free or double-free).
+ * We need to first check that the pointer to the object is valid, and
+ * only then unpoison the memory. The report printed from ksize() is
+ * more useful, then when it's printed later when the behaviour could
+ * be undefined due to a potential use-after-free or double-free.
+ *
+ * We use kasan_check_byte(), which is supported for hardware tag-based
+ * KASAN mode, unlike kasan_check_read/write().
*
* If the pointed to memory is invalid we return 0, to avoid users of
* ksize() writing to and potentially corrupting the memory region.
@@ -1169,7 +1171,8 @@ size_t ksize(const void *objp)
* We want to perform the check before __ksize(), to avoid potentially
* crashing in __ksize() due to accessing invalid metadata.
*/
- if (unlikely(ZERO_OR_NULL_PTR(objp)) || !__kasan_check_read(objp, 1))
+ if (unlikely(ZERO_OR_NULL_PTR(objp)) ||
+ !kasan_check_byte(objp, _RET_IP_))
return 0;

size = __ksize(objp);
--
2.30.0.284.gd98b1dd5eaa7-goog

Andrey Konovalov

unread,

Jan 13, 2021, 11:22:17 AM1/13/21

to Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

The currently existing page allocator tests rely on kmalloc fallback
with large sizes that is only present for SLUB. Add proper tests that
use alloc/free_pages().

Link: https://linux-review.googlesource.com/id/Ia173d5a1b215fe6b2548d814ef0f4433cf983570

Reviewed-by: Marco Elver <el...@google.com>
Reviewed-by: Alexander Potapenko <gli...@google.com>

Signed-off-by: Andrey Konovalov <andre...@google.com>
---

lib/test_kasan.c | 51 +++++++++++++++++++++++++++++++++++++++++++-----
1 file changed, 46 insertions(+), 5 deletions(-)

diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index 710e714dc0cb..5e3d054e5b8c 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c
@@ -147,6 +147,12 @@ static void kmalloc_node_oob_right(struct kunit *test)
kfree(ptr);
}

+/*
+ * These kmalloc_pagealloc_* tests try allocating a memory chunk that doesn't
+ * fit into a slab cache and therefore is allocated via the page allocator
+ * fallback. Since this kind of fallback is only implemented for SLUB, these
+ * tests are limited to that allocator.
+ */
static void kmalloc_pagealloc_oob_right(struct kunit *test)
{
char *ptr;
@@ -154,14 +160,11 @@ static void kmalloc_pagealloc_oob_right(struct kunit *test)

KASAN_TEST_NEEDS_CONFIG_ON(test, CONFIG_SLUB);

- /*
- * Allocate a chunk that does not fit into a SLUB cache to trigger
- * the page allocator fallback.
- */

ptr = kmalloc(size, GFP_KERNEL);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);

KUNIT_EXPECT_KASAN_FAIL(test, ptr[size + OOB_TAG_OFF] = 0);
+
kfree(ptr);
}

@@ -174,8 +177,8 @@ static void kmalloc_pagealloc_uaf(struct kunit *test)

ptr = kmalloc(size, GFP_KERNEL);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);

-
kfree(ptr);
+
KUNIT_EXPECT_KASAN_FAIL(test, ptr[0] = 0);
}

@@ -192,6 +195,42 @@ static void kmalloc_pagealloc_invalid_free(struct kunit *test)
KUNIT_EXPECT_KASAN_FAIL(test, kfree(ptr + 1));
}

+static void pagealloc_oob_right(struct kunit *test)
+{
+ char *ptr;
+ struct page *pages;
+ size_t order = 4;
+ size_t size = (1UL << (PAGE_SHIFT + order));
+
+ /*
+ * With generic KASAN page allocations have no redzones, thus
+ * out-of-bounds detection is not guaranteed.
+ * See https://bugzilla.kernel.org/show_bug.cgi?id=210503.
+ */
+ KASAN_TEST_NEEDS_CONFIG_OFF(test, CONFIG_KASAN_GENERIC);
+

+ pages = alloc_pages(GFP_KERNEL, order);
+ ptr = page_address(pages);
+ KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);
+

+ KUNIT_EXPECT_KASAN_FAIL(test, ptr[size] = 0);

+ free_pages((unsigned long)ptr, order);
+}
+

+static void pagealloc_uaf(struct kunit *test)
+{
+ char *ptr;
+ struct page *pages;
+ size_t order = 4;
+

+ pages = alloc_pages(GFP_KERNEL, order);
+ ptr = page_address(pages);
+ KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);

+ free_pages((unsigned long)ptr, order);
+

+ KUNIT_EXPECT_KASAN_FAIL(test, ptr[0] = 0);
+}
+
static void kmalloc_large_oob_right(struct kunit *test)
{
char *ptr;
@@ -903,6 +942,8 @@ static struct kunit_case kasan_kunit_test_cases[] = {
KUNIT_CASE(kmalloc_pagealloc_oob_right),
KUNIT_CASE(kmalloc_pagealloc_uaf),
KUNIT_CASE(kmalloc_pagealloc_invalid_free),
+ KUNIT_CASE(pagealloc_oob_right),
+ KUNIT_CASE(pagealloc_uaf),
KUNIT_CASE(kmalloc_large_oob_right),
KUNIT_CASE(kmalloc_oob_krealloc_more),
KUNIT_CASE(kmalloc_oob_krealloc_less),
--
2.30.0.284.gd98b1dd5eaa7-goog

Andrey Konovalov

unread,

Jan 13, 2021, 11:22:19 AM1/13/21

to Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

Add a test for kmem_cache_alloc/free_bulk to make sure there are now
false-positives when these functions are used.

Link: https://linux-review.googlesource.com/id/I2a8bf797aecf81baeac61380c567308f319e263d

Signed-off-by: Andrey Konovalov <andre...@google.com>
---

lib/test_kasan.c | 39 ++++++++++++++++++++++++++++++++++-----
1 file changed, 34 insertions(+), 5 deletions(-)

diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index 5e3d054e5b8c..d9f9a93922d5 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c
@@ -479,10 +479,11 @@ static void kmem_cache_oob(struct kunit *test)
{
char *p;
size_t size = 200;
- struct kmem_cache *cache = kmem_cache_create("test_cache",
- size, 0,
- 0, NULL);
+ struct kmem_cache *cache;
+
+ cache = kmem_cache_create("test_cache", size, 0, 0, NULL);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, cache);
+
p = kmem_cache_alloc(cache, GFP_KERNEL);
if (!p) {
kunit_err(test, "Allocation failed: %s\n", __func__);
@@ -491,11 +492,12 @@ static void kmem_cache_oob(struct kunit *test)
}

KUNIT_EXPECT_KASAN_FAIL(test, *p = p[size + OOB_TAG_OFF]);
+
kmem_cache_free(cache, p);
kmem_cache_destroy(cache);
}

-static void memcg_accounted_kmem_cache(struct kunit *test)
+static void kmem_cache_accounted(struct kunit *test)
{
int i;
char *p;
@@ -522,6 +524,32 @@ static void memcg_accounted_kmem_cache(struct kunit *test)
kmem_cache_destroy(cache);
}

+static void kmem_cache_bulk(struct kunit *test)
+{
+ struct kmem_cache *cache;
+ size_t size = 200;
+ size_t p_size = 10;
+ char *p[10];
+ bool ret;
+ int i;
+
+ cache = kmem_cache_create("test_cache", size, 0, 0, NULL);
+ KUNIT_ASSERT_NOT_ERR_OR_NULL(test, cache);
+
+ ret = kmem_cache_alloc_bulk(cache, GFP_KERNEL, p_size, (void **)&p);
+ if (!ret) {
+ kunit_err(test, "Allocation failed: %s\n", __func__);
+ kmem_cache_destroy(cache);
+ return;
+ }
+
+ for (i = 0; i < p_size; i++)
+ p[i][0] = p[i][size - 1] = 42;
+
+ kmem_cache_free_bulk(cache, p_size, (void **)&p);
+ kmem_cache_destroy(cache);
+}
+
static char global_array[10];

static void kasan_global_oob(struct kunit *test)

@@ -961,7 +989,8 @@ static struct kunit_case kasan_kunit_test_cases[] = {
KUNIT_CASE(kfree_via_page),
KUNIT_CASE(kfree_via_phys),
KUNIT_CASE(kmem_cache_oob),
- KUNIT_CASE(memcg_accounted_kmem_cache),
+ KUNIT_CASE(kmem_cache_accounted),
+ KUNIT_CASE(kmem_cache_bulk),
KUNIT_CASE(kasan_global_oob),
KUNIT_CASE(kasan_stack_oob),
KUNIT_CASE(kasan_alloca_oob_left),
--
2.30.0.284.gd98b1dd5eaa7-goog

Andrey Konovalov

unread,

Jan 13, 2021, 11:22:21 AM1/13/21

to Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

Don't run KASAN tests when it's disabled with kasan.mode=off to avoid
corrupting kernel memory.

Link: https://linux-review.googlesource.com/id/I6447af436a69a94bfc35477f6bf4e2122948355e

Signed-off-by: Andrey Konovalov <andre...@google.com>
---

lib/test_kasan.c | 3 +++
1 file changed, 3 insertions(+)

diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index d9f9a93922d5..0c8279d9907e 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c
@@ -47,6 +47,9 @@ static bool multishot;

*/
static int kasan_test_init(struct kunit *test)
{

+ if (!kasan_enabled())
+ return -1;
+
multishot = kasan_save_enable_multi_shot();
hw_set_tagging_report_once(false);
return 0;
--
2.30.0.284.gd98b1dd5eaa7-goog

unread,

Jan 13, 2021, 11:28:52 AM1/13/21

to Andrey Konovalov, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Andrew Morton, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

On Wed, 13 Jan 2021 at 17:22, Andrey Konovalov <andre...@google.com> wrote:
>
> It might not be obvious to the compiler that the expression must be
> executed between writing and reading to fail_data. In this case, the
> compiler might reorder or optimize away some of the accesses, and
> the tests will fail.
>
> Add compiler barriers around the expression in KUNIT_EXPECT_KASAN_FAIL
> and use READ/WRITE_ONCE() for accessing fail_data fields.
>
> Link: https://linux-review.googlesource.com/id/I046079f48641a1d36fe627fc8827a9249102fd50
> Signed-off-by: Andrey Konovalov <andre...@google.com>

Reviewed-by: Marco Elver <el...@google.com>

Marco Elver

unread,

Jan 13, 2021, 11:31:35 AM1/13/21

to Andrey Konovalov, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Andrew Morton, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

On Wed, 13 Jan 2021 at 17:22, Andrey Konovalov <andre...@google.com> wrote:
>

> In the kmalloc_uaf2() test, the pointers to the two allocated memory
> blocks might happen to be the same, and the test will fail. With the
> software tag-based mode, the probability of the that is 1/254, so it's
> hard to observe the failure. For the hardware tag-based mode though,
> the probablity is 1/14, which is quite noticable.
>
> Allow up to 16 attempts at generating different tags for the tag-based
> modes.
>
> Link: https://linux-review.googlesource.com/id/Ibfa458ef2804ff465d8eb07434a300bf36388d55
> Signed-off-by: Andrey Konovalov <andre...@google.com>

Reviewed-by: Marco Elver <el...@google.com>

Alexander Potapenko

unread,

Jan 13, 2021, 11:33:08 AM1/13/21

to Andrey Konovalov, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

On Wed, Jan 13, 2021 at 5:22 PM 'Andrey Konovalov' via kasan-dev
<kasa...@googlegroups.com> wrote:
>
> In the kmalloc_uaf2() test, the pointers to the two allocated memory
> blocks might happen to be the same, and the test will fail. With the
> software tag-based mode, the probability of the that is 1/254, so it's
> hard to observe the failure. For the hardware tag-based mode though,
> the probablity is 1/14, which is quite noticable.
>
> Allow up to 16 attempts at generating different tags for the tag-based
> modes.
>
> Link: https://linux-review.googlesource.com/id/Ibfa458ef2804ff465d8eb07434a300bf36388d55
> Signed-off-by: Andrey Konovalov <andre...@google.com>

Reviewed-by: Alexander Potapenko <gli...@google.com>

> --
> You received this message because you are subscribed to the Google Groups "kasan-dev" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to kasan-dev+...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/kasan-dev/1b884616c85091d6d173f7c1a8647d25424f1e7e.1610554432.git.andreyknvl%40google.com.

--
Alexander Potapenko
Software Engineer

Google Germany GmbH
Erika-Mann-Straße, 33
80636 München

Geschäftsführer: Paul Manicle, Halimah DeLaine Prado
Registergericht und -nummer: Hamburg, HRB 86891
Sitz der Gesellschaft: Hamburg

Marco Elver

unread,

Jan 13, 2021, 11:38:11 AM1/13/21

to Andrey Konovalov, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Andrew Morton, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

On Wed, 13 Jan 2021 at 17:22, Andrey Konovalov <andre...@google.com> wrote:
>

> Add a test for kmem_cache_alloc/free_bulk to make sure there are now
> false-positives when these functions are used.

s/now/no/ (but by itself doesn't necessarily demand a v3)

s/p_size/ARRAY_SIZE(p)/
?

Marco Elver

unread,

Jan 13, 2021, 11:40:00 AM1/13/21

to Andrey Konovalov, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Andrew Morton, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

On Wed, 13 Jan 2021 at 17:22, Andrey Konovalov <andre...@google.com> wrote:
>

> Don't run KASAN tests when it's disabled with kasan.mode=off to avoid
> corrupting kernel memory.
>
> Link: https://linux-review.googlesource.com/id/I6447af436a69a94bfc35477f6bf4e2122948355e
> Signed-off-by: Andrey Konovalov <andre...@google.com>
> ---
> lib/test_kasan.c | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff --git a/lib/test_kasan.c b/lib/test_kasan.c
> index d9f9a93922d5..0c8279d9907e 100644
> --- a/lib/test_kasan.c
> +++ b/lib/test_kasan.c
> @@ -47,6 +47,9 @@ static bool multishot;
> */
> static int kasan_test_init(struct kunit *test)
> {
> + if (!kasan_enabled())
> + return -1;

This should WARN_ON() or pr_err(). Otherwise it's impossible to say
why the test couldn't initialize.

Alexander Potapenko

unread,

Jan 13, 2021, 11:46:05 AM1/13/21

to Marco Elver, Andrey Konovalov, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Andrew Morton, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

On Wed, Jan 13, 2021 at 5:25 PM Marco Elver <el...@google.com> wrote:
>
> On Wed, 13 Jan 2021 at 17:21, Andrey Konovalov <andre...@google.com> wrote:
> >
> > Some KASAN tests require specific kernel configs to be enabled.
> > Instead of copy-pasting the checks for these configs add a few helper
> > macros and use them.
> >
> > Link: https://linux-review.googlesource.com/id/I237484a7fddfedf4a4aae9cc61ecbcdbe85a0a63
> > Suggested-by: Alexander Potapenko <gli...@google.com>
> > Signed-off-by: Andrey Konovalov <andre...@google.com>
>
> Nice!
>
> Reviewed-by: Marco Elver <el...@google.com>

Reviewed-by: Alexander Potapenko <gli...@google.com>

Marco Elver

unread,

Jan 13, 2021, 11:54:50 AM1/13/21

to Andrey Konovalov, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Andrew Morton, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

On Wed, 13 Jan 2021 at 17:22, Andrey Konovalov <andre...@google.com> wrote:
>

> The currently existing kasan_check_read/write() annotations are intended
> to be used for kernel modules that have KASAN compiler instrumentation
> disabled. Thus, they are only relevant for the software KASAN modes that
> rely on compiler instrumentation.
>
> However there's another use case for these annotations: ksize() checks
> that the object passed to it is indeed accessible before unpoisoning the
> whole object. This is currently done via __kasan_check_read(), which is
> compiled away for the hardware tag-based mode that doesn't rely on
> compiler instrumentation. This leads to KASAN missing detecting some
> memory corruptions.
>
> Provide another annotation called kasan_check_byte() that is available
> for all KASAN modes. As the implementation rename and reuse
> kasan_check_invalid_free(). Use this new annotation in ksize().
>
> Also add a new ksize_uaf() test that checks that a use-after-free is
> detected via ksize() itself, and via plain accesses that happen later.
>
> Link: https://linux-review.googlesource.com/id/Iaabf771881d0f9ce1b969f2a62938e99d3308ec5
> Signed-off-by: Andrey Konovalov <andre...@google.com>

Added some additional questions below, as I'm not sure about these
points yet. Otherwise looks good.

Why was this not added to kasan-checks.h? I'd assume including all of
kasan.h is also undesirable for tag-based modes if we just want to do
a kasan_check_byte().

Was requiring 'ip' intentional? Unlike the other
kasan_check-functions, this takes an explicit 'ip'. In the case of
ksize() usage, this is an advantage, so I'd probably keep it, but the
rationale to introducing 'ip' vs. before wasn't mentioned.

Like the other __kasan_check*, should this have been EXPORT_SYMBOL()?
Or was it intentional to not export as it's currently only used by
non-modules?

Andrey Konovalov

unread,

Jan 14, 2021, 10:32:44 AM1/14/21

to Marco Elver, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Andrew Morton, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

On Wed, Jan 13, 2021 at 5:39 PM Marco Elver <el...@google.com> wrote:
>
> On Wed, 13 Jan 2021 at 17:22, Andrey Konovalov <andre...@google.com> wrote:
> >
> > Don't run KASAN tests when it's disabled with kasan.mode=off to avoid
> > corrupting kernel memory.
> >
> > Link: https://linux-review.googlesource.com/id/I6447af436a69a94bfc35477f6bf4e2122948355e
> > Signed-off-by: Andrey Konovalov <andre...@google.com>
> > ---
> > lib/test_kasan.c | 3 +++
> > 1 file changed, 3 insertions(+)
> >
> > diff --git a/lib/test_kasan.c b/lib/test_kasan.c
> > index d9f9a93922d5..0c8279d9907e 100644
> > --- a/lib/test_kasan.c
> > +++ b/lib/test_kasan.c
> > @@ -47,6 +47,9 @@ static bool multishot;
> > */
> > static int kasan_test_init(struct kunit *test)
> > {
> > + if (!kasan_enabled())
> > + return -1;
>
> This should WARN_ON() or pr_err(). Otherwise it's impossible to say
> why the test couldn't initialize.

Will do in v3, thanks!

Andrey Konovalov

unread,

Jan 14, 2021, 10:33:06 AM1/14/21

to Marco Elver, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Andrew Morton, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

Will fix all in v3, thanks!

Andrey Konovalov

unread,

Jan 14, 2021, 12:59:09 PM1/14/21

to Marco Elver, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Andrew Morton, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

On Wed, Jan 13, 2021 at 5:54 PM Marco Elver <el...@google.com> wrote:
>
> > +bool __kasan_check_byte(const void *addr, unsigned long ip);
> > +static __always_inline bool kasan_check_byte(const void *addr, unsigned long ip)
> > +{
> > + if (kasan_enabled())
> > + return __kasan_check_byte(addr, ip);
> > + return true;
> > +}
>
> Why was this not added to kasan-checks.h? I'd assume including all of
> kasan.h is also undesirable for tag-based modes if we just want to do
> a kasan_check_byte().
>
> Was requiring 'ip' intentional? Unlike the other
> kasan_check-functions, this takes an explicit 'ip'. In the case of
> ksize() usage, this is an advantage, so I'd probably keep it, but the
> rationale to introducing 'ip' vs. before wasn't mentioned.

Yes, to avoid having a ksize() frame in the report. However, I'll move
_RET_IP_ inside of kasan_check_byte() as it's inline.

> > +bool __kasan_check_byte(const void *address, unsigned long ip)
> > +{
> > + if (!kasan_byte_accessible(address)) {
> > + kasan_report((unsigned long)address, 1, false, ip);
> > + return false;
> > + }
> > + return true;
> > +}
>
> Like the other __kasan_check*, should this have been EXPORT_SYMBOL()?
> Or was it intentional to not export as it's currently only used by
> non-modules?

We can add EXPORT_SYMBOL as soon as there's a need for it.

Thanks!

Andrey Konovalov

unread,

Jan 14, 2021, 1:01:52 PM1/14/21

to Marco Elver, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Andrew Morton, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

On Wed, Jan 13, 2021 at 5:54 PM Marco Elver <el...@google.com> wrote:
>

> > +/*
> > + * Unlike kasan_check_read/write(), kasan_check_byte() is performed even for
> > + * the hardware tag-based mode that doesn't rely on compiler instrumentation.
> > + */
> > +bool __kasan_check_byte(const void *addr, unsigned long ip);
> > +static __always_inline bool kasan_check_byte(const void *addr, unsigned long ip)
> > +{
> > + if (kasan_enabled())
> > + return __kasan_check_byte(addr, ip);
> > + return true;
> > +}
>
> Why was this not added to kasan-checks.h? I'd assume including all of
> kasan.h is also undesirable for tag-based modes if we just want to do
> a kasan_check_byte().

It requires kasan_enabled() definition. I can move both to
kasan-checks.h if you prefer. However, the only place where
kasan_check_byte() is currently used includes kasan.h anyway.

Andrey Konovalov

unread,

Jan 14, 2021, 2:36:36 PM1/14/21

to Andrew Morton, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

This patchset adds support for running KASAN-KUnit tests with the
hardware tag-based mode and also contains a few fixes.

Changes v2->v3:
- Don't call kmalloc(0) when generating random size.
- Use ARRAY_SIZE() in kmem_cache_bulk_alloc() test.
- Print error message when tests are being ran with kasan.mode=off.
- Move _RET_IP_ to inline wrappers for kasan annotations.

Andrey Konovalov (15):
kasan: prefix global functions with kasan_
kasan: clarify HW_TAGS impact on TBI
kasan: clean up comments in tests
kasan: add macros to simplify checking test constraints
kasan: add match-all tag tests
kasan, arm64: allow using KUnit tests with HW_TAGS mode
kasan: rename CONFIG_TEST_KASAN_MODULE
kasan: add compiler barriers to KUNIT_EXPECT_KASAN_FAIL
kasan: adapt kmalloc_uaf2 test to HW_TAGS mode
kasan: fix memory corruption in kasan_bitops_tags test
kasan: move _RET_IP_ to inline wrappers
kasan: fix bug detection via ksize for HW_TAGS mode
kasan: add proper page allocator tests
kasan: add a test for kmem_cache_alloc/free_bulk
kasan: don't run tests when KASAN is not enabled

Documentation/dev-tools/kasan.rst | 24 +-

arch/arm64/include/asm/memory.h | 1 +
arch/arm64/include/asm/mte-kasan.h | 12 +

Andrey Konovalov

unread,

Jan 14, 2021, 2:36:38 PM1/14/21

to Andrew Morton, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

There's a number of internal KASAN functions that are used across multiple
source code files and therefore aren't marked as static inline. To avoid
littering the kernel function names list with generic function names,
prefix all such KASAN functions with kasan_.

As a part of this change:

- Rename internal (un)poison_range() to kasan_(un)poison() (no _range)
to avoid name collision with a public kasan_unpoison_range().

- Rename check_memory_region() to kasan_check_range(), as it's a more
fitting name.

Link: https://linux-review.googlesource.com/id/I719cc93483d4ba288a634dba80ee6b7f2809cd26
Suggested-by: Marco Elver <el...@google.com>

Reviewed-by: Marco Elver <el...@google.com>
Reviewed-by: Alexander Potapenko <gli...@google.com>

Signed-off-by: Andrey Konovalov <andre...@google.com>
---

mm/kasan/common.c | 47 +++++++++++++++++++-------------------
mm/kasan/generic.c | 36 ++++++++++++++---------------
mm/kasan/kasan.h | 48 +++++++++++++++++++--------------------
mm/kasan/quarantine.c | 22 +++++++++---------
mm/kasan/report.c | 13 ++++++-----
mm/kasan/report_generic.c | 8 +++----
mm/kasan/report_hw_tags.c | 8 +++----
mm/kasan/report_sw_tags.c | 8 +++----
mm/kasan/shadow.c | 26 ++++++++++-----------
mm/kasan/sw_tags.c | 16 ++++++-------
tools/objtool/check.c | 2 +-
11 files changed, 117 insertions(+), 117 deletions(-)

diff --git a/mm/kasan/common.c b/mm/kasan/common.c
index b25167664ead..eedc3e0fe365 100644
--- a/mm/kasan/common.c
+++ b/mm/kasan/common.c
@@ -60,7 +60,7 @@ void kasan_disable_current(void)

void __kasan_unpoison_range(const void *address, size_t size)
{
- unpoison_range(address, size);
+ kasan_unpoison(address, size);
}

#if CONFIG_KASAN_STACK
@@ -69,7 +69,7 @@ void kasan_unpoison_task_stack(struct task_struct *task)
{
void *base = task_stack_page(task);

- unpoison_range(base, THREAD_SIZE);
+ kasan_unpoison(base, THREAD_SIZE);
}

/* Unpoison the stack for the current task beyond a watermark sp value. */
@@ -82,7 +82,7 @@ asmlinkage void kasan_unpoison_task_stack_below(const void *watermark)
*/
void *base = (void *)((unsigned long)watermark & ~(THREAD_SIZE - 1));

- unpoison_range(base, watermark - base);
+ kasan_unpoison(base, watermark - base);
}
#endif /* CONFIG_KASAN_STACK */

@@ -105,18 +105,17 @@ void __kasan_alloc_pages(struct page *page, unsigned int order)
if (unlikely(PageHighMem(page)))
return;

- tag = random_tag();
+ tag = kasan_random_tag();
for (i = 0; i < (1 << order); i++)
page_kasan_tag_set(page + i, tag);
- unpoison_range(page_address(page), PAGE_SIZE << order);
+ kasan_unpoison(page_address(page), PAGE_SIZE << order);
}

void __kasan_free_pages(struct page *page, unsigned int order)
{
if (likely(!PageHighMem(page)))
- poison_range(page_address(page),
- PAGE_SIZE << order,
- KASAN_FREE_PAGE);
+ kasan_poison(page_address(page), PAGE_SIZE << order,
+ KASAN_FREE_PAGE);
}

/*
@@ -246,18 +245,18 @@ void __kasan_poison_slab(struct page *page)

for (i = 0; i < compound_nr(page); i++)
page_kasan_tag_reset(page + i);
- poison_range(page_address(page), page_size(page),
+ kasan_poison(page_address(page), page_size(page),
KASAN_KMALLOC_REDZONE);
}

void __kasan_unpoison_object_data(struct kmem_cache *cache, void *object)
{
- unpoison_range(object, cache->object_size);
+ kasan_unpoison(object, cache->object_size);
}

void __kasan_poison_object_data(struct kmem_cache *cache, void *object)
{
- poison_range(object, cache->object_size, KASAN_KMALLOC_REDZONE);
+ kasan_poison(object, cache->object_size, KASAN_KMALLOC_REDZONE);
}

/*
@@ -294,7 +293,7 @@ static u8 assign_tag(struct kmem_cache *cache, const void *object,
* set, assign a tag when the object is being allocated (init == false).
*/
if (!cache->ctor && !(cache->flags & SLAB_TYPESAFE_BY_RCU))
- return init ? KASAN_TAG_KERNEL : random_tag();
+ return init ? KASAN_TAG_KERNEL : kasan_random_tag();

/* For caches that either have a constructor or SLAB_TYPESAFE_BY_RCU: */
#ifdef CONFIG_SLAB
@@ -305,7 +304,7 @@ static u8 assign_tag(struct kmem_cache *cache, const void *object,
* For SLUB assign a random tag during slab creation, otherwise reuse
* the already assigned tag.
*/
- return init ? random_tag() : get_tag(object);
+ return init ? kasan_random_tag() : get_tag(object);
#endif
}

@@ -346,12 +345,12 @@ static bool ____kasan_slab_free(struct kmem_cache *cache, void *object,

if (unlikely(cache->flags & SLAB_TYPESAFE_BY_RCU))
return false;

- if (check_invalid_free(tagged_object)) {
+ if (kasan_check_invalid_free(tagged_object)) {
kasan_report_invalid_free(tagged_object, ip);
return true;
}

- poison_range(object, cache->object_size, KASAN_KMALLOC_FREE);
+ kasan_poison(object, cache->object_size, KASAN_KMALLOC_FREE);

if (!kasan_stack_collection_enabled())
return false;
@@ -361,7 +360,7 @@ static bool ____kasan_slab_free(struct kmem_cache *cache, void *object,

kasan_set_free_info(cache, object, tag);

- return quarantine_put(cache, object);
+ return kasan_quarantine_put(cache, object);
}

bool __kasan_slab_free(struct kmem_cache *cache, void *object, unsigned long ip)
@@ -386,7 +385,7 @@ void __kasan_slab_free_mempool(void *ptr, unsigned long ip)
kasan_report_invalid_free(ptr, ip);
return;
}
- poison_range(ptr, page_size(page), KASAN_FREE_PAGE);
+ kasan_poison(ptr, page_size(page), KASAN_FREE_PAGE);
} else {
____kasan_slab_free(page->slab_cache, ptr, ip, false);
}
@@ -409,7 +408,7 @@ static void *____kasan_kmalloc(struct kmem_cache *cache, const void *object,
u8 tag;

if (gfpflags_allow_blocking(flags))
- quarantine_reduce();
+ kasan_quarantine_reduce();

if (unlikely(object == NULL))
return NULL;
@@ -421,9 +420,9 @@ static void *____kasan_kmalloc(struct kmem_cache *cache, const void *object,
tag = assign_tag(cache, object, false, keep_tag);

/* Tag is ignored in set_tag without CONFIG_KASAN_SW/HW_TAGS */
- unpoison_range(set_tag(object, tag), size);
- poison_range((void *)redzone_start, redzone_end - redzone_start,
- KASAN_KMALLOC_REDZONE);
+ kasan_unpoison(set_tag(object, tag), size);
+ kasan_poison((void *)redzone_start, redzone_end - redzone_start,
+ KASAN_KMALLOC_REDZONE);

if (kasan_stack_collection_enabled())
set_alloc_info(cache, (void *)object, flags);
@@ -452,7 +451,7 @@ void * __must_check __kasan_kmalloc_large(const void *ptr, size_t size,
unsigned long redzone_end;

if (gfpflags_allow_blocking(flags))
- quarantine_reduce();
+ kasan_quarantine_reduce();

if (unlikely(ptr == NULL))
return NULL;
@@ -462,8 +461,8 @@ void * __must_check __kasan_kmalloc_large(const void *ptr, size_t size,
KASAN_GRANULE_SIZE);
redzone_end = (unsigned long)ptr + page_size(page);

- unpoison_range(ptr, size);
- poison_range((void *)redzone_start, redzone_end - redzone_start,
+ kasan_unpoison(ptr, size);
+ kasan_poison((void *)redzone_start, redzone_end - redzone_start,
KASAN_PAGE_REDZONE);

return (void *)ptr;
diff --git a/mm/kasan/generic.c b/mm/kasan/generic.c
index 5106b84b07d4..acab8862dc67 100644
--- a/mm/kasan/generic.c
+++ b/mm/kasan/generic.c
@@ -158,7 +158,7 @@ static __always_inline bool memory_is_poisoned(unsigned long addr, size_t size)
return memory_is_poisoned_n(addr, size);
}

-static __always_inline bool check_memory_region_inline(unsigned long addr,
+static __always_inline bool check_region_inline(unsigned long addr,
size_t size, bool write,
unsigned long ret_ip)
{
@@ -179,13 +179,13 @@ static __always_inline bool check_memory_region_inline(unsigned long addr,
return !kasan_report(addr, size, write, ret_ip);
}

-bool check_memory_region(unsigned long addr, size_t size, bool write,
- unsigned long ret_ip)
+bool kasan_check_range(unsigned long addr, size_t size, bool write,
+ unsigned long ret_ip)
{
- return check_memory_region_inline(addr, size, write, ret_ip);
+ return check_region_inline(addr, size, write, ret_ip);
}

-bool check_invalid_free(void *addr)
+bool kasan_check_invalid_free(void *addr)

{
s8 shadow_byte = READ_ONCE(*(s8 *)kasan_mem_to_shadow(addr));

@@ -194,22 +194,22 @@ bool check_invalid_free(void *addr)

void kasan_cache_shrink(struct kmem_cache *cache)
{
- quarantine_remove_cache(cache);
+ kasan_quarantine_remove_cache(cache);
}

void kasan_cache_shutdown(struct kmem_cache *cache)
{
if (!__kmem_cache_empty(cache))
- quarantine_remove_cache(cache);
+ kasan_quarantine_remove_cache(cache);
}

static void register_global(struct kasan_global *global)
{
size_t aligned_size = round_up(global->size, KASAN_GRANULE_SIZE);

- unpoison_range(global->beg, global->size);
+ kasan_unpoison(global->beg, global->size);

- poison_range(global->beg + aligned_size,
+ kasan_poison(global->beg + aligned_size,
global->size_with_redzone - aligned_size,
KASAN_GLOBAL_REDZONE);
}
@@ -231,7 +231,7 @@ EXPORT_SYMBOL(__asan_unregister_globals);
#define DEFINE_ASAN_LOAD_STORE(size) \
void __asan_load##size(unsigned long addr) \
{ \
- check_memory_region_inline(addr, size, false, _RET_IP_);\
+ check_region_inline(addr, size, false, _RET_IP_); \
} \
EXPORT_SYMBOL(__asan_load##size); \
__alias(__asan_load##size) \
@@ -239,7 +239,7 @@ EXPORT_SYMBOL(__asan_unregister_globals);
EXPORT_SYMBOL(__asan_load##size##_noabort); \
void __asan_store##size(unsigned long addr) \
{ \
- check_memory_region_inline(addr, size, true, _RET_IP_); \
+ check_region_inline(addr, size, true, _RET_IP_); \
} \
EXPORT_SYMBOL(__asan_store##size); \
__alias(__asan_store##size) \
@@ -254,7 +254,7 @@ DEFINE_ASAN_LOAD_STORE(16);

void __asan_loadN(unsigned long addr, size_t size)
{
- check_memory_region(addr, size, false, _RET_IP_);
+ kasan_check_range(addr, size, false, _RET_IP_);
}
EXPORT_SYMBOL(__asan_loadN);

@@ -264,7 +264,7 @@ EXPORT_SYMBOL(__asan_loadN_noabort);

void __asan_storeN(unsigned long addr, size_t size)
{
- check_memory_region(addr, size, true, _RET_IP_);
+ kasan_check_range(addr, size, true, _RET_IP_);
}
EXPORT_SYMBOL(__asan_storeN);

@@ -290,11 +290,11 @@ void __asan_alloca_poison(unsigned long addr, size_t size)

WARN_ON(!IS_ALIGNED(addr, KASAN_ALLOCA_REDZONE_SIZE));

- unpoison_range((const void *)(addr + rounded_down_size),
- size - rounded_down_size);
- poison_range(left_redzone, KASAN_ALLOCA_REDZONE_SIZE,
+ kasan_unpoison((const void *)(addr + rounded_down_size),
+ size - rounded_down_size);
+ kasan_poison(left_redzone, KASAN_ALLOCA_REDZONE_SIZE,
KASAN_ALLOCA_LEFT);
- poison_range(right_redzone, padding_size + KASAN_ALLOCA_REDZONE_SIZE,
+ kasan_poison(right_redzone, padding_size + KASAN_ALLOCA_REDZONE_SIZE,
KASAN_ALLOCA_RIGHT);
}
EXPORT_SYMBOL(__asan_alloca_poison);
@@ -305,7 +305,7 @@ void __asan_allocas_unpoison(const void *stack_top, const void *stack_bottom)
if (unlikely(!stack_top || stack_top > stack_bottom))
return;

- unpoison_range(stack_top, stack_bottom - stack_top);
+ kasan_unpoison(stack_top, stack_bottom - stack_top);
}
EXPORT_SYMBOL(__asan_allocas_unpoison);

diff --git a/mm/kasan/kasan.h b/mm/kasan/kasan.h
index cc4d9e1d49b1..3b38baddec47 100644
--- a/mm/kasan/kasan.h
+++ b/mm/kasan/kasan.h
@@ -195,14 +195,14 @@ static inline bool addr_has_metadata(const void *addr)
}

/**
- * check_memory_region - Check memory region, and report if invalid access.
+ * kasan_check_range - Check memory region, and report if invalid access.
* @addr: the accessed address
* @size: the accessed size
* @write: true if access is a write access
* @ret_ip: return address
* @return: true if access was valid, false if invalid
*/
-bool check_memory_region(unsigned long addr, size_t size, bool write,
+bool kasan_check_range(unsigned long addr, size_t size, bool write,
unsigned long ret_ip);

#else /* CONFIG_KASAN_GENERIC || CONFIG_KASAN_SW_TAGS */
@@ -215,19 +215,19 @@ static inline bool addr_has_metadata(const void *addr)
#endif /* CONFIG_KASAN_GENERIC || CONFIG_KASAN_SW_TAGS */

#if defined(CONFIG_KASAN_SW_TAGS) || defined(CONFIG_KASAN_HW_TAGS)
-void print_tags(u8 addr_tag, const void *addr);
+void kasan_print_tags(u8 addr_tag, const void *addr);
#else
-static inline void print_tags(u8 addr_tag, const void *addr) { }
+static inline void kasan_print_tags(u8 addr_tag, const void *addr) { }
#endif

-void *find_first_bad_addr(void *addr, size_t size);
-const char *get_bug_type(struct kasan_access_info *info);
-void metadata_fetch_row(char *buffer, void *row);
+void *kasan_find_first_bad_addr(void *addr, size_t size);
+const char *kasan_get_bug_type(struct kasan_access_info *info);
+void kasan_metadata_fetch_row(char *buffer, void *row);

#if defined(CONFIG_KASAN_GENERIC) && CONFIG_KASAN_STACK
-void print_address_stack_frame(const void *addr);
+void kasan_print_address_stack_frame(const void *addr);
#else
-static inline void print_address_stack_frame(const void *addr) { }
+static inline void kasan_print_address_stack_frame(const void *addr) { }
#endif

bool kasan_report(unsigned long addr, size_t size,
@@ -244,13 +244,13 @@ struct kasan_track *kasan_get_free_track(struct kmem_cache *cache,

#if defined(CONFIG_KASAN_GENERIC) && \
(defined(CONFIG_SLAB) || defined(CONFIG_SLUB))
-bool quarantine_put(struct kmem_cache *cache, void *object);
-void quarantine_reduce(void);
-void quarantine_remove_cache(struct kmem_cache *cache);
+bool kasan_quarantine_put(struct kmem_cache *cache, void *object);
+void kasan_quarantine_reduce(void);
+void kasan_quarantine_remove_cache(struct kmem_cache *cache);
#else
-static inline bool quarantine_put(struct kmem_cache *cache, void *object) { return false; }
-static inline void quarantine_reduce(void) { }
-static inline void quarantine_remove_cache(struct kmem_cache *cache) { }
+static inline bool kasan_quarantine_put(struct kmem_cache *cache, void *object) { return false; }
+static inline void kasan_quarantine_reduce(void) { }
+static inline void kasan_quarantine_remove_cache(struct kmem_cache *cache) { }
#endif

#ifndef arch_kasan_set_tag
@@ -293,28 +293,28 @@ static inline const void *arch_kasan_set_tag(const void *addr, u8 tag)

#endif /* CONFIG_KASAN_HW_TAGS */

#ifdef CONFIG_KASAN_SW_TAGS

-u8 random_tag(void);
+u8 kasan_random_tag(void);
#elif defined(CONFIG_KASAN_HW_TAGS)
-static inline u8 random_tag(void) { return hw_get_random_tag(); }
+static inline u8 kasan_random_tag(void) { return hw_get_random_tag(); }
#else
-static inline u8 random_tag(void) { return 0; }
+static inline u8 kasan_random_tag(void) { return 0; }
#endif

#ifdef CONFIG_KASAN_HW_TAGS

-static inline void poison_range(const void *address, size_t size, u8 value)
+static inline void kasan_poison(const void *address, size_t size, u8 value)
{
hw_set_mem_tag_range(kasan_reset_tag(address),
round_up(size, KASAN_GRANULE_SIZE), value);
}

-static inline void unpoison_range(const void *address, size_t size)
+static inline void kasan_unpoison(const void *address, size_t size)
{
hw_set_mem_tag_range(kasan_reset_tag(address),
round_up(size, KASAN_GRANULE_SIZE), get_tag(address));
}

-static inline bool check_invalid_free(void *addr)
+static inline bool kasan_check_invalid_free(void *addr)
{
u8 ptr_tag = get_tag(addr);
u8 mem_tag = hw_get_mem_tag(addr);
@@ -325,9 +325,9 @@ static inline bool check_invalid_free(void *addr)

#else /* CONFIG_KASAN_HW_TAGS */

-void poison_range(const void *address, size_t size, u8 value);
-void unpoison_range(const void *address, size_t size);
-bool check_invalid_free(void *addr);
+void kasan_poison(const void *address, size_t size, u8 value);
+void kasan_unpoison(const void *address, size_t size);
+bool kasan_check_invalid_free(void *addr);

#endif /* CONFIG_KASAN_HW_TAGS */

diff --git a/mm/kasan/quarantine.c b/mm/kasan/quarantine.c
index 55783125a767..728fb24c5683 100644
--- a/mm/kasan/quarantine.c
+++ b/mm/kasan/quarantine.c
@@ -168,7 +168,7 @@ static void qlist_free_all(struct qlist_head *q, struct kmem_cache *cache)
qlist_init(q);
}

-bool quarantine_put(struct kmem_cache *cache, void *object)
+bool kasan_quarantine_put(struct kmem_cache *cache, void *object)
{
unsigned long flags;
struct qlist_head *q;
@@ -184,11 +184,11 @@ bool quarantine_put(struct kmem_cache *cache, void *object)

/*
* Note: irq must be disabled until after we move the batch to the
- * global quarantine. Otherwise quarantine_remove_cache() can miss
- * some objects belonging to the cache if they are in our local temp
- * list. quarantine_remove_cache() executes on_each_cpu() at the
- * beginning which ensures that it either sees the objects in per-cpu
- * lists or in the global quarantine.
+ * global quarantine. Otherwise kasan_quarantine_remove_cache() can
+ * miss some objects belonging to the cache if they are in our local
+ * temp list. kasan_quarantine_remove_cache() executes on_each_cpu()
+ * at the beginning which ensures that it either sees the objects in
+ * per-cpu lists or in the global quarantine.
*/
local_irq_save(flags);

@@ -222,7 +222,7 @@ bool quarantine_put(struct kmem_cache *cache, void *object)
return true;
}

-void quarantine_reduce(void)
+void kasan_quarantine_reduce(void)
{
size_t total_size, new_quarantine_size, percpu_quarantines;
unsigned long flags;
@@ -234,7 +234,7 @@ void quarantine_reduce(void)
return;

/*
- * srcu critical section ensures that quarantine_remove_cache()
+ * srcu critical section ensures that kasan_quarantine_remove_cache()
* will not miss objects belonging to the cache while they are in our
* local to_free list. srcu is chosen because (1) it gives us private
* grace period domain that does not interfere with anything else,
@@ -309,15 +309,15 @@ static void per_cpu_remove_cache(void *arg)
}

/* Free all quarantined objects belonging to cache. */
-void quarantine_remove_cache(struct kmem_cache *cache)
+void kasan_quarantine_remove_cache(struct kmem_cache *cache)
{
unsigned long flags, i;
struct qlist_head to_free = QLIST_INIT;

/*
* Must be careful to not miss any objects that are being moved from
- * per-cpu list to the global quarantine in quarantine_put(),
- * nor objects being freed in quarantine_reduce(). on_each_cpu()
+ * per-cpu list to the global quarantine in kasan_quarantine_put(),
+ * nor objects being freed in kasan_quarantine_reduce(). on_each_cpu()
* achieves the first goal, while synchronize_srcu() achieves the
* second.
*/
diff --git a/mm/kasan/report.c b/mm/kasan/report.c
index c0fb21797550..e93d7973792e 100644
--- a/mm/kasan/report.c
+++ b/mm/kasan/report.c
@@ -61,7 +61,7 @@ __setup("kasan_multi_shot", kasan_set_multi_shot);
static void print_error_description(struct kasan_access_info *info)
{
pr_err("BUG: KASAN: %s in %pS\n",
- get_bug_type(info), (void *)info->ip);
+ kasan_get_bug_type(info), (void *)info->ip);
if (info->access_size)
pr_err("%s of size %zu at addr %px by task %s/%d\n",
info->is_write ? "Write" : "Read", info->access_size,
@@ -247,7 +247,7 @@ static void print_address_description(void *addr, u8 tag)
dump_page(page, "kasan: bad access detected");
}

- print_address_stack_frame(addr);
+ kasan_print_address_stack_frame(addr);
}

static bool meta_row_is_guilty(const void *row, const void *addr)
@@ -293,7 +293,7 @@ static void print_memory_metadata(const void *addr)
* function, because generic functions may try to
* access kasan mapping for the passed address.
*/
- metadata_fetch_row(&metadata[0], row);
+ kasan_metadata_fetch_row(&metadata[0], row);

print_hex_dump(KERN_ERR, buffer,
DUMP_PREFIX_NONE, META_BYTES_PER_ROW, 1,
@@ -350,7 +350,7 @@ void kasan_report_invalid_free(void *object, unsigned long ip)

start_report(&flags);
pr_err("BUG: KASAN: double-free or invalid-free in %pS\n", (void *)ip);
- print_tags(tag, object);
+ kasan_print_tags(tag, object);
pr_err("\n");
print_address_description(object, tag);
pr_err("\n");
@@ -378,7 +378,8 @@ static void __kasan_report(unsigned long addr, size_t size, bool is_write,

info.access_addr = tagged_addr;
if (addr_has_metadata(untagged_addr))
- info.first_bad_addr = find_first_bad_addr(tagged_addr, size);
+ info.first_bad_addr =
+ kasan_find_first_bad_addr(tagged_addr, size);
else
info.first_bad_addr = untagged_addr;
info.access_size = size;
@@ -389,7 +390,7 @@ static void __kasan_report(unsigned long addr, size_t size, bool is_write,

print_error_description(&info);
if (addr_has_metadata(untagged_addr))
- print_tags(get_tag(tagged_addr), info.first_bad_addr);
+ kasan_print_tags(get_tag(tagged_addr), info.first_bad_addr);
pr_err("\n");

if (addr_has_metadata(untagged_addr)) {
diff --git a/mm/kasan/report_generic.c b/mm/kasan/report_generic.c
index 8a9c889872da..41f374585144 100644
--- a/mm/kasan/report_generic.c
+++ b/mm/kasan/report_generic.c
@@ -30,7 +30,7 @@
#include "kasan.h"
#include "../slab.h"

-void *find_first_bad_addr(void *addr, size_t size)
+void *kasan_find_first_bad_addr(void *addr, size_t size)
{
void *p = addr;

@@ -105,7 +105,7 @@ static const char *get_wild_bug_type(struct kasan_access_info *info)
return bug_type;
}

-const char *get_bug_type(struct kasan_access_info *info)
+const char *kasan_get_bug_type(struct kasan_access_info *info)
{
/*
* If access_size is a negative number, then it has reason to be
@@ -123,7 +123,7 @@ const char *get_bug_type(struct kasan_access_info *info)
return get_wild_bug_type(info);
}

-void metadata_fetch_row(char *buffer, void *row)
+void kasan_metadata_fetch_row(char *buffer, void *row)
{
memcpy(buffer, kasan_mem_to_shadow(row), META_BYTES_PER_ROW);
}
@@ -263,7 +263,7 @@ static bool __must_check get_address_stack_frame_info(const void *addr,
return true;
}

-void print_address_stack_frame(const void *addr)
+void kasan_print_address_stack_frame(const void *addr)
{
unsigned long offset;
const char *frame_descr;
diff --git a/mm/kasan/report_hw_tags.c b/mm/kasan/report_hw_tags.c
index 57114f0e14d1..42b2168755d6 100644
--- a/mm/kasan/report_hw_tags.c
+++ b/mm/kasan/report_hw_tags.c
@@ -15,17 +15,17 @@

#include "kasan.h"

-const char *get_bug_type(struct kasan_access_info *info)
+const char *kasan_get_bug_type(struct kasan_access_info *info)
{
return "invalid-access";
}

-void *find_first_bad_addr(void *addr, size_t size)
+void *kasan_find_first_bad_addr(void *addr, size_t size)
{
return kasan_reset_tag(addr);
}

-void metadata_fetch_row(char *buffer, void *row)
+void kasan_metadata_fetch_row(char *buffer, void *row)
{
int i;

@@ -33,7 +33,7 @@ void metadata_fetch_row(char *buffer, void *row)
buffer[i] = hw_get_mem_tag(row + i * KASAN_GRANULE_SIZE);
}

-void print_tags(u8 addr_tag, const void *addr)
+void kasan_print_tags(u8 addr_tag, const void *addr)
{
u8 memory_tag = hw_get_mem_tag((void *)addr);

diff --git a/mm/kasan/report_sw_tags.c b/mm/kasan/report_sw_tags.c
index 1b026793ad57..3d20d3451d9e 100644
--- a/mm/kasan/report_sw_tags.c
+++ b/mm/kasan/report_sw_tags.c
@@ -29,7 +29,7 @@
#include "kasan.h"
#include "../slab.h"

-const char *get_bug_type(struct kasan_access_info *info)
+const char *kasan_get_bug_type(struct kasan_access_info *info)
{
#ifdef CONFIG_KASAN_SW_TAGS_IDENTIFY
struct kasan_alloc_meta *alloc_meta;
@@ -72,7 +72,7 @@ const char *get_bug_type(struct kasan_access_info *info)
return "invalid-access";
}

-void *find_first_bad_addr(void *addr, size_t size)
+void *kasan_find_first_bad_addr(void *addr, size_t size)
{
u8 tag = get_tag(addr);
void *p = kasan_reset_tag(addr);
@@ -83,12 +83,12 @@ void *find_first_bad_addr(void *addr, size_t size)
return p;
}

-void metadata_fetch_row(char *buffer, void *row)
+void kasan_metadata_fetch_row(char *buffer, void *row)
{
memcpy(buffer, kasan_mem_to_shadow(row), META_BYTES_PER_ROW);
}

-void print_tags(u8 addr_tag, const void *addr)
+void kasan_print_tags(u8 addr_tag, const void *addr)
{
u8 *shadow = (u8 *)kasan_mem_to_shadow(addr);

diff --git a/mm/kasan/shadow.c b/mm/kasan/shadow.c
index 7c2c08c55f32..38958eb0d653 100644
--- a/mm/kasan/shadow.c
+++ b/mm/kasan/shadow.c
@@ -27,20 +27,20 @@

bool __kasan_check_read(const volatile void *p, unsigned int size)
{
- return check_memory_region((unsigned long)p, size, false, _RET_IP_);
+ return kasan_check_range((unsigned long)p, size, false, _RET_IP_);
}
EXPORT_SYMBOL(__kasan_check_read);

bool __kasan_check_write(const volatile void *p, unsigned int size)
{
- return check_memory_region((unsigned long)p, size, true, _RET_IP_);
+ return kasan_check_range((unsigned long)p, size, true, _RET_IP_);
}
EXPORT_SYMBOL(__kasan_check_write);

#undef memset
void *memset(void *addr, int c, size_t len)
{
- if (!check_memory_region((unsigned long)addr, len, true, _RET_IP_))
+ if (!kasan_check_range((unsigned long)addr, len, true, _RET_IP_))
return NULL;

return __memset(addr, c, len);
@@ -50,8 +50,8 @@ void *memset(void *addr, int c, size_t len)
#undef memmove
void *memmove(void *dest, const void *src, size_t len)
{
- if (!check_memory_region((unsigned long)src, len, false, _RET_IP_) ||
- !check_memory_region((unsigned long)dest, len, true, _RET_IP_))
+ if (!kasan_check_range((unsigned long)src, len, false, _RET_IP_) ||
+ !kasan_check_range((unsigned long)dest, len, true, _RET_IP_))
return NULL;

return __memmove(dest, src, len);
@@ -61,8 +61,8 @@ void *memmove(void *dest, const void *src, size_t len)
#undef memcpy
void *memcpy(void *dest, const void *src, size_t len)
{
- if (!check_memory_region((unsigned long)src, len, false, _RET_IP_) ||
- !check_memory_region((unsigned long)dest, len, true, _RET_IP_))
+ if (!kasan_check_range((unsigned long)src, len, false, _RET_IP_) ||
+ !kasan_check_range((unsigned long)dest, len, true, _RET_IP_))
return NULL;

return __memcpy(dest, src, len);
@@ -72,7 +72,7 @@ void *memcpy(void *dest, const void *src, size_t len)
* Poisons the shadow memory for 'size' bytes starting from 'addr'.
* Memory addresses should be aligned to KASAN_GRANULE_SIZE.
*/
-void poison_range(const void *address, size_t size, u8 value)
+void kasan_poison(const void *address, size_t size, u8 value)
{
void *shadow_start, *shadow_end;

@@ -90,7 +90,7 @@ void poison_range(const void *address, size_t size, u8 value)
__memset(shadow_start, value, shadow_end - shadow_start);
}

-void unpoison_range(const void *address, size_t size)
+void kasan_unpoison(const void *address, size_t size)
{
u8 tag = get_tag(address);

@@ -101,7 +101,7 @@ void unpoison_range(const void *address, size_t size)
*/
address = kasan_reset_tag(address);

- poison_range(address, size, tag);
+ kasan_poison(address, size, tag);

if (size & KASAN_GRANULE_MASK) {
u8 *shadow = (u8 *)kasan_mem_to_shadow(address + size);
@@ -286,7 +286,7 @@ int kasan_populate_vmalloc(unsigned long addr, unsigned long size)
* // vmalloc() allocates memory
* // let a = area->addr
* // we reach kasan_populate_vmalloc
- * // and call unpoison_range:
+ * // and call kasan_unpoison:
* STORE shadow(a), unpoison_val
* ...
* STORE shadow(a+99), unpoison_val x = LOAD p
@@ -321,7 +321,7 @@ void kasan_poison_vmalloc(const void *start, unsigned long size)
return;

size = round_up(size, KASAN_GRANULE_SIZE);
- poison_range(start, size, KASAN_VMALLOC_INVALID);
+ kasan_poison(start, size, KASAN_VMALLOC_INVALID);
}

void kasan_unpoison_vmalloc(const void *start, unsigned long size)
@@ -329,7 +329,7 @@ void kasan_unpoison_vmalloc(const void *start, unsigned long size)
if (!is_vmalloc_or_module_addr(start))
return;

- unpoison_range(start, size);
+ kasan_unpoison(start, size);
}

static int kasan_depopulate_vmalloc_pte(pte_t *ptep, unsigned long addr,
diff --git a/mm/kasan/sw_tags.c b/mm/kasan/sw_tags.c
index 5dcd830805b2..cc271fceb5d5 100644
--- a/mm/kasan/sw_tags.c
+++ b/mm/kasan/sw_tags.c
@@ -57,7 +57,7 @@ void __init kasan_init_sw_tags(void)
* sequence has in fact positive effect, since interrupts that randomly skew
* PRNG at unpredictable points do only good.
*/
-u8 random_tag(void)
+u8 kasan_random_tag(void)
{
u32 state = this_cpu_read(prng_state);

@@ -67,7 +67,7 @@ u8 random_tag(void)
return (u8)(state % (KASAN_TAG_MAX + 1));
}

-bool check_memory_region(unsigned long addr, size_t size, bool write,
+bool kasan_check_range(unsigned long addr, size_t size, bool write,
unsigned long ret_ip)
{
u8 tag;
@@ -118,7 +118,7 @@ bool check_memory_region(unsigned long addr, size_t size, bool write,
return true;
}

-bool check_invalid_free(void *addr)
+bool kasan_check_invalid_free(void *addr)

{
u8 tag = get_tag(addr);
u8 shadow_byte = READ_ONCE(*(u8 *)kasan_mem_to_shadow(kasan_reset_tag(addr)));

@@ -130,12 +130,12 @@ bool check_invalid_free(void *addr)
#define DEFINE_HWASAN_LOAD_STORE(size) \
void __hwasan_load##size##_noabort(unsigned long addr) \
{ \
- check_memory_region(addr, size, false, _RET_IP_); \
+ kasan_check_range(addr, size, false, _RET_IP_); \
} \
EXPORT_SYMBOL(__hwasan_load##size##_noabort); \
void __hwasan_store##size##_noabort(unsigned long addr) \
{ \
- check_memory_region(addr, size, true, _RET_IP_); \
+ kasan_check_range(addr, size, true, _RET_IP_); \
} \
EXPORT_SYMBOL(__hwasan_store##size##_noabort)

@@ -147,19 +147,19 @@ DEFINE_HWASAN_LOAD_STORE(16);

void __hwasan_loadN_noabort(unsigned long addr, unsigned long size)
{
- check_memory_region(addr, size, false, _RET_IP_);
+ kasan_check_range(addr, size, false, _RET_IP_);
}
EXPORT_SYMBOL(__hwasan_loadN_noabort);

void __hwasan_storeN_noabort(unsigned long addr, unsigned long size)
{
- check_memory_region(addr, size, true, _RET_IP_);
+ kasan_check_range(addr, size, true, _RET_IP_);
}
EXPORT_SYMBOL(__hwasan_storeN_noabort);

void __hwasan_tag_memory(unsigned long addr, u8 tag, unsigned long size)
{
- poison_range((void *)addr, size, tag);
+ kasan_poison((void *)addr, size, tag);
}
EXPORT_SYMBOL(__hwasan_tag_memory);

diff --git a/tools/objtool/check.c b/tools/objtool/check.c
index 5f8d3eed78a1..5b2a22591ea7 100644
--- a/tools/objtool/check.c
+++ b/tools/objtool/check.c
@@ -576,7 +576,7 @@ static void add_ignores(struct objtool_file *file)
static const char *uaccess_safe_builtin[] = {
/* KASAN */
"kasan_report",
- "check_memory_region",
+ "kasan_check_range",
/* KASAN out-of-line */
"__asan_loadN_noabort",
"__asan_load1_noabort",
--
2.30.0.284.gd98b1dd5eaa7-goog

Andrey Konovalov

unread,

Jan 14, 2021, 2:36:41 PM1/14/21

to Andrew Morton, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

Mention in the documentation that enabling CONFIG_KASAN_HW_TAGS
always results in in-kernel TBI (Top Byte Ignore) being enabled.

Also do a few minor documentation cleanups.

Link: https://linux-review.googlesource.com/id/Iba2a6697e3c6304cb53f89ec61dedc77fa29e3ae

Reviewed-by: Marco Elver <el...@google.com>
Reviewed-by: Alexander Potapenko <gli...@google.com>
Signed-off-by: Andrey Konovalov <andre...@google.com>
---

Documentation/dev-tools/kasan.rst | 16 +++++++++++-----
1 file changed, 11 insertions(+), 5 deletions(-)

diff --git a/Documentation/dev-tools/kasan.rst b/Documentation/dev-tools/kasan.rst
index 0fc3fb1860c4..26c99852a852 100644
--- a/Documentation/dev-tools/kasan.rst
+++ b/Documentation/dev-tools/kasan.rst
@@ -147,15 +147,14 @@ negative values to distinguish between different kinds of inaccessible memory
like redzones or freed memory (see mm/kasan/kasan.h).

In the report above the arrows point to the shadow byte 03, which means that
-the accessed address is partially accessible.
-
-For tag-based KASAN this last report section shows the memory tags around the
-accessed address (see `Implementation details`_ section).
+the accessed address is partially accessible. For tag-based KASAN modes this
+last report section shows the memory tags around the accessed address
+(see the `Implementation details`_ section).

Boot parameters
~~~~~~~~~~~~~~~

-Hardware tag-based KASAN mode (see the section about different mode below) is
+Hardware tag-based KASAN mode (see the section about various modes below) is
intended for use in production as a security mitigation. Therefore it supports
boot parameters that allow to disable KASAN competely or otherwise control
particular KASAN features.
@@ -305,6 +304,13 @@ reserved to tag freed memory regions.
Hardware tag-based KASAN currently only supports tagging of
kmem_cache_alloc/kmalloc and page_alloc memory.

+If the hardware doesn't support MTE (pre ARMv8.5), hardware tag-based KASAN
+won't be enabled. In this case all boot parameters are ignored.
+
+Note, that enabling CONFIG_KASAN_HW_TAGS always results in in-kernel TBI being
+enabled. Even when kasan.mode=off is provided, or when the hardware doesn't
+support MTE (but supports TBI).
+
What memory accesses are sanitised by KASAN?
--------------------------------------------

--
2.30.0.284.gd98b1dd5eaa7-goog

Andrey Konovalov

unread,

Jan 14, 2021, 2:36:42 PM1/14/21

to Andrew Morton, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

Clarify and update comments in KASAN tests.

Link: https://linux-review.googlesource.com/id/I6c816c51fa1e0eb7aa3dead6bda1f339d2af46c8

Reviewed-by: Marco Elver <el...@google.com>

Signed-off-by: Andrey Konovalov <andre...@google.com>
---

lib/test_kasan.c | 59 +++++++++++++++++++++++++----------------
lib/test_kasan_module.c | 5 ++--

2 files changed, 39 insertions(+), 25 deletions(-)

diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index 2947274cc2d3..6f46e27c2af7 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c

@@ -28,10 +28,9 @@
#define OOB_TAG_OFF (IS_ENABLED(CONFIG_KASAN_GENERIC) ? 0 : KASAN_GRANULE_SIZE)

/*
- * We assign some test results to these globals to make sure the tests
- * are not eliminated as dead code.
+ * Some tests use these global variables to store return values from function
+ * calls that could otherwise be eliminated by the compiler as dead code.
*/
-
void *kasan_ptr_result;
int kasan_int_result;

@@ -39,14 +38,13 @@ static struct kunit_resource resource;
static struct kunit_kasan_expectation fail_data;
static bool multishot;

+/*
+ * Temporarily enable multi-shot mode. Otherwise, KASAN would only report the
+ * first detected bug and panic the kernel if panic_on_warn is enabled.

+ */

static int kasan_test_init(struct kunit *test)
{

- /*
- * Temporarily enable multi-shot mode and set panic_on_warn=0.
- * Otherwise, we'd only get a report for the first case.
- */
multishot = kasan_save_enable_multi_shot();
-
return 0;
}

@@ -56,12 +54,12 @@ static void kasan_test_exit(struct kunit *test)
}

/**
- * KUNIT_EXPECT_KASAN_FAIL() - Causes a test failure when the expression does
- * not cause a KASAN error. This uses a KUnit resource named "kasan_data." Do
- * Do not use this name for a KUnit resource outside here.
- *
+ * KUNIT_EXPECT_KASAN_FAIL() - check that the executed expression produces a
+ * KASAN report; causes a test failure otherwise. This relies on a KUnit
+ * resource named "kasan_data". Do not use this name for KUnit resources
+ * outside of KASAN tests.
*/
-#define KUNIT_EXPECT_KASAN_FAIL(test, condition) do { \
+#define KUNIT_EXPECT_KASAN_FAIL(test, expression) do { \
fail_data.report_expected = true; \
fail_data.report_found = false; \
kunit_add_named_resource(test, \

@@ -69,7 +67,7 @@ static void kasan_test_exit(struct kunit *test)

NULL, \
&resource, \
"kasan_data", &fail_data); \

- condition; \
+ expression; \
KUNIT_EXPECT_EQ(test, \
fail_data.report_expected, \
fail_data.report_found); \
@@ -121,7 +119,8 @@ static void kmalloc_pagealloc_oob_right(struct kunit *test)
return;
}

- /* Allocate a chunk that does not fit into a SLUB cache to trigger
+ /*
+ * Allocate a chunk that does not fit into a SLUB cache to trigger
* the page allocator fallback.
*/
ptr = kmalloc(size, GFP_KERNEL);

@@ -168,7 +167,9 @@ static void kmalloc_large_oob_right(struct kunit *test)
{
char *ptr;

size_t size = KMALLOC_MAX_CACHE_SIZE - 256;
- /* Allocate a chunk that is large enough, but still fits into a slab
+
+ /*
+ * Allocate a chunk that is large enough, but still fits into a slab
* and does not trigger the page allocator fallback in SLUB.
*/
ptr = kmalloc(size, GFP_KERNEL);

@@ -469,10 +470,13 @@ static void ksize_unpoisons_memory(struct kunit *test)

ptr = kmalloc(size, GFP_KERNEL);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);

real_size = ksize(ptr);
- /* This access doesn't trigger an error. */

+
+ /* This access shouldn't trigger a KASAN report. */

ptr[size] = 'x';
- /* This one does. */
+
+ /* This one must. */
KUNIT_EXPECT_KASAN_FAIL(test, ptr[real_size] = 'y');
+
kfree(ptr);
}

@@ -568,7 +572,7 @@ static void kmem_cache_invalid_free(struct kunit *test)
return;
}

- /* Trigger invalid free, the object doesn't get freed */
+ /* Trigger invalid free, the object doesn't get freed. */
KUNIT_EXPECT_KASAN_FAIL(test, kmem_cache_free(cache, p + 1));

/*

@@ -585,7 +589,10 @@ static void kasan_memchr(struct kunit *test)
char *ptr;

size_t size = 24;

- /* See https://bugzilla.kernel.org/show_bug.cgi?id=206337 */
+ /*
+ * str* functions are not instrumented with CONFIG_AMD_MEM_ENCRYPT.
+ * See https://bugzilla.kernel.org/show_bug.cgi?id=206337 for details.
+ */
if (IS_ENABLED(CONFIG_AMD_MEM_ENCRYPT)) {
kunit_info(test,

"str* functions are not instrumented with CONFIG_AMD_MEM_ENCRYPT");

@@ -610,7 +617,10 @@ static void kasan_memcmp(struct kunit *test)
size_t size = 24;
int arr[9];

- /* See https://bugzilla.kernel.org/show_bug.cgi?id=206337 */
+ /*

+ * str* functions are not instrumented with CONFIG_AMD_MEM_ENCRYPT.
+ * See https://bugzilla.kernel.org/show_bug.cgi?id=206337 for details.
+ */
if (IS_ENABLED(CONFIG_AMD_MEM_ENCRYPT)) {
kunit_info(test,

"str* functions are not instrumented with CONFIG_AMD_MEM_ENCRYPT");

@@ -634,7 +644,10 @@ static void kasan_strings(struct kunit *test)
char *ptr;

size_t size = 24;

- /* See https://bugzilla.kernel.org/show_bug.cgi?id=206337 */
+ /*
+ * str* functions are not instrumented with CONFIG_AMD_MEM_ENCRYPT.
+ * See https://bugzilla.kernel.org/show_bug.cgi?id=206337 for details.
+ */
if (IS_ENABLED(CONFIG_AMD_MEM_ENCRYPT)) {
kunit_info(test,

"str* functions are not instrumented with CONFIG_AMD_MEM_ENCRYPT");

Andrey Konovalov

unread,

Jan 14, 2021, 2:36:45 PM1/14/21

to Andrew Morton, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

Some KASAN tests require specific kernel configs to be enabled.
Instead of copy-pasting the checks for these configs add a few helper
macros and use them.

Link: https://linux-review.googlesource.com/id/I237484a7fddfedf4a4aae9cc61ecbcdbe85a0a63
Suggested-by: Alexander Potapenko <gli...@google.com>

Reviewed-by: Marco Elver <el...@google.com>
Reviewed-by: Alexander Potapenko <gli...@google.com>

Signed-off-by: Andrey Konovalov <andre...@google.com>
---

lib/test_kasan.c | 101 +++++++++++++++--------------------------------
1 file changed, 31 insertions(+), 70 deletions(-)

diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index 6f46e27c2af7..714ea27fcc3e 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c

@@ -73,6 +73,20 @@ static void kasan_test_exit(struct kunit *test)
fail_data.report_found); \
} while (0)

+#define KASAN_TEST_NEEDS_CONFIG_ON(test, config) do { \
+ if (!IS_ENABLED(config)) { \
+ kunit_info((test), "skipping, " #config " required"); \
+ return; \
+ } \
+} while (0)
+
+#define KASAN_TEST_NEEDS_CONFIG_OFF(test, config) do { \
+ if (IS_ENABLED(config)) { \
+ kunit_info((test), "skipping, " #config " enabled"); \
+ return; \
+ } \
+} while (0)
+

static void kmalloc_oob_right(struct kunit *test)
{
char *ptr;
@@ -114,10 +128,7 @@ static void kmalloc_pagealloc_oob_right(struct kunit *test)
char *ptr;

size_t size = KMALLOC_MAX_CACHE_SIZE + 10;

- if (!IS_ENABLED(CONFIG_SLUB)) {
- kunit_info(test, "CONFIG_SLUB is not enabled.");
- return;
- }
+ KASAN_TEST_NEEDS_CONFIG_ON(test, CONFIG_SLUB);

/*

* Allocate a chunk that does not fit into a SLUB cache to trigger

@@ -135,10 +146,7 @@ static void kmalloc_pagealloc_uaf(struct kunit *test)
char *ptr;

size_t size = KMALLOC_MAX_CACHE_SIZE + 10;

- if (!IS_ENABLED(CONFIG_SLUB)) {
- kunit_info(test, "CONFIG_SLUB is not enabled.");
- return;
- }
+ KASAN_TEST_NEEDS_CONFIG_ON(test, CONFIG_SLUB);

ptr = kmalloc(size, GFP_KERNEL);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);

@@ -152,10 +160,7 @@ static void kmalloc_pagealloc_invalid_free(struct kunit *test)
char *ptr;

size_t size = KMALLOC_MAX_CACHE_SIZE + 10;

- if (!IS_ENABLED(CONFIG_SLUB)) {
- kunit_info(test, "CONFIG_SLUB is not enabled.");
- return;
- }
+ KASAN_TEST_NEEDS_CONFIG_ON(test, CONFIG_SLUB);

ptr = kmalloc(size, GFP_KERNEL);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);

@@ -593,11 +575,7 @@ static void kasan_memchr(struct kunit *test)

* str* functions are not instrumented with CONFIG_AMD_MEM_ENCRYPT.

* See https://bugzilla.kernel.org/show_bug.cgi?id=206337 for details.

*/
- if (IS_ENABLED(CONFIG_AMD_MEM_ENCRYPT)) {
- kunit_info(test,
- "str* functions are not instrumented with CONFIG_AMD_MEM_ENCRYPT");
- return;
- }
+ KASAN_TEST_NEEDS_CONFIG_OFF(test, CONFIG_AMD_MEM_ENCRYPT);

if (OOB_TAG_OFF)
size = round_up(size, OOB_TAG_OFF);

@@ -621,11 +599,7 @@ static void kasan_memcmp(struct kunit *test)

* str* functions are not instrumented with CONFIG_AMD_MEM_ENCRYPT.

* See https://bugzilla.kernel.org/show_bug.cgi?id=206337 for details.

*/
- if (IS_ENABLED(CONFIG_AMD_MEM_ENCRYPT)) {
- kunit_info(test,

- "str* functions are not instrumented with CONFIG_AMD_MEM_ENCRYPT");
- return;
- }
+ KASAN_TEST_NEEDS_CONFIG_OFF(test, CONFIG_AMD_MEM_ENCRYPT);

if (OOB_TAG_OFF)
size = round_up(size, OOB_TAG_OFF);

@@ -648,11 +622,7 @@ static void kasan_strings(struct kunit *test)

* str* functions are not instrumented with CONFIG_AMD_MEM_ENCRYPT.

* See https://bugzilla.kernel.org/show_bug.cgi?id=206337 for details.

*/
- if (IS_ENABLED(CONFIG_AMD_MEM_ENCRYPT)) {
- kunit_info(test,

- "str* functions are not instrumented with CONFIG_AMD_MEM_ENCRYPT");
- return;
- }
+ KASAN_TEST_NEEDS_CONFIG_OFF(test, CONFIG_AMD_MEM_ENCRYPT);

ptr = kmalloc(size, GFP_KERNEL | __GFP_ZERO);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);
@@ -713,10 +683,7 @@ static void kasan_bitops_generic(struct kunit *test)
long *bits;

/* This test is specifically crafted for the generic mode. */
- if (!IS_ENABLED(CONFIG_KASAN_GENERIC)) {
- kunit_info(test, "CONFIG_KASAN_GENERIC required\n");
- return;
- }
+ KASAN_TEST_NEEDS_CONFIG_ON(test, CONFIG_KASAN_GENERIC);

/*

* Allocate 1 more byte, which causes kzalloc to round up to 16 bytes;

Andrey Konovalov

unread,

Jan 14, 2021, 2:36:47 PM1/14/21

to Andrew Morton, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

Add 3 new tests for tag-based KASAN modes:

1. Check that match-all pointer tag is not assigned randomly.
2. Check that 0xff works as a match-all pointer tag.
3. Check that there are no match-all memory tags.

Note, that test #3 causes a significant number (255) of KASAN reports
to be printed during execution for the SW_TAGS mode.

Link: https://linux-review.googlesource.com/id/I78f1375efafa162b37f3abcb2c5bc2f3955dfd8e

Reviewed-by: Marco Elver <el...@google.com>

Signed-off-by: Andrey Konovalov <andre...@google.com>
---

lib/test_kasan.c | 92 ++++++++++++++++++++++++++++++++++++++++++++++++
mm/kasan/kasan.h | 6 ++++
2 files changed, 98 insertions(+)

diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index 714ea27fcc3e..c344fe506ffc 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c

@@ -13,6 +13,7 @@
#include <linux/mman.h>
#include <linux/module.h>
#include <linux/printk.h>
+#include <linux/random.h>
#include <linux/slab.h>
#include <linux/string.h>
#include <linux/uaccess.h>

@@ -754,6 +755,94 @@ static void vmalloc_oob(struct kunit *test)
vfree(area);
}

+/*

+ * Check that the assigned pointer tag falls within the [KASAN_TAG_MIN,
+ * KASAN_TAG_KERNEL) range (note: excluding the match-all tag) for tag-based
+ * modes.
+ */

+static void match_all_not_assigned(struct kunit *test)
+{
+ char *ptr;

+ struct page *pages;
+ int i, size, order;
+
+ KASAN_TEST_NEEDS_CONFIG_OFF(test, CONFIG_KASAN_GENERIC);
+
+ for (i = 0; i < 256; i++) {

+ size = (get_random_int() % 1024) + 1;

+ ptr = kmalloc(size, GFP_KERNEL);
+ KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);

+ KUNIT_EXPECT_GE(test, (u8)get_tag(ptr), (u8)KASAN_TAG_MIN);
+ KUNIT_EXPECT_LT(test, (u8)get_tag(ptr), (u8)KASAN_TAG_KERNEL);
+ kfree(ptr);
+ }
+
+ for (i = 0; i < 256; i++) {

+ order = (get_random_int() % 4) + 1;

+ pages = alloc_pages(GFP_KERNEL, order);
+ ptr = page_address(pages);
+ KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);
+ KUNIT_EXPECT_GE(test, (u8)get_tag(ptr), (u8)KASAN_TAG_MIN);
+ KUNIT_EXPECT_LT(test, (u8)get_tag(ptr), (u8)KASAN_TAG_KERNEL);
+ free_pages((unsigned long)ptr, order);
+ }
+}
+
+/* Check that 0xff works as a match-all pointer tag for tag-based modes. */

+static void match_all_ptr_tag(struct kunit *test)
+{
+ char *ptr;

+ u8 tag;
+
+ KASAN_TEST_NEEDS_CONFIG_OFF(test, CONFIG_KASAN_GENERIC);
+

+ ptr = kmalloc(128, GFP_KERNEL);
+ KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);
+

+ /* Backup the assigned tag. */
+ tag = get_tag(ptr);
+ KUNIT_EXPECT_NE(test, tag, (u8)KASAN_TAG_KERNEL);
+
+ /* Reset the tag to 0xff.*/
+ ptr = set_tag(ptr, KASAN_TAG_KERNEL);

+
+ /* This access shouldn't trigger a KASAN report. */

+ *ptr = 0;
+
+ /* Recover the pointer tag and free. */
+ ptr = set_tag(ptr, tag);
+ kfree(ptr);
+}
+
+/* Check that there are no match-all memory tags for tag-based modes. */

+static void match_all_mem_tag(struct kunit *test)
+{
+ char *ptr;
+ int tag;

+
+ KASAN_TEST_NEEDS_CONFIG_OFF(test, CONFIG_KASAN_GENERIC);
+

+ ptr = kmalloc(128, GFP_KERNEL);
+ KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);

+ KUNIT_EXPECT_NE(test, (u8)get_tag(ptr), (u8)KASAN_TAG_KERNEL);
+
+ /* For each possible tag value not matching the pointer tag. */
+ for (tag = KASAN_TAG_MIN; tag <= KASAN_TAG_KERNEL; tag++) {
+ if (tag == get_tag(ptr))
+ continue;
+
+ /* Mark the first memory granule with the chosen memory tag. */
+ kasan_poison(ptr, KASAN_GRANULE_SIZE, (u8)tag);

+

+ /* This access must cause a KASAN report. */
+ KUNIT_EXPECT_KASAN_FAIL(test, *ptr = 0);
+ }
+
+ /* Recover the memory tag and free. */
+ kasan_poison(ptr, KASAN_GRANULE_SIZE, get_tag(ptr));
+ kfree(ptr);
+}
+

static struct kunit_case kasan_kunit_test_cases[] = {

KUNIT_CASE(kmalloc_oob_right),
KUNIT_CASE(kmalloc_oob_left),
@@ -793,6 +882,9 @@ static struct kunit_case kasan_kunit_test_cases[] = {
KUNIT_CASE(kasan_bitops_tags),
KUNIT_CASE(kmalloc_double_kzfree),
KUNIT_CASE(vmalloc_oob),
+ KUNIT_CASE(match_all_not_assigned),
+ KUNIT_CASE(match_all_ptr_tag),
+ KUNIT_CASE(match_all_mem_tag),
{}
};

diff --git a/mm/kasan/kasan.h b/mm/kasan/kasan.h
index 3b38baddec47..c3fb9bf241d3 100644
--- a/mm/kasan/kasan.h
+++ b/mm/kasan/kasan.h

Andrey Konovalov

unread,

Jan 14, 2021, 2:36:49 PM1/14/21

to Andrew Morton, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

On a high level, this patch allows running KUnit KASAN tests with the
hardware tag-based KASAN mode.

Internally, this change reenables tag checking at the end of each KASAN
test that triggers a tag fault and leads to tag checking being disabled.

With this patch KASAN tests are still failing for the hardware tag-based
mode; fixes come in the next few patches.

Link: https://linux-review.googlesource.com/id/Id94dc9eccd33b23cda4950be408c27f879e474c8

Signed-off-by: Andrey Konovalov <andre...@google.com>
---

arch/arm64/include/asm/memory.h | 1 +
arch/arm64/include/asm/mte-kasan.h | 12 +++++++++
arch/arm64/kernel/mte.c | 12 +++++++++
arch/arm64/mm/fault.c | 16 +++++++-----
lib/Kconfig.kasan | 4 +--
lib/test_kasan.c | 42 +++++++++++++++++++++---------
mm/kasan/kasan.h | 9 +++++++
7 files changed, 75 insertions(+), 21 deletions(-)

diff --git a/arch/arm64/include/asm/memory.h b/arch/arm64/include/asm/memory.h
index 18fce223b67b..cedfc9e97bcc 100644
--- a/arch/arm64/include/asm/memory.h
+++ b/arch/arm64/include/asm/memory.h
@@ -232,6 +232,7 @@ static inline const void *__tag_set(const void *addr, u8 tag)

#ifdef CONFIG_KASAN_HW_TAGS
#define arch_enable_tagging() mte_enable_kernel()
+#define arch_set_tagging_report_once(state) mte_set_report_once(state)
#define arch_init_tags(max_tag) mte_init_tags(max_tag)
#define arch_get_random_tag() mte_get_random_tag()
#define arch_get_mem_tag(addr) mte_get_mem_tag(addr)
diff --git a/arch/arm64/include/asm/mte-kasan.h b/arch/arm64/include/asm/mte-kasan.h
index 26349a4b5e2e..3748d5bb88c0 100644
--- a/arch/arm64/include/asm/mte-kasan.h
+++ b/arch/arm64/include/asm/mte-kasan.h
@@ -32,6 +32,9 @@ void *mte_set_mem_tag_range(void *addr, size_t size, u8 tag);
void mte_enable_kernel(void);
void mte_init_tags(u64 max_tag);

+void mte_set_report_once(bool state);
+bool mte_report_once(void);
+
#else /* CONFIG_ARM64_MTE */

static inline u8 mte_get_ptr_tag(void *ptr)
@@ -60,6 +63,15 @@ static inline void mte_init_tags(u64 max_tag)
{
}

+static inline void mte_set_report_once(bool state)
+{
+}
+
+static inline bool mte_report_once(void)
+{

+ return false;
+}
+

#endif /* CONFIG_ARM64_MTE */

#endif /* __ASSEMBLY__ */
diff --git a/arch/arm64/kernel/mte.c b/arch/arm64/kernel/mte.c
index dc9ada64feed..c63b3d7a3cd9 100644
--- a/arch/arm64/kernel/mte.c
+++ b/arch/arm64/kernel/mte.c
@@ -25,6 +25,8 @@

u64 gcr_kernel_excl __ro_after_init;

+static bool report_fault_once = true;
+
static void mte_sync_page_tags(struct page *page, pte_t *ptep, bool check_swap)
{
pte_t old_pte = READ_ONCE(*ptep);
@@ -158,6 +160,16 @@ void mte_enable_kernel(void)
isb();
}

+void mte_set_report_once(bool state)
+{
+ WRITE_ONCE(report_fault_once, state);
+}
+
+bool mte_report_once(void)
+{
+ return READ_ONCE(report_fault_once);
+}
+
static void update_sctlr_el1_tcf0(u64 tcf0)
{
/* ISB required for the kernel uaccess routines */
diff --git a/arch/arm64/mm/fault.c b/arch/arm64/mm/fault.c
index a218f6f2fdc8..f1b77dc79948 100644
--- a/arch/arm64/mm/fault.c
+++ b/arch/arm64/mm/fault.c
@@ -302,7 +302,14 @@ static void die_kernel_fault(const char *msg, unsigned long addr,
static void report_tag_fault(unsigned long addr, unsigned int esr,
struct pt_regs *regs)
{
- bool is_write = ((esr & ESR_ELx_WNR) >> ESR_ELx_WNR_SHIFT) != 0;
+ static bool reported;
+ bool is_write;
+
+ if (READ_ONCE(reported))
+ return;
+
+ if (mte_report_once())
+ WRITE_ONCE(reported, true);

/* The format of KASAN tags is 0xF<x>. */
addr |= (0xF0UL << MTE_TAG_SHIFT);
@@ -310,6 +317,7 @@ static void report_tag_fault(unsigned long addr, unsigned int esr,
* SAS bits aren't set for all faults reported in EL1, so we can't
* find out access size.
*/
+ is_write = ((esr & ESR_ELx_WNR) >> ESR_ELx_WNR_SHIFT) != 0;
kasan_report(addr, 0, is_write, regs->pc);
}
#else
@@ -321,12 +329,8 @@ static inline void report_tag_fault(unsigned long addr, unsigned int esr,
static void do_tag_recovery(unsigned long addr, unsigned int esr,
struct pt_regs *regs)
{
- static bool reported;

- if (!READ_ONCE(reported)) {
- report_tag_fault(addr, esr, regs);
- WRITE_ONCE(reported, true);
- }
+ report_tag_fault(addr, esr, regs);

/*
* Disable MTE Tag Checking on the local CPU for the current EL.
diff --git a/lib/Kconfig.kasan b/lib/Kconfig.kasan
index f5fa4ba126bf..3091432acb0a 100644
--- a/lib/Kconfig.kasan
+++ b/lib/Kconfig.kasan
@@ -190,11 +190,11 @@ config KASAN_KUNIT_TEST
kernel debugging features like KASAN.

For more information on KUnit and unit tests in general, please refer
- to the KUnit documentation in Documentation/dev-tools/kunit
+ to the KUnit documentation in Documentation/dev-tools/kunit.

config TEST_KASAN_MODULE
tristate "KUnit-incompatible tests of KASAN bug detection capabilities"
- depends on m && KASAN
+ depends on m && KASAN && !KASAN_HW_TAGS
help
This is a part of the KASAN test suite that is incompatible with
KUnit. Currently includes tests that do bad copy_from/to_user
diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index c344fe506ffc..ef663bcf83e5 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c
@@ -41,16 +41,20 @@ static bool multishot;

/*

* Temporarily enable multi-shot mode. Otherwise, KASAN would only report the

- * first detected bug and panic the kernel if panic_on_warn is enabled.
+ * first detected bug and panic the kernel if panic_on_warn is enabled. For
+ * hardware tag-based KASAN also allow tag checking to be reenabled for each
+ * test, see the comment for KUNIT_EXPECT_KASAN_FAIL().

*/
static int kasan_test_init(struct kunit *test)
{

multishot = kasan_save_enable_multi_shot();
+ hw_set_tagging_report_once(false);
return 0;

}

static void kasan_test_exit(struct kunit *test)

{
+ hw_set_tagging_report_once(true);
kasan_restore_multi_shot(multishot);
}

@@ -59,19 +63,31 @@ static void kasan_test_exit(struct kunit *test)

* KASAN report; causes a test failure otherwise. This relies on a KUnit

* resource named "kasan_data". Do not use this name for KUnit resources

* outside of KASAN tests.

+ *
+ * For hardware tag-based KASAN, when a tag fault happens, tag checking is
+ * normally auto-disabled. When this happens, this test handler reenables
+ * tag checking. As tag checking can be only disabled or enabled per CPU, this
+ * handler disables migration (preemption).
*/
-#define KUNIT_EXPECT_KASAN_FAIL(test, expression) do { \

- fail_data.report_expected = true; \
- fail_data.report_found = false; \

- kunit_add_named_resource(test, \
- NULL, \
- NULL, \
- &resource, \
- "kasan_data", &fail_data); \
- expression; \
- KUNIT_EXPECT_EQ(test, \
- fail_data.report_expected, \
- fail_data.report_found); \
+#define KUNIT_EXPECT_KASAN_FAIL(test, expression) do { \
+ if (IS_ENABLED(CONFIG_KASAN_HW_TAGS)) \
+ migrate_disable(); \
+ fail_data.report_expected = true; \
+ fail_data.report_found = false; \
+ kunit_add_named_resource(test, \
+ NULL, \
+ NULL, \
+ &resource, \
+ "kasan_data", &fail_data); \
+ expression; \
+ KUNIT_EXPECT_EQ(test, \
+ fail_data.report_expected, \
+ fail_data.report_found); \
+ if (IS_ENABLED(CONFIG_KASAN_HW_TAGS)) { \
+ if (fail_data.report_found) \
+ hw_enable_tagging(); \
+ migrate_enable(); \
+ } \
} while (0)

#define KASAN_TEST_NEEDS_CONFIG_ON(test, config) do { \
diff --git a/mm/kasan/kasan.h b/mm/kasan/kasan.h
index c3fb9bf241d3..292dfbc37deb 100644
--- a/mm/kasan/kasan.h
+++ b/mm/kasan/kasan.h
@@ -280,6 +280,9 @@ static inline const void *arch_kasan_set_tag(const void *addr, u8 tag)
#ifndef arch_init_tags
#define arch_init_tags(max_tag)
#endif
+#ifndef arch_set_tagging_report_once
+#define arch_set_tagging_report_once(state)
+#endif
#ifndef arch_get_random_tag
#define arch_get_random_tag() (0xFF)
#endif
@@ -292,10 +295,16 @@ static inline const void *arch_kasan_set_tag(const void *addr, u8 tag)

#define hw_enable_tagging() arch_enable_tagging()
#define hw_init_tags(max_tag) arch_init_tags(max_tag)
+#define hw_set_tagging_report_once(state) arch_set_tagging_report_once(state)
#define hw_get_random_tag() arch_get_random_tag()
#define hw_get_mem_tag(addr) arch_get_mem_tag(addr)
#define hw_set_mem_tag_range(addr, size, tag) arch_set_mem_tag_range((addr), (size), (tag))

+#else /* CONFIG_KASAN_HW_TAGS */
+
+#define hw_enable_tagging()
+#define hw_set_tagging_report_once(state)
+

#endif /* CONFIG_KASAN_HW_TAGS */

#ifdef CONFIG_KASAN_SW_TAGS

--
2.30.0.284.gd98b1dd5eaa7-goog

Andrey Konovalov

unread,

Jan 14, 2021, 2:36:52 PM1/14/21

to Andrew Morton, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

Rename CONFIG_TEST_KASAN_MODULE to CONFIG_KASAN_MODULE_TEST.

This naming is more consistent with the existing CONFIG_KASAN_KUNIT_TEST.

Link: https://linux-review.googlesource.com/id/Id347dfa5fe8788b7a1a189863e039f409da0ae5f

Reviewed-by: Marco Elver <el...@google.com>
Reviewed-by: Alexander Potapenko <gli...@google.com>

Signed-off-by: Andrey Konovalov <andre...@google.com>
---

Documentation/dev-tools/kasan.rst | 8 ++++----
lib/Kconfig.kasan | 2 +-
lib/Makefile | 2 +-
3 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/Documentation/dev-tools/kasan.rst b/Documentation/dev-tools/kasan.rst
index 26c99852a852..b25ae43d683e 100644
--- a/Documentation/dev-tools/kasan.rst
+++ b/Documentation/dev-tools/kasan.rst
@@ -374,17 +374,17 @@ unmapped. This will require changes in arch-specific code.
This allows ``VMAP_STACK`` support on x86, and can simplify support of
architectures that do not have a fixed module region.

-CONFIG_KASAN_KUNIT_TEST & CONFIG_TEST_KASAN_MODULE
---------------------------------------------------
+CONFIG_KASAN_KUNIT_TEST and CONFIG_KASAN_MODULE_TEST
+----------------------------------------------------

-KASAN tests consist on two parts:
+KASAN tests consist of two parts:

1. Tests that are integrated with the KUnit Test Framework. Enabled with
``CONFIG_KASAN_KUNIT_TEST``. These tests can be run and partially verified
automatically in a few different ways, see the instructions below.

2. Tests that are currently incompatible with KUnit. Enabled with
-``CONFIG_TEST_KASAN_MODULE`` and can only be run as a module. These tests can
+``CONFIG_KASAN_MODULE_TEST`` and can only be run as a module. These tests can
only be verified manually, by loading the kernel module and inspecting the
kernel log for KASAN reports.

diff --git a/lib/Kconfig.kasan b/lib/Kconfig.kasan
index 3091432acb0a..624ae1df7984 100644
--- a/lib/Kconfig.kasan
+++ b/lib/Kconfig.kasan
@@ -192,7 +192,7 @@ config KASAN_KUNIT_TEST

For more information on KUnit and unit tests in general, please refer

to the KUnit documentation in Documentation/dev-tools/kunit.

-config TEST_KASAN_MODULE
+config KASAN_MODULE_TEST

tristate "KUnit-incompatible tests of KASAN bug detection capabilities"

depends on m && KASAN && !KASAN_HW_TAGS
help

diff --git a/lib/Makefile b/lib/Makefile
index afeff05fa8c5..122f25d6407e 100644
--- a/lib/Makefile
+++ b/lib/Makefile
@@ -68,7 +68,7 @@ obj-$(CONFIG_TEST_IDA) += test_ida.o
obj-$(CONFIG_KASAN_KUNIT_TEST) += test_kasan.o
CFLAGS_test_kasan.o += -fno-builtin
CFLAGS_test_kasan.o += $(call cc-disable-warning, vla)
-obj-$(CONFIG_TEST_KASAN_MODULE) += test_kasan_module.o
+obj-$(CONFIG_KASAN_MODULE_TEST) += test_kasan_module.o
CFLAGS_test_kasan_module.o += -fno-builtin
obj-$(CONFIG_TEST_UBSAN) += test_ubsan.o
CFLAGS_test_ubsan.o += $(call cc-disable-warning, vla)
--
2.30.0.284.gd98b1dd5eaa7-goog

Andrey Konovalov

unread,

Jan 14, 2021, 2:36:53 PM1/14/21

to Andrew Morton, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

It might not be obvious to the compiler that the expression must be
executed between writing and reading to fail_data. In this case, the
compiler might reorder or optimize away some of the accesses, and
the tests will fail.

Add compiler barriers around the expression in KUNIT_EXPECT_KASAN_FAIL
and use READ/WRITE_ONCE() for accessing fail_data fields.

Link: https://linux-review.googlesource.com/id/I046079f48641a1d36fe627fc8827a9249102fd50

Reviewed-by: Marco Elver <el...@google.com>

Signed-off-by: Andrey Konovalov <andre...@google.com>
---

lib/test_kasan.c | 17 ++++++++++++-----
mm/kasan/report.c | 2 +-

2 files changed, 13 insertions(+), 6 deletions(-)

diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index ef663bcf83e5..2419e36e117b 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c
@@ -68,23 +68,30 @@ static void kasan_test_exit(struct kunit *test)

* normally auto-disabled. When this happens, this test handler reenables

* tag checking. As tag checking can be only disabled or enabled per CPU, this

* handler disables migration (preemption).

+ *
+ * Since the compiler doesn't see that the expression can change the fail_data
+ * fields, it can reorder or optimize away the accesses to those fields.
+ * Use READ/WRITE_ONCE() for the accesses and compiler barriers around the
+ * expression to prevent that.
*/
#define KUNIT_EXPECT_KASAN_FAIL(test, expression) do { \
if (IS_ENABLED(CONFIG_KASAN_HW_TAGS)) \
migrate_disable(); \

- fail_data.report_expected = true; \
- fail_data.report_found = false; \

+ WRITE_ONCE(fail_data.report_expected, true); \
+ WRITE_ONCE(fail_data.report_found, false); \
kunit_add_named_resource(test, \
NULL, \

NULL, \
&resource, \
"kasan_data", &fail_data); \

+ barrier(); \
expression; \
+ barrier(); \

KUNIT_EXPECT_EQ(test, \
- fail_data.report_expected, \
- fail_data.report_found); \

+ READ_ONCE(fail_data.report_expected), \
+ READ_ONCE(fail_data.report_found)); \
if (IS_ENABLED(CONFIG_KASAN_HW_TAGS)) { \
- if (fail_data.report_found) \
+ if (READ_ONCE(fail_data.report_found)) \
hw_enable_tagging(); \
migrate_enable(); \
} \

diff --git a/mm/kasan/report.c b/mm/kasan/report.c
index e93d7973792e..234f35a84f19 100644
--- a/mm/kasan/report.c
+++ b/mm/kasan/report.c

Andrey Konovalov

unread,

Jan 14, 2021, 2:36:57 PM1/14/21

to Andrew Morton, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

In the kmalloc_uaf2() test, the pointers to the two allocated memory
blocks might happen to be the same, and the test will fail. With the
software tag-based mode, the probability of the that is 1/254, so it's
hard to observe the failure. For the hardware tag-based mode though,
the probablity is 1/14, which is quite noticable.

Allow up to 16 attempts at generating different tags for the tag-based
modes.

Link: https://linux-review.googlesource.com/id/Ibfa458ef2804ff465d8eb07434a300bf36388d55

Reviewed-by: Marco Elver <el...@google.com>
Reviewed-by: Alexander Potapenko <gli...@google.com>

Signed-off-by: Andrey Konovalov <andre...@google.com>
---

lib/test_kasan.c | 11 +++++++++++
1 file changed, 11 insertions(+)

diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index 2419e36e117b..0cda4a1ff394 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c

Andrey Konovalov

unread,

Jan 14, 2021, 2:36:58 PM1/14/21

to Andrew Morton, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

Since the hardware tag-based KASAN mode might not have a redzone that
comes after an allocated object (when kasan.mode=prod is enabled), the
kasan_bitops_tags() test ends up corrupting the next object in memory.

Change the test so it always accesses the redzone that lies within the
allocated object's boundaries.

Link: https://linux-review.googlesource.com/id/I67f51d1ee48f0a8d0fe2658c2a39e4879fe0832a

Reviewed-by: Marco Elver <el...@google.com>
Reviewed-by: Alexander Potapenko <gli...@google.com>
Signed-off-by: Andrey Konovalov <andre...@google.com>
---

lib/test_kasan.c | 10 +++++-----
1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index 0cda4a1ff394..a06e7946f581 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c
@@ -749,13 +749,13 @@ static void kasan_bitops_tags(struct kunit *test)

/* This test is specifically crafted for tag-based modes. */

KASAN_TEST_NEEDS_CONFIG_OFF(test, CONFIG_KASAN_GENERIC);

- /* Allocation size will be rounded to up granule size, which is 16. */
- bits = kzalloc(sizeof(*bits), GFP_KERNEL);
+ /* kmalloc-64 cache will be used and the last 16 bytes will be the redzone. */
+ bits = kzalloc(48, GFP_KERNEL);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, bits);

- /* Do the accesses past the 16 allocated bytes. */
- kasan_bitops_modify(test, BITS_PER_LONG, &bits[1]);
- kasan_bitops_test_and_modify(test, BITS_PER_LONG + BITS_PER_BYTE, &bits[1]);
+ /* Do the accesses past the 48 allocated bytes, but within the redone. */
+ kasan_bitops_modify(test, BITS_PER_LONG, (void *)bits + 48);
+ kasan_bitops_test_and_modify(test, BITS_PER_LONG + BITS_PER_BYTE, (void *)bits + 48);

kfree(bits);
}
--
2.30.0.284.gd98b1dd5eaa7-goog

Andrey Konovalov

unread,

Jan 14, 2021, 2:37:01 PM1/14/21

to Andrew Morton, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

Generic mm functions that call KASAN annotations that might report a bug
pass _RET_IP_ to them as an argument. This allows KASAN to include the
name of the function that called the mm function in its report's header.

Now that KASAN has inline wrappers for all of its annotations, move
_RET_IP_ to those wrappers to simplify annotation call sites.

Link: https://linux-review.googlesource.com/id/I8fb3c06d49671305ee184175a39591bc26647a67

Signed-off-by: Andrey Konovalov <andre...@google.com>
---

include/linux/kasan.h | 20 +++++++++-----------

mm/mempool.c | 2 +-
mm/slab.c | 2 +-

mm/slub.c | 4 ++--
4 files changed, 13 insertions(+), 15 deletions(-)

diff --git a/include/linux/kasan.h b/include/linux/kasan.h
index 5e0655fb2a6f..bba1637827c3 100644
--- a/include/linux/kasan.h
+++ b/include/linux/kasan.h
@@ -181,19 +181,18 @@ static __always_inline void * __must_check kasan_init_slab_obj(
}

bool __kasan_slab_free(struct kmem_cache *s, void *object, unsigned long ip);
-static __always_inline bool kasan_slab_free(struct kmem_cache *s, void *object,
- unsigned long ip)
+static __always_inline bool kasan_slab_free(struct kmem_cache *s, void *object)
{
if (kasan_enabled())
- return __kasan_slab_free(s, object, ip);
+ return __kasan_slab_free(s, object, _RET_IP_);
return false;
}

void __kasan_slab_free_mempool(void *ptr, unsigned long ip);
-static __always_inline void kasan_slab_free_mempool(void *ptr, unsigned long ip)
+static __always_inline void kasan_slab_free_mempool(void *ptr)
{
if (kasan_enabled())
- __kasan_slab_free_mempool(ptr, ip);
+ __kasan_slab_free_mempool(ptr, _RET_IP_);
}

void * __must_check __kasan_slab_alloc(struct kmem_cache *s,
@@ -237,10 +236,10 @@ static __always_inline void * __must_check kasan_krealloc(const void *object,
}

void __kasan_kfree_large(void *ptr, unsigned long ip);
-static __always_inline void kasan_kfree_large(void *ptr, unsigned long ip)
+static __always_inline void kasan_kfree_large(void *ptr)
{
if (kasan_enabled())
- __kasan_kfree_large(ptr, ip);
+ __kasan_kfree_large(ptr, _RET_IP_);
}

bool kasan_save_enable_multi_shot(void);
@@ -273,12 +272,11 @@ static inline void *kasan_init_slab_obj(struct kmem_cache *cache,
{
return (void *)object;
}
-static inline bool kasan_slab_free(struct kmem_cache *s, void *object,
- unsigned long ip)
+static inline bool kasan_slab_free(struct kmem_cache *s, void *object)
{
return false;
}
-static inline void kasan_slab_free_mempool(void *ptr, unsigned long ip) {}
+static inline void kasan_slab_free_mempool(void *ptr) {}
static inline void *kasan_slab_alloc(struct kmem_cache *s, void *object,
gfp_t flags)
{
@@ -298,7 +296,7 @@ static inline void *kasan_krealloc(const void *object, size_t new_size,
{
return (void *)object;
}
-static inline void kasan_kfree_large(void *ptr, unsigned long ip) {}
+static inline void kasan_kfree_large(void *ptr) {}

#endif /* CONFIG_KASAN */

diff --git a/mm/mempool.c b/mm/mempool.c
index 624ed51b060f..79959fac27d7 100644
--- a/mm/mempool.c
+++ b/mm/mempool.c
@@ -104,7 +104,7 @@ static inline void poison_element(mempool_t *pool, void *element)
static __always_inline void kasan_poison_element(mempool_t *pool, void *element)
{
if (pool->alloc == mempool_alloc_slab || pool->alloc == mempool_kmalloc)
- kasan_slab_free_mempool(element, _RET_IP_);
+ kasan_slab_free_mempool(element);
else if (pool->alloc == mempool_alloc_pages)
kasan_free_pages(element, (unsigned long)pool->pool_data);
}
diff --git a/mm/slab.c b/mm/slab.c
index d7c8da9319c7..afeb6191fb1e 100644
--- a/mm/slab.c
+++ b/mm/slab.c
@@ -3421,7 +3421,7 @@ static __always_inline void __cache_free(struct kmem_cache *cachep, void *objp,
memset(objp, 0, cachep->object_size);

/* Put the object into the quarantine, don't touch it for now. */
- if (kasan_slab_free(cachep, objp, _RET_IP_))
+ if (kasan_slab_free(cachep, objp))
return;

/* Use KCSAN to help debug racy use-after-free. */
diff --git a/mm/slub.c b/mm/slub.c
index 75fb097d990d..0afb53488238 100644
--- a/mm/slub.c
+++ b/mm/slub.c
@@ -1514,7 +1514,7 @@ static inline void *kmalloc_large_node_hook(void *ptr, size_t size, gfp_t flags)
static __always_inline void kfree_hook(void *x)
{
kmemleak_free(x);
- kasan_kfree_large(x, _RET_IP_);
+ kasan_kfree_large(x);
}

static __always_inline bool slab_free_hook(struct kmem_cache *s, void *x)
@@ -1544,7 +1544,7 @@ static __always_inline bool slab_free_hook(struct kmem_cache *s, void *x)
KCSAN_ACCESS_WRITE | KCSAN_ACCESS_ASSERT);

/* KASAN might put x into memory quarantine, delaying its reuse */
- return kasan_slab_free(s, x, _RET_IP_);
+ return kasan_slab_free(s, x);
}

static inline bool slab_free_freelist_hook(struct kmem_cache *s,
--
2.30.0.284.gd98b1dd5eaa7-goog

Andrey Konovalov

unread,

Jan 14, 2021, 2:37:03 PM1/14/21

to Andrew Morton, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

The currently existing kasan_check_read/write() annotations are intended
to be used for kernel modules that have KASAN compiler instrumentation
disabled. Thus, they are only relevant for the software KASAN modes that
rely on compiler instrumentation.

However there's another use case for these annotations: ksize() checks
that the object passed to it is indeed accessible before unpoisoning the
whole object. This is currently done via __kasan_check_read(), which is
compiled away for the hardware tag-based mode that doesn't rely on
compiler instrumentation. This leads to KASAN missing detecting some
memory corruptions.

Provide another annotation called kasan_check_byte() that is available
for all KASAN modes. As the implementation rename and reuse
kasan_check_invalid_free(). Use this new annotation in ksize().

To avoid having ksize() as the top frame in the reported stack trace
pass _RET_IP_ to __kasan_check_byte().

Also add a new ksize_uaf() test that checks that a use-after-free is
detected via ksize() itself, and via plain accesses that happen later.

Link: https://linux-review.googlesource.com/id/Iaabf771881d0f9ce1b969f2a62938e99d3308ec5

Signed-off-by: Andrey Konovalov <andre...@google.com>
---

include/linux/kasan-checks.h | 6 ++++++
include/linux/kasan.h | 17 +++++++++++++++++

mm/slab_common.c | 16 +++++++++-------
8 files changed, 72 insertions(+), 18 deletions(-)

diff --git a/include/linux/kasan-checks.h b/include/linux/kasan-checks.h
index ca5e89fb10d3..3d6d22a25bdc 100644
--- a/include/linux/kasan-checks.h
+++ b/include/linux/kasan-checks.h
@@ -4,6 +4,12 @@

#include <linux/types.h>

+/*
+ * The annotations present in this file are only relevant for the software
+ * KASAN modes that rely on compiler instrumentation, and will be optimized
+ * away for the hardware tag-based KASAN mode. Use kasan_check_byte() instead.
+ */
+
/*
* __kasan_check_*: Always available when KASAN is enabled. This may be used
* even in compilation units that selectively disable KASAN, but must use KASAN

diff --git a/include/linux/kasan.h b/include/linux/kasan.h
index bba1637827c3..5bedd5ee481f 100644
--- a/include/linux/kasan.h
+++ b/include/linux/kasan.h
@@ -242,6 +242,19 @@ static __always_inline void kasan_kfree_large(void *ptr)
__kasan_kfree_large(ptr, _RET_IP_);

}

+/*
+ * Unlike kasan_check_read/write(), kasan_check_byte() is performed even for
+ * the hardware tag-based mode that doesn't rely on compiler instrumentation.
+ */

+bool __kasan_check_byte(const void *addr, unsigned long ip);
+static __always_inline bool kasan_check_byte(const void *addr)
+{
+ if (kasan_enabled())
+ return __kasan_check_byte(addr, _RET_IP_);
+ return true;
+}
+
+

bool kasan_save_enable_multi_shot(void);
void kasan_restore_multi_shot(bool enabled);

@@ -297,6 +310,10 @@ static inline void *kasan_krealloc(const void *object, size_t new_size,
return (void *)object;
}
static inline void kasan_kfree_large(void *ptr) {}
+static inline bool kasan_check_byte(const void *address)

+{
+ return true;
+}

#endif /* CONFIG_KASAN */

diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index a06e7946f581..566d894ba20b 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c

@@ -496,6 +496,7 @@ static void kasan_global_oob(struct kunit *test)
KUNIT_EXPECT_KASAN_FAIL(test, *(volatile char *)p);
}

+/* Check that ksize() makes the whole object accessible. */

static void ksize_unpoisons_memory(struct kunit *test)
{
char *ptr;
@@ -514,6 +515,24 @@ static void ksize_unpoisons_memory(struct kunit *test)
kfree(ptr);
}

+/*

+ * Check that a use-after-free is detected by ksize() and via normal accesses
+ * after it.
+ */

+static void ksize_uaf(struct kunit *test)
+{
+ char *ptr;

+ int size = 128 - KASAN_GRANULE_SIZE;
+

+ ptr = kmalloc(size, GFP_KERNEL);
+ KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);

+ kfree(ptr);
+
+ KUNIT_EXPECT_KASAN_FAIL(test, ksize(ptr));
+ KUNIT_EXPECT_KASAN_FAIL(test, kasan_int_result = *ptr);
+ KUNIT_EXPECT_KASAN_FAIL(test, kasan_int_result = *(ptr + size));
+}
+

static void kasan_stack_oob(struct kunit *test)

{
char stack_array[10];
@@ -907,6 +926,7 @@ static struct kunit_case kasan_kunit_test_cases[] = {
KUNIT_CASE(kasan_alloca_oob_left),
KUNIT_CASE(kasan_alloca_oob_right),
KUNIT_CASE(ksize_unpoisons_memory),
+ KUNIT_CASE(ksize_uaf),
KUNIT_CASE(kmem_cache_double_free),
KUNIT_CASE(kmem_cache_invalid_free),
KUNIT_CASE(kasan_memchr),

diff --git a/mm/kasan/common.c b/mm/kasan/common.c
index eedc3e0fe365..b18189ef3a92 100644
--- a/mm/kasan/common.c
+++ b/mm/kasan/common.c
@@ -345,7 +345,7 @@ static bool ____kasan_slab_free(struct kmem_cache *cache, void *object,

if (unlikely(cache->flags & SLAB_TYPESAFE_BY_RCU))
return false;

- if (kasan_check_invalid_free(tagged_object)) {
+ if (!kasan_byte_accessible(tagged_object)) {
kasan_report_invalid_free(tagged_object, ip);
return true;
}
@@ -490,3 +490,12 @@ void __kasan_kfree_large(void *ptr, unsigned long ip)
kasan_report_invalid_free(ptr, ip);
/* The object will be poisoned by kasan_free_pages(). */
}
+

+bool __kasan_check_byte(const void *address, unsigned long ip)
+{

+ if (!kasan_byte_accessible(address)) {
+ kasan_report((unsigned long)address, 1, false, ip);

+ return false;
+ }
+ return true;
+}
diff --git a/mm/kasan/generic.c b/mm/kasan/generic.c
index acab8862dc67..3f17a1218055 100644
--- a/mm/kasan/generic.c
+++ b/mm/kasan/generic.c
@@ -185,11 +185,11 @@ bool kasan_check_range(unsigned long addr, size_t size, bool write,

return check_region_inline(addr, size, write, ret_ip);
}

-bool kasan_check_invalid_free(void *addr)
+bool kasan_byte_accessible(const void *addr)

{
s8 shadow_byte = READ_ONCE(*(s8 *)kasan_mem_to_shadow(addr));

- return shadow_byte < 0 || shadow_byte >= KASAN_GRANULE_SIZE;
+ return shadow_byte >= 0 && shadow_byte < KASAN_GRANULE_SIZE;
}

void kasan_cache_shrink(struct kmem_cache *cache)

diff --git a/mm/kasan/kasan.h b/mm/kasan/kasan.h
index 292dfbc37deb..bd4ee6fab648 100644
--- a/mm/kasan/kasan.h
+++ b/mm/kasan/kasan.h
@@ -329,20 +329,20 @@ static inline void kasan_unpoison(const void *address, size_t size)
round_up(size, KASAN_GRANULE_SIZE), get_tag(address));
}

-static inline bool kasan_check_invalid_free(void *addr)

+static inline bool kasan_byte_accessible(const void *addr)
{
u8 ptr_tag = get_tag(addr);

- u8 mem_tag = hw_get_mem_tag(addr);
+ u8 mem_tag = hw_get_mem_tag((void *)addr);

- return (mem_tag == KASAN_TAG_INVALID) ||
- (ptr_tag != KASAN_TAG_KERNEL && ptr_tag != mem_tag);
+ return (mem_tag != KASAN_TAG_INVALID) &&
+ (ptr_tag == KASAN_TAG_KERNEL || ptr_tag == mem_tag);
}

#else /* CONFIG_KASAN_HW_TAGS */

void kasan_poison(const void *address, size_t size, u8 value);
void kasan_unpoison(const void *address, size_t size);
-bool kasan_check_invalid_free(void *addr);
+bool kasan_byte_accessible(const void *addr);

#endif /* CONFIG_KASAN_HW_TAGS */

diff --git a/mm/kasan/sw_tags.c b/mm/kasan/sw_tags.c
index cc271fceb5d5..94c2d33be333 100644
--- a/mm/kasan/sw_tags.c
+++ b/mm/kasan/sw_tags.c

@@ -118,13 +118,13 @@ bool kasan_check_range(unsigned long addr, size_t size, bool write,
return true;
}

-bool kasan_check_invalid_free(void *addr)

+bool kasan_byte_accessible(const void *addr)

{
u8 tag = get_tag(addr);
u8 shadow_byte = READ_ONCE(*(u8 *)kasan_mem_to_shadow(kasan_reset_tag(addr)));

- return (shadow_byte == KASAN_TAG_INVALID) ||
- (tag != KASAN_TAG_KERNEL && tag != shadow_byte);
+ return (shadow_byte != KASAN_TAG_INVALID) &&
+ (tag == KASAN_TAG_KERNEL || tag == shadow_byte);
}

#define DEFINE_HWASAN_LOAD_STORE(size) \
diff --git a/mm/slab_common.c b/mm/slab_common.c

index e981c80d216c..9c12cf4212ea 100644
--- a/mm/slab_common.c
+++ b/mm/slab_common.c
@@ -1157,19 +1157,21 @@ size_t ksize(const void *objp)

size_t size;

/*
- * We need to check that the pointed to object is valid, and only then
- * unpoison the shadow memory below. We use __kasan_check_read(), to
- * generate a more useful report at the time ksize() is called (rather
- * than later where behaviour is undefined due to potential
- * use-after-free or double-free).
+ * We need to first check that the pointer to the object is valid, and
+ * only then unpoison the memory. The report printed from ksize() is
+ * more useful, then when it's printed later when the behaviour could
+ * be undefined due to a potential use-after-free or double-free.

*
- * If the pointed to memory is invalid we return 0, to avoid users of
+ * We use kasan_check_byte(), which is supported for the hardware
+ * tag-based KASAN mode, unlike kasan_check_read/write().
+ *
+ * If the pointed to memory is invalid, we return 0 to avoid users of

* ksize() writing to and potentially corrupting the memory region.

*

* We want to perform the check before __ksize(), to avoid potentially
* crashing in __ksize() due to accessing invalid metadata.
*/
- if (unlikely(ZERO_OR_NULL_PTR(objp)) || !__kasan_check_read(objp, 1))

+ if (unlikely(ZERO_OR_NULL_PTR(objp)) || !kasan_check_byte(objp))

Andrey Konovalov

unread,

Jan 14, 2021, 2:37:05 PM1/14/21

to Andrew Morton, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

The currently existing page allocator tests rely on kmalloc fallback
with large sizes that is only present for SLUB. Add proper tests that
use alloc/free_pages().

Link: https://linux-review.googlesource.com/id/Ia173d5a1b215fe6b2548d814ef0f4433cf983570

Reviewed-by: Marco Elver <el...@google.com>
Reviewed-by: Alexander Potapenko <gli...@google.com>

Signed-off-by: Andrey Konovalov <andre...@google.com>
---

lib/test_kasan.c | 51 +++++++++++++++++++++++++++++++++++++++++++-----
1 file changed, 46 insertions(+), 5 deletions(-)

diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index 566d894ba20b..ab22a653762e 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c
@@ -147,6 +147,12 @@ static void kmalloc_node_oob_right(struct kunit *test)
kfree(ptr);
}

+/*
+ * These kmalloc_pagealloc_* tests try allocating a memory chunk that doesn't
+ * fit into a slab cache and therefore is allocated via the page allocator
+ * fallback. Since this kind of fallback is only implemented for SLUB, these
+ * tests are limited to that allocator.
+ */
static void kmalloc_pagealloc_oob_right(struct kunit *test)
{
char *ptr;
@@ -154,14 +160,11 @@ static void kmalloc_pagealloc_oob_right(struct kunit *test)

KASAN_TEST_NEEDS_CONFIG_ON(test, CONFIG_SLUB);

- /*
- * Allocate a chunk that does not fit into a SLUB cache to trigger
- * the page allocator fallback.
- */

ptr = kmalloc(size, GFP_KERNEL);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);

KUNIT_EXPECT_KASAN_FAIL(test, ptr[size + OOB_TAG_OFF] = 0);
+
kfree(ptr);
}

@@ -174,8 +177,8 @@ static void kmalloc_pagealloc_uaf(struct kunit *test)

ptr = kmalloc(size, GFP_KERNEL);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);

-
kfree(ptr);
+
KUNIT_EXPECT_KASAN_FAIL(test, ptr[0] = 0);
}

@@ -192,6 +195,42 @@ static void kmalloc_pagealloc_invalid_free(struct kunit *test)
KUNIT_EXPECT_KASAN_FAIL(test, kfree(ptr + 1));
}

+static void pagealloc_oob_right(struct kunit *test)
+{
+ char *ptr;
+ struct page *pages;
+ size_t order = 4;
+ size_t size = (1UL << (PAGE_SHIFT + order));
+
+ /*
+ * With generic KASAN page allocations have no redzones, thus
+ * out-of-bounds detection is not guaranteed.
+ * See https://bugzilla.kernel.org/show_bug.cgi?id=210503.
+ */
+ KASAN_TEST_NEEDS_CONFIG_OFF(test, CONFIG_KASAN_GENERIC);
+

+ pages = alloc_pages(GFP_KERNEL, order);
+ ptr = page_address(pages);
+ KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);
+

+ KUNIT_EXPECT_KASAN_FAIL(test, ptr[size] = 0);

+ free_pages((unsigned long)ptr, order);
+}
+

+static void pagealloc_uaf(struct kunit *test)
+{
+ char *ptr;
+ struct page *pages;
+ size_t order = 4;
+

+ pages = alloc_pages(GFP_KERNEL, order);
+ ptr = page_address(pages);
+ KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ptr);

+ free_pages((unsigned long)ptr, order);
+

+ KUNIT_EXPECT_KASAN_FAIL(test, ptr[0] = 0);
+}
+
static void kmalloc_large_oob_right(struct kunit *test)
{
char *ptr;
@@ -903,6 +942,8 @@ static struct kunit_case kasan_kunit_test_cases[] = {
KUNIT_CASE(kmalloc_pagealloc_oob_right),
KUNIT_CASE(kmalloc_pagealloc_uaf),
KUNIT_CASE(kmalloc_pagealloc_invalid_free),
+ KUNIT_CASE(pagealloc_oob_right),
+ KUNIT_CASE(pagealloc_uaf),
KUNIT_CASE(kmalloc_large_oob_right),
KUNIT_CASE(kmalloc_oob_krealloc_more),
KUNIT_CASE(kmalloc_oob_krealloc_less),
--
2.30.0.284.gd98b1dd5eaa7-goog

Andrey Konovalov

unread,

Jan 14, 2021, 2:37:08 PM1/14/21

to Andrew Morton, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

Add a test for kmem_cache_alloc/free_bulk to make sure there are no

false-positives when these functions are used.

Link: https://linux-review.googlesource.com/id/I2a8bf797aecf81baeac61380c567308f319e263d

Signed-off-by: Andrey Konovalov <andre...@google.com>
---

lib/test_kasan.c | 38 +++++++++++++++++++++++++++++++++-----
1 file changed, 33 insertions(+), 5 deletions(-)

diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index ab22a653762e..a96376aa7293 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c

@@ -479,10 +479,11 @@ static void kmem_cache_oob(struct kunit *test)
{
char *p;
size_t size = 200;
- struct kmem_cache *cache = kmem_cache_create("test_cache",
- size, 0,
- 0, NULL);
+ struct kmem_cache *cache;
+
+ cache = kmem_cache_create("test_cache", size, 0, 0, NULL);
KUNIT_ASSERT_NOT_ERR_OR_NULL(test, cache);
+
p = kmem_cache_alloc(cache, GFP_KERNEL);
if (!p) {
kunit_err(test, "Allocation failed: %s\n", __func__);
@@ -491,11 +492,12 @@ static void kmem_cache_oob(struct kunit *test)
}

KUNIT_EXPECT_KASAN_FAIL(test, *p = p[size + OOB_TAG_OFF]);
+
kmem_cache_free(cache, p);
kmem_cache_destroy(cache);
}

-static void memcg_accounted_kmem_cache(struct kunit *test)
+static void kmem_cache_accounted(struct kunit *test)
{
int i;
char *p;

@@ -522,6 +524,31 @@ static void memcg_accounted_kmem_cache(struct kunit *test)

kmem_cache_destroy(cache);
}

+static void kmem_cache_bulk(struct kunit *test)
+{
+ struct kmem_cache *cache;
+ size_t size = 200;

+ char *p[10];
+ bool ret;
+ int i;
+
+ cache = kmem_cache_create("test_cache", size, 0, 0, NULL);
+ KUNIT_ASSERT_NOT_ERR_OR_NULL(test, cache);
+

+ ret = kmem_cache_alloc_bulk(cache, GFP_KERNEL, ARRAY_SIZE(p), (void **)&p);

+ if (!ret) {
+ kunit_err(test, "Allocation failed: %s\n", __func__);
+ kmem_cache_destroy(cache);
+ return;

+ }
+
+ for (i = 0; i < ARRAY_SIZE(p); i++)

+ p[i][0] = p[i][size - 1] = 42;
+

+ kmem_cache_free_bulk(cache, ARRAY_SIZE(p), (void **)&p);

+ kmem_cache_destroy(cache);
+}
+
static char global_array[10];

static void kasan_global_oob(struct kunit *test)

@@ -961,7 +988,8 @@ static struct kunit_case kasan_kunit_test_cases[] = {

Andrey Konovalov

unread,

Jan 14, 2021, 2:37:10 PM1/14/21

to Andrew Morton, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasa...@googlegroups.com, linux-ar...@lists.infradead.org, linu...@kvack.org, linux-...@vger.kernel.org, Andrey Konovalov

Don't run KASAN tests when it's disabled with kasan.mode=off to avoid
corrupting kernel memory.

Link: https://linux-review.googlesource.com/id/I6447af436a69a94bfc35477f6bf4e2122948355e

Signed-off-by: Andrey Konovalov <andre...@google.com>
---

lib/test_kasan.c | 5 +++++
1 file changed, 5 insertions(+)

diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index a96376aa7293..6238b56127f8 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c
@@ -47,6 +47,11 @@ static bool multishot;

*/
static int kasan_test_init(struct kunit *test)
{

+ if (!kasan_enabled()) {
+ kunit_err(test, "can't run KASAN tests with KASAN disabled");
+ return -1;
+ }
+

Marco Elver

unread,

Jan 14, 2021, 2:56:33 PM1/14/21

to Andrey Konovalov, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Andrew Morton, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

Hmm, if it makes too much of a mess, let's just leave as-is.

Thanks,
-- Marco

Andrey Konovalov

unread,

Jan 15, 2021, 8:11:50 AM1/15/21

to Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Alexander Potapenko, Marco Elver, Andrew Morton, Will Deacon, Andrey Ryabinin, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

On Tue, Jan 12, 2021 at 8:01 PM Catalin Marinas <catalin...@arm.com> wrote:

>
> On Tue, Jan 05, 2021 at 07:27:49PM +0100, Andrey Konovalov wrote:
> > diff --git a/arch/arm64/mm/fault.c b/arch/arm64/mm/fault.c

> > index 3c40da479899..57d3f165d907 100644
> > --- a/arch/arm64/mm/fault.c
> > +++ b/arch/arm64/mm/fault.c
> > @@ -302,12 +302,20 @@ static void die_kernel_fault(const char *msg, unsigned long addr,

> > static void report_tag_fault(unsigned long addr, unsigned int esr,
> > struct pt_regs *regs)
> > {
> > - bool is_write = ((esr & ESR_ELx_WNR) >> ESR_ELx_WNR_SHIFT) != 0;
> > + static bool reported;
> > + bool is_write;
> > +
> > + if (READ_ONCE(reported))
> > + return;
> > +
> > + if (mte_report_once())
> > + WRITE_ONCE(reported, true);
>

> I guess the assumption here is that you don't get any report before the
> tests start and temporarily set report_once to false. It's probably
> fine, if we get a tag check failure we'd notice in the logs anyway.

Good point. I'll add a note in a comment in v4.

> > /*

> > * SAS bits aren't set for all faults reported in EL1, so we can't
> > * find out access size.
> > */
> > + is_write = ((esr & ESR_ELx_WNR) >> ESR_ELx_WNR_SHIFT) != 0;
>

> I now noticed, you could write this in a shorter way:
>
> is_write = !!(esr & ESR_ELx_WNR);

>
> > kasan_report(addr, 0, is_write, regs->pc);
> > }

Will do in v4.

> The patch looks fine to me.
>
> Reviewed-by: Catalin Marinas <catalin...@arm.com>

Thanks!

unread,

Jan 15, 2021, 8:26:19 AM1/15/21

to Andrey Konovalov, Andrew Morton, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Marco Elver, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

On Thu, Jan 14, 2021 at 8:36 PM Andrey Konovalov <andre...@google.com> wrote:
>
> Clarify and update comments in KASAN tests.
>
> Link: https://linux-review.googlesource.com/id/I6c816c51fa1e0eb7aa3dead6bda1f339d2af46c8
> Reviewed-by: Marco Elver <el...@google.com>
> Signed-off-by: Andrey Konovalov <andre...@google.com>

Reviewed-by: Alexander Potapenko <gli...@google.com>

Alexander Potapenko

unread,

Jan 15, 2021, 8:33:17 AM1/15/21

to Andrey Konovalov, Andrew Morton, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Marco Elver, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

On Thu, Jan 14, 2021 at 8:36 PM Andrey Konovalov <andre...@google.com> wrote:
>

> Add 3 new tests for tag-based KASAN modes:
>
> 1. Check that match-all pointer tag is not assigned randomly.
> 2. Check that 0xff works as a match-all pointer tag.
> 3. Check that there are no match-all memory tags.
>
> Note, that test #3 causes a significant number (255) of KASAN reports
> to be printed during execution for the SW_TAGS mode.
>
> Link: https://linux-review.googlesource.com/id/I78f1375efafa162b37f3abcb2c5bc2f3955dfd8e
> Reviewed-by: Marco Elver <el...@google.com>
> Signed-off-by: Andrey Konovalov <andre...@google.com>

Reviewed-by: Alexander Potapenko <gli...@google.com>

Alexander Potapenko

unread,

Jan 15, 2021, 8:44:58 AM1/15/21

to Marco Elver, Andrey Konovalov, Andrew Morton, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

On Fri, Jan 15, 2021 at 2:13 PM Marco Elver <el...@google.com> wrote:
>
> On Thu, Jan 14, 2021 at 08:36PM +0100, Andrey Konovalov wrote:
> > Don't run KASAN tests when it's disabled with kasan.mode=off to avoid
> > corrupting kernel memory.
> >
> > Link: https://linux-review.googlesource.com/id/I6447af436a69a94bfc35477f6bf4e2122948355e
> > Signed-off-by: Andrey Konovalov <andre...@google.com>
>
> Reviewed-by: Marco Elver <el...@google.com>

Reviewed-by: Alexander Potapenko <gli...@google.com>

Alexander Potapenko

unread,

Jan 15, 2021, 8:49:37 AM1/15/21

to Marco Elver, Andrey Konovalov, Andrew Morton, Catalin Marinas, Vincenzo Frascino, Dmitry Vyukov, Will Deacon, Andrey Ryabinin, Peter Collingbourne, Evgenii Stepanov, Branislav Rankov, Kevin Brodsky, kasan-dev, Linux ARM, Linux Memory Management List, LKML

On Fri, Jan 15, 2021 at 2:16 PM Marco Elver <el...@google.com> wrote:
>
> On Thu, Jan 14, 2021 at 08:36PM +0100, Andrey Konovalov wrote:
> > Add a test for kmem_cache_alloc/free_bulk to make sure there are no
> > false-positives when these functions are used.
> >
> > Link: https://linux-review.googlesource.com/id/I2a8bf797aecf81baeac61380c567308f319e263d
> > Signed-off-by: Andrey Konovalov <andre...@google.com>
>
> Reviewed-by: Marco Elver <el...@google.com>

Reviewed-by: Alexander Potapenko <gli...@google.com>

(see a nit below)

> > + cache = kmem_cache_create("test_cache", size, 0, 0, NULL);

Looks like there's a tab between "test_cache" and size, please double-check.