linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
* [PATCH 0/4] kasan: add clang support
@ 2017-07-06 22:01 Greg Hackmann
  2017-07-06 22:01 ` [PATCH 1/4] kasan: support alloca() poisoning Greg Hackmann
                   ` (3 more replies)
  0 siblings, 4 replies; 17+ messages in thread
From: Greg Hackmann @ 2017-07-06 22:01 UTC (permalink / raw)
  To: Andrey Ryabinin, Alexander Potapenko, Dmitry Vyukov,
	Masahiro Yamada, Michal Marek
  Cc: linux-kernel, kasan-dev, linux-mm, linux-kbuild,
	Matthias Kaehlcke, Michael Davidson, Greg Hackmann

This patch series adds support for building KASAN-enabled kernels with clang.
This mostly involves adding callbacks for a couple of new features in LLVM's
AddressSanitizer implementation.  We also need to probe for the (slightly
different) CFLAGS used to configure ASAN with clang.

*** BLURB HERE ***

Alexander Potapenko (1):
  kasan: added functions for unpoisoning stack variables

Greg Hackmann (3):
  kasan: support alloca() poisoning
  kasan: support LLVM-style asan parameters
  kasan: add compiler support for clang

 include/linux/compiler-clang.h | 10 ++++++++++
 lib/test_kasan.c               | 22 ++++++++++++++++++++++
 mm/kasan/kasan.c               | 41 +++++++++++++++++++++++++++++++++++++++++
 mm/kasan/kasan.h               |  8 ++++++++
 mm/kasan/report.c              |  3 +++
 scripts/Makefile.kasan         | 10 +++++++++-
 6 files changed, 93 insertions(+), 1 deletion(-)

-- 
2.13.2.725.g09c95d1e9-goog

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

^ permalink raw reply	[flat|nested] 17+ messages in thread

* [PATCH 1/4] kasan: support alloca() poisoning
  2017-07-06 22:01 [PATCH 0/4] kasan: add clang support Greg Hackmann
@ 2017-07-06 22:01 ` Greg Hackmann
  2017-07-07  0:09   ` Greg Hackmann
                     ` (2 more replies)
  2017-07-06 22:01 ` [PATCH 2/4] kasan: added functions for unpoisoning stack variables Greg Hackmann
                   ` (2 subsequent siblings)
  3 siblings, 3 replies; 17+ messages in thread
From: Greg Hackmann @ 2017-07-06 22:01 UTC (permalink / raw)
  To: Andrey Ryabinin, Alexander Potapenko, Dmitry Vyukov,
	Masahiro Yamada, Michal Marek
  Cc: linux-kernel, kasan-dev, linux-mm, linux-kbuild,
	Matthias Kaehlcke, Michael Davidson, Greg Hackmann

clang's AddressSanitizer implementation adds redzones on either side of
alloca()ed buffers.  These redzones are 32-byte aligned and at least 32
bytes long.

__asan_alloca_poison() is passed the size and address of the allocated
buffer, *excluding* the redzones on either side.  The left redzone will
always be to the immediate left of this buffer; but AddressSanitizer may
need to add padding between the end of the buffer and the right redzone.
If there are any 8-byte chunks inside this padding, we should poison
those too.

__asan_allocas_unpoison() is just passed the top and bottom of the
dynamic stack area, so unpoisoning is simpler.

Signed-off-by: Greg Hackmann <ghackmann@google.com>
---
 lib/test_kasan.c  | 22 ++++++++++++++++++++++
 mm/kasan/kasan.c  | 26 ++++++++++++++++++++++++++
 mm/kasan/kasan.h  |  8 ++++++++
 mm/kasan/report.c |  3 +++
 4 files changed, 59 insertions(+)

diff --git a/lib/test_kasan.c b/lib/test_kasan.c
index a25c9763fce1..f774fcafb696 100644
--- a/lib/test_kasan.c
+++ b/lib/test_kasan.c
@@ -473,6 +473,26 @@ static noinline void __init use_after_scope_test(void)
 	p[1023] = 1;
 }
 
+static noinline void __init kasan_alloca_oob_left(void)
+{
+	volatile int i = 10;
+	char alloca_array[i];
+	char *p = alloca_array - 1;
+
+	pr_info("out-of-bounds to left on alloca\n");
+	*(volatile char *)p;
+}
+
+static noinline void __init kasan_alloca_oob_right(void)
+{
+	volatile int i = 10;
+	char alloca_array[i];
+	char *p = alloca_array + round_up(i, 8);
+
+	pr_info("out-of-bounds to right on alloca\n");
+	*(volatile char *)p;
+}
+
 static int __init kmalloc_tests_init(void)
 {
 	/*
@@ -503,6 +523,8 @@ static int __init kmalloc_tests_init(void)
 	memcg_accounted_kmem_cache();
 	kasan_stack_oob();
 	kasan_global_oob();
+	kasan_alloca_oob_left();
+	kasan_alloca_oob_right();
 	ksize_unpoisons_memory();
 	copy_user_test();
 	use_after_scope_test();
diff --git a/mm/kasan/kasan.c b/mm/kasan/kasan.c
index c81549d5c833..892b626f564b 100644
--- a/mm/kasan/kasan.c
+++ b/mm/kasan/kasan.c
@@ -802,6 +802,32 @@ void __asan_unpoison_stack_memory(const void *addr, size_t size)
 }
 EXPORT_SYMBOL(__asan_unpoison_stack_memory);
 
+/* Emitted by compiler to poison alloca()ed objects. */
+void __asan_alloca_poison(unsigned long addr, size_t size)
+{
+	size_t rounded_up_size = round_up(size, KASAN_SHADOW_SCALE_SIZE);
+	size_t padding_size = round_up(size, KASAN_ALLOCA_REDZONE_SIZE) -
+			round_up(size, KASAN_SHADOW_SCALE_SIZE);
+
+	const void *left_redzone = (const void *)(addr -
+			KASAN_ALLOCA_REDZONE_SIZE);
+	const void *right_redzone = (const void *)(addr + rounded_up_size);
+
+	kasan_poison_shadow(left_redzone, KASAN_ALLOCA_REDZONE_SIZE,
+			KASAN_ALLOCA_LEFT);
+	kasan_poison_shadow(right_redzone,
+			padding_size + KASAN_ALLOCA_REDZONE_SIZE,
+			KASAN_ALLOCA_RIGHT);
+}
+EXPORT_SYMBOL(__asan_alloca_poison);
+
+/* Emitted by compiler to unpoison alloca()ed areas when the stack unwinds. */
+void __asan_allocas_unpoison(const void *stack_top, const void *stack_bottom)
+{
+	kasan_unpoison_shadow(stack_top, stack_bottom - stack_top);
+}
+EXPORT_SYMBOL(__asan_allocas_unpoison);
+
 #ifdef CONFIG_MEMORY_HOTPLUG
 static int kasan_mem_notifier(struct notifier_block *nb,
 			unsigned long action, void *data)
diff --git a/mm/kasan/kasan.h b/mm/kasan/kasan.h
index 1229298cce64..b857dc70d6a2 100644
--- a/mm/kasan/kasan.h
+++ b/mm/kasan/kasan.h
@@ -23,6 +23,14 @@
 #define KASAN_STACK_PARTIAL     0xF4
 #define KASAN_USE_AFTER_SCOPE   0xF8
 
+/*
+ * alloca redzone shadow values
+ */
+#define KASAN_ALLOCA_LEFT	0xCA
+#define KASAN_ALLOCA_RIGHT	0xCB
+
+#define KASAN_ALLOCA_REDZONE_SIZE	32
+
 /* Don't break randconfig/all*config builds */
 #ifndef KASAN_ABI_VERSION
 #define KASAN_ABI_VERSION 1
diff --git a/mm/kasan/report.c b/mm/kasan/report.c
index beee0e980e2d..c6a5b7ab9e3a 100644
--- a/mm/kasan/report.c
+++ b/mm/kasan/report.c
@@ -101,6 +101,9 @@ static const char *get_shadow_bug_type(struct kasan_access_info *info)
 		break;
 	case KASAN_USE_AFTER_SCOPE:
 		bug_type = "use-after-scope";
+	case KASAN_ALLOCA_LEFT:
+	case KASAN_ALLOCA_RIGHT:
+		bug_type = "alloca-out-of-bounds";
 		break;
 	}
 
-- 
2.13.2.725.g09c95d1e9-goog

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

^ permalink raw reply related	[flat|nested] 17+ messages in thread

* [PATCH 2/4] kasan: added functions for unpoisoning stack variables
  2017-07-06 22:01 [PATCH 0/4] kasan: add clang support Greg Hackmann
  2017-07-06 22:01 ` [PATCH 1/4] kasan: support alloca() poisoning Greg Hackmann
@ 2017-07-06 22:01 ` Greg Hackmann
  2017-07-10  8:46   ` Dmitry Vyukov
  2017-07-10 10:31   ` Andrey Ryabinin
  2017-07-06 22:01 ` [PATCH 3/4] kasan: support LLVM-style asan parameters Greg Hackmann
  2017-07-06 22:01 ` [PATCH 4/4] kasan: add compiler support for clang Greg Hackmann
  3 siblings, 2 replies; 17+ messages in thread
From: Greg Hackmann @ 2017-07-06 22:01 UTC (permalink / raw)
  To: Andrey Ryabinin, Alexander Potapenko, Dmitry Vyukov,
	Masahiro Yamada, Michal Marek
  Cc: linux-kernel, kasan-dev, linux-mm, linux-kbuild,
	Matthias Kaehlcke, Michael Davidson, Greg Hackmann

From: Alexander Potapenko <glider@google.com>

As a code-size optimization, LLVM builds since r279383 may
bulk-manipulate the shadow region when (un)poisoning large memory
blocks.  This requires new callbacks that simply do an uninstrumented
memset().

This fixes linking the Clang-built kernel when using KASAN.

Signed-off-by: Alexander Potapenko <glider@google.com>
[ghackmann@google.com: fix memset() parameters, and tweak
 commit message to describe new callbacks]
Signed-off-by: Greg Hackmann <ghackmann@google.com>
---
 mm/kasan/kasan.c | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/mm/kasan/kasan.c b/mm/kasan/kasan.c
index 892b626f564b..89911e5c69f9 100644
--- a/mm/kasan/kasan.c
+++ b/mm/kasan/kasan.c
@@ -828,6 +828,21 @@ void __asan_allocas_unpoison(const void *stack_top, const void *stack_bottom)
 }
 EXPORT_SYMBOL(__asan_allocas_unpoison);
 
+/* Emitted by the compiler to [un]poison local variables. */
+#define DEFINE_ASAN_SET_SHADOW(byte) \
+	void __asan_set_shadow_##byte(const void *addr, size_t size)	\
+	{								\
+		__memset((void *)addr, 0x##byte, size);			\
+	}								\
+	EXPORT_SYMBOL(__asan_set_shadow_##byte)
+
+DEFINE_ASAN_SET_SHADOW(00);
+DEFINE_ASAN_SET_SHADOW(f1);
+DEFINE_ASAN_SET_SHADOW(f2);
+DEFINE_ASAN_SET_SHADOW(f3);
+DEFINE_ASAN_SET_SHADOW(f5);
+DEFINE_ASAN_SET_SHADOW(f8);
+
 #ifdef CONFIG_MEMORY_HOTPLUG
 static int kasan_mem_notifier(struct notifier_block *nb,
 			unsigned long action, void *data)
-- 
2.13.2.725.g09c95d1e9-goog

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

^ permalink raw reply related	[flat|nested] 17+ messages in thread

* [PATCH 3/4] kasan: support LLVM-style asan parameters
  2017-07-06 22:01 [PATCH 0/4] kasan: add clang support Greg Hackmann
  2017-07-06 22:01 ` [PATCH 1/4] kasan: support alloca() poisoning Greg Hackmann
  2017-07-06 22:01 ` [PATCH 2/4] kasan: added functions for unpoisoning stack variables Greg Hackmann
@ 2017-07-06 22:01 ` Greg Hackmann
  2017-07-10  8:47   ` Dmitry Vyukov
  2017-07-06 22:01 ` [PATCH 4/4] kasan: add compiler support for clang Greg Hackmann
  3 siblings, 1 reply; 17+ messages in thread
From: Greg Hackmann @ 2017-07-06 22:01 UTC (permalink / raw)
  To: Andrey Ryabinin, Alexander Potapenko, Dmitry Vyukov,
	Masahiro Yamada, Michal Marek
  Cc: linux-kernel, kasan-dev, linux-mm, linux-kbuild,
	Matthias Kaehlcke, Michael Davidson, Greg Hackmann

Use cc-option to figure out whether the compiler's sanitizer uses
LLVM-style parameters ("-mllvm -asan-foo=bar") or GCC-style parameters
("--param asan-foo=bar").

Signed-off-by: Greg Hackmann <ghackmann@google.com>
---
 scripts/Makefile.kasan | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/scripts/Makefile.kasan b/scripts/Makefile.kasan
index 9576775a86f6..b66ae4b4546b 100644
--- a/scripts/Makefile.kasan
+++ b/scripts/Makefile.kasan
@@ -9,11 +9,19 @@ KASAN_SHADOW_OFFSET ?= $(CONFIG_KASAN_SHADOW_OFFSET)
 
 CFLAGS_KASAN_MINIMAL := -fsanitize=kernel-address
 
-CFLAGS_KASAN := $(call cc-option, -fsanitize=kernel-address \
+CFLAGS_KASAN_GCC := $(call cc-option, -fsanitize=kernel-address \
 		-fasan-shadow-offset=$(KASAN_SHADOW_OFFSET) \
 		--param asan-stack=1 --param asan-globals=1 \
 		--param asan-instrumentation-with-call-threshold=$(call_threshold))
 
+CFLAGS_KASAN_LLVM := $(call cc-option, -fsanitize=kernel-address \
+		-mllvm -asan-mapping-offset=$(KASAN_SHADOW_OFFSET) \
+		-mllvm -asan-stack=1 -mllvm -asan-globals=1 \
+		-mllvm -asan-use-after-scope=1 \
+		-mllvm -asan-instrumentation-with-call-threshold=$(call_threshold))
+
+CFLAGS_KASAN := $(CFLAGS_KASAN_GCC) $(CFLAGS_KASAN_LLVM)
+
 ifeq ($(call cc-option, $(CFLAGS_KASAN_MINIMAL) -Werror),)
    ifneq ($(CONFIG_COMPILE_TEST),y)
         $(warning Cannot use CONFIG_KASAN: \
-- 
2.13.2.725.g09c95d1e9-goog

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

^ permalink raw reply related	[flat|nested] 17+ messages in thread

* [PATCH 4/4] kasan: add compiler support for clang
  2017-07-06 22:01 [PATCH 0/4] kasan: add clang support Greg Hackmann
                   ` (2 preceding siblings ...)
  2017-07-06 22:01 ` [PATCH 3/4] kasan: support LLVM-style asan parameters Greg Hackmann
@ 2017-07-06 22:01 ` Greg Hackmann
  2017-07-10  8:48   ` Dmitry Vyukov
  2017-07-10 10:34   ` Andrey Ryabinin
  3 siblings, 2 replies; 17+ messages in thread
From: Greg Hackmann @ 2017-07-06 22:01 UTC (permalink / raw)
  To: Andrey Ryabinin, Alexander Potapenko, Dmitry Vyukov,
	Masahiro Yamada, Michal Marek
  Cc: linux-kernel, kasan-dev, linux-mm, linux-kbuild,
	Matthias Kaehlcke, Michael Davidson, Greg Hackmann

For now we can hard-code ASAN ABI level 5, since historical clang builds
can't build the kernel anyway.  We also need to emulate gcc's
__SANITIZE_ADDRESS__ flag, or memset() calls won't be instrumented.

Signed-off-by: Greg Hackmann <ghackmann@google.com>
---
 include/linux/compiler-clang.h | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/include/linux/compiler-clang.h b/include/linux/compiler-clang.h
index d614c5ea1b5e..8153f793b22a 100644
--- a/include/linux/compiler-clang.h
+++ b/include/linux/compiler-clang.h
@@ -23,3 +23,13 @@
  */
 #undef inline
 #define inline inline __attribute__((unused)) notrace
+
+/* all clang versions usable with the kernel support KASAN ABI version 5
+ */
+#define KASAN_ABI_VERSION 5
+
+/* emulate gcc's __SANITIZE_ADDRESS__ flag
+ */
+#if __has_feature(address_sanitizer)
+#define __SANITIZE_ADDRESS__
+#endif
-- 
2.13.2.725.g09c95d1e9-goog

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

^ permalink raw reply related	[flat|nested] 17+ messages in thread

* Re: [PATCH 1/4] kasan: support alloca() poisoning
  2017-07-06 22:01 ` [PATCH 1/4] kasan: support alloca() poisoning Greg Hackmann
@ 2017-07-07  0:09   ` Greg Hackmann
  2017-07-10  8:44   ` Dmitry Vyukov
  2017-07-10 10:30   ` Andrey Ryabinin
  2 siblings, 0 replies; 17+ messages in thread
From: Greg Hackmann @ 2017-07-07  0:09 UTC (permalink / raw)
  To: Andrey Ryabinin, Alexander Potapenko, Dmitry Vyukov,
	Masahiro Yamada, Michal Marek
  Cc: linux-kernel, kasan-dev, linux-mm, linux-kbuild,
	Matthias Kaehlcke, Michael Davidson

On 07/06/2017 03:01 PM, Greg Hackmann wrote:
> @@ -101,6 +101,9 @@ static const char *get_shadow_bug_type(struct kasan_access_info *info)
>   		break;
>   	case KASAN_USE_AFTER_SCOPE:
>   		bug_type = "use-after-scope";
> +	case KASAN_ALLOCA_LEFT:
> +	case KASAN_ALLOCA_RIGHT:
> +		bug_type = "alloca-out-of-bounds";
>   		break;
>   	}

There needs to be a "break" above the new case statements.  I'll wait to 
see if there's any other feedback, then send out a V2 patch that fixes this.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

^ permalink raw reply	[flat|nested] 17+ messages in thread

* Re: [PATCH 1/4] kasan: support alloca() poisoning
  2017-07-06 22:01 ` [PATCH 1/4] kasan: support alloca() poisoning Greg Hackmann
  2017-07-07  0:09   ` Greg Hackmann
@ 2017-07-10  8:44   ` Dmitry Vyukov
  2017-07-13 22:40     ` Greg Hackmann
  2017-07-10 10:30   ` Andrey Ryabinin
  2 siblings, 1 reply; 17+ messages in thread
From: Dmitry Vyukov @ 2017-07-10  8:44 UTC (permalink / raw)
  To: Greg Hackmann
  Cc: Andrey Ryabinin, Alexander Potapenko, Masahiro Yamada,
	Michal Marek, LKML, kasan-dev, linux-mm,
	open list:KERNEL BUILD + fi...,
	Matthias Kaehlcke, Michael Davidson

On Fri, Jul 7, 2017 at 12:01 AM, Greg Hackmann <ghackmann@google.com> wrote:
> clang's AddressSanitizer implementation adds redzones on either side of
> alloca()ed buffers.  These redzones are 32-byte aligned and at least 32
> bytes long.
>
> __asan_alloca_poison() is passed the size and address of the allocated
> buffer, *excluding* the redzones on either side.  The left redzone will
> always be to the immediate left of this buffer; but AddressSanitizer may
> need to add padding between the end of the buffer and the right redzone.
> If there are any 8-byte chunks inside this padding, we should poison
> those too.
>
> __asan_allocas_unpoison() is just passed the top and bottom of the
> dynamic stack area, so unpoisoning is simpler.
>
> Signed-off-by: Greg Hackmann <ghackmann@google.com>
> ---
>  lib/test_kasan.c  | 22 ++++++++++++++++++++++
>  mm/kasan/kasan.c  | 26 ++++++++++++++++++++++++++
>  mm/kasan/kasan.h  |  8 ++++++++
>  mm/kasan/report.c |  3 +++
>  4 files changed, 59 insertions(+)
>
> diff --git a/lib/test_kasan.c b/lib/test_kasan.c
> index a25c9763fce1..f774fcafb696 100644
> --- a/lib/test_kasan.c
> +++ b/lib/test_kasan.c
> @@ -473,6 +473,26 @@ static noinline void __init use_after_scope_test(void)
>         p[1023] = 1;
>  }
>
> +static noinline void __init kasan_alloca_oob_left(void)
> +{
> +       volatile int i = 10;
> +       char alloca_array[i];
> +       char *p = alloca_array - 1;
> +
> +       pr_info("out-of-bounds to left on alloca\n");
> +       *(volatile char *)p;
> +}
> +
> +static noinline void __init kasan_alloca_oob_right(void)
> +{
> +       volatile int i = 10;
> +       char alloca_array[i];
> +       char *p = alloca_array + round_up(i, 8);
> +
> +       pr_info("out-of-bounds to right on alloca\n");
> +       *(volatile char *)p;
> +}
> +
>  static int __init kmalloc_tests_init(void)
>  {
>         /*
> @@ -503,6 +523,8 @@ static int __init kmalloc_tests_init(void)
>         memcg_accounted_kmem_cache();
>         kasan_stack_oob();
>         kasan_global_oob();
> +       kasan_alloca_oob_left();
> +       kasan_alloca_oob_right();
>         ksize_unpoisons_memory();
>         copy_user_test();
>         use_after_scope_test();
> diff --git a/mm/kasan/kasan.c b/mm/kasan/kasan.c
> index c81549d5c833..892b626f564b 100644
> --- a/mm/kasan/kasan.c
> +++ b/mm/kasan/kasan.c
> @@ -802,6 +802,32 @@ void __asan_unpoison_stack_memory(const void *addr, size_t size)
>  }
>  EXPORT_SYMBOL(__asan_unpoison_stack_memory);
>
> +/* Emitted by compiler to poison alloca()ed objects. */
> +void __asan_alloca_poison(unsigned long addr, size_t size)
> +{
> +       size_t rounded_up_size = round_up(size, KASAN_SHADOW_SCALE_SIZE);
> +       size_t padding_size = round_up(size, KASAN_ALLOCA_REDZONE_SIZE) -
> +                       round_up(size, KASAN_SHADOW_SCALE_SIZE);

Perhaps s/round_up(size, KASAN_SHADOW_SCALE_SIZE)/rounded_up_size/
because we already calculated that.

> +
> +       const void *left_redzone = (const void *)(addr -
> +                       KASAN_ALLOCA_REDZONE_SIZE);
> +       const void *right_redzone = (const void *)(addr + rounded_up_size);

Please check that size is rounded to KASAN_ALLOCA_REDZONE_SIZE. That's
the expectation, right? That can change is clang silently.

> +       kasan_poison_shadow(left_redzone, KASAN_ALLOCA_REDZONE_SIZE,
> +                       KASAN_ALLOCA_LEFT);
> +       kasan_poison_shadow(right_redzone,
> +                       padding_size + KASAN_ALLOCA_REDZONE_SIZE,
> +                       KASAN_ALLOCA_RIGHT);

We also need to poison the unaligned part at the end of the object
from size to rounded_up_size. You can see how we do it for heap
objects.

> +}
> +EXPORT_SYMBOL(__asan_alloca_poison);
> +/* Emitted by compiler to unpoison alloca()ed areas when the stack unwinds. */
> +void __asan_allocas_unpoison(const void *stack_top, const void *stack_bottom)
> +{
> +       kasan_unpoison_shadow(stack_top, stack_bottom - stack_top);
> +}
> +EXPORT_SYMBOL(__asan_allocas_unpoison);
> +
>  #ifdef CONFIG_MEMORY_HOTPLUG
>  static int kasan_mem_notifier(struct notifier_block *nb,
>                         unsigned long action, void *data)
> diff --git a/mm/kasan/kasan.h b/mm/kasan/kasan.h
> index 1229298cce64..b857dc70d6a2 100644
> --- a/mm/kasan/kasan.h
> +++ b/mm/kasan/kasan.h
> @@ -23,6 +23,14 @@
>  #define KASAN_STACK_PARTIAL     0xF4
>  #define KASAN_USE_AFTER_SCOPE   0xF8
>
> +/*
> + * alloca redzone shadow values
> + */
> +#define KASAN_ALLOCA_LEFT      0xCA
> +#define KASAN_ALLOCA_RIGHT     0xCB
> +
> +#define KASAN_ALLOCA_REDZONE_SIZE      32
> +
>  /* Don't break randconfig/all*config builds */
>  #ifndef KASAN_ABI_VERSION
>  #define KASAN_ABI_VERSION 1
> diff --git a/mm/kasan/report.c b/mm/kasan/report.c
> index beee0e980e2d..c6a5b7ab9e3a 100644
> --- a/mm/kasan/report.c
> +++ b/mm/kasan/report.c
> @@ -101,6 +101,9 @@ static const char *get_shadow_bug_type(struct kasan_access_info *info)
>                 break;
>         case KASAN_USE_AFTER_SCOPE:
>                 bug_type = "use-after-scope";
> +       case KASAN_ALLOCA_LEFT:
> +       case KASAN_ALLOCA_RIGHT:
> +               bug_type = "alloca-out-of-bounds";
>                 break;
>         }
>
> --
> 2.13.2.725.g09c95d1e9-goog
>

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

^ permalink raw reply	[flat|nested] 17+ messages in thread

* Re: [PATCH 2/4] kasan: added functions for unpoisoning stack variables
  2017-07-06 22:01 ` [PATCH 2/4] kasan: added functions for unpoisoning stack variables Greg Hackmann
@ 2017-07-10  8:46   ` Dmitry Vyukov
  2017-07-10 10:31   ` Andrey Ryabinin
  1 sibling, 0 replies; 17+ messages in thread
From: Dmitry Vyukov @ 2017-07-10  8:46 UTC (permalink / raw)
  To: Greg Hackmann
  Cc: Andrey Ryabinin, Alexander Potapenko, Masahiro Yamada,
	Michal Marek, LKML, kasan-dev, linux-mm,
	open list:KERNEL BUILD + fi...,
	Matthias Kaehlcke, Michael Davidson

On Fri, Jul 7, 2017 at 12:01 AM, Greg Hackmann <ghackmann@google.com> wrote:
> From: Alexander Potapenko <glider@google.com>
>
> As a code-size optimization, LLVM builds since r279383 may
> bulk-manipulate the shadow region when (un)poisoning large memory
> blocks.  This requires new callbacks that simply do an uninstrumented
> memset().
>
> This fixes linking the Clang-built kernel when using KASAN.
>
> Signed-off-by: Alexander Potapenko <glider@google.com>
> [ghackmann@google.com: fix memset() parameters, and tweak
>  commit message to describe new callbacks]
> Signed-off-by: Greg Hackmann <ghackmann@google.com>
> ---
>  mm/kasan/kasan.c | 15 +++++++++++++++
>  1 file changed, 15 insertions(+)
>
> diff --git a/mm/kasan/kasan.c b/mm/kasan/kasan.c
> index 892b626f564b..89911e5c69f9 100644
> --- a/mm/kasan/kasan.c
> +++ b/mm/kasan/kasan.c
> @@ -828,6 +828,21 @@ void __asan_allocas_unpoison(const void *stack_top, const void *stack_bottom)
>  }
>  EXPORT_SYMBOL(__asan_allocas_unpoison);
>
> +/* Emitted by the compiler to [un]poison local variables. */
> +#define DEFINE_ASAN_SET_SHADOW(byte) \
> +       void __asan_set_shadow_##byte(const void *addr, size_t size)    \
> +       {                                                               \
> +               __memset((void *)addr, 0x##byte, size);                 \
> +       }                                                               \
> +       EXPORT_SYMBOL(__asan_set_shadow_##byte)
> +
> +DEFINE_ASAN_SET_SHADOW(00);
> +DEFINE_ASAN_SET_SHADOW(f1);
> +DEFINE_ASAN_SET_SHADOW(f2);
> +DEFINE_ASAN_SET_SHADOW(f3);
> +DEFINE_ASAN_SET_SHADOW(f5);
> +DEFINE_ASAN_SET_SHADOW(f8);
> +
>  #ifdef CONFIG_MEMORY_HOTPLUG
>  static int kasan_mem_notifier(struct notifier_block *nb,
>                         unsigned long action, void *data)

Reviewed-by: Dmitry Vyukov <dvyukov@google.com>

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

^ permalink raw reply	[flat|nested] 17+ messages in thread

* Re: [PATCH 3/4] kasan: support LLVM-style asan parameters
  2017-07-06 22:01 ` [PATCH 3/4] kasan: support LLVM-style asan parameters Greg Hackmann
@ 2017-07-10  8:47   ` Dmitry Vyukov
  0 siblings, 0 replies; 17+ messages in thread
From: Dmitry Vyukov @ 2017-07-10  8:47 UTC (permalink / raw)
  To: Greg Hackmann
  Cc: Andrey Ryabinin, Alexander Potapenko, Masahiro Yamada,
	Michal Marek, LKML, kasan-dev, linux-mm,
	open list:KERNEL BUILD + fi...,
	Matthias Kaehlcke, Michael Davidson

On Fri, Jul 7, 2017 at 12:01 AM, Greg Hackmann <ghackmann@google.com> wrote:
> Use cc-option to figure out whether the compiler's sanitizer uses
> LLVM-style parameters ("-mllvm -asan-foo=bar") or GCC-style parameters
> ("--param asan-foo=bar").
>
> Signed-off-by: Greg Hackmann <ghackmann@google.com>
> ---
>  scripts/Makefile.kasan | 10 +++++++++-
>  1 file changed, 9 insertions(+), 1 deletion(-)
>
> diff --git a/scripts/Makefile.kasan b/scripts/Makefile.kasan
> index 9576775a86f6..b66ae4b4546b 100644
> --- a/scripts/Makefile.kasan
> +++ b/scripts/Makefile.kasan
> @@ -9,11 +9,19 @@ KASAN_SHADOW_OFFSET ?= $(CONFIG_KASAN_SHADOW_OFFSET)
>
>  CFLAGS_KASAN_MINIMAL := -fsanitize=kernel-address
>
> -CFLAGS_KASAN := $(call cc-option, -fsanitize=kernel-address \
> +CFLAGS_KASAN_GCC := $(call cc-option, -fsanitize=kernel-address \
>                 -fasan-shadow-offset=$(KASAN_SHADOW_OFFSET) \
>                 --param asan-stack=1 --param asan-globals=1 \
>                 --param asan-instrumentation-with-call-threshold=$(call_threshold))
>
> +CFLAGS_KASAN_LLVM := $(call cc-option, -fsanitize=kernel-address \
> +               -mllvm -asan-mapping-offset=$(KASAN_SHADOW_OFFSET) \
> +               -mllvm -asan-stack=1 -mllvm -asan-globals=1 \
> +               -mllvm -asan-use-after-scope=1 \
> +               -mllvm -asan-instrumentation-with-call-threshold=$(call_threshold))
> +
> +CFLAGS_KASAN := $(CFLAGS_KASAN_GCC) $(CFLAGS_KASAN_LLVM)
> +
>  ifeq ($(call cc-option, $(CFLAGS_KASAN_MINIMAL) -Werror),)
>     ifneq ($(CONFIG_COMPILE_TEST),y)
>          $(warning Cannot use CONFIG_KASAN: \

Reviewed-by: Dmitry Vyukov <dvyukov@google.com>

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

^ permalink raw reply	[flat|nested] 17+ messages in thread

* Re: [PATCH 4/4] kasan: add compiler support for clang
  2017-07-06 22:01 ` [PATCH 4/4] kasan: add compiler support for clang Greg Hackmann
@ 2017-07-10  8:48   ` Dmitry Vyukov
  2017-07-10 10:34   ` Andrey Ryabinin
  1 sibling, 0 replies; 17+ messages in thread
From: Dmitry Vyukov @ 2017-07-10  8:48 UTC (permalink / raw)
  To: Greg Hackmann
  Cc: Andrey Ryabinin, Alexander Potapenko, Masahiro Yamada,
	Michal Marek, LKML, kasan-dev, linux-mm,
	open list:KERNEL BUILD + fi...,
	Matthias Kaehlcke, Michael Davidson

On Fri, Jul 7, 2017 at 12:01 AM, Greg Hackmann <ghackmann@google.com> wrote:
> For now we can hard-code ASAN ABI level 5, since historical clang builds
> can't build the kernel anyway.  We also need to emulate gcc's
> __SANITIZE_ADDRESS__ flag, or memset() calls won't be instrumented.
>
> Signed-off-by: Greg Hackmann <ghackmann@google.com>
> ---
>  include/linux/compiler-clang.h | 10 ++++++++++
>  1 file changed, 10 insertions(+)
>
> diff --git a/include/linux/compiler-clang.h b/include/linux/compiler-clang.h
> index d614c5ea1b5e..8153f793b22a 100644
> --- a/include/linux/compiler-clang.h
> +++ b/include/linux/compiler-clang.h
> @@ -23,3 +23,13 @@
>   */
>  #undef inline
>  #define inline inline __attribute__((unused)) notrace
> +
> +/* all clang versions usable with the kernel support KASAN ABI version 5
> + */
> +#define KASAN_ABI_VERSION 5
> +
> +/* emulate gcc's __SANITIZE_ADDRESS__ flag
> + */
> +#if __has_feature(address_sanitizer)
> +#define __SANITIZE_ADDRESS__
> +#endif


Reviewed-by: Dmitry Vyukov <dvyukov@google.com>

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

^ permalink raw reply	[flat|nested] 17+ messages in thread

* Re: [PATCH 1/4] kasan: support alloca() poisoning
  2017-07-06 22:01 ` [PATCH 1/4] kasan: support alloca() poisoning Greg Hackmann
  2017-07-07  0:09   ` Greg Hackmann
  2017-07-10  8:44   ` Dmitry Vyukov
@ 2017-07-10 10:30   ` Andrey Ryabinin
  2017-07-13 22:49     ` Greg Hackmann
  2 siblings, 1 reply; 17+ messages in thread
From: Andrey Ryabinin @ 2017-07-10 10:30 UTC (permalink / raw)
  To: Greg Hackmann, Alexander Potapenko, Dmitry Vyukov,
	Masahiro Yamada, Michal Marek
  Cc: linux-kernel, kasan-dev, linux-mm, linux-kbuild,
	Matthias Kaehlcke, Michael Davidson

On 07/07/2017 01:01 AM, Greg Hackmann wrote:
> clang's AddressSanitizer implementation adds redzones on either side of
> alloca()ed buffers.  These redzones are 32-byte aligned and at least 32
> bytes long.

gcc now supports this too. So I think this patch should enable it.
It's off by default so you'll have to add --param asan-instrument-allocas=1 into cflags
to make it work


> 
> __asan_alloca_poison() is passed the size and address of the allocated
> buffer, *excluding* the redzones on either side.  The left redzone will
> always be to the immediate left of this buffer; but AddressSanitizer may
> need to add padding between the end of the buffer and the right redzone.
> If there are any 8-byte chunks inside this padding, we should poison
> those too.
> 
> __asan_allocas_unpoison() is just passed the top and bottom of the
> dynamic stack area, so unpoisoning is simpler.
> 
> Signed-off-by: Greg Hackmann <ghackmann@google.com>
> ---
>  lib/test_kasan.c  | 22 ++++++++++++++++++++++

Tests would be better as a separate patch.


>  mm/kasan/kasan.c  | 26 ++++++++++++++++++++++++++
>  mm/kasan/kasan.h  |  8 ++++++++
>  mm/kasan/report.c |  3 +++
>  4 files changed, 59 insertions(+)
> 
> diff --git a/lib/test_kasan.c b/lib/test_kasan.c
> index a25c9763fce1..f774fcafb696 100644
> --- a/lib/test_kasan.c
> +++ b/lib/test_kasan.c
> @@ -473,6 +473,26 @@ static noinline void __init use_after_scope_test(void)
>  	p[1023] = 1;
>  }
>  
> +static noinline void __init kasan_alloca_oob_left(void)
> +{
> +	volatile int i = 10;
> +	char alloca_array[i];
> +	char *p = alloca_array - 1;
> +
> +	pr_info("out-of-bounds to left on alloca\n");
> +	*(volatile char *)p;
> +}
> +
> +static noinline void __init kasan_alloca_oob_right(void)
> +{
> +	volatile int i = 10;
> +	char alloca_array[i];
> +	char *p = alloca_array + round_up(i, 8);

Why round_up() ?

> +
> +	pr_info("out-of-bounds to right on alloca\n");
> +	*(volatile char *)p;
> +}
> +
>  static int __init kmalloc_tests_init(void)
>  {
>  	/*
> @@ -503,6 +523,8 @@ static int __init kmalloc_tests_init(void)
>  	memcg_accounted_kmem_cache();
>  	kasan_stack_oob();
>  	kasan_global_oob();
> +	kasan_alloca_oob_left();
> +	kasan_alloca_oob_right();
>  	ksize_unpoisons_memory();
>  	copy_user_test();
>  	use_after_scope_test();
> diff --git a/mm/kasan/kasan.c b/mm/kasan/kasan.c
> index c81549d5c833..892b626f564b 100644
> --- a/mm/kasan/kasan.c
> +++ b/mm/kasan/kasan.c
> @@ -802,6 +802,32 @@ void __asan_unpoison_stack_memory(const void *addr, size_t size)
>  }
>  EXPORT_SYMBOL(__asan_unpoison_stack_memory);
>  
> +/* Emitted by compiler to poison alloca()ed objects. */
> +void __asan_alloca_poison(unsigned long addr, size_t size)
> +{
> +	size_t rounded_up_size = round_up(size, KASAN_SHADOW_SCALE_SIZE);
> +	size_t padding_size = round_up(size, KASAN_ALLOCA_REDZONE_SIZE) -
> +			round_up(size, KASAN_SHADOW_SCALE_SIZE);
> +
> +	const void *left_redzone = (const void *)(addr -
> +			KASAN_ALLOCA_REDZONE_SIZE);
> +	const void *right_redzone = (const void *)(addr + rounded_up_size);
> +
> +	kasan_poison_shadow(left_redzone, KASAN_ALLOCA_REDZONE_SIZE,
> +			KASAN_ALLOCA_LEFT);
> +	kasan_poison_shadow(right_redzone,
> +			padding_size + KASAN_ALLOCA_REDZONE_SIZE,
> +			KASAN_ALLOCA_RIGHT);

As Dmitry pointed out, the memory between [addr+size, addr+rounded_up_size) is left
unpoisoned. kasan_alloca_oob_right() without round_up() would have caught this.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

^ permalink raw reply	[flat|nested] 17+ messages in thread

* Re: [PATCH 2/4] kasan: added functions for unpoisoning stack variables
  2017-07-06 22:01 ` [PATCH 2/4] kasan: added functions for unpoisoning stack variables Greg Hackmann
  2017-07-10  8:46   ` Dmitry Vyukov
@ 2017-07-10 10:31   ` Andrey Ryabinin
  1 sibling, 0 replies; 17+ messages in thread
From: Andrey Ryabinin @ 2017-07-10 10:31 UTC (permalink / raw)
  To: Greg Hackmann, Alexander Potapenko, Dmitry Vyukov,
	Masahiro Yamada, Michal Marek
  Cc: linux-kernel, kasan-dev, linux-mm, linux-kbuild,
	Matthias Kaehlcke, Michael Davidson



On 07/07/2017 01:01 AM, Greg Hackmann wrote:
> From: Alexander Potapenko <glider@google.com>
> 
> As a code-size optimization, LLVM builds since r279383 may
> bulk-manipulate the shadow region when (un)poisoning large memory
> blocks.  This requires new callbacks that simply do an uninstrumented
> memset().
> 
> This fixes linking the Clang-built kernel when using KASAN.
> 
> Signed-off-by: Alexander Potapenko <glider@google.com>
> [ghackmann@google.com: fix memset() parameters, and tweak
>  commit message to describe new callbacks]
> Signed-off-by: Greg Hackmann <ghackmann@google.com>
> ---
>  mm/kasan/kasan.c | 15 +++++++++++++++
>  1 file changed, 15 insertions(+)
> 
> diff --git a/mm/kasan/kasan.c b/mm/kasan/kasan.c
> index 892b626f564b..89911e5c69f9 100644
> --- a/mm/kasan/kasan.c
> +++ b/mm/kasan/kasan.c
> @@ -828,6 +828,21 @@ void __asan_allocas_unpoison(const void *stack_top, const void *stack_bottom)
>  }
>  EXPORT_SYMBOL(__asan_allocas_unpoison);
>  
> +/* Emitted by the compiler to [un]poison local variables. */
> +#define DEFINE_ASAN_SET_SHADOW(byte) \
> +	void __asan_set_shadow_##byte(const void *addr, size_t size)	\
> +	{								\
> +		__memset((void *)addr, 0x##byte, size);			\
> +	}								\
> +	EXPORT_SYMBOL(__asan_set_shadow_##byte)
> +
> +DEFINE_ASAN_SET_SHADOW(00);
> +DEFINE_ASAN_SET_SHADOW(f1);
> +DEFINE_ASAN_SET_SHADOW(f2);
> +DEFINE_ASAN_SET_SHADOW(f3);
> +DEFINE_ASAN_SET_SHADOW(f5);
> +DEFINE_ASAN_SET_SHADOW(f8);

I think we can remove f8 as it should be used only by use-after-return instrumentation.
We don't use it in the kernel

> +
>  #ifdef CONFIG_MEMORY_HOTPLUG
>  static int kasan_mem_notifier(struct notifier_block *nb,
>  			unsigned long action, void *data)
> 

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

^ permalink raw reply	[flat|nested] 17+ messages in thread

* Re: [PATCH 4/4] kasan: add compiler support for clang
  2017-07-06 22:01 ` [PATCH 4/4] kasan: add compiler support for clang Greg Hackmann
  2017-07-10  8:48   ` Dmitry Vyukov
@ 2017-07-10 10:34   ` Andrey Ryabinin
  1 sibling, 0 replies; 17+ messages in thread
From: Andrey Ryabinin @ 2017-07-10 10:34 UTC (permalink / raw)
  To: Greg Hackmann, Alexander Potapenko, Dmitry Vyukov,
	Masahiro Yamada, Michal Marek
  Cc: linux-kernel, kasan-dev, linux-mm, linux-kbuild,
	Matthias Kaehlcke, Michael Davidson

On 07/07/2017 01:01 AM, Greg Hackmann wrote:
> For now we can hard-code ASAN ABI level 5, since historical clang builds
> can't build the kernel anyway.  We also need to emulate gcc's
> __SANITIZE_ADDRESS__ flag, or memset() calls won't be instrumented.
> 
> Signed-off-by: Greg Hackmann <ghackmann@google.com>
> ---
>  include/linux/compiler-clang.h | 10 ++++++++++
>  1 file changed, 10 insertions(+)
> 
> diff --git a/include/linux/compiler-clang.h b/include/linux/compiler-clang.h
> index d614c5ea1b5e..8153f793b22a 100644
> --- a/include/linux/compiler-clang.h
> +++ b/include/linux/compiler-clang.h
> @@ -23,3 +23,13 @@
>   */
>  #undef inline
>  #define inline inline __attribute__((unused)) notrace
> +
> +/* all clang versions usable with the kernel support KASAN ABI version 5
> + */

 Enclosing */ should be on the same line for single-line comments.

> +#define KASAN_ABI_VERSION 5
> +
> +/* emulate gcc's __SANITIZE_ADDRESS__ flag
> + */

Ditto.

> +#if __has_feature(address_sanitizer)
> +#define __SANITIZE_ADDRESS__
> +#endif
> 

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

^ permalink raw reply	[flat|nested] 17+ messages in thread

* Re: [PATCH 1/4] kasan: support alloca() poisoning
  2017-07-10  8:44   ` Dmitry Vyukov
@ 2017-07-13 22:40     ` Greg Hackmann
  2017-07-14  6:13       ` Dmitry Vyukov
  0 siblings, 1 reply; 17+ messages in thread
From: Greg Hackmann @ 2017-07-13 22:40 UTC (permalink / raw)
  To: Dmitry Vyukov
  Cc: Andrey Ryabinin, Alexander Potapenko, Masahiro Yamada,
	Michal Marek, LKML, kasan-dev, linux-mm,
	open list:KERNEL BUILD + fi...,
	Matthias Kaehlcke, Michael Davidson

Hi,

Thanks for taking a look at this patchstack.  I apologize for the delay 
in responding.

On 07/10/2017 01:44 AM, Dmitry Vyukov wrote:
>> +
>> +       const void *left_redzone = (const void *)(addr -
>> +                       KASAN_ALLOCA_REDZONE_SIZE);
>> +       const void *right_redzone = (const void *)(addr + rounded_up_size);
> 
> Please check that size is rounded to KASAN_ALLOCA_REDZONE_SIZE. That's
> the expectation, right? That can change is clang silently.
> 
>> +       kasan_poison_shadow(left_redzone, KASAN_ALLOCA_REDZONE_SIZE,
>> +                       KASAN_ALLOCA_LEFT);
>> +       kasan_poison_shadow(right_redzone,
>> +                       padding_size + KASAN_ALLOCA_REDZONE_SIZE,
>> +                       KASAN_ALLOCA_RIGHT);
> 
> We also need to poison the unaligned part at the end of the object
> from size to rounded_up_size. You can see how we do it for heap
> objects.

The expectation is that `size' is the exact size of the alloca()ed 
object.  `rounded_up_size' then adds the 0-7 bytes needed to adjust the 
size to the ASAN shadow scale.  So `addr + rounded_up_size' should be 
the correct place to start poisoning.

In retrospect this part of the code was pretty confusing.  How about 
this?  I think its intent is clearer, plus it's a closer match for the 
description in my commit message:

	unsigned long left_redzone_start;
	unsigned long object_end;
	unsigned long right_redzone_start, right_redzone_end;

	left_redzone_start = addr - KASAN_ALLOCA_REDZONE_SIZE;
	kasan_poison_shadow((const void *)left_redzone_start,
			KASAN_ALLOCA_REDZONE_SIZE,
			KASAN_ALLOCA_LEFT);

	object_end = round_up(addr + size, KASAN_SHADOW_SCALE_SIZE);
	right_redzone_start = round_up(object_end, KASAN_ALLOCA_REDZONE_SIZE);
	right_redzone_end = right_redzone_start + KASAN_ALLOCA_REDZONE_SIZE;
	kasan_poison_shadow((const void *)object_end,
			right_redzone_end - object_end,
			KASAN_ALLOCA_RIGHT);

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

^ permalink raw reply	[flat|nested] 17+ messages in thread

* Re: [PATCH 1/4] kasan: support alloca() poisoning
  2017-07-10 10:30   ` Andrey Ryabinin
@ 2017-07-13 22:49     ` Greg Hackmann
  2017-07-14 16:52       ` Andrey Ryabinin
  0 siblings, 1 reply; 17+ messages in thread
From: Greg Hackmann @ 2017-07-13 22:49 UTC (permalink / raw)
  To: Andrey Ryabinin, Alexander Potapenko, Dmitry Vyukov,
	Masahiro Yamada, Michal Marek
  Cc: linux-kernel, kasan-dev, linux-mm, linux-kbuild,
	Matthias Kaehlcke, Michael Davidson

On 07/10/2017 03:30 AM, Andrey Ryabinin wrote:
> gcc now supports this too. So I think this patch should enable it.
> It's off by default so you'll have to add --param asan-instrument-allocas=1 into cflags
> to make it work

Thanks, will fix.  For now, it looks like I'll need to build gcc from 
git to test this?

>>   lib/test_kasan.c  | 22 ++++++++++++++++++++++
> 
> Tests would be better as a separate patch.

I was following the precedent in 828347f8f9a5 ("kasan: support 
use-after-scope detection") which added both at the same time.  But I 
can split the test off into a separate patch if you feel really strongly 
about it.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

^ permalink raw reply	[flat|nested] 17+ messages in thread

* Re: [PATCH 1/4] kasan: support alloca() poisoning
  2017-07-13 22:40     ` Greg Hackmann
@ 2017-07-14  6:13       ` Dmitry Vyukov
  0 siblings, 0 replies; 17+ messages in thread
From: Dmitry Vyukov @ 2017-07-14  6:13 UTC (permalink / raw)
  To: Greg Hackmann
  Cc: Andrey Ryabinin, Alexander Potapenko, Masahiro Yamada,
	Michal Marek, LKML, kasan-dev, linux-mm,
	open list:KERNEL BUILD + fi...,
	Matthias Kaehlcke, Michael Davidson

On Fri, Jul 14, 2017 at 12:40 AM, Greg Hackmann <ghackmann@google.com> wrote:
> Hi,
>
> Thanks for taking a look at this patchstack.  I apologize for the delay in
> responding.
>
> On 07/10/2017 01:44 AM, Dmitry Vyukov wrote:
>>>
>>> +
>>> +       const void *left_redzone = (const void *)(addr -
>>> +                       KASAN_ALLOCA_REDZONE_SIZE);
>>> +       const void *right_redzone = (const void *)(addr +
>>> rounded_up_size);
>>
>>
>> Please check that size is rounded to KASAN_ALLOCA_REDZONE_SIZE. That's
>> the expectation, right? That can change is clang silently.
>>
>>> +       kasan_poison_shadow(left_redzone, KASAN_ALLOCA_REDZONE_SIZE,
>>> +                       KASAN_ALLOCA_LEFT);
>>> +       kasan_poison_shadow(right_redzone,
>>> +                       padding_size + KASAN_ALLOCA_REDZONE_SIZE,
>>> +                       KASAN_ALLOCA_RIGHT);
>>
>>
>> We also need to poison the unaligned part at the end of the object
>> from size to rounded_up_size. You can see how we do it for heap
>> objects.
>
>
> The expectation is that `size' is the exact size of the alloca()ed object.
> `rounded_up_size' then adds the 0-7 bytes needed to adjust the size to the
> ASAN shadow scale.  So `addr + rounded_up_size' should be the correct place
> to start poisoning.


We need to start poisoning at addr+size exactly.
Asan shadow scheme supports this. It's not possible to poison
beginning of an aligned 8-byte block, but leave tail unpoisoned. But
it is possible to poison tail of an aligned 8-byte block and leave
beginning unpoisoned. Look at what we do for kmalloc.


> In retrospect this part of the code was pretty confusing.  How about this?
> I think its intent is clearer, plus it's a closer match for the description
> in my commit message:
>
>         unsigned long left_redzone_start;
>         unsigned long object_end;
>         unsigned long right_redzone_start, right_redzone_end;
>
>         left_redzone_start = addr - KASAN_ALLOCA_REDZONE_SIZE;
>         kasan_poison_shadow((const void *)left_redzone_start,
>                         KASAN_ALLOCA_REDZONE_SIZE,
>                         KASAN_ALLOCA_LEFT);
>
>         object_end = round_up(addr + size, KASAN_SHADOW_SCALE_SIZE);
>         right_redzone_start = round_up(object_end,
> KASAN_ALLOCA_REDZONE_SIZE);
>         right_redzone_end = right_redzone_start + KASAN_ALLOCA_REDZONE_SIZE;
>         kasan_poison_shadow((const void *)object_end,
>                         right_redzone_end - object_end,
>                         KASAN_ALLOCA_RIGHT);

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

^ permalink raw reply	[flat|nested] 17+ messages in thread

* Re: [PATCH 1/4] kasan: support alloca() poisoning
  2017-07-13 22:49     ` Greg Hackmann
@ 2017-07-14 16:52       ` Andrey Ryabinin
  0 siblings, 0 replies; 17+ messages in thread
From: Andrey Ryabinin @ 2017-07-14 16:52 UTC (permalink / raw)
  To: Greg Hackmann, Alexander Potapenko, Dmitry Vyukov,
	Masahiro Yamada, Michal Marek
  Cc: linux-kernel, kasan-dev, linux-mm, linux-kbuild,
	Matthias Kaehlcke, Michael Davidson

On 07/14/2017 01:49 AM, Greg Hackmann wrote:
> On 07/10/2017 03:30 AM, Andrey Ryabinin wrote:
>> gcc now supports this too. So I think this patch should enable it.
>> It's off by default so you'll have to add --param asan-instrument-allocas=1 into cflags
>> to make it work
> 
> Thanks, will fix.  For now, it looks like I'll need to build gcc from git to test this?
> 

Right, you'll need quite fresh revision >= 250032

>>>   lib/test_kasan.c  | 22 ++++++++++++++++++++++
>>
>> Tests would be better as a separate patch.
> 
> I was following the precedent in 828347f8f9a5 ("kasan: support use-after-scope detection") which added both at the same time. But I can split the test off into a separate patch if you feel really strongly about it.

Please, do the split.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

^ permalink raw reply	[flat|nested] 17+ messages in thread

end of thread, other threads:[~2017-07-14 16:49 UTC | newest]

Thread overview: 17+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2017-07-06 22:01 [PATCH 0/4] kasan: add clang support Greg Hackmann
2017-07-06 22:01 ` [PATCH 1/4] kasan: support alloca() poisoning Greg Hackmann
2017-07-07  0:09   ` Greg Hackmann
2017-07-10  8:44   ` Dmitry Vyukov
2017-07-13 22:40     ` Greg Hackmann
2017-07-14  6:13       ` Dmitry Vyukov
2017-07-10 10:30   ` Andrey Ryabinin
2017-07-13 22:49     ` Greg Hackmann
2017-07-14 16:52       ` Andrey Ryabinin
2017-07-06 22:01 ` [PATCH 2/4] kasan: added functions for unpoisoning stack variables Greg Hackmann
2017-07-10  8:46   ` Dmitry Vyukov
2017-07-10 10:31   ` Andrey Ryabinin
2017-07-06 22:01 ` [PATCH 3/4] kasan: support LLVM-style asan parameters Greg Hackmann
2017-07-10  8:47   ` Dmitry Vyukov
2017-07-06 22:01 ` [PATCH 4/4] kasan: add compiler support for clang Greg Hackmann
2017-07-10  8:48   ` Dmitry Vyukov
2017-07-10 10:34   ` Andrey Ryabinin

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).