[PATCH 0/5] net/packet: fix multiple overflow issues in ring buffers

STALE3413d

Revision v1 of 2 in this series.

20 messages, 2 authors, 2017-03-28 · open the first message on its own page

[PATCH 0/5] net/packet: fix multiple overflow issues in ring buffers

From: Andrey Konovalov <hidden>
Date: 2017-03-28 14:01:15

This patchset addresses multiple overflows and signedness-related issues
in packet socket ring buffers.

Andrey Konovalov (5):
  net/packet: fix overflow in check for priv area size
  net/packet: add explicit checks for tp_frame_size
  net/packet: fix overflow in check for tp_frame_nr
  net/packet: fix overflow in check for tp_reserve
  net/packet: reorder checks for ring buffer parameters

 net/packet/af_packet.c | 31 +++++++++++++++++++++----------
 1 file changed, 21 insertions(+), 10 deletions(-)

-- 
2.12.2.564.g063fe858b8-goog

[PATCH v4 3/9] kasan: change allocation and freeing stack traces headers

From: Andrey Konovalov <hidden>
Date: 2017-03-28 14:01:07

Change stack traces headers from:

Allocated:
PID = 42

to:

Allocated by task 42:

Makes the report one line shorter and look better.

Signed-off-by: Andrey Konovalov <redacted>
---
 mm/kasan/report.c | 10 ++++------
 1 file changed, 4 insertions(+), 6 deletions(-)

diff --git a/mm/kasan/report.c b/mm/kasan/report.c
index fc0577d15671..382d4d2b9052 100644
--- a/mm/kasan/report.c
+++ b/mm/kasan/report.c

@@ -175,9 +175,9 @@ static void kasan_end_report(unsigned long *flags)
 	kasan_enable_current();
 }
 
-static void print_track(struct kasan_track *track)
+static void print_track(struct kasan_track *track, const char *prefix)
 {
-	pr_err("PID = %u\n", track->pid);
+	pr_err("%s by task %u:\n", prefix, track->pid);
 	if (track->stack) {
 		struct stack_trace trace;

@@ -199,10 +199,8 @@ static void kasan_object_err(struct kmem_cache *cache, void *object)
 	if (!(cache->flags & SLAB_KASAN))
 		return;
 
-	pr_err("Allocated:\n");
-	print_track(&alloc_info->alloc_track);
-	pr_err("Freed:\n");
-	print_track(&alloc_info->free_track);
+	print_track(&alloc_info->alloc_track, "Allocated");
+	print_track(&alloc_info->free_track, "Freed");
 }
 
 void kasan_report_double_free(struct kmem_cache *cache, void *object,

-- 
2.12.1.578.ge9c3154ca4-goog

[PATCH 5/5] net/packet: reorder checks for ring buffer parameters

From: Andrey Konovalov <hidden>
Date: 2017-03-28 14:01:09

No semantic changes.

Improves readability.

Signed-off-by: Andrey Konovalov <redacted>
---
 net/packet/af_packet.c | 17 +++++++++--------
 1 file changed, 9 insertions(+), 8 deletions(-)

diff --git a/net/packet/af_packet.c b/net/packet/af_packet.c
index 28b49749d1af..de25736a7988 100644
--- a/net/packet/af_packet.c
+++ b/net/packet/af_packet.c

@@ -4191,6 +4191,11 @@ static int packet_set_ring(struct sock *sk, union tpacket_req_u *req_u,
 
 		err = -EINVAL;
 
+		if (unlikely(!PAGE_ALIGNED(req->tp_block_size)))
+			goto out;
+		if (unlikely(req->tp_frame_size & (TPACKET_ALIGNMENT - 1)))
+			goto out;
+
 		if (unlikely(req->tp_block_size > INT_MAX))
 			goto out;
 		if (unlikely(req->tp_block_size == 0))

@@ -4205,19 +4210,15 @@ static int packet_set_ring(struct sock *sk, union tpacket_req_u *req_u,
 		if (unlikely(po->tp_reserve >= req->tp_frame_size))
 			goto out;
 
-		if (unlikely(!PAGE_ALIGNED(req->tp_block_size)))
+		rb->frames_per_block = req->tp_block_size / req->tp_frame_size;
+		if (unlikely((rb->frames_per_block * req->tp_block_nr) !=
+					req->tp_frame_nr))
 			goto out;
+
 		if (po->tp_version >= TPACKET_V3 &&
 		    req->tp_block_size <=
 			  BLK_PLUS_PRIV((u64)req_u->req3.tp_sizeof_priv))
 			goto out;
-		if (unlikely(req->tp_frame_size & (TPACKET_ALIGNMENT - 1)))
-			goto out;
-
-		rb->frames_per_block = req->tp_block_size / req->tp_frame_size;
-		if (unlikely((rb->frames_per_block * req->tp_block_nr) !=
-					req->tp_frame_nr))
-			goto out;
 
 		err = -ENOMEM;
 		order = get_order(req->tp_block_size);

-- 
2.12.2.564.g063fe858b8-goog

[PATCH 4/5] net/packet: fix overflow in check for tp_reserve

From: Andrey Konovalov <hidden>
Date: 2017-03-28 14:01:11

When calculating po->tp_hdrlen + po->tp_reserve the result can overflow.

Fix by checking that tp_reserve <= INT_MAX on assign.

This also takes cared of an overflow when calculating
 macoff = TPACKET_ALIGN(po->tp_hdrlen) + 16 + po->tp_reserve
 snaplen = skb->len
 macoff + snaplen
since macoff ~ INT_MAX and snaplen < SKB_MAX_ALLOC.

Signed-off-by: Andrey Konovalov <redacted>
---
 net/packet/af_packet.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/net/packet/af_packet.c b/net/packet/af_packet.c
index c5c43fff8c01..28b49749d1af 100644
--- a/net/packet/af_packet.c
+++ b/net/packet/af_packet.c

@@ -3665,6 +3665,8 @@ packet_setsockopt(struct socket *sock, int level, int optname, char __user *optv
 			return -EBUSY;
 		if (copy_from_user(&val, optval, sizeof(val)))
 			return -EFAULT;
+		if (val > INT_MAX)
+			return -EINVAL;
 		po->tp_reserve = val;
 		return 0;
 	}

@@ -4200,6 +4202,8 @@ static int packet_set_ring(struct sock *sk, union tpacket_req_u *req_u,
 		if (unlikely((u64)req->tp_block_size * req->tp_block_nr >
 					UINT_MAX))
 			goto out;
+		if (unlikely(po->tp_reserve >= req->tp_frame_size))
+			goto out;
 
 		if (unlikely(!PAGE_ALIGNED(req->tp_block_size)))
 			goto out;

@@ -4207,9 +4211,6 @@ static int packet_set_ring(struct sock *sk, union tpacket_req_u *req_u,
 		    req->tp_block_size <=
 			  BLK_PLUS_PRIV((u64)req_u->req3.tp_sizeof_priv))
 			goto out;
-		if (unlikely(req->tp_frame_size < po->tp_hdrlen +
-					po->tp_reserve))
-			goto out;
 		if (unlikely(req->tp_frame_size & (TPACKET_ALIGNMENT - 1)))
 			goto out;

-- 
2.12.2.564.g063fe858b8-goog

Re: [PATCH 4/5] net/packet: fix overflow in check for tp_reserve

From: Willem de Bruijn <willemdebruijn.kernel@gmail.com>
Date: 2017-03-28 15:02:18

On Tue, Mar 28, 2017 at 10:00 AM, Andrey Konovalov
[off-list ref] wrote:

When calculating po->tp_hdrlen + po->tp_reserve the result can overflow.

Fix by checking that tp_reserve <= INT_MAX on assign.

This also takes cared of an overflow when calculating
 macoff = TPACKET_ALIGN(po->tp_hdrlen) + 16 + po->tp_reserve
 snaplen = skb->len
 macoff + snaplen
since macoff ~ INT_MAX and snaplen < SKB_MAX_ALLOC.

This refers to the overflow of macoff + snaplen?

Note that macoff is unsigned short, so will truncate any overflow from
tp_reserve.

quoted hunk

Signed-off-by: Andrey Konovalov <redacted>
---
 net/packet/af_packet.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/net/packet/af_packet.c b/net/packet/af_packet.c
index c5c43fff8c01..28b49749d1af 100644
--- a/net/packet/af_packet.c
+++ b/net/packet/af_packet.c

@@ -3665,6 +3665,8 @@ packet_setsockopt(struct socket *sock, int level, int optname, char __user *optv
                        return -EBUSY;
                if (copy_from_user(&val, optval, sizeof(val)))
                        return -EFAULT;
+               if (val > INT_MAX)
+                       return -EINVAL;

This change on its own is sufficient to avoid the overflow. For net
and backports to stable, this minimal patch is preferable.

quoted hunk

                po->tp_reserve = val;
                return 0;
        }

@@ -4200,6 +4202,8 @@ static int packet_set_ring(struct sock *sk, union tpacket_req_u *req_u,
                if (unlikely((u64)req->tp_block_size * req->tp_block_nr >
                                        UINT_MAX))
                        goto out;
+               if (unlikely(po->tp_reserve >= req->tp_frame_size))
+                       goto out;

                if (unlikely(!PAGE_ALIGNED(req->tp_block_size)))
                        goto out;
@@ -4207,9 +4211,6 @@ static int packet_set_ring(struct sock *sk, union tpacket_req_u *req_u,
                    req->tp_block_size <=
                          BLK_PLUS_PRIV((u64)req_u->req3.tp_sizeof_priv))
                        goto out;
-               if (unlikely(req->tp_frame_size < po->tp_hdrlen +
-                                       po->tp_reserve))
-                       goto out;

Is there a reason that the test is moved up? It is probably not
correct to remove tp_hdrlen from the test.

                if (unlikely(req->tp_frame_size & (TPACKET_ALIGNMENT - 1)))
                        goto out;

--
2.12.2.564.g063fe858b8-goog

Re: [PATCH 4/5] net/packet: fix overflow in check for tp_reserve

From: Andrey Konovalov <hidden>
Date: 2017-03-28 15:12:49

On Tue, Mar 28, 2017 at 5:00 PM, Willem de Bruijn
[off-list ref] wrote:

On Tue, Mar 28, 2017 at 10:00 AM, Andrey Konovalov
[off-list ref] wrote:

quoted

When calculating po->tp_hdrlen + po->tp_reserve the result can overflow.

Fix by checking that tp_reserve <= INT_MAX on assign.

This also takes cared of an overflow when calculating
 macoff = TPACKET_ALIGN(po->tp_hdrlen) + 16 + po->tp_reserve
 snaplen = skb->len
 macoff + snaplen
since macoff ~ INT_MAX and snaplen < SKB_MAX_ALLOC.

This refers to the overflow of macoff + snaplen?

Note that macoff is unsigned short, so will truncate any overflow from
tp_reserve.

Yes, you're right.
Should I make macoff unsigned int to fix this?

quoted

Signed-off-by: Andrey Konovalov <redacted>
---
 net/packet/af_packet.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/net/packet/af_packet.c b/net/packet/af_packet.c
index c5c43fff8c01..28b49749d1af 100644
--- a/net/packet/af_packet.c
+++ b/net/packet/af_packet.c

@@ -3665,6 +3665,8 @@ packet_setsockopt(struct socket *sock, int level, int optname, char __user *optv
                        return -EBUSY;
                if (copy_from_user(&val, optval, sizeof(val)))
                        return -EFAULT;
+               if (val > INT_MAX)
+                       return -EINVAL;

This change on its own is sufficient to avoid the overflow. For net
and backports to stable, this minimal patch is preferable.

I will put it into a separate patch then.

quoted

                po->tp_reserve = val;
                return 0;
        }

@@ -4200,6 +4202,8 @@ static int packet_set_ring(struct sock *sk, union tpacket_req_u *req_u,
                if (unlikely((u64)req->tp_block_size * req->tp_block_nr >
                                        UINT_MAX))
                        goto out;
+               if (unlikely(po->tp_reserve >= req->tp_frame_size))
+                       goto out;

                if (unlikely(!PAGE_ALIGNED(req->tp_block_size)))
                        goto out;
@@ -4207,9 +4211,6 @@ static int packet_set_ring(struct sock *sk, union tpacket_req_u *req_u,
                    req->tp_block_size <=
                          BLK_PLUS_PRIV((u64)req_u->req3.tp_sizeof_priv))
                        goto out;
-               if (unlikely(req->tp_frame_size < po->tp_hdrlen +
-                                       po->tp_reserve))
-                       goto out;

Is there a reason that the test is moved up? It is probably not
correct to remove tp_hdrlen from the test.

Just to group together all checks of tp_frame_size and tp_block_size.

I'm not sure there's any difference between checking against
po->tp_hdrlen + po->tp_reserve and just po->tp_reserve.
I guess the correct check should be against
TPACKET_ALIGN(po->tp_hdrlen) + 16 + po->tp_reserve.

Should I use this value?

quoted

                if (unlikely(req->tp_frame_size & (TPACKET_ALIGNMENT - 1)))
                        goto out;

--
2.12.2.564.g063fe858b8-goog

Re: [PATCH 4/5] net/packet: fix overflow in check for tp_reserve

From: Willem de Bruijn <willemdebruijn.kernel@gmail.com>
Date: 2017-03-28 15:21:57

On Tue, Mar 28, 2017 at 11:11 AM, Andrey Konovalov
[off-list ref] wrote:

On Tue, Mar 28, 2017 at 5:00 PM, Willem de Bruijn
[off-list ref] wrote:

quoted

On Tue, Mar 28, 2017 at 10:00 AM, Andrey Konovalov
[off-list ref] wrote:

quoted

When calculating po->tp_hdrlen + po->tp_reserve the result can overflow.

Fix by checking that tp_reserve <= INT_MAX on assign.

This also takes cared of an overflow when calculating
 macoff = TPACKET_ALIGN(po->tp_hdrlen) + 16 + po->tp_reserve
 snaplen = skb->len
 macoff + snaplen
since macoff ~ INT_MAX and snaplen < SKB_MAX_ALLOC.

This refers to the overflow of macoff + snaplen?

Note that macoff is unsigned short, so will truncate any overflow from
tp_reserve.

Yes, you're right.
Should I make macoff unsigned int to fix this?

This is an unrelated issue. On first read, it seems quite harmless as
a process can
cause data to be placed at an offset that causes it to be overwritten
by the tpacket_hdr
later. Worth looking into more closely separately.

quoted

Signed-off-by: Andrey Konovalov <redacted>
---
 net/packet/af_packet.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/net/packet/af_packet.c b/net/packet/af_packet.c
index c5c43fff8c01..28b49749d1af 100644
--- a/net/packet/af_packet.c
+++ b/net/packet/af_packet.c

@@ -3665,6 +3665,8 @@ packet_setsockopt(struct socket *sock, int level, int optname, char __user *optv
                        return -EBUSY;
                if (copy_from_user(&val, optval, sizeof(val)))
                        return -EFAULT;
+               if (val > INT_MAX)
+                       return -EINVAL;

This change on its own is sufficient to avoid the overflow. For net
and backports to stable, this minimal patch is preferable.

I will put it into a separate patch then.

Thanks.

quoted

                po->tp_reserve = val;
                return 0;
        }

@@ -4200,6 +4202,8 @@ static int packet_set_ring(struct sock *sk, union tpacket_req_u *req_u,
                if (unlikely((u64)req->tp_block_size * req->tp_block_nr >
                                        UINT_MAX))
                        goto out;
+               if (unlikely(po->tp_reserve >= req->tp_frame_size))
+                       goto out;

                if (unlikely(!PAGE_ALIGNED(req->tp_block_size)))
                        goto out;
@@ -4207,9 +4211,6 @@ static int packet_set_ring(struct sock *sk, union tpacket_req_u *req_u,
                    req->tp_block_size <=
                          BLK_PLUS_PRIV((u64)req_u->req3.tp_sizeof_priv))
                        goto out;
-               if (unlikely(req->tp_frame_size < po->tp_hdrlen +
-                                       po->tp_reserve))
-                       goto out;

Is there a reason that the test is moved up? It is probably not
correct to remove tp_hdrlen from the test.

Just to group together all checks of tp_frame_size and tp_block_size.

That makes sense, but indeed more for net-next. I would then send a single patch
that includes the other new block and frame tests.

I'm not sure there's any difference between checking against
po->tp_hdrlen + po->tp_reserve and just po->tp_reserve.
I guess the correct check should be against
TPACKET_ALIGN(po->tp_hdrlen) + 16 + po->tp_reserve.

Should I use this value?

Yes, for net-next this seems like a good tightening of the test.

[PATCH 1/5] net/packet: fix overflow in check for priv area size

From: Andrey Konovalov <hidden>
Date: 2017-03-28 14:01:17

Subtracting tp_sizeof_priv from tp_block_size and casting to int
to check whether one is less then the other doesn't always work
(both of them are unsigned ints).

Compare them as is instead.

Also cast tp_sizeof_priv to u64 before using BLK_PLUS_PRIV, as
it can overflow inside BLK_PLUS_PRIV otherwise.

Signed-off-by: Andrey Konovalov <redacted>
---
 net/packet/af_packet.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/net/packet/af_packet.c b/net/packet/af_packet.c
index a0dbe7ca8f72..2323ee35dc09 100644
--- a/net/packet/af_packet.c
+++ b/net/packet/af_packet.c

@@ -4193,8 +4193,8 @@ static int packet_set_ring(struct sock *sk, union tpacket_req_u *req_u,
 		if (unlikely(!PAGE_ALIGNED(req->tp_block_size)))
 			goto out;
 		if (po->tp_version >= TPACKET_V3 &&
-		    (int)(req->tp_block_size -
-			  BLK_PLUS_PRIV(req_u->req3.tp_sizeof_priv)) <= 0)
+		    req->tp_block_size <=
+			  BLK_PLUS_PRIV((u64)req_u->req3.tp_sizeof_priv))
 			goto out;
 		if (unlikely(req->tp_frame_size < po->tp_hdrlen +
 					po->tp_reserve))

-- 
2.12.2.564.g063fe858b8-goog

[PATCH v4 1/9] kasan: introduce helper functions for determining bug type

From: Andrey Konovalov <hidden>
Date: 2017-03-28 14:01:21

Introduce get_shadow_bug_type() function, which determines bug type
based on the shadow value for a particular kernel address.
Introduce get_wild_bug_type() function, which determines bug type
for addresses which don't have a corresponding shadow value.

Signed-off-by: Andrey Konovalov <redacted>
---
 mm/kasan/report.c | 40 ++++++++++++++++++++++++++++++----------
 1 file changed, 30 insertions(+), 10 deletions(-)

diff --git a/mm/kasan/report.c b/mm/kasan/report.c
index f479365530b6..e3af37b7a74c 100644
--- a/mm/kasan/report.c
+++ b/mm/kasan/report.c

@@ -49,7 +49,13 @@ static const void *find_first_bad_addr(const void *addr, size_t size)
 	return first_bad_addr;
 }
 
-static void print_error_description(struct kasan_access_info *info)
+static bool addr_has_shadow(struct kasan_access_info *info)
+{
+	return (info->access_addr >=
+		kasan_shadow_to_mem((void *)KASAN_SHADOW_START));
+}
+
+static const char *get_shadow_bug_type(struct kasan_access_info *info)
 {
 	const char *bug_type = "unknown-crash";
 	u8 *shadow_addr;

@@ -96,6 +102,27 @@ static void print_error_description(struct kasan_access_info *info)
 		break;
 	}
 
+	return bug_type;
+}
+
+static const char *get_wild_bug_type(struct kasan_access_info *info)
+{
+	const char *bug_type;
+
+	if ((unsigned long)info->access_addr < PAGE_SIZE)
+		bug_type = "null-ptr-deref";
+	else if ((unsigned long)info->access_addr < TASK_SIZE)
+		bug_type = "user-memory-access";
+	else
+		bug_type = "wild-memory-access";
+
+	return bug_type;
+}
+
+static void print_error_description(struct kasan_access_info *info)
+{
+	const char *bug_type = get_shadow_bug_type(info);
+
 	pr_err("BUG: KASAN: %s in %pS at addr %p\n",
 		bug_type, (void *)info->ip,
 		info->access_addr);

@@ -265,18 +292,11 @@ static void print_shadow_for_address(const void *addr)
 static void kasan_report_error(struct kasan_access_info *info)
 {
 	unsigned long flags;
-	const char *bug_type;
 
 	kasan_start_report(&flags);
 
-	if (info->access_addr <
-			kasan_shadow_to_mem((void *)KASAN_SHADOW_START)) {
-		if ((unsigned long)info->access_addr < PAGE_SIZE)
-			bug_type = "null-ptr-deref";
-		else if ((unsigned long)info->access_addr < TASK_SIZE)
-			bug_type = "user-memory-access";
-		else
-			bug_type = "wild-memory-access";
+	if (!addr_has_shadow(info)) {
+		const char *bug_type = get_wild_bug_type(info);
 		pr_err("BUG: KASAN: %s on address %p\n",
 			bug_type, info->access_addr);
 		pr_err("%s of size %zu by task %s/%d\n",

-- 
2.12.1.578.ge9c3154ca4-goog

[PATCH v4 4/9] kasan: simplify address description logic

From: Andrey Konovalov <hidden>
Date: 2017-03-28 14:01:21

Simplify logic for describing a memory address.
Add addr_to_page() helper function.

Makes the code easier to follow.

Signed-off-by: Andrey Konovalov <redacted>
---
 mm/kasan/report.c | 36 ++++++++++++++++++++----------------
 1 file changed, 20 insertions(+), 16 deletions(-)

diff --git a/mm/kasan/report.c b/mm/kasan/report.c
index 382d4d2b9052..f77341979dae 100644
--- a/mm/kasan/report.c
+++ b/mm/kasan/report.c

@@ -188,11 +188,17 @@ static void print_track(struct kasan_track *track, const char *prefix)
 	}
 }
 
-static void kasan_object_err(struct kmem_cache *cache, void *object)
+static struct page *addr_to_page(const void *addr)
+{
+	if ((addr >= (void *)PAGE_OFFSET) && (addr < high_memory))
+		return virt_to_head_page(addr);
+	return NULL;
+}
+
+static void describe_object(struct kmem_cache *cache, void *object)
 {
 	struct kasan_alloc_meta *alloc_info = get_alloc_info(cache, object);
 
-	dump_stack();
 	pr_err("Object at %p, in cache %s size: %d\n", object, cache->name,
 		cache->object_size);

@@ -211,34 +217,32 @@ void kasan_report_double_free(struct kmem_cache *cache, void *object,
 	kasan_start_report(&flags);
 	pr_err("BUG: Double free or freeing an invalid pointer\n");
 	pr_err("Unexpected shadow byte: 0x%hhX\n", shadow);
-	kasan_object_err(cache, object);
+	dump_stack();
+	describe_object(cache, object);
 	kasan_end_report(&flags);
 }
 
 static void print_address_description(struct kasan_access_info *info)
 {
 	const void *addr = info->access_addr;
+	struct page *page = addr_to_page(addr);
 
-	if ((addr >= (void *)PAGE_OFFSET) &&
-		(addr < high_memory)) {
-		struct page *page = virt_to_head_page(addr);
-
-		if (PageSlab(page)) {
-			void *object;
-			struct kmem_cache *cache = page->slab_cache;
-			object = nearest_obj(cache, page,
-						(void *)info->access_addr);
-			kasan_object_err(cache, object);
-			return;
-		}
+	if (page)
 		dump_page(page, "kasan: bad access detected");
+
+	dump_stack();
+
+	if (page && PageSlab(page)) {
+		struct kmem_cache *cache = page->slab_cache;
+		void *object = nearest_obj(cache, page,	(void *)addr);
+
+		describe_object(cache, object);
 	}
 
 	if (kernel_or_module_addr(addr)) {
 		if (!init_task_stack_addr(addr))
 			pr_err("Address belongs to variable %pS\n", addr);
 	}
-	dump_stack();
 }
 
 static bool row_is_guilty(const void *row, const void *guilty)

-- 
2.12.1.578.ge9c3154ca4-goog

[PATCH v4 5/9] kasan: change report header

From: Andrey Konovalov <hidden>
Date: 2017-03-28 14:01:22

Change report header format from:

BUG: KASAN: use-after-free in unwind_get_return_address+0x28a/0x2c0 at addr ffff880069437950
Read of size 8 by task insmod/3925

to:

BUG: KASAN: use-after-free in unwind_get_return_address+0x28a/0x2c0
Read of size 8 at addr ffff880069437950 by task insmod/3925

The exact access address is not usually important, so move it to the
second line. This also makes the header look visually balanced.

Signed-off-by: Andrey Konovalov <redacted>
---
 mm/kasan/report.c | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

diff --git a/mm/kasan/report.c b/mm/kasan/report.c
index f77341979dae..156f998199e2 100644
--- a/mm/kasan/report.c
+++ b/mm/kasan/report.c

@@ -130,11 +130,10 @@ static void print_error_description(struct kasan_access_info *info)
 {
 	const char *bug_type = get_bug_type(info);
 
-	pr_err("BUG: KASAN: %s in %pS at addr %p\n",
-		bug_type, (void *)info->ip, info->access_addr);
-	pr_err("%s of size %zu by task %s/%d\n",
+	pr_err("BUG: KASAN: %s in %pS\n", bug_type, (void *)info->ip);
+	pr_err("%s of size %zu at addr %p by task %s/%d\n",
 		info->is_write ? "Write" : "Read", info->access_size,
-		current->comm, task_pid_nr(current));
+		info->access_addr, current->comm, task_pid_nr(current));
 }
 
 static inline bool kernel_or_module_addr(const void *addr)

-- 
2.12.1.578.ge9c3154ca4-goog

[PATCH 2/5] net/packet: add explicit checks for tp_frame_size

From: Andrey Konovalov <hidden>
Date: 2017-03-28 14:01:22

tp_frame_size can't be 0 or be larger than tp_block_size.

As a result the check for frames_per_block == 0 is not needed any more.

Also do explicit checks for tp_block_size, instead of casting to int.

Signed-off-by: Andrey Konovalov <redacted>
---
 net/packet/af_packet.c | 12 +++++++++---
 1 file changed, 9 insertions(+), 3 deletions(-)

diff --git a/net/packet/af_packet.c b/net/packet/af_packet.c
index 2323ee35dc09..506348abdf2f 100644
--- a/net/packet/af_packet.c
+++ b/net/packet/af_packet.c

@@ -4188,8 +4188,16 @@ static int packet_set_ring(struct sock *sk, union tpacket_req_u *req_u,
 		}
 
 		err = -EINVAL;
-		if (unlikely((int)req->tp_block_size <= 0))
+
+		if (unlikely(req->tp_block_size > INT_MAX))
+			goto out;
+		if (unlikely(req->tp_block_size == 0))
+			goto out;
+		if (unlikely(req->tp_frame_size > req->tp_block_size))
 			goto out;
+		if (unlikely(req->tp_frame_size == 0))
+			goto out;
+
 		if (unlikely(!PAGE_ALIGNED(req->tp_block_size)))
 			goto out;
 		if (po->tp_version >= TPACKET_V3 &&

@@ -4203,8 +4211,6 @@ static int packet_set_ring(struct sock *sk, union tpacket_req_u *req_u,
 			goto out;
 
 		rb->frames_per_block = req->tp_block_size / req->tp_frame_size;
-		if (unlikely(rb->frames_per_block == 0))
-			goto out;
 		if (unlikely((rb->frames_per_block * req->tp_block_nr) !=
 					req->tp_frame_nr))
 			goto out;

-- 
2.12.2.564.g063fe858b8-goog

[PATCH v4 8/9] kasan: improve double-free report format

From: Andrey Konovalov <hidden>
Date: 2017-03-28 14:01:31

Changes double-free report header from:

BUG: Double free or freeing an invalid pointer
Unexpected shadow byte: 0xFB

to:

BUG: KASAN: double-free or invalid-free in kmalloc_oob_left+0xe5/0xef

This makes a bug uniquely identifiable by the first report line.
To account for removing of the unexpected shadow value, print shadow
bytes at the end of the report as in reports for other kinds of bugs.

To print caller funtion name in the report header, the caller address
is passed from SLUB/SLAB free handlers.

Signed-off-by: Andrey Konovalov <redacted>
---
 include/linux/kasan.h |  2 +-
 mm/kasan/kasan.c      |  5 +++--
 mm/kasan/kasan.h      |  2 +-
 mm/kasan/report.c     | 30 ++++++++++++++----------------
 mm/slab.c             |  2 +-
 mm/slub.c             | 12 +++++++-----
 6 files changed, 27 insertions(+), 26 deletions(-)

diff --git a/include/linux/kasan.h b/include/linux/kasan.h
index 5734480c9590..554f843f1625 100644
--- a/include/linux/kasan.h
+++ b/include/linux/kasan.h

@@ -62,7 +62,7 @@ void kasan_kmalloc(struct kmem_cache *s, const void *object, size_t size,
 void kasan_krealloc(const void *object, size_t new_size, gfp_t flags);
 
 void kasan_slab_alloc(struct kmem_cache *s, void *object, gfp_t flags);
-bool kasan_slab_free(struct kmem_cache *s, void *object);
+bool kasan_slab_free(struct kmem_cache *s, void *object, unsigned long pc);
 
 struct kasan_cache {
 	int alloc_meta_offset;

diff --git a/mm/kasan/kasan.c b/mm/kasan/kasan.c
index 98b27195e38b..83cc011bb9bc 100644
--- a/mm/kasan/kasan.c
+++ b/mm/kasan/kasan.c

@@ -567,7 +567,8 @@ static void kasan_poison_slab_free(struct kmem_cache *cache, void *object)
 	kasan_poison_shadow(object, rounded_up_size, KASAN_KMALLOC_FREE);
 }
 
-bool kasan_slab_free(struct kmem_cache *cache, void *object)
+bool kasan_slab_free(struct kmem_cache *cache, void *object,
+		     unsigned long pc)
 {
 	s8 shadow_byte;

@@ -577,7 +578,7 @@ bool kasan_slab_free(struct kmem_cache *cache, void *object)
 
 	shadow_byte = READ_ONCE(*(s8 *)kasan_mem_to_shadow(object));
 	if (shadow_byte < 0 || shadow_byte >= KASAN_SHADOW_SCALE_SIZE) {
-		kasan_report_double_free(cache, object, shadow_byte);
+		kasan_report_double_free(cache, object, pc);
 		return true;
 	}

diff --git a/mm/kasan/kasan.h b/mm/kasan/kasan.h
index 1c260e6b3b3c..75729173ade9 100644
--- a/mm/kasan/kasan.h
+++ b/mm/kasan/kasan.h

@@ -104,7 +104,7 @@ static inline bool kasan_report_enabled(void)
 void kasan_report(unsigned long addr, size_t size,
 		bool is_write, unsigned long ip);
 void kasan_report_double_free(struct kmem_cache *cache, void *object,
-			s8 shadow);
+					void *ip);
 
 #if defined(CONFIG_SLAB) || defined(CONFIG_SLUB)
 void quarantine_put(struct kasan_free_meta *info, struct kmem_cache *cache);

diff --git a/mm/kasan/report.c b/mm/kasan/report.c
index e0f7dbf9e883..2368b8cf5f95 100644
--- a/mm/kasan/report.c
+++ b/mm/kasan/report.c

@@ -224,22 +224,8 @@ static void describe_object(struct kmem_cache *cache, void *object,
 	describe_object_addr(cache, object, addr);
 }
 
-void kasan_report_double_free(struct kmem_cache *cache, void *object,
-			s8 shadow)
-{
-	unsigned long flags;
-
-	kasan_start_report(&flags);
-	pr_err("BUG: Double free or freeing an invalid pointer\n");
-	pr_err("Unexpected shadow byte: 0x%hhX\n", shadow);
-	dump_stack();
-	describe_object(cache, object, NULL);
-	kasan_end_report(&flags);
-}
-
-static void print_address_description(struct kasan_access_info *info)
+static void print_address_description(void *addr)
 {
-	void *addr = (void *)info->access_addr;
 	struct page *page = addr_to_page(addr);
 
 	dump_stack();

@@ -314,6 +300,18 @@ static void print_shadow_for_address(const void *addr)
 	}
 }
 
+void kasan_report_double_free(struct kmem_cache *cache, void *object,
+				void *ip)
+{
+	unsigned long flags;
+
+	kasan_start_report(&flags);
+	pr_err("BUG: KASAN: double-free or invalid-free in %pS\n", ip);
+	print_address_description(object);
+	print_shadow_for_address(object);
+	kasan_end_report(&flags);
+}
+
 static void kasan_report_error(struct kasan_access_info *info)
 {
 	unsigned long flags;

@@ -325,7 +323,7 @@ static void kasan_report_error(struct kasan_access_info *info)
 	if (!addr_has_shadow(info)) {
 		dump_stack();
 	} else {
-		print_address_description(info);
+		print_address_description((void *)info->access_addr);
 		print_shadow_for_address(info->first_bad_addr);
 	}

diff --git a/mm/slab.c b/mm/slab.c
index 807d86c76908..aba5f30ea63e 100644
--- a/mm/slab.c
+++ b/mm/slab.c

@@ -3508,7 +3508,7 @@ static inline void __cache_free(struct kmem_cache *cachep, void *objp,
 				unsigned long caller)
 {
 	/* Put the object into the quarantine, don't touch it for now. */
-	if (kasan_slab_free(cachep, objp))
+	if (kasan_slab_free(cachep, objp, caller))
 		return;
 
 	___cache_free(cachep, objp, caller);

diff --git a/mm/slub.c b/mm/slub.c
index 7f4bc7027ed5..763570a0b15e 100644
--- a/mm/slub.c
+++ b/mm/slub.c

@@ -1325,7 +1325,8 @@ static inline void kfree_hook(const void *x)
 	kasan_kfree_large(x);
 }
 
-static inline void *slab_free_hook(struct kmem_cache *s, void *x)
+static inline void *slab_free_hook(struct kmem_cache *s, void *x,
+				   unsigned long addr)
 {
 	void *freeptr;

@@ -1354,12 +1355,13 @@ static inline void *slab_free_hook(struct kmem_cache *s, void *x)
 	 * kasan_slab_free() may put x into memory quarantine, delaying its
 	 * reuse. In this case the object's freelist pointer is changed.
 	 */
-	kasan_slab_free(s, x);
+	kasan_slab_free(s, x, addr);
 	return freeptr;
 }
 
 static inline void slab_free_freelist_hook(struct kmem_cache *s,
-					   void *head, void *tail)
+					   void *head, void *tail,
+					   unsigned long addr)
 {
 /*
  * Compiler cannot detect this function can be removed if slab_free_hook()

@@ -1376,7 +1378,7 @@ static inline void slab_free_freelist_hook(struct kmem_cache *s,
 	void *freeptr;
 
 	do {
-		freeptr = slab_free_hook(s, object);
+		freeptr = slab_free_hook(s, object, addr);
 	} while ((object != tail_obj) && (object = freeptr));
 #endif
 }

@@ -2958,7 +2960,7 @@ static __always_inline void slab_free(struct kmem_cache *s, struct page *page,
 				      void *head, void *tail, int cnt,
 				      unsigned long addr)
 {
-	slab_free_freelist_hook(s, head, tail);
+	slab_free_freelist_hook(s, head, tail, addr);
 	/*
 	 * slab_free_freelist_hook() could have put the items into quarantine.
 	 * If so, no need to free them.

-- 
2.12.1.578.ge9c3154ca4-goog

[PATCH v4 9/9] kasan: separate report parts by empty lines

From: Andrey Konovalov <hidden>
Date: 2017-03-28 14:01:32

Makes the report easier to read.

Signed-off-by: Andrey Konovalov <redacted>
---
 mm/kasan/report.c | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/mm/kasan/report.c b/mm/kasan/report.c
index 2368b8cf5f95..a79fc1036161 100644
--- a/mm/kasan/report.c
+++ b/mm/kasan/report.c

@@ -218,7 +218,9 @@ static void describe_object(struct kmem_cache *cache, void *object,
 
 	if (cache->flags & SLAB_KASAN) {
 		print_track(&alloc_info->alloc_track, "Allocated");
+		pr_err("\n");
 		print_track(&alloc_info->free_track, "Freed");
+		pr_err("\n");
 	}
 
 	describe_object_addr(cache, object, addr);

@@ -229,6 +231,7 @@ static void print_address_description(void *addr)
 	struct page *page = addr_to_page(addr);
 
 	dump_stack();
+	pr_err("\n");
 
 	if (page && PageSlab(page)) {
 		struct kmem_cache *cache = page->slab_cache;

@@ -307,7 +310,9 @@ void kasan_report_double_free(struct kmem_cache *cache, void *object,
 
 	kasan_start_report(&flags);
 	pr_err("BUG: KASAN: double-free or invalid-free in %pS\n", ip);
+	pr_err("\n");
 	print_address_description(object);
+	pr_err("\n");
 	print_shadow_for_address(object);
 	kasan_end_report(&flags);
 }

@@ -319,11 +324,13 @@ static void kasan_report_error(struct kasan_access_info *info)
 	kasan_start_report(&flags);
 
 	print_error_description(info);
+	pr_err("\n");
 
 	if (!addr_has_shadow(info)) {
 		dump_stack();
 	} else {
 		print_address_description((void *)info->access_addr);
+		pr_err("\n");
 		print_shadow_for_address(info->first_bad_addr);
 	}

-- 
2.12.1.578.ge9c3154ca4-goog

[PATCH v4 7/9] kasan: print page description after stacks

From: Andrey Konovalov <hidden>
Date: 2017-03-28 14:01:32

Moves page description after the stacks since it's less important.

Signed-off-by: Andrey Konovalov <redacted>
---
 mm/kasan/report.c | 14 ++++++++------
 1 file changed, 8 insertions(+), 6 deletions(-)

diff --git a/mm/kasan/report.c b/mm/kasan/report.c
index 06e27a342d1d..e0f7dbf9e883 100644
--- a/mm/kasan/report.c
+++ b/mm/kasan/report.c

@@ -242,9 +242,6 @@ static void print_address_description(struct kasan_access_info *info)
 	void *addr = (void *)info->access_addr;
 	struct page *page = addr_to_page(addr);
 
-	if (page)
-		dump_page(page, "kasan: bad access detected");
-
 	dump_stack();
 
 	if (page && PageSlab(page)) {

@@ -254,9 +251,14 @@ static void print_address_description(struct kasan_access_info *info)
 		describe_object(cache, object, addr);
 	}
 
-	if (kernel_or_module_addr(addr)) {
-		if (!init_task_stack_addr(addr))
-			pr_err("Address belongs to variable %pS\n", addr);
+	if (kernel_or_module_addr(addr) && !init_task_stack_addr(addr)) {
+		pr_err("The buggy address belongs to the variable:\n");
+		pr_err(" %pS\n", addr);
+	}
+
+	if (page) {
+		pr_err("The buggy address belongs to the page:\n");
+		dump_page(page, "kasan: bad access detected");
 	}
 }

-- 
2.12.1.578.ge9c3154ca4-goog

[PATCH v4 2/9] kasan: unify report headers

From: Andrey Konovalov <hidden>
Date: 2017-03-28 14:01:32

Unify KASAN report header format for different kinds of bad memory
accesses. Makes the code simpler.

Signed-off-by: Andrey Konovalov <redacted>
---
 mm/kasan/report.c | 26 +++++++++++++-------------
 1 file changed, 13 insertions(+), 13 deletions(-)

diff --git a/mm/kasan/report.c b/mm/kasan/report.c
index e3af37b7a74c..fc0577d15671 100644
--- a/mm/kasan/report.c
+++ b/mm/kasan/report.c

@@ -119,16 +119,22 @@ static const char *get_wild_bug_type(struct kasan_access_info *info)
 	return bug_type;
 }
 
+static const char *get_bug_type(struct kasan_access_info *info)
+{
+	if (addr_has_shadow(info))
+		return get_shadow_bug_type(info);
+	return get_wild_bug_type(info);
+}
+
 static void print_error_description(struct kasan_access_info *info)
 {
-	const char *bug_type = get_shadow_bug_type(info);
+	const char *bug_type = get_bug_type(info);
 
 	pr_err("BUG: KASAN: %s in %pS at addr %p\n",
-		bug_type, (void *)info->ip,
-		info->access_addr);
+		bug_type, (void *)info->ip, info->access_addr);
 	pr_err("%s of size %zu by task %s/%d\n",
-		info->is_write ? "Write" : "Read",
-		info->access_size, current->comm, task_pid_nr(current));
+		info->is_write ? "Write" : "Read", info->access_size,
+		current->comm, task_pid_nr(current));
 }
 
 static inline bool kernel_or_module_addr(const void *addr)

@@ -295,17 +301,11 @@ static void kasan_report_error(struct kasan_access_info *info)
 
 	kasan_start_report(&flags);
 
+	print_error_description(info);
+
 	if (!addr_has_shadow(info)) {
-		const char *bug_type = get_wild_bug_type(info);
-		pr_err("BUG: KASAN: %s on address %p\n",
-			bug_type, info->access_addr);
-		pr_err("%s of size %zu by task %s/%d\n",
-			info->is_write ? "Write" : "Read",
-			info->access_size, current->comm,
-			task_pid_nr(current));
 		dump_stack();
 	} else {
-		print_error_description(info);
 		print_address_description(info);
 		print_shadow_for_address(info->first_bad_addr);
 	}

-- 
2.12.1.578.ge9c3154ca4-goog

[PATCH v4 6/9] kasan: improve slab object description

From: Andrey Konovalov <hidden>
Date: 2017-03-28 14:01:36

Changes slab object description from:

Object at ffff880068388540, in cache kmalloc-128 size: 128

to:

Object at ffff88006a2d5a80 belongs to cache kmalloc-128 of size 128
 accessed at offset 123

This adds information about relative offset of the accessed address to
the start of the object.

Signed-off-by: Andrey Konovalov <redacted>
---
 mm/kasan/report.c | 38 +++++++++++++++++++++++++++-----------
 1 file changed, 27 insertions(+), 11 deletions(-)

diff --git a/mm/kasan/report.c b/mm/kasan/report.c
index 156f998199e2..06e27a342d1d 100644
--- a/mm/kasan/report.c
+++ b/mm/kasan/report.c

@@ -194,18 +194,34 @@ static struct page *addr_to_page(const void *addr)
 	return NULL;
 }
 
-static void describe_object(struct kmem_cache *cache, void *object)
+static void describe_object_addr(struct kmem_cache *cache, void *object,
+				const void *addr)
 {
-	struct kasan_alloc_meta *alloc_info = get_alloc_info(cache, object);
+	unsigned long access_addr = (unsigned long)addr;
+	unsigned long object_addr = (unsigned long)object;
+	const char *rel_type;
+	int rel_bytes;
 
-	pr_err("Object at %p, in cache %s size: %d\n", object, cache->name,
-		cache->object_size);
+	pr_err("Object at %p belongs to cache %s of size %d\n",
+		object, cache->name, cache->object_size);
 
-	if (!(cache->flags & SLAB_KASAN))
+	if (!addr)
 		return;
 
-	print_track(&alloc_info->alloc_track, "Allocated");
-	print_track(&alloc_info->free_track, "Freed");
+	pr_err(" accessed at offset %d\n", access_addr - object_addr);
+}
+
+static void describe_object(struct kmem_cache *cache, void *object,
+				const void *addr)
+{
+	struct kasan_alloc_meta *alloc_info = get_alloc_info(cache, object);
+
+	if (cache->flags & SLAB_KASAN) {
+		print_track(&alloc_info->alloc_track, "Allocated");
+		print_track(&alloc_info->free_track, "Freed");
+	}
+
+	describe_object_addr(cache, object, addr);
 }
 
 void kasan_report_double_free(struct kmem_cache *cache, void *object,

@@ -217,13 +233,13 @@ void kasan_report_double_free(struct kmem_cache *cache, void *object,
 	pr_err("BUG: Double free or freeing an invalid pointer\n");
 	pr_err("Unexpected shadow byte: 0x%hhX\n", shadow);
 	dump_stack();
-	describe_object(cache, object);
+	describe_object(cache, object, NULL);
 	kasan_end_report(&flags);
 }
 
 static void print_address_description(struct kasan_access_info *info)
 {
-	const void *addr = info->access_addr;
+	void *addr = (void *)info->access_addr;
 	struct page *page = addr_to_page(addr);
 
 	if (page)

@@ -233,9 +249,9 @@ static void print_address_description(struct kasan_access_info *info)
 
 	if (page && PageSlab(page)) {
 		struct kmem_cache *cache = page->slab_cache;
-		void *object = nearest_obj(cache, page,	(void *)addr);
+		void *object = nearest_obj(cache, page,	addr);
 
-		describe_object(cache, object);
+		describe_object(cache, object, addr);
 	}
 
 	if (kernel_or_module_addr(addr)) {

-- 
2.12.1.578.ge9c3154ca4-goog

[PATCH 3/5] net/packet: fix overflow in check for tp_frame_nr

From: Andrey Konovalov <hidden>
Date: 2017-03-28 14:01:36

When calculating rb->frames_per_block * req->tp_block_nr the result
can overflow.

Add a check that tp_block_size * tp_block_nr <= UINT_MAX.

Since frames_per_block <= tp_block_size, the expression would
never overflow.

Signed-off-by: Andrey Konovalov <redacted>
---
 net/packet/af_packet.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/net/packet/af_packet.c b/net/packet/af_packet.c
index 506348abdf2f..c5c43fff8c01 100644
--- a/net/packet/af_packet.c
+++ b/net/packet/af_packet.c

@@ -4197,6 +4197,9 @@ static int packet_set_ring(struct sock *sk, union tpacket_req_u *req_u,
 			goto out;
 		if (unlikely(req->tp_frame_size == 0))
 			goto out;
+		if (unlikely((u64)req->tp_block_size * req->tp_block_nr >
+					UINT_MAX))
+			goto out;
 
 		if (unlikely(!PAGE_ALIGNED(req->tp_block_size)))
 			goto out;

-- 
2.12.2.564.g063fe858b8-goog

Re: [PATCH 0/5] net/packet: fix multiple overflow issues in ring buffers

From: Willem de Bruijn <willemdebruijn.kernel@gmail.com>
Date: 2017-03-28 15:07:14

On Tue, Mar 28, 2017 at 10:00 AM, Andrey Konovalov
[off-list ref] wrote:

This patchset addresses multiple overflows and signedness-related issues
in packet socket ring buffers.

Andrey Konovalov (5):
  net/packet: fix overflow in check for priv area size
  net/packet: add explicit checks for tp_frame_size
  net/packet: fix overflow in check for tp_frame_nr
  net/packet: fix overflow in check for tp_reserve
  net/packet: reorder checks for ring buffer parameters

These are a lot of changes to backport to stable kernels.

Can we separate the minimal patch set needed to address known overflow
to send to net (with annotation [PATCH net]) and follow up with the larger
cleanup to net-next.

 net/packet/af_packet.c | 31 +++++++++++++++++++++----------
 1 file changed, 21 insertions(+), 10 deletions(-)

--
2.12.2.564.g063fe858b8-goog

Re: [PATCH 0/5] net/packet: fix multiple overflow issues in ring buffers

From: Andrey Konovalov <hidden>
Date: 2017-03-28 15:16:00

On Tue, Mar 28, 2017 at 5:06 PM, Willem de Bruijn
[off-list ref] wrote:

On Tue, Mar 28, 2017 at 10:00 AM, Andrey Konovalov
[off-list ref] wrote:

quoted

This patchset addresses multiple overflows and signedness-related issues
in packet socket ring buffers.

Andrey Konovalov (5):
  net/packet: fix overflow in check for priv area size
  net/packet: add explicit checks for tp_frame_size
  net/packet: fix overflow in check for tp_frame_nr
  net/packet: fix overflow in check for tp_reserve
  net/packet: reorder checks for ring buffer parameters

These are a lot of changes to backport to stable kernels.

Can we separate the minimal patch set needed to address known overflow
to send to net (with annotation [PATCH net]) and follow up with the larger
cleanup to net-next.

Sure, I can put patches 2 and 5 to a separate patchset.

quoted

 net/packet/af_packet.c | 31 +++++++++++++++++++++----------
 1 file changed, 21 insertions(+), 10 deletions(-)

--
2.12.2.564.g063fe858b8-goog

`h`	back out one level
`j`	next message in thread
`k`	previous message in thread
`l`	drill in
`Esc`	close help / fold thread tree
`?`	toggle this help