[PATCH V4 1/2] ptr_ring: fail early if queue occupies more than KMALLOC_MAX_SIZE

Subsystems: the rest

STALE3136d

9 messages, 4 authors, 2018-02-11 · open the first message on its own page

[PATCH V4 1/2] ptr_ring: fail early if queue occupies more than KMALLOC_MAX_SIZE

From: Jason Wang <hidden>
Date: 2018-02-09 09:45:56

To avoid slab to warn about exceeded size, fail early if queue
occupies more than KMALLOC_MAX_SIZE.

Reported-by: syzbot+e4d4f9ddd4295539735d@syzkaller.appspotmail.com
Fixes: 2e0ab8ca83c12 ("ptr_ring: array based FIFO for pointers")
Signed-off-by: Jason Wang <redacted>
---
 include/linux/ptr_ring.h | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/include/linux/ptr_ring.h b/include/linux/ptr_ring.h
index 1883d61..6051a5f 100644
--- a/include/linux/ptr_ring.h
+++ b/include/linux/ptr_ring.h

@@ -466,6 +466,8 @@ static inline int ptr_ring_consume_batched_bh(struct ptr_ring *r,
 
 static inline void **__ptr_ring_init_queue_alloc(unsigned int size, gfp_t gfp)
 {
+	if (size * sizeof(void *) > KMALLOC_MAX_SIZE)
+		return NULL;
 	return kcalloc(size, sizeof(void *), gfp);
 }

-- 
2.7.4

[PATCH V4 2/2] ptr_ring: try vmalloc() when kmalloc() fails

From: Jason Wang <hidden>
Date: 2018-02-09 09:46:00

This patch switch to use kvmalloc_array() for using a vmalloc()
fallback to help in case kmalloc() fails.

Reported-by: syzbot+e4d4f9ddd4295539735d@syzkaller.appspotmail.com
Fixes: 2e0ab8ca83c12 ("ptr_ring: array based FIFO for pointers")
Signed-off-by: Jason Wang <redacted>
---
 include/linux/ptr_ring.h | 13 ++++++++-----
 1 file changed, 8 insertions(+), 5 deletions(-)

diff --git a/include/linux/ptr_ring.h b/include/linux/ptr_ring.h
index 6051a5f..b884b77 100644
--- a/include/linux/ptr_ring.h
+++ b/include/linux/ptr_ring.h

@@ -464,11 +464,14 @@ static inline int ptr_ring_consume_batched_bh(struct ptr_ring *r,
 	__PTR_RING_PEEK_CALL_v; \
 })
 
+/* Not all gfp_t flags (besides GFP_KERNEL) are allowed. See
+ * documentation for vmalloc for which of them are legal.
+ */
 static inline void **__ptr_ring_init_queue_alloc(unsigned int size, gfp_t gfp)
 {
 	if (size * sizeof(void *) > KMALLOC_MAX_SIZE)
 		return NULL;
-	return kcalloc(size, sizeof(void *), gfp);
+	return kvmalloc_array(size, sizeof(void *), gfp | __GFP_ZERO);
 }
 
 static inline void __ptr_ring_set_size(struct ptr_ring *r, int size)

@@ -603,7 +606,7 @@ static inline int ptr_ring_resize(struct ptr_ring *r, int size, gfp_t gfp,
 	spin_unlock(&(r)->producer_lock);
 	spin_unlock_irqrestore(&(r)->consumer_lock, flags);
 
-	kfree(old);
+	kvfree(old);
 
 	return 0;
 }

@@ -643,7 +646,7 @@ static inline int ptr_ring_resize_multiple(struct ptr_ring **rings,
 	}
 
 	for (i = 0; i < nrings; ++i)
-		kfree(queues[i]);
+		kvfree(queues[i]);
 
 	kfree(queues);

@@ -651,7 +654,7 @@ static inline int ptr_ring_resize_multiple(struct ptr_ring **rings,
 
 nomem:
 	while (--i >= 0)
-		kfree(queues[i]);
+		kvfree(queues[i]);
 
 	kfree(queues);

@@ -666,7 +669,7 @@ static inline void ptr_ring_cleanup(struct ptr_ring *r, void (*destroy)(void *))
 	if (destroy)
 		while ((ptr = ptr_ring_consume(r)))
 			destroy(ptr);
-	kfree(r->queue);
+	kvfree(r->queue);
 }
 
 #endif /* _LINUX_PTR_RING_H  */

-- 
2.7.4

Re: [PATCH V4 1/2] ptr_ring: fail early if queue occupies more than KMALLOC_MAX_SIZE

From: "Michael S. Tsirkin" <mst@redhat.com>
Date: 2018-02-09 15:39:13

On Fri, Feb 09, 2018 at 05:45:49PM +0800, Jason Wang wrote:

To avoid slab to warn about exceeded size, fail early if queue
occupies more than KMALLOC_MAX_SIZE.

Reported-by: syzbot+e4d4f9ddd4295539735d@syzkaller.appspotmail.com
Fixes: 2e0ab8ca83c12 ("ptr_ring: array based FIFO for pointers")
Signed-off-by: Jason Wang <redacted>

Acked-by: Michael S. Tsirkin <mst@redhat.com>

quoted hunk

---
 include/linux/ptr_ring.h | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/include/linux/ptr_ring.h b/include/linux/ptr_ring.h
index 1883d61..6051a5f 100644
--- a/include/linux/ptr_ring.h
+++ b/include/linux/ptr_ring.h

@@ -466,6 +466,8 @@ static inline int ptr_ring_consume_batched_bh(struct ptr_ring *r,
 
 static inline void **__ptr_ring_init_queue_alloc(unsigned int size, gfp_t gfp)
 {
+	if (size * sizeof(void *) > KMALLOC_MAX_SIZE)
+		return NULL;
 	return kcalloc(size, sizeof(void *), gfp);
 }

-- 
2.7.4

Re: [PATCH V4 2/2] ptr_ring: try vmalloc() when kmalloc() fails

From: "Michael S. Tsirkin" <mst@redhat.com>
Date: 2018-02-09 15:53:53

On Fri, Feb 09, 2018 at 05:45:50PM +0800, Jason Wang wrote:

This patch switch to use kvmalloc_array() for using a vmalloc()
fallback to help in case kmalloc() fails.

Above isn't really saying anything about the motivation, it
just explains what kvmalloc_array does.

How about:

Switch ptr_ring from kmalloc to kvmalloc. This way it can support larger
ring sizes.

Reported-by: syzbot+e4d4f9ddd4295539735d@syzkaller.appspotmail.com
Fixes: 2e0ab8ca83c12 ("ptr_ring: array based FIFO for pointers")

I don't see this as a bugfix - it's more of an enhancement.

Do we need this in net or can we defer this to net-next?

I'm not sure myself. For now

Acked-by: Michael S. Tsirkin <mst@redhat.com>

Signed-off-by: Jason Wang <redacted>

quoted hunk

---
 include/linux/ptr_ring.h | 13 ++++++++-----
 1 file changed, 8 insertions(+), 5 deletions(-)

diff --git a/include/linux/ptr_ring.h b/include/linux/ptr_ring.h
index 6051a5f..b884b77 100644
--- a/include/linux/ptr_ring.h
+++ b/include/linux/ptr_ring.h

@@ -464,11 +464,14 @@ static inline int ptr_ring_consume_batched_bh(struct ptr_ring *r,
 	__PTR_RING_PEEK_CALL_v; \
 })
 
+/* Not all gfp_t flags (besides GFP_KERNEL) are allowed. See
+ * documentation for vmalloc for which of them are legal.
+ */
 static inline void **__ptr_ring_init_queue_alloc(unsigned int size, gfp_t gfp)
 {
 	if (size * sizeof(void *) > KMALLOC_MAX_SIZE)
 		return NULL;
-	return kcalloc(size, sizeof(void *), gfp);
+	return kvmalloc_array(size, sizeof(void *), gfp | __GFP_ZERO);
 }
 
 static inline void __ptr_ring_set_size(struct ptr_ring *r, int size)

@@ -603,7 +606,7 @@ static inline int ptr_ring_resize(struct ptr_ring *r, int size, gfp_t gfp,
 	spin_unlock(&(r)->producer_lock);
 	spin_unlock_irqrestore(&(r)->consumer_lock, flags);
 
-	kfree(old);
+	kvfree(old);
 
 	return 0;
 }

@@ -643,7 +646,7 @@ static inline int ptr_ring_resize_multiple(struct ptr_ring **rings,
 	}
 
 	for (i = 0; i < nrings; ++i)
-		kfree(queues[i]);
+		kvfree(queues[i]);
 
 	kfree(queues);

@@ -651,7 +654,7 @@ static inline int ptr_ring_resize_multiple(struct ptr_ring **rings,
 
 nomem:
 	while (--i >= 0)
-		kfree(queues[i]);
+		kvfree(queues[i]);
 
 	kfree(queues);

@@ -666,7 +669,7 @@ static inline void ptr_ring_cleanup(struct ptr_ring *r, void (*destroy)(void *))
 	if (destroy)
 		while ((ptr = ptr_ring_consume(r)))
 			destroy(ptr);
-	kfree(r->queue);
+	kvfree(r->queue);
 }
 
 #endif /* _LINUX_PTR_RING_H  */

-- 
2.7.4

Re: [PATCH V4 1/2] ptr_ring: fail early if queue occupies more than KMALLOC_MAX_SIZE

From: David Miller <davem@davemloft.net>
Date: 2018-02-09 19:29:24

From: Jason Wang <redacted>
Date: Fri,  9 Feb 2018 17:45:49 +0800

To avoid slab to warn about exceeded size, fail early if queue
occupies more than KMALLOC_MAX_SIZE.

Reported-by: syzbot+e4d4f9ddd4295539735d@syzkaller.appspotmail.com
Fixes: 2e0ab8ca83c12 ("ptr_ring: array based FIFO for pointers")
Signed-off-by: Jason Wang <redacted>

Applied.

Re: [PATCH V4 2/2] ptr_ring: try vmalloc() when kmalloc() fails

From: David Miller <davem@davemloft.net>
Date: 2018-02-09 19:29:33

From: Jason Wang <redacted>
Date: Fri,  9 Feb 2018 17:45:50 +0800

This patch switch to use kvmalloc_array() for using a vmalloc()
fallback to help in case kmalloc() fails.

Reported-by: syzbot+e4d4f9ddd4295539735d@syzkaller.appspotmail.com
Fixes: 2e0ab8ca83c12 ("ptr_ring: array based FIFO for pointers")
Signed-off-by: Jason Wang <redacted>

Applied.

Re: [PATCH V4 1/2] ptr_ring: fail early if queue occupies more than KMALLOC_MAX_SIZE

From: Eric Biggers <hidden>
Date: 2018-02-10 19:32:07

Hi Jason,

On Fri, Feb 09, 2018 at 05:45:49PM +0800, Jason Wang wrote:

quoted hunk

To avoid slab to warn about exceeded size, fail early if queue
occupies more than KMALLOC_MAX_SIZE.

Reported-by: syzbot+e4d4f9ddd4295539735d@syzkaller.appspotmail.com
Fixes: 2e0ab8ca83c12 ("ptr_ring: array based FIFO for pointers")
Signed-off-by: Jason Wang <redacted>
---
 include/linux/ptr_ring.h | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/include/linux/ptr_ring.h b/include/linux/ptr_ring.h
index 1883d61..6051a5f 100644
--- a/include/linux/ptr_ring.h
+++ b/include/linux/ptr_ring.h

@@ -466,6 +466,8 @@ static inline int ptr_ring_consume_batched_bh(struct ptr_ring *r,
 
 static inline void **__ptr_ring_init_queue_alloc(unsigned int size, gfp_t gfp)
 {
+	if (size * sizeof(void *) > KMALLOC_MAX_SIZE)
+		return NULL;

Are you sure that size can't be over 0x40000000?  The proper way to write this
(safe from integer overflow) would be:

	if (size > KMALLOC_MAX_SIZE / sizeof(void *))
		return NULL;

- Eric

Re: [PATCH V4 2/2] ptr_ring: try vmalloc() when kmalloc() fails

From: Jason Wang <hidden>
Date: 2018-02-11 02:46:01


On 2018年02月09日 23:53, Michael S. Tsirkin wrote:

On Fri, Feb 09, 2018 at 05:45:50PM +0800, Jason Wang wrote:

quoted

This patch switch to use kvmalloc_array() for using a vmalloc()
fallback to help in case kmalloc() fails.

Above isn't really saying anything about the motivation, it
just explains what kvmalloc_array does.

How about:

Switch ptr_ring from kmalloc to kvmalloc. This way it can support larger
ring sizes.

quoted

Reported-by:syzbot+e4d4f9ddd4295539735d@syzkaller.appspotmail.com
Fixes: 2e0ab8ca83c12 ("ptr_ring: array based FIFO for pointers")

I don't see this as a bugfix - it's more of an enhancement.

Do we need this in net or can we defer this to net-next?

Consider they are users of tx_queue_len of more than 1000000 from google 
searching. We'd better have this in -net.

Thanks

I'm not sure myself. For now

Acked-by: Michael S. Tsirkin<mst@redhat.com>

Re: [PATCH V4 1/2] ptr_ring: fail early if queue occupies more than KMALLOC_MAX_SIZE

From: Jason Wang <hidden>
Date: 2018-02-11 02:46:40


On 2018年02月11日 03:32, Eric Biggers wrote:

Hi Jason,

On Fri, Feb 09, 2018 at 05:45:49PM +0800, Jason Wang wrote:

quoted

To avoid slab to warn about exceeded size, fail early if queue
occupies more than KMALLOC_MAX_SIZE.

Reported-by: syzbot+e4d4f9ddd4295539735d@syzkaller.appspotmail.com
Fixes: 2e0ab8ca83c12 ("ptr_ring: array based FIFO for pointers")
Signed-off-by: Jason Wang <redacted>
---
  include/linux/ptr_ring.h | 2 ++
  1 file changed, 2 insertions(+)

diff --git a/include/linux/ptr_ring.h b/include/linux/ptr_ring.h
index 1883d61..6051a5f 100644
--- a/include/linux/ptr_ring.h
+++ b/include/linux/ptr_ring.h

@@ -466,6 +466,8 @@ static inline int ptr_ring_consume_batched_bh(struct ptr_ring *r,
  
  static inline void **__ptr_ring_init_queue_alloc(unsigned int size, gfp_t gfp)
  {
+	if (size * sizeof(void *) > KMALLOC_MAX_SIZE)
+		return NULL;

Are you sure that size can't be over 0x40000000?  The proper way to write this
(safe from integer overflow) would be:

	if (size > KMALLOC_MAX_SIZE / sizeof(void *))
		return NULL;

- Eric

Good catch.

Will post a fix.

Thanks

`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