Skip to contentSkip to conversation

sctp: fix incorrect overflow check on autoclose

PATCH RESEND Superseded Xi Wang 2 participants 14 messages Dec 9, 2011 – Jan 3, 2012

Subsystems (3)
Networking Network drivers SCTP protocol
lore
Share & tools
lore

Xi Wang

author
#0 lore raw

PATCH RESEND 1/1 sctp: fix incorrect overflow check on autoclose

The commit 8ffd3208 voids the previous patches f6778aab and 810c0719 for limiting the maximum autoclose value. If userspace passes in -1 on 32-bit platform, the overflow check didn't work and autoclose would be set to 0xffffffff.

Signed-off-by Xi Wang <[email protected]>
Cc Andrew Morton <[email protected]>
Cc Andrei Pelinescu-Onciul <[email protected]>
Cc Vlad Yasevich <[email protected]>
Cc David S. Miller <[email protected]>
---
 net/sctp/socket.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)
net/sctp/socket.c +2 -1
index 13bf5fc..bb91281 100644
@@ -2201,7 +2201,8 @@ static int sctp_setsockopt_autoclose(struct sock *sk, char __user *optval,
 	if (copy_from_user(&sp->autoclose, optval, optlen))
 		return -EFAULT;
 	/* make sure it won't exceed MAX_SCHEDULE_TIMEOUT */
-	sp->autoclose = min_t(long, sp->autoclose, MAX_SCHEDULE_TIMEOUT / HZ);
+	sp->autoclose = min_t(unsigned long, sp->autoclose,
+			      MAX_SCHEDULE_TIMEOUT / HZ);
 
 	return 0;
 }
Signature · 1 line
1.7.5.4
#1 lore raw
Quoted 19 lines · Xi Wang

The commit 8ffd3208 voids the previous patches f6778aab and 810c0719 for limiting the maximum autoclose value. If userspace passes in -1 on 32-bit platform, the overflow check didn't work and autoclose would be set to 0xffffffff.

Signed-off-by Xi Wang <[email protected]>
Cc Andrew Morton <[email protected]>
Cc Andrei Pelinescu-Onciul <[email protected]>
Cc Vlad Yasevich <[email protected]>
Cc David S. Miller <[email protected]>
---
 net/sctp/socket.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)
net/sctp/socket.c +0 -0
index 13bf5fc..bb91281 100644

net/sctp/socket.c static int sctp_setsockopt_autoclose(struct sock *sk, char __user *optval, · line 2205

Quoted 6 lines
net/sctp/socket.c +2 -1
@@ -2201,7 +2201,8 @@ static int sctp_setsockopt_autoclose(struct sock *sk, char __user *optval,
 	if (copy_from_user(&sp->autoclose, optval, optlen))
 		return -EFAULT;
 	/* make sure it won't exceed MAX_SCHEDULE_TIMEOUT */
-	sp->autoclose = min_t(long, sp->autoclose, MAX_SCHEDULE_TIMEOUT / HZ);
+	sp->autoclose = min_t(unsigned long, sp->autoclose,
+			      MAX_SCHEDULE_TIMEOUT / HZ);

I think this should be u32 since that's what sp->autoclose is.

-vlad

Quoted 3 lines
net/sctp/socket.c +0 -0
 	return 0;
 }
author
#2 lore raw
Quoted 1 line · Vladislav Yasevich

I think this should be u32 since that's what sp->autoclose is.

Do you mean something like this?

	min_t(u32, sp->autoclose, MAX_SCHEDULE_TIMEOUT / HZ)

On 64-bit platform this would limit autoclose for no good reason to

	(u32)(MAX_SCHEDULE_TIMEOUT / HZ),

which is basically 0x978d4fdf (assuming HZ is 250). I guess the intention was to allow autoclose to be any u32 on 64-bit platform.

- xi

#3 lore raw
Quoted 7 lines · Xi Wang
Quoted 1 line · Vladislav Yasevich

I think this should be u32 since that's what sp->autoclose is.

Do you mean something like this?

	min_t(u32, sp->autoclose, MAX_SCHEDULE_TIMEOUT / HZ)
Quoted 8 lines

On 64-bit platform this would limit autoclose for no good reason to

	(u32)(MAX_SCHEDULE_TIMEOUT / HZ),

which is basically 0x978d4fdf (assuming HZ is 250). I guess the intention was to allow autoclose to be any u32 on 64-bit platform.

Hm.. this is a bit strange. This makes it so that on 32 bit platforms we have one upper bound for autoclose and on 64 we have another even though the type is platform dependent. This could be considered a regression by applications.

In addition this would result in confusion to user since the values between setsockopt() and getsockopt() for autoclose would be different.

Looking at the latest spec, it actually looks like we are completely mis-interpreting autoclose. It's supposed to be in seconds.

For now, I'd suggest to make this consistent between 32 and 64 bits. Having inconsistent values seems strange.

As a next set the api needs to be fixed to accept seconds as argument.

-vlad

author
#4 lore raw
Quoted 4 lines · Vladislav Yasevich

Hm.. this is a bit strange. This makes it so that on 32 bit platforms we have one upper bound for autoclose and on 64 we have another even though the type is platform dependent. This could be considered a regression by applications.

Either looks good to me. Timeout limit is essentially different on 32/64 platforms.

Another (probably uglier) option is to limit the value on 32-bit platform only, like sock_setsockopt() in net/core/sock.c.

#if (BITS_PER_LONG = 32)
	if (sp->autoclose > MAX_SCHEDULE_TIMEOUT / HZ)
		sp->autoclose = MAX_SCHEDULE_TIMEOUT / HZ;

#endif

Quoted 2 lines

In addition this would result in confusion to user since the values between setsockopt() and getsockopt() for autoclose would be different.

Are you suggesting to reject the value and return -EINVAL, rather than silently limiting the autoclose value?

- xi

#5 lore raw
Quoted 8 lines · Xi Wang
Quoted 4 lines · Vladislav Yasevich

Hm.. this is a bit strange. This makes it so that on 32 bit platforms we have one upper bound for autoclose and on 64 we have another even though the type is platform dependent. This could be considered a regression by applications.

Either looks good to me. Timeout limit is essentially different on 32/64 platforms.

I don't think it really should be different. Notice that our rto values remain consistent. I really thing that this should be consistent from the user's point of view.

Quoted 8 lines

Another (probably uglier) option is to limit the value on 32-bit platform only, like sock_setsockopt() in net/core/sock.c.

#if (BITS_PER_LONG = 32)
	if (sp->autoclose > MAX_SCHEDULE_TIMEOUT / HZ)
		sp->autoclose = MAX_SCHEDULE_TIMEOUT / HZ;

#endif

I agree, this is ugly. It might make more sense to define a max autoclose value and expose it through /sys. That way the values remains consistent.

-vlad

Quoted 9 lines
Quoted 3 lines

In addition this would result in confusion to user since the values between setsockopt() and getsockopt() for autoclose would be different.

Are you suggesting to reject the value and return -EINVAL, rather than silently limiting the autoclose value?

- xi

author
#6 lore raw
Quoted 2 lines · Vladislav Yasevich

I agree, this is ugly. It might make more sense to define a max autoclose value and expose it through /sys. That way the values remains consistent.

Hmm, this sounds nice. I will give it a try in v2.

- xi

author
#7 lore raw

PATCH v2 1/1 sctp: fix incorrect overflow check on autoclose

In reply to Xi Wang

Commit 8ffd3208 voids the previous patches f6778aab and 810c0719 for limiting the maximum autoclose value. If userspace passes in -1 on 32-bit platform, the overflow check didn't work and autoclose would be set to 0xffffffff.

This patch defines a max_autoclose for limiting the value and exposes it through sysctl.

Suggested-by Vlad Yasevich <[email protected]>
Signed-off-by Xi Wang <[email protected]>
---
 include/net/sctp/structs.h |    4 ++++
 net/sctp/protocol.c        |    3 +++
 net/sctp/socket.c          |    4 ++--
 net/sctp/sysctl.c          |    7 +++++++
 4 files changed, 16 insertions(+), 2 deletions(-)
include/net/sctp/structs.h +4 -0
index e90e7a9..781f16d 100644
@@ -241,6 +241,9 @@ extern struct sctp_globals {
 	 * bits is an indicator of when to send and window update SACK.
 	 */
 	int rwnd_update_shift;
+
+	/* Threshold for autoclose timeout. */
+	unsigned int max_autoclose;
 } sctp_globals;
 
 #define sctp_rto_initial		(sctp_globals.rto_initial)

@@ -281,6 +284,7 @@ extern struct sctp_globals {
 #define sctp_auth_enable		(sctp_globals.auth_enable)
 #define sctp_checksum_disable		(sctp_globals.checksum_disable)
 #define sctp_rwnd_upd_shift		(sctp_globals.rwnd_update_shift)
+#define sctp_max_autoclose		(sctp_globals.max_autoclose)
 
 /* SCTP Socket type: UDP or TCP style. */
 typedef enum {
net/sctp/protocol.c +3 -0
index 61b9fca..4e07b18 100644
@@ -1285,6 +1285,9 @@ SCTP_STATIC __init int sctp_init(void)
 	sctp_max_instreams    		= SCTP_DEFAULT_INSTREAMS;
 	sctp_max_outstreams   		= SCTP_DEFAULT_OUTSTREAMS;
 
+	/* Initialize maximum autoclose timeout. */
+	sctp_max_autoclose		= INT_MAX;
+
 	/* Initialize handle used for association ids. */
 	idr_init(&sctp_assocs_id);
net/sctp/socket.c +2 -2
index 13bf5fc..f8c8e66 100644
@@ -2200,8 +2200,8 @@ static int sctp_setsockopt_autoclose(struct sock *sk, char __user *optval,
 		return -EINVAL;
 	if (copy_from_user(&sp->autoclose, optval, optlen))
 		return -EFAULT;
-	/* make sure it won't exceed MAX_SCHEDULE_TIMEOUT */
-	sp->autoclose = min_t(long, sp->autoclose, MAX_SCHEDULE_TIMEOUT / HZ);
+	/* make sure it won't exceed sctp_max_autoclose / HZ */
+	sp->autoclose = min(sp->autoclose, sctp_max_autoclose / HZ);
 
 	return 0;
 }
net/sctp/sysctl.c +7 -0
index 6b39529..b74686e 100644
@@ -258,6 +258,13 @@ static ctl_table sctp_table[] = {
 		.extra1		= &one,
 		.extra2		= &rwnd_scale_max,
 	},
+	{
+		.procname	= "max_autoclose",
+		.data		= &sctp_max_autoclose,
+		.maxlen		= sizeof(unsigned int),
+		.mode		= 0644,
+		.proc_handler	= &proc_dointvec_jiffies,
+	},
 
 	{ /* sentinel */ }
 };
Signature · 1 line
1.7.5.4
#8 lore raw
Quoted 74 lines · Xi Wang

Commit 8ffd3208 voids the previous patches f6778aab and 810c0719 for limiting the maximum autoclose value. If userspace passes in -1 on 32-bit platform, the overflow check didn't work and autoclose would be set to 0xffffffff.

This patch defines a max_autoclose for limiting the value and exposes it through sysctl.

Suggested-by Vlad Yasevich <[email protected]>
Signed-off-by Xi Wang <[email protected]>
---
 include/net/sctp/structs.h |    4 ++++
 net/sctp/protocol.c        |    3 +++
 net/sctp/socket.c          |    4 ++--
 net/sctp/sysctl.c          |    7 +++++++
 4 files changed, 16 insertions(+), 2 deletions(-)
include/net/sctp/structs.h +4 -0
index e90e7a9..781f16d 100644
@@ -241,6 +241,9 @@ extern struct sctp_globals {
 	 * bits is an indicator of when to send and window update SACK.
 	 */
 	int rwnd_update_shift;
+
+	/* Threshold for autoclose timeout. */
+	unsigned int max_autoclose;
 } sctp_globals;
 
 #define sctp_rto_initial		(sctp_globals.rto_initial)

@@ -281,6 +284,7 @@ extern struct sctp_globals {
 #define sctp_auth_enable		(sctp_globals.auth_enable)
 #define sctp_checksum_disable		(sctp_globals.checksum_disable)
 #define sctp_rwnd_upd_shift		(sctp_globals.rwnd_update_shift)
+#define sctp_max_autoclose		(sctp_globals.max_autoclose)
 
 /* SCTP Socket type: UDP or TCP style. */
 typedef enum {
net/sctp/protocol.c +3 -0
index 61b9fca..4e07b18 100644
@@ -1285,6 +1285,9 @@ SCTP_STATIC __init int sctp_init(void)
 	sctp_max_instreams    		= SCTP_DEFAULT_INSTREAMS;
 	sctp_max_outstreams   		= SCTP_DEFAULT_OUTSTREAMS;
 
+	/* Initialize maximum autoclose timeout. */
+	sctp_max_autoclose		= INT_MAX;
+
 	/* Initialize handle used for association ids. */
 	idr_init(&sctp_assocs_id);
net/sctp/socket.c +2 -2
index 13bf5fc..f8c8e66 100644
@@ -2200,8 +2200,8 @@ static int sctp_setsockopt_autoclose(struct sock *sk, char __user *optval,
 		return -EINVAL;
 	if (copy_from_user(&sp->autoclose, optval, optlen))
 		return -EFAULT;
-	/* make sure it won't exceed MAX_SCHEDULE_TIMEOUT */
-	sp->autoclose = min_t(long, sp->autoclose, MAX_SCHEDULE_TIMEOUT / HZ);
+	/* make sure it won't exceed sctp_max_autoclose / HZ */
+	sp->autoclose = min(sp->autoclose, sctp_max_autoclose / HZ);
 
 	return 0;
 }
net/sctp/sysctl.c +0 -0
index 6b39529..b74686e 100644
@@ -258,6 +258,13 @@ static ctl_table sctp_table[] = {
 		.extra1		= &one,

net/sctp/sysctl.c static ctl_table sctp_table[] = { · line 268

Quoted 10 lines
net/sctp/sysctl.c +7 -0
 		.extra2		= &rwnd_scale_max,
 	},
+	{
+		.procname	= "max_autoclose",
+		.data		= &sctp_max_autoclose,
+		.maxlen		= sizeof(unsigned int),
+		.mode		= 0644,
+		.proc_handler	= &proc_dointvec_jiffies,
+	},

I think it would be better to keep this value in seconds and get rid of division in the setsockopt code. We could then have a min and max values where max value could be something like 2 days. I really don't see an autoclose value that is bigger then that being very useful. In fact, most of the time these values are very small as one wants to close out idle associations.

-vlad

Quoted 2 lines
net/sctp/sysctl.c +0 -0
 	{ /* sentinel */ }
 };
author
#9 lore raw
Quoted 6 lines · Vlad Yasevich

I think it would be better to keep this value in seconds and get rid of division in the setsockopt code. We could then have a min and max values where max value could be something like 2 days. I really don't see an autoclose value that is bigger then that being very useful. In fact, most of the time these values are very small as one wants to close out idle associations.

Now I start to think exposing a new sysctl option might be a little overkill since autoclose is often small.

How about this?

1) Simply store autoclose in seconds in setsockopt.

2) Avoid overflow in associola.c.

	asoc->timeouts[SCTP_EVENT_TIMEOUT_AUTOCLOSE] 		(sp->autoclose > MAX_SCHEDULE_TIMEOUT / HZ)
		? MAX_SCHEDULE_TIMEOUT
		: (unsigned long)sp->autoclose * HZ;

Or we could use INT_MAX instead of MAX_SCHEDULE_TIMEOUT if you want to keep that value consistent across 32/64 bits.

- xi

#10 lore raw
Quoted 15 lines · Xi Wang
Quoted 6 lines · Vlad Yasevich

I think it would be better to keep this value in seconds and get rid of division in the setsockopt code. We could then have a min and max values where max value could be something like 2 days. I really don't see an autoclose value that is bigger then that being very useful. In fact, most of the time these values are very small as one wants to close out idle associations.

Now I start to think exposing a new sysctl option might be a little overkill since autoclose is often small.

How about this?

1) Simply store autoclose in seconds in setsockopt.

Quoted 8 lines

2) Avoid overflow in associola.c.

	asoc->timeouts[SCTP_EVENT_TIMEOUT_AUTOCLOSE] > 		(sp->autoclose > MAX_SCHEDULE_TIMEOUT / HZ)
		? MAX_SCHEDULE_TIMEOUT
		: (unsigned long)sp->autoclose * HZ;

Or we could use INT_MAX instead of MAX_SCHEDULE_TIMEOUT if you want to keep that value consistent across 32/64 bits.

This would work as well. I do like the max configurable though as it might be a nice feature, but the above code is exactly what I was thinking about too.

-vlad

Quoted 3 lines

- xi

author
#11 lore raw
Quoted 3 lines · Vlad Yasevich

This would work as well. I do like the max configurable though as it might be a nice feature, but the above code is exactly what I was thinking about too.

Okay I will do a v3 based on that with the max autoclose. ;-)

- xi

author
#12 lore raw

PATCH v3 1/1 sctp: fix incorrect overflow check on autoclose

In reply to Xi Wang

Commit 8ffd3208 voids the previous patches f6778aab and 810c0719 for limiting the autoclose value. If userspace passes in -1 on 32-bit platform, the overflow check didn't work and autoclose would be set to 0xffffffff.

This patch defines a max_autoclose (in seconds) for limiting the value and exposes it through sysctl, with the following intentions.

1) Avoid overflowing autoclose * HZ.

2) Keep the default autoclose bound consistent across 32- and 64-bit platforms (INT_MAX / HZ in this patch).

3) Keep the autoclose value consistent between setsockopt() and getsockopt() calls.

Suggested-by Vlad Yasevich <[email protected]>
Signed-off-by Xi Wang <[email protected]>
---
 include/net/sctp/structs.h |    4 ++++
 net/sctp/associola.c       |    2 +-
 net/sctp/protocol.c        |    3 +++
 net/sctp/socket.c          |    2 --
 net/sctp/sysctl.c          |   13 +++++++++++++
 5 files changed, 21 insertions(+), 3 deletions(-)
include/net/sctp/structs.h +4 -0
index e90e7a9..a15432da 100644
@@ -241,6 +241,9 @@ extern struct sctp_globals {
 	 * bits is an indicator of when to send and window update SACK.
 	 */
 	int rwnd_update_shift;
+
+	/* Threshold for autoclose timeout, in seconds. */
+	unsigned long max_autoclose;
 } sctp_globals;
 
 #define sctp_rto_initial		(sctp_globals.rto_initial)

@@ -281,6 +284,7 @@ extern struct sctp_globals {
 #define sctp_auth_enable		(sctp_globals.auth_enable)
 #define sctp_checksum_disable		(sctp_globals.checksum_disable)
 #define sctp_rwnd_upd_shift		(sctp_globals.rwnd_update_shift)
+#define sctp_max_autoclose		(sctp_globals.max_autoclose)
 
 /* SCTP Socket type: UDP or TCP style. */
 typedef enum {
net/sctp/associola.c +1 -0
index 152b5b3..acd2edb 100644
@@ -173,7 +173,7 @@ static struct sctp_association *sctp_association_init(struct sctp_association *a
 	asoc->timeouts[SCTP_EVENT_TIMEOUT_HEARTBEAT] = 0;
 	asoc->timeouts[SCTP_EVENT_TIMEOUT_SACK] = asoc->sackdelay;
 	asoc->timeouts[SCTP_EVENT_TIMEOUT_AUTOCLOSE] -		(unsigned long)sp->autoclose * HZ;
+		min_t(unsigned long, sp->autoclose, sctp_max_autoclose) * HZ;
 
 	/* Initializes the timers */
 	for (i = SCTP_EVENT_TIMEOUT_NONE; i < SCTP_NUM_TIMEOUT_TYPES; ++i)
net/sctp/protocol.c +3 -0
index 61b9fca..6f6ad86 100644
@@ -1285,6 +1285,9 @@ SCTP_STATIC __init int sctp_init(void)
 	sctp_max_instreams    		= SCTP_DEFAULT_INSTREAMS;
 	sctp_max_outstreams   		= SCTP_DEFAULT_OUTSTREAMS;
 
+	/* Initialize maximum autoclose timeout. */
+	sctp_max_autoclose		= INT_MAX / HZ;
+
 	/* Initialize handle used for association ids. */
 	idr_init(&sctp_assocs_id);
net/sctp/socket.c +0 -2
index 13bf5fc..54a7cd2 100644
@@ -2200,8 +2200,6 @@ static int sctp_setsockopt_autoclose(struct sock *sk, char __user *optval,
 		return -EINVAL;
 	if (copy_from_user(&sp->autoclose, optval, optlen))
 		return -EFAULT;
-	/* make sure it won't exceed MAX_SCHEDULE_TIMEOUT */
-	sp->autoclose = min_t(long, sp->autoclose, MAX_SCHEDULE_TIMEOUT / HZ);
 
 	return 0;
 }
net/sctp/sysctl.c +12 -0
index 6b39529..60ffbd0 100644
@@ -53,6 +53,10 @@ static int sack_timer_min = 1;
 static int sack_timer_max = 500;
 static int addr_scope_max = 3; /* check sctp_scope_policy_t in include/net/sctp/constants.h for max entries */
 static int rwnd_scale_max = 16;
+static unsigned long max_autoclose_min = 0;
+static unsigned long max_autoclose_max +	(MAX_SCHEDULE_TIMEOUT / HZ > UINT_MAX)
+	? UINT_MAX : MAX_SCHEDULE_TIMEOUT / HZ;
 
 extern long sysctl_sctp_mem[3];
 extern int sysctl_sctp_rmem[3];

@@ -258,6 +262,15 @@ static ctl_table sctp_table[] = {
 		.extra1		= &one,
 		.extra2		= &rwnd_scale_max,
 	},
+	{
+		.procname	= "max_autoclose",
+		.data		= &sctp_max_autoclose,
+		.maxlen		= sizeof(unsigned long),
+		.mode		= 0644,
+		.proc_handler	= &proc_doulongvec_minmax,
+		.extra1		= &max_autoclose_min,
+		.extra2		= &max_autoclose_max,
+	},
 
 	{ /* sentinel */ }
 };
Signature · 1 line
1.7.5.4
Acked-by on v3 1/1
#13 lore raw
Quoted 10 lines · Xi Wang

Commit 8ffd3208 voids the previous patches f6778aab and 810c0719 for limiting the autoclose value. If userspace passes in -1 on 32-bit platform, the overflow check didn't work and autoclose would be set to 0xffffffff.

This patch defines a max_autoclose (in seconds) for limiting the value and exposes it through sysctl, with the following intentions.

1) Avoid overflowing autoclose * HZ.

Quoted 8 lines

2) Keep the default autoclose bound consistent across 32- and 64-bit platforms (INT_MAX / HZ in this patch).

3) Keep the autoclose value consistent between setsockopt() and getsockopt() calls.

Suggested-by Vlad Yasevich <[email protected]>
Signed-off-by Xi Wang <[email protected]>

Looks good to me.

Acked-by Vlad Yasevich <[email protected]>

-vlad

P.S. Sorry, took so long. Just got back from a long vacation.

Quoted 99 lines
---
 include/net/sctp/structs.h |    4 ++++
 net/sctp/associola.c       |    2 +-
 net/sctp/protocol.c        |    3 +++
 net/sctp/socket.c          |    2 --
 net/sctp/sysctl.c          |   13 +++++++++++++
 5 files changed, 21 insertions(+), 3 deletions(-)
include/net/sctp/structs.h +4 -0
index e90e7a9..a15432da 100644
@@ -241,6 +241,9 @@ extern struct sctp_globals {
 	 * bits is an indicator of when to send and window update SACK.
 	 */
 	int rwnd_update_shift;
+
+	/* Threshold for autoclose timeout, in seconds. */
+	unsigned long max_autoclose;
 } sctp_globals;
 
 #define sctp_rto_initial		(sctp_globals.rto_initial)

@@ -281,6 +284,7 @@ extern struct sctp_globals {
 #define sctp_auth_enable		(sctp_globals.auth_enable)
 #define sctp_checksum_disable		(sctp_globals.checksum_disable)
 #define sctp_rwnd_upd_shift		(sctp_globals.rwnd_update_shift)
+#define sctp_max_autoclose		(sctp_globals.max_autoclose)
 
 /* SCTP Socket type: UDP or TCP style. */
 typedef enum {
net/sctp/associola.c +1 -0
index 152b5b3..acd2edb 100644
@@ -173,7 +173,7 @@ static struct sctp_association *sctp_association_init(struct sctp_association *a
 	asoc->timeouts[SCTP_EVENT_TIMEOUT_HEARTBEAT] = 0;
 	asoc->timeouts[SCTP_EVENT_TIMEOUT_SACK] = asoc->sackdelay;
 	asoc->timeouts[SCTP_EVENT_TIMEOUT_AUTOCLOSE] > -		(unsigned long)sp->autoclose * HZ;
+		min_t(unsigned long, sp->autoclose, sctp_max_autoclose) * HZ;
 
 	/* Initializes the timers */
 	for (i = SCTP_EVENT_TIMEOUT_NONE; i < SCTP_NUM_TIMEOUT_TYPES; ++i)
net/sctp/protocol.c +3 -0
index 61b9fca..6f6ad86 100644
@@ -1285,6 +1285,9 @@ SCTP_STATIC __init int sctp_init(void)
 	sctp_max_instreams    		= SCTP_DEFAULT_INSTREAMS;
 	sctp_max_outstreams   		= SCTP_DEFAULT_OUTSTREAMS;
 
+	/* Initialize maximum autoclose timeout. */
+	sctp_max_autoclose		= INT_MAX / HZ;
+
 	/* Initialize handle used for association ids. */
 	idr_init(&sctp_assocs_id);
net/sctp/socket.c +0 -2
index 13bf5fc..54a7cd2 100644
@@ -2200,8 +2200,6 @@ static int sctp_setsockopt_autoclose(struct sock *sk, char __user *optval,
 		return -EINVAL;
 	if (copy_from_user(&sp->autoclose, optval, optlen))
 		return -EFAULT;
-	/* make sure it won't exceed MAX_SCHEDULE_TIMEOUT */
-	sp->autoclose = min_t(long, sp->autoclose, MAX_SCHEDULE_TIMEOUT / HZ);
 
 	return 0;
 }
net/sctp/sysctl.c +12 -0
index 6b39529..60ffbd0 100644
@@ -53,6 +53,10 @@ static int sack_timer_min = 1;
 static int sack_timer_max = 500;
 static int addr_scope_max = 3; /* check sctp_scope_policy_t in include/net/sctp/constants.h for max entries */
 static int rwnd_scale_max = 16;
+static unsigned long max_autoclose_min = 0;
+static unsigned long max_autoclose_max > +	(MAX_SCHEDULE_TIMEOUT / HZ > UINT_MAX)
+	? UINT_MAX : MAX_SCHEDULE_TIMEOUT / HZ;
 
 extern long sysctl_sctp_mem[3];
 extern int sysctl_sctp_rmem[3];

@@ -258,6 +262,15 @@ static ctl_table sctp_table[] = {
 		.extra1		= &one,
 		.extra2		= &rwnd_scale_max,
 	},
+	{
+		.procname	= "max_autoclose",
+		.data		= &sctp_max_autoclose,
+		.maxlen		= sizeof(unsigned long),
+		.mode		= 0644,
+		.proc_handler	= &proc_doulongvec_minmax,
+		.extra1		= &max_autoclose_min,
+		.extra2		= &max_autoclose_max,
+	},
 
 	{ /* sentinel */ }
 };