[PATCH] net: prevent user from passing illegal stab size

We observed below report when playing with netlink sock:

  UBSAN: shift-out-of-bounds in net/sched/sch_api.c:580:10
  shift exponent 249 is too large for 32-bit type
  CPU: 0 PID: 685 Comm: a.out Not tainted
  Call Trace:
   dump_stack_lvl+0x8d/0xcf
   ubsan_epilogue+0xa/0x4e
   __ubsan_handle_shift_out_of_bounds+0x161/0x182
   __qdisc_calculate_pkt_len+0xf0/0x190
   __dev_queue_xmit+0x2ed/0x15b0

it seems like kernel won't check the stab size_log passing from
user, and will use the insane value later to calculate pkt_len.

This patch just add a check on the size_log to avoid insane
calculation.

Reported-by: Abaci <redacted>
Signed-off-by: Michael Wang <redacted>
---
 include/uapi/linux/pkt_sched.h | 1 +
 net/sched/sch_api.c            | 3 +++
 2 files changed, 4 insertions(+)

diff --git a/include/uapi/linux/pkt_sched.h b/include/uapi/linux/pkt_sched.h
index ec88590..fa194a0 100644
--- a/include/uapi/linux/pkt_sched.h
+++ b/include/uapi/linux/pkt_sched.h

@@ -98,6 +98,7 @@ struct tc_ratespec {
 };

 #define TC_RTAB_SIZE	1024
+#define TC_LOG_MAX	30

 struct tc_sizespec {
 	unsigned char	cell_log;

diff --git a/net/sched/sch_api.c b/net/sched/sch_api.c
index 5e90e9b..1b6b8f8 100644
--- a/net/sched/sch_api.c
+++ b/net/sched/sch_api.c

@@ -513,6 +513,9 @@ static struct qdisc_size_table *qdisc_get_stab(struct nlattr *opt,
 		return stab;
 	}

+	if (s->size_log > TC_LOG_MAX)
+		return ERR_PTR(-EINVAL);
+
 	stab = kmalloc(sizeof(*stab) + tsize * sizeof(u16), GFP_KERNEL);
 	if (!stab)
 		return ERR_PTR(-ENOMEM);

-- 
1.8.3.1

On Thu, 23 Sep 2021 17:08:13 +0800 王贇 wrote:

Adding a uAPI define is not necessary.

Looks sane, please add an extack message.

Why not cover cell_log as well while at it? 

 	stab = kmalloc(sizeof(*stab) + tsize * sizeof(u16), GFP_KERNEL);
 	if (!stab)
 		return ERR_PTR(-ENOMEM);

On 2021/9/24 上午12:00, Jakub Kicinski wrote:

On Thu, 23 Sep 2021 17:08:13 +0800 王贇 wrote:

quoted

We observed below report when playing with netlink sock:

  UBSAN: shift-out-of-bounds in net/sched/sch_api.c:580:10
  shift exponent 249 is too large for 32-bit type
  CPU: 0 PID: 685 Comm: a.out Not tainted
  Call Trace:
   dump_stack_lvl+0x8d/0xcf
   ubsan_epilogue+0xa/0x4e
   __ubsan_handle_shift_out_of_bounds+0x161/0x182
   __qdisc_calculate_pkt_len+0xf0/0x190
   __dev_queue_xmit+0x2ed/0x15b0

it seems like kernel won't check the stab size_log passing from
user, and will use the insane value later to calculate pkt_len.

This patch just add a check on the size_log to avoid insane
calculation.

Reported-by: Abaci <redacted>
Signed-off-by: Michael Wang <redacted>
---
 include/uapi/linux/pkt_sched.h | 1 +
 net/sched/sch_api.c            | 3 +++
 2 files changed, 4 insertions(+)

diff --git a/include/uapi/linux/pkt_sched.h b/include/uapi/linux/pkt_sched.h
index ec88590..fa194a0 100644
--- a/include/uapi/linux/pkt_sched.h
+++ b/include/uapi/linux/pkt_sched.h

@@ -98,6 +98,7 @@ struct tc_ratespec {
 };

 #define TC_RTAB_SIZE	1024
+#define TC_LOG_MAX	30

Adding a uAPI define is not necessary.

Yes, I'll move this to somewhere else.

quoted

 struct tc_sizespec {
 	unsigned char	cell_log;

diff --git a/net/sched/sch_api.c b/net/sched/sch_api.c
index 5e90e9b..1b6b8f8 100644
--- a/net/sched/sch_api.c
+++ b/net/sched/sch_api.c

@@ -513,6 +513,9 @@ static struct qdisc_size_table *qdisc_get_stab(struct nlattr *opt,
 		return stab;
 	}

+	if (s->size_log > TC_LOG_MAX)
+		return ERR_PTR(-EINVAL);

Looks sane, please add an extack message.

Why not cover cell_log as well while at it? 

You're right, will add message and check this one too in v2 :-)

Regards,
Michael Wang

quoted

 	stab = kmalloc(sizeof(*stab) + tsize * sizeof(u16), GFP_KERNEL);
 	if (!stab)
 		return ERR_PTR(-ENOMEM);

We observed below report when playing with netlink sock:

  UBSAN: shift-out-of-bounds in net/sched/sch_api.c:580:10
  shift exponent 249 is too large for 32-bit type
  CPU: 0 PID: 685 Comm: a.out Not tainted
  Call Trace:
   dump_stack_lvl+0x8d/0xcf
   ubsan_epilogue+0xa/0x4e
   __ubsan_handle_shift_out_of_bounds+0x161/0x182
   __qdisc_calculate_pkt_len+0xf0/0x190
   __dev_queue_xmit+0x2ed/0x15b0

it seems like kernel won't check the stab log value passing from
user, and will use the insane value later to calculate pkt_len.

This patch just add a check on the size/cell_log to avoid insane
calculation.

Reported-by: Abaci <redacted>
Signed-off-by: Michael Wang <redacted>
---
 include/net/pkt_sched.h | 1 +
 net/sched/sch_api.c     | 6 ++++++
 2 files changed, 7 insertions(+)

diff --git a/include/net/pkt_sched.h b/include/net/pkt_sched.h
index 6d7b12c..bf79f3a 100644
--- a/include/net/pkt_sched.h
+++ b/include/net/pkt_sched.h

@@ -11,6 +11,7 @@
 #include <uapi/linux/pkt_sched.h>

 #define DEFAULT_TX_QUEUE_LEN	1000
+#define STAB_SIZE_LOG_MAX	30

 struct qdisc_walker {
 	int	stop;

diff --git a/net/sched/sch_api.c b/net/sched/sch_api.c
index 5e90e9b..12f39a2 100644
--- a/net/sched/sch_api.c
+++ b/net/sched/sch_api.c

@@ -513,6 +513,12 @@ static struct qdisc_size_table *qdisc_get_stab(struct nlattr *opt,
 		return stab;
 	}

+	if (s->size_log > STAB_SIZE_LOG_MAX ||
+	    s->cell_log > STAB_SIZE_LOG_MAX) {
+		NL_SET_ERR_MSG(extack, "Invalid logarithmic size of size table");
+		return ERR_PTR(-EINVAL);
+	}
+
 	stab = kmalloc(sizeof(*stab) + tsize * sizeof(u16), GFP_KERNEL);
 	if (!stab)
 		return ERR_PTR(-ENOMEM);

-- 
1.8.3.1