diff mbox series

[1/2] IB/iser: Fix possible NULL deref at iser_inv_desc()

Message ID 1537885006-2535-1-git-send-email-israelr@mellanox.com (mailing list archive)
State Changes Requested
Headers show
Series [1/2] IB/iser: Fix possible NULL deref at iser_inv_desc() | expand

Commit Message

Israel Rukshin Sept. 25, 2018, 2:16 p.m. UTC
In case target remote invalidates bogus rkey and signature is not used,
pi_ctx is NULL deref.
The commit also fails the connection on bogus remote invalidation.

Signed-off-by: Israel Rukshin <israelr@mellanox.com>
Reviewed-by: Max Gurtovoy <maxg@mellanox.com>
---
 drivers/infiniband/ulp/iser/iser_initiator.c | 18 +++++++++++++-----
 1 file changed, 13 insertions(+), 5 deletions(-)

Comments

Leon Romanovsky Sept. 25, 2018, 5:39 p.m. UTC | #1
On Tue, Sep 25, 2018 at 02:16:46PM +0000, Israel Rukshin wrote:
> In case target remote invalidates bogus rkey and signature is not used,
> pi_ctx is NULL deref.
> The commit also fails the connection on bogus remote invalidation.
>
> Signed-off-by: Israel Rukshin <israelr@mellanox.com>
> Reviewed-by: Max Gurtovoy <maxg@mellanox.com>
> ---

There is a need to add Fixes line.

>  drivers/infiniband/ulp/iser/iser_initiator.c | 18 +++++++++++++-----
>  1 file changed, 13 insertions(+), 5 deletions(-)
>
> diff --git a/drivers/infiniband/ulp/iser/iser_initiator.c b/drivers/infiniband/ulp/iser/iser_initiator.c
> index 2f63885..96af06c 100644
> --- a/drivers/infiniband/ulp/iser/iser_initiator.c
> +++ b/drivers/infiniband/ulp/iser/iser_initiator.c
> @@ -589,13 +589,19 @@ void iser_login_rsp(struct ib_cq *cq, struct ib_wc *wc)
>  	ib_conn->post_recv_buf_count--;
>  }
>
> -static inline void
> +static inline int
>  iser_inv_desc(struct iser_fr_desc *desc, u32 rkey)
>  {
> -	if (likely(rkey == desc->rsc.mr->rkey))
> +	if (likely(rkey == desc->rsc.mr->rkey)) {
>  		desc->rsc.mr_valid = 0;
> -	else if (likely(rkey == desc->pi_ctx->sig_mr->rkey))
> +	} else if (likely(desc->pi_ctx && rkey == desc->pi_ctx->sig_mr->rkey)) {
>  		desc->pi_ctx->sig_mr_valid = 0;
> +	} else {
> +		iser_err("Bogus remote invalidation for rkey %#x\n", rkey);

Does it mean that remote user can spam local dmesg?

> +		return -EINVAL;
> +	}
> +
> +	return 0;
>  }
>
>  static int
> @@ -623,12 +629,14 @@ void iser_login_rsp(struct ib_cq *cq, struct ib_wc *wc)
>
>  			if (iser_task->dir[ISER_DIR_IN]) {
>  				desc = iser_task->rdma_reg[ISER_DIR_IN].mem_h;
> -				iser_inv_desc(desc, rkey);
> +				if (unlikely(iser_inv_desc(desc, rkey)))
> +					return -EINVAL;
>  			}
>
>  			if (iser_task->dir[ISER_DIR_OUT]) {
>  				desc = iser_task->rdma_reg[ISER_DIR_OUT].mem_h;
> -				iser_inv_desc(desc, rkey);
> +				if (unlikely(iser_inv_desc(desc, rkey)))
> +					return -EINVAL;
>  			}
>  		} else {
>  			iser_err("failed to get task for itt=%d\n", hdr->itt);
> --
> 1.8.3.1
>
Israel Rukshin Sept. 26, 2018, 9:42 a.m. UTC | #2
Hi Leon,

On 9/25/2018 8:39 PM, Leon Romanovsky wrote:
> Does it mean that remote user can spam local dmesg?

No, because the connection is closed right away.

regards,
Israel
Leon Romanovsky Sept. 26, 2018, 11:03 a.m. UTC | #3
On Wed, Sep 26, 2018 at 12:42:55PM +0300, Israel Rukshin wrote:
> Hi Leon,
>
> On 9/25/2018 8:39 PM, Leon Romanovsky wrote:
> > Does it mean that remote user can spam local dmesg?
>
> No, because the connection is closed right away.

But error is printed before you close the connection, so what will stop
from me to send bogus data in the loop?

Thanks

>
> regards,
> Israel
Max Gurtovoy Sept. 26, 2018, 11:48 a.m. UTC | #4
On 9/26/2018 2:03 PM, Leon Romanovsky wrote:
> On Wed, Sep 26, 2018 at 12:42:55PM +0300, Israel Rukshin wrote:
>> Hi Leon,
>>
>> On 9/25/2018 8:39 PM, Leon Romanovsky wrote:
>>> Does it mean that remote user can spam local dmesg?
>>
>> No, because the connection is closed right away.
> 
> But error is printed before you close the connection, so what will stop
> from me to send bogus data in the loop?
> 
> Thanks

I guess in the worst case it will print <queue_depth> messages.
This will stop the loop.

> 
>>
>> regards,
>> Israel
Sagi Grimberg Sept. 28, 2018, 8:46 p.m. UTC | #5
>>> Hi Leon,
>>>
>>> On 9/25/2018 8:39 PM, Leon Romanovsky wrote:
>>>> Does it mean that remote user can spam local dmesg?
>>>
>>> No, because the connection is closed right away.
>>
>> But error is printed before you close the connection, so what will stop
>> from me to send bogus data in the loop?
>>
>> Thanks
> 
> I guess in the worst case it will print <queue_depth> messages.
> This will stop the loop.

Yes, its bound... We can switch it to pr_warn_ratelimited if this turns
out to be a real spammer.

Otherwise, looks good,
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
diff mbox series

Patch

diff --git a/drivers/infiniband/ulp/iser/iser_initiator.c b/drivers/infiniband/ulp/iser/iser_initiator.c
index 2f63885..96af06c 100644
--- a/drivers/infiniband/ulp/iser/iser_initiator.c
+++ b/drivers/infiniband/ulp/iser/iser_initiator.c
@@ -589,13 +589,19 @@  void iser_login_rsp(struct ib_cq *cq, struct ib_wc *wc)
 	ib_conn->post_recv_buf_count--;
 }
 
-static inline void
+static inline int
 iser_inv_desc(struct iser_fr_desc *desc, u32 rkey)
 {
-	if (likely(rkey == desc->rsc.mr->rkey))
+	if (likely(rkey == desc->rsc.mr->rkey)) {
 		desc->rsc.mr_valid = 0;
-	else if (likely(rkey == desc->pi_ctx->sig_mr->rkey))
+	} else if (likely(desc->pi_ctx && rkey == desc->pi_ctx->sig_mr->rkey)) {
 		desc->pi_ctx->sig_mr_valid = 0;
+	} else {
+		iser_err("Bogus remote invalidation for rkey %#x\n", rkey);
+		return -EINVAL;
+	}
+
+	return 0;
 }
 
 static int
@@ -623,12 +629,14 @@  void iser_login_rsp(struct ib_cq *cq, struct ib_wc *wc)
 
 			if (iser_task->dir[ISER_DIR_IN]) {
 				desc = iser_task->rdma_reg[ISER_DIR_IN].mem_h;
-				iser_inv_desc(desc, rkey);
+				if (unlikely(iser_inv_desc(desc, rkey)))
+					return -EINVAL;
 			}
 
 			if (iser_task->dir[ISER_DIR_OUT]) {
 				desc = iser_task->rdma_reg[ISER_DIR_OUT].mem_h;
-				iser_inv_desc(desc, rkey);
+				if (unlikely(iser_inv_desc(desc, rkey)))
+					return -EINVAL;
 			}
 		} else {
 			iser_err("failed to get task for itt=%d\n", hdr->itt);