In the alloc_ep_skb_list() failure path, the c4iw_put_ep() is
incorrectly used instead of the kfree(). Since the child_ep's
reference count hasn't been properly established at this point,
the c4iw_put_ep() won't actually free the memory, resulting in
permanent memory leak.

Fix by releasing child_ep correctly in the fail path.

Fixes: 4a740838bf44 ("RDMA/iw_cxgb4: Low resource fixes for connection manager")
Cc: stable@vger.kernel.org
Signed-off-by: Wentao Liang <vulab@iscas.ac.cn>
---
 drivers/infiniband/hw/cxgb4/cm.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/infiniband/hw/cxgb4/cm.c b/drivers/infiniband/hw/cxgb4/cm.c
index b3b45c49077d..a09eeb48775f 100644
--- a/drivers/infiniband/hw/cxgb4/cm.c
+++ b/drivers/infiniband/hw/cxgb4/cm.c
@@ -2665,7 +2665,7 @@ static int pass_accept_req(struct c4iw_dev *dev, struct sk_buff *skb)
 	}
 	goto out;
 fail:
-	c4iw_put_ep(&child_ep->com);
+	kfree(child_ep);
 reject:
 	reject_cr(dev, hwtid, skb);
 out:
-- 
2.34.1

On Mon, Dec 15, 2025 at 03:56:07PM +0000, Wentao Liang wrote:
> In the alloc_ep_skb_list() failure path, the c4iw_put_ep() is
> incorrectly used instead of the kfree(). Since the child_ep's
> reference count hasn't been properly established at this point,
> the c4iw_put_ep() won't actually free the memory, resulting in
> permanent memory leak.
> 
> Fix by releasing child_ep correctly in the fail path.
> 
> Fixes: 4a740838bf44 ("RDMA/iw_cxgb4: Low resource fixes for connection manager")
> Cc: stable@vger.kernel.org
> Signed-off-by: Wentao Liang <vulab@iscas.ac.cn>
> ---
>  drivers/infiniband/hw/cxgb4/cm.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/infiniband/hw/cxgb4/cm.c b/drivers/infiniband/hw/cxgb4/cm.c
> index b3b45c49077d..a09eeb48775f 100644
> --- a/drivers/infiniband/hw/cxgb4/cm.c
> +++ b/drivers/infiniband/hw/cxgb4/cm.c
> @@ -2665,7 +2665,7 @@ static int pass_accept_req(struct c4iw_dev *dev, struct sk_buff *skb)
>  	}
>  	goto out;
>  fail:
> -	c4iw_put_ep(&child_ep->com);
> +	kfree(child_ep);

You should squash with your previous patch and get Acked-by from cxgb4
maintainers.

Thanks

>  reject:
>  	reject_cr(dev, hwtid, skb);
>  out:
> -- 
> 2.34.1
>