[PATCH] infiniband: cma: fix the dev refcnt leak

[PATCH] infiniband: cma: fix the dev refcnt leak

[Wang Yufen]

Syzbot report the following issue:
  infiniband syj1: RDMA CMA: cma_listen_on_dev, error -98
  unregister_netdevice: waiting for vlan0 to become free. Usage count = 2

The causes are as follows:

rdma_listen()
  rdma_bind_addr()
    cma_acquire_dev_by_src_ip()
      cma_attach_to_dev()
        _cma_attach_to_dev()
          cma_dev_get()

  cma_check_port()
  <--The return value is -98， goto err

err:
<-- The error handling here is missing the operation of cma_release_dev.

To fix, add cma_release_dev to error handing.

Fixes: e51060f08a61 ("IB: IP address based RDMA connection manager")
Reported-by: syzbot+5e70d01ee8985ae62a3b@syzkaller.appspotmail.com
Signed-off-by: Wang Yufen <wangyufen@huawei.com>
---
 drivers/infiniband/core/cma.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c
index 26d1772..3a50a8e 100644
--- a/drivers/infiniband/core/cma.c
+++ b/drivers/infiniband/core/cma.c
@@ -4049,6 +4049,9 @@ int rdma_listen(struct rdma_cm_id *id, int backlog)
 	return 0;
 err:
 	id_priv->backlog = 0;
+	if (id_priv->cma_dev)
+		cma_release_dev(id_priv);
+
 	/*
 	 * All the failure paths that lead here will not allow the req_handler's
 	 * to have run.
-- 
1.8.3.1

Re: [PATCH] infiniband: cma: fix the dev refcnt leak

[Jason Gunthorpe]

On Tue, Nov 22, 2022 at 02:47:53PM +0800, Wang Yufen wrote:
> Syzbot report the following issue:
>   infiniband syj1: RDMA CMA: cma_listen_on_dev, error -98
>   unregister_netdevice: waiting for vlan0 to become free. Usage count = 2
> 
> The causes are as follows:
> 
> rdma_listen()
>   rdma_bind_addr()
>     cma_acquire_dev_by_src_ip()
>       cma_attach_to_dev()
>         _cma_attach_to_dev()
>           cma_dev_get()
> 
>   cma_check_port()
>   <--The return value is -98， goto err
> 
> err:
> <-- The error handling here is missing the operation of cma_release_dev.
> 
> To fix, add cma_release_dev to error handing.
> 
> Fixes: e51060f08a61 ("IB: IP address based RDMA connection manager")
> Reported-by: syzbot+5e70d01ee8985ae62a3b@syzkaller.appspotmail.com
> Signed-off-by: Wang Yufen <wangyufen@huawei.com>
> ---
>  drivers/infiniband/core/cma.c | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c
> index 26d1772..3a50a8e 100644
> --- a/drivers/infiniband/core/cma.c
> +++ b/drivers/infiniband/core/cma.c
> @@ -4049,6 +4049,9 @@ int rdma_listen(struct rdma_cm_id *id, int backlog)
>  	return 0;
>  err:
>  	id_priv->backlog = 0;
> +	if (id_priv->cma_dev)
> +		cma_release_dev(id_priv);
> +

I'm not sure about this, the dev is released during _destroy_id()

Is something leaking an entire CM id or is there something wrong with
_destroy_id() ?

Jason