Re: [IB/srpt] c804af2c1d: last_state.test.blktests.exit_code.143

From: Bart Van Assche <bvanassche@acm.org>
To: Jason Gunthorpe <jgg@nvidia.com>
Cc: Sagi Grimberg <sagi@grimberg.me>,
	Yamin Friedman <yaminf@mellanox.com>,
	kernel test robot <rong.a.chen@intel.com>,
	Max Gurtovoy <maxg@mellanox.com>,
	LKML <linux-kernel@vger.kernel.org>,
	Doug Ledford <dledford@redhat.com>,
	Jason Gunthorpe <jgg+lists@ziepe.ca>,
	linux-rdma@vger.kernel.org, lkp@lists.01.org
Subject: Re: [IB/srpt] c804af2c1d: last_state.test.blktests.exit_code.143
Date: Tue, 8 Sep 2020 19:01:54 -0700	[thread overview]
Message-ID: <e8a240aa-9e9b-3dca-062f-9130b787f29b@acm.org> (raw)
In-Reply-To: <20200908182232.GP9166@nvidia.com>

On 2020-09-08 11:22, Jason Gunthorpe wrote:
> It is reasonable to consider the cq_pool as a built-in client, so I
> would suggest moving it to right around the time the dynamic clients
> are handled. Something like this:
> 
> diff --git a/drivers/infiniband/core/device.c b/drivers/infiniband/core/device.c
> index c36b4d2b61e0c0..e3651dacad1da6 100644
> --- a/drivers/infiniband/core/device.c
> +++ b/drivers/infiniband/core/device.c
> @@ -1285,6 +1285,8 @@ static void disable_device(struct ib_device *device)
>  		remove_client_context(device, cid);
>  	}
>  
> +	ib_cq_pool_destroy(ib_dev);
> +
>  	/* Pairs with refcount_set in enable_device */
>  	ib_device_put(device);
>  	wait_for_completion(&device->unreg_completion);
> @@ -1328,6 +1330,8 @@ static int enable_device_and_get(struct ib_device *device)
>  			goto out;
>  	}
>  
> +	ib_cq_pool_init(device);
> +
>  	down_read(&clients_rwsem);
>  	xa_for_each_marked (&clients, index, client, CLIENT_REGISTERED) {
>  		ret = add_client_context(device, client);
> @@ -1400,7 +1404,6 @@ int ib_register_device(struct ib_device *device, const char *name)
>  		goto dev_cleanup;
>  	}
>  
> -	ib_cq_pool_init(device);
>  	ret = enable_device_and_get(device);
>  	dev_set_uevent_suppress(&device->dev, false);
>  	/* Mark for userspace that device is ready */
> @@ -1455,7 +1458,6 @@ static void __ib_unregister_device(struct ib_device *ib_dev)
>  		goto out;
>  
>  	disable_device(ib_dev);
> -	ib_cq_pool_destroy(ib_dev);
>  
>  	/* Expedite removing unregistered pointers from the hash table */
>  	free_netdevs(ib_dev);

The above patch didn't compile, but the patch below does and makes the hang
disappear. So feel free to add the following to the patch below:

Tested-by: Bart Van Assche <bvanassche@acm.org>

diff --git a/drivers/infiniband/core/device.c b/drivers/infiniband/core/device.c
index c36b4d2b61e0..23ee65a9185f 100644
--- a/drivers/infiniband/core/device.c
+++ b/drivers/infiniband/core/device.c
@@ -1285,6 +1285,8 @@ static void disable_device(struct ib_device *device)
 		remove_client_context(device, cid);
 	}

+	ib_cq_pool_destroy(device);
+
 	/* Pairs with refcount_set in enable_device */
 	ib_device_put(device);
 	wait_for_completion(&device->unreg_completion);
@@ -1328,6 +1330,8 @@ static int enable_device_and_get(struct ib_device *device)
 			goto out;
 	}

+	ib_cq_pool_init(device);
+
 	down_read(&clients_rwsem);
 	xa_for_each_marked (&clients, index, client, CLIENT_REGISTERED) {
 		ret = add_client_context(device, client);
@@ -1400,7 +1404,6 @@ int ib_register_device(struct ib_device *device, const char *name)
 		goto dev_cleanup;
 	}

-	ib_cq_pool_init(device);
 	ret = enable_device_and_get(device);
 	dev_set_uevent_suppress(&device->dev, false);
 	/* Mark for userspace that device is ready */
@@ -1455,7 +1458,6 @@ static void __ib_unregister_device(struct ib_device *ib_dev)
 		goto out;

 	disable_device(ib_dev);
-	ib_cq_pool_destroy(ib_dev);

 	/* Expedite removing unregistered pointers from the hash table */
 	free_netdevs(ib_dev);

