[dpdk-dev] [PATCH v3 44/44] net/virtio: handle Virtio-user setup failure properly

[Maxime Coquelin]

On 1/26/21 7:36 AM, Xia, Chenbo wrote:
> Hi Maxime,
> 
>> -----Original Message-----
>> From: Maxime Coquelin <maxime.coquelin at redhat.com>
>> Sent: Tuesday, January 26, 2021 1:15 AM
>> To: dev at dpdk.org; Xia, Chenbo <chenbo.xia at intel.com>; olivier.matz at 6wind.com;
>> amorenoz at redhat.com; david.marchand at redhat.com
>> Cc: Maxime Coquelin <maxime.coquelin at redhat.com>
>> Subject: [PATCH v3 44/44] net/virtio: handle Virtio-user setup failure
>> properly
>>
>> This patch fixes virtio_user_dev_setup() error path,
>> by cleaning all resources it allocates. It introduces
>> virtio_user_dev_uninit_notify() that cleans all open
>> FDs. It implies assigning all FDs to -1 at init time.
>>
>> With these changes done, virtio_user_dev_init_notify()
>> can be simplified.
>>
>> Suggested-by: Adrian Moreno <amorenoz at redhat.com>
>> Signed-off-by: Maxime Coquelin <maxime.coquelin at redhat.com>
>> ---
>>  .../net/virtio/virtio_user/virtio_user_dev.c  | 68 +++++++++++++------
>>  1 file changed, 47 insertions(+), 21 deletions(-)
>>
>> diff --git a/drivers/net/virtio/virtio_user/virtio_user_dev.c
>> b/drivers/net/virtio/virtio_user/virtio_user_dev.c
>> index a1e32158bb..7c60022a26 100644
>> --- a/drivers/net/virtio/virtio_user/virtio_user_dev.c
>> +++ b/drivers/net/virtio/virtio_user/virtio_user_dev.c
>> @@ -283,13 +283,7 @@ virtio_user_dev_init_notify(struct virtio_user_dev *dev)
>>  	int callfd;
>>  	int kickfd;
>>
>> -	for (i = 0; i < VIRTIO_MAX_VIRTQUEUES; ++i) {
>> -		if (i >= dev->max_queue_pairs * 2) {
>> -			dev->kickfds[i] = -1;
>> -			dev->callfds[i] = -1;
>> -			continue;
>> -		}
>> -
>> +	for (i = 0; i < dev->max_queue_pairs * 2; i++) {
>>  		/* May use invalid flag, but some backend uses kickfd and
>>  		 * callfd as criteria to judge if dev is alive. so finally we
>>  		 * use real event_fd.
>> @@ -297,25 +291,48 @@ virtio_user_dev_init_notify(struct virtio_user_dev *dev)
>>  		callfd = eventfd(0, EFD_CLOEXEC | EFD_NONBLOCK);
>>  		if (callfd < 0) {
>>  			PMD_DRV_LOG(ERR, "(%s) callfd error, %s", dev->path,
>> strerror(errno));
>> -			break;
>> +			goto err;
>>  		}
>>  		kickfd = eventfd(0, EFD_CLOEXEC | EFD_NONBLOCK);
>>  		if (kickfd < 0) {
>>  			close(callfd);
>>  			PMD_DRV_LOG(ERR, "(%s) kickfd error, %s", dev->path,
>> strerror(errno));
>> -			break;
>> +			goto err;
>>  		}
>>  		dev->callfds[i] = callfd;
>>  		dev->kickfds[i] = kickfd;
>>  	}
>>
>> -	if (i < VIRTIO_MAX_VIRTQUEUES) {
>> -		for (j = 0; j < i; ++j) {
>> -			close(dev->callfds[j]);
>> +	return 0;
>> +err:
>> +	for (j = 0; j < i; j++) {
>> +		if (dev->kickfds[j] >= 0) {
>>  			close(dev->kickfds[j]);
>> +			dev->kickfds[j] = -1;
>>  		}
>> +		if (dev->callfds[j] >= 0) {
>> +			close(dev->callfds[j]);
>> +			dev->callfds[j] = -1;
>> +		}
>> +	}
>>
>> -		return -1;
>> +	return -1;
>> +}
>> +
>> +static int
> 
> Use void?

yes.

>> +virtio_user_dev_uninit_notify(struct virtio_user_dev *dev)
>> +{
>> +	uint32_t i;
>> +
>> +	for (i = 0; i < VIRTIO_MAX_VIRTQUEUES; ++i) {
> 
> Do we need to loop for VIRTIO_MAX_VIRTQUEUES? Kick/call fds would only be set
> valid in virtio_user_dev_init_notify, and since virtio_user_dev_init_notify only
> loop for dev->max_queue_pairs * 2, this function will also loop for dev->max_queue_pairs * 2?
> Other fds should be -1, right?

Yes, given max_queue_pairs is never updated during the lifecycle, it is
safe to loop for max_queue_pairs.

I'll fix it in v4.

Thanks,
Maxime

> Thanks,
> Chenbo
> 
>> +		if (dev->kickfds[i] >= 0) {
>> +			close(dev->kickfds[i]);
>> +			dev->kickfds[i] = -1;
>> +		}
>> +		if (dev->callfds[i] >= 0) {
>> +			close(dev->callfds[i]);
>> +			dev->callfds[i] = -1;
>> +		}
>>  	}
>>
>>  	return 0;
>> @@ -427,15 +444,22 @@ virtio_user_dev_setup(struct virtio_user_dev *dev)
>>
>>  	if (virtio_user_dev_init_notify(dev) < 0) {
>>  		PMD_INIT_LOG(ERR, "(%s) Failed to init notifiers\n", dev->path);
>> -		return -1;
>> +		goto destroy;
>>  	}
>>
>>  	if (virtio_user_fill_intr_handle(dev) < 0) {
>>  		PMD_INIT_LOG(ERR, "(%s) Failed to init interrupt handler\n", dev-
>>> path);
>> -		return -1;
>> +		goto uninit;
>>  	}
>>
>>  	return 0;
>> +
>> +uninit:
>> +	virtio_user_dev_uninit_notify(dev);
>> +destroy:
>> +	dev->ops->destroy(dev);
>> +
>> +	return -1;
>>  }
>>
>>  /* Use below macro to filter features from vhost backend */
>> @@ -466,9 +490,16 @@ virtio_user_dev_init(struct virtio_user_dev *dev, char
>> *path, int queues,
>>  		     enum virtio_user_backend_type backend_type)
>>  {
>>  	uint64_t backend_features;
>> +	int i;
>>
>>  	pthread_mutex_init(&dev->mutex, NULL);
>>  	strlcpy(dev->path, path, PATH_MAX);
>> +
>> +	for (i = 0; i < VIRTIO_MAX_VIRTQUEUES; i++) {
>> +		dev->kickfds[i] = -1;
>> +		dev->callfds[i] = -1;
>> +	}
>> +
>>  	dev->started = 0;
>>  	dev->max_queue_pairs = queues;
>>  	dev->queue_pairs = 1; /* mq disabled by default */
>> @@ -565,16 +596,11 @@ virtio_user_dev_init(struct virtio_user_dev *dev, char
>> *path, int queues,
>>  void
>>  virtio_user_dev_uninit(struct virtio_user_dev *dev)
>>  {
>> -	uint32_t i;
>> -
>>  	virtio_user_stop_device(dev);
>>
>>  	rte_mem_event_callback_unregister(VIRTIO_USER_MEM_EVENT_CLB_NAME, dev);
>>
>> -	for (i = 0; i < dev->max_queue_pairs * 2; ++i) {
>> -		close(dev->callfds[i]);
>> -		close(dev->kickfds[i]);
>> -	}
>> +	virtio_user_dev_uninit_notify(dev);
>>
>>  	free(dev->ifname);
>>
>> --
>> 2.29.2
>