Re: [PATCH net V3] vhost: log dirty page correctly

[Jason Wang <jasowang@redhat.com>]

On 2019/1/15 上午2:04, Michael S. Tsirkin wrote:
> On Fri, Jan 11, 2019 at 12:00:36PM +0800, Jason Wang wrote:
>> Vhost dirty page logging API is designed to sync through GPA. But we
>> try to log GIOVA when device IOTLB is enabled. This is wrong and may
>> lead to missing data after migration.
>>
>> To solve this issue, when logging with device IOTLB enabled, we will:
>>
>> 1) reuse the device IOTLB translation result of GIOVA->HVA mapping to
>>     get HVA, for writable descriptor, get HVA through iovec. For used
>>     ring update, translate its GIOVA to HVA
>> 2) traverse the GPA->HVA mapping to get the possible GPA and log
>>     through GPA. Pay attention this reverse mapping is not guaranteed
>>     to be unique, so we should log each possible GPA in this case.
>>
>> This fix the failure of scp to guest during migration. In -next, we
>> will probably support passing GIOVA->GPA instead of GIOVA->HVA.
>>
>> Fixes: 6b1e6cc7855b ("vhost: new device IOTLB API")
>> Reported-by: Jintack Lim<jintack@cs.columbia.edu>
>> Cc: Jintack Lim<jintack@cs.columbia.edu>
>> Signed-off-by: Jason Wang<jasowang@redhat.com>
>> ---
>> Changes from V2:
>> - check and log the case of range overlap
>> - remove unnecessary u64 cast
>> - use smp_wmb() for the case of device IOTLB as well
>> Changes from V1:
>> - return error instead of warn
>> ---
>>   drivers/vhost/net.c   |  3 +-
>>   drivers/vhost/vhost.c | 88 ++++++++++++++++++++++++++++++++++++-------
>>   drivers/vhost/vhost.h |  3 +-
>>   3 files changed, 78 insertions(+), 16 deletions(-)
>>
>> diff --git a/drivers/vhost/net.c b/drivers/vhost/net.c
>> index 36f3d0f49e60..bca86bf7189f 100644
>> --- a/drivers/vhost/net.c
>> +++ b/drivers/vhost/net.c
>> @@ -1236,7 +1236,8 @@ static void handle_rx(struct vhost_net *net)
>>   		if (nvq->done_idx > VHOST_NET_BATCH)
>>   			vhost_net_signal_used(nvq);
>>   		if (unlikely(vq_log))
>> -			vhost_log_write(vq, vq_log, log, vhost_len);
>> +			vhost_log_write(vq, vq_log, log, vhost_len,
>> +					vq->iov, in);
>>   		total_len += vhost_len;
>>   		if (unlikely(vhost_exceeds_weight(++recv_pkts, total_len))) {
>>   			vhost_poll_queue(&vq->poll);
>> diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c
>> index 9f7942cbcbb2..55a2e8f9f8ca 100644
>> --- a/drivers/vhost/vhost.c
>> +++ b/drivers/vhost/vhost.c
>> @@ -1733,13 +1733,78 @@ static int log_write(void __user *log_base,
>>   	return r;
>>   }
>>   
>> +static int log_write_hva(struct vhost_virtqueue *vq, u64 hva, u64 len)
>> +{
>> +	struct vhost_umem *umem = vq->umem;
>> +	struct vhost_umem_node *u;
>> +	u64 start, end;
>> +	int r;
>> +	bool hit = false;
>> +
>> +	/* More than one GPAs can be mapped into a single HVA. So
>> +	 * iterate all possible umems here to be safe.
>> +	 */
>> +	list_for_each_entry(u, &umem->umem_list, link) {
>> +		if (u->userspace_addr > hva - 1 + len ||
>> +		    u->userspace_addr - 1 + u->size < hva)
>> +			continue;
>> +		start = max(u->userspace_addr, hva);
>> +		end = min(u->userspace_addr - 1 + u->size, hva - 1 + len);
>> +		r = log_write(vq->log_base,
>> +			      u->start + start - u->userspace_addr,
>> +			      end - start + 1);
>> +		if (r < 0)
>> +			return r;
>> +		hit = true;
>> +	}
>> +
>> +	if (!hit)
>> +		return -EFAULT;
>> +
>> +	return 0;
>> +}
>> +
> I definitely like the simplicity.
>
> But there's one point left here: if len crosses a region boundary,
> but doesn't all fall within a region, we don't consistently report -EFAULT.
>
> So I suspect we need to start by finding a region that contains hva.
> If there are many of these - move right to the end of the
> leftmost one and then repeat until we run out of len
> or fail to find a region and exit with -EFAULT.
>

Ok, will do it in V4.

Thanks


>
>