Hi Peter,
On 2/3/20 2:55 PM, Peter Xu wrote:
> On Sat, Jan 25, 2020 at 06:19:51PM +0100, Eric Auger wrote:
>
> [...]
>
>> +static void virtio_iommu_report_fault(VirtIOIOMMU *viommu, uint8_t reason,
>> + int flags, uint32_t endpoint,
>> + uint64_t address)
>> +{
>> + VirtIODevice *vdev = &viommu->parent_obj;
>> + VirtQueue *vq = viommu->event_vq;
>> + struct virtio_iommu_fault fault;
>> + VirtQueueElement *elem;
>> + size_t sz;
>> +
>> + memset(&fault, 0, sizeof(fault));
>> + fault.reason = reason;
>> + fault.flags = cpu_to_le32(flags);
>> + fault.endpoint = cpu_to_le32(endpoint);
>> + fault.address = cpu_to_le64(address);
>> +
>> + for (;;) {
>> + elem = virtqueue_pop(vq, sizeof(VirtQueueElement));
>> +
>> + if (!elem) {
>> + error_report_once(
>> + "no buffer available in event queue to report event");
>> + return;
>> + }
>> +
>> + if (iov_size(elem->in_sg, elem->in_num) < sizeof(fault)) {
>> + virtio_error(vdev, "error buffer of wrong size");
>> + virtqueue_detach_element(vq, elem, 0);
>> + g_free(elem);
>> + return;
>> + }
>> + break;
>
> This for loop is not needed any more? Other than that:
hum yes indeed.
>
> Reviewed-by: Peter Xu <pet...@redhat.com>
Thanks!
Eric
>
