On 21-10-21, 13:55, Jie Deng wrote: > Can the backend driver control the time point of interrupt injection ? I > can't think of > > why the backend has to send an early interrupt. This operation should be > avoided > > in the backend driver if possible. However, this change make sense if early > interrupt > > can't be avoid. The backend driver probably won't send an event, but the notification event, if it comes, shouldn't have side effects like what it currently have (where we finish the ongoing transfer by calling complete()). -- viresh _______________________________________________ Virtualization mailing list Virtualization@xxxxxxxxxxxxxxxxxxxxxxxxxx https://lists.linuxfoundation.org/mailman/listinfo/virtualization
