Is there any way to schedule NAPI on a specific CPU, e.g. similar to what schedule_work_on() workqueue context API does but in a NAPI context?

Asking this because I’m writing a Linux network driver for a NIC that makes use of a “indirect interrupts ring“, for which I’d end up having just one MSI-X irq vector. The related hardirq handler would iterate over this “indirect interrupts ring” whose entries/descriptors tell me exactly which queue requires serving (if RX or TX, and the queue id). Assuming a NAPI instance per RX/TX queue (one per CPU), I’m now confused about how to schedule the RX/TX queue’s NAPI on the queue’s assigned CPU, cuz this way (i.e. via this “indirect interrupts ring“) the RX/TX queues don’t have direct irq vectors associated with them.