On Fri, May 21, 2021 at 5:48 PM Thomas Gleixner [off-list ref] wrote:

On Fri, May 21 2021 at 12:13, Nitesh Lal wrote:

quoted

On Fri, May 21, 2021 at 8:03 AM Thomas Gleixner [off-list ref] wrote:

quoted

Provide new interfaces:

  irq_update_affinity_hint() - Only sets the affinity hint pointer
  irq_apply_affinity_hint()  - Set the pointer and apply the affinity to
                               the interrupt

Any reason why you ruled out the usage of irq_set_affinity_and_hint()?
IMHO the latter makes it very clear what the function is meant to do.

You're right. I was trying to phase the existing hint setter out, but
that's probably pointless overengineering for no real value. Let me redo
that.

Thomas, are you planning to send a v2 for this soon or did I somehow miss
it?

Since your other patch "genirq: Export affinity setter for modules" is
already in linux-next, I have started looking into the drivers where
we can use that.

On thinking about this whole chunk a little more, I do wonder about the
reason why we are still sticking with the hints.

The two reasons that I could come up with are:
- We are not entirely sure if irqbalance still consumes this or not
- For future use by some other userspace daemon  (?)

Does that sound right?

-- 
Thanks
Nitesh