On Tue, 12 Feb 2019, Ming Lei wrote:

Currently pre-caculated set vectors are provided by driver for
allocating & spread vectors. This way only works when drivers passes
same 'max_vecs' and 'min_vecs' to pci_alloc_irq_vectors_affinity(),
also requires driver to retry the allocating & spread.

As Bjorn and Keith mentioned, the current usage & interface for irq sets
is a bit awkward because the retrying should have been avoided by providing
one resonable 'min_vecs'. However, if 'min_vecs' isn't same with
'max_vecs', number of the allocated vectors is unknown before calling
pci_alloc_irq_vectors_affinity(), then each set's vectors can't be
pre-caculated.

Add a new callback of .calc_sets into 'struct irq_affinity' so that
driver can caculate set vectors after IRQ vector is allocated and
before spread IRQ vectors. Add 'priv' so that driver may retrieve
its private data via the 'struct irq_affinity'.

Nice changelog!

 	 * Spread on present CPUs starting from affd->pre_vectors. If we
 	 * have multiple sets, build each sets affinity mask separately.
 	 */
-	nr_sets = affd->nr_sets;
-	if (!nr_sets) {
+	if (affd->calc_sets) {
+		affd->calc_sets(affd, nvecs);
+		nr_sets = affd->nr_sets;
+	} else if (!affd->nr_sets) {
 		nr_sets = 1;
 		affd->set_vectors[0] = affvecs;
-	}
+	} else
+		nr_sets = affd->nr_sets;

Lacks curly brackets.

Aside of that it might make sense to get rid of the local variable, but
that should be done in patch 1 if at all. No strong opinion though.

Thanks,

	tglx