iommu/amd: Allocate IOMMU irqs using numa locality info

Use numa information to allocate irq resources and also to set
irq affinity. This optimizes the IOMMU interrupt handling.

Reviewed-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Reviewed-by: Alexey Kardashevskiy <aik@amd.com>
Signed-off-by: Vasant Hegde <vasant.hegde@amd.com>
Link: https://lore.kernel.org/r/20230321092348.6127-3-vasant.hegde@amd.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
This commit is contained in:
Vasant Hegde
2023-03-22 15:43:40 +01:00
committed by Joerg Roedel
parent 0d571dcbe7
commit 4d4a0dbab2
+2 -1
View File
@@ -2383,6 +2383,7 @@ static int iommu_setup_intcapxt(struct amd_iommu *iommu)
struct irq_domain *domain;
struct irq_alloc_info info;
int irq, ret;
int node = dev_to_node(&iommu->dev->dev);
domain = iommu_get_irqdomain();
if (!domain)
@@ -2392,7 +2393,7 @@ static int iommu_setup_intcapxt(struct amd_iommu *iommu)
info.type = X86_IRQ_ALLOC_TYPE_AMDVI;
info.data = iommu;
irq = irq_domain_alloc_irqs(domain, 1, NUMA_NO_NODE, &info);
irq = irq_domain_alloc_irqs(domain, 1, node, &info);
if (irq < 0) {
irq_domain_remove(domain);
return irq;