Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
From:   Dou Liyang <douliyangs@gmail.com>
To:     linux-kernel@vger.kernel.org, linux-pci@vger.kernel.org
Cc:     tglx@linutronix.de, kashyap.desai@broadcom.com,
        shivasharan.srikanteshwara@broadcom.com, sumit.saxena@broadcom.com,
        ming.lei@redhat.com, hch@lst.de, bhelgaas@google.com,
        douliyang1@huawei.com, Dou Liyang <douliyangs@gmail.com>
Subject: [PATCH 2/3] irq/affinity: Add is_managed into struct irq_affinity_desc
Date:   Tue,  4 Dec 2018 23:51:21 +0800
Message-Id: <20181204155122.6327-3-douliyangs@gmail.com>
In-Reply-To: <20181204155122.6327-1-douliyangs@gmail.com>
References: <20181204155122.6327-1-douliyangs@gmail.com>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

Now, Linux uses the irq_affinity_desc to convey information.

As Kashyap and Sumit reported, in MSI/-x subsystem, the pre/post vectors
may be used to some extra reply queues for performance.

  https://marc.info/?l=linux-kernel&m=153543887027997&w=2

Their affinities are not NULL, but, they should be mapped as unmanaged
interrupts. So, only transfering the irq affinity assignments is not enough.

Add a new bit "is_managed" to convey the info in irq_affinity_desc and use
it in alloc_descs().

Reported-by: Kashyap Desai <kashyap.desai@broadcom.com>
Reported-by: Sumit Saxena <sumit.saxena@broadcom.com>
Signed-off-by: Dou Liyang <douliyangs@gmail.com>
---
 include/linux/interrupt.h | 1 +
 kernel/irq/affinity.c     | 7 +++++++
 kernel/irq/irqdesc.c      | 9 +++++++--
 3 files changed, 15 insertions(+), 2 deletions(-)

diff --git a/include/linux/interrupt.h b/include/linux/interrupt.h
index 71be303231e9..a12b3dbbc45e 100644
--- a/include/linux/interrupt.h
+++ b/include/linux/interrupt.h
@@ -263,6 +263,7 @@ struct irq_affinity {
  */
 struct irq_affinity_desc {
 	struct cpumask	mask;
+	unsigned int	is_managed : 1;
 };
 
 #if defined(CONFIG_SMP)
diff --git a/kernel/irq/affinity.c b/kernel/irq/affinity.c
index 1562a36e7c0f..d122575ba1b4 100644
--- a/kernel/irq/affinity.c
+++ b/kernel/irq/affinity.c
@@ -289,6 +289,13 @@ irq_create_affinity_masks(int nvecs, const struct irq_affinity *affd)
 	for (; curvec < nvecs; curvec++)
 		cpumask_copy(&masks[curvec].mask, irq_default_affinity);
 
+	/*  Setup complementary information */
+	for (i = 0; i < nvecs; i++) {
+		if (i >= affd->pre_vectors && i < nvecs - affd->post_vectors)
+			masks[i].is_managed = 1;
+		else
+			masks[i].is_managed = 0;
+	}
 outnodemsk:
 	free_node_to_cpumask(node_to_cpumask);
 	return masks;
diff --git a/kernel/irq/irqdesc.c b/kernel/irq/irqdesc.c
index f87fa2b9935a..6b0821c144c0 100644
--- a/kernel/irq/irqdesc.c
+++ b/kernel/irq/irqdesc.c
@@ -455,7 +455,7 @@ static int alloc_descs(unsigned int start, unsigned int cnt, int node,
 	const struct irq_affinity_desc *cur_affinity= affinity;
 	const struct cpumask *mask = NULL;
 	struct irq_desc *desc;
-	unsigned int flags;
+	unsigned int flags = 0;
 	int i;
 
 	/* Validate affinity mask(s) */
@@ -468,11 +468,16 @@ static int alloc_descs(unsigned int start, unsigned int cnt, int node,
 		}
 	}
 
-	flags = affinity ? IRQD_AFFINITY_MANAGED | IRQD_MANAGED_SHUTDOWN : 0;
 	mask = NULL;
 
 	for (i = 0; i < cnt; i++) {
 		if (affinity) {
+			if (affinity->is_managed) {
+				flags = IRQD_AFFINITY_MANAGED |
+					IRQD_MANAGED_SHUTDOWN;
+			} else {
+				flags = 0;
+			}
 			mask = &affinity->mask;
 			node = cpu_to_node(cpumask_first(mask));
 			affinity++;
-- 
2.17.2