summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorDavid Woodhouse <dwmw@amazon.co.uk>2021-01-06 15:39:55 +0000
committerJuergen Gross <jgross@suse.com>2021-01-13 16:12:03 +0100
commit8f4fd86aa5d6aa122619623910065d236592e37c (patch)
tree5e297925c38d026b2f6a35c863bbc2c6688959e9
parent3499ba8198cad47b731792e5e56b9ec2a78a83a2 (diff)
xen: Set platform PCI device INTX affinity to CPU0
With INTX or GSI delivery, Xen uses the event channel structures of CPU0. If the interrupt gets handled by Linux on a different CPU, then no events are seen as pending. Rather than introducing locking to allow other CPUs to process CPU0's events, just ensure that the PCI interrupts happens only on CPU0. Signed-off-by: David Woodhouse <dwmw@amazon.co.uk> Reviewed-by: Boris Ostrovsky <boris.ostrovsky@oracle.com> Link: https://lore.kernel.org/r/20210106153958.584169-3-dwmw2@infradead.org Signed-off-by: Juergen Gross <jgross@suse.com>
-rw-r--r--drivers/xen/platform-pci.c7
1 files changed, 7 insertions, 0 deletions
diff --git a/drivers/xen/platform-pci.c b/drivers/xen/platform-pci.c
index 9db557b76511..18f0ed8b1f93 100644
--- a/drivers/xen/platform-pci.c
+++ b/drivers/xen/platform-pci.c
@@ -132,6 +132,13 @@ static int platform_pci_probe(struct pci_dev *pdev,
dev_warn(&pdev->dev, "request_irq failed err=%d\n", ret);
goto out;
}
+ /*
+ * It doesn't strictly *have* to run on CPU0 but it sure
+ * as hell better process the event channel ports delivered
+ * to CPU0.
+ */
+ irq_set_affinity(pdev->irq, cpumask_of(0));
+
callback_via = get_callback_via(pdev);
ret = xen_set_callback_via(callback_via);
if (ret) {