Re: [PATCH] genirq/msi: Make sure PCI MSIs are activated early

From: Thomas Gleixner
Date: Tue Jul 26 2016 - 07:44:41 EST

On Mon, 25 Jul 2016, Bjorn Helgaas wrote:
> On Mon, Jul 25, 2016 at 09:45:13AM +0200, Thomas Gleixner wrote:
> I thought the original issue [1] was that PCI_MSI_FLAGS_ENABLE was being
> written before PCI_MSI_ADDRESS_LO. That doesn't sound like a good
> idea to me.

Well. That's only a problem if the PCI device does not support masking. But
yes, we missed that case back then.

> That does seem like a problem. Maybe it would be better to delay
> setting PCI_MSI_FLAGS_ENABLE until after the MSI address & data bits
> have been set?

I thought about that, but that gets ugly pretty fast. Here is an alternative

I think that's the proper place to do it _AFTER_ the hierarchical allocation
took place. On x86 Marc's ACTIVATE_EARLY flag would not work because the
message is not yet ready to be assembled.


diff --git a/drivers/pci/msi.c b/drivers/pci/msi.c
index a080f4496fe2..142341f8331b 100644
--- a/drivers/pci/msi.c
+++ b/drivers/pci/msi.c
@@ -645,6 +645,15 @@ static int msi_capability_init(struct pci_dev *dev, int nvec)
return ret;

+ /*
+ * The mask can be ignored and PCI 2.3 does not specify mask bits for
+ * each MSI interrupt. So in case of hierarchical irqdomains we need
+ * to make sure that if masking is not available that the msi message
+ * is written prior to setting the MSI enable bit in the device.
+ */
+ if (pci_msi_ignore_mask || !entry->msi_attrib.maskbit)
+ irq_domain_activate_irq(irq_get_irq_data(entry->irq));
/* Set MSI enabled bits */
pci_intx_for_msi(dev, 0);
pci_msi_set_enable(dev, 1);