On 2014/11/6 6:45, Bjorn Helgaas wrote:

On Tue, Nov 04, 2014 at 08:01:54PM +0800, Jiang Liu wrote:

quoted

It's arch_setup_msi_irq()/arch_setup_msi_irqs()'s responsibility to call
irq_set_msi_desc() to associate IRQ descriptors and MSI descriptors,
so kill the redundant call of irq_set_msi_desc() for MSIx interrupts
in PCI MSI core.

"MSI-X" in English text, "msix" in code.

The default arch_setup_msi_irq() in drivers/pci/msi.c doesn't call
irq_set_msi_desc().  Does it happen somewhere inside chip->setup_irq()?

Yes.

I also found this.
http://www.spinics.net/lists/linux-pci/msg34256.html

I don't know how to verify that there are calls in all the places needed.
That makes me wonder if the factoring is wrong -- maybe irq_set_msi_desc()
could be done in some common place.

In my idea, place the irq_set_msi_desc() in common MSI core is ok, but currently almost
all MSI arch code call irq_set_msi_desc() in arch code. So a lot of code need to change.
And arch code setup MSI and MSI-X in the same way, so if MSI work happy without irq_set_msi_desc(entry->irq, entry)
in common MSI code, MSI-X should be the same.

quoted

Signed-off-by: Jiang Liu <redacted>
---
 drivers/pci/msi.c |    1 -
 1 file changed, 1 deletion(-)

diff --git a/drivers/pci/msi.c b/drivers/pci/msi.c
index afe974600c7d..da181c59394b 100644
--- a/drivers/pci/msi.c
+++ b/drivers/pci/msi.c

@@ -685,7 +685,6 @@ static void msix_program_entries(struct pci_dev *dev,
 						PCI_MSIX_ENTRY_VECTOR_CTRL;
 
 		entries[i].vector = entry->irq;
-		irq_set_msi_desc(entry->irq, entry);
 		entry->masked = readl(entry->mask_base + offset);
 		msix_mask_irq(entry, 1);
 		i++;

-- 
1.7.10.4

--
To unsubscribe from this list: send the line "unsubscribe linux-acpi" in
the body of a message to majordomo at vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

.

-- 
Thanks!
Yijing