On Mon, Jun 13, 2016 at 3:40 AM, Lorenzo Pieralisi
[off-list ref] wrote:

On Fri, Jun 10, 2016 at 06:36:12PM -0500, Bjorn Helgaas wrote:

quoted

On Fri, Jun 10, 2016 at 09:55:17PM +0200, Tomasz Nowicki wrote:

quoted

To enable PCI legacy IRQs on platforms booting with ACPI, arch code
should include ACPI specific callbacks that parse and set-up the
device IRQ number, equivalent to the DT boot path. Owing to the current
ACPI core scan handlers implementation, ACPI PCI legacy IRQs bindings
cannot be parsed at device add time, since that would trigger ACPI scan
handlers ordering issues depending on how the ACPI tables are defined.

Uh, OK :)  I can't figure out exactly what the problem is here -- I
don't know where to look if I wanted to fix the scan handler ordering
issues, and I don't know how I could tell if it would ever be safe to
move this from driver probe-time back to device add-time.

Right, the commit log could have been more informative.

pcibios_add_device() was added in:

commit d1e6dc91b532 ("arm64: Add architectural support for PCI")

whose commit log does not specify why legacy IRQ parsing should
be done at pcibios_add_device() either, so honestly we had to
do with the information we have at hand.

quoted

I also notice that x86 and ia64 call acpi_pci_irq_enable() even later,
when the driver *enables* the device.  Is there a reason you didn't do
it at the same time as x86 and ia64?  This is another of those pcibios
hooks that really don't do anything arch-specific, so I can imagine
refactoring this somehow, someday.

Yes, with [1], that was the goal, that stopped because [1] does not
work on x86.

Only DT platform(s) affected by this change are all platforms relying on
drivers/pci/host/pci-xgene.c (others rely on pci_fixup_irqs() that
should be removed too), if on those platforms probing IRQs at device
enable time works ok I can update this patch (it can be done through [1]
once we figure out what to do with it on x86) and move the IRQ set-up at
pcibios_enable_device() time.

@Duc: any feedback on this ?

Hi Lorenzo,

The changes to add pcibios_alloc_irq works fine on X-Gene PCIe

I also tried to remove pcibios_alloc_irq and move its code into
pcibios_enable_device
after pci_enable_resource call and legacy IRQ also works.

Can you also point me to the discussion thread or some info. about the
issue on x86 with [1]?
I want to check if there is any more test case I need to verify.

Regards,
Duc Dang.

Thanks,
Lorenzo

[1] http://www.spinics.net/lists/linux-pci/msg45950.html

quoted

Did we have this conversation before?  It seems vaguely familiar, so I
apologize if you already explained this once.

quoted

To solve this problem and consolidate FW PCI legacy IRQs parsing in
one single pcibios callback (pending final removal), this patch moves
DT PCI IRQ parsing to the pcibios_alloc_irq() callback (called by
PCI core code at device probe time) and adds ACPI PCI legacy IRQs
parsing to the same callback too, so that FW PCI legacy IRQs parsing
is confined in one single arch callback that can be easily removed
when code parsing PCI legacy IRQs is consolidated and moved to core
PCI code.

Signed-off-by: Tomasz Nowicki <redacted>
Suggested-by: Lorenzo Pieralisi <redacted>
---
 arch/arm64/kernel/pci.c | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)

diff --git a/arch/arm64/kernel/pci.c b/arch/arm64/kernel/pci.c
index d5d3d26..b3b8a2c 100644
--- a/arch/arm64/kernel/pci.c
+++ b/arch/arm64/kernel/pci.c

@@ -51,11 +51,16 @@ int pcibios_enable_device(struct pci_dev *dev, int mask)
 }

 /*
- * Try to assign the IRQ number from DT when adding a new device
+ * Try to assign the IRQ number when probing a new device
  */
-int pcibios_add_device(struct pci_dev *dev)
+int pcibios_alloc_irq(struct pci_dev *dev)
 {
-   dev->irq = of_irq_parse_and_map_pci(dev, 0, 0);
+   if (acpi_disabled)
+           dev->irq = of_irq_parse_and_map_pci(dev, 0, 0);
+#ifdef CONFIG_ACPI
+   else
+           return acpi_pci_irq_enable(dev);
+#endif

    return 0;
 }
--

1.9.1