These 2 patches correct a bogus behaviour introduced by commit 1851617cd2
("PCI/MSI: Disable MSI at enumeration even if kernel doesn't support MSI").
The commit moved the logic responsible to disable MSI/MSI-X interrupts
at PCI probe time to a new function, named pci_msi_setup_pci_dev(), that
is not reachable in the code path of PowerPC pSeries platform.
Since then, devices aren't able to activate MSI/MSI-X capability, even
after boot. The first patch makes the function pci_msi_setup_pci_dev()
non-static. The second patch inserts a call to the function in powerpc
code, so it explicitly disables MSI/MSI-X interrupts at PCI probe time.
Guilherme G. Piccoli (2):
PCI: Make pci_msi_setup_pci_dev() non-static for use by arch code
powerpc/PCI: Disable MSI/MSI-X interrupts at PCI probe time in OF case
arch/powerpc/kernel/pci_of_scan.c | 3 +++
drivers/pci/probe.c | 2 +-
include/linux/pci.h | 1 +
3 files changed, 5 insertions(+), 1 deletion(-)
--
2.1.0
Commit 1851617cd2 ("PCI/MSI: Disable MSI at enumeration even if kernel
doesn't support MSI") changed the location of the code that disables
MSI/MSI-X interrupts at PCI probe time in devices that have this flag set.
It moved the code from pci_msi_init_pci_dev() to a new function named
pci_msi_setup_pci_dev(), called by pci_setup_device().
Since then, the pSeries platform of the powerpc architecture needs to
disable MSI at PCI probe time manually, as the code flow doesn't
reach pci_setup_device(). For doing so, it wants to call
pci_msi_setup_pci_dev(). This patch makes the required function
non-static, so that it will be called on PCI probe path on powerpc pSeries
platform in next patch.
Signed-off-by: Guilherme G. Piccoli <redacted>
---
drivers/pci/probe.c | 2 +-
include/linux/pci.h | 1 +
2 files changed, 2 insertions(+), 1 deletion(-)
diff --git a/drivers/pci/probe.c b/drivers/pci/probe.c
index cefd636..520c5b6 100644
--- a/drivers/pci/probe.c
+++ b/drivers/pci/probe.c
@@ -1103,7 +1103,7 @@ int pci_cfg_space_size(struct pci_dev *dev)
#define LEGACY_IO_RESOURCE (IORESOURCE_IO | IORESOURCE_PCI_FIXED)
-static void pci_msi_setup_pci_dev(struct pci_dev *dev)
+void pci_msi_setup_pci_dev(struct pci_dev *dev)
{
/*
* Disable the MSI hardware to avoid screaming interruptsdiff --git a/include/linux/pci.h b/include/linux/pci.h
index 8a0321a..860c751 100644
--- a/include/linux/pci.h
+++ b/include/linux/pci.h
@@ -1202,6 +1202,7 @@ struct msix_entry {
u16 entry; /* driver uses to specify entry, OS writes */
};
+void pci_msi_setup_pci_dev(struct pci_dev *dev);
#ifdef CONFIG_PCI_MSI
int pci_msi_vec_count(struct pci_dev *dev);--
2.1.0
Since the commit 1851617cd2 ("PCI/MSI: Disable MSI at enumeration even if
kernel doesn't support MSI"), MSI/MSI-X interrupts aren't being disabled
at PCI probe time, as the logic responsible for this was moved in the
aforementioned commit from pci_device_add() to pci_setup_device(). The
latter function is not reachable on PowerPC pSeries platform during
Open Firmware PCI probing time.
This patch calls pci_msi_setup_pci_dev() explicitly to disable MSI/MSI-X
during PCI probe time on pSeries platform.
Signed-off-by: Guilherme G. Piccoli <redacted>
---
arch/powerpc/kernel/pci_of_scan.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/arch/powerpc/kernel/pci_of_scan.c b/arch/powerpc/kernel/pci_of_scan.c
index 42e02a2..0e920f3 100644
--- a/arch/powerpc/kernel/pci_of_scan.c
+++ b/arch/powerpc/kernel/pci_of_scan.c
@@ -191,6 +191,9 @@ struct pci_dev *of_create_pci_dev(struct device_node *node,
pci_device_add(dev, bus);
+ /* Disable MSI/MSI-X here to avoid bogus interrupts */
+ pci_msi_setup_pci_dev(dev);
+
return dev;
}
EXPORT_SYMBOL(of_create_pci_dev);
--
2.1.0
Hi Guilherme,
Thanks for the patches.
On Tue, 2015-08-18 at 18:13 -0300, Guilherme G. Piccoli wrote:
Commit 1851617cd2 ("PCI/MSI: Disable MSI at enumeration even if kernel
doesn't support MSI") changed the location of the code that disables
MSI/MSI-X interrupts at PCI probe time in devices that have this flag set.
It moved the code from pci_msi_init_pci_dev() to a new function named
pci_msi_setup_pci_dev(), called by pci_setup_device().
OK.
Since then, the pSeries platform of the powerpc architecture needs to
disable MSI at PCI probe time manually, as the code flow doesn't
reach pci_setup_device().
For doing so, it wants to call
pci_msi_setup_pci_dev(). This patch makes the required function
non-static, so that it will be called on PCI probe path on powerpc pSeries
platform in next patch.
I didn't follow that entirely, I think you mean something like:
The pseries PCI probing code does not call pci_setup_device(), so since
commit 1851617cd2 pci_msi_setup_pci_dev() is not called and MSIs are left
enabled, which is a bug.
To fix this the pseries PCI probe should manually call
pci_msi_setup_pci_dev(), so make it non-static.
Does that look OK?
Also you haven't CC'ed the original author of the commit, or the PCI
maintainer, or the relevant lists.
That would be:
Michael S. Tsirkin [off-list ref]
Bjorn Helgaas [off-list ref]
linux-pci@vger.kernel.org
linux-kernel@vger.kernel.org
And finally both patches should have a fixes line, such as:
Fixes: 1851617cd2da ("PCI/MSI: Disable MSI at enumeration even if kernel doesn't support MSI")
cheers