linux-mips
[Top] [All Lists]

Re: [PATCH 02/12] MIPS: Netlogic: MSI enable fix for XLS

To: "Sergei Shtylyov" <sshtylyov@mvista.com>
Subject: Re: [PATCH 02/12] MIPS: Netlogic: MSI enable fix for XLS
From: "Jayachandran C." <jayachandranc@netlogicmicro.com>
Date: Mon, 30 Apr 2012 14:59:57 +0530
Cc: linux-mips@linux-mips.org, ralf@linux-mips.org, "Ganesan Ramalingam" <ganesanr@broadcom.com>
In-reply-to: <4F9D8E12.9090006@mvista.com>
References: <1335618738-4679-1-git-send-email-jayachandranc@netlogicmicro.com> <1335618738-4679-3-git-send-email-jayachandranc@netlogicmicro.com> <4F9D8E12.9090006@mvista.com>
Sender: linux-mips-bounce@linux-mips.org
User-agent: Mutt/1.5.21 (2010-09-15)
On Sun, Apr 29, 2012 at 10:53:06PM +0400, Sergei Shtylyov wrote:
> Hello.
> 
> On 28-04-2012 17:12, Jayachandran C wrote:
> 
> >From: Ganesan Ramalingam<ganesanr@broadcom.com>
> 
> >MSI interrupts do not work on XLS after commit a776c49, because
> 
>    Please also specify that commit's summary in parens.
> 
> >the change disables MSI interrupts on the XLS PCIe bridges at boot-up.
> 
> >Fix this by enabling MSI interrupts on the bridge in the
> >arch_setup_msi_irq() function. Earlier, this was done from firmware
> >and we did not need to change the configuration in linux.
> 
> >Signed-off-by: Ganesan Ramalingam<ganesanr@broadcom.com>
> >Signed-off-by: Jayachandran C<jayachandranc@netlogicmicro.com>
> >---
> >  arch/mips/pci/pci-xlr.c |   30 +++++++++++++++++++++++++-----
> >  1 file changed, 25 insertions(+), 5 deletions(-)
> >
> >diff --git a/arch/mips/pci/pci-xlr.c b/arch/mips/pci/pci-xlr.c
> >index 50ff4dc..003e053 100644
> >--- a/arch/mips/pci/pci-xlr.c
> >+++ b/arch/mips/pci/pci-xlr.c
> [...]
> >@@ -168,17 +169,17 @@ static int get_irq_vector(const struct pci_dev *dev)
> >     if (dev->bus->self == NULL)
> >             return 0;
> >
> >-    switch  (dev->bus->self->devfn) {
> >-    case 0x0:
> >+    switch  (PCI_SLOT(dev->bus->self->devfn)) {
> >+    case 0:
> >             return PIC_PCIE_LINK0_IRQ;
> >-    case 0x8:
> >+    case 1:
> >             return PIC_PCIE_LINK1_IRQ;
> >-    case 0x10:
> >+    case 2:
> >             if (nlm_chip_is_xls_b())
> >                     return PIC_PCIE_XLSB0_LINK2_IRQ;
> >             else
> >                     return PIC_PCIE_LINK2_IRQ;
> >-    case 0x18:
> >+    case 3:
> >             if (nlm_chip_is_xls_b())
> >                     return PIC_PCIE_XLSB0_LINK3_IRQ;
> >             else
> 
>    This seems like un unrelated change...

This is just a slightly better way of writing the same logic, it does
not fix any issue.

We should have added this to the commit description, but I'm not sure
if it requires a separate patch.

JC.


<Prev in Thread] Current Thread [Next in Thread>