Re: [PATCH v4] Add pcie_hp=nomsi to disable MSI/MSI-X for pciehp driver

[MUNEDA Takahiro <muneda.takahiro@jp.fujitsu.com>]

On Thu, 19 Jan 2012 15:41:09 -0800,
Yinghai Lu <yinghai@kernel.org> wrote:

snip

>> Index: linux-3.2/drivers/pci/quirks.c
>> ===================================================================
>> --- linux-3.2.orig/drivers/pci/quirks.c
>> +++ linux-3.2/drivers/pci/quirks.c
>> @@ -2665,6 +2665,35 @@ DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_AT
>>   DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_ATI, 0x4375,
>>                         quirk_msi_intx_disable_bug);
>>
>> +/*
>> + * The PCI bridge from Integrated Device Technology(IDT) has a problem
>> + * with MSI/MSI-X.
>> + * An un-wanted interrupt is generated when PCI driver switches from
>> + * MSI/MSI-X to INTx while shutting down the PCI Express port, if it
>> + * provides the Native Hotplug service.  The interrupt does not happen
>> + * if MSI/MSI-X is not used on it.  Therefore, turn off MSI/MSI-X for
>> + * this device.
>> + */
>> +static void __devinit idt_bridge_msi_disable(struct pci_dev *dev)
>> +{
>> +       int pos;
>> +       u16 reg16;
>> +       u32 reg32;
>> +
>> +       pos = pci_pcie_cap(dev);
>> +       if (!pos)
>> +               return;
>> +       pci_read_config_word(dev, pos + PCI_EXP_FLAGS,&reg16);
>> +       if (!(reg16&  PCI_EXP_FLAGS_SLOT))
>> +               return;
>> +       pci_read_config_dword(dev, pos + PCI_EXP_SLTCAP,&reg32);
>> +       if (reg32&  PCI_EXP_SLTCAP_HPC) {
>> +               pci_msi_off(dev);
>> +               dev->no_msi = 1;
>> +       }
>> +}
>> +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_IDT, 0x807f, idt_bridge_msi_disable);
>
> do you have statement for IDT, says that is chipset problem ?

I'm trying to contact to them, but not yet.

> it could be your platform fw problem.
>
> if that is the case, you will disable MSI/MSIX for other platform that
> do not have fw problem.

Yinghai,
Btw, how about other pieces?  I'm thinking to drop the quirk code 
temporary, until I get the comment from them.
Can you give an ACK if I remove the hardware specific quirk code?

Thanks,
Takahiro