Re: [Qemu-devel] [PATCH v4 2/8] ahci: MSI capability should be at 0x80, not 0x50.

[John Snow <jsnow@redhat.com>]

On 09/17/2014 12:42 PM, Michael S. Tsirkin wrote:
> On Wed, Sep 17, 2014 at 05:54:28PM +0200, Jan Kiszka wrote:
>> On 2014-08-21 19:44, John Snow wrote:
>>> In the Intel ICH9 data sheet, the MSI capability offset
>>> in the PCI configuration space for ICH9 AHCI devices is
>>> specified to be 0x80.
>>>
>>> Further, the PCI capability pointer should always point
>>> to 0x80 in ICH9 devices, despite the fact that AHCI 1.3
>>> specifies that it should be pointing to PMCAP (Which in
>>> this instance would be 0x70) to maintain adherence to
>>> the Intel data sheet specifications and real observed behavior.
>>>
>>> Signed-off-by: John Snow <jsnow@redhat.com>
>>> ---
>>>   hw/ide/ich.c | 7 ++++++-
>>>   1 file changed, 6 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/hw/ide/ich.c b/hw/ide/ich.c
>>> index a2f1639..8eb77a1 100644
>>> --- a/hw/ide/ich.c
>>> +++ b/hw/ide/ich.c
>>> @@ -71,6 +71,7 @@
>>>   #include <hw/ide/pci.h>
>>>   #include <hw/ide/ahci.h>
>>>
>>> +#define ICH9_MSI_CAP_OFFSET     0x80
>>>   #define ICH9_SATA_CAP_OFFSET    0xA8
>>>
>>>   #define ICH9_IDP_BAR            4
>>> @@ -115,7 +116,6 @@ static int pci_ich9_ahci_init(PCIDevice *dev)
>>>       /* XXX Software should program this register */
>>>       dev->config[0x90]   = 1 << 6; /* Address Map Register - AHCI mode */
>>>
>>> -    msi_init(dev, 0x50, 1, true, false);
>>>       d->ahci.irq = pci_allocate_irq(dev);
>>>
>>>       pci_register_bar(dev, ICH9_IDP_BAR, PCI_BASE_ADDRESS_SPACE_IO,
>>> @@ -135,6 +135,11 @@ static int pci_ich9_ahci_init(PCIDevice *dev)
>>>                    (ICH9_IDP_BAR + 0x4) | (ICH9_IDP_INDEX_LOG2 << 4));
>>>       d->ahci.idp_offset = ICH9_IDP_INDEX;
>>>
>>> +    /* Although the AHCI 1.3 specification states that the first capability
>>> +     * should be PMCAP, the Intel ICH9 data sheet specifies that the ICH9
>>> +     * AHCI device puts the MSI capability first, pointing to 0x80. */
>>> +    msi_init(dev, ICH9_MSI_CAP_OFFSET, 1, true, false);
>>> +
>>>       return 0;
>>>   }
>>>
>>>
>>
>> I did this for HDA recently, and it became clear that this requires a
>> compat switch to expose the old, broken layout to older guests.
>> Otherwise you can't migrate from them.
>>
>> Jan
>
> Absolutely but ahci still disables migration :)

I was halfway to preparing a v5, but you're right ... We've disabled 
AHCI migration until we iron out the most egregious kinks.

>
>> --
>> Siemens AG, Corporate Technology, CT RTC ITP SES-DE
>> Corporate Competence Center Embedded Linux

-- 
—js