Re: [Xen-devel] [PATCH] OvmfPkg: AcpiPlatformDxe: PCI enumeration may be disabled

[Laszlo Ersek <lersek@xxxxxxxxxx>]

On 02/12/15 13:29, Wei Liu wrote:
> On Thu, Feb 12, 2015 at 01:16:07PM +0100, Laszlo Ersek wrote:
>> SVN r16411 delayed ACPI table installation until PCI enumeration was
>> complete, because on QEMU the ACPI-related fw_cfg files should only be
>> downloaded after PCI enumeration.
>>
>> However, InitializeXen() in "OvmfPkg/PlatformPei/Xen.c" sets
>> PcdPciDisableBusEnumeration to TRUE. This causes
>> PciBusDriverBindingStart() in "MdeModulePkg/Bus/Pci/PciBusDxe/PciBus.c" to
>> set gFullEnumeration to FALSE, which in turn makes PciEnumerator() in
>> "MdeModulePkg/Bus/Pci/PciBusDxe/PciEnumerator.c" branch to
>> PciEnumeratorLight(). The installation of
>> EFI_PCI_ENUMERATION_COMPLETE_PROTOCOL at the end of PciEnumerator() is not
>> reached.
>>
>> Which means that starting with SVN r16411, AcpiPlatformDxe is never
>> dispatched on Xen.
>>
>> This patch replaces the EFI_PCI_ENUMERATION_COMPLETE_PROTOCOL depex with a
>> matching protocol registration callback for the PCI enumeration enabled
>> (ie. QEMU) case. When PCI enumeration is disabled (ie. when running on
>> Xen), AcpiPlatformDxe doesn't wait for
>> EFI_PCI_ENUMERATION_COMPLETE_PROTOCOL.
>>
>> Contributed-under: TianoCore Contribution Agreement 1.0
>> Signed-off-by: Laszlo Ersek <lersek@xxxxxxxxxx>
> 
> This patch fixes the bug I have. Thanks!
> 
> Tested-by: Wei Liu <wei.liu2@xxxxxxxxxx>

Thanks. If Jordan is okay with the patch too, I'll push it.

Sorry about the regression, too.

Cheers
Laszlo


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel