On 2015/5/14 12:05, Hanjun Guo wrote:

On 2015?05?14? 09:09, Jiang Liu wrote:

quoted

On 2015/5/13 21:25, Hanjun Guo wrote:

quoted

On 2015?05?13? 20:24, Jiang Liu wrote:

quoted

On 2015/5/13 17:29, Hanjun Guo wrote:

quoted

Hi Jiang,

On 2015?05?05? 10:46, Jiang Liu wrote:

struct pci_controller {
          struct acpi_device *companion;
          void *iommu;
          int segment;
          int node;               /* nearest node with memory or
NUMA_NO_NODE for global allocation */

          void *platform_data;
};

except void *platform_data;

On ARM64, the structure is almost the same, so how about
introduce

struct pci_controller {
          struct acpi_device *companion;  /* ACPI companion device */
          void            *iommu;         /* IOMMU private data */
          int             segment;        /* PCI domain */
          int             node;           /* NUMA node */
#ifdef CONFIG_IA64
      void *platform_data;
#endif
};

in this file, then can be used for all architectures?

Current mode is that architecture defines its own version of
struct pci_controller. It would be better to keep this pattern.

OK, thanks for the clarify :) So how about add my basic
PCI support patch for ARM64 on top of you patch set to fix
this problem?

Sure, please send me the patches and I will send out v3 to
cover your review comments.

OK, I need to rework my patches because my patch set is dependent
on top of another MMCFG refactor patch set [1], so I need to remove
MMCONFIG first then will speed up the upstream process of your patch
set, will send you the patches soon.

Hi Hanjun,
	I will send out v3 soon , so you could rebase onto v3.
There are changes which will affect your rebase.
Thanks!
Gerry

[1]: https://lkml.org/lkml/2015/4/17/29

Thanks
Hanjun