Re: [PATCH 3/4] drivers/vfio: New IOCTL command VFIO_EEH_INFO

[Alexander Graf <agraf@suse.de>]

On 20.05.14 13:40, Gavin Shan wrote:
> On Tue, May 20, 2014 at 01:28:40PM +0200, Alexander Graf wrote:
>> On 20.05.14 13:21, Alexander Graf wrote:
>>> On 20.05.14 10:30, Gavin Shan wrote:
>>>> The patch adds new IOCTL command VFIO_EEH_OP to VFIO PCI device
>>>> to support EEH functionality for PCI devices, which have been
>>>> passed from host to guest via VFIO.
>>>>
>>>> Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
>>>> ---
>>>>   arch/powerpc/platforms/powernv/Makefile   |   1 +
>>>>   arch/powerpc/platforms/powernv/eeh-vfio.c | 445
>>>> ++++++++++++++++++++++++++++++
>>>>   drivers/vfio/pci/vfio_pci.c               |  24 +-
>>>>   drivers/vfio/pci/vfio_pci_private.h       |  16 ++
>>>>   include/uapi/linux/vfio.h                 |  43 +++
>>>>   5 files changed, 523 insertions(+), 6 deletions(-)
>>>>   create mode 100644 arch/powerpc/platforms/powernv/eeh-vfio.c
>>> Why doesn't this code live inside the vfio module? If I don't load
>>> the vfio module, I don't need that code to waste memory in my
>>> kernel, no?
> Yes, It saves some memory.
>
>> So I think from a modeling point of view, you want VFIO code that
>> calls reasonably generic helpers inside the kernel to deal with
>> errors.
>>
>> The "generic helpers" don't have anything to do with VFIO. Everything
>> that interfaces via ioctls with user space is 100% VFIO code.
>>
>> The latter should be tristate inside vfio.ko, the former can be =y.
>>
> The main reason I put eeh-vfio.c to arch/powerpc/platforms/powernv/ is
> the source file needs access data structures (struct pnv_phb) defined
> in "pci.h" under that directory.

Then create a good in-kernel framework from that directory and make use 
of it from the VFIO code :). But please don't mesh together VFIO, 
powernv EEH handling and RTAS.


Alex