Re: [PATCH V9 9/9] virtio: Intel IFC VF driver for VDPA

["Zhu, Lingshan" <lingshan.zhu@intel.com>]

On 4/9/2020 8:43 PM, Jason Wang wrote:
>
> On 2020/4/9 下午6:41, Arnd Bergmann wrote:
>> On Thu, Mar 26, 2020 at 3:08 PM Jason Wang <jasowang@redhat.com> wrote:
>>> From: Zhu Lingshan <lingshan.zhu@intel.com>
>>>
>>> This commit introduced two layers to drive IFC VF:
>>>
>>> (1) ifcvf_base layer, which handles IFC VF NIC hardware operations and
>>>      configurations.
>>>
>>> (2) ifcvf_main layer, which complies to VDPA bus framework,
>>>      implemented device operations for VDPA bus, handles device probe,
>>>      bus attaching, vring operations, etc.
>>>
>>> Signed-off-by: Zhu Lingshan <lingshan.zhu@intel.com>
>>> Signed-off-by: Bie Tiwei <tiwei.bie@intel.com>
>>> Signed-off-by: Wang Xiao <xiao.w.wang@intel.com>
>>> Signed-off-by: Jason Wang <jasowang@redhat.com>
>>> +
>>> +#define IFCVF_QUEUE_ALIGNMENT  PAGE_SIZE
>>> +#define IFCVF_QUEUE_MAX                32768
>>> +static u16 ifcvf_vdpa_get_vq_align(struct vdpa_device *vdpa_dev)
>>> +{
>>> +       return IFCVF_QUEUE_ALIGNMENT;
>>> +}
>> This fails to build on arm64 with 64kb page size (found in linux-next):
>>
>> /drivers/vdpa/ifcvf/ifcvf_main.c: In function 'ifcvf_vdpa_get_vq_align':
>> arch/arm64/include/asm/page-def.h:17:20: error: conversion from 'long
>> unsigned int' to 'u16' {aka 'short unsigned int'} changes value from
>> '65536' to '0' [-Werror=overflow]
>>     17 | #define PAGE_SIZE  (_AC(1, UL) << PAGE_SHIFT)
>>        |                    ^
>> drivers/vdpa/ifcvf/ifcvf_base.h:37:31: note: in expansion of macro 
>> 'PAGE_SIZE'
>>     37 | #define IFCVF_QUEUE_ALIGNMENT PAGE_SIZE
>>        |                               ^~~~~~~~~
>> drivers/vdpa/ifcvf/ifcvf_main.c:231:9: note: in expansion of macro
>> 'IFCVF_QUEUE_ALIGNMENT'
>>    231 |  return IFCVF_QUEUE_ALIGNMENT;
>>        |         ^~~~~~~~~~~~~~~~~~~~~
>>
>> It's probably good enough to just not allow the driver to be built in 
>> that
>> configuration as it's fairly rare but unfortunately there is no 
>> simple Kconfig
>> symbol for it.
>
>
> Or I think the 64KB alignment is probably more than enough.
>
> Ling Shan, can we use smaller value here?
>
> Thanks

sure, we just need it page aligned, this value is only used in 
get_vq_align(). I will try to find a arm64 machine and debug on this issue

Thanks!

>
>
>>
>> In a similar driver, we did
>>
>> config VMXNET3
>>          tristate "VMware VMXNET3 ethernet driver"
>>          depends on PCI && INET
>>          depends on !(PAGE_SIZE_64KB || ARM64_64K_PAGES || \
>>                       IA64_PAGE_SIZE_64KB || MICROBLAZE_64K_PAGES || \
>>                       PARISC_PAGE_SIZE_64KB || PPC_64K_PAGES)
>>
>> I think we should probably make PAGE_SIZE_64KB a global symbol
>> in arch/Kconfig and have it selected by the other symbols so drivers
>> like yours can add a dependency for it.
>>
>>           Arnd
>>
>