[dpdk-dev] [PATCH] vhost: fix packed ring defines declaration

[Ferruh Yigit]

On 11/22/2018 7:50 PM, Maxime Coquelin wrote:
> 
> 
> On 11/22/18 8:20 PM, Ferruh Yigit wrote:
>> On 11/22/2018 6:51 PM, Maxime Coquelin wrote:
>>>
>>>
>>> On 11/22/18 7:23 PM, Ferruh Yigit wrote:
>>>> On 11/22/2018 5:09 PM, Maxime Coquelin wrote:
>>>>> The packed ring defines were declared only if kernel
>>>>> header does not declare them.
>>>>> The problem is that they are not applied in upstream kernel,
>>>>> and some changes in the names have been required.
>>>>>
>>>>> This patch declares the defines unconditionally, which
>>>>> fixes potential build issues.
>>>>
>>>> +1 to address possible build issues.
>>>>
>>>>>
>>>>> Fixes: 297b1e7350f6 ("vhost: add virtio packed virtqueue defines")
>>>>> Cc: stable at dpdk.org
>>>>>
>>>>> Signed-off-by: Maxime Coquelin <maxime.coquelin at redhat.com>
>>>>> ---
>>>>>    lib/librte_vhost/vhost.h | 22 +++++++++++-----------
>>>>>    1 file changed, 11 insertions(+), 11 deletions(-)
>>>>>
>>>>> diff --git a/lib/librte_vhost/vhost.h b/lib/librte_vhost/vhost.h
>>>>> index 760f42192..5218f1b12 100644
>>>>> --- a/lib/librte_vhost/vhost.h
>>>>> +++ b/lib/librte_vhost/vhost.h
>>>>> @@ -219,13 +219,6 @@ struct vhost_msg {
>>>>>    
>>>>>    #define VIRTIO_F_RING_PACKED 34
>>>>>    
>>>>> -#define VRING_DESC_F_NEXT	1
>>>>> -#define VRING_DESC_F_WRITE	2
>>>>> -#define VRING_DESC_F_INDIRECT	4
>>>>
>>>> Why these are not re-defined below? Not used?
>>>
>>> These defines are in kernel headers since the beginning of virtio
>>> support, so there are no reasons to keep them (and they aren't packed-
>>> ring specific).
>>>
>>> I can let them if you prefer, it does not hurt but shouldn't be here.
>>
>> Not required, I just want to confirm nothing missed.
>>
>>>
>>>>> -
>>>>> -#define VRING_DESC_F_AVAIL	(1ULL << 7)
>>>>> -#define VRING_DESC_F_USED	(1ULL << 15)
>>>>> -
>>>>>    struct vring_packed_desc {
>>>>>    	uint64_t addr;
>>>>>    	uint32_t len;
>>>>> @@ -233,16 +226,23 @@ struct vring_packed_desc {
>>>>>    	uint16_t flags;
>>>>>    };
>>>>>    
>>>>> -#define VRING_EVENT_F_ENABLE 0x0
>>>>> -#define VRING_EVENT_F_DISABLE 0x1
>>>>> -#define VRING_EVENT_F_DESC 0x2
>>>>> -
>>>>>    struct vring_packed_desc_event {
>>>>>    	uint16_t off_wrap;
>>>>>    	uint16_t flags;
>>>>>    };
>>>>>    #endif
>>>>>    
>>>>> +/*
>>>>> + * Declare below packed ring defines unconditionally
>>>>> + * as Kernel header might use different names.
>>>>> + */
>>>>> +#define VRING_DESC_F_AVAIL	(1ULL << 7)
>>>>> +#define VRING_DESC_F_USED	(1ULL << 15)
>>>>> +
>>>>> +#define VRING_EVENT_F_ENABLE 0x0
>>>>> +#define VRING_EVENT_F_DISABLE 0x1
>>>>> +#define VRING_EVENT_F_DESC 0x2
>>>>
>>>> What if Linux changes mind and uses old names again, build will fail again. If
>>>> related part in Linux is not released yet, what do you think being on safe side
>>>> and adding these defines with "#ifndef" wrap?
>>>
>>> There are the "old" ones.
>>> In any case it will work (I tested it).
>>
>> I see they are the old ones but is there any chance that kernel uses them again
>> (for some odd reason) ? Which can break the build. Should we add a protection?
> 
> Yes, there is a chance, but that would not break build, it will just
> override it.

It will give redefined warning if values are different but right, won't cause
any issue with same number.

> 
>>
>>>
>>> In a future release, I plan to get rid of this dependency with Kernel
>>> which only causes problems.
>>>
>>> There was a benefit at the beginning when all needed declarations where
>>> in the Kernel headers. But now, every time we add a new virtio feature,
>>> we have to check kernel supports it and if not define it.
>>>
>>> Thanks,
>>> Maxime
>>>>> +
>>>>>    /*
>>>>>     * Available and used descs are in same order
>>>>>     */
>>>>>
>>>>
>>