Re: [PATCH v6] net/core: Introduce netdev_core_stats_inc()

[Yajun Deng <yajun.deng@linux.dev>]

On 2023/9/29 00:32, Yajun Deng wrote:
>
> On 2023/9/29 00:23, Eric Dumazet wrote:
>> On Thu, Sep 28, 2023 at 6:16 PM Yajun Deng <yajun.deng@linux.dev> wrote:
>>>
>>> On 2023/9/28 23:44, Eric Dumazet wrote:
>>>> On Thu, Sep 28, 2023 at 5:40 PM Yajun Deng <yajun.deng@linux.dev> 
>>>> wrote:
>>>>> On 2023/9/28 22:18, Eric Dumazet wrote:
>>>>>> On Thu, Sep 28, 2023 at 12:04 PM Yajun Deng 
>>>>>> <yajun.deng@linux.dev> wrote:
>>>>>>> Although there is a kfree_skb_reason() helper function that can 
>>>>>>> be used to
>>>>>>> find the reason why this skb is dropped, but most callers didn't 
>>>>>>> increase
>>>>>>> one of rx_dropped, tx_dropped, rx_nohandler and 
>>>>>>> rx_otherhost_dropped.
>>>>>>>
>>>>>>> For the users, people are more concerned about why the dropped 
>>>>>>> in ip
>>>>>>> is increasing.
>>>>>>>
>>>>>>> Introduce netdev_core_stats_inc() for trace the caller of the 
>>>>>>> dropped
>>>>>>> skb. Also, add __code to netdev_core_stats_alloc(), as it's called
>>>>>>> unlinkly.
>>>>>>>
>>>>>>> Signed-off-by: Yajun Deng <yajun.deng@linux.dev>
>>>>>>> Suggested-by: Alexander Lobakin <aleksander.lobakin@intel.com>
>>>>>>> ---
>>>>>>> v6: merge netdev_core_stats and netdev_core_stats_inc together
>>>>>>> v5: Access the per cpu pointer before reach the relevant offset.
>>>>>>> v4: Introduce netdev_core_stats_inc() instead of export 
>>>>>>> dev_core_stats_*_inc()
>>>>>>> v3: __cold should be added to the netdev_core_stats_alloc().
>>>>>>> v2: use __cold instead of inline in dev_core_stats().
>>>>>>> v1: 
>>>>>>> https://lore.kernel.org/netdev/20230911082016.3694700-1-yajun.deng@linux.dev/
>>>>>>> ---
>>>>>>>     include/linux/netdevice.h | 21 ++++-----------------
>>>>>>>     net/core/dev.c            | 17 +++++++++++++++--
>>>>>>>     2 files changed, 19 insertions(+), 19 deletions(-)
>>>>>>>
>>>>>>> diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
>>>>>>> index 7e520c14eb8c..eb1fa04fbccc 100644
>>>>>>> --- a/include/linux/netdevice.h
>>>>>>> +++ b/include/linux/netdevice.h
>>>>>>> @@ -4002,32 +4002,19 @@ static __always_inline bool 
>>>>>>> __is_skb_forwardable(const struct net_device *dev,
>>>>>>>            return false;
>>>>>>>     }
>>>>>>>
>>>>>>> -struct net_device_core_stats __percpu 
>>>>>>> *netdev_core_stats_alloc(struct net_device *dev);
>>>>>>> -
>>>>>>> -static inline struct net_device_core_stats __percpu 
>>>>>>> *dev_core_stats(struct net_device *dev)
>>>>>>> -{
>>>>>>> -       /* This READ_ONCE() pairs with the write in 
>>>>>>> netdev_core_stats_alloc() */
>>>>>>> -       struct net_device_core_stats __percpu *p = 
>>>>>>> READ_ONCE(dev->core_stats);
>>>>>>> -
>>>>>>> -       if (likely(p))
>>>>>>> -               return p;
>>>>>>> -
>>>>>>> -       return netdev_core_stats_alloc(dev);
>>>>>>> -}
>>>>>>> +void netdev_core_stats_inc(struct net_device *dev, u32 offset);
>>>>>>>
>>>>>>>     #define DEV_CORE_STATS_INC(FIELD) \
>>>>>>>     static inline void dev_core_stats_##FIELD##_inc(struct 
>>>>>>> net_device *dev)                \
>>>>>>> { \
>>>>>>> -       struct net_device_core_stats __percpu 
>>>>>>> *p;                               \
>>>>>>> - \
>>>>>>> -       p = dev_core_stats(dev); \
>>>>>>> -       if (p) \
>>>>>>> - this_cpu_inc(p->FIELD); \
>>>>>> Note that we were using this_cpu_inc() which implied :
>>>>>> - IRQ safety, and
>>>>>> - a barrier paired with :
>>>>>>
>>>>>> net/core/dev.c:10548: storage->rx_dropped +=
>>>>>> READ_ONCE(core_stats->rx_dropped);
>>>>>> net/core/dev.c:10549: storage->tx_dropped +=
>>>>>> READ_ONCE(core_stats->tx_dropped);
>>>>>> net/core/dev.c:10550: storage->rx_nohandler +=
>>>>>> READ_ONCE(core_stats->rx_nohandler);
>>>>>> net/core/dev.c:10551: storage->rx_otherhost_dropped
>>>>>> += READ_ONCE(core_stats->rx_otherhost_dropped);
>>>>>>
>>>>>>
>>>>>>> + netdev_core_stats_inc(dev, \
>>>>>>> +                       offsetof(struct net_device_core_stats, 
>>>>>>> FIELD));         \
>>>>>>>     }
>>>>>>>     DEV_CORE_STATS_INC(rx_dropped)
>>>>>>>     DEV_CORE_STATS_INC(tx_dropped)
>>>>>>>     DEV_CORE_STATS_INC(rx_nohandler)
>>>>>>>     DEV_CORE_STATS_INC(rx_otherhost_dropped)
>>>>>>> +#undef DEV_CORE_STATS_INC
>>>>>>>
>>>>>>>     static __always_inline int ____dev_forward_skb(struct 
>>>>>>> net_device *dev,
>>>>>>> struct sk_buff *skb,
>>>>>>> diff --git a/net/core/dev.c b/net/core/dev.c
>>>>>>> index 606a366cc209..88a32c392c1d 100644
>>>>>>> --- a/net/core/dev.c
>>>>>>> +++ b/net/core/dev.c
>>>>>>> @@ -10497,7 +10497,8 @@ void netdev_stats_to_stats64(struct 
>>>>>>> rtnl_link_stats64 *stats64,
>>>>>>>     }
>>>>>>>     EXPORT_SYMBOL(netdev_stats_to_stats64);
>>>>>>>
>>>>>>> -struct net_device_core_stats __percpu 
>>>>>>> *netdev_core_stats_alloc(struct net_device *dev)
>>>>>>> +static __cold struct net_device_core_stats __percpu 
>>>>>>> *netdev_core_stats_alloc(
>>>>>>> +               struct net_device *dev)
>>>>>>>     {
>>>>>>>            struct net_device_core_stats __percpu *p;
>>>>>>>
>>>>>>> @@ -10510,7 +10511,19 @@ struct net_device_core_stats __percpu 
>>>>>>> *netdev_core_stats_alloc(struct net_device
>>>>>>>            /* This READ_ONCE() pairs with the cmpxchg() above */
>>>>>>>            return READ_ONCE(dev->core_stats);
>>>>>>>     }
>>>>>>> -EXPORT_SYMBOL(netdev_core_stats_alloc);
>>>>>>> +
>>>>>>> +void netdev_core_stats_inc(struct net_device *dev, u32 offset)
>>>>>>> +{
>>>>>>> +       /* This READ_ONCE() pairs with the write in 
>>>>>>> netdev_core_stats_alloc() */
>>>>>>> +       struct net_device_core_stats __percpu *p = 
>>>>>>> READ_ONCE(dev->core_stats);
>>>>>>> +
>>>>>>> +       if (unlikely(!p))
>>>>>>> +               p = netdev_core_stats_alloc(dev);
>>>>>>> +
>>>>>>> +       if (p)
>>>>>>> +               (*(unsigned long *)((void *)this_cpu_ptr(p) + 
>>>>>>> offset))++;
>>>>>> While here you are using a ++ operation that :
>>>>>>
>>>>>> - is not irq safe
>>>>>> - might cause store-tearing.
>>>>>>
>>>>>> I would suggest a preliminary patch converting the "unsigned 
>>>>>> long" fields in
>>>>>> struct net_device_core_stats to local_t
>>>>> Do you mean it needs to revert the commit 6510ea973d8d ("net: Use
>>>>> this_cpu_inc() to increment
>>>>>
>>>>> net->core_stats") first? But it would allocate memory which breaks on
>>>>> PREEMPT_RT.
>>>> I think I provided an (untested) alternative.
>>>>
>>>> unsigned long __percpu *field = (__force unsigned long __percpu *)
>>>> ((__force u8 *)p + offset);
>>>> this_cpu_inc(field);
>>> unsigned long __percpu *field = (__force unsigned long __percpu *)
>>> ((__force u8 *)p + offset);
>>> this_cpu_inc(*(int *)field);
>>>
>>> This would compiler success. But I didn't test it.
>>> This cold look complex.
>> Why exactly ? Not very different from the cast you already had.
> Okay, I'll test it.


It seems something wrong.

"ip -s a" would see the 'dropped' is increasing. But I cann't trace 
anything by the following cmd.

"sudo  python3  /usr/share/bcc/tools/trace netdev_core_stats_inc"

If I change back to "(*(unsigned long *)((void *)this_cpu_ptr(p) + 
offset))++; ", I can trace the caller.

So the following code would accidentally change somthing.

unsigned long __percpu *field = (__force unsigned long __percpu *) 
((__force u8 *)p + offset);

this_cpu_inc(*field);

>>
>>> Shoud I base v3? Export dev_core_stats_*_inc() intead of introduce 
>>> netdev_core_stats_inc().
>>> That would be easy.
>> Well, you tell me, but this does not look incremental to me.
>>
>> I do not think we need 4 different (and maybe more to come if struct
>> net_device_core_stats
>> grows in the future) functions for some hardly used path.