linux-kernel - Re: [PATCH v2 net-next 0/2] net: snmp: tracepoint support for snmp

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <CADxym3ZfBVAecK-oFdMVV2gkOV6iUrq5XGkRZx3yXCuXDOS=2A@mail.gmail.com>
Date:   Fri, 19 Nov 2021 11:45:40 +0800
From:   Menglong Dong <menglong8.dong@...il.com>
To:     David Ahern <dsahern@...il.com>
Cc:     Jakub Kicinski <kuba@...nel.org>,
        Steven Rostedt <rostedt@...dmis.org>,
        David Miller <davem@...emloft.net>, mingo@...hat.com,
        Hideaki YOSHIFUJI <yoshfuji@...ux-ipv6.org>,
        dsahern@...nel.org, Menglong Dong <imagedong@...cent.com>,
        Yuchung Cheng <ycheng@...gle.com>, kuniyu@...zon.co.jp,
        LKML <linux-kernel@...r.kernel.org>,
        netdev <netdev@...r.kernel.org>
Subject: Re: [PATCH v2 net-next 0/2] net: snmp: tracepoint support for snmp

Hello~

On Thu, Nov 18, 2021 at 11:36 PM David Ahern <dsahern@...il.com> wrote:
>
[...]
>
> there is already good infrastructure around kfree_skb - e.g., drop watch
> monitor. Why not extend that in a way that other drop points can benefit
> over time?
>

Thanks for your advice.

In fact, I don't think that this is a perfect idea. This way may have benefit
of reuse the existing kfree_skb event, but this will do plentiful modification
to the current code. For example, in tcp_v4_rcv(), you need to introduce the
new variate 'int free_reason' and record the drop reason in it, and pass
it to 'kfree_skb_with_reason()' in 'discard_it:'. Many places need this kind
modification. What's more, some statistics don't use 'kfree_skb()'.

However, with the tracepoint for snmp, we just need to pass 'skb' to
'UDP_INC_STATS()/TCP_INC_STATS()', the reason is already included.
This way, the modification is more simple and easier to maintain.

Thanks!
Menglong Dong

> e.g., something like this (uncompiled and not tested; and to which
> Steven is going to suggest strings for the reason):
>
> diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
> index 0bd6520329f6..e66e634acad0 100644
> --- a/include/linux/skbuff.h
> +++ b/include/linux/skbuff.h
> @@ -1075,8 +1075,13 @@ static inline bool skb_unref(struct sk_buff *skb)
>         return true;
>  }
>
> +enum skb_drop_reason {
> +       SKB_DROP_REASON_NOT_SPECIFIED,
> +       SKB_DROP_REASON_CSUM,
> +}
>  void skb_release_head_state(struct sk_buff *skb);
>  void kfree_skb(struct sk_buff *skb);
> +void kfree_skb_with_reason(struct sk_buff *skb, enum skb_drop_reason);
>  void kfree_skb_list(struct sk_buff *segs);
>  void skb_dump(const char *level, const struct sk_buff *skb, bool full_pkt);
>  void skb_tx_error(struct sk_buff *skb);
> diff --git a/include/trace/events/skb.h b/include/trace/events/skb.h
> index 9e92f22eb086..2a2d263f9d46 100644
> --- a/include/trace/events/skb.h
> +++ b/include/trace/events/skb.h
> @@ -14,7 +14,7 @@
>   */
>  TRACE_EVENT(kfree_skb,
>
> -       TP_PROTO(struct sk_buff *skb, void *location),
> +       TP_PROTO(struct sk_buff *skb, void *location, enum
> skb_drop_reason reason),
>
>         TP_ARGS(skb, location),
>
> @@ -22,16 +22,18 @@ TRACE_EVENT(kfree_skb,
>                 __field(        void *,         skbaddr         )
>                 __field(        void *,         location        )
>                 __field(        unsigned short, protocol        )
> +               __field(        unsigned int,   reason          )
>         ),
>
>         TP_fast_assign(
>                 __entry->skbaddr = skb;
>                 __entry->location = location;
>                 __entry->protocol = ntohs(skb->protocol);
> +               __entry->reason = reason;
>         ),
>
> -       TP_printk("skbaddr=%p protocol=%u location=%p",
> -               __entry->skbaddr, __entry->protocol, __entry->location)
> +       TP_printk("skbaddr=%p protocol=%u location=%p reason %u",
> +               __entry->skbaddr, __entry->protocol, __entry->location,
> __entry->reason)
>  );
>
>  TRACE_EVENT(consume_skb,
> diff --git a/net/core/skbuff.c b/net/core/skbuff.c
> index 67a9188d8a49..388059bda3d1 100644
> --- a/net/core/skbuff.c
> +++ b/net/core/skbuff.c
> @@ -770,11 +770,29 @@ void kfree_skb(struct sk_buff *skb)
>         if (!skb_unref(skb))
>                 return;
>
> -       trace_kfree_skb(skb, __builtin_return_address(0));
> +       trace_kfree_skb(skb, __builtin_return_address(0),
> SKB_DROP_REASON_NOT_SPECIFIED);
>         __kfree_skb(skb);
>  }
>  EXPORT_SYMBOL(kfree_skb);
>
> +/**
> + *     kfree_skb_with_reason - free an sk_buff
> + *     @skb: buffer to free
> + *     @reason: enum describing why the skb is dropped
> + *
> + *     Drop a reference to the buffer and free it if the usage count has
> + *     hit zero.
> + */
> +void kfree_skb_with_reason(struct sk_buff *skb, enum skb_drop_reason
> reason);
> +{
> +       if (!skb_unref(skb))
> +               return;
> +
> +       trace_kfree_skb(skb, __builtin_return_address(0), reason);
> +       __kfree_skb(skb);
> +}
> +EXPORT_SYMBOL(kfree_skb_with_reason);
> +
>  void kfree_skb_list(struct sk_buff *segs)
>  {
>         while (segs) {