From: Dongli Zhang <dongli.zhang@oracle.com>
To: netdev@vger.kernel.org, bpf@vger.kernel.org
Cc: linux-kernel@vger.kernel.org, rostedt@goodmis.org,
mingo@redhat.com, davem@davemloft.net, kuba@kernel.org,
edumazet@google.com, yoshfuji@linux-ipv6.org, dsahern@kernel.org,
ast@kernel.org, daniel@iogearbox.net, andrii@kernel.org,
imagedong@tencent.com, joao.m.martins@oracle.com,
joe.jin@oracle.com
Subject: [PATCH RFC 0/4] net: skb: to use (function, line) pair as reason for kfree_skb_reason()
Date: Thu, 3 Feb 2022 07:37:27 -0800 [thread overview]
Message-ID: <20220203153731.8992-1-dongli.zhang@oracle.com> (raw)
This RFC is to seek for suggestion to track the reason that the sk_buff is
dropped.
Sometimes the kernel may not directly call kfree_skb() to drop the sk_buff.
Instead, it "goto drop" and call kfree_skb() at 'drop'. This makes it
difficult to track the reason that the sk_buff is dropped.
The commit c504e5c2f964 ("net: skb: introduce kfree_skb_reason()") has
introduced the kfree_skb_reason() to help track the reason. However, we may
need to define many reasons for each driver/subsystem.
I am going to trace the "goto drop" in TUN and TAP drivers. However, I will
need to introduce many new reasons if I re-use kfree_skb_reason().
There are some other options.
1. The 1st option is to introduce a new tracepoint, e.g., trace_drop_skb()
as below to track the function and line number. We would call
trace_drop_skb() before "goto drop".
TP_PROTO(struct sk_buff *skb, struct net_device *dev,
const char *function, unsigned int line),
2. The 2nd option is to directly call trace_kfree_skb() before "goto drop".
And we may replace kfree_skb() with below kfree_skb_notrace() as suggested
by Joao Martins.
/**
* kfree_skb_notrace - free an sk_buff without tracing
* @skb: buffer to free
*
* Drop a reference to the buffer and free it if the usage count has
* hit zero.
*/
void kfree_skb_notrace(struct sk_buff *skb)
{
if (!skb_unref(skb))
return;
__kfree_skb(skb);
}
3. The last option is this RFC. To avoid introducing so many new reasons,
we use (__func__, __LINE__) to uniquely identify the location of
each "goto drop". The 'reason' introduced by the
commit c504e5c2f964 ("net: skb: introduce kfree_skb_reason()") is replaced
by the (function, line) pair.
The below is the sample output from trace_pipe by this RFC, when the
sk_buff is dropped by TUN driver.
<idle>-0 [016] ..s1. 432.701987: kfree_skb: skbaddr=00000000a65c0a72 protocol=0 location=000000008a49d80c function=none line=0
<idle>-0 [003] b.s2. 432.704397: kfree_skb: skbaddr=00000000665e5ccd protocol=2048 location=00000000ec3b7129 function=tun_net_xmit line=1116
<idle>-0 [003] ..s1. 432.704400: kfree_skb: skbaddr=00000000e4c806f8 protocol=2048 location=000000002929642d function=none line=0
<idle>-0 [002] b.s2. 432.734617: kfree_skb: skbaddr=00000000079749b3 protocol=2048 location=00000000ec3b7129 function=tun_net_xmit line=1116
<idle>-0 [015] b.s2. 432.880571: kfree_skb: skbaddr=00000000e1542f1e protocol=34525 location=00000000ec3b7129 function=tun_net_xmit line=1116
<idle>-0 [015] ..s1. 432.880577: kfree_skb: skbaddr=000000004f3022b6 protocol=34525 location=00000000547c5c25 function=none line=0
<idle>-0 [002] b.s2. 432.886247: kfree_skb: skbaddr=0000000062990a71 protocol=2054 location=00000000ec3b7129 function=tun_net_xmit line=1116
drivers/net/tap.c | 30 ++++++++++++++++++++++--------
drivers/net/tun.c | 33 +++++++++++++++++++++++++--------
include/linux/skbuff.h | 24 +++++++-----------------
include/trace/events/skb.h | 37 ++++++++-----------------------------
net/core/dev.c | 3 ++-
net/core/skbuff.c | 10 ++++++----
net/ipv4/tcp_ipv4.c | 14 +++++++-------
net/ipv4/udp.c | 14 +++++++-------
8 files changed, 84 insertions(+), 81 deletions(-)
Would you please share your suggestion and feedback?
Thank you very much!
Dongli Zhang
next reply other threads:[~2022-02-03 15:38 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-02-03 15:37 Dongli Zhang [this message]
2022-02-03 15:37 ` [PATCH RFC 1/4] net: skb: use line number to trace dropped skb Dongli Zhang
2022-02-03 15:48 ` David Ahern
2022-02-03 17:13 ` Dongli Zhang
2022-02-03 15:59 ` Eric Dumazet
2022-02-03 17:14 ` Dongli Zhang
2022-02-03 15:37 ` [PATCH RFC 2/4] net: skb: add function name as part of reason Dongli Zhang
2022-02-03 15:37 ` [PATCH RFC 3/4] net: tun: track dropped skb via kfree_skb_reason() Dongli Zhang
2022-02-03 15:37 ` [PATCH RFC 4/4] net: tap: " Dongli Zhang
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20220203153731.8992-1-dongli.zhang@oracle.com \
--to=dongli.zhang@oracle.com \
--cc=andrii@kernel.org \
--cc=ast@kernel.org \
--cc=bpf@vger.kernel.org \
--cc=daniel@iogearbox.net \
--cc=davem@davemloft.net \
--cc=dsahern@kernel.org \
--cc=edumazet@google.com \
--cc=imagedong@tencent.com \
--cc=joao.m.martins@oracle.com \
--cc=joe.jin@oracle.com \
--cc=kuba@kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=mingo@redhat.com \
--cc=netdev@vger.kernel.org \
--cc=rostedt@goodmis.org \
--cc=yoshfuji@linux-ipv6.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®