RE: [PATCH v4] skmsg: lose offset info in sk_psock_skb_ingress

-----Original Message-----
From: John Fastabend [mailto:john.fastabend@gmail.com]
Sent: Friday, October 1, 2021 6:48 AM
To: liujian (CE) <redacted>; john.fastabend@gmail.com;
daniel@iogearbox.net; jakub@cloudflare.com; lmb@cloudflare.com;
davem@davemloft.net; kuba@kernel.org; ast@kernel.org;
andrii@kernel.org; kafai@fb.com; songliubraving@fb.com; yhs@fb.com;
kpsingh@kernel.org; netdev@vger.kernel.org; bpf@vger.kernel.org;
xiyou.wangcong@gmail.com
Cc: liujian (CE) <redacted>
Subject: RE: [PATCH v4] skmsg: lose offset info in sk_psock_skb_ingress

Liu Jian wrote:

quoted

If sockmap enable strparser, there are lose offset info in
sk_psock_skb_ingress. If the length determined by parse_msg function
is not skb->len, the skb will be converted to sk_msg multiple times,
and userspace app will get the data multiple times.

Fix this by get the offset and length from strp_msg.
And as Cong suggestion, add one bit in skb->_sk_redir to distinguish
enable or disable strparser.

Signed-off-by: Liu Jian <redacted>
---

Thanks. Please add Fixes tags so we can track these I've added it here.

This has been broken from the initial patches and after a quick glance I
suspect this will need manual backports if we need it. Also all the I use and all
the selftests set parser to a nop by returning skb->len.

Can you also create a test so we can ensure we don't break this again?

Okay, I will do this after the holiday.

Fixes: 604326b41a6fb ("bpf, sockmap: convert to generic sk_msg interface")
Acked-by: John Fastabend <john.fastabend@gmail.com>

Thank you for reviewing this patch again.