Re: [PATCH v2 net 2/2] af_unix: Don't peek OOB data without MSG_OOB.

[Kuniyuki Iwashima <kuniyu@amazon.com>]

From: Rao Shoaib <rao.shoaib@oracle.com>
Date: Tue, 16 Apr 2024 13:11:09 -0700
> The proposed fix is not the correct fix as among other things it does 
> not allow going pass the OOB if data is present. TCP allows that.

Ugh, exactly.

But the behaviour was broken initially, so the tag is

Fixes: 314001f0bf92 ("af_unix: Add OOB support")


TCP:
---8<---
>>> from socket import *
>>> s = socket()
>>> s.listen()
>>> c1 = socket()
>>> c1.connect(s.getsockname())
>>> c2, _ = s.accept()
>>> 
>>> c1.send(b'h', MSG_OOB)
1
>>> c1.send(b'ello')
4
>>> c2.recv(5, MSG_PEEK)
b'ello'
---8<---

Latest net.git
---8<---
>>> from socket import *
>>> c1, c2 = socketpair(AF_UNIX)
>>> c1.send(b'h', MSG_OOB)
1
>>> c1.send(b'ello')
4
>>> 
>>> c2.recv(5, MSG_PEEK)
^C
---8<---

314001f0bf92:
---8<---
>>> from socket import *
>>> c1, c2 = socketpair(AF_UNIX)
>>> c1.send(b'h', MSG_OOB)
1
>>> c1.send(b'ello')
4
>>> c2.recv(5, MSG_PEEK)
b'hello'
---8<---


> 
> I have attached a patch that fixes this issue and other issues that I 
> encountered in my testing. I compared TCP.

Could you post patches formally on top of the latest net.git ?
It seems one of my patch is squashed.

Also, please note that one patch should fix one issue.
The change in queue_oob() should be another patch.

Thanks!