netdev - Odd behaviour with SO

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>] [day] [month] [year] [list]

Date:   Fri, 15 Jul 2022 19:43:12 -0500
From:   Zeb Figura <zfigura@...eweavers.com>
To:     netdev@...r.kernel.org
Subject: Odd behaviour with SO_OOBINLINE

Hello all,

I found what seems like a bug related to SO_OOBINLINE handling with TCP 
sockets. I couldn't easily find any mention of this on the Internet, so 
I'm asking about it here.

The basic problem is best expressed by reading and running the attached 
.c file. As a summary, setting SO_OOBINLINE on a socket after sending 
and receiving OOB data seems to cause that data to be received again. 
This is especially suspicious since making an (unsuccessful) recv() call 
before setting SO_OOBINLINE [commented out with if(0)] will cause said 
recv to return normally.

Is this a bug? Is there anything that should be done about this behaviour?

Thanks,
Zeb
View attachment "socktest_oob.c" of type "text/x-csrc" (2783 bytes)