linux-kernel - Re: INFO: task hung in chmod

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 for Android: free password hash cracker in your pocket

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

Message-ID: <20250403-teppich-geist-a877dd4158f2@brauner>
Date: Thu, 3 Apr 2025 12:30:49 +0200
From: Christian Brauner <brauner@...nel.org>
To: 白烁冉 <baishuoran@...eu.edu.cn>
Cc: Alexander Viro <viro@...iv.linux.org.uk>, Jan Kara <jack@...e.cz>, 
	linux-fsdevel@...r.kernel.org, Kun Hu <huk23@...udan.edu.cn>, 
	Jiaji Qin <jjtan24@...udan.edu.cn>, linux-kernel@...r.kernel.org, syzkaller@...glegroups.com
Subject: Re: INFO: task hung in chmod_common

On Thu, Apr 03, 2025 at 05:33:31PM +0800, 白烁冉 wrote:
> Dear Maintainers,
> 
> When using our customized Syzkaller to fuzz the latest Linux kernel, the following crash (95th)was triggered.
> 
> 
> HEAD commit: 6537cfb395f352782918d8ee7b7f10ba2cc3cbf2
> git tree: upstream
> Output:https://github.com/pghk13/Kernel-Bug/tree/main/0305_6.14rc5/95-INFO_%20rcu%20detected%20stall%20in%20sys_chdir
> Kernel config:https://github.com/pghk13/Kernel-Bug/blob/main/0305_6.14rc5/config.txt
> C reproducer:https://github.com/pghk13/Kernel-Bug/blob/main/0305_6.14rc5/95-INFO_%20rcu%20detected%20stall%20in%20sys_chdir/95repro.c
> Syzlang reproducer: https://github.com/pghk13/Kernel-Bug/blob/main/0305_6.14rc5/95-INFO_%20rcu%20detected%20stall%20in%20sys_chdir/95call_trace.txt

We generally ignore any reports from non-official syzbot instances.

> 
> 
> 
> Our reproducer uses mounts a constructed filesystem image.
> This could be a file system deadlock issue that occurs on lines
> 547-548 of the chmod_common function. When this function is called,
> the code already holds the inode lock (via the inode_lock (inode)),
> but the notify_change may need to perform RCU-protected operations
> internally. The core of the problem is that the chmod_common function
> calls the notify_change while holding the inode lock, and the
> notify_change internally relies on the RCU protection mechanism. At a
> specific path to the SYSV file system, this combination results in a
> deadlock.

I'm not following at all but also sysv is removed in v6.15.