[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <aJnJ2jnFdu9V-j1Z@infradead.org>
Date: Mon, 11 Aug 2025 03:45:46 -0700
From: Christoph Hellwig <hch@...radead.org>
To: cen zhang <zzzccc427@...il.com>
Cc: cem@...nel.org, linux-kernel@...r.kernel.org, baijiaju1990@...il.com,
zhenghaoran154@...il.com, r33s3n6@...il.com, gality365@...il.com,
linux-xfs@...r.kernel.org
Subject: Re: [BUG] xfs: Assertion failure in dio_write( flags &
IOMAP_DIO_OVERWRITE_ONLY) with a UAF
On Wed, Aug 06, 2025 at 07:40:19PM +0800, cen zhang wrote:
> Hello maintainers,
>
> I would like to report a kernel panic found using syzkaller on a 6.16.0-rc6.
>
> The kernel log shows two distinct but closely timed crash reports,
> which I guess are related.
>
> 1. An XFS assertion failure: Assertion failed: flags &
> IOMAP_DIO_OVERWRITE_ONLY, file: fs/xfs/xfs_file.c, line: 876 triggered
> by a write() system call in xfs_file_dio_write_unaligned.
>
> 2. A KASAN use-after-free report on a task_struct object, triggered
> during an ioctl() call (likely FICLONE or FIDEDUPERANGE). The crash
> occurs in rwsem_down_write_slowpath when trying to lock an inode via
> xfs_reflink_remap_prep.
>
> Unfortunately, I have not been able to create a standalone C
> reproducer, and attempts to use syzkaller's repro tool on the syz-prog
> have not reliably triggered the bug again.
Thanks for the report, but it will be really hard to do anything without
a reproducer. I case you are still trying to create one it would be
great to hear if you have one!
Powered by blists - more mailing lists