Date: Wed, 17 Oct 2007 20:33:21 +0100
From: Al Viro <viro@ftp.linux.org.uk>
To: Roland Dreier <rdreier@cisco.com>
Cc: Andrew Morton <akpm@linux-foundation.org>, linux-kernel@vger.kernel.org
Subject: Re: [RESEND] file operations: release can race with read/write?
Message-ID: <20071017193320.GK8181@ftp.linux.org.uk>
References: <adave96ki0x.fsf@cisco.com> <adaodexh9uz.fsf@cisco.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <adaodexh9uz.fsf@cisco.com>
User-Agent: Mutt/1.4.1i
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1054
Lines: 31

On Wed, Oct 17, 2007 at 11:30:44AM -0700, Roland Dreier wrote:
> static unsigned long flag;
> 
> static int cw_open(struct inode *inode, struct file *filp)
> {
> 	return 0;
> }
> 
> static int cw_close(struct inode *inode, struct file *filp)
> {
> 	clear_bit(1, &flag);
> 
> 	msleep(1);
> 
> 	if (test_and_set_bit(1, &flag))
> 		printk(KERN_ERR "Write raced with close?\n");

Of _course_ you can trigger that - just open two files and have write
on one of them while another gets closed.

That is certainly possible, but it has nothing whatsoever with the
race you've described - you have two different objects and method
call on one of them might happen when destructor is called on another.

You have protection against ->release() for a struct file called
while you are in ->write() for _same_ struct file.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/