Re: [PATCH][RESEND 3] disassociate_ctty SMP fix

[Russell King <rmk@arm.linux.org.uk>]

On Tue, Feb 04, 2003 at 11:58:52PM -0200, Rik van Riel wrote:
> I guess it's time to fix the caller of this function then,
> since something strange is going on here:
> 
> http://bugme.osdl.org/show_bug.cgi?id=54

I don't think the tty is null there.  It'll be a filp->private_data
being null.

>From the trace, my guess is that a file descriptor is being left
with a null filp->private_data, yet its on the file descriptor list
for the tty, and its fops is set to tty_fops.

I'll see if I can reproduce it here and work out what's going on.
However, note the following (and I think this is the crux of the problem):
release_dev is _supposed_ to run under the BKL.  So how the fsck are we
getting into tty_do_hangup's BKL protected region while also being in
release_dev's BKL protected region?  Is it the BKL which has broken?

-- 
Russell King (rmk@arm.linux.org.uk)                The developer of ARM Linux
             http://www.arm.linux.org.uk/personal/aboutme.html