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, 17 Oct 2008 01:43:15 +0200
From:	Bodo Eggert <7eggert@....de>
To:	Hugh Dickins <hugh@...itas.com>,
	Peter Zijlstra <peterz@...radead.org>,
	Peter Cordes <peter@...des.ca>, linux-kernel@...r.kernel.org,
	Christoph Hellwig <hch@...radead.org>,
	linux-mm <linux-mm@...ck.org>
Subject: Re: no way to swapoff a deleted swap file?

Hugh Dickins <hugh@...itas.com> wrote:
> On Thu, 16 Oct 2008, Peter Zijlstra wrote:
>> On Wed, 2008-10-15 at 17:21 -0300, Peter Cordes wrote:

>> > I unlinked a swapfile without realizing I was still swapping on it.
>> > Now my /proc/swaps looks like this:
>> > Filename                                Type            Size    Used
>> > Priority
>> > /var/tmp/EXP/cache/swap/1\040(deleted)  file            1288644 1448       -1
>> > /var/tmp/EXP/cache/swap/2\040(deleted)  file            1433368 0  -2

>> >  If kswapd0 had a fd open on the swap files, swapoff /proc/$PID/fd/3
>> > could possibly work.  But it looks like the files are open but with no
>> > user-space accessable file descriptors to them.  Which makes sense,
>> > except for this case.
>> 
>> Right, except that kswapd is per node, so we'd either have to add it to
>> all kswapd instances or a random one. Also, kthreads don't seem to have
>> a files table afaict.
>> 
>> But yes, I see your problem and it makes sense to look for a nice
>> solution.
> 
> No immediate answer springs to my mind.
> 
> It's not something I'd want to add a new system call for.
> I guess we could put a magic file for each swap area
> somewhere down in /sys, and allow swapoff to act upon that.

I think the original idea of something like /proc/$PID/fd/ is not too bad.
I don't know if it's possible to have the same mechanism in sysfs. I guess
not, but with the rest of the vm knobs being in /proc, I would not be too sad.

Maybe it's possible to clone(CLONE_FILES) the kswapds. This would allow to
have /proc/sys/vm/swapfiles point to one of the correct /proc/$kwapd/fd/.

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ