linux-kernel - Re: [linux-pm] Freezer: Don't count threads waiting for frozen filesystems.

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

Message-Id: <1225238069.9661.61.camel@nigel-laptop>
Date:	Wed, 29 Oct 2008 10:54:29 +1100
From:	Nigel Cunningham <ncunningham@...a.org.au>
To:	Miklos Szeredi <miklos@...redi.hu>
Cc:	linux-pm@...ts.linux-foundation.org, linux-kernel@...r.kernel.org
Subject: Re: [linux-pm] Freezer: Don't count threads waiting for
	frozen	filesystems.

Hi Miklos.

On Wed, 2008-10-29 at 00:45 +0100, Miklos Szeredi wrote:
> On Wed, 29 Oct 2008, Nigel Cunningham wrote:
> > Hi.
> > 
> > On Wed, 2008-10-29 at 00:24 +0100, Miklos Szeredi wrote:
> > > On Wed, 29 Oct 2008, Nigel Cunningham wrote:
> > > > Remember, though, that we're only freezing fuse at the moment, and
> > > > strictly one filesystem at a time. We can thus happily wait for the
> > > > i_mutex taken by some other process to be released.
> > > 
> > > Not going to work: you need to wait for all requests to be finished,
> > > but those might depend on some other fuse filesystem which has already
> > > been frozen.
> > 
> > Okay. In that case, am I right in thinking that the request waiting on
> > the frozen filesystem will be stuck in request_wait_answer,
> 
> Yes.
> 
> > and the
> > userspace process that was trying to satisfy the request will be stuck
> > in the FUSE_MIGHT_FREEZE call that was invoked for the frozen
> > filesystem?
> 
> No, it already passed that, before the filesystem got frozen.  But it
> doesn't matter, in either case i_mutex will already have been taken by
> the VFS and it won't be released until the request completely
> finishes.

I think we're making different assumptions. I'm assuming that one of
those solutions we already discussed is implemented, such that we don't
start freezing a new filesystem until all the requests for the current
filesystem are dealt with. Perhaps I should come up with a new version
of the patch that implements that.

Nigel

--
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/