linux-kernel - Re: [PATCH 1/3] dcache: add a new enum type for 'dentry_d_lock

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]

Date:   Fri, 15 Nov 2019 14:17:54 +0000
From:   Al Viro <viro@...iv.linux.org.uk>
To:     Steven Rostedt <rostedt@...dmis.org>
Cc:     Greg KH <gregkh@...uxfoundation.org>, yu kuai <yukuai3@...wei.com>,
        rafael@...nel.org, oleg@...hat.com, mchehab+samsung@...nel.org,
        corbet@....net, tytso@....edu, jmorris@...ei.org,
        linux-kernel@...r.kernel.org, linux-fsdevel@...r.kernel.org,
        zhengbin13@...wei.com, yi.zhang@...wei.com,
        chenxiang66@...ilicon.com, xiexiuqi@...wei.com
Subject: Re: [PATCH 1/3] dcache: add a new enum type for 'dentry_d_lock_class'

On Fri, Nov 15, 2019 at 08:58:05AM -0500, Steven Rostedt wrote:
> On Fri, 15 Nov 2019 13:48:23 +0000
> Al Viro <viro@...iv.linux.org.uk> wrote:
> 
> > > BTW, what do you mean by "can debugfs_remove_recursive() rely upon the
> > > lack of attempts to create new entries inside the subtree it's trying
> > > to kill?"  
> > 
> > Is it possible for something to call e.g. debugfs_create_dir() (or any
> > similar primitive) with parent inside the subtree that has been
> > passed to debugfs_remove_recursive() call that is still in progress?
> > 
> > If debugfs needs to cope with that, debugfs_remove_recursive() needs
> > considerably heavier locking, to start with.
> 
> I don't know about debugfs, but at least tracefs (which cut and pasted
> from debugfs) does not allow that. At least in theory it doesn't allow
> that (and if it does, it's a bug in the locking at the higher levels).
> 
> And perhaps debugfs shouldn't allow that either. As it is only suppose
> to be a light weight way to interact with the kernel, hence the name
> "debugfs".
> 
> Yu, do you have a test case for the "infinite loop" case?

Infinite loop, AFAICS, is reasonably easy to trigger - just open
a non-empty subdirectory and lseek to e.g. next-to-last element
in it.  Again, list_empty() use in there is quite wrong - it can
give false negatives just on the cursors.  No arguments about
that part...