linux-kernel - Re: [PATCH 0/3 RESEND] namei: add follow_up

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: <20180405174455.GA27462@gmail.com>
Date:   Thu, 5 Apr 2018 19:45:15 +0200
From:   Christian Brauner <christian.brauner@...onical.com>
To:     Linus Torvalds <torvalds@...ux-foundation.org>
Cc:     Al Viro <viro@...iv.linux.org.uk>,
        "Eric W. Biederman" <ebiederm@...ssion.com>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH 0/3 RESEND] namei: add follow_up_bind()

On Thu, Apr 05, 2018 at 09:28:56AM -0700, Linus Torvalds wrote:
> On Thu, Apr 5, 2018 at 3:51 AM, Christian Brauner
> <christian.brauner@...ntu.com> wrote:
> >
> > This series adds:
> > - follow_up_bind() to namei.{c,h}
> > - switches fs/nfsd/vfs.c:follow_to_parent() to use follow_up_bind()
> > - switches fs/devpts/inode.c:devpts_mntget() to use follow_up_bind()
> 
> Hmm. Seems fair enough to me, although I wonder how much this really
> helps. It does get rid of a duplicate code pattern, but:
> 
>  4 files changed, 14 insertions(+), 5 deletions(-)
> 
> and while some of that is just the new comment, some of it is just "overhead".

Fwiw, it does get read of these while loops in two places but I
personally see the biggest value in making it obvious what bind-mount
resolution means.

> 
> It's also a bit odd how the new helper is marked "inline", but nobody
> will inline it because it's not actually in the header file or any of
> the isers in the same C file. So instead, it has to be exported. I
> wonder if it should just be a trivial inline in <linux/namei.h>? Maybe
> it originally was, and that's where the inline came from, and then
> Christian decided to make it be by the regular "follow_up()" instead?

I head it inline first but it would have required to forward declare
struct vfsmount in the head and I wasn't sure if that was going to fly.
But I explicitly left the inline in there because I was following
user_path_create() ([1], [2]) which does the same. But if that's an
issue I can make it static inline in the header like I had, forward
declare struct vfsmount and remove the unnecessary inline from
user_path_create() in a separate patch unless there's a specific reason
to leave it in there.

[1]: https://elixir.bootlin.com/linux/latest/source/include/linux/namei.h#L79
[2]: https://elixir.bootlin.com/linux/latest/source/fs/namei.c#L3680

> 
> But with all that said, I certainly don't *mind* the patch series.

Cool.

Thanks!
Christian

> 
> Al, I'm leaving this up to you, and expect to get it from your vfs
> tree eventually. Or not.
> 
>                       Linus