mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Richard Guy Briggs <rgb@redhat.com>
To: Oleg Nesterov <oleg@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>,
	linux-audit@redhat.com, linux-kernel@vger.kernel.org,
	Eric Paris <eparis@redhat.com>,
	"Eric W. Biederman" <ebiederm@xmission.com>,
	"Serge E. Hallyn" <serge@hallyn.com>,
	Ingo Molnar <mingo@redhat.com>,
	Andrew Morton <akpm@linux-foundation.org>
Subject: Re: [PATCH 11/12] pid: rewrite task helper functions avoiding task->pid and task->tgid
Date: Mon, 26 Aug 2013 23:04:48 -0400	[thread overview]
Message-ID: <20130827030448.GE21110@madcap2.tricolour.ca> (raw)
In-Reply-To: <20130823192807.GA6504@redhat.com>

On Fri, Aug 23, 2013 at 09:28:07PM +0200, Oleg Nesterov wrote:
> On 08/22, Richard Guy Briggs wrote:
> >
> > On Thu, Aug 22, 2013 at 10:05:55PM +0200, Peter Zijlstra wrote:
> > >
> > > Why would you ever want to do this? It just makes these tests more
> > > expensive for no gain what so ff'ing ever.
> >
> > Backups are generally considered a good idea, but in this case, I'd
> > quote:
> 
> And perhaps you are right. At least we can probably kill task->tgid.
> And I agree, it would be nice to kill task->pid.
> 
> But: I also agree with Peter, we should try to not make the current
> code more expensive.

I don't disagree.  I was given some hope by Eric Biederman that it might
not cause cache-line misses...

> Anyway. Imho, you should not mix the different things in one series.
> If you want to fix audit, do not add the patches like 10/12 or 11/12
> into this series.

They were included to gain reassurance that PIDs reported in audit logs
were accurate.  If those assurances can be made, then I can limit my
work to audit itself.

> Or 3/12.

That is a cleanup to make clear what parts are actually pid-related and
what isn't.

> OK, I agree sys_getppid() in audit_log_task_info() looks
> strange at least. Just fix it using the helpers we already have and
> add the new helpers later. Or send the patch(es) which adds the new
> helpers first.

Patch 04/12 is that helper.  It is used in only two places in audit (and
once in apparmor), so I could have duplicated the code, but since it
needs rcu locking, an inline funciton in the audit namespace seemed
somewhat pointless when it could just as easily be shared with other
subsystems.

> Or task_pid_nr_init_ns()... For what? We already have task_pid_nr().
> Use the helper we already have, or introduce the new one first and
> change the current users of task_pid_nr().

If task_struct::pid is definitely not going away, then that whole part
is moot and we'll just use task_pid_nr() as is.

> In short. Fortunately you do not need to convince me, I do not
> maintain audit or namespaces ;) But imho this series looks a bit
> confusing.

It is incomplete.  The last step(s) were intended to purge
task_struct::pid (or abstract it using task_pid_nr()), which haven't
been submitted yet.  The whole point of the patchset was to give
confidence in the PIDs reported in any audit logs.

> Oleg.

- RGB

--
Richard Guy Briggs <rbriggs@redhat.com>
Senior Software Engineer
Kernel Security
AMER ENG Base Operating Systems
Remote, Ottawa, Canada
Voice: +1.647.777.2635
Internal: (81) 32635
Alt: +1.613.693.0684x3545

  reply	other threads:[~2013-08-27  3:05 UTC|newest]

Thread overview: 107+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2013-08-20 21:31 [PATCH 00/12] RFC: steps to make audit pid namespace-safe Richard Guy Briggs
2013-08-20 21:31 ` [PATCH 01/12] audit: Kill the unused struct audit_aux_data_capset Richard Guy Briggs
2013-08-20 21:31 ` [PATCH 02/12] audit: fix netlink portid naming and types Richard Guy Briggs
2013-08-20 21:31 ` [PATCH 03/12] pid: get ppid pid_t of task in init_pid_ns safely Richard Guy Briggs
2013-08-27 17:21   ` Oleg Nesterov
2013-08-30 19:56     ` Richard Guy Briggs
2013-08-30 20:37       ` John Johansen
2013-08-30 22:41         ` [PATCH 1/3] apparmor: fix capability to not use the current task, during reporting John Johansen
2013-08-30 22:42         ` [PATCH 2/3] apparmor: remove tsk field from the apparmor_audit_struct John Johansen
2013-08-30 22:43         ` [PATCH 03/3] apparmor: remove parent task info from audit logging John Johansen
2013-09-03 18:31         ` [PATCH 03/12] pid: get ppid pid_t of task in init_pid_ns safely Richard Guy Briggs
2013-12-11 14:47           ` Richard Guy Briggs
2013-12-11 16:44             ` John Johansen
2013-12-11 17:19               ` Richard Guy Briggs
2013-08-20 21:31 ` [PATCH 04/12] audit: convert PPIDs to the inital PID namespace Richard Guy Briggs
2013-08-20 21:31 ` [PATCH 05/12] pid: get pid_t of task in init_pid_ns correctly Richard Guy Briggs
2013-08-20 21:31 ` [PATCH 06/12] audit: Simplify and correct audit_log_capset Richard Guy Briggs
2013-08-20 21:31 ` [PATCH 07/12] audit: store audit_pid as a struct pid pointer Richard Guy Briggs
2013-08-20 21:32 ` [PATCH 08/12] audit: anchor all pid references in the initial pid namespace Richard Guy Briggs
2013-08-20 21:32 ` [PATCH 09/12] pid: modify task_pid_nr to work without task->pid Richard Guy Briggs
2013-12-16 21:03   ` [PATCH] pid: change task_struct::pid to read-only Richard Guy Briggs
2013-12-17  9:58     ` Peter Zijlstra
2013-12-20  4:48       ` Richard Guy Briggs
2013-12-20  8:58         ` Peter Zijlstra
2013-12-20 14:04           ` Richard Guy Briggs
2014-01-23 19:32           ` [PATCH 0/7][RFC] pid: changes to support audit Richard Guy Briggs
2014-01-23 19:32             ` [PATCH 1/7] pid: change task_struct::pid to read-only Richard Guy Briggs
2014-01-23 19:32             ` [PATCH 2/7] compiler: CONST_CAST makes writing const vars easier and obvious Richard Guy Briggs
2014-01-23 19:32             ` [PATCH 3/7] pid: use the CONST_CAST macro instead to write to const task_struct::pid Richard Guy Briggs
2014-01-23 19:32             ` [PATCH 4/7] pid: modify task_tgid_nr to work without task->tgid Richard Guy Briggs
2014-02-20 18:35               ` Oleg Nesterov
2014-02-21 20:47                 ` Richard Guy Briggs
2014-02-24 18:40                   ` Oleg Nesterov
2014-01-23 19:32             ` [PATCH 5/7] pid: rewrite task helper function is_global_init() avoiding task->pid Richard Guy Briggs
2014-02-20 18:39               ` Oleg Nesterov
2014-02-21 16:10                 ` Richard Guy Briggs
2014-01-23 19:32             ` [PATCH 6/7] pid: mark struct task const in helper functions Richard Guy Briggs
2014-01-23 19:32             ` [PATCH 7/7] pid: get pid_t ppid of task in init_pid_ns Richard Guy Briggs
2014-02-20 19:01               ` Oleg Nesterov
2014-02-21 18:10                 ` Richard Guy Briggs
2014-02-24 18:32                   ` Oleg Nesterov
2014-03-17 20:14               ` Tony Luck
2014-03-17 20:15                 ` Eric Paris
2014-01-23 21:25             ` [PATCH 0/7][RFC] pid: changes to support audit Peter Zijlstra
2014-01-24  6:14               ` Richard Guy Briggs
2014-01-24  8:52                 ` Peter Zijlstra
2014-01-24 14:31                   ` Richard Guy Briggs
2014-02-19 16:18             ` Richard Guy Briggs
2014-02-19 17:47               ` Oleg Nesterov
2014-02-19 18:15                 ` Richard Guy Briggs
2014-02-20 19:08                   ` Oleg Nesterov
2013-12-17  9:59     ` [PATCH] pid: change task_struct::pid to read-only Peter Zijlstra
2013-12-17 15:36     ` Oleg Nesterov
2013-12-17 15:40       ` Oleg Nesterov
2013-12-20 19:01         ` Oleg Nesterov
2013-12-20 20:19           ` Richard Guy Briggs
2013-12-20 21:33           ` Peter Zijlstra
2013-12-22 16:03             ` Oleg Nesterov
2014-01-23 19:24               ` Richard Guy Briggs
2013-08-20 21:32 ` [PATCH 10/12] pid: modify task_tgid_nr to work without task->tgid Richard Guy Briggs
2013-08-20 21:32 ` [PATCH 11/12] pid: rewrite task helper functions avoiding task->pid and task->tgid Richard Guy Briggs
2013-08-22 19:08   ` Oleg Nesterov
2013-08-26 22:07     ` Richard Guy Briggs
2013-08-27 16:15       ` Oleg Nesterov
2013-12-16 17:35       ` Richard Guy Briggs
2013-12-16 21:05         ` Oleg Nesterov
2013-12-16 22:20           ` Richard Guy Briggs
2013-12-17  9:34             ` Peter Zijlstra
2013-12-17  9:48               ` Peter Zijlstra
2013-12-20  4:54               ` Richard Guy Briggs
2013-08-22 20:05   ` Peter Zijlstra
2013-08-22 21:43     ` Richard Guy Briggs
2013-08-23  6:36       ` Peter Zijlstra
2013-08-27  2:37         ` Richard Guy Briggs
2013-08-27 12:11           ` Peter Zijlstra
2013-08-27 21:35             ` Eric W. Biederman
2013-08-28  8:16               ` Peter Zijlstra
2013-08-23 19:28       ` Oleg Nesterov
2013-08-27  3:04         ` Richard Guy Briggs [this message]
2013-08-27 17:11           ` Oleg Nesterov
2013-08-30 19:06             ` audit looks unmaintained? [was: Re: [PATCH 11/12] pid: rewrite task helper functions avoiding task->pid and task->tgid] Richard Guy Briggs
2013-08-30 19:54               ` Steve Grubb
2013-09-08 15:54                 ` Oleg Nesterov
2013-09-10 17:20                   ` Oleg Nesterov
2013-09-13 18:42                     ` Steve Grubb
2013-09-14 18:10                       ` Oleg Nesterov
2013-09-13 18:28                   ` Steve Grubb
2013-09-14 18:08                     ` Oleg Nesterov
2013-08-20 21:32 ` [PATCH 12/12] pid: mark struct task const in helper functions Richard Guy Briggs
2013-12-23 22:27 ` [PATCH 0/5][RFC][v2] steps to make audit pid namespace-safe Richard Guy Briggs
2013-12-23 22:27   ` [PATCH 1/5] pid: get pid_t ppid of task in init_pid_ns Richard Guy Briggs
2013-12-30 17:04     ` Oleg Nesterov
2013-12-23 22:27   ` [PATCH 2/5] audit: convert PPIDs to the inital PID namespace Richard Guy Briggs
2013-12-30 17:07     ` Oleg Nesterov
2013-12-23 22:27   ` [PATCH 3/5] audit: store audit_pid as a struct pid pointer Richard Guy Briggs
2013-12-30 17:51     ` Oleg Nesterov
2014-01-21 23:37       ` Richard Guy Briggs
2013-12-23 22:27   ` [PATCH 4/5] audit: anchor all pid references in the initial pid namespace Richard Guy Briggs
2013-12-30 18:06     ` Oleg Nesterov
2014-02-19 20:28       ` Richard Guy Briggs
2013-12-23 22:27   ` [PATCH 5/5] audit: allow user processes to log from another PID namespace Richard Guy Briggs
2014-02-19 20:57   ` [PATCH 0/5][RFC][v3] steps to make audit pid namespace-safe Richard Guy Briggs
2014-02-19 20:57     ` [PATCH 1/5] pid: get pid_t ppid of task in init_pid_ns Richard Guy Briggs
2014-02-19 20:57     ` [PATCH 2/5] audit: convert PPIDs to the inital PID namespace Richard Guy Briggs
2014-02-19 20:57     ` [PATCH 3/5] audit: store audit_pid as a struct pid pointer Richard Guy Briggs
2014-02-19 20:57     ` [PATCH 4/5] audit: anchor all pid references in the initial pid namespace Richard Guy Briggs
2014-02-19 20:57     ` [PATCH 5/5] audit: allow user processes to log from another PID namespace Richard Guy Briggs

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20130827030448.GE21110@madcap2.tricolour.ca \
    --to=rgb@redhat.com \
    --cc=akpm@linux-foundation.org \
    --cc=ebiederm@xmission.com \
    --cc=eparis@redhat.com \
    --cc=linux-audit@redhat.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@redhat.com \
    --cc=oleg@redhat.com \
    --cc=peterz@infradead.org \
    --cc=serge@hallyn.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®