linux-kernel - Re: [PATCH v4 0/2] proc: Ensure we see the exit of each process tid exactly

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: <875zdj8mq4.fsf@x220.int.ebiederm.org>
Date:   Tue, 28 Apr 2020 13:54:43 -0500
From:   ebiederm@...ssion.com (Eric W. Biederman)
To:     Oleg Nesterov <oleg@...hat.com>
Cc:     Linus Torvalds <torvalds@...ux-foundation.org>,
        LKML <linux-kernel@...r.kernel.org>,
        Linux FS Devel <linux-fsdevel@...r.kernel.org>,
        Alexey Dobriyan <adobriyan@...il.com>,
        Alexey Gladkov <legion@...nel.org>,
        Andrew Morton <akpm@...ux-foundation.org>,
        Alexey Gladkov <gladkov.alexey@...il.com>,
        Thomas Gleixner <tglx@...utronix.de>,
        "Paul E. McKenney" <paulmck@...nel.org>
Subject: Re: [PATCH v4 0/2] proc: Ensure we see the exit of each process tid exactly

Oleg Nesterov <oleg@...hat.com> writes:

> On 04/28, Eric W. Biederman wrote:
>>
>> In short I don't think this change will introduce any regressions.
>>
>> Eric W. Biederman (2):
>>       rculist: Add hlists_swap_heads_rcu
>>       proc: Ensure we see the exit of each process tid exactly once
>
> Eric, sorry, I got lost.
>
> Both changes look good to me, feel free to add my ack, but I presume
> this is on top of next_tgid/lookup_task changes ? If yes, why did not
> you remove has_group_leader_pid?

On top of next_tgid.

Upon a close examination there are not any current bugs in
posix-cpu-timers nor is there anything that exchange_tids
will make worse.

I am preparing a follow on patchset to kill has_group_leader_pid.

I am preparing a further follow on patchset to see if I can get that
code to start returning pids, because that is cheaper and clearer.

I pushed those changes a little farther out so I could maintain focus on
what I am accomplishing.

Adding exchange_tids was difficult because I had to audit pretty much
all of the pid use in the kernel to see if the small change in behavior
would make anything worse.   The rest of the changes should be simpler
and more localized so I hope they go faster.

Eric