linux-kernel - Re: [PATCH 1/1][v2] Add reboot_pid

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: <20111204154543.GA23805@redhat.com>
Date:	Sun, 4 Dec 2011 16:45:43 +0100
From:	Oleg Nesterov <oleg@...hat.com>
To:	Daniel Lezcano <daniel.lezcano@...e.fr>
Cc:	akpm@...ux-foundation.org, serge.hallyn@...onical.com,
	containers@...ts.linux-foundation.org, gkurz@...ibm.com,
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH 1/1][v2] Add reboot_pid_ns to handle the reboot syscall

On 12/04, Daniel Lezcano wrote:
>
> @@ -32,6 +32,8 @@ struct pid_namespace {
>  #endif
>  	gid_t pid_gid;
>  	int hide_pid;
> +	int reboot;
> +	spinlock_t reboot_lock;
>  };

Well. I was thinking about the serialization too, but this
->reboot_lock asks for v3 imho ;)

First of all, do we really care? force_sig(SIGKILL, child_reaper)
can't race with itself, it does nothing if init is already killed.

So why it is important to protect pid_ns->reboot? Yes, it is possible
to change it again if two callers do sys_reboot() "at the same time".
But in this case we can't predict which caller wins anyway, so why
should we worry?

IOW. Say, we have the 2 tasks doing HALT and RESTART in parallel.
It is possible that HALT sets ->reboot and kills init first, then
RESTART changes ->reboot and the second force_sig() does nothing.
In this case we can simply pretend that RESTART wins and the dying
init kills HALT before it calls sys_reboot().

In any case. Even if you want to serialize, instead of adding the
new lock reboot_pid_ns() can simply do:

	if (cmpxchg(&pid_ns->reboot, 0, reboot) != 0)
		return -EBUSY;

this looks much simpler to me.

Oleg.

--
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/