linux-kernel - Re: [PATCH 1/2] cgroup_pids: track highwater mark of pids

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] [day] [month] [year] [list]

Message-ID: <20160719190708.GO3078@mtj.duckdns.org>
Date:	Tue, 19 Jul 2016 15:07:08 -0400
From:	Tejun Heo <tj@...nel.org>
To:	Topi Miettinen <toiwoton@...il.com>
Cc:	linux-kernel@...r.kernel.org, Li Zefan <lizefan@...wei.com>,
	Johannes Weiner <hannes@...xchg.org>,
	"open list:CONTROL GROUP (CGROUP)" <cgroups@...r.kernel.org>
Subject: Re: [PATCH 1/2] cgroup_pids: track highwater mark of pids

Hello,

On Sun, Jul 17, 2016 at 11:03:38PM +0300, Topi Miettinen wrote:
> +static void pids_update_highwater_mark(struct pids_cgroup *p)
> +{
> +	while (1) {
> +		int64_t old_mark, new_mark, cur_mark;
> +
> +		old_mark = atomic64_read(&p->highwater_mark);
> +		new_mark = atomic64_read(&p->counter);
> +		if (old_mark >= new_mark)
> +			return;
> +		cur_mark = atomic64_cmpxchg(&p->highwater_mark, old_mark,
> +					    new_mark);
> +
> +		/* It's OK if the counter was decreased meanwhile */
> +		if (cur_mark == old_mark &&
> +		    atomic64_read(&p->counter) <= new_mark)
> +			return;
> +	}
> +}

I think it'd be better to make this part of pids_charge() - maybe use
atomic64_add_return() to get the current value and track the maximum?

> @@ -300,6 +341,11 @@ static struct cftype pids_files[] = {
>  		.read_s64 = pids_current_read,
>  		.flags = CFTYPE_NOT_ON_ROOT,
>  	},
> +	{
> +		.name = "highwater_mark",
> +		.read_s64 = pids_highwater_mark_read,
> +		.flags = CFTYPE_NOT_ON_ROOT,
> +	},

This should be an entry in the pids.stats file.  Please also update
the documentation accordingly.

Thanks.

-- 
tejun