linux-kernel - Re: [PATCH v1 2/2] sysctl: handle overflow for file-max

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: <20181016152100.zg2pmyk6puc5qsr5@brauner.io>
Date:   Tue, 16 Oct 2018 17:21:01 +0200
From:   Christian Brauner <christian@...uner.io>
To:     Waiman Long <longman@...hat.com>
Cc:     keescook@...omium.org, linux-kernel@...r.kernel.org,
        ebiederm@...ssion.com, mcgrof@...nel.org,
        akpm@...ux-foundation.org, joe.lawrence@...hat.com,
        linux@...inikbrodowski.net, viro@...iv.linux.org.uk
Subject: Re: [PATCH v1 2/2] sysctl: handle overflow for file-max

On Tue, Oct 16, 2018 at 11:13:28AM -0400, Waiman Long wrote:
> On 10/15/2018 06:55 AM, Christian Brauner wrote:
> > Currently, when writing
> >
> > echo 18446744073709551616 > /proc/sys/fs/file-max
> >
> > /proc/sys/fs/file-max will overflow and be set to 0. That quickly
> > crashes the system.
> > This commit explicitly caps the value for file-max to ULONG_MAX.
> >
> > Note, this isn't technically necessary since proc_get_long() will already
> > return ULONG_MAX. However, two reason why we still should do this:
> > 1. it makes it explicit what the upper bound of file-max is instead of
> >    making readers of the code infer it from proc_get_long() themselves
> > 2. other tunebles than file-max may want to set a lower max value than
> >    ULONG_MAX and we need to enable __do_proc_doulongvec_minmax() to handle
> >    such cases too
> >
> > Cc: Kees Cook <keescook@...omium.org>
> > Signed-off-by: Christian Brauner <christian@...uner.io>
> > ---
> > v0->v1:
> > - if max value is < than ULONG_MAX use max as upper bound
> > - (Dominik) remove double "the" from commit message
> > ---
> >  kernel/sysctl.c | 4 ++++
> >  1 file changed, 4 insertions(+)
> >
> > diff --git a/kernel/sysctl.c b/kernel/sysctl.c
> > index 97551eb42946..226d4eaf4b0e 100644
> > --- a/kernel/sysctl.c
> > +++ b/kernel/sysctl.c
> > @@ -127,6 +127,7 @@ static int __maybe_unused one = 1;
> >  static int __maybe_unused two = 2;
> >  static int __maybe_unused four = 4;
> >  static unsigned long one_ul = 1;
> > +static unsigned long ulong_max = ULONG_MAX;
> >  static int one_hundred = 100;
> >  static int one_thousand = 1000;
> >  #ifdef CONFIG_PRINTK
> > @@ -1696,6 +1697,7 @@ static struct ctl_table fs_table[] = {
> >  		.maxlen		= sizeof(files_stat.max_files),
> >  		.mode		= 0644,
> >  		.proc_handler	= proc_doulongvec_minmax,
> > +		.extra2		= &ulong_max,
> 
> What is the point of having a maximum value of ULONG_MAX anyway? No
> value you can put into a ulong type can be bigger than that.

This is changed in the new code to LONG_MAX. See the full thread for
context. There's also an additional explantion in the commit message.

> 
> >  	},
> >  	{
> >  		.procname	= "nr_open",
> > @@ -2795,6 +2797,8 @@ static int __do_proc_doulongvec_minmax(void *data, struct ctl_table *table, int
> >  				break;
> >  			if (neg)
> >  				continue;
> > +			if (max && val > *max)
> > +				val = *max;
> >  			val = convmul * val / convdiv;
> >  			if ((min && val < *min) || (max && val > *max))
> >  				continue;
> 
> This does introduce a change in behavior. Previously the out-of-bound
> value is ignored, now it is capped at its maximum. This is a
> user-visible change.

Not completely true though. Try

echo 18446744073709551616 > /proc/sys/fs/file-max

on a system you find acceptable loosing.
So this is an acceptable user-visible change I'd say. But I'm open to
other suggestions.

> 
> Cheers,
> Longman
> 
>