Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Date:   Fri, 8 Jan 2021 12:47:18 +0100
From:   Michal Hocko <mhocko@suse.com>
To:     Xiaoming Ni <nixiaoming@huawei.com>
Cc:     linux-kernel@vger.kernel.org, mcgrof@kernel.org,
        keescook@chromium.org, yzaikin@google.com, adobriyan@gmail.com,
        linux-fsdevel@vger.kernel.org, vbabka@suse.cz,
        akpm@linux-foundation.org, wangle6@huawei.com
Subject: Re: [PATCH v2] proc_sysctl: fix oops caused by incorrect command
 parameters.
Message-ID: <20210108114718.GA13207@dhcp22.suse.cz>
References: <20210108023339.55917-1-nixiaoming@huawei.com>
 <20210108092145.GX13207@dhcp22.suse.cz>
 <829bbba0-d3bb-a114-af81-df7390082958@huawei.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <829bbba0-d3bb-a114-af81-df7390082958@huawei.com>
Precedence: bulk

On Fri 08-01-21 18:01:52, Xiaoming Ni wrote:
> On 2021/1/8 17:21, Michal Hocko wrote:
> > On Fri 08-01-21 10:33:39, Xiaoming Ni wrote:
> > > The process_sysctl_arg() does not check whether val is empty before
> > >   invoking strlen(val). If the command line parameter () is incorrectly
> > >   configured and val is empty, oops is triggered.
> > > 
> > > For example, "hung_task_panic=1" is incorrectly written as "hung_task_panic".
> > > 
> > > log:
> > > 	Kernel command line: .... hung_task_panic
> > > 	....
> > > 	[000000000000000n] user address but active_mm is swapper
> > > 	Internal error: Oops: 96000005 [#1] SMP
> > > 	Modules linked in:
> > > 	CPU: 0 PID: 1 Comm: swapper/0 Not tainted 5.10.1 #1
> > > 	Hardware name: linux,dummy-virt (DT)
> > > 	pstate: 40000005 (nZcv daif -PAN -UAO -TCO BTYPE=--)
> > > 	pc : __pi_strlen+0x10/0x98
> > > 	lr : process_sysctl_arg+0x1e4/0x2ac
> > > 	sp : ffffffc01104bd40
> > > 	x29: ffffffc01104bd40 x28: 0000000000000000
> > > 	x27: ffffff80c0a4691e x26: ffffffc0102a7c8c
> > > 	x25: 0000000000000000 x24: ffffffc01104be80
> > > 	x23: ffffff80c22f0b00 x22: ffffff80c02e28c0
> > > 	x21: ffffffc0109f9000 x20: 0000000000000000
> > > 	x19: ffffffc0107c08de x18: 0000000000000003
> > > 	x17: ffffffc01105d000 x16: 0000000000000054
> > > 	x15: ffffffffffffffff x14: 3030253078413830
> > > 	x13: 000000000000ffff x12: 0000000000000000
> > > 	x11: 0101010101010101 x10: 0000000000000005
> > > 	x9 : 0000000000000003 x8 : ffffff80c0980c08
> > > 	x7 : 0000000000000000 x6 : 0000000000000002
> > > 	x5 : ffffff80c0235000 x4 : ffffff810f7c7ee0
> > > 	x3 : 000000000000043a x2 : 00bdcc4ebacf1a54
> > > 	x1 : 0000000000000000 x0 : 0000000000000000
> > > 	Call trace:
> > > 	 __pi_strlen+0x10/0x98
> > > 	 parse_args+0x278/0x344
> > > 	 do_sysctl_args+0x8c/0xfc
> > > 	 kernel_init+0x5c/0xf4
> > > 	 ret_from_fork+0x10/0x30
> > > 	Code: b200c3eb 927cec01 f2400c07 54000301 (a8c10c22)
> > > 
> > > Fixes: 3db978d480e2843 ("kernel/sysctl: support setting sysctl parameters
> > >   from kernel command line")
> > > Signed-off-by: Xiaoming Ni <nixiaoming@huawei.com>
> > 
> > Thanks for catching this!
> > 
> > > ---------
> > > v2:
> > >     Added log output of the failure branch based on the review comments of Kees Cook.
> > > v1: https://lore.kernel.org/lkml/20201224074256.117413-1-nixiaoming@huawei.com/
> > > ---------
> > > ---
> > >   fs/proc/proc_sysctl.c | 5 +++++
> > >   1 file changed, 5 insertions(+)
> > > 
> > > diff --git a/fs/proc/proc_sysctl.c b/fs/proc/proc_sysctl.c
> > > index 317899222d7f..dc1a56515e86 100644
> > > --- a/fs/proc/proc_sysctl.c
> > > +++ b/fs/proc/proc_sysctl.c
> > > @@ -1757,6 +1757,11 @@ static int process_sysctl_arg(char *param, char *val,
> > >   	loff_t pos = 0;
> > >   	ssize_t wret;
> > > +	if (!val) {
> > > +		pr_err("Missing param value! Expected '%s=...value...'\n", param);
> > > +		return 0;
> I may need to move the validation code for val to the end of the validation
> code for param to prevent non-sysctl arguments from triggering the current
> print.

Why would that matter? A missing value is clearly a error path and it
should be reported.

> Or delete the print and keep it silent for a little better performance.
> Which is better?

I do not think there is a performance argument on the table. The generic
code is returning EINVAL on a missing value where it is needed. Sysctl
all require a value IIRC so EINVAL would be the right way to report
this and let the generic code to complain.

-- 
Michal Hocko
SUSE Labs