Re: IPv4: sysctl table check failed [was: mmotm 2010-10-07-14-08 uploaded]

[ebiederm@xmission.com (Eric W. Biederman)]

Andrew Morton <akpm@linux-foundation.org> writes:

> On Fri, 08 Oct 2010 00:22:15 +0200
> Eric Dumazet <eric.dumazet@gmail.com> wrote:
>
>> Le vendredi 08 octobre 2010 __ 00:06 +0200, Jiri Slaby a __crit :
>> > On 10/07/2010 11:08 PM, akpm@linux-foundation.org wrote:
>> > > The mm-of-the-moment snapshot 2010-10-07-14-08 has been uploaded to
>> > 
>> > Hi, I got bunch of "sysctl table check failed" below. All seem to be
>> > related to ipv4:
>> 
>> I would say, sysctl check is buggy :(
>> 
>> min/max are optional
>> 
>> [PATCH] sysctl: min/max bounds are optional
>> 
>> sysctl check complains when proc_doulongvec_minmax or
>> proc_doulongvec_ms_jiffies_minmax are used by a vector of longs (with
>> more than one element), with no min or max value specified.
>> 
>> This is unexpected, given we had a bug on this min/max handling :)
>> 
>> Reported-by: Jiri Slaby <jirislaby@gmail.com>
>> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
>> ---
>>  kernel/sysctl_check.c |    9 ---------
>>  1 file changed, 9 deletions(-)
>> 
>> diff --git a/kernel/sysctl_check.c b/kernel/sysctl_check.c
>> index 04cdcf7..10b90d8 100644
>> --- a/kernel/sysctl_check.c
>> +++ b/kernel/sysctl_check.c
>> @@ -143,15 +143,6 @@ int sysctl_check_table(struct nsproxy *namespaces, struct ctl_table *table)
>>  				if (!table->maxlen)
>>  					set_fail(&fail, table, "No maxlen");
>>  			}
>> -			if ((table->proc_handler == proc_doulongvec_minmax) ||
>> -			    (table->proc_handler == proc_doulongvec_ms_jiffies_minmax)) {
>> -				if (table->maxlen > sizeof (unsigned long)) {
>> -					if (!table->extra1)
>> -						set_fail(&fail, table, "No min");
>> -					if (!table->extra2)
>> -						set_fail(&fail, table, "No max");
>> -				}
>> -			}
>>  #ifdef CONFIG_PROC_SYSCTL
>>  			if (table->procname && !table->proc_handler)
>>  				set_fail(&fail, table, "No proc_handler");
>
> That will probably fix it ;)
>
> net-avoid-limits-overflow.patch is dependent on this patch.  Unless
> Eric B squeaks I'll plan on sending this patch in for 2.6.37.

Oh.  I see. I actually had a sanity check for the case that was failing.
I probably spotted the buggy code and wanted to see if there was
anything that cared.

So sysctl_check was perfectly correct until the bug was removed from
proc_doulongvec_minmax.  Which also means we have been auditing the
kernel for quite a while to make certain that it is safe not to
increment min and max.

Eric