svn commit: r278323 - in head: etc/rc.d usr.sbin/jail

[James Gritton]

On 2015-02-06 22:31, Garrett Cooper wrote:
> On Feb 6, 2015, at 21:27, James Gritton <jamie at freebsd.org> wrote:
> 
>> On 2015-02-06 22:18, Garrett Cooper wrote:
>>> On Feb 6, 2015, at 9:54, Jamie Gritton <jamie at freebsd.org> wrote:
>>>> Modified: head/usr.sbin/jail/command.c
>>>> ==============================================================================
>>>> --- head/usr.sbin/jail/command.c	Fri Feb  6 17:43:13 2015	(r278322)
>>>> +++ head/usr.sbin/jail/command.c	Fri Feb  6 17:54:53 2015	(r278323)
>>>> @@ -112,6 +112,12 @@ next_command(struct cfjail *j)
>>>> 				if (!bool_param(j->intparams[IP_MOUNT_FDESCFS]))
>>>> 					continue;
>>>> 				j->comstring = &dummystring;
>>>> +				break;
>>>> +			case IP_MOUNT_PROCFS:
>>>> +				if (!bool_param(j->intparams[IP_MOUNT_PROCFS]))
>>>> +					continue;
>>>> +				j->comstring = &dummystring;
>>>> +				break;
>>> Did you intend on adding another break? The code would previously 
>>> fall
>>> through to the next case statement...
>>>> 			case IP__OP:
>>>> 			case IP_STOP_TIMEOUT:
>>>> 				j->comstring = &dummystring;
>> 
>> Yes.  The code did indeed previously fall to the next case, but it was 
>> a no-op: the next case only had the same exact assignment that had 
>> just taken place (j->comstring = &dummystring).  The lack of a break 
>> that had existed before was just some sloppy coding that I didn't 
>> notice at the time because it didn't actually change any behavior.  
>> Nonetheless it seemed worth correcting when I noticed it.
> 
> True. I looked at the code afterwards and it looks ok. mount.procfs
> doesn’t exist in my environment. Is that command correct?
> 
> $ which mount.procfs; echo $?
> 1

I added mount.procfs as a jail parameter, but it's not a command.  Just 
like the existing mount.devfs and mount.fdescfs aren't commands either.  
The reason these jail parameters exist is to ease the backward 
compatibility with the old rc-based jail system.  It should be a simple 
case of doing for procfs exactly what I did for the other two, but 
apparently it isn't.  It's likely related to something I'm missing in 
the proper way of modifying rc scripts.

- Jamie