[Bug 196474] jls causes kernel panic

bugzilla-noreply at freebsd.org bugzilla-noreply at freebsd.org
Mon Jan 12 22:24:25 UTC 2015


https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=196474

Jamie Gritton <jamie at FreeBSD.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Assignee|freebsd-bugs at FreeBSD.org    |jamie at FreeBSD.org
             Status|New                         |In Progress

--- Comment #10 from Jamie Gritton <jamie at FreeBSD.org> ---
[A repeat from the database going down]

I'm with Bjoern in that I see the problem on the jail_set end, and not the
jail_get end.  But I wouldn't want to make the command fail.  It seems a
reasonable analog to (the correct) "ip4=disable", and in fact ends up treating
it the same way in every respect except the jail_get panic.

It's never proper for pr_ip4 to have a negative value.  It used to be, before
the PR_IP4_DISABLE flag, and I would test for it in the proper places.  But I
was incorrect to keep that vestige when I added the flag.  I must have
considered it proper at the time since I took pains to keep it, but a
years-later code review concludes differently.

My patch merely gets rid of the -1 that doesn't belong.

After that, the only difference between an ip4-disabled jail and one that is
set to no addresses is the PR_IP4_DISABLED flag.  Both have no addresses, and
will not admit to supporting IPV4.  In fact, PR_IP4_DISABLED, which is only
ever referenced in kern_jail.c, becomes write-only and useless.  I plan to
issue a second patch which removes it entirely.  But that isn't quite part of
this bug.

On the matter of signed versus unsigned ints in vfs_get/setopt, that's a whole
'nother can of worms, and one I don't think we need to open up just for this
one bug.

-- 
You are receiving this mail because:
You are the assignee for the bug.


More information about the freebsd-bugs mailing list