Re: kern/51877: carp related panic during shutdown

[christos%zoulas.com@localhost (Christos Zoulas)]

On Jan 18,  7:05am, ozaki-r%netbsd.org@localhost (Ryota Ozaki) wrote:
-- Subject: Re: kern/51877: carp related panic during shutdown

| The following reply was made to PR kern/51877; it has been noted by GNATS.
| 
| From: Ryota Ozaki <ozaki-r%netbsd.org@localhost>
| To: Hauke Fath <hf%spg.tu-darmstadt.de@localhost>
| Cc: "gnats-bugs%NetBSD.org@localhost" <gnats-bugs%netbsd.org@localhost>, kern-bug-people%netbsd.org@localhost, 
| 	gnats-admin%netbsd.org@localhost
| Subject: Re: kern/51877: carp related panic during shutdown
| Date: Wed, 18 Jan 2017 16:00:00 +0900
| 
|  On Tue, Jan 17, 2017 at 7:24 PM, Hauke Fath <hf%spg.tu-darmstadt.de@localhost> wrote:
|  > On Tue, 17 Jan 2017 19:15:36 +0900, Ryota Ozaki wrote:
|  >> Could you check where mutex_tryenter+0x12 is?
|  >> (objdump -d netbsd |grep -A 30 'mutex_tryenter>:' or something)
|  >
|  > Attached.
|  
|  Thanks.
|  
|  The fault happened at MUTEX_SPIN_P(mtx) in mutex_tryenter perhaps
|  because mtx is an invalid pointer (address). The mtx comes from
|  this:
|    l = curlwp;  // or l = owner below
|    owner = (*l->l_syncobj->sobj_owner)(l->l_wchan);
|    lwp_trylock(owner);
|    mutex_tryenter(owner->l_mutex);
|  
|  IIUC, owner->l_mutex can be invalid if the adaptive mutex in question
|  is destroyed or the owner of the mutex is disappeared holding the mutex
|  for some reasons (or the data of the mutex is corrupted somehow).
|  The former doesn't happen in this case as I said and also the latter
|  is unlikely to happen.
|  
|  I don't have any ideas :-/ I hope clean build solves the issue.
|    ozaki-r


Can you put a printf in carp_detach carp_ifdetach and carp_clone_destroy
and see when they are called during shutdown?

christos