Re: find stuck in state "clock"

["Simon 'corecode' Schubert" <corecode@xxxxxxxxxxxx>]

Matthew Dillon wrote:
> :hey,
> :
> :while looking around on chlamydia why it is so slow I found a dozen 
> :find(1)s stuck in state "clock", and all periodic scripts etc dangling 
> :on that.  Dmesg tells me:
> :
> :Aug 28 03:01:04 chlamydia kernel: [diagnostic] cache_lock: blocked on 
> :0xc9630970 "ttyp5"
> :
> :the system is
> :
> :DragonFly chlamydia.fs.ei.tum.de 1.2.2-RELEASE DragonFly 1.2.2-RELEASE 
> :#3: Tue Jun 21 23:57:56 CEST 2005 
> :root@xxxxxxxxxxxxxxxxxxxxxx:/usr/obj/usr/src/sys/CHLAMYDIA  i386
> :
> :Has this been fixed since?
>
>     I believe it was fixed.  I'm not entirely sure if it was MFC'd but
>     I think it was.

I couldn't find a commit that I can relate to a fix for that problem, 
neither in -release nor in -devel.

Ok, I tried to play a bit Sherlock Holmes (on the live system).

I found:

find blocks in cache_lock() on "/dev/ttyp5"

This entry is locked by sshd in state vnode, waiting for an exclusive 
lock on a vnode which is being locked by tcsh, which is in state ttywai. 
 Guess on which tty: ttyp5.  Seems like a deadlock situation.

Is there something I can look at?  I could try to get a dump tomorrow, 
but I'd rather not, I should be busy at the moment.

cheers
  simon

--
Serve - BSD     +++  RENT this banner advert  +++    ASCII Ribbon   /"\
Work - Mac      +++  space for low $$$ NOW!1  +++      Campaign     \ /
Party Enjoy Relax   |   http://dragonflybsd.org      Against  HTML   \
Dude 2c 2 the max   !   http://golden-apple.biz       Mail + News   / \