possible race on HEAD?

Thomas Nikolajsen thomas.nikolajsen at mail.dk
Sat May 12 03:15:21 PDT 2007


::'ps alx' looks like:
::
::  UID   PID  PPID CPU PRI  NI   VSZ  RSS WCHAN  STAT  TT       TIME COMMAND
::    0  7027  1084   0 152   0  1632 1036 clock  DL+   p5    0:00.09 cvsup -L 2 /root/DragonFly-supfile
::
::
::Some of the other cvsups have vnode instead of clock.
::
::ylem:~# uname -a
::DragonFly ylem.theshell.com 1.9.0-DEVELOPMENT DragonFly 1.9.0-DEVELOPMENT #10: Wed May  9 18:32:59 EDT 2007     root at ylem.theshell.com:/usr/obj/usr/src/sys/YLEM  i386
::(sources from May 9th)
. .
:    "clock" is a namecache lock.  It's possible that you found a deadlock
:    but it is also possible that a low level block device or NFS mount got
:    stuck or something of that ilk.
:
:    Drop into the debugger, panic the system, and see if you can get a
:    kernel core dump.  If you can, upload the core and the kernel to leaf.

I have similar problem:
kernel from May 5 OK; May 9 (and newer) problem:
doing quickworld (or buildworld) hangs within a few minutes; ctl-T shows 'clock' state.

This happens when /usr/src and /usr/obj NFS mounted;
when only /usr/src NFS mounted (/usr/obj UFS): no problem.

When build proc. is hung: accessing UFS files works as expected;
but proces accessing other NFS mounted file usually also hangs.

 -thomas






More information about the Kernel mailing list