Panic on 1.2.6: "getnewbuf: locked buf"

Matthew Dillon dillon at apollo.backplane.com
Tue Nov 15 09:38:10 PST 2005


:I came across this while doing a large file copy between partitions,but otherwisenot much in particular was going on.  Since I was logged in via ssh,I'm not surewhen exactly the panic occurred, only that the cp never returned.  Thetarget filemay have been in use by a running program.
:I can provide a copy of the kernel and core if necessary.  The trace follows.
:Chris
:..
:syncing disks... panic: getnewbuf: locked bufmp_lock = 00000001; cpuid = 1; lapic.id = 0c000000boot() called on cpu#1Uptime: 19d15h34m18s
:...

    I just yesterday hit this exact panic on my own test box and track it 
    down.  I committed a workaround to HEAD yesterday.  It will be a while
    before it can be MFC'd to 1.2.x, however, because it's a pretty bad hack
    and only meant to be temporary until I can fix the root cause.

    However, if you want to try to manually patch the workaround into your
    own source tree, the commit was to 1.52 of kern/vfs_bio.c:

    http://www.dragonflybsd.org/cvsweb/src/sys/kern/vfs_bio.c.diff?r1=1.51&r2=1.52&f=u

    There are at least three fixes recently put in HEAD that need to be MFC'd
    to 1.2.x, but it probably isn't going to go into 1.2 officially until
    we release 1.4 around Christmas time.

						-Matt





More information about the Kernel mailing list