mptutil(8) causes VMware VM to crash hard

[Charles Owens]

Hello,

We've been doing some work with mpt(4) and mptutil(8).  We noticed that
in VMware (both Workstation and ESX) the SCSI controller is showing up
as mpt0... so it seemed only natural to probe it with mptutil.  For
about 1 second, it looks ok (see below)... but what happened, in fact,
is that the VM went down pretty much instantly (nothing on console... no
panic, nothing).

[root at newercastle ~]# mptutil show adapter
mpt0 Adapter:
       Board Name: 0
   Board Assembly: 0
        Chip Name: C1030
    Chip Revision: 0
Read from remote host 169.254.222.65: Connection reset by peer
Connection to 169.254.222.65 closed.

A VMware log file has this to say:

Feb 11 16:06:50: vcpu-0| LSI: ProcessIOCPageRequest: unsupported page 6
action 0
Feb 11 16:06:50: vcpu-0| NOT_IMPLEMENTED
C:/ob/bora-44386/bora/devices/lsilogic/
lsilogic.c:2429
Feb 11 16:06:50: vcpu-0| Backtrace:
...

Behavior confirmed in both ESX and Workstation.  Any help with this
would be appreciated.  I'd be glad to test patches or whatever.

Also FYI, we're working with IBM hardware (xSeries 3550) with an actual
mpt-based RAID setup, and am seeing symptoms similar to that described
in this thread http://docs.freebsd.org/cgi/mid.cgi?4B56CD4C.80503 .  It
actually seemed fine at first (we could create and remove volumes at
will, fail drives, etc.) but now most commands except "show adapter"
result in segfault.  We'll be gathering debug info possibly testing the
patch that John Baldwin posted.  Preemptive suggestions welcome.

All of the above using FreeBSD 8.0-RELEASE-p2.


Thank you,
Charles


-- 
 Charles Owens
 Great Bay Software, Inc.