Re: i386_set_ioperm() problems

[Matthew Dillon <dillon@xxxxxxxxxxxxxxxxxxxx>]

    Ok, I believe I have fixed the private TSS issues that were causing
    the ioperm map to break.  I have committed the fixes, please update and
    test!  I'll hold off MFCing to -RELEASE and I'll hold off rolling 1.4.2
    until you've had a chance to test it.

    The previous code was:

    * Preventing the iomap from being properly activated by the
      initial call to i386_set_ioperm().

    * Allowing interrupts or blockages in kmem_free() to switch a thread
      before the TSS was properly validated or invalidated.

    Both issues should now be fixed.

						-Matt