explanation of hackvideo patch to proc.c?

[David C. <da...@co...>]

I'm working on a problem where I need to be able to have multiple X
servers responding concurrently (the video driver is not actually
writing to a framebuffer device), and so I was trying out
the /proc/bus/pci/hackvideo patch provided by linuxconsole/ruby.  It
does not seem to solve my problem, but I would like to understand what
it is supposed to do (to hopefully give me better insight into where my
problem is).  I believe the following code in proc_bus_pci_write() in
drivers/pci/proc.c is the only modification to the 2.6 kernel code
(other than reading & writing the proc entry) for the hackvideo patch:

if (pci_hackvideo && (pos == PCI_CMD_STAT_REG) &&
           (buf[0] <= PCI_CMD_MASTER_ENABLE) &&
            ((dev->class>>8) == 0x0300 ||
             (dev->class>>8) == 0x0400 ||
             (dev->class>>8) == 0x0000   ) )
                return -EINVAL;

It is not clear to me what this code is doing, however.  Can anyone
offer an explanation?  Does it relate to XFree86's disabling of port I/O
access when switching VTs?  If so, is it designed to block the
functionality of a particular function called within xf86VTSwitch() in
xc/programs/Xserver/hw/xfree86/common/xf86Events.c?


TIA for any insight you can offer,
David Coulthart