Article 13287 of comp.os.ms-windows.programmer.nt.kernel-mode:
John, it's a bad pointer in your driver.  Here is some email that I
exchanged with Tony Mason of OSR (giving him credit for this answer),

"When Lou designed the VM System he divvied up the system portion of the
address space into a variety of different pieces - there's a piece for the
OS itself, a piece for various data structures, etc.  One area he set up is
for the Cache Manager (with a max size of 256MB) and another was for "non
paged" addresses (with a max size of 192MB.)  There's other parts used for
paged pool, and other parts used for non-cached memory addresses
(typically, these are used for device memory, for instance, so the CPU's on
chip cache doesn't cache those memory accesses.)
When the VM system gets a page fault it actually uses information about
where the address is in figuring out how to handle it.  Thus, if a page
fault occurs anywhere within the 192MB range marked as "non-paged pool" the
VM system knows its done - after all, that's a NON PAGED address and we're
handling a page fault for it.  You can see the code for this in the
MmAccessFault.
Hence, the standard rule of thumb - "PAGE_FAULT_IN_NONPAGED_AREA" just
means "bad pointer".  It's just that it happens to be a bad pointer in the
non-paged portion of the virtual address."

-Paul
jvasquez@getntds.com wrote in message
<34844483.3052714@news.connectnet.com>...
>I'm doing a loopback test using my newly created Windows NT Device
>Driver. The loopback program will run for awhile and then I get the
>famous "Blue Screen" with
>PAGE_FAULT_IN_NONPAGED_AREA 0X50
>(BAD0B0F8, 0, 0, 2)
>Does anybody know what this may mean? Is it my driver or the
>application?
>
>John V.
>