Re: [PATCH 11/16] IB uverbs: add mthca mmap support

[Roland Dreier <rolandd@cisco.com>]

    Andrew> What's the thinking behind the VM_DONTCOPY there?

I think that was my paranoia about something like a process doing a
fork, the original process exiting, and the new process having page
still mapped even though the file has been released.  This is bad
because then we could map the same page to a different process and
have them collide.  But it seems that there will always be a reference
to the underlying struct file as long as someone has this mapping, so
I don't really need to worry about this and the VM_DONTCOPY can go.

    Andrew> What's actually being mapped here?  Hardware?  If so, is
    Andrew> VM_IO not needed?

Yes, this is a page from a PCI BAR.  However, we use remap_pfn_range()
to map the page, which sets VM_IO already.

 - R.