On 23 July 2015 at 22:10, Michael S. Tsirkin <m...@redhat.com> wrote: > On Thu, Jul 23, 2015 at 10:00:30PM +0100, Peter Maydell wrote: >> (Also, none of our PCI device models actually try to do >> the "BAR at zero means I won't respond" behaviour, which >> presumably they might do in real life.)
> Maybe some devices do this, but I'm guessing not all of them, > since there's no hint in the pci spec that they should. I think this depends on which version of the spec you read. The PCI 2.1 spec has an implementation note that says: # Note: A Base Address register does not contain a valid # address when it is equal to "0". which could be taken to mean "0 is invalid". A 'note' isn't part of the formal spec, of course (and the text got deleted from later spec revisions), but that doesn't mean implementers necessarily ignored it. There's an ancient Torvalds rant where he claims in passing that "a lot" of devices do this: http://yarchive.net/comp/linux/zero.html -- PMM
