Hi,
I made the changed in the device mapper to implement the offset, also updated my branches of hurd and libpciaccess accordingly. It wasn't as trivial as it seemed, there were some /* HACK */ lines which in fact limited the number of pagers per device to 1, so once the system got a pager for "mem" at offset 0, I couldn't give it another offset without that affecting all other memory mappings for that device. I updated the structure of the hash table to get rid of the /* HACK */ lines and support many pagers per device, one pager per [device,offset]. (BTW, why is the the hash table limited to 127 slots?) On the other hand, I couldn't think of any other way to get rid of the struct pci_user_data to get the pager from the arbiter. Any ideas? I'll work on the new libpciaccess interface for the Hurd to get the pager unless somebody has a better idea.
