Re: [PATCH v5 11/15] coco: host: arm64: Connect RMM pdev streams for IDE devices

[Ankit Agrawal]

Hi Aneesh

It isn't clear to me why does pci_dev_addr_range() derive addr_range
from the upstream bridge PCI_BRIDGE_MEM_WINDOW /
PCI_BRIDGE_PREF_MEM_WINDOW rather than from pdev's own
regions? Should it be filled up with device regions instead?

IIUC, there may be a couple of issues with the current approach:
1. The bridge's window can be wider than what pdev actually owns due
to alignment padding.
2. If the bridge has multiple downstream devices (are there any?),
the window spans address space belonging to sibling devices too
and not just this pdev. So RMM would get a range that isn't
exclusive to the pdev/stream being connected.