Cloud Hypervisor extra PCI buses no longer reserve slot 0
Brief
Anyone still on v53.0 still plants the dummy on every bus.
Notes
fd86afefe9057fcbfc46619aa81b112df733a4f5 (Sebastien Boeuf, 2026-08-19). pci, vmm: create a host bridge only on PCI segment 0. Extra segments leave device 0 free for an endpoint. Hotplug on those segments now lands at 00:00.0 instead of 00:01.0. The guest sees one host bridge.
37b2a078311e589a1be469496ba9abf6ac487e82 is the matching config check. NVIDIA Grace GPU drivers reject a non-zero device number in the SRAT Generic Initiator entry, so the device has to be 00:00.0. Segment 0 still reserves that slot. Same author, same push.
No new release tag. Latest release remains v53.0. No GHSA in the messages. Not a recast of 01f1e783, b7f19864, 3fa4e75c, or the tap-EIO card. Not a crate bump.
Commit fd86afefe9057fcbfc46619aa81b112df733a4f5