On Thu, Jan 27, 2011 at 10:42:53AM +0200, Michael S. Tsirkin wrote: > On Thu, Jan 27, 2011 at 03:56:34PM +0900, Isaku Yamahata wrote: > > Changes v1 -> v2: > > - dropped PCI_DEVFN_MAX > > - use uint8_t for devfn instead of int > > - move pcie_check_slot into pci.h and made it static inline > > - minor clean ups > > So I put this on the pci branch but
Thank you. > we need to at least consider the following: Yes, there are several issues to consider. What I have in mind at the moment for the first step is to directly assign a physical pcie ARI device to a guest as a while. > - command line/monitor/qmp: do we need to tweak it to force slot 0? > Or just declare that slot >= 1 should be used to support > insane # of functions? So far I've thought of slot >= 1 way. But I'm not sure it's good or not. Anyway we will see. > - management needs to learn about ARI anyway, so > maybe a clean change is better? > - multifunction: need to detect and support ARI: > see pci_init_multifunction. > - And, we still did not implement the idea > where parent id for the device is specified. Parent id is definitely necessary for pcie hot plug too. thanks, > So while I don't think these patches make anything > worse, I think I'll give these a bit of time to stew > on my branch so we can hopefully see a bit more of the > full picture. > > > Isaku Yamahata (5): > > pci: replace the magic, 256, for the maximum of devfn > > pci: use devfn for pci_find_device() instead of (slot, fn) pair > > pci/pcie: make pci_find_device() ARI aware. > > pci: use PCI_SLOT in pci_get_bus_devfn() > > pci: use uint8_t for devfn_min > > > > hw/pci-hotplug.c | 5 +++-- > > hw/pci.c | 18 ++++++++++++------ > > hw/pci.h | 51 > > +++++++++++++++++++++++++++++++++++++++++++++++---- > > hw/pci_host.c | 2 +- > > hw/pci_internals.h | 4 ++-- > > hw/pcie.c | 13 ------------- > > hw/pcie.h | 1 - > > hw/pcie_host.c | 3 +-- > > 8 files changed, 66 insertions(+), 31 deletions(-) > -- yamahata