On 05/27/2012 02:15 AM, Andreas Färber wrote:
Hello,
This mini-series, based on master, starts fixing issues in the way of calling
recursive object_realize() at machine-level (vl.c).
Patch 1 is a cherry-pick from Anthony's / Wan Peng's i440fx series, fixing
Coding Style issues and dropping parts dependent on MemoryRegion QOM'ification.
Patch 2 avoids two allocations, the PCIBus and the controller's PCIDevice,
by allocating them as part of the host controller state. This requires access
to struct PCIBus, which Anthony and Paolo have refactored on qom-next, so for
simplicity I just #include pci_internals.h for now.
Looks good to me.
Reviewed-by: Anthony Liguori aligu...@us.ibm.com
Anthony, do you have further refactorings of PCIBus queued already, moving the
struct to pci.h?
qom-rebase on github has my work to date. Much of it is in a pretty raw state
though.
Regards,
Anthony Liguori
Regards,
Andreas
Cc: Anthony Liguorianth...@codemonkey.ws
Cc: Paolo Bonzinipbonz...@redhat.com
Cc: Alexander Grafag...@suse.de
Andreas Färber (2):
pci-host: Turn into SysBus-derived QOM type
prep_pci: Create PCIBus and PCIDevice in-place
hw/pci_host.c | 11 ++
hw/pci_host.h |3 ++
hw/prep_pci.c | 62
3 files changed, 58 insertions(+), 18 deletions(-)