δΊ 2013/9/3 20:32, Paolo Bonzini ει:
> QOM splits the destruction of a device in two phases:
>
> - unrealize, also known as "exit" from qdev times, should isolate
> the device from the guest. After unrealize returns, the guest
> should not be able to issue new requests.
>
> - instance_finalize will reclaim the memory. This is only called
> after all requests terminate and drop the references on the
> device.
>
> Though overlooked, this is important even now: QEMU's little secret is
> that devices already do access memory out of the iothread mutex (with
> address_space_map/unmap and AIO), and this can be MMIO memory too
> through a bounce buffer. This series prepares things so that, once
> we'll put the memory_region_ref/unref infrastructure to complete use,
> things will just work.
>
> Of course this split will be particularly important for devices that
> will be able to do unlocked MMIO.
>
> This series changes all PCI devices (the sole to support hotplug _and_
> use MemoryRegions) to do memory_region_del_subregion at unrealize time,
> and memory_region_destroy at instance_finalize time. As it is mostly
> a PCI patch, it should go through mst's tree.
>
> Paolo
>
> Paolo Bonzini (38):
> qdev: document assumption that unrealize is followed by finalize
> pci: split exit and finalize
> ac97: use instance_finalize instead of exit
> es1370: use instance_finalize instead of exit
> hda: reclaim memory in instance_finalize instead of exit
> serial: reclaim memory in instance_finalize instead of exit
> tpci200: use instance_finalize instead of exit
> pci-assign: reclaim memory in instance_finalize instead of exit
> ahci: reclaim memory in instance_finalize instead of exit
> msix: split msix_free from msix_uninit
> cmd646: use instance_finalize instead of exit
> ide/piix: use instance_finalize instead of exit
> ide/via: use instance_finalize instead of exit
> ivshmem: reclaim memory in instance_finalize instead of exit
> pci-testdev: use instance_finalize instead of exit
> vfio: reclaim memory in instance_finalize instead of exit
> e1000: use instance_finalize instead of exit
> eepro100: use instance_finalize instead of exit
> ne2000: use instance_finalize instead of exit
> pcnet: use instance_finalize instead of exit
> rtl8139: use instance_finalize instead of exit
> vmxnet3: reclaim memory in instance_finalize instead of exit
> shpc: split shpc_free from shpc_cleanup
> pci_bridge: split pci_bridge_free from pci_bridge_exitfn
> pcie_aer: pcie_aer_exit really frees stuff
> pci_bridge: reclaim memory in instance_finalize instead of exit
> ioh4320: reclaim memory in instance_finalize instead of exit
> xio3130-downstream: reclaim memory in instance_finalize instead of
> exit
> xio3130-upstream: reclaim memory in instance_finalize instead of exit
> pcie: do not recreate mmcfg I/O region, use an alias instead
> esp: use instance_finalize instead of exit
> lsi: use instance_finalize instead of exit
> pvscsi: reclaim memory in instance_finalize instead of exit
> usb-uhci: use instance_finalize instead of exit
> virtio-pci: reclaim memory in instance_finalize instead of exit
> wdt_i6300esb: use instance_finalize instead of exit
> xen_pt: reclaim memory in instance_finalize instead of exit
> tpm: move add/del_subregion to realize/unrealize
>
> hw/audio/ac97.c | 5 ++--
> hw/audio/es1370.c | 5 ++--
> hw/audio/intel-hda.c | 8 ++++++
> hw/char/serial-pci.c | 24 ++++++++++++++++++
> hw/char/tpci200.c | 5 ++--
> hw/i386/kvm/pci-assign.c | 8 ++++++
> hw/ide/ahci.c | 2 +-
> hw/ide/ahci.h | 2 +-
> hw/ide/cmd646.c | 6 ++---
> hw/ide/ich.c | 12 ++++++---
> hw/ide/piix.c | 9 ++++---
> hw/ide/via.c | 6 ++---
> hw/misc/ivshmem.c | 13 +++++++---
> hw/misc/pci-testdev.c | 6 ++---
> hw/misc/vfio.c | 52
> +++++++++++++++++++++++++++++++++++---
> hw/net/e1000.c | 6 ++---
> hw/net/eepro100.c | 5 ++--
> hw/net/ne2000.c | 5 ++--
> hw/net/pcnet-pci.c | 6 ++---
> hw/net/rtl8139.c | 6 ++---
> hw/net/vmxnet3.c | 14 ++++++++--
> hw/pci-bridge/i82801b11.c | 1 +
> hw/pci-bridge/ioh3420.c | 11 +++++++-
> hw/pci-bridge/pci_bridge_dev.c | 13 +++++++++-
> hw/pci-bridge/xio3130_downstream.c | 11 +++++++-
> hw/pci-bridge/xio3130_upstream.c | 11 +++++++-
> hw/pci/msix.c | 22 +++++++++++-----
> hw/pci/pci.c | 15 ++++++++---
> hw/pci/pci_bridge.c | 5 ++++
> hw/pci/pcie_aer.c | 3 ++-
> hw/pci/pcie_host.c | 23 ++++++++++++-----
> hw/pci/shpc.c | 8 +++++-
> hw/scsi/esp-pci.c | 6 ++---
> hw/scsi/lsi53c895a.c | 6 ++---
> hw/scsi/vmw_pvscsi.c | 12 ++++++++-
> hw/tpm/tpm_tis.c | 17 +++++++++----
> hw/usb/hcd-uhci.c | 5 ++--
> hw/virtio/virtio-pci.c | 10 +++++++-
> hw/watchdog/wdt_i6300esb.c | 5 ++--
> hw/xen/xen_pt.c | 10 ++++++++
> hw/xen/xen_pt_config_init.c | 3 ---
> hw/xen/xen_pt_msi.c | 8 +++++-
> include/hw/pci/msix.h | 1 +
> include/hw/pci/pci_bridge.h | 1 +
> include/hw/pci/pcie_aer.h | 2 +-
> include/hw/pci/pcie_host.h | 1 +
> include/hw/pci/shpc.h | 1 +
> include/hw/qdev-core.h | 4 +++
> 48 files changed, 329 insertions(+), 91 deletions(-)
>
Just one question: where is the caller of .instance_finalize(), did
I missed that patch?
