From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:38174) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1UkxID-0002aN-F0 for qemu-devel@nongnu.org; Fri, 07 Jun 2013 10:05:26 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1UkxIC-0006kw-7s for qemu-devel@nongnu.org; Fri, 07 Jun 2013 10:05:25 -0400 Received: from mail-oa0-x22c.google.com ([2607:f8b0:4003:c02::22c]:56020) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1UkxIC-0006kn-32 for qemu-devel@nongnu.org; Fri, 07 Jun 2013 10:05:24 -0400 Received: by mail-oa0-f44.google.com with SMTP id n12so3379875oag.3 for ; Fri, 07 Jun 2013 07:05:23 -0700 (PDT) From: Anthony Liguori In-Reply-To: <1370371954-8479-4-git-send-email-pbonzini@redhat.com> References: <1370371954-8479-1-git-send-email-pbonzini@redhat.com> <1370371954-8479-4-git-send-email-pbonzini@redhat.com> Date: Fri, 07 Jun 2013 09:05:15 -0500 Message-ID: <87y5amt3s4.fsf@codemonkey.ws> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Subject: Re: [Qemu-devel] [PATCH 03/39] pci: split exit and finalize List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Paolo Bonzini , qemu-devel@nongnu.org Cc: mst@redhat.com Paolo Bonzini writes: > To properly support devices that do DMA out of the BQL, destruction > needs to be done in two phases. First, the device is unrealized; > at this point, pending memory accesses can still be completed, but > no new accesses will be started. The second part is freeing the > device, which happens only after the reference count drops to zero; > this means that all memory accesses are complete. > > This patch changes the PCI core to delay destruction of the > bus-master address space and root region. > > Signed-off-by: Paolo Bonzini Reviewed-by: Anthony Liguori Regards, Anthony Liguori > --- > hw/pci/pci.c | 15 +++++++++++---- > 1 file changed, 11 insertions(+), 4 deletions(-) > > diff --git a/hw/pci/pci.c b/hw/pci/pci.c > index 776ad96..b60d9d0d 100644 > --- a/hw/pci/pci.c > +++ b/hw/pci/pci.c > @@ -779,6 +779,16 @@ static void pci_config_free(PCIDevice *pci_dev) > g_free(pci_dev->used); > } > > +static void pci_device_instance_finalize(Object *obj) > +{ > + PCIDevice *pci_dev = PCI_DEVICE(obj); > + > + qemu_free_irqs(pci_dev->irq); > + > + address_space_destroy(&pci_dev->bus_master_as); > + memory_region_destroy(&pci_dev->bus_master_enable_region); > +} > + > /* -1 for devfn means auto assign */ > static PCIDevice *do_pci_register_device(PCIDevice *pci_dev, PCIBus *bus, > const char *name, int devfn) > @@ -866,12 +876,8 @@ static PCIDevice *do_pci_register_device(PCIDevice *pci_dev, PCIBus *bus, > > static void do_pci_unregister_device(PCIDevice *pci_dev) > { > - qemu_free_irqs(pci_dev->irq); > pci_dev->bus->devices[pci_dev->devfn] = NULL; > pci_config_free(pci_dev); > - > - address_space_destroy(&pci_dev->bus_master_as); > - memory_region_destroy(&pci_dev->bus_master_enable_region); > } > > static void pci_unregister_io_regions(PCIDevice *pci_dev) > @@ -2243,6 +2249,7 @@ static const TypeInfo pci_device_type_info = { > .abstract = true, > .class_size = sizeof(PCIDeviceClass), > .class_init = pci_device_class_init, > + .instance_finalize = pci_device_instance_finalize, > }; > > static void pci_register_types(void) > -- > 1.8.1.4