| Message ID | 1306333427-26186-7-git-send-email-levinsasha928@gmail.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
* Sasha Levin <levinsasha928@gmail.com> wrote: > void virtio_console__init(struct kvm *kvm) > { > u8 dev, line, pin; > + u16 console_base_addr; > > if (irq__register_device(VIRTIO_ID_CONSOLE, &dev, &pin, &line) < 0) > return; > > virtio_console_pci_device.irq_pin = pin; > virtio_console_pci_device.irq_line = line; > + console_base_addr = ioport__find_free_range(); > + virtio_console_pci_device.bar[0] = console_base_addr | PCI_BASE_ADDRESS_SPACE_IO; > + cdev.base_addr = console_base_addr; > pci__register(&virtio_console_pci_device, dev); > - ioport__register(IOPORT_VIRTIO_CONSOLE, &virtio_console_io_ops, IOPORT_VIRTIO_CONSOLE_SIZE); > + ioport__register(console_base_addr, &virtio_console_io_ops, IOPORT_SIZE); Why is the ioport registration done in two steps? Wouldnt a better sequence be something like: > + console_base_addr = ioport__register(&virtio_console_io_ops, IOPORT_SIZE); > > + virtio_console_pci_device.bar[0] = console_base_addr | PCI_BASE_ADDRESS_SPACE_IO; > + cdev.base_addr = console_base_addr; > pci__register(&virtio_console_pci_device, dev); I.e. first register the ioport range - this would also get a free range for you, and then register the PCI driver? Or something even more compact could be done i suspect - all of the drivers seem to be using the same registration sequence. Thanks, Ingo -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Wed, 2011-05-25 at 16:42 +0200, Ingo Molnar wrote: > * Sasha Levin <levinsasha928@gmail.com> wrote: > > > void virtio_console__init(struct kvm *kvm) > > { > > u8 dev, line, pin; > > + u16 console_base_addr; > > > > if (irq__register_device(VIRTIO_ID_CONSOLE, &dev, &pin, &line) < 0) > > return; > > > > virtio_console_pci_device.irq_pin = pin; > > virtio_console_pci_device.irq_line = line; > > + console_base_addr = ioport__find_free_range(); > > + virtio_console_pci_device.bar[0] = console_base_addr | PCI_BASE_ADDRESS_SPACE_IO; > > + cdev.base_addr = console_base_addr; > > pci__register(&virtio_console_pci_device, dev); > > - ioport__register(IOPORT_VIRTIO_CONSOLE, &virtio_console_io_ops, IOPORT_VIRTIO_CONSOLE_SIZE); > > + ioport__register(console_base_addr, &virtio_console_io_ops, IOPORT_SIZE); > > Why is the ioport registration done in two steps? > > Wouldnt a better sequence be something like: > > > + console_base_addr = ioport__register(&virtio_console_io_ops, IOPORT_SIZE); > > > > + virtio_console_pci_device.bar[0] = console_base_addr | PCI_BASE_ADDRESS_SPACE_IO; > > + cdev.base_addr = console_base_addr; > > pci__register(&virtio_console_pci_device, dev); > > I.e. first register the ioport range - this would also get a free > range for you, and then register the PCI driver? > > Or something even more compact could be done i suspect - all of the > drivers seem to be using the same registration sequence. Some drivers need explicit IO ports, such as the serial driver. It was a question of whether I create another registration function (which may be as simple as calling 'ioport__find_free_range()' before calling the real registration) but it would lead to two 'registration' functions behaving very differently - one returning a new port and one that just does nothing except registering.
diff --git a/tools/kvm/virtio/console.c b/tools/kvm/virtio/console.c index affff0b..eeb5dc9 100644 --- a/tools/kvm/virtio/console.c +++ b/tools/kvm/virtio/console.c @@ -36,7 +36,6 @@ static struct pci_device_header virtio_console_pci_device = { .class = 0x078000, .subsys_vendor_id = PCI_SUBSYSTEM_VENDOR_ID_REDHAT_QUMRANET, .subsys_id = VIRTIO_ID_CONSOLE, - .bar[0] = IOPORT_VIRTIO_CONSOLE | PCI_BASE_ADDRESS_SPACE_IO, }; struct con_dev { @@ -50,6 +49,7 @@ struct con_dev { u8 status; u8 isr; u16 queue_selector; + u16 base_addr; void *jobs[VIRTIO_CONSOLE_NUM_QUEUES]; }; @@ -113,7 +113,7 @@ static bool virtio_console_pci_io_device_specific_in(void *data, unsigned long o static bool virtio_console_pci_io_in(struct kvm *kvm, u16 port, void *data, int size, u32 count) { - unsigned long offset = port - IOPORT_VIRTIO_CONSOLE; + unsigned long offset = port - cdev.base_addr; bool ret = true; mutex_lock(&cdev.mutex); @@ -181,7 +181,7 @@ static void virtio_console_handle_callback(struct kvm *kvm, void *param) static bool virtio_console_pci_io_out(struct kvm *kvm, u16 port, void *data, int size, u32 count) { - unsigned long offset = port - IOPORT_VIRTIO_CONSOLE; + unsigned long offset = port - cdev.base_addr; bool ret = true; mutex_lock(&cdev.mutex); @@ -243,12 +243,16 @@ static struct ioport_operations virtio_console_io_ops = { void virtio_console__init(struct kvm *kvm) { u8 dev, line, pin; + u16 console_base_addr; if (irq__register_device(VIRTIO_ID_CONSOLE, &dev, &pin, &line) < 0) return; virtio_console_pci_device.irq_pin = pin; virtio_console_pci_device.irq_line = line; + console_base_addr = ioport__find_free_range(); + virtio_console_pci_device.bar[0] = console_base_addr | PCI_BASE_ADDRESS_SPACE_IO; + cdev.base_addr = console_base_addr; pci__register(&virtio_console_pci_device, dev); - ioport__register(IOPORT_VIRTIO_CONSOLE, &virtio_console_io_ops, IOPORT_VIRTIO_CONSOLE_SIZE); + ioport__register(console_base_addr, &virtio_console_io_ops, IOPORT_SIZE); }
Signed-off-by: Sasha Levin <levinsasha928@gmail.com> --- tools/kvm/virtio/console.c | 12 ++++++++---- 1 files changed, 8 insertions(+), 4 deletions(-)