On Thu, Jun 28, 2012 at 04:46:34AM +0300, Uri Lublin wrote:
> Note that this change may affect performance a bit, as we now need to look
> for the libusb_device if we need it. Likely it's negligible.
> ---
> gtk/Makefile.am | 4 +
> gtk/channel-usbredir-priv.h | 4 +-
> gtk/channel-usbredir.c | 46 +++++-----
> gtk/map-file | 7 ++
> gtk/usb-device-manager-priv.h | 3 +
> gtk/usb-device-manager.c | 194 +++++++++++++++++++++++++++++------------
> gtk/usb-device-manager.h | 5 +-
> gtk/usb-device-widget.c | 9 +--
> spice-common | 2 +-
> 9 files changed, 182 insertions(+), 92 deletions(-)
>
> @@ -395,7 +396,7 @@ static void spice_usb_device_manager_class_init(SpiceUsbDeviceManagerClass *klas
> g_cclosure_marshal_VOID__BOXED,
> G_TYPE_NONE,
> 1,
> - SPICE_TYPE_USB_DEVICE);
> + G_TYPE_POINTER);
Why are you not keeping SPICE_TYPE_USB_DEVICE here and in the calls below?
>
> /**
> * SpiceUsbDeviceManager::device-removed:
> @@ -414,7 +415,7 @@ static void spice_usb_device_manager_class_init(SpiceUsbDeviceManagerClass *klas
> g_cclosure_marshal_VOID__BOXED,
> G_TYPE_NONE,
> 1,
> - SPICE_TYPE_USB_DEVICE);
> + G_TYPE_POINTER);
>
> /**
> * SpiceUsbDeviceManager::auto-connect-failed:
> @@ -435,7 +436,7 @@ static void spice_usb_device_manager_class_init(SpiceUsbDeviceManagerClass *klas
> g_cclosure_user_marshal_VOID__BOXED_BOXED,
> G_TYPE_NONE,
> 2,
> - SPICE_TYPE_USB_DEVICE,
> + G_TYPE_POINTER,
> G_TYPE_ERROR);
>
> /**
> @@ -457,7 +458,7 @@ static void spice_usb_device_manager_class_init(SpiceUsbDeviceManagerClass *klas
> g_cclosure_user_marshal_VOID__BOXED_BOXED,
> G_TYPE_NONE,
> 2,
> - SPICE_TYPE_USB_DEVICE,
> + G_TYPE_POINTER,
> G_TYPE_ERROR);
>
> g_type_class_add_private(klass, sizeof(SpiceUsbDeviceManagerPrivate));
> @@ -1009,34 +1019,32 @@ spice_usb_device_manager_can_redirect_device(SpiceUsbDeviceManager *self,
> *
> * Returns: a newly-allocated string holding the description, or %NULL if failed
> */
> -gchar *spice_usb_device_get_description(SpiceUsbDevice *_device, const gchar *format)
> +gchar *spice_usb_device_get_description(SpiceUsbDevice *device, const gchar *format)
> {
> #ifdef USE_USBREDIR
> - libusb_device *device = (libusb_device *)_device;
> - struct libusb_device_descriptor desc;
> - int bus, address;
> + int bus, addr, vid, pid;
> gchar *description, *descriptor, *manufacturer = NULL, *product = NULL;
>
> - g_return_val_if_fail(device != NULL, NULL);
> + g_return_val_if_fail(SPICE_IS_USB_DEVICE(device), NULL);
>
> - bus = libusb_get_bus_number(device);
> - address = libusb_get_device_address(device);
> + bus = spice_usb_device_get_busnum(device);
> + addr = spice_usb_device_get_devaddr(device);
> + vid = spice_usb_device_get_vid(device);
> + pid = spice_usb_device_get_pid(device);
>
> - if (libusb_get_device_descriptor(device, &desc) == LIBUSB_SUCCESS) {
> - spice_usb_util_get_device_strings(bus, address,
> - desc.idVendor, desc.idProduct,
> - &manufacturer, &product);
> - descriptor = g_strdup_printf("[%04x:%04x]", desc.idVendor, desc.idProduct);
> + if ((vid > 0) && (pid > 0)) {
> + descriptor = g_strdup_printf("[%04x:%04x]", vid, pid);
I don't think this will do the right thing in error cases, get_vid returns
an uint16_t which value is 0xffff when there's an error, it will be turned
into 65535 when it's cast to int, it won't be -1. I haven't looked at the
rest of the patch for similar issues.
Christophe
Attachment:
pgpy8_hzWtC26.pgp
Description: PGP signature
_______________________________________________ Spice-devel mailing list Spice-devel@xxxxxxxxxxxxxxxxxxxxx http://lists.freedesktop.org/mailman/listinfo/spice-devel