We can't easily enter digits other than 1 through 4 (inclusive)
so leave it up to the agent to figure out a good passcode
for the iCade.
Note that we can not use the VID/PID of the device, as it is not
yet known at that point.
---
plugins/autopair.c | 12 +++++++++++-
1 file changed, 11 insertions(+), 1 deletion(-)
diff --git a/plugins/autopair.c b/plugins/autopair.c
index 8c98c12..5d2f6f7 100644
--- a/plugins/autopair.c
+++ b/plugins/autopair.c
@@ -57,13 +57,23 @@ static ssize_t autopair_pincb(struct btd_adapter *adapter,
{
char addr[18];
char pinstr[7];
+ char name[25];
uint32_t class;
ba2str(device_get_address(device), addr);
class = btd_device_get_class(device);
- DBG("device %s 0x%x", addr, class);
+ device_get_name(device, name, sizeof(name));
+ name[sizeof(name) - 1] = 0;
+
+ DBG("device %s (%s) 0x%x", addr, name, class);
+
+ g_message ("vendor 0x%X product: 0x%X", btd_device_get_vendor (device), btd_device_get_product (device));
+
+ /* The iCade shouldn't use random PINs like normal keyboards */
+ if (name != NULL && strstr(name, "iCade") != NULL)
+ return 0;
/* This is a class-based pincode guesser. Ignore devices with an
* unknown class.
--
1.8.4.2
On Thu, 2017-09-07 at 16:46 +0900, ERAMOTO Masaya wrote:
> Hi Bastien,
>
> On 09/05/2017 02:54 AM, Bastien Nocera wrote:
> > We can't easily enter digits other than 1 through 4 (inclusive)
> > so leave it up to the agent to figure out a good passcode
> > for the iCade.
> >
> > Note that we can not use the VID/PID of the device, as it is not
> > yet known at that point.
>
> Does not the device class of iCade include its unique value or a
> value
> for similar devices?
No, the iCade is configured as a keyboard so it can be paired with
Apple iOS devices which only recognise keyboards. Each keypress and key
release is a separate keyboard key press/release.
In mainline Linux, you can find the hid-icade driver that'll take care
of transforming those key presses and key releases into joypad events:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/hid/hid-icade.c
As we don't want the autopair keyboard to generate impossible to "type"
numbers, we need to do this somewhere else. It's done in gnome-
bluetooth's pairing agent as the device was deemed too niche to have
this handled in bluez. We still need the quirk however.
Hi Bastien,
On 09/05/2017 02:54 AM, Bastien Nocera wrote:
> We can't easily enter digits other than 1 through 4 (inclusive)
> so leave it up to the agent to figure out a good passcode
> for the iCade.
>
> Note that we can not use the VID/PID of the device, as it is not
> yet known at that point.
Does not the device class of iCade include its unique value or a value
for similar devices?
> ---
> plugins/autopair.c | 13 ++++++++++++-
> 1 file changed, 12 insertions(+), 1 deletion(-)
>
> diff --git a/plugins/autopair.c b/plugins/autopair.c
> index d44cf539b..a9d2a930a 100644
> --- a/plugins/autopair.c
> +++ b/plugins/autopair.c
> @@ -60,13 +60,24 @@ static ssize_t autopair_pincb(struct btd_adapter *adapter,
> {
> char addr[18];
> char pinstr[7];
> + char name[25];
> uint32_t class;
>
> ba2str(device_get_address(device), addr);
>
> class = btd_device_get_class(device);
>
> - DBG("device %s 0x%x", addr, class);
> + device_get_name(device, name, sizeof(name));
> + name[sizeof(name) - 1] = 0;
> +
> + DBG("device '%s' (%s) class: 0x%x vid/pid: 0x%X/0x%X",
> + name, addr, class,
> + btd_device_get_vendor (device),
> + btd_device_get_product (device));
> +
> + /* The iCade shouldn't use random PINs like normal keyboards */
> + if (name != NULL && strstr(name, "iCade") != NULL)
> + return 0;
If there is not a value identified as iCade in the device class,
I think that it is better to use simply device->name instead of
copying since the full device name is used.
Regards,
Eramoto
>
> /* This is a class-based pincode guesser. Ignore devices with an
> * unknown class.
>