by Bastien Nocera

[permalink] [raw]

Subject: Re: [PATCH 8/9] plugins/sixaxis: Add support for DualShock 4/PS4 cable pairing

On Wed, 2017-09-27 at 11:12 +0200, Szymon Janc wrote:
> Hi Bastien,
>
> On Monday, 4 September 2017 20:12:11 CEST Bastien Nocera wrote:
> > From: Juha Kuikka <[email protected]>
> >
> > This patch adds support for "pairing" a Dualshock4 controller over
> > USB
> > into the sixaxis plugin, similarly to what the Sixaxis/PS3
> > controller
> > supported.
> >
> > Actual bonding happens on first connection, but the device will be
> > marked as trusted when the agent replies.
> >
> > This patch is based on DS4 supprt for sixpair tool by t0xicCode:
> > https://github.com/t0xicCode/sixpair/commit/975c38cb6cd61a2f0be3507
> > 14f0f64ce
> > f5f0432c ---
> > plugins/sixaxis.c | 73
> > ++++++++++++++++++++++++++++++++++++++++++++++++++++++- 1 file
> > changed, 72
> > insertions(+), 1 deletion(-)
> >
> > diff --git a/plugins/sixaxis.c b/plugins/sixaxis.c
> > index b62834d72..eb15acb92 100644
> > --- a/plugins/sixaxis.c
> > +++ b/plugins/sixaxis.c
> > @@ -83,10 +83,34 @@ static int sixaxis_get_device_bdaddr(int fd,
> > bdaddr_t
> > *bdaddr) return 0;
> > }
> >
> > +static int ds4_get_device_bdaddr(int fd, bdaddr_t *bdaddr)
> > +{
> > + uint8_t buf[7];
> > + int ret;
> > +
> > + memset(buf, 0, sizeof(buf));
> > +
> > + buf[0] = 0x81;
> > +
> > + ret = ioctl(fd, HIDIOCGFEATURE(sizeof(buf)), buf);
> > + if (ret < 0) {
> > + error("ds4: failed to read device address (%s)",
> > + strerror(errno));
>
> Lets prefix all sixaxis error/info logs with "sixaxis:"

Sure, I'll make some changes to mention whether the sixaxis or DS4
variant of the call was made.

Nothing else?

2017-10-04 12:38:08

by Bastien Nocera

[permalink] [raw]

Subject: Re: [PATCH 3/9] sixaxis: Ask user whether cable configuration should be allowed

On Wed, 2017-09-27 at 11:14 +0200, Szymon Janc wrote:
> Hi Bastien,
>
> On Monday, 4 September 2017 20:12:06 CEST Bastien Nocera wrote:
> > Previously, users doing cable configuration of Sixaxis PS3
> > controllers
> > would only get asked whether a device was allowed to connect to the
> > computer when switching it to Bluetooth mode: unplugging it, and
> > pressing the PS button.
> >
> > Instead, we should ask the user straight away, through the agent,
> > whether the pad should be allowed to connect.
> >
> > This makes it easier to setup those devices, while keeping
> > security.
> > ---
> > plugins/sixaxis.c | 83
> > +++++++++++++++++++++++++++++++++++++++++-------------- 1 file
> > changed, 63
> > insertions(+), 20 deletions(-)
> >
> > diff --git a/plugins/sixaxis.c b/plugins/sixaxis.c
> > index 7d3a8f3ac..c8305d52f 100644
> > --- a/plugins/sixaxis.c
> > +++ b/plugins/sixaxis.c
> > @@ -44,6 +44,7 @@
> >
> > #include "src/adapter.h"
> > #include "src/device.h"
> > +#include "src/agent.h"
> > #include "src/plugin.h"
> > #include "src/log.h"
> > #include "src/shared/util.h"
> > @@ -71,6 +72,13 @@ static const struct {
> > },
> > };
> >
> > +struct authentication_closure {
> > + struct btd_adapter *adapter;
> > + struct btd_device *device;
> > + int fd;
> > + char device_addr[18];
> > +};
> > +
> > static struct udev *ctx = NULL;
> > static struct udev_monitor *monitor = NULL;
> > static guint watch_id = 0;
> > @@ -135,19 +143,51 @@ static int set_master_bdaddr(int fd, const
> > bdaddr_t
> > *bdaddr) return ret;
> > }
> >
> > +static void agent_auth_cb(DBusError *derr,
> > + void *user_data)
> > +{
> > + struct authentication_closure *closure = user_data;
> > + char master_addr[18], adapter_addr[18];
> > + bdaddr_t master_bdaddr;
> > + const bdaddr_t *adapter_bdaddr;
> > +
> > + if (derr != NULL) {
> > + DBG("Agent replied negatively, removing temporary
> > device");
> > + goto error;
> > + }
> > +
> > + btd_device_set_temporary(closure->device, false);
> > +
> > + if (get_master_bdaddr(closure->fd, &master_bdaddr) < 0)
> > + goto error;
> > +
> > + adapter_bdaddr = btd_adapter_get_address(closure-
> > >adapter);
> > + if (bacmp(adapter_bdaddr, &master_bdaddr)) {
> > + if (set_master_bdaddr(closure->fd, adapter_bdaddr)
> > < 0)
> > + goto error;
> > + }
> > +
> > + ba2str(&master_bdaddr, master_addr);
> > + ba2str(adapter_bdaddr, adapter_addr);
> > + DBG("remote %s old_master %s new_master %s",
> > + closure->device_addr, master_addr,
> > adapter_addr);
> > +
> > +error:
> > + close(closure->fd);
> > + g_free(closure);
> > +}
> > +
> > static bool setup_device(int fd, int index, struct btd_adapter
> > *adapter)
> > {
> > - char device_addr[18], master_addr[18], adapter_addr[18];
> > - bdaddr_t device_bdaddr, master_bdaddr;
> > + char device_addr[18];
> > + bdaddr_t device_bdaddr;
> > const bdaddr_t *adapter_bdaddr;
> > struct btd_device *device;
> > + struct authentication_closure *closure;
> >
> > if (get_device_bdaddr(fd, &device_bdaddr) < 0)
> > return false;
> >
> > - if (get_master_bdaddr(fd, &master_bdaddr) < 0)
> > - return false;
> > -
> > /* This can happen if controller was plugged while already
> > connected
> > * eg. to charge up battery. */
> > device = btd_adapter_find_device(adapter, &device_bdaddr,
> > @@ -155,25 +195,14 @@ static bool setup_device(int fd, int index,
> > struct
> > btd_adapter *adapter) if (device &&
> > btd_device_is_connected(device))
> > return false;
> >
> > - adapter_bdaddr = btd_adapter_get_address(adapter);
> > -
> > - if (bacmp(adapter_bdaddr, &master_bdaddr)) {
> > - if (set_master_bdaddr(fd, adapter_bdaddr) < 0)
> > - return false;
> > - }
> > -
> > ba2str(&device_bdaddr, device_addr);
> > - ba2str(&master_bdaddr, master_addr);
> > - ba2str(adapter_bdaddr, adapter_addr);
> > - DBG("remote %s old_master %s new_master %s",
> > - device_addr, master_addr,
> > adapter_addr);
> >
> > device = btd_adapter_get_device(adapter, &device_bdaddr,
> > BDADDR_BREDR);
> >
> > if (g_slist_find_custom(btd_device_get_uuids(device),
> > HID_UUID,
> > (GCompareFunc)strc
> > asecmp)) {
> > DBG("device %s already known, skipping",
> > device_addr);
> > - return true;
> > + return false;
> > }
> >
> > info("sixaxis: setting up new device");
> > @@ -181,7 +210,20 @@ static bool setup_device(int fd, int index,
> > struct
> > btd_adapter *adapter) btd_device_device_set_name(device,
> > devices[index].name);
> > btd_device_set_pnpid(device, devices[index].source,
> > devices[index].vid,
> > devices[index].pid,
> > devices[index].version);
> > - btd_device_set_temporary(device, false);
> > + btd_device_set_temporary(device, true);
> > +
> > + closure = g_try_new0(struct authentication_closure, 1);
> > + if (!closure) {
> > + btd_adapter_remove_device(adapter, device);
> > + return false;
> > + }
> > + closure->adapter = adapter;
> > + closure->device = device;
> > + closure->fd = fd;
> > + memcpy(&closure->device_addr, device_addr,
> > sizeof(device_addr));
> > + adapter_bdaddr = btd_adapter_get_address(adapter);
> > + btd_request_authorization_cable_configured(adapter_bdaddr,
> > &device_bdaddr,
> > + HID_UUID, agent_auth_cb, closure);
> >
> > return true;
> > }
> > @@ -236,8 +278,9 @@ static void device_added(struct udev_device
> > *udevice)
> > if (fd < 0)
> > return;
> >
> > - setup_device(fd, index, adapter);
> > - close(fd);
> > + /* Only close the fd if an authentication is not pending
> > */
> > + if (!setup_device(fd, index, adapter))
> > + close(fd);
> > }
> >
> > static gboolean monitor_watch(GIOChannel *source, GIOCondition
> > condition,
>
> So user will be asked for consent after DS3 was configured? Not that
> I have
> strong opinion on this but did you consider configuring it only after
> user
> gave consent?

We need to create a temporary device so that it's visible through the
D-Bus API. The flow is:
- joypad gets plugged in
- sixaxis plugin's udev signal watch gets triggered
- we fetch the device's details (Bluetooth address, name)
- we create a pairing request which gets passed to front-ends (such as
the Bluetooth panel in GNOME's Settings)
- we get the answer back, if the pairing is cancelled, or there wasn't
an agent to answer the question, we remove the temporary device (the
device is now charging, and usable through USB), otherwise we write the
new master to the joypad, so it connects to us when wireless, and mark
it as trusted so it can connect without any further questions.

There's still no way to pair the device without user interaction, as
before this patchset, but we avoid overwriting the device's
configuration when we just want to charge it, or use it wired, and the
"can it connect yes/no" happens when we plug it via USB rather than
when we finally disconnect it and press the "P" button to connect
wirelessly.