How does one find the valid signatures in D-Bus for an interface method?

Platform
,
1

From BlueZ documentation for the GATT API’s:

		void WriteValue(array{byte} value, dict options)

			Issues a request to write the value of the
			characteristic.

This is the error when I try to connect to the above API:

Method "WriteValue" with signature "aya{sv}" on interface "org.bluez.GattCharacteristic1" doesn't exist

How do I determine exactly what the signature should be? The error message has the parameter list described as I intended.

2

You can use

$ gdbus introspect --system --dest org.bluez --object-path /org/bluez --recurse

to get information on all interfaces of all objects under the /org/bluez hierarchy.

3 
gdbus introspect --system --dest org.bluez --object-path /org/bluez --recurse | grep -i gatt
    interface org.bluez.GattManager1 {

But org.bluez.GattCharacteristic1 does not exist in the output. That is unlikely, since it is used in the python-dbus example that comes with BlueZ (I have tried it and it works), and I have successfully done ReadValue:

 gatt_objects = g_dbus_connection_call_sync(con,
                "org.bluez",
                charp,
                "org.bluez.GattCharacteristic1",
                "ReadValue",
                g_variant_new_tuple(&v1, 1),  // Empty variant.
                NULL,
                G_DBUS_CALL_FLAGS_NONE,  // Flags
                10000, // 10 second timeout
                NULL, // Cancellable
                &errorO  // error
            );  // User data for callback
4

This stuff is apparently dynamic. So you cannot introspect until the device is connected:

        interface org.bluez.GattCharacteristic1 {
            methods:
              ReadValue(in  a{sv} options,
                        out ay value);
              WriteValue(in  ay value,
                         in  a{sv} options);
              AcquireWrite(in  a{sv} options,
                           out h fd,
                           out q mtu);
              AcquireNotify(in  a{sv} options,
                            out h fd,
                            out q mtu);
              StartNotify();
              StopNotify();
            signals:

So the first parameter for WriteValue is :
ay value

The 2nd is:
a{sv} options

Isn’t that what the error message is telling me is wrong? Or is the problem not in my code?

5

Hard to tell without seeing the code. Maybe you forgot to wrap the parameters in a tuple?

6

The function g_variant_get_type_string() can be used to get the formatting string for the GVariant.

So for sanity, I first tried this on org.freedesktop.DBus.Properties method called “Set”.

   Set(in  s interface,
            in  s name,
            in  v value);

This results in g_variant_get_type_string() on its parameter returns:

(ssv)

This function is working correctly.

For my org.bluez.GattCharacteristic1 method “WriteValue”, g_variant_get_type_string() returns:

(aya{sv})

The gdbus introspect returns:

         interface org.bluez.GattCharacteristic1 {
            methods:
              ReadValue(in  a{sv} options,
                        out ay value);
              WriteValue(in  ay value,
                         in  a{sv} options);

The error I see is:

Error message: Method "WriteValue" with signature "aya{sv}" on interface "org.bluez.GattCharacteristic1" doesn't exist

Code:

            GVariantBuilder *builder;
            GVariant *v1=g_variant_new("(q)", 0);
            GVariant *options;
            GVariant *parameters;
            const gchar *typestring;
            
            builder = g_variant_builder_new( G_VARIANT_TYPE("a{sv}") );
            g_variant_builder_add(builder, "{sv}", "offset", v1);
            options = g_variant_new ("a{sv}", builder);
            g_variant_builder_unref (builder);
            GBytes *bytes = g_bytes_new(RSU2_PASSWORD, strlen(RSU2_PASSWORD)+1);
            GVariant *pwd = g_variant_new_from_bytes(G_VARIANT_TYPE_BYTESTRING, bytes, TRUE);
            GVariant *args = g_variant_new("(@ay@a{sv})",pwd,options);

            typestring = g_variant_get_type_string(args);
            g_print("%p: %s: typestring for WriteValue: %s\n",g_thread_self(),__func__,typestring);

            gatt_objects = g_dbus_connection_call_sync(con,
                "org.bluez",
                charp,
                "org.bluez.GattCharacteristic1",
                "WriteValue",
                args,
                NULL,
                G_DBUS_CALL_FLAGS_NONE,  // Flags
                10000, // 10 second timeout
                NULL, // Cancellable
                &errorO  // error
            );  // User data for callback
7

I solved this one. The problem was that errorO was not cleared in my outer loop. I eventually intend to add a wait for the characteristic to appear, but for now I called the function in a loop, and errorO has to be set to NULL before invoking the function.

Apparently g_dbus_callsync will not overwrite &errorO with a new error.

So the “doesn’t exist” message means that it really did not exist yet, and now after clearing errorO I see:

0x7fd10c0082a0: print_error: Error message: GDBus.Error:org.bluez.Error.InvalidArguments: Invalid arguments in method call
0x7fd10c0082a0: print_error: Remote error
0x7fd10c0082a0: print_error: Raw message: org.bluez.Error.InvalidArguments
0x7fd10c0082a0: print_error: Returning

Rather strange semantics that the error parameter is not overwritten on the next call.

8

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.