From mboxrd@z Thu Jan 1 00:00:00 1970 Content-Type: multipart/mixed; boundary="===============6251689078236229860==" MIME-Version: 1.0 From: Gustavo F. Padovan Subject: Re: [PATCH 1/8] bluetooth: add server support Date: Thu, 03 Feb 2011 14:30:39 -0200 Message-ID: <20110203163038.GA2168@joana> In-Reply-To: <4D4AD01E.9040800@linux.intel.com> List-Id: To: ofono@ofono.org --===============6251689078236229860== Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Hi Frederic, * Frederic Danis [2011-02-03 16:56:14 +0= 100]: > Hello Gustavo, > = > Le 02/02/2011 19:21, Gustavo F. Padovan a =C3=A9crit : > > Hi Frederic, > > > > * Frederic Danis [2011-02-01 15:17:5= 6 +0100]: > > > >> Hello Gustavo, > >> > >> Le 31/01/2011 21:51, Gustavo F. Padovan a =C3=A9crit : > >>> Initial code to have support to listen over a RFCOMM socket for incom= ing > >>> connections. > >>> --- > >>> Makefile.am | 1 + > >>> plugins/bluetooth.c | 165 +++++++++++++++++++++++++++++++++++++++= ++++++++++++ > >>> plugins/bluetooth.h | 11 ++++ > >>> 3 files changed, 177 insertions(+), 0 deletions(-) > >>> > >>> diff --git a/Makefile.am b/Makefile.am > >>> index a38fcb9..77b1453 100644 > >>> --- a/Makefile.am > >>> +++ b/Makefile.am > >>> @@ -321,6 +321,7 @@ builtin_sources +=3D plugins/bluetooth.c plugins/= bluetooth.h > >>> builtin_modules +=3D hfp > >>> builtin_sources +=3D plugins/hfp.c plugins/bluetooth.h > >>> > >>> +builtin_sources +=3D $(btio_sources) > >>> builtin_cflags +=3D @BLUEZ_CFLAGS@ > >>> builtin_libadd +=3D @BLUEZ_LIBS@ > >>> endif > >>> diff --git a/plugins/bluetooth.c b/plugins/bluetooth.c > >>> index 93dd7a1..dcf75e6 100644 > >>> --- a/plugins/bluetooth.c > >>> +++ b/plugins/bluetooth.c > >>> @@ -35,13 +35,58 @@ > >>> > >>> #include > >>> > >>> +#include > >>> #include "bluetooth.h" > >>> > >>> static DBusConnection *connection; > >>> static GHashTable *uuid_hash =3D NULL; > >>> static GHashTable *adapter_address_hash =3D NULL; > >>> +static GSList *server_list =3D NULL; > >>> static gint bluetooth_refcount; > >>> > >>> +struct server { > >>> + guint16 service; > >>> + gchar *name; > >>> + guint8 channel; > >>> + GIOChannel *io; > >>> + char *adapter; > >>> + guint handle; > >>> + ConnectFunc connect_cb; > >>> + gpointer user_data; > >>> +}; > >>> + > >>> +typedef struct { > >>> + guint8 b[6]; > >>> +} __attribute__((packed)) bdaddr_t; > >>> + > >>> +static void baswap(bdaddr_t *dst, const bdaddr_t *src) > >>> +{ > >>> + register unsigned char *d =3D (unsigned char *) dst; > >>> + register const unsigned char *s =3D (const unsigned char *) src; > >>> + register int i; > >>> + > >>> + for (i =3D 0; i< 6; i++) > >>> + d[i] =3D s[5-i]; > >>> +} > >>> + > >>> +static int str2ba(const char *str, bdaddr_t *ba) > >>> +{ > >>> + guint8 b[6]; > >>> + const char *ptr =3D str; > >>> + int i; > >>> + > >>> + for (i =3D 0; i< 6; i++) { > >>> + b[i] =3D (guint8) strtol(ptr, NULL, 16); > >>> + if (i !=3D 5&& !(ptr =3D strchr(ptr, ':'))) > >>> + ptr =3D ":00:00:00:00:00"; > >>> + ptr++; > >>> + } > >>> + > >>> + baswap(ba, (bdaddr_t *) b); > >>> + > >>> + return 0; > >>> +} > >>> + > >>> void bluetooth_create_path(const char *dev_addr, const char *adapt= er_addr, > >>> char *buf, int size) > >>> { > >>> @@ -371,6 +416,70 @@ static gboolean property_changed(DBusConnection = *connection, DBusMessage *msg, > >>> return TRUE; > >>> } > >>> > >>> +static void server_stop(gpointer data) > >>> +{ > >>> + struct server *server =3D data; > >>> + > >>> + if (server->handle> 0) { > >>> + DBusMessage *msg; > >>> + > >>> + msg =3D dbus_message_new_method_call(BLUEZ_SERVICE, > >>> + server->adapter, > >>> + BLUEZ_SERVICE_INTERFACE, > >>> + "RemoveRecord"); > >>> + dbus_message_append_args(msg, DBUS_TYPE_UINT32,&server->handle, > >>> + DBUS_TYPE_INVALID); > >>> + g_dbus_send_message(connection, msg); > >>> + > >>> + server->handle =3D 0; > >>> + } > >>> + > >>> + if (server->io !=3D NULL) { > >>> + g_io_channel_shutdown(server->io, TRUE, NULL); > >>> + g_io_channel_unref(server->io); > >>> + server->io =3D NULL; > >>> + } > >>> + > >>> + g_free(server->adapter); > >>> + server->adapter =3D NULL; > >>> +} > >>> + > >>> +static void new_connection(GIOChannel *io, gpointer user_data) > >>> +{ > >>> + struct server *server =3D user_data; > >>> + > >>> + DBG("%p", server); > >>> +} > >>> + > >>> +static void server_start(gpointer data, gpointer user_data) > >>> +{ > >>> + struct server *server =3D data; > >>> + char *addr, *path =3D user_data; > >>> + bdaddr_t baddr; > >>> + GError *err =3D NULL; > >>> + > >>> + if (server->handle !=3D 0) > >>> + return; > >>> + > >>> + addr =3D g_hash_table_lookup(adapter_address_hash, path); > >>> + str2ba(addr,&baddr); > >>> + server->io =3D bt_io_listen(BT_IO_RFCOMM, NULL, new_connection, > >>> + server, NULL,&err, > >>> + BT_IO_OPT_SOURCE_BDADDR,&baddr, > >>> + BT_IO_OPT_CHANNEL, server->channel, > >>> + BT_IO_OPT_SEC_LEVEL, BT_IO_SEC_MEDIUM, > >>> + BT_IO_OPT_INVALID); > >>> + if (server->io =3D=3D NULL) { > >>> + ofono_error("Bluetooth %s register failed: %s", > >>> + server->name, err->message); > >>> + g_error_free(err); > >>> + server_stop(server); > >>> + return; > >>> + } > >>> + > >>> + server->adapter =3D g_strdup(path); > >>> +} > >>> + > >> This will not allows to start server on multiple adapters, as the test > >> of the SDP Record handle will prevent to bt_io_listen on other adapter= s. > > > > Yes, it will. The SDP record is per adapter, so I call bt_io_listen on = that > > adapter and and add the SDP record to the adapter records. The code doe= s that > > for all adapters. > > > >> > >> I do not understand why you pass the adapter address, if it is omitted > >> the bt_io_listen will be done for all adapters. > > > > See above. > > > = > Without the second patch (which add the SDP record), I agree with you = > that bt_io_listen will be called for each adapter path. > But, you will lost reference to the io channel and to the path adapter = > (resulting in memory leak). Then we need a per adapter structure here. > = > With the second patch, this will only work if adapters are present = > during ofono launch. But, if the adapter is added later, after the sdp = > record has been added for the first adapters, bt_io_listen will not be = > called as server->handle is non-null. Another reason for the per adapter structure. > = > One other thing, I do not understand why you use BT_IO_OPT_SOURCE_BDADDR = > instead of BT_IO_OPT_SOURCE (this one accept Bluetooth address as string). Hum, we need to fix this. -- = Gustavo F. Padovan http://profusion.mobi --===============6251689078236229860==--