This patch stores the address type while storing the primary services
of LE Device which is later used for re-creating device from primary
service database.
---
src/adapter.c | 10 ++++++++--
src/device.c | 9 +++++++--
2 files changed, 15 insertions(+), 4 deletions(-)
diff --git a/src/adapter.c b/src/adapter.c
index 12b6aeb..5df49e3 100644
--- a/src/adapter.c
+++ b/src/adapter.c
@@ -1920,13 +1920,19 @@ static void create_stored_device_from_primary(char *key, char *value,
struct btd_adapter *adapter = user_data;
struct btd_device *device;
GSList *services, *uuids, *l;
+ /* Assume address type to be public by default */
+ addr_type_t addr_type = ADDR_TYPE_LE_PUBLIC;
if (g_slist_find_custom(adapter->devices,
key, (GCompareFunc) device_address_cmp))
return;
- /* FIXME: Get the correct LE addr type (public/random) */
- device = device_create(connection, adapter, key, ADDR_TYPE_LE_PUBLIC);
+ /* Check if address type can be parsed properly */
+ if (sscanf(value, "%d", &addr_type))
+ /* Increment by 2 (address type and space) */
+ value += 2;
+
+ device = device_create(connection, adapter, key, addr_type);
if (!device)
return;
diff --git a/src/device.c b/src/device.c
index ea6fec2..9c15d28 100644
--- a/src/device.c
+++ b/src/device.c
@@ -1748,14 +1748,19 @@ static void store_services(struct btd_device *device)
{
struct btd_adapter *adapter = device->adapter;
bdaddr_t dba, sba;
- char *str = primary_list_to_string(device->primaries);
+ addr_type_t addr_type;
+ char *primary = primary_list_to_string(device->primaries);
+ /* Allocate extra space for address type, space and \0 */
+ char *str = g_malloc(strlen(primary) + 2 + 1);
adapter_get_address(adapter, &sba);
- device_get_address(device, &dba, NULL);
+ device_get_address(device, &dba, &addr_type);
+ sprintf(str, "%d %s", addr_type, primary);
write_device_services(&sba, &dba, str);
g_free(str);
+ g_free(primary);
}
static void attio_connected(gpointer data, gpointer user_data)
--
1.7.0.4
Hi Claudio,
On Thu, Apr 26, 2012, Claudio Takahasi wrote:
> On Thu, Apr 26, 2012 at 6:32 AM, Johan Hedberg <[email protected]> wrote:
> > Hi Hemant,
> >
> > On Wed, Apr 25, 2012, Hemant Gupta wrote:
> >> This patch stores the address type while storing the primary services
> >> of LE Device which is later used for re-creating device from primary
> >> service database.
> >> ---
> >> ?src/adapter.c | ? 10 ++++++++--
> >> ?src/device.c ?| ? ?9 +++++++--
> >> ?2 files changed, 15 insertions(+), 4 deletions(-)
> >
> > Would it make more sense (and I'm not asking just you but also the
> > INdTers) to have the address type as part of the key instead of the
> > value. After all it's basically an extension to the address and that way
> > tightly bound to it. Maybe something like the following:
> >
> > 11:22:33:44:55:66|1 <value1>
> > aa:bb:cc:dd:ee:ff|0 <value2>
> >
> > The same should then also be considered for the longtermkeys file.
> >
> > Johan
>
> I agree if we append the address type for all LE/GATT storage
> files(not only for primary and longtermkeys).
> Probably we will have minor problems between BlueZ releases, but IMO
> it is acceptable since LE it is not officially supported.
> btw: the separator commonly used in the storage files is "#" instead of "|".
I'm fine with using # and yes all LE related files should use this
convention. And if possible these changes should go in before the next
BlueZ release.
Johan
Hi Hemant/Johan:
On Thu, Apr 26, 2012 at 6:32 AM, Johan Hedberg <[email protected]> wrote:
> Hi Hemant,
>
> On Wed, Apr 25, 2012, Hemant Gupta wrote:
>> This patch stores the address type while storing the primary services
>> of LE Device which is later used for re-creating device from primary
>> service database.
>> ---
>> src/adapter.c | 10 ++++++++--
>> src/device.c | 9 +++++++--
>> 2 files changed, 15 insertions(+), 4 deletions(-)
>
> Would it make more sense (and I'm not asking just you but also the
> INdTers) to have the address type as part of the key instead of the
> value. After all it's basically an extension to the address and that way
> tightly bound to it. Maybe something like the following:
>
> 11:22:33:44:55:66|1 <value1>
> aa:bb:cc:dd:ee:ff|0 <value2>
>
> The same should then also be considered for the longtermkeys file.
>
> Johan
I agree if we append the address type for all LE/GATT storage
files(not only for primary and longtermkeys).
Probably we will have minor problems between BlueZ releases, but IMO
it is acceptable since LE it is not officially supported.
btw: the separator commonly used in the storage files is "#" instead of "|".
BR,
Claudio
Hi Hemant,
On Wed, Apr 25, 2012, Hemant Gupta wrote:
> This patch stores the address type while storing the primary services
> of LE Device which is later used for re-creating device from primary
> service database.
> ---
> src/adapter.c | 10 ++++++++--
> src/device.c | 9 +++++++--
> 2 files changed, 15 insertions(+), 4 deletions(-)
Would it make more sense (and I'm not asking just you but also the
INdTers) to have the address type as part of the key instead of the
value. After all it's basically an extension to the address and that way
tightly bound to it. Maybe something like the following:
11:22:33:44:55:66|1 <value1>
aa:bb:cc:dd:ee:ff|0 <value2>
The same should then also be considered for the longtermkeys file.
Johan