Re: [PATCH 1/2] core: don't set BR/EDR support when no flags present

[Johan Hedberg <johan.hedberg@gmail.com>]

Hi Scott,

On Thu, Apr 03, 2014, Scott James Remnant wrote:
> The logic for setting the BR/EDR supported flag checked for the
> non-presence of the BR/EDR Unsupported flag without checking whether
> any flags were present at all.
> 
> This meant all LE-only devices that returned Scan Response Data were
> being marked as supporting BR/EDR, since the flag was only set in the
> initial AD and not the SRD.
> ---
>  src/adapter.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/src/adapter.c b/src/adapter.c
> index e396a3d..b8b9a9c 100644
> --- a/src/adapter.c
> +++ b/src/adapter.c
> @@ -4342,7 +4342,8 @@ static void update_found_devices(struct btd_adapter *adapter,
>  
>  	device_update_last_seen(dev, bdaddr_type);
>  
> -	if (bdaddr_type != BDADDR_BREDR && !(eir_data.flags & EIR_BREDR_UNSUP))
> +	if (bdaddr_type != BDADDR_BREDR && eir_data.flags &&
> +					!(eir_data.flags & EIR_BREDR_UNSUP))
>  		device_set_bredr_support(dev, true);
>  
>  	if (eir_data.name != NULL && eir_data.name_complete)

Both patches have been applied. Thanks.

One thing to note regarding this first one: this is actually not a
problem with current bluetooth-next since there we merge the adv_ind and
scan_rsp into a single mgmt device found event. I.e. with such a kernel
we will not need to have this extra non-zero flags check. I added an
extra code comment to this place so we don't forget to fix it.

Johan