On Thu, Jun 18, 2020 at 06:02:59PM +0200, Garrit Franke wrote:
> This should prevent possible overflowing bits by using the BIT macro in
> vchiq_core
There is no reason to think that these will overflow. For that to
happen we would need to be using a 64bit with a 1 << 31 shift.
if (flags & BIT(i)) {
^^^^^
Is "flags" a 64 bit and can "i" go up to 31? Just say that it's a clean
up.
>
> Signed-off-by: Garrit Franke <garritfranke@xxxxxxxxx>
> ---
> .../interface/vchiq_arm/vchiq_core.c | 22 +++++++++----------
> 1 file changed, 11 insertions(+), 11 deletions(-)
>
> diff --git a/drivers/staging/vc04_services/interface/vchiq_arm/vchiq_core.c b/drivers/staging/vc04_services/interface/vchiq_arm/vchiq_core.c
> index ae9183db44ee..5a6d2bd59ec0 100644
> --- a/drivers/staging/vc04_services/interface/vchiq_arm/vchiq_core.c
> +++ b/drivers/staging/vc04_services/interface/vchiq_arm/vchiq_core.c
> @@ -39,9 +39,9 @@ struct vchiq_openack_payload {
> };
>
> enum {
> - QMFLAGS_IS_BLOCKING = (1 << 0),
> - QMFLAGS_NO_MUTEX_LOCK = (1 << 1),
> - QMFLAGS_NO_MUTEX_UNLOCK = (1 << 2)
> + QMFLAGS_IS_BLOCKING = BIT(0),
> + QMFLAGS_NO_MUTEX_LOCK = BIT(1),
> + QMFLAGS_NO_MUTEX_UNLOCK = BIT(2)
> };
>
> /* we require this for consistency between endpoints */
> @@ -526,14 +526,14 @@ request_poll(struct vchiq_state *state, struct vchiq_service *service,
> do {
> value = atomic_read(&service->poll_flags);
> } while (atomic_cmpxchg(&service->poll_flags, value,
> - value | (1 << poll_type)) != value);
> + value | BIT(poll_type)) != value);
>
> do {
> value = atomic_read(&state->poll_services[
> service->localport>>5]);
> } while (atomic_cmpxchg(
> &state->poll_services[service->localport>>5],
> - value, value | (1 << (service->localport & 0x1f)))
> + value, value | BIT((service->localport & 0x1f)))
^ ^
Too many parentheses.
Otherwise it looks fine.
regards,
dan carpenter
_______________________________________________
devel mailing list
devel@xxxxxxxxxxxxxxxxxxxxxx
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel
