> Automatically cleaned up pointers need to be initialized before exiting
> their scope. In this case, they need to be initialized to NULL before
> any return statement.
Will any adjustments become relevant also for this change description
if scope reductions would become more appealing for affected local variables?
How much can a small script (like the following) for the semantic patch language
(Coccinelle software) help to achieve a better common understanding for
possible source code transformations?
// See also:
// drivers/net/ethernet/intel/ice/ice_common.c
@movement1@
attribute name __free;
@@
-struct ice_aqc_get_phy_caps_data *pcaps __free(kfree);
... when any
+struct ice_aqc_get_phy_caps_data *
pcaps
+__free(kfree)
= kzalloc(sizeof(*pcaps), ...);
@movement2@
attribute name __free;
@@
-void *mac_buf __free(kfree);
... when any
+void *
mac_buf
+__free(kfree)
= kcalloc(2, sizeof(struct ice_aqc_manage_mac_read_resp), ...);
// See also:
// drivers/net/ethernet/intel/ice/ice_ethtool.c
@movement3@
attribute name __free;
@@
-u8 *tx_frame __free(kfree);
int i;
... when any
if (ice_fltr_add_mac(test_vsi, ...))
{ ... }
+
+{
+u8 *tx_frame __free(kfree) = NULL;
if (ice_lbtest_create_frame(pf, &tx_frame, ...))
{ ... }
... when any
+}
+
valid_frames = ice_lbtest_receive_frames(...);
Regards,
Markus
