mpu: fixing the rnr bug in SHM mapping

same fix as dev mapping

kernel/src/managers/memory/memory_mpu.c

@@ -290,10 +290,17 @@ kstatus_t mgr_mm_map_shm(taskh_t tsk, shmh_t shm)
     if (unlikely((status = mgr_task_get_layout_from_handle(tsk, &layout_tab)) != K_STATUS_OKAY)) {
         goto err;
     }
+    /**
+     * NOTE: here, we ask for the first empty field in task layout.
+     * This field id must be incremented of 2 (<<1) as the first region of th layout_tab
+     * correspond to region 2 (see kernel and task memory mapping) as the kernel has locked
+     * regions 0 and 1 for itself.
+     */
     if (unlikely((status = mpu_get_free_id(layout_tab, TASK_MAX_RESSOURCES_NUM, &mpu_cfg.id)) != K_STATUS_OKAY)) {
         status = K_ERROR_BUSY;
         goto err;
     }
+    mpu_cfg.id += 2; /* as layout starts at task TXT, defined as reg2, it must be incremented */
     mpu_cfg.addr = (uint32_t)shm_meta->baseaddr;
     mpu_cfg.size = mpu_convert_size_to_region(shm_meta->size);