On 6/30/2017 2:53 AM, Dan Carpenter wrote:
> 122 if (pps->main.state != IB_PORT_PKEY_NOT_VALID) {
> 123 get_pkey_and_subnet_prefix(&pps->main,
> 124 &pkey,
> ^^^^^
> Can get_pkey_and_subnet_prefix() fail? None of the callers seem to be
> error checked so my static checker complains a lot.
The only way it could fail is if some memory corruption happened, because the port number is checked and must be valid before this could happen. I submitted a patch for it though, it's definitely not clear it can't fail in that case.
>
> 125 &subnet_prefix);
> 126
> 127 ret = enforce_qp_pkey_security(pkey,
> 128 subnet_prefix,
> 129 sec);
> 130 }
> 131 if (ret)
> 132 return ret;
>
> Also:
>
> drivers/infiniband/core/device.c
> 362 static void ib_policy_change_task(struct work_struct *work)
> 363 {
> 364 struct ib_device *dev;
> 365
> 366 down_read(&lists_rwsem);
> 367 list_for_each_entry(dev, &device_list, core_list) {
> 368 int i;
> 369
> 370 for (i = rdma_start_port(dev); i <= rdma_end_port(dev); i++) {
> 371 u64 sp;
> 372 int ret = ib_get_cached_subnet_prefix(dev,
> 373 i,
> 374 &sp);
> 375
> 376 WARN_ONCE(ret,
> 377 "ib_get_cached_subnet_prefix err: %d, this should never happen here\n",
> 378 ret);
> 379 ib_security_cache_change(dev, i, sp);
> ^^
> Potentially uninitialized.
This in the only check that can fail in ib_get_cached_subnet_prefix:
if (port_num < rdma_start_port(device) ||
port_num > rdma_end_port(device))
return -EINVAL;
So this one won't even fail if memory is corrupted. I submitted a patch for it though, to resolve your warning, and in case that ever changes.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
