netvsc PMD crashes in secondary process in dev_info_get

https://bugs.dpdk.org/show_bug.cgi?id=384

            Bug ID: 384
           Summary: netvsc PMD crashes in secondary process in
                    dev_info_get
           Product: DPDK
           Version: 19.08
          Hardware: x86
                OS: Linux
            Status: UNCONFIRMED
          Severity: normal
          Priority: Normal
         Component: ethdev
          Assignee: [email protected]
          Reporter: [email protected]

Target Milestone: ---

Call rte_ethdev_dev_info_get in a secondary process will crash with netvsc PMD.

The data structure mapping for the vmbus is not mapped at same address. It is crashing here because hv->vmbus is not the correct in secondary.

static void *hn_rndis_alloc(struct hn_data *hv, size_t size) {
return rte_zmalloc_socket("RNDIS", size, PAGE_SIZE,
hv->vmbus->device.numa_node);
}

The numa_node is not needed for control messages.
But fixing that just exposes the next crash when sending the control message.

You are receiving this mail because:
You are the assignee for the bug.

https://bugs.dpdk.org/show_bug.cgi?id=384

Stephen Hemminger ([email protected]) changed:

What |Removed |Added
----------------------------------------------------------------------------

             Status|UNCONFIRMED                 |RESOLVED
         Resolution|---                         |FIXED

--- Comment #1 from Stephen Hemminger ([email protected]) --- Fixed in 20.02 by
7214dda61361 ("net/netvsc: eliminate NUMA specific allocation")

You are receiving this mail because:
You are the assignee for the bug.