ibmvnic: fix rx buffer tracking and index management in replenish_rx_pool partial...

We observed that in the error case for batched send_subcrq_indirect() the
driver does not account for the partial success case. This caused Linux to
crash when free_map and pool index are inconsistent.

Driver needs to update the rx pools "available" count when some batched
sends worked but an error was encountered as part of the whole operation.
Also track replenish_add_buff_failure for statistic purposes.

Fixes: 4f0b6812e9b9a ("ibmvnic: Introduce batched RX buffer descriptor transmission")
Signed-off-by: Dwip N. Banerjee <dnbanerg@us.ibm.com>
Reviewed-by: Dany Madden <drt@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

diff --git a/drivers/net/ethernet/ibm/ibmvnic.c b/drivers/net/ethernet/ibm/ibmvnic.c
index cb701a6..a219139 100644 (file)
--- a/drivers/net/ethernet/ibm/ibmvnic.c
+++ b/drivers/net/ethernet/ibm/ibmvnic.c
@@ -398,6 +398,8 @@ failure:
                dev_kfree_skb_any(pool->rx_buff[index].skb);
                pool->rx_buff[index].skb = NULL;
        }
+       adapter->replenish_add_buff_failure += ind_bufp->index;
+       atomic_add(buffers_added, &pool->available);
        ind_bufp->index = 0;
        if (lpar_rc == H_CLOSED || adapter->failover_pending) {
                /* Disable buffer pool replenishment and report carrier off if