i40e: Use smp_rmb rather than read_barrier_depends
authorBrian King <brking@linux.vnet.ibm.com>
Fri, 17 Nov 2017 17:05:44 +0000 (11:05 -0600)
committerJeff Kirsher <jeffrey.t.kirsher@intel.com>
Wed, 22 Nov 2017 07:43:21 +0000 (23:43 -0800)
The original issue being fixed in this patch was seen with the ixgbe
driver, but the same issue exists with i40e as well, as the code is
very similar. read_barrier_depends is not sufficient to ensure
loads following it are not speculatively loaded out of order
by the CPU, which can result in stale data being loaded, causing
potential system crashes.

Cc: stable <stable@vger.kernel.org>
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Acked-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
drivers/net/ethernet/intel/i40e/i40e_main.c
drivers/net/ethernet/intel/i40e/i40e_txrx.c

index 775d5a1..4c08cc8 100644 (file)
@@ -3966,7 +3966,7 @@ static bool i40e_clean_fdir_tx_irq(struct i40e_ring *tx_ring, int budget)
                        break;
 
                /* prevent any other reads prior to eop_desc */
-               read_barrier_depends();
+               smp_rmb();
 
                /* if the descriptor isn't done, no work yet to do */
                if (!(eop_desc->cmd_type_offset_bsz &
index d6d352a..4566d66 100644 (file)
@@ -759,7 +759,7 @@ static bool i40e_clean_tx_irq(struct i40e_vsi *vsi,
                        break;
 
                /* prevent any other reads prior to eop_desc */
-               read_barrier_depends();
+               smp_rmb();
 
                i40e_trace(clean_tx_irq, tx_ring, tx_desc, tx_buf);
                /* we have caught up to head, no work left to do */