patches/kernel/0009-i40e-prevent-overlapping-tx_timeout-recover.patch

   1 From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
   2 From: Alan Brady <alan.brady@intel.com>
   3 Date: Mon, 29 Oct 2018 11:27:21 -0700
   4 Subject: [PATCH] i40e: prevent overlapping tx_timeout recover
   5
   6 If a TX hang occurs, we attempt to recover by incrementally resetting.
   7 If we're starved for CPU time, it's possible the reset doesn't actually
   8 complete (or even fire) before another tx_timeout fires causing us to
   9 fly through the different resets without actually doing them.
  10
  11 This adds a bit to set and check if a timeout recovery is already
  12 pending and, if so, bail out of tx_timeout.  The bit will get cleared at
  13 the end of i40e_rebuild when reset is complete.
  14
  15 Signed-off-by: Alan Brady <alan.brady@intel.com>
  16 Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
  17 Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
  18 Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
  19 ---
  20  drivers/net/ethernet/intel/i40e/i40e.h      | 1 +
  21  drivers/net/ethernet/intel/i40e/i40e_main.c | 5 +++++
  22  2 files changed, 6 insertions(+)
  23
  24 diff --git a/drivers/net/ethernet/intel/i40e/i40e.h b/drivers/net/ethernet/intel/i40e/i40e.h
  25 index e019baa905c5..80114d6a910a 100644
  26 --- a/drivers/net/ethernet/intel/i40e/i40e.h
  27 +++ b/drivers/net/ethernet/intel/i40e/i40e.h
  28 @@ -145,6 +145,7 @@ enum i40e_state_t {
  29         __I40E_MDD_EVENT_PENDING,
  30         __I40E_VFLR_EVENT_PENDING,
  31         __I40E_RESET_RECOVERY_PENDING,
  32 +       __I40E_TIMEOUT_RECOVERY_PENDING,
  33         __I40E_MISC_IRQ_REQUESTED,
  34         __I40E_RESET_INTR_RECEIVED,
  35         __I40E_REINIT_REQUESTED,
  36 diff --git a/drivers/net/ethernet/intel/i40e/i40e_main.c b/drivers/net/ethernet/intel/i40e/i40e_main.c
  37 index 7895a0af37e6..874fd143c351 100644
  38 --- a/drivers/net/ethernet/intel/i40e/i40e_main.c
  39 +++ b/drivers/net/ethernet/intel/i40e/i40e_main.c
  40 @@ -365,6 +365,10 @@ static void i40e_tx_timeout(struct net_device *netdev)
  41                       (pf->tx_timeout_last_recovery + netdev->watchdog_timeo)))
  42                 return;   /* don't do any new action before the next timeout */
  43
  44 +       /* don't kick off another recovery if one is already pending */
  45 +       if (test_and_set_bit(__I40E_TIMEOUT_RECOVERY_PENDING, pf->state))
  46 +               return;
  47 +
  48         if (tx_ring) {
  49                 head = i40e_get_head(tx_ring);
  50                 /* Read interrupt register */
  51 @@ -9478,6 +9482,7 @@ static void i40e_rebuild(struct i40e_pf *pf, bool reinit, bool lock_acquired)
  52         clear_bit(__I40E_RESET_FAILED, pf->state);
  53  clear_recovery:
  54         clear_bit(__I40E_RESET_RECOVERY_PENDING, pf->state);
  55 +       clear_bit(__I40E_TIMEOUT_RECOVERY_PENDING, pf->state);
  56  }
  57
  58  /**