i40e: Fix removing driver while bare-metal VFs pass traffic

Prevent VFs from resetting when PF driver is being unloaded:
- introduce new pf state: __I40E_VF_RESETS_DISABLED;
- check if pf state has __I40E_VF_RESETS_DISABLED state set,
  if so, disable any further VFLR event notifications;
- when i40e_remove (rmmod i40e) is called, disable any resets on
  the VFs;

Previously if there were bare-metal VFs passing traffic and PF
driver was removed, there was a possibility of VFs triggering a Tx
timeout right before iavf_remove. This was causing iavf_close to
not be called because there is a check in the beginning of  iavf_remove
that bails out early if adapter->state < IAVF_DOWN_PENDING. This
makes it so some resources do not get cleaned up.

Fixes: 6a9ddb36ee ("i40e: disable IOV before freeing resources")
Signed-off-by: Slawomir Laba <slawomirx.laba@intel.com>
Signed-off-by: Brett Creeley <brett.creeley@intel.com>
Signed-off-by: Sylwester Dziedziuch <sylwesterx.dziedziuch@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Link: https://lore.kernel.org/r/20201120180640.3654474-1-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
This commit is contained in:
Sylwester Dziedziuch 2020-11-20 10:06:40 -08:00 committed by Jakub Kicinski
parent 3fe356d58e
commit 2980cbd4dc
3 changed files with 31 additions and 18 deletions

View File

@ -140,6 +140,7 @@ enum i40e_state_t {
__I40E_CLIENT_RESET,
__I40E_VIRTCHNL_OP_PENDING,
__I40E_RECOVERY_MODE,
__I40E_VF_RESETS_DISABLED, /* disable resets during i40e_remove */
/* This must be last as it determines the size of the BITMAP */
__I40E_STATE_SIZE__,
};

View File

@ -4010,8 +4010,16 @@ static irqreturn_t i40e_intr(int irq, void *data)
}
if (icr0 & I40E_PFINT_ICR0_VFLR_MASK) {
ena_mask &= ~I40E_PFINT_ICR0_ENA_VFLR_MASK;
set_bit(__I40E_VFLR_EVENT_PENDING, pf->state);
/* disable any further VFLR event notifications */
if (test_bit(__I40E_VF_RESETS_DISABLED, pf->state)) {
u32 reg = rd32(hw, I40E_PFINT_ICR0_ENA);
reg &= ~I40E_PFINT_ICR0_VFLR_MASK;
wr32(hw, I40E_PFINT_ICR0_ENA, reg);
} else {
ena_mask &= ~I40E_PFINT_ICR0_ENA_VFLR_MASK;
set_bit(__I40E_VFLR_EVENT_PENDING, pf->state);
}
}
if (icr0 & I40E_PFINT_ICR0_GRST_MASK) {
@ -15311,6 +15319,11 @@ static void i40e_remove(struct pci_dev *pdev)
while (test_bit(__I40E_RESET_RECOVERY_PENDING, pf->state))
usleep_range(1000, 2000);
if (pf->flags & I40E_FLAG_SRIOV_ENABLED) {
set_bit(__I40E_VF_RESETS_DISABLED, pf->state);
i40e_free_vfs(pf);
pf->flags &= ~I40E_FLAG_SRIOV_ENABLED;
}
/* no more scheduling of any task */
set_bit(__I40E_SUSPENDED, pf->state);
set_bit(__I40E_DOWN, pf->state);
@ -15337,11 +15350,6 @@ static void i40e_remove(struct pci_dev *pdev)
*/
i40e_notify_client_of_netdev_close(pf->vsi[pf->lan_vsi], false);
if (pf->flags & I40E_FLAG_SRIOV_ENABLED) {
i40e_free_vfs(pf);
pf->flags &= ~I40E_FLAG_SRIOV_ENABLED;
}
i40e_fdir_teardown(pf);
/* If there is a switch structure or any orphans, remove them.

View File

@ -1403,7 +1403,8 @@ static void i40e_cleanup_reset_vf(struct i40e_vf *vf)
* @vf: pointer to the VF structure
* @flr: VFLR was issued or not
*
* Returns true if the VF is reset, false otherwise.
* Returns true if the VF is in reset, resets successfully, or resets
* are disabled and false otherwise.
**/
bool i40e_reset_vf(struct i40e_vf *vf, bool flr)
{
@ -1413,11 +1414,14 @@ bool i40e_reset_vf(struct i40e_vf *vf, bool flr)
u32 reg;
int i;
if (test_bit(__I40E_VF_RESETS_DISABLED, pf->state))
return true;
/* If the VFs have been disabled, this means something else is
* resetting the VF, so we shouldn't continue.
*/
if (test_and_set_bit(__I40E_VF_DISABLE, pf->state))
return false;
return true;
i40e_trigger_vf_reset(vf, flr);
@ -1581,6 +1585,15 @@ void i40e_free_vfs(struct i40e_pf *pf)
i40e_notify_client_of_vf_enable(pf, 0);
/* Disable IOV before freeing resources. This lets any VF drivers
* running in the host get themselves cleaned up before we yank
* the carpet out from underneath their feet.
*/
if (!pci_vfs_assigned(pf->pdev))
pci_disable_sriov(pf->pdev);
else
dev_warn(&pf->pdev->dev, "VFs are assigned - not disabling SR-IOV\n");
/* Amortize wait time by stopping all VFs at the same time */
for (i = 0; i < pf->num_alloc_vfs; i++) {
if (test_bit(I40E_VF_STATE_INIT, &pf->vf[i].vf_states))
@ -1596,15 +1609,6 @@ void i40e_free_vfs(struct i40e_pf *pf)
i40e_vsi_wait_queues_disabled(pf->vsi[pf->vf[i].lan_vsi_idx]);
}
/* Disable IOV before freeing resources. This lets any VF drivers
* running in the host get themselves cleaned up before we yank
* the carpet out from underneath their feet.
*/
if (!pci_vfs_assigned(pf->pdev))
pci_disable_sriov(pf->pdev);
else
dev_warn(&pf->pdev->dev, "VFs are assigned - not disabling SR-IOV\n");
/* free up VF resources */
tmp = pf->num_alloc_vfs;
pf->num_alloc_vfs = 0;