nagai/kernel
3
0
Fork 0
forked from mirrors/linux
Code Pull requests Activity

iavf: get rid of the crit lock

Browse source 
Get rid of the crit lock.
That frees us from the error prone logic of try_locks.

Thanks to netdev_lock() by Jakub it is now easy, and in most cases we were
protected by it already - replace crit lock by netdev lock when it was not
the case.

Lockdep reports that we should cancel the work under crit_lock [splat1],
and that was the scheme we have mostly followed since [1] by Slawomir.
But when that is done we still got into deadlocks [splat2]. So instead
we should look at the bigger problem, namely "weird locking/scheduling"
of the iavf. The first step to fix that is to remove the crit lock.
I will followup with a -next series that simplifies scheduling/tasks.

Cancel the work without netdev lock (weird unlock+lock scheme),
to fix the [splat2] (which would be totally ugly if we would kept
the crit lock).

Extend protected part of iavf_watchdog_task() to include scheduling
more work.

Note that the removed comment in iavf_reset_task() was misplaced,
it belonged to inside of the removed if condition, so it's gone now.

[splat1] - w/o this patch - The deadlock during VF removal:
     WARNING: possible circular locking dependency detected
     sh/3825 is trying to acquire lock:
      ((work_completion)(&(&adapter->watchdog_task)->work)){+.+.}-{0:0}, at: start_flush_work+0x1a1/0x470
          but task is already holding lock:
      (&adapter->crit_lock){+.+.}-{4:4}, at: iavf_remove+0xd1/0x690 [iavf]
          which lock already depends on the new lock.

[splat2] - when cancelling work under crit lock, w/o this series,
	   see [2] for the band aid attempt
    WARNING: possible circular locking dependency detected
    sh/3550 is trying to acquire lock:
    ((wq_completion)iavf){+.+.}-{0:0}, at: touch_wq_lockdep_map+0x26/0x90
        but task is already holding lock:
    (&dev->lock){+.+.}-{4:4}, at: iavf_remove+0xa6/0x6e0 [iavf]
        which lock already depends on the new lock.

[1] fc2e6b3b13 ("iavf: Rework mutexes for better synchronisation")
[2] https://github.com/pkitszel/linux/commit/52dddbfc2bb60294083f5711a158a

Fixes: d1639a1731 ("iavf: fix a deadlock caused by rtnl and driver's lock circular dependencies")
Signed-off-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
This commit is contained in:
Przemek Kitszel 2025-04-04 12:23:21 +02:00 committed by Tony Nguyen
parent 05702b5c94
commit 120f28a6f3
3 changed files with 38 additions and 151 deletions
repo.diff.show_diff_stats Download patch file Download diff file Expand all files Collapse all files

1
drivers/net/ethernet/intel/iavf/iavf.h
View file

@ -268,7 +268,6 @@ struct iavf_adapter {
struct list_head vlan_filter_list;
int num_vlan_filters;
struct list_head mac_filter_list;
struct mutex crit_lock;
/* Lock to protect accesses to MAC and VLAN lists */
spinlock_t mac_vlan_list_lock;
char misc_vector_name[IFNAMSIZ + 9];

23
drivers/net/ethernet/intel/iavf/iavf_ethtool.c
View file

@ -1258,7 +1258,6 @@ static int iavf_add_fdir_ethtool(struct iavf_adapter *adapter, struct ethtool_rx
{
struct ethtool_rx_flow_spec *fsp = &cmd->fs;
struct iavf_fdir_fltr *fltr;
int count = 50;
int err;
netdev_assert_locked(adapter->netdev);
@ -1281,14 +1280,6 @@ static int iavf_add_fdir_ethtool(struct iavf_adapter *adapter, struct ethtool_rx
if (!fltr)
return -ENOMEM;
while (!mutex_trylock(&adapter->crit_lock)) {
if (--count == 0) {
kfree(fltr);
return -EINVAL;
}
udelay(1);
}
err = iavf_add_fdir_fltr_info(adapter, fsp, fltr);
if (!err)
err = iavf_fdir_add_fltr(adapter, fltr);
@ -1296,7 +1287,6 @@ static int iavf_add_fdir_ethtool(struct iavf_adapter *adapter, struct ethtool_rx
if (err)
kfree(fltr);
mutex_unlock(&adapter->crit_lock);
return err;
}
@ -1439,9 +1429,9 @@ iavf_set_adv_rss_hash_opt(struct iavf_adapter *adapter,
{
struct iavf_adv_rss *rss_old, *rss_new;
bool rss_new_add = false;
int count = 50, err = 0;
bool symm = false;
u64 hash_flds;
int err = 0;
u32 hdrs;
netdev_assert_locked(adapter->netdev);
@ -1469,15 +1459,6 @@ iavf_set_adv_rss_hash_opt(struct iavf_adapter *adapter,
return -EINVAL;
}
while (!mutex_trylock(&adapter->crit_lock)) {
if (--count == 0) {
kfree(rss_new);
return -EINVAL;
}
udelay(1);
}
spin_lock_bh(&adapter->adv_rss_lock);
rss_old = iavf_find_adv_rss_cfg_by_hdrs(adapter, hdrs);
if (rss_old) {
@ -1506,8 +1487,6 @@ iavf_set_adv_rss_hash_opt(struct iavf_adapter *adapter,
if (!err)
iavf_schedule_aq_request(adapter, IAVF_FLAG_AQ_ADD_ADV_RSS_CFG);
mutex_unlock(&adapter->crit_lock);
if (!rss_new_add)
kfree(rss_new);

165
drivers/net/ethernet/intel/iavf/iavf_main.c
View file

@ -1287,9 +1287,7 @@ static void iavf_configure(struct iavf_adapter *adapter)
/**
* iavf_up_complete - Finish the last steps of bringing up a connection
* @adapter: board private structure
*
* Expects to be called while holding crit_lock.
**/
*/
static void iavf_up_complete(struct iavf_adapter *adapter)
{
netdev_assert_locked(adapter->netdev);
@ -1412,9 +1410,7 @@ static void iavf_clear_adv_rss_conf(struct iavf_adapter *adapter)
/**
* iavf_down - Shutdown the connection processing
* @adapter: board private structure
*
* Expects to be called while holding crit_lock.
**/
*/
void iavf_down(struct iavf_adapter *adapter)
{
struct net_device *netdev = adapter->netdev;
@ -2029,22 +2025,21 @@ static int iavf_reinit_interrupt_scheme(struct iavf_adapter *adapter, bool runni
* iavf_finish_config - do all netdev work that needs RTNL
* @work: our work_struct
*
* Do work that needs both RTNL and crit_lock.
**/
* Do work that needs RTNL.
*/
static void iavf_finish_config(struct work_struct *work)
{
struct iavf_adapter *adapter;
bool locks_released = false;
bool netdev_released = false;
int pairs, err;
adapter = container_of(work, struct iavf_adapter, finish_config);
/* Always take RTNL first to prevent circular lock dependency;
* The dev->lock is needed to update the queue number
* the dev->lock (== netdev lock) is needed to update the queue number.
*/
rtnl_lock();
netdev_lock(adapter->netdev);
mutex_lock(&adapter->crit_lock);
if ((adapter->flags & IAVF_FLAG_SETUP_NETDEV_FEATURES) &&
adapter->netdev->reg_state == NETREG_REGISTERED &&
@ -2063,22 +2058,21 @@ static void iavf_finish_config(struct work_struct *work)
netif_set_real_num_tx_queues(adapter->netdev, pairs);
if (adapter->netdev->reg_state != NETREG_REGISTERED) {
mutex_unlock(&adapter->crit_lock);
netdev_unlock(adapter->netdev);
locks_released = true;
netdev_released = true;
err = register_netdevice(adapter->netdev);
if (err) {
dev_err(&adapter->pdev->dev, "Unable to register netdev (%d)\n",
err);
/* go back and try again.*/
mutex_lock(&adapter->crit_lock);
netdev_lock(adapter->netdev);
iavf_free_rss(adapter);
iavf_free_misc_irq(adapter);
iavf_reset_interrupt_capability(adapter);
iavf_change_state(adapter,
__IAVF_INIT_CONFIG_ADAPTER);
mutex_unlock(&adapter->crit_lock);
netdev_unlock(adapter->netdev);
goto out;
}
}
@ -2094,10 +2088,8 @@ static void iavf_finish_config(struct work_struct *work)
}
out:
if (!locks_released) {
mutex_unlock(&adapter->crit_lock);
if (!netdev_released)
netdev_unlock(adapter->netdev);
}
rtnl_unlock();
}
@ -2924,7 +2916,6 @@ static int iavf_watchdog_step(struct iavf_adapter *adapter)
u32 reg_val;
netdev_assert_locked(adapter->netdev);
lockdep_assert_held(&adapter->crit_lock);
if (adapter->flags & IAVF_FLAG_PF_COMMS_FAILED)
iavf_change_state(adapter, __IAVF_COMM_FAILED);
@ -3044,22 +3035,7 @@ static void iavf_watchdog_task(struct work_struct *work)
int msec_delay;
netdev_lock(netdev);
if (!mutex_trylock(&adapter->crit_lock)) {
if (adapter->state == __IAVF_REMOVE) {
netdev_unlock(netdev);
return;
}
msec_delay = 20;
goto restart_watchdog;
}
msec_delay = iavf_watchdog_step(adapter);
mutex_unlock(&adapter->crit_lock);
restart_watchdog:
netdev_unlock(netdev);
/* note that we schedule a different task */
if (adapter->state >= __IAVF_DOWN)
queue_work(adapter->wq, &adapter->adminq_task);
@ -3067,6 +3043,7 @@ static void iavf_watchdog_task(struct work_struct *work)
if (msec_delay != IAVF_NO_RESCHED)
queue_delayed_work(adapter->wq, &adapter->watchdog_task,
msecs_to_jiffies(msec_delay));
netdev_unlock(netdev);
}
/**
@ -3074,8 +3051,7 @@ static void iavf_watchdog_task(struct work_struct *work)
* @adapter: board private structure
*
* Set communication failed flag and free all resources.
* NOTE: This function is expected to be called with crit_lock being held.
**/
*/
static void iavf_disable_vf(struct iavf_adapter *adapter)
{
struct iavf_mac_filter *f, *ftmp;
@ -3183,17 +3159,7 @@ static void iavf_reset_task(struct work_struct *work)
int i = 0, err;
bool running;
/* When device is being removed it doesn't make sense to run the reset
* task, just return in such a case.
*/
netdev_lock(netdev);
if (!mutex_trylock(&adapter->crit_lock)) {
if (adapter->state != __IAVF_REMOVE)
queue_work(adapter->wq, &adapter->reset_task);
netdev_unlock(netdev);
return;
}
iavf_misc_irq_disable(adapter);
if (adapter->flags & IAVF_FLAG_RESET_NEEDED) {
@ -3238,7 +3204,6 @@ static void iavf_reset_task(struct work_struct *work)
dev_err(&adapter->pdev->dev, "Reset never finished (%x)\n",
reg_val);
iavf_disable_vf(adapter);
mutex_unlock(&adapter->crit_lock);
netdev_unlock(netdev);
return; /* Do not attempt to reinit. It's dead, Jim. */
}
@ -3382,7 +3347,6 @@ static void iavf_reset_task(struct work_struct *work)
adapter->flags &= ~IAVF_FLAG_REINIT_ITR_NEEDED;
wake_up(&adapter->reset_waitqueue);
mutex_unlock(&adapter->crit_lock);
netdev_unlock(netdev);
return;
@ -3393,7 +3357,6 @@ static void iavf_reset_task(struct work_struct *work)
}
iavf_disable_vf(adapter);
mutex_unlock(&adapter->crit_lock);
netdev_unlock(netdev);
dev_err(&adapter->pdev->dev, "failed to allocate resources during reinit\n");
}
@ -3406,6 +3369,7 @@ static void iavf_adminq_task(struct work_struct *work)
{
struct iavf_adapter *adapter =
container_of(work, struct iavf_adapter, adminq_task);
struct net_device *netdev = adapter->netdev;
struct iavf_hw *hw = &adapter->hw;
struct iavf_arq_event_info event;
enum virtchnl_ops v_op;
@ -3413,13 +3377,7 @@ static void iavf_adminq_task(struct work_struct *work)
u32 val, oldval;
u16 pending;
if (!mutex_trylock(&adapter->crit_lock)) {
if (adapter->state == __IAVF_REMOVE)
return;
queue_work(adapter->wq, &adapter->adminq_task);
goto out;
}
netdev_lock(netdev);
if (adapter->flags & IAVF_FLAG_PF_COMMS_FAILED)
goto unlock;
@ -3486,8 +3444,7 @@ static void iavf_adminq_task(struct work_struct *work)
freedom:
kfree(event.msg_buf);
unlock:
mutex_unlock(&adapter->crit_lock);
out:
netdev_unlock(netdev);
/* re-enable Admin queue interrupt cause */
iavf_misc_irq_enable(adapter);
}
@ -4180,8 +4137,8 @@ static int iavf_configure_clsflower(struct iavf_adapter *adapter,
struct flow_cls_offload *cls_flower)
{
int tc = tc_classid_to_hwtc(adapter->netdev, cls_flower->classid);
struct iavf_cloud_filter *filter = NULL;
int err = -EINVAL, count = 50;
struct iavf_cloud_filter *filter;
int err;
if (tc < 0) {
dev_err(&adapter->pdev->dev, "Invalid traffic class\n");
@ -4191,17 +4148,10 @@ static int iavf_configure_clsflower(struct iavf_adapter *adapter,
filter = kzalloc(sizeof(*filter), GFP_KERNEL);
if (!filter)
return -ENOMEM;
while (!mutex_trylock(&adapter->crit_lock)) {
if (--count == 0) {
kfree(filter);
return err;
}
udelay(1);
}
filter->cookie = cls_flower->cookie;
netdev_lock(adapter->netdev);
/* bail out here if filter already exists */
spin_lock_bh(&adapter->cloud_filter_list_lock);
if (iavf_find_cf(adapter, &cls_flower->cookie)) {
@ -4235,7 +4185,7 @@ static int iavf_configure_clsflower(struct iavf_adapter *adapter,
if (err)
kfree(filter);
mutex_unlock(&adapter->crit_lock);
netdev_unlock(adapter->netdev);
return err;
}
@ -4539,28 +4489,13 @@ static int iavf_open(struct net_device *netdev)
return -EIO;
}
while (!mutex_trylock(&adapter->crit_lock)) {
/* If we are in __IAVF_INIT_CONFIG_ADAPTER state the crit_lock
* is already taken and iavf_open is called from an upper
* device's notifier reacting on NETDEV_REGISTER event.
* We have to leave here to avoid dead lock.
*/
if (adapter->state == __IAVF_INIT_CONFIG_ADAPTER)
return -EBUSY;
usleep_range(500, 1000);
}
if (adapter->state != __IAVF_DOWN) {
err = -EBUSY;
goto err_unlock;
}
if (adapter->state != __IAVF_DOWN)
return -EBUSY;
if (adapter->state == __IAVF_RUNNING &&
!test_bit(__IAVF_VSI_DOWN, adapter->vsi.state)) {
dev_dbg(&adapter->pdev->dev, "VF is already open.\n");
err = 0;
goto err_unlock;
return 0;
}
/* allocate transmit descriptors */
@ -4579,9 +4514,7 @@ static int iavf_open(struct net_device *netdev)
goto err_req_irq;
spin_lock_bh(&adapter->mac_vlan_list_lock);
iavf_add_filter(adapter, adapter->hw.mac.addr);
spin_unlock_bh(&adapter->mac_vlan_list_lock);
/* Restore filters that were removed with IFF_DOWN */
@ -4594,8 +4527,6 @@ static int iavf_open(struct net_device *netdev)
iavf_irq_enable(adapter, true);
mutex_unlock(&adapter->crit_lock);
return 0;
err_req_irq:
@ -4605,8 +4536,6 @@ static int iavf_open(struct net_device *netdev)
iavf_free_all_rx_resources(adapter);
err_setup_tx:
iavf_free_all_tx_resources(adapter);
err_unlock:
mutex_unlock(&adapter->crit_lock);
return err;
}
@ -4630,12 +4559,8 @@ static int iavf_close(struct net_device *netdev)
netdev_assert_locked(netdev);
mutex_lock(&adapter->crit_lock);
if (adapter->state <= __IAVF_DOWN_PENDING) {
mutex_unlock(&adapter->crit_lock);
if (adapter->state <= __IAVF_DOWN_PENDING)
return 0;
}
set_bit(__IAVF_VSI_DOWN, adapter->vsi.state);
/* We cannot send IAVF_FLAG_AQ_GET_OFFLOAD_VLAN_V2_CAPS before
@ -4666,7 +4591,6 @@ static int iavf_close(struct net_device *netdev)
iavf_change_state(adapter, __IAVF_DOWN_PENDING);
iavf_free_traffic_irqs(adapter);
mutex_unlock(&adapter->crit_lock);
netdev_unlock(netdev);
/* We explicitly don't free resources here because the hardware is
@ -4685,11 +4609,10 @@ static int iavf_close(struct net_device *netdev)
msecs_to_jiffies(500));
if (!status)
netdev_warn(netdev, "Device resources not yet released\n");
netdev_lock(netdev);
mutex_lock(&adapter->crit_lock);
adapter->aq_required |= aq_to_restore;
mutex_unlock(&adapter->crit_lock);
return 0;
}
@ -5198,17 +5121,16 @@ iavf_shaper_set(struct net_shaper_binding *binding,
struct iavf_adapter *adapter = netdev_priv(binding->netdev);
const struct net_shaper_handle *handle = &shaper->handle;
struct iavf_ring *tx_ring;
int ret = 0;
int ret;
netdev_assert_locked(adapter->netdev);
mutex_lock(&adapter->crit_lock);
if (handle->id >= adapter->num_active_queues)
goto unlock;
return 0;
ret = iavf_verify_shaper(binding, shaper, extack);
if (ret)
goto unlock;
return ret;
tx_ring = &adapter->tx_rings[handle->id];
@ -5218,9 +5140,7 @@ iavf_shaper_set(struct net_shaper_binding *binding,
adapter->aq_required |= IAVF_FLAG_AQ_CONFIGURE_QUEUES_BW;
unlock:
mutex_unlock(&adapter->crit_lock);
return ret;
return 0;
}
static int iavf_shaper_del(struct net_shaper_binding *binding,
@ -5232,9 +5152,8 @@ static int iavf_shaper_del(struct net_shaper_binding *binding,
netdev_assert_locked(adapter->netdev);
mutex_lock(&adapter->crit_lock);
if (handle->id >= adapter->num_active_queues)
goto unlock;
return 0;
tx_ring = &adapter->tx_rings[handle->id];
tx_ring->q_shaper.bw_min = 0;
@ -5243,8 +5162,6 @@ static int iavf_shaper_del(struct net_shaper_binding *binding,
adapter->aq_required |= IAVF_FLAG_AQ_CONFIGURE_QUEUES_BW;
unlock:
mutex_unlock(&adapter->crit_lock);
return 0;
}
@ -5505,10 +5422,6 @@ static int iavf_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
goto err_alloc_qos_cap;
}
/* set up the locks for the AQ, do this only once in probe
* and destroy them only once in remove
*/
mutex_init(&adapter->crit_lock);
mutex_init(&hw->aq.asq_mutex);
mutex_init(&hw->aq.arq_mutex);
@ -5578,9 +5491,7 @@ static int iavf_suspend(struct device *dev_d)
running = netif_running(netdev);
if (running)
rtnl_lock();
netdev_lock(netdev);
mutex_lock(&adapter->crit_lock);
if (running)
iavf_down(adapter);
@ -5588,7 +5499,6 @@ static int iavf_suspend(struct device *dev_d)
iavf_free_misc_irq(adapter);
iavf_reset_interrupt_capability(adapter);
mutex_unlock(&adapter->crit_lock);
netdev_unlock(netdev);
if (running)
rtnl_unlock();
@ -5668,20 +5578,20 @@ static void iavf_remove(struct pci_dev *pdev)
* There are flows where register/unregister netdev may race.
*/
while (1) {
mutex_lock(&adapter->crit_lock);
netdev_lock(netdev);
if (adapter->state == __IAVF_RUNNING ||
adapter->state == __IAVF_DOWN ||
adapter->state == __IAVF_INIT_FAILED) {
mutex_unlock(&adapter->crit_lock);
netdev_unlock(netdev);
break;
}
/* Simply return if we already went through iavf_shutdown */
if (adapter->state == __IAVF_REMOVE) {
mutex_unlock(&adapter->crit_lock);
netdev_unlock(netdev);
return;
}
mutex_unlock(&adapter->crit_lock);
netdev_unlock(netdev);
usleep_range(500, 1000);
}
cancel_delayed_work_sync(&adapter->watchdog_task);
@ -5691,7 +5601,6 @@ static void iavf_remove(struct pci_dev *pdev)
unregister_netdev(netdev);
netdev_lock(netdev);
mutex_lock(&adapter->crit_lock);
dev_info(&adapter->pdev->dev, "Removing device\n");
iavf_change_state(adapter, __IAVF_REMOVE);
@ -5707,9 +5616,11 @@ static void iavf_remove(struct pci_dev *pdev)
iavf_misc_irq_disable(adapter);
/* Shut down all the garbage mashers on the detention level */
netdev_unlock(netdev);
cancel_work_sync(&adapter->reset_task);
cancel_delayed_work_sync(&adapter->watchdog_task);
cancel_work_sync(&adapter->adminq_task);
netdev_lock(netdev);
adapter->aq_required = 0;
adapter->flags &= ~IAVF_FLAG_REINIT_ITR_NEEDED;
@ -5727,8 +5638,6 @@ static void iavf_remove(struct pci_dev *pdev)
/* destroy the locks only once, here */
mutex_destroy(&hw->aq.arq_mutex);
mutex_destroy(&hw->aq.asq_mutex);
mutex_unlock(&adapter->crit_lock);
mutex_destroy(&adapter->crit_lock);
netdev_unlock(netdev);
iounmap(hw->hw_addr);