drivers/net/ethernet/intel/iavf/iavf.h | 2 +- .../net/ethernet/intel/iavf/iavf_ethtool.c | 21 +++--- drivers/net/ethernet/intel/iavf/iavf_main.c | 72 +++++++------------ 3 files changed, 33 insertions(+), 62 deletions(-)
Three driver callbacks schedule a reset and wait for its completion:
ndo_change_mtu(), ethtool set_ringparam(), and ethtool set_channels().
Waiting for reset in ndo_change_mtu() and set_ringparam() was added by
commit c2ed2403f12c ("iavf: Wait for reset in callbacks which trigger
it") to fix a race condition where adding an interface to bonding
immediately after MTU or ring parameter change failed because the
interface was still in __RESETTING state. The same commit also added
waiting in iavf_set_priv_flags(), which was later removed by commit
53844673d555 ("iavf: kill "legacy-rx" for good").
Waiting in set_channels() was introduced earlier by commit 4e5e6b5d9d13
("iavf: Fix return of set the new channel count") to ensure the PF has
enough time to complete the VF reset when changing channel count, and to
return correct error codes to userspace.
Commit ef490bbb2267 ("iavf: Add net_shaper_ops support") added
net_shaper_ops to iavf, which required reset_task to use _locked NAPI
variants (napi_enable_locked, napi_disable_locked) that need the netdev
instance lock.
Later, commit 7e4d784f5810 ("net: hold netdev instance lock during
rtnetlink operations") and commit 2bcf4772e45a ("net: ethtool: try to
protect all callback with netdev instance lock") started holding the
netdev instance lock during ndo and ethtool callbacks for drivers with
net_shaper_ops.
Finally, commit 120f28a6f314 ("iavf: get rid of the crit lock")
replaced the driver's crit_lock with netdev_lock in reset_task, making
the deadlock manifest: the callback holds netdev_lock and waits for
reset_task, but reset_task needs the same lock:
Thread 1 (callback) Thread 2 (reset_task)
------------------- ---------------------
netdev_lock() [blocked on workqueue]
ndo_change_mtu() or ethtool op
iavf_schedule_reset()
iavf_wait_for_reset() iavf_reset_task()
waiting... netdev_lock() <- DEADLOCK
Fix this by extracting the reset logic from iavf_reset_task() into a new
iavf_reset_step() function that expects netdev_lock to be already held.
The three callbacks now call iavf_reset_step() directly instead of
scheduling the work and waiting, performing the reset synchronously in
the caller's context which already holds netdev_lock. This eliminates
both the deadlock and the need for iavf_wait_for_reset(), which is
removed.
The workqueue-based iavf_reset_task() becomes a thin wrapper that
acquires netdev_lock and calls iavf_reset_step(), preserving its use
for PF-initiated resets.
The callbacks may block for several seconds while iavf_reset_step()
polls hardware registers, but this is acceptable since netdev_lock is a
per-device mutex and only serializes operations on the same interface.
Fixes: 120f28a6f314 ("iavf: get rid of the crit lock")
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Signed-off-by: Petr Oros <poros@redhat.com>
---
drivers/net/ethernet/intel/iavf/iavf.h | 2 +-
.../net/ethernet/intel/iavf/iavf_ethtool.c | 21 +++---
drivers/net/ethernet/intel/iavf/iavf_main.c | 72 +++++++------------
3 files changed, 33 insertions(+), 62 deletions(-)
diff --git a/drivers/net/ethernet/intel/iavf/iavf.h b/drivers/net/ethernet/intel/iavf/iavf.h
index d552f912e8a947..0c3844b3ff1c86 100644
--- a/drivers/net/ethernet/intel/iavf/iavf.h
+++ b/drivers/net/ethernet/intel/iavf/iavf.h
@@ -625,5 +625,5 @@ void iavf_add_adv_rss_cfg(struct iavf_adapter *adapter);
void iavf_del_adv_rss_cfg(struct iavf_adapter *adapter);
struct iavf_mac_filter *iavf_add_filter(struct iavf_adapter *adapter,
const u8 *macaddr);
-int iavf_wait_for_reset(struct iavf_adapter *adapter);
+void iavf_reset_step(struct iavf_adapter *adapter);
#endif /* _IAVF_H_ */
diff --git a/drivers/net/ethernet/intel/iavf/iavf_ethtool.c b/drivers/net/ethernet/intel/iavf/iavf_ethtool.c
index 2cc21289a70779..9b0f47f9340942 100644
--- a/drivers/net/ethernet/intel/iavf/iavf_ethtool.c
+++ b/drivers/net/ethernet/intel/iavf/iavf_ethtool.c
@@ -492,7 +492,6 @@ static int iavf_set_ringparam(struct net_device *netdev,
{
struct iavf_adapter *adapter = netdev_priv(netdev);
u32 new_rx_count, new_tx_count;
- int ret = 0;
if ((ring->rx_mini_pending) || (ring->rx_jumbo_pending))
return -EINVAL;
@@ -537,13 +536,11 @@ static int iavf_set_ringparam(struct net_device *netdev,
}
if (netif_running(netdev)) {
- iavf_schedule_reset(adapter, IAVF_FLAG_RESET_NEEDED);
- ret = iavf_wait_for_reset(adapter);
- if (ret)
- netdev_warn(netdev, "Changing ring parameters timeout or interrupted waiting for reset");
+ adapter->flags |= IAVF_FLAG_RESET_NEEDED;
+ iavf_reset_step(adapter);
}
- return ret;
+ return 0;
}
/**
@@ -1723,7 +1720,6 @@ static int iavf_set_channels(struct net_device *netdev,
{
struct iavf_adapter *adapter = netdev_priv(netdev);
u32 num_req = ch->combined_count;
- int ret = 0;
if ((adapter->vf_res->vf_cap_flags & VIRTCHNL_VF_OFFLOAD_ADQ) &&
adapter->num_tc) {
@@ -1745,13 +1741,12 @@ static int iavf_set_channels(struct net_device *netdev,
adapter->num_req_queues = num_req;
adapter->flags |= IAVF_FLAG_REINIT_ITR_NEEDED;
- iavf_schedule_reset(adapter, IAVF_FLAG_RESET_NEEDED);
-
- ret = iavf_wait_for_reset(adapter);
- if (ret)
- netdev_warn(netdev, "Changing channel count timeout or interrupted waiting for reset");
+ if (netif_running(netdev)) {
+ adapter->flags |= IAVF_FLAG_RESET_NEEDED;
+ iavf_reset_step(adapter);
+ }
- return ret;
+ return 0;
}
/**
diff --git a/drivers/net/ethernet/intel/iavf/iavf_main.c b/drivers/net/ethernet/intel/iavf/iavf_main.c
index 8aa6e92c16431f..9c8d6125106f5a 100644
--- a/drivers/net/ethernet/intel/iavf/iavf_main.c
+++ b/drivers/net/ethernet/intel/iavf/iavf_main.c
@@ -185,31 +185,6 @@ static bool iavf_is_reset_in_progress(struct iavf_adapter *adapter)
return false;
}
-/**
- * iavf_wait_for_reset - Wait for reset to finish.
- * @adapter: board private structure
- *
- * Returns 0 if reset finished successfully, negative on timeout or interrupt.
- */
-int iavf_wait_for_reset(struct iavf_adapter *adapter)
-{
- int ret = wait_event_interruptible_timeout(adapter->reset_waitqueue,
- !iavf_is_reset_in_progress(adapter),
- msecs_to_jiffies(5000));
-
- /* If ret < 0 then it means wait was interrupted.
- * If ret == 0 then it means we got a timeout while waiting
- * for reset to finish.
- * If ret > 0 it means reset has finished.
- */
- if (ret > 0)
- return 0;
- else if (ret < 0)
- return -EINTR;
- else
- return -EBUSY;
-}
-
/**
* iavf_allocate_dma_mem_d - OS specific memory alloc for shared code
* @hw: pointer to the HW structure
@@ -3100,18 +3075,16 @@ static void iavf_reconfig_qs_bw(struct iavf_adapter *adapter)
}
/**
- * iavf_reset_task - Call-back task to handle hardware reset
- * @work: pointer to work_struct
+ * iavf_reset_step - Perform the VF reset sequence
+ * @adapter: board private structure
*
- * During reset we need to shut down and reinitialize the admin queue
- * before we can use it to communicate with the PF again. We also clear
- * and reinit the rings because that context is lost as well.
- **/
-static void iavf_reset_task(struct work_struct *work)
+ * Requests a reset from PF, polls for completion, and reconfigures
+ * the driver. Caller must hold the netdev instance lock.
+ *
+ * This can sleep for several seconds while polling HW registers.
+ */
+void iavf_reset_step(struct iavf_adapter *adapter)
{
- struct iavf_adapter *adapter = container_of(work,
- struct iavf_adapter,
- reset_task);
struct virtchnl_vf_resource *vfres = adapter->vf_res;
struct net_device *netdev = adapter->netdev;
struct iavf_hw *hw = &adapter->hw;
@@ -3122,7 +3095,7 @@ static void iavf_reset_task(struct work_struct *work)
int i = 0, err;
bool running;
- netdev_lock(netdev);
+ netdev_assert_locked(netdev);
iavf_misc_irq_disable(adapter);
if (adapter->flags & IAVF_FLAG_RESET_NEEDED) {
@@ -3167,7 +3140,6 @@ 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);
- netdev_unlock(netdev);
return; /* Do not attempt to reinit. It's dead, Jim. */
}
@@ -3179,7 +3151,6 @@ static void iavf_reset_task(struct work_struct *work)
iavf_startup(adapter);
queue_delayed_work(adapter->wq, &adapter->watchdog_task,
msecs_to_jiffies(30));
- netdev_unlock(netdev);
return;
}
@@ -3321,7 +3292,6 @@ static void iavf_reset_task(struct work_struct *work)
adapter->flags &= ~IAVF_FLAG_REINIT_ITR_NEEDED;
wake_up(&adapter->reset_waitqueue);
- netdev_unlock(netdev);
return;
reset_err:
@@ -3331,10 +3301,21 @@ static void iavf_reset_task(struct work_struct *work)
}
iavf_disable_vf(adapter);
- netdev_unlock(netdev);
dev_err(&adapter->pdev->dev, "failed to allocate resources during reinit\n");
}
+static void iavf_reset_task(struct work_struct *work)
+{
+ struct iavf_adapter *adapter = container_of(work,
+ struct iavf_adapter,
+ reset_task);
+ struct net_device *netdev = adapter->netdev;
+
+ netdev_lock(netdev);
+ iavf_reset_step(adapter);
+ netdev_unlock(netdev);
+}
+
/**
* iavf_adminq_task - worker thread to clean the admin queue
* @work: pointer to work_struct containing our data
@@ -4600,22 +4581,17 @@ static int iavf_close(struct net_device *netdev)
static int iavf_change_mtu(struct net_device *netdev, int new_mtu)
{
struct iavf_adapter *adapter = netdev_priv(netdev);
- int ret = 0;
netdev_dbg(netdev, "changing MTU from %d to %d\n",
netdev->mtu, new_mtu);
WRITE_ONCE(netdev->mtu, new_mtu);
if (netif_running(netdev)) {
- iavf_schedule_reset(adapter, IAVF_FLAG_RESET_NEEDED);
- ret = iavf_wait_for_reset(adapter);
- if (ret < 0)
- netdev_warn(netdev, "MTU change interrupted waiting for reset");
- else if (ret)
- netdev_warn(netdev, "MTU change timed out waiting for reset");
+ adapter->flags |= IAVF_FLAG_RESET_NEEDED;
+ iavf_reset_step(adapter);
}
- return ret;
+ return 0;
}
/**
--
2.52.0© 2016 - 2026 Red Hat, Inc.