idpf: skip getting/setting ring params if vport is NULL during HW reset
Subsystems (3)
PATCH 1/5 idpf: skip getting/setting ring params if vport is NULL during HW reset
When an idpf HW reset is triggered, it clears the vport but does not clear the netdev held by vport:
// In idpf_vport_dealloc() called by idpf_init_hard_reset(),
// idpf_init_hard_reset() sets IDPF_HR_RESET_IN_PROG, so
// idpf_decfg_netdev() doesn't get called.
if (!test_bit(IDPF_HR_RESET_IN_PROG, adapter->flags))
idpf_decfg_netdev(vport);
// idpf_decfg_netdev() would clear netdev but it isn't called:
unregister_netdev(vport->netdev);
free_netdev(vport->netdev);
vport->netdev = NULL;
// Later in idpf_init_hard_reset(), the vport is cleared:
kfree(adapter->vports);
adapter->vports = NULL;During an idpf HW reset, when "ethtool -g/-G" is called on the netdev, the vport associated with the netdev is NULL, and so a kernel panic would happen:
[ 513.185327] BUG: kernel NULL pointer dereference, address: 0000000000000038 ... [ 513.232756] RIP: 0010:idpf_get_ringparam+0x45/0x80
This can be reproduced reliably by injecting a TX timeout to cause an idpf HW reset, and injecting a virtchnl error to cause the HW reset to fail and retry, while calling "ethtool -g/-G" on the netdev at the same time.
With this patch applied, we see the following error but no kernel panics anymore:
[ 476.323630] idpf 0000:05:00.0 eth1: failed to get ring params due to no vport in netdev
--- drivers/net/ethernet/intel/idpf/idpf_ethtool.c | 12 ++++++++++++ 1 file changed, 12 insertions(+)
drivers/net/ethernet/intel/idpf/idpf_ethtool.c +10 -0
@@ -639,6 +638,10 @@ static void idpf_get_ringparam(struct net_device *netdev,idpf_vport_ctrl_lock (netdev); vport =idpf_netdev_to_vport (netdev); +if (!vport) { +netdev_err (netdev,"failed to get ring params due to no vport in netdev\n" ); + goto unlock; + } ring->rx_max_pending =IDPF_MAX_RXQ_DESC ; ring->tx_max_pending =IDPF_MAX_TXQ_DESC ;
@@ -647,6 +651,7 @@ static void idpf_get_ringparam(struct net_device *netdev, kring->tcp_data_split =idpf_vport_get_hsplit (vport); +unlock:idpf_vport_ctrl_unlock (netdev); }
@@ -673,6 +674,11 @@ static int idpf_set_ringparam(struct net_device *netdev,idpf_vport_ctrl_lock (netdev); vport =idpf_netdev_to_vport (netdev); +if (!vport) { +netdev_err (netdev,"ring params not changed due to no vport in netdev\n" ); + err = -EFAULT ; + goto unlock_mutex; + } idx = vport->idx;
Signature · 1 line
2.52.0.351.gbe84eed79e-goog
PATCH 2/5 idpf: skip changing MTU if vport is NULL during HW reset
When an idpf HW reset is triggered, it clears the vport but does not clear the netdev held by vport:
// In idpf_vport_dealloc() called by idpf_init_hard_reset(),
// idpf_init_hard_reset() sets IDPF_HR_RESET_IN_PROG, so
// idpf_decfg_netdev() doesn't get called.
if (!test_bit(IDPF_HR_RESET_IN_PROG, adapter->flags))
idpf_decfg_netdev(vport);
// idpf_decfg_netdev() would clear netdev but it isn't called:
unregister_netdev(vport->netdev);
free_netdev(vport->netdev);
vport->netdev = NULL;
// Later in idpf_init_hard_reset(), the vport is cleared:
kfree(adapter->vports);
adapter->vports = NULL;During an idpf HW reset, when userspace changes the MTU of the netdev, the vport associated with the netdev is NULL, and so a kernel panic would happen:
[ 2081.955742] BUG: kernel NULL pointer dereference, address: 0000000000000068 ... [ 2082.002739] RIP: 0010:idpf_initiate_soft_reset+0x19/0x190
This can be reproduced reliably by injecting a TX timeout to cause an idpf HW reset, and injecting a virtchnl error to cause the HW reset to fail and retry, while changing the MTU of the netdev in userspace.
With this patch applied, we see the following error but no kernel panics anymore:
[ 304.291346] idpf 0000:05:00.0 eth1: mtu not changed due to no vport innetdev
RTNETLINK answers: Bad address
--- drivers/net/ethernet/intel/idpf/idpf_lib.c | 7 +++++++ 1 file changed, 7 insertions(+)
drivers/net/ethernet/intel/idpf/idpf_lib.c +6 -0
@@ -2328,11 +2327,17 @@ static int idpf_change_mtu(struct net_device *netdev, int new_mtu)idpf_vport_ctrl_lock (netdev); vport =idpf_netdev_to_vport (netdev); +if (!vport) { +netdev_err (netdev,"mtu not changed due to no vport in netdev\n" ); + err = -EFAULT ; + goto unlock; + }WRITE_ONCE (netdev->mtu, new_mtu); err =idpf_initiate_soft_reset (vport,IDPF_SR_MTU_CHANGE ); +unlock:idpf_vport_ctrl_unlock (netdev);return err;
Signature · 1 line
2.52.0.351.gbe84eed79e-goog
PATCH 3/5 idpf: skip getting RX flow rules if vport is NULL during HW reset
In reply to Li LiWhen an idpf HW reset is triggered, it clears the vport but does not clear the netdev held by vport:
// In idpf_vport_dealloc() called by idpf_init_hard_reset(),
// idpf_init_hard_reset() sets IDPF_HR_RESET_IN_PROG, so
// idpf_decfg_netdev() doesn't get called.
if (!test_bit(IDPF_HR_RESET_IN_PROG, adapter->flags))
idpf_decfg_netdev(vport);
// idpf_decfg_netdev() would clear netdev but it isn't called:
unregister_netdev(vport->netdev);
free_netdev(vport->netdev);
vport->netdev = NULL;
// Later in idpf_init_hard_reset(), the vport is cleared:
kfree(adapter->vports);
adapter->vports = NULL;During an idpf HW reset, when userspace gets RX flow classification rules of the netdev, the vport associated with the netdev is NULL, and so a kernel panic would happen:
[ 1466.308592] BUG: kernel NULL pointer dereference, address: 0000000000000032 ... [ 1466.356222] RIP: 0010:idpf_get_rxnfc+0x3b/0x70
This can be reproduced reliably by injecting a TX timeout to cause an idpf HW reset, and injecting a virtchnl error to cause the HW reset to fail and retry, while running "ethtool -n" in userspace.
With this patch applied, we see the following error but no kernel panics anymore:
[ 312.476576] idpf 0000:05:00.0 eth1: failed to get rules due to no vport in netdev
Cannot get RX rings: Bad address
rxclass: Cannot get RX class rule count: Bad address RX classification rule retrieval failed
--- drivers/net/ethernet/intel/idpf/idpf_ethtool.c | 7 +++++++ 1 file changed, 7 insertions(+)
drivers/net/ethernet/intel/idpf/idpf_ethtool.c +6 -0
@@ -45,6 +44,11 @@ static int idpf_get_rxnfc(struct net_device *netdev, struct ethtool_rxnfc *cmd,idpf_vport_ctrl_lock (netdev); vport =idpf_netdev_to_vport (netdev); +if (!vport) { +netdev_err (netdev,"failed to get rules due to no vport in netdev\n" ); + err = -EFAULT ; + goto unlock; + } vport_config = np->adapter->vport_config[np->vport_idx]; user_config = &vport_config->user_config;
@@ -85,6 +90,7 @@ static int idpf_get_rxnfc(struct net_device *netdev, struct ethtool_rxnfc *cmd,break ; } +unlock:idpf_vport_ctrl_unlock (netdev);return err;
Signature · 1 line
2.52.0.351.gbe84eed79e-goog
RE: [Intel-wired-lan] [PATCH 3/5] idpf: skip getting RX flow rules if vport is NULL during HW reset
Quoted 86 lines
-----Original Message-----
From: Intel-wired-lan <[email protected]> On Behalf Of Li Li via Intel-wired-lan
Sent: Wednesday, January 7, 2026 2:05 AM
To: Nguyen, Anthony L <[email protected]>; Kitszel, Przemyslaw <[email protected]>; David S. Miller <[email protected]>; Jakub Kicinski <[email protected]>; Eric Dumazet <[email protected]>; [email protected]
Cc: [email protected]; [email protected]; David Decotigny <[email protected]>; Singhai, Anjali <[email protected]>; Samudrala, Sridhar <[email protected]>; Brian Vazquez <[email protected]>; Li Li <[email protected]>; Tantilov, Emil S <[email protected]>
Subject: [Intel-wired-lan] [PATCH 3/5] idpf: skip getting RX flow rules if vport is NULL during HW resetWhen an idpf HW reset is triggered, it clears the vport but does not clear the netdev held by vport:
// In idpf_vport_dealloc() called by idpf_init_hard_reset(), // idpf_init_hard_reset() sets IDPF_HR_RESET_IN_PROG, so // idpf_decfg_netdev() doesn't get called. if (!test_bit(IDPF_HR_RESET_IN_PROG, adapter->flags)) idpf_decfg_netdev(vport); // idpf_decfg_netdev() would clear netdev but it isn't called: unregister_netdev(vport->netdev); free_netdev(vport->netdev); vport->netdev = NULL; // Later in idpf_init_hard_reset(), the vport is cleared: kfree(adapter->vports); adapter->vports = NULL;During an idpf HW reset, when userspace gets RX flow classification rules of the netdev, the vport associated with the netdev is NULL, and so a kernel panic would happen:
[ 1466.308592] BUG: kernel NULL pointer dereference, address: 0000000000000032 ... [ 1466.356222] RIP: 0010:idpf_get_rxnfc+0x3b/0x70This can be reproduced reliably by injecting a TX timeout to cause an idpf HW reset, and injecting a virtchnl error to cause the HW reset to fail and retry, while running "ethtool -n" in userspace.
With this patch applied, we see the following error but no kernel panics anymore:
[ 312.476576] idpf 0000:05:00.0 eth1: failed to get rules due to novport in netdev Cannot get RX rings: Bad address
rxclass: Cannot get RX class rule count: Bad address RX classification rule retrieval failedSigned-off-by Li Li <[email protected]>--- drivers/net/ethernet/intel/idpf/idpf_ethtool.c | 7 +++++++ 1 file changed, 7 insertions(+)drivers/net/ethernet/intel/idpf/idpf_ethtool.c +5
-0@@ -45,6 +44,11 @@ static int idpf_get_rxnfc(struct net_device *netdev, struct ethtool_rxnfc *cmd,idpf_vport_ctrl_lock (netdev); vport =idpf_netdev_to_vport (netdev); +if (!vport) { + netdev_err(netdev, "failed to get rules due to no vport in netdev\n"); + err = -EFAULT; + goto unlock; + } vport_config = np->adapter->vport_config[np->vport_idx];drivers/net/ethernet/intel/idpf/idpf_ethtool.c +1
-0user_config = &vport_config->user_config;@@ -85,6 +90,7 @@ static int idpf_get_rxnfc(struct net_device *netdev,struct ethtool_rxnfc *cmd, break; } +unlock:idpf_vport_ctrl_unlock (netdev);
Quoted 3 lines
return err;
--
2.52.0.351.gbe84eed79e-goog
PATCH 4/5 idpf: skip setting channels if vport is NULL during HW reset
In reply to Li LiWhen an idpf HW reset is triggered, it clears the vport but does not clear the netdev held by vport:
// In idpf_vport_dealloc() called by idpf_init_hard_reset(),
// idpf_init_hard_reset() sets IDPF_HR_RESET_IN_PROG, so
// idpf_decfg_netdev() doesn't get called.
if (!test_bit(IDPF_HR_RESET_IN_PROG, adapter->flags))
idpf_decfg_netdev(vport);
// idpf_decfg_netdev() would clear netdev but it isn't called:
unregister_netdev(vport->netdev);
free_netdev(vport->netdev);
vport->netdev = NULL;
// Later in idpf_init_hard_reset(), the vport is cleared:
kfree(adapter->vports);
adapter->vports = NULL;During an idpf HW reset, when userspace changes the netdev channels, the vport associated with the netdev is NULL, and so a kernel panic would happen:
[ 2245.795117] BUG: kernel NULL pointer dereference, address: 0000000000000088 ... [ 2245.842720] RIP: 0010:idpf_set_channels+0x40/0x120
This can be reproduced reliably by injecting a TX timeout to cause an idpf HW reset, and injecting a virtchnl error to cause the HW reset to fail and retry, while running "ethtool -L" in userspace.
With this patch applied, we see the following error but no kernel panics anymore:
[ 1176.743096] idpf 0000:05:00.0 eth1: channels not changed due to no vport in netdev
netlink error: Bad address
--- drivers/net/ethernet/intel/idpf/idpf_ethtool.c | 6 ++++++ 1 file changed, 6 insertions(+)
drivers/net/ethernet/intel/idpf/idpf_ethtool.c +5 -0
@@ -580,6 +579,11 @@ static int idpf_set_channels(struct net_device *netdev,idpf_vport_ctrl_lock (netdev); vport =idpf_netdev_to_vport (netdev); +if (!vport) { +netdev_err (netdev,"channels not changed due to no vport in netdev\n" ); + err = -EFAULT ; + goto unlock_mutex; + } idx = vport->idx; vport_config = vport->adapter->vport_config[idx];
Signature · 1 line
2.52.0.351.gbe84eed79e-goog
PATCH 5/5 idpf: skip stopping/opening vport if it is NULL during HW reset
In reply to Li LiWhen an idpf HW reset is triggered, it clears the vport but does not clear the netdev held by vport:
// In idpf_vport_dealloc() called by idpf_init_hard_reset(),
// idpf_init_hard_reset() sets IDPF_HR_RESET_IN_PROG, so
// idpf_decfg_netdev() doesn't get called.
if (!test_bit(IDPF_HR_RESET_IN_PROG, adapter->flags))
idpf_decfg_netdev(vport);
// idpf_decfg_netdev() would clear netdev but it isn't called:
unregister_netdev(vport->netdev);
free_netdev(vport->netdev);
vport->netdev = NULL;
// Later in idpf_init_hard_reset(), the vport is cleared:
kfree(adapter->vports);
adapter->vports = NULL;During an idpf HW reset, when userspace restarts the network service, the vport associated with the netdev is NULL, and so a kernel panic would happen:
[ 1791.669339] BUG: kernel NULL pointer dereference, address: 0000000000000070 ... [ 1791.717130] RIP: 0010:idpf_vport_stop+0x16/0x1c0
This can be reproduced reliably by injecting a TX timeout to cause an idpf HW reset, and injecting a virtchnl error to cause the HW reset to fail and retry, while running "service network restart" in userspace.
With this patch applied, we see the following error but no kernel panics anymore:
[ 181.409483] idpf 0000:05:00.0 eth1: mtu not changed due to no vport innetdev RTNETLINK answers: Bad address ... [ 181.913644] idpf 0000:05:00.0 eth1: not stopping vport because it is NULL [ 181.938675] idpf 0000:05:00.0 eth1: mtu not changed due to no vport in netdev ... [ 242.849499] idpf 0000:05:00.0 eth1: not opening vport because it is NULL ... [ 304.289364] idpf 0000:05:00.0 eth0: not opening vport because it is NULL
--- drivers/net/ethernet/intel/idpf/idpf_lib.c | 12 ++++++++++++ 1 file changed, 12 insertions(+)
drivers/net/ethernet/intel/idpf/idpf_lib.c +12 -0
@@ -1021,6 +1021,8 @@ static void idpf_vport_stop(struct idpf_vport *vport, bool rtnl)*/ static int idpf_stop (struct net_device *netdev) { +if (!netdev) +return 0 ;struct idpf_netdev_priv *np =netdev_priv (netdev);struct idpf_vport *vport;
@@ -1029,9 +1031,14 @@ static int idpf_stop(struct net_device *netdev)idpf_vport_ctrl_lock (netdev); vport =idpf_netdev_to_vport (netdev); +if (!vport) { +netdev_err (netdev,"not stopping vport because it is NULL" ); + goto unlock; + }idpf_vport_stop (vport, false); +unlock:idpf_vport_ctrl_unlock (netdev);return 0 ;
@@ -2301,6 +2308,11 @@ static int idpf_open(struct net_device *netdev)idpf_vport_ctrl_lock (netdev); vport =idpf_netdev_to_vport (netdev); +if (!vport) { +netdev_err (netdev,"not opening vport because it is NULL" ); + err = -EFAULT ; + goto unlock; + } err =idpf_set_real_num_queues (vport);if (err)
Signature · 1 line
2.52.0.351.gbe84eed79e-goog
RE: [Intel-wired-lan] [PATCH 5/5] idpf: skip stopping/opening vport if it is NULL during HW reset
Quoted 82 lines
-----Original Message-----
From: Intel-wired-lan <[email protected]> On Behalf Of Li Li via Intel-wired-lan
Sent: Wednesday, January 7, 2026 2:05 AM
To: Nguyen, Anthony L <[email protected]>; Kitszel, Przemyslaw <[email protected]>; David S. Miller <[email protected]>; Jakub Kicinski <[email protected]>; Eric Dumazet <[email protected]>; [email protected]
Cc: [email protected]; [email protected]; David Decotigny <[email protected]>; Singhai, Anjali <[email protected]>; Samudrala, Sridhar <[email protected]>; Brian Vazquez <[email protected]>; Li Li <[email protected]>; Tantilov, Emil S <[email protected]>
Subject: [Intel-wired-lan] [PATCH 5/5] idpf: skip stopping/opening vport if it is NULL during HW resetWhen an idpf HW reset is triggered, it clears the vport but does not clear the netdev held by vport:
// In idpf_vport_dealloc() called by idpf_init_hard_reset(), // idpf_init_hard_reset() sets IDPF_HR_RESET_IN_PROG, so // idpf_decfg_netdev() doesn't get called. if (!test_bit(IDPF_HR_RESET_IN_PROG, adapter->flags)) idpf_decfg_netdev(vport); // idpf_decfg_netdev() would clear netdev but it isn't called: unregister_netdev(vport->netdev); free_netdev(vport->netdev); vport->netdev = NULL; // Later in idpf_init_hard_reset(), the vport is cleared: kfree(adapter->vports); adapter->vports = NULL;During an idpf HW reset, when userspace restarts the network service, the vport associated with the netdev is NULL, and so a kernel panic would
happen:[ 1791.669339] BUG: kernel NULL pointer dereference, address: 0000000000000070 ... [ 1791.717130] RIP: 0010:idpf_vport_stop+0x16/0x1c0This can be reproduced reliably by injecting a TX timeout to cause an idpf HW reset, and injecting a virtchnl error to cause the HW reset to fail and retry, while running "service network restart" in userspace.
With this patch applied, we see the following error but no kernel panics anymore:
[ 181.409483] idpf 0000:05:00.0 eth1: mtu not changed due to no vport innetdev RTNETLINK answers: Bad address ... [ 181.913644] idpf 0000:05:00.0 eth1: not stopping vport because it is NULL [ 181.938675] idpf 0000:05:00.0 eth1: mtu not changed due to no vport in netdev ... [ 242.849499] idpf 0000:05:00.0 eth1: not opening vport because it is NULL ... [ 304.289364] idpf 0000:05:00.0 eth0: not opening vport because itis NULL
Signed-off-by Li Li <[email protected]>--- drivers/net/ethernet/intel/idpf/idpf_lib.c | 12 ++++++++++++ 1 file changed, 12 insertions(+)drivers/net/ethernet/intel/idpf/idpf_lib.c +2
-0@@ -1021,6 +1021,8 @@ static void idpf_vport_stop(struct idpf_vport*vport, bool rtnl) */ static int idpf_stop (struct net_device *netdev) { +if (!netdev) +return 0 ;struct idpf_netdev_priv *np =netdev_priv (netdev);struct idpf_vport *vport;
drivers/net/ethernet/intel/idpf/idpf_lib.c static int idpf_stop(struct net_device · line 1037
Quoted 7 lines
drivers/net/ethernet/intel/idpf/idpf_lib.c +2
-0@@ -1029,9 +1031,14 @@ static int idpf_stop(struct net_device *netdev)idpf_vport_ctrl_lock (netdev); vport =idpf_netdev_to_vport (netdev); +if (!vport) { +netdev_err (netdev,"not stopping vport because it is NULL");
Please don't forget to add trailing '\n'.
Quoted 10 lines
drivers/net/ethernet/intel/idpf/idpf_lib.c +3
-0+ goto unlock; + }idpf_vport_stop (vport, false); +unlock:idpf_vport_ctrl_unlock (netdev);drivers/net/ethernet/intel/idpf/idpf_lib.c +0
-0return 0 ;
drivers/net/ethernet/intel/idpf/idpf_lib.c static int idpf_open(struct net_device · line 2314
Quoted 7 lines
drivers/net/ethernet/intel/idpf/idpf_lib.c +2
-0@@ -2301,6 +2308,11 @@ static int idpf_open(struct net_device *netdev)idpf_vport_ctrl_lock (netdev); vport =idpf_netdev_to_vport (netdev); +if (!vport) { +netdev_err (netdev,"not opening vport because it is NULL");
Please don't forget to add trailing '\n', here too.
Quoted 8 lines
drivers/net/ethernet/intel/idpf/idpf_lib.c +3
-0+ err = -EFAULT ; + goto unlock; + }drivers/net/ethernet/intel/idpf/idpf_lib.c +0
-1err =idpf_set_real_num_queues (vport);if (err) --2.52.0.351.gbe84eed79e-goog
RE: [Intel-wired-lan] [PATCH 5/5] idpf: skip stopping/opening vport if it is NULL during HW reset
Quoted 69 lines
-----Original Message-----
From: Intel-wired-lan <[email protected]> On Behalf Of Loktionov, Aleksandr
Sent: Friday, January 9, 2026 7:07 AM
To: Li Li <[email protected]>; Nguyen, Anthony L <[email protected]>; Kitszel, Przemyslaw <[email protected]>; David S. Miller <[email protected]>; Jakub Kicinski <[email protected]>; Eric Dumazet <[email protected]>; [email protected]
Cc: [email protected]; [email protected]; David Decotigny <[email protected]>; Singhai, Anjali <[email protected]>; Samudrala, Sridhar <[email protected]>; Brian Vazquez <[email protected]>; Tantilov, Emil S <[email protected]>
Subject: Re: [Intel-wired-lan] [PATCH 5/5] idpf: skip stopping/opening vport if it is NULL during HW resetQuoted 7 lines
-----Original Message-----
From: Intel-wired-lan <[email protected]> On Behalf Of Li Li via Intel-wired-lan
Sent: Wednesday, January 7, 2026 2:05 AM
To: Nguyen, Anthony L <[email protected]>; Kitszel, Przemyslaw <[email protected]>; David S. Miller <[email protected]>; Jakub Kicinski <[email protected]>; EricDumazet
Quoted 5 lines
<[email protected]>; [email protected]
Cc: [email protected]; [email protected]; David Decotigny <[email protected]>; Singhai, Anjali <[email protected]>; Samudrala, Sridhar <[email protected]>; Brian Vazquez <[email protected]>;Li
Quoted 20 lines
Li <[email protected]>; Tantilov, Emil S <[email protected]>
Subject: [Intel-wired-lan] [PATCH 5/5] idpf: skip stopping/opening vport if it is NULL during HW resetWhen an idpf HW reset is triggered, it clears the vport but does not clear the netdev held by vport:
// In idpf_vport_dealloc() called by idpf_init_hard_reset(), // idpf_init_hard_reset() sets IDPF_HR_RESET_IN_PROG, so // idpf_decfg_netdev() doesn't get called. if (!test_bit(IDPF_HR_RESET_IN_PROG, adapter->flags)) idpf_decfg_netdev(vport); // idpf_decfg_netdev() would clear netdev but it isn't called: unregister_netdev(vport->netdev); free_netdev(vport->netdev); vport->netdev = NULL; // Later in idpf_init_hard_reset(), the vport is cleared: kfree(adapter->vports); adapter->vports = NULL;Quoted 1 line
During an idpf HW reset, when userspace restarts the network
service,
Quoted 9 lines
the vport associated with the netdev is NULL, and so a kernel panic would
happen:[ 1791.669339] BUG: kernel NULL pointer dereference, address: 0000000000000070 ... [ 1791.717130] RIP: 0010:idpf_vport_stop+0x16/0x1c0This can be reproduced reliably by injecting a TX timeout to cause
an
Quoted 1 line
idpf HW reset, and injecting a virtchnl error to cause the HW reset
to
Quoted 9 lines
Quoted 1 line
fail and retry, while running "service network restart" in
userspace.
Quoted 5 lines
With this patch applied, we see the following error but no kernel panics anymore:
[ 181.409483] idpf 0000:05:00.0 eth1: mtu not changed due to novport
Quoted 1 line
innetdev RTNETLINK answers: Bad address ...
"innetdev" -> "in netdev"
Quoted 69 lines
Quoted 2 lines
[ 181.913644] idpf 0000:05:00.0 eth1: not stopping vport because itis NULL [ 181.938675] idpf 0000:05:00.0 eth1: mtu not changed due
to
Quoted 2 lines
no vport in netdev ...
[ 242.849499] idpf 0000:05:00.0 eth1: not opening vport because itis
Quoted 2 lines
NULL ...
[ 304.289364] idpf 0000:05:00.0 eth0: not opening vport because itis
Quoted 22 lines
NULL
Signed-off-by Li Li <[email protected]>--- drivers/net/ethernet/intel/idpf/idpf_lib.c | 12 ++++++++++++ 1 file changed, 12 insertions(+)drivers/net/ethernet/intel/idpf/idpf_lib.c +2
-0@@ -1021,6 +1021,8 @@ static void idpf_vport_stop(struct idpf_vport*vport, bool rtnl) */ static int idpf_stop (struct net_device *netdev) { +if (!netdev) +return 0 ;struct idpf_netdev_priv *np =netdev_priv (netdev);struct idpf_vport *vport;drivers/net/ethernet/intel/idpf/idpf_lib.c static int idpf_stop(struct net_device · line 1037
Quoted 7 lines
drivers/net/ethernet/intel/idpf/idpf_lib.c +2
-0@@ -1029,9 +1031,14 @@ static int idpf_stop(struct net_device *netdev)idpf_vport_ctrl_lock (netdev); vport =idpf_netdev_to_vport (netdev); +if (!vport) { +netdev_err (netdev,"not stopping vport because it is NULL");Please don't forget to add trailing '\n'.
Quoted 10 lines
drivers/net/ethernet/intel/idpf/idpf_lib.c +3
-0+ goto unlock; + }idpf_vport_stop (vport, false); +unlock:idpf_vport_ctrl_unlock (netdev);drivers/net/ethernet/intel/idpf/idpf_lib.c +0
-0return 0 ;drivers/net/ethernet/intel/idpf/idpf_lib.c static int idpf_open(struct net_device · line 2314
Quoted 7 lines
drivers/net/ethernet/intel/idpf/idpf_lib.c +2
-0@@ -2301,6 +2308,11 @@ static int idpf_open(struct net_device *netdev)idpf_vport_ctrl_lock (netdev); vport =idpf_netdev_to_vport (netdev); +if (!vport) { +netdev_err (netdev,"not opening vport because it is NULL");Please don't forget to add trailing '\n', here too.
Quoted 8 lines
drivers/net/ethernet/intel/idpf/idpf_lib.c +3
-0+ err = -EFAULT ; + goto unlock; + }drivers/net/ethernet/intel/idpf/idpf_lib.c +0
-1err =idpf_set_real_num_queues (vport);if (err) --2.52.0.351.gbe84eed79e-goog
Reviewed-by Aleksandr Loktionov <[email protected]>
Dear Li,
Thank you for your patch.
Am 07.01.26 um 02:04 schrieb Li Li via Intel-wired-lan:
Quoted 6 lines
When an idpf HW reset is triggered, it clears the vport but does not clear the netdev held by vport:
// In idpf_vport_dealloc() called by idpf_init_hard_reset(), // idpf_init_hard_reset() sets IDPF_HR_RESET_IN_PROG, so // idpf_decfg_netdev() doesn't get called.
No need to format this as code comments. At least it confused me a little.
Quoted 14 lines
if (!test_bit(IDPF_HR_RESET_IN_PROG, adapter->flags)) idpf_decfg_netdev(vport); // idpf_decfg_netdev() would clear netdev but it isn't called: unregister_netdev(vport->netdev); free_netdev(vport->netdev); vport->netdev = NULL; // Later in idpf_init_hard_reset(), the vport is cleared: kfree(adapter->vports); adapter->vports = NULL;During an idpf HW reset, when "ethtool -g/-G" is called on the netdev, the vport associated with the netdev is NULL, and so a kernel panic would happen:
Quoted 8 lines
[ 513.185327] BUG: kernel NULL pointer dereference, address: 0000000000000038 ... [ 513.232756] RIP: 0010:idpf_get_ringparam+0x45/0x80This can be reproduced reliably by injecting a TX timeout to cause an idpf HW reset, and injecting a virtchnl error to cause the HW reset to fail and retry, while calling "ethtool -g/-G" on the netdev at the same time.
If you shared the commands, how to do that, it would make reproducing the issue easier.
Quoted 15 lines
With this patch applied, we see the following error but no kernel panics anymore:
[ 476.323630] idpf 0000:05:00.0 eth1: failed to get ring params due to no vport in netdevSigned-off-by Li Li <[email protected]>--- drivers/net/ethernet/intel/idpf/idpf_ethtool.c | 12 ++++++++++++ 1 file changed, 12 insertions(+)drivers/net/ethernet/intel/idpf/idpf_ethtool.c +0
-0
drivers/net/ethernet/intel/idpf/idpf_ethtool.c static void idpf_get_ringparam(struct net_device *netdev, · line 642
Quoted 5 lines
drivers/net/ethernet/intel/idpf/idpf_ethtool.c +2
-0@@ -639,6 +638,10 @@ static void idpf_get_ringparam(struct net_device *netdev,idpf_vport_ctrl_lock (netdev); vport =idpf_netdev_to_vport (netdev); +if (!vport) { +netdev_err (netdev,"failed to get ring params due to no vport in netdev\n" );
If vport == NULL is expected, why log it as an error. What should the user do? Wait until reset is done?
Quoted 15 lines
drivers/net/ethernet/intel/idpf/idpf_ethtool.c +3
-0+ goto unlock; + } ring->rx_max_pending =IDPF_MAX_RXQ_DESC ; ring->tx_max_pending =IDPF_MAX_TXQ_DESC ;@@ -647,6 +651,7 @@ static void idpf_get_ringparam(struct net_device *netdev, kring->tcp_data_split =idpf_vport_get_hsplit (vport); +unlock:idpf_vport_ctrl_unlock (netdev); }@@ -673,6 +674,11 @@ static int idpf_set_ringparam(struct net_device *netdev,
drivers/net/ethernet/intel/idpf/idpf_ethtool.c static int idpf_set_ringparam(struct net_device *netdev, · line 684
Quoted 10 lines
drivers/net/ethernet/intel/idpf/idpf_ethtool.c +5
-0idpf_vport_ctrl_lock (netdev); vport =idpf_netdev_to_vport (netdev); +if (!vport) { +netdev_err (netdev,"ring params not changed due to no vport in netdev\n" ); + err = -EFAULT ; + goto unlock_mutex; + } idx = vport->idx;
Is there another – possible more involved – solution possible to wait until the hardware reset finished?
Kind regards,
Paul
Quoted 7 lines · Paul Menzel
Dear Li,
Thank you for your patch.
Am 07.01.26 um 02:04 schrieb Li Li via Intel-wired-lan:
Quoted 8 lines
Quoted 6 lines
When an idpf HW reset is triggered, it clears the vport but does not clear the netdev held by vport:
// In idpf_vport_dealloc() called by idpf_init_hard_reset(), // idpf_init_hard_reset() sets IDPF_HR_RESET_IN_PROG, so // idpf_decfg_netdev() doesn't get called.No need to format this as code comments. At least it confused me a little.
Thanks for the pointer. Will drop the comment format in the future.
Quoted 19 lines
Quoted 19 lines
if (!test_bit(IDPF_HR_RESET_IN_PROG, adapter->flags)) idpf_decfg_netdev(vport); // idpf_decfg_netdev() would clear netdev but it isn't called: unregister_netdev(vport->netdev); free_netdev(vport->netdev); vport->netdev = NULL; // Later in idpf_init_hard_reset(), the vport is cleared: kfree(adapter->vports); adapter->vports = NULL;During an idpf HW reset, when "ethtool -g/-G" is called on the netdev, the vport associated with the netdev is NULL, and so a kernel panic would happen:
[ 513.185327] BUG: kernel NULL pointer dereference, address: 0000000000000038 ... [ 513.232756] RIP: 0010:idpf_get_ringparam+0x45/0x80
Quoted 7 lines
Quoted 4 lines
This can be reproduced reliably by injecting a TX timeout to cause an idpf HW reset, and injecting a virtchnl error to cause the HW reset to fail and retry, while calling "ethtool -g/-G" on the netdev at the same time.
If you shared the commands, how to do that, it would make reproducing the issue easier.
Here's what I did to introduce TX timeouts and virtchnl timeouts at run time:
drivers/net/ethernet/intel/idpf/idpf_txrx.c +3 -0
@@ -15,6 +15,9 @@ struct idpf_tx_stash {#define idpf_tx_buf_compl_tag (buf) (*(u32 *)&(buf)->priv)LIBETH_SQE_CHECK_PRIV (u32); +static bool SIMULATE_TX_TIMEOUT ; +module_param (SIMULATE_TX_TIMEOUT , bool,0644 ); +/** * idpf_chk_linearize - Check if skb exceeds max descriptors per packet * @skb: send buffer
@@ -79,6 +82,8 @@ void idpf_tx_timeout(struct net_device *netdev,
unsigned int txqueue)
adapter->tx_timeout_count++;
+ SIMULATE_TX_TIMEOUT = false;
+
netdev_err(netdev, "Detected Tx timeout: Count %d, Queue %d\n",
adapter->tx_timeout_count, txqueue);
if (!idpf_is_reset_in_prog(adapter)) {
@@ -2028,6 +2033,12 @@ static bool idpf_tx_clean_complq(struct
idpf_compl_queue *complq, int budget,
}
tx_q = complq->txq_grp->txqs[rel_tx_qid];+ if (unlikely(SIMULATE_TX_TIMEOUT && (tx_q->idx == 1))) {
+ netdev_err(tx_q->netdev, "boolli test:
triggering TX timeout for TX queue id %d\n", tx_q->idx);
+ goto fetch_next_desc;
+ }
+
+
/* Determine completion type */
ctype = le16_get_bits(tx_desc->qid_comptype_gen,
IDPF_TXD_COMPLQ_COMPL_TYPE_M);drivers/net/ethernet/intel/idpf/idpf_virtchnl.c +5 -0
@@ -6,6 +6,9 @@#include "idpf.h" #include "idpf_virtchnl.h" +static bool SIMULATE_VC_TIMEOUT ; +module_param (SIMULATE_VC_TIMEOUT , bool,0644 ); +#define IDPF_VC_XN_MIN_TIMEOUT_MSEC 2000#define IDPF_VC_XN_DEFAULT_TIMEOUT_MSEC (60 * 1000)#define IDPF_VC_XN_IDX_M GENMASK(7, 0)
@@ -800,6 +803,10 @@ static int idpf_send_ver_msg(struct idpf_adapter *adapter) xn_params.timeout_ms =IDPF_VC_XN_DEFAULT_TIMEOUT_MSEC ; reply_sz =idpf_vc_xn_exec (adapter, &xn_params); +if (SIMULATE_VC_TIMEOUT ) { + dev_err(&adapter->pdev->dev, "boolli test: simulating
VC timeout by returning -ETIME in idpf_send_ver_msg");
+ reply_sz = -ETIME;
+ }
if (reply_sz < 0)
return reply_sz;
if (reply_sz < sizeof(vvi))Then after the kernel is booted, we can introduce the TX timeout that lasts forever by doing the following:
echo 1 | tee /sys/module/idpf/parameters/SIMULATE_TX_TIMEOUT && echo 1 | tee /sys/module/idpf/parameters/SIMULATE_VC_TIMEOUT
All my experiments in this patch series were performed after the kernel was put in such a state.
Quoted 48 lines
Quoted 16 lines
With this patch applied, we see the following error but no kernel panics anymore:
[ 476.323630] idpf 0000:05:00.0 eth1: failed to get ring params due to no vport in netdevSigned-off-by Li Li <[email protected]>--- drivers/net/ethernet/intel/idpf/idpf_ethtool.c | 12 ++++++++++++ 1 file changed, 12 insertions(+)drivers/net/ethernet/intel/idpf/idpf_ethtool.c +0
-0drivers/net/ethernet/intel/idpf/idpf_ethtool.c static void idpf_get_ringparam(struct net_device *netdev, · line 642
Quoted 5 lines
drivers/net/ethernet/intel/idpf/idpf_ethtool.c +2
-0@@ -639,6 +638,10 @@ static void idpf_get_ringparam(struct net_device *netdev,idpf_vport_ctrl_lock (netdev); vport =idpf_netdev_to_vport (netdev); +if (!vport) { +netdev_err (netdev,"failed to get ring params due to no vport in netdev\n" );If vport == NULL is expected, why log it as an error. What should the user do? Wait until reset is done?
Quoted 23 lines
drivers/net/ethernet/intel/idpf/idpf_ethtool.c +8
-0+ goto unlock; + } ring->rx_max_pending =IDPF_MAX_RXQ_DESC ; ring->tx_max_pending =IDPF_MAX_TXQ_DESC ;@@ -647,6 +651,7 @@ static void idpf_get_ringparam(struct net_device *netdev, kring->tcp_data_split =idpf_vport_get_hsplit (vport); +unlock:idpf_vport_ctrl_unlock (netdev); }@@ -673,6 +674,11 @@ static int idpf_set_ringparam(struct net_device *netdev,idpf_vport_ctrl_lock (netdev); vport =idpf_netdev_to_vport (netdev); +if (!vport) { +netdev_err (netdev,"ring params not changed due to no vport in netdev\n" ); + err = -EFAULT ; + goto unlock_mutex; + }
Quoted 5 lines
drivers/net/ethernet/intel/idpf/idpf_ethtool.c static int idpf_set_ringparam(struct net_device *netdev, · line 684
Quoted 2 lines
drivers/net/ethernet/intel/idpf/idpf_ethtool.c +0
-0idx = vport->idx;Is there another – possible more involved – solution possible to wait until the hardware reset finished?
Please see Emil's patch series at
https://lore.kernel.org/intel-wired-lan/[email protected]/,
in which https://lore.kernel.org/intel-wired-lan/[email protected]/
detaches the netdev at the start of a HW reset, which I also think is
a more elegant solution than mine.
I'm going to drop this patch series in favor of Emil's solution above.
Quoted 5 lines
Kind regards,
Paul
Quoted 20 lines · Li Li via Intel-wired-lan
When an idpf HW reset is triggered, it clears the vport but does not clear the netdev held by vport:
// In idpf_vport_dealloc() called by idpf_init_hard_reset(), // idpf_init_hard_reset() sets IDPF_HR_RESET_IN_PROG, so // idpf_decfg_netdev() doesn't get called. if (!test_bit(IDPF_HR_RESET_IN_PROG, adapter->flags)) idpf_decfg_netdev(vport); // idpf_decfg_netdev() would clear netdev but it isn't called: unregister_netdev(vport->netdev); free_netdev(vport->netdev); vport->netdev = NULL; // Later in idpf_init_hard_reset(), the vport is cleared: kfree(adapter->vports); adapter->vports = NULL;During an idpf HW reset, when "ethtool -g/-G" is called on the netdev, the vport associated with the netdev is NULL, and so a kernel panic would happen:
Quoted 8 lines
[ 513.185327] BUG: kernel NULL pointer dereference, address: 0000000000000038 ... [ 513.232756] RIP: 0010:idpf_get_ringparam+0x45/0x80This can be reproduced reliably by injecting a TX timeout to cause an idpf HW reset, and injecting a virtchnl error to cause the HW reset to fail and retry, while calling "ethtool -g/-G" on the netdev at the same time.
I have posted series that resolves these issues in the reset path by reshuffling the flow a bit and adding netif_device_detach/attach to make sure the netdevs are better protected in the middle of a reset:
https://lore.kernel.org/intel-wired-lan/[email protected]/
If you are still seeing issues with the above applied, let me know and I can take a look.
Quoted 16 lines
With this patch applied, we see the following error but no kernel panics anymore:
[ 476.323630] idpf 0000:05:00.0 eth1: failed to get ring params due to no vport in netdevSigned-off-by Li Li <[email protected]>--- drivers/net/ethernet/intel/idpf/idpf_ethtool.c | 12 ++++++++++++ 1 file changed, 12 insertions(+)drivers/net/ethernet/intel/idpf/idpf_ethtool.c +0
-0
drivers/net/ethernet/intel/idpf/idpf_ethtool.c static void idpf_get_ringparam(struct net_device *netdev, · line 641
Quoted 4 lines
drivers/net/ethernet/intel/idpf/idpf_ethtool.c +1
-0@@ -639,6 +638,10 @@ static void idpf_get_ringparam(struct net_device *netdev,idpf_vport_ctrl_lock (netdev); vport =idpf_netdev_to_vport (netdev); +if (!vport) {
We used to have these all over the place, but the code was changed to rely on idpf_vport_ctrl_lock() for the protection of the vport state. Still some issues remain with the error paths (hence the series above), but in general we don't want to resort to vport NULL checks and rather fix the reset flows to rely on cleaner logic and locks.
Thanks,
Emil
Quoted 26 lines
drivers/net/ethernet/intel/idpf/idpf_ethtool.c +9
-0+netdev_err (netdev,"failed to get ring params due to no vport in netdev\n" ); + goto unlock; + } ring->rx_max_pending =IDPF_MAX_RXQ_DESC ; ring->tx_max_pending =IDPF_MAX_TXQ_DESC ;@@ -647,6 +651,7 @@ static void idpf_get_ringparam(struct net_device *netdev, kring->tcp_data_split =idpf_vport_get_hsplit (vport); +unlock:idpf_vport_ctrl_unlock (netdev); }@@ -673,6 +674,11 @@ static int idpf_set_ringparam(struct net_device *netdev,idpf_vport_ctrl_lock (netdev); vport =idpf_netdev_to_vport (netdev); +if (!vport) { +netdev_err (netdev,"ring params not changed due to no vport in netdev\n" ); + err = -EFAULT ; + goto unlock_mutex; + } idx = vport->idx;
Please reject this patch series given the underlying issue is fixed in an earlier patch
series already, thanks.
Quoted 33 lines · Emil S
Quoted 20 lines · Li Li via Intel-wired-lan
When an idpf HW reset is triggered, it clears the vport but does not clear the netdev held by vport:
// In idpf_vport_dealloc() called by idpf_init_hard_reset(), // idpf_init_hard_reset() sets IDPF_HR_RESET_IN_PROG, so // idpf_decfg_netdev() doesn't get called. if (!test_bit(IDPF_HR_RESET_IN_PROG, adapter->flags)) idpf_decfg_netdev(vport); // idpf_decfg_netdev() would clear netdev but it isn't called: unregister_netdev(vport->netdev); free_netdev(vport->netdev); vport->netdev = NULL; // Later in idpf_init_hard_reset(), the vport is cleared: kfree(adapter->vports); adapter->vports = NULL;During an idpf HW reset, when "ethtool -g/-G" is called on the netdev, the vport associated with the netdev is NULL, and so a kernel panic would happen:
Quoted 8 lines
[ 513.185327] BUG: kernel NULL pointer dereference, address: 0000000000000038 ... [ 513.232756] RIP: 0010:idpf_get_ringparam+0x45/0x80This can be reproduced reliably by injecting a TX timeout to cause an idpf HW reset, and injecting a virtchnl error to cause the HW reset to fail and retry, while calling "ethtool -g/-G" on the netdev at the same time.
Quoted 7 lines
I have posted series that resolves these issues in the reset path by reshuffling the flow a bit and adding netif_device_detach/attach to make sure the netdevs are better protected in the middle of a reset:
https://lore.kernel.org/intel-wired-lan/[email protected]/If you are still seeing issues with the above applied, let me know and I can take a look.
Thanks Emil! Yes I performed the experiment at a commit past your patch series above, and it
does look like the kernel panic does appear anymore. Now performing ethtool commands during
HW resets would result in "netlink error: No such device", which is expected because we are detaching
the netdev at the start of the HW reset.
Please reject this patch series, thanks!
Quoted 58 lines
Quoted 17 lines
With this patch applied, we see the following error but no kernel panics anymore:
[ 476.323630] idpf 0000:05:00.0 eth1: failed to get ring params due to no vport in netdevSigned-off-by Li Li <[email protected]>--- drivers/net/ethernet/intel/idpf/idpf_ethtool.c | 12 ++++++++++++ 1 file changed, 12 insertions(+)drivers/net/ethernet/intel/idpf/idpf_ethtool.c +0
-0drivers/net/ethernet/intel/idpf/idpf_ethtool.c static void idpf_get_ringparam(struct net_device *netdev, · line 641
Quoted 4 lines
drivers/net/ethernet/intel/idpf/idpf_ethtool.c +1
-0@@ -639,6 +638,10 @@ static void idpf_get_ringparam(struct net_device *netdev,idpf_vport_ctrl_lock (netdev); vport =idpf_netdev_to_vport (netdev); +if (!vport) {We used to have these all over the place, but the code was changed to rely on
idpf_vport_ctrl_lock()for the protection of the vport state. Still some issues remain with the error paths (hence the series above), but in general we don't want to resort to vport NULL checks and rather fix the reset flows to rely on cleaner logic and locks.Thanks,
EmilQuoted 26 lines
drivers/net/ethernet/intel/idpf/idpf_ethtool.c +9
-0+netdev_err (netdev,"failed to get ring params due to no vport in netdev\n" ); + goto unlock; + } ring->rx_max_pending =IDPF_MAX_RXQ_DESC ; ring->tx_max_pending =IDPF_MAX_TXQ_DESC ;@@ -647,6 +651,7 @@ static void idpf_get_ringparam(struct net_device *netdev, kring->tcp_data_split =idpf_vport_get_hsplit (vport); +unlock:idpf_vport_ctrl_unlock (netdev); }@@ -673,6 +674,11 @@ static int idpf_set_ringparam(struct net_device *netdev,idpf_vport_ctrl_lock (netdev); vport =idpf_netdev_to_vport (netdev); +if (!vport) { +netdev_err (netdev,"ring params not changed due to no vport in netdev\n" ); + err = -EFAULT ; + goto unlock_mutex; + } idx = vport->idx;