Skip to contentSkip to conversation

i40e: fix napi_disable hang in i40e_down() during firmware update

PATCH In review [email protected] 3 participants 3 messages Jul 27 – Oct 1

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 i40e: fix napi_disable hang in i40e_down() during firmware update

From: Mohammad Heib <[email protected]>

i40e_down() expects its caller to set __I40E_VSI_DOWN before calling it, as noted in the comment at the top of the function. All callers do this except i40e_vsi_reinit_locked(), which calls i40e_down() directly without setting the flag.

Normally this doesn't cause problems because NAPI drains cleanly on its own. But when the hardware is in a bad state, like during an NVM firmware update, the ring descriptors can contain garbage that i40e_napi_poll() sees as real work. Since __I40E_VSI_DOWN isn't set, the poll function has no reason to bail out. It keeps returning budget, napi_disable() keeps waiting, and the whole thing hangs. The thread stuck in napi_disable() is usually NetworkManager, and because it holds the RTNL lock, the system can end up completely frozen.

Set __I40E_VSI_DOWN in i40e_vsi_reinit_locked() before calling i40e_down(). i40e_up_complete() already clears the flag when the VSI comes back up, so no additional cleanup is needed.

Fixes 41c445ff0f48 ("i40e: main driver core")
---
 drivers/net/ethernet/intel/i40e/i40e_main.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/net/ethernet/intel/i40e/i40e_main.c +2 -0
index 878e0c46e5bb..78f95f63b588 100644
@@ -7463,6 +7463,8 @@ static void i40e_vsi_reinit_locked(struct i40e_vsi *vsi)
 
 	while (test_and_set_bit(__I40E_CONFIG_BUSY, pf->state))
 		usleep_range(1000, 2000);
+	/* required by i40e_down() */
+	set_bit(__I40E_VSI_DOWN, vsi->state);
 	i40e_down(vsi);
 
 	i40e_up(vsi);
Signature · 1 line
2.53.0
Reviewed-by on v1 1/1
#1 lore raw
Quoted 20 lines · On Mon, Jul 27, 2026 at 02:57:56PM +0300, [email protected] wrote:

From: Mohammad Heib <[email protected]>

i40e_down() expects its caller to set __I40E_VSI_DOWN before calling it, as noted in the comment at the top of the function. All callers do this except i40e_vsi_reinit_locked(), which calls i40e_down() directly without setting the flag.

Normally this doesn't cause problems because NAPI drains cleanly on its own. But when the hardware is in a bad state, like during an NVM firmware update, the ring descriptors can contain garbage that i40e_napi_poll() sees as real work. Since __I40E_VSI_DOWN isn't set, the poll function has no reason to bail out. It keeps returning budget, napi_disable() keeps waiting, and the whole thing hangs. The thread stuck in napi_disable() is usually NetworkManager, and because it holds the RTNL lock, the system can end up completely frozen.

Set __I40E_VSI_DOWN in i40e_vsi_reinit_locked() before calling i40e_down(). i40e_up_complete() already clears the flag when the VSI comes back up, so no additional cleanup is needed.

Quoted 6 lines
Fixes 41c445ff0f48 ("i40e: main driver core")
---
 drivers/net/ethernet/intel/i40e/i40e_main.c | 2 ++
 1 file changed, 2 insertions(+)
Tested-by on v1 1/1
#2 lore raw
In reply to [email protected]
Quoted 15 lines

-----Original Message-----
From: Intel-wired-lan <[email protected]> On Behalf Of [email protected]
Sent: Monday, July 27, 2026 4:58 AM
To: [email protected]
Cc: [email protected]; Nguyen, Anthony L <[email protected]>; Kitszel, Przemyslaw <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; Mohammad Heib <[email protected]>
Subject: [Intel-wired-lan] [PATCH iwl-net] i40e: fix napi_disable hang in i40e_down() during firmware update

From: Mohammad Heib <[email protected]>

i40e_down() expects its caller to set __I40E_VSI_DOWN before calling it, as noted in the comment at the top of the function. All callers do this except i40e_vsi_reinit_locked(), which calls i40e_down() directly without setting the flag.

Normally this doesn't cause problems because NAPI drains cleanly on its own. But when the hardware is in a bad state, like during an NVM firmware update, the ring descriptors can contain garbage that i40e_napi_poll() sees as real work. Since __I40E_VSI_DOWN isn't set, the poll function has no reason to bail out. It keeps returning budget,
napi_disable() keeps waiting, and the whole thing hangs. The thread stuck in napi_disable() is usually NetworkManager, and because it holds the RTNL lock, the system can end up completely frozen.

Quoted 7 lines

Set __I40E_VSI_DOWN in i40e_vsi_reinit_locked() before calling i40e_down(). i40e_up_complete() already clears the flag when the VSI comes back up, so no additional cleanup is needed.

Fixes 41c445ff0f48 ("i40e: main driver core")
---
 drivers/net/ethernet/intel/i40e/i40e_main.c | 2 ++
 1 file changed, 2 insertions(+)