iavf: Fix close path on shutdown and remove in iavf driver.
Subsystems (2)
v2 cover iavf: Fix close path on shutdown and remove in iavf driver.
From: Slawomir Laba <[email protected]>
iavf_shutdown was implementing an incomplete version of iavf_remove. It misses several calls to the kernel like iavf_free_misc_irq, iavf_reset_interrupt_capability, iounmap that might break the system on reboot or hibernation.
Fix a deadlock introduced by commit974578017fc1f ("iavf: Add waiting so the port is initialized in remove") due to race condition between iavf_shutdown and iavf_remove, where iavf_remove stucks forever in while loop since iavf_shutdown already set __IAVF_REMOVE adapter state.
---
v2: Fixed author
---
SlawomirX Laba (2):
iavf: Fix shutdown pci callback to match the remove one
iavf: Fix race condition between iavf_shutdown and iavf_remove
drivers/net/ethernet/intel/iavf/iavf_main.c | 59 ++++++++------------- 1 file changed, 22 insertions(+), 37 deletions(-)
Signature · 6 lines
2.27.0 _______________________________________________ Intel-wired-lan mailing list [email protected] https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
PATCH v2 1/2 iavf: Fix shutdown pci callback to match the remove one
From: Slawomir Laba <[email protected]>
Make the flow for pci shutdown be the same to the pci remove.
iavf_shutdown was implementing an incomplete version of iavf_remove. It misses several calls to the kernel like iavf_free_misc_irq, iavf_reset_interrupt_capability, iounmap that might break the system on reboot or hibernation.
Implement the call of iavf_remove directly in iavf_shutdown to close this gap.
---
v2: Fixed author
--- drivers/net/ethernet/intel/iavf/iavf_main.c | 40 +++++++-------------- 1 file changed, 12 insertions(+), 28 deletions(-)
drivers/net/ethernet/intel/iavf/iavf_main.c +12 -28
@@ -4647,34 +4647,6 @@ int iavf_process_config(struct iavf_adapter *adapter)return 0 ; } -/** -* iavf_shutdown - Shutdown the device in preparation for a reboot -* @pdev: pci device structure -**/ -static void iavf_shutdown (struct pci_dev *pdev) -{ -struct iavf_adapter *adapter =iavf_pdev_to_adapter (pdev); -struct net_device *netdev = adapter->netdev; - -netif_device_detach (netdev); - -if (netif_running (netdev)) -iavf_close (netdev); - -if (iavf_lock_timeout (&adapter->crit_lock,5000 )) -dev_warn (&adapter->pdev->dev,"failed to acquire crit_lock in %s\n" , __FUNCTION__); -/* Prevent the watchdog from running. */ -iavf_change_state (adapter, __IAVF_REMOVE); - adapter->aq_required =0 ; -mutex_unlock (&adapter->crit_lock); - -#ifdef CONFIG_PM -pci_save_state (pdev); - -#endif -pci_disable_device (pdev); -} -/** * iavf_probe - Device Initialization Routine * @pdev: PCI device information struct
@@ -5011,6 +4983,18 @@ static void iavf_remove(struct pci_dev *pdev)pci_disable_device (pdev); } +/** +* iavf_shutdown - Shutdown the device in preparation for a reboot +* @pdev: pci device structure +**/ +static void iavf_shutdown (struct pci_dev *pdev) +{ +iavf_remove (pdev); + +if (system_state ==SYSTEM_POWER_OFF ) +pci_set_power_state (pdev, PCI_D3hot); +} + static SIMPLE_DEV_PM_OPS(iavf_pm_ops, iavf_suspend, iavf_resume);static struct pci_driver iavf_driver = {
Signature · 6 lines
2.27.0 _______________________________________________ Intel-wired-lan mailing list [email protected] https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
Re: [Intel-wired-lan] [PATCH net v2 1/2] iavf: Fix shutdown pci callback to match the remove one
drivers/net/ethernet/intel/iavf/iavf_main.c
Quoted 36 lines
-----Original Message-----
From: Intel-wired-lan <[email protected]> On Behalf Of Mateusz Palczewski
Sent: wtorek, 2 sierpnia 2022 13:52
To: [email protected]
Cc: Laba, SlawomirX <[email protected]>
Subject: [Intel-wired-lan] [PATCH net v2 1/2] iavf: Fix shutdown pci callback to match the remove oneFrom: Slawomir Laba <[email protected]>
Make the flow for pci shutdown be the same to the pci remove.
iavf_shutdownwas implementing an incomplete version ofiavf_remove. It misses several calls to the kernel likeiavf_free_misc_irq,iavf_reset_interrupt_capability, iounmap that might break the system on reboot or hibernation.Implement the call of
iavf_removedirectly iniavf_shutdownto close this gap.Fixes 5eae00c57f5e ("i40evf: main driver core")Signed-off-by Slawomir Laba <[email protected]>Signed-off-by Mateusz Palczewski <[email protected]>---
v2: Fixed author--- drivers/net/ethernet/intel/iavf/iavf_main.c | 40 +++++++-------------- 1 file changed, 12 insertions(+), 28 deletions(-)drivers/net/ethernet/intel/iavf/iavf_main.c +0
-0
_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
PATCH v2 2/2 iavf: Fix race condition between iavf_shutdown and iavf_remove
In reply to Mateusz PalczewskiFrom: Slawomir Laba <[email protected]>
Fix a deadlock introduced by commit974578017fc1 ("iavf: Add waiting so the port is initialized in remove") due to race condition between iavf_shutdown and iavf_remove, where iavf_remove stucks forever in while loop since iavf_shutdown already set __IAVF_REMOVE adapter state.
Fix this by checking if the __IAVF_IN_REMOVE_TASK has already been set and return if so.
---
v2: Fixed author
--- drivers/net/ethernet/intel/iavf/iavf_main.c | 19 ++++++++++--------- 1 file changed, 10 insertions(+), 9 deletions(-)
drivers/net/ethernet/intel/iavf/iavf_main.c +10 -9
@@ -4846,23 +4846,24 @@ static int __maybe_unused iavf_resume(struct device *dev_d)static void iavf_remove (struct pci_dev *pdev) {struct iavf_adapter *adapter =iavf_pdev_to_adapter (pdev); -struct net_device *netdev = adapter->netdev;struct iavf_fdir_fltr *fdir, *fdirtmp;struct iavf_vlan_filter *vlf, *vlftmp; +struct iavf_cloud_filter *cf, *cftmp;struct iavf_adv_rss *rss, *rsstmp;struct iavf_mac_filter *f, *ftmp; -struct iavf_cloud_filter *cf, *cftmp; -struct iavf_hw *hw = &adapter->hw; +struct net_device *netdev; +struct iavf_hw *hw;int err; -/* When reboot/shutdown is in progress no need to do anything -* as the adapter is already REMOVE state that was set during -* iavf_shutdown() callback. -*/ -if (adapter->state == __IAVF_REMOVE) +if (!adapter) +return ; + + netdev = adapter->netdev; + hw = &adapter->hw; + +if (test_and_set_bit (__IAVF_IN_REMOVE_TASK, &adapter->crit_section))return ; -set_bit (__IAVF_IN_REMOVE_TASK, &adapter->crit_section);/* Wait until port initialization is complete. * There are flows where register/unregister netdev may race. */
Signature · 6 lines
2.27.0 _______________________________________________ Intel-wired-lan mailing list [email protected] https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
Re: [Intel-wired-lan] [PATCH net v2 2/2] iavf: Fix race condition between iavf_shutdown and iavf_remove
drivers/net/ethernet/intel/iavf/iavf_main.c
Quoted 36 lines
-----Original Message-----
From: Intel-wired-lan <[email protected]> On Behalf Of Mateusz Palczewski
Sent: wtorek, 2 sierpnia 2022 13:52
To: [email protected]
Cc: Laba, SlawomirX <[email protected]>
Subject: [Intel-wired-lan] [PATCH net v2 2/2] iavf: Fix race condition betweeniavf_shutdownandiavf_removeFrom: Slawomir Laba <[email protected]>
Fix a deadlock introduced by commit
974578017fc1("iavf: Add waiting so the port is initialized in remove") due to race condition betweeniavf_shutdownandiavf_remove, whereiavf_removestucks forever in while loop sinceiavf_shutdownalready set__IAVF_REMOVEadapter state.Fix this by checking if the
__IAVF_IN_REMOVE_TASKhas already been set and return if so.Fixes 974578017fc1 ("iavf: Add waiting so the port is initialized in remove")Signed-off-by Slawomir Laba <[email protected]>Signed-off-by Mateusz Palczewski <[email protected]>---
v2: Fixed author--- drivers/net/ethernet/intel/iavf/iavf_main.c | 19 ++++++++++--------- 1 file changed, 10 insertions(+), 9 deletions(-)drivers/net/ethernet/intel/iavf/iavf_main.c +0
-0
_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
Re: [Intel-wired-lan] [PATCH net v2 2/2] iavf: Fix race condition between iavf_shutdown and iavf_remove
In reply to Mateusz PalczewskiQuoted 41 lines · Mateusz Palczewski
From: Slawomir Laba <[email protected]>
Fix a deadlock introduced by commit
974578017fc1("iavf: Add waiting so the port is initialized in remove") due to race condition betweeniavf_shutdownandiavf_remove, whereiavf_removestucks forever in while loop sinceiavf_shutdownalready set__IAVF_REMOVEadapter state.Fix this by checking if the
__IAVF_IN_REMOVE_TASKhas already been set and return if so.Fixes 974578017fc1 ("iavf: Add waiting so the port is initialized inremove")
Signed-off-by Slawomir Laba <[email protected]>Signed-off-by Mateusz Palczewski <[email protected]>---
v2: Fixed author--- drivers/net/ethernet/intel/iavf/iavf_main.c | 19 ++++++++++--------- 1 file changed, 10 insertions(+), 9 deletions(-)drivers/net/ethernet/intel/iavf/iavf_main.c +3
-3@@ -4846,23 +4846,24 @@ static int __maybe_unused iavf_resume(struct device *dev_d)static void iavf_remove (struct pci_dev *pdev) {struct iavf_adapter *adapter =iavf_pdev_to_adapter (pdev); -struct net_device *netdev = adapter->netdev;struct iavf_fdir_fltr *fdir, *fdirtmp;struct iavf_vlan_filter *vlf, *vlftmp; +struct iavf_cloud_filter *cf, *cftmp;struct iavf_adv_rss *rss, *rsstmp;struct iavf_mac_filter *f, *ftmp; -struct iavf_cloud_filter *cf, *cftmp; -struct iavf_hw *hw = &adapter->hw; +struct net_device *netdev; +struct iavf_hw *hw;
drivers/net/ethernet/intel/iavf/iavf_main.c static int __maybe_unused iavf_resume(struct · line 4861
Quoted 10 lines
drivers/net/ethernet/intel/iavf/iavf_main.c +2
-5int err; -/* When reboot/shutdown is in progress no need to do anything -* as the adapter is already REMOVE state that was set during -* iavf_shutdown() callback. -*/ -if (adapter->state == __IAVF_REMOVE) +if (!adapter) +return ;
The check for !adapter cannot work. iavf_pdev_to_adapter(pdev) will never return NULL. It does:
return netdev_priv(pci_get_drvdata(pdev));
Even if pci_get_drvdata(pdev) somehow returns NULL (which I don't think it will, because the driver never resets the drvdata before freeing netdev), netdev_priv() would make a non-NULL value out of it (it adds the aligned size of struct net_device to the pointer).
CCing Ivan, who will have more comments about the adapter states. Michal
+
Quoted 14 lines
drivers/net/ethernet/intel/iavf/iavf_main.c +4
-2+ netdev = adapter->netdev; + hw = &adapter->hw; + +if (test_and_set_bit (__IAVF_IN_REMOVE_TASK, &adapter->crit_section))return ; -set_bit (__IAVF_IN_REMOVE_TASK, &adapter->crit_section);/* Wait until port initialization is complete. * There are flows where register/unregister netdev may race. */ --2.27.0
Quoted 6 lines
_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
Re: [Intel-wired-lan] [PATCH net v2 2/2] iavf: Fix race condition between iavf_shutdown and iavf_remove
Quoted 57 lines · On Thu, 4 Aug 2022 11:14:58 +0200 Michal Schmidt <[email protected]> wrote:
Quoted 41 lines · Mateusz Palczewski
From: Slawomir Laba <[email protected]>
Fix a deadlock introduced by commit
974578017fc1("iavf: Add waiting so the port is initialized in remove") due to race condition betweeniavf_shutdownandiavf_remove, whereiavf_removestucks forever in while loop sinceiavf_shutdownalready set__IAVF_REMOVEadapter state.Fix this by checking if the
__IAVF_IN_REMOVE_TASKhas already been set and return if so.Fixes 974578017fc1 ("iavf: Add waiting so the port is initialized inremove")
Signed-off-by Slawomir Laba <[email protected]>Signed-off-by Mateusz Palczewski <[email protected]>---
v2: Fixed author--- drivers/net/ethernet/intel/iavf/iavf_main.c | 19 ++++++++++--------- 1 file changed, 10 insertions(+), 9 deletions(-)drivers/net/ethernet/intel/iavf/iavf_main.c +3
-3@@ -4846,23 +4846,24 @@ static int __maybe_unused iavf_resume(struct device *dev_d)static void iavf_remove (struct pci_dev *pdev) {struct iavf_adapter *adapter =iavf_pdev_to_adapter (pdev); -struct net_device *netdev = adapter->netdev;struct iavf_fdir_fltr *fdir, *fdirtmp;struct iavf_vlan_filter *vlf, *vlftmp; +struct iavf_cloud_filter *cf, *cftmp;struct iavf_adv_rss *rss, *rsstmp;struct iavf_mac_filter *f, *ftmp; -struct iavf_cloud_filter *cf, *cftmp; -struct iavf_hw *hw = &adapter->hw; +struct net_device *netdev; +struct iavf_hw *hw;drivers/net/ethernet/intel/iavf/iavf_main.c static int __maybe_unused iavf_resume(struct · line 4861
Quoted 10 lines
drivers/net/ethernet/intel/iavf/iavf_main.c +2
-5int err; -/* When reboot/shutdown is in progress no need to do anything -* as the adapter is already REMOVE state that was set during -* iavf_shutdown() callback. -*/ -if (adapter->state == __IAVF_REMOVE) +if (!adapter) +return ;The check for !adapter cannot work.
iavf_pdev_to_adapter(pdev)will never return NULL. It does:
Quoted 8 lines
return netdev_priv(pci_get_drvdata(pdev));Even if
pci_get_drvdata(pdev)somehow returns NULL (which I don't think it will, because the driver never resets the drvdata before freeing netdev),netdev_priv()would make a non-NULL value out of it (it adds the aligned size of structnet_deviceto the pointer).CCing Ivan, who will have more comments about the adapter states. Michal
Yes, to make your patch working correctly you need to adjust iavf_pdev_to_adapter() appropriately and also set pci drvdata to NULL prior free_netdev():
drivers/net/ethernet/intel/iavf/iavf_main.c +4 -1
@@ -164,7 +164,9 @@ int virtchnl_status_to_errno(enum virtchnl_status_code v_status)*/ static struct iavf_adapter *iavf_pdev_to_adapter (struct pci_dev *pdev) { -return netdev_priv (pci_get_drvdata (pdev)); +struct net_device *netdev =pci_get_drvdata (pdev); + +return netdev ?netdev_priv (netdev) :NULL ; }/**
@@ -4899,6 +4901,7 @@ static void iavf_remove(struct pci_dev *pdev) }spin_unlock_bh (&adapter->adv_rss_lock); +pci_set_drvdata (pdev,NULL );free_netdev (netdev);pci_disable_pcie_error_reporting (pdev);
Regarding adapter states... __IAVF_REMOVE can be removed as redundant and instead only use __IAVF_IN_REMOVE_TASK bit.
Ivan
_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
Re: [Intel-wired-lan] [PATCH net v2 2/2] iavf: Fix race condition between iavf_shutdown and iavf_remove
Quoted 77 lines
-----Original Message-----
From: Ivan Vecera <[email protected]>
Sent: Thursday, August 4, 2022 11:38 AM
To: mschmidt <[email protected]>
Cc: Palczewski, Mateusz <[email protected]>; intel-wired- [email protected]; Laba, SlawomirX <[email protected]>
Subject: Re: [Intel-wired-lan] [PATCH net v2 2/2] iavf: Fix race condition betweeniavf_shutdownandiavf_removeQuoted 26 lines · On Thu, 4 Aug 2022 11:14:58 +0200 Michal Schmidt <[email protected]> wrote:
Quoted 23 lines · Mateusz Palczewski
From: Slawomir Laba <[email protected]>
Fix a deadlock introduced by commit
974578017fc1("iavf: Add waiting so the port is initialized in remove") due to race condition betweeniavf_shutdownandiavf_remove, whereiavf_removestucks forever in while loop sinceiavf_shutdownalready set__IAVF_REMOVEadapter state.Fix this by checking if the
__IAVF_IN_REMOVE_TASKhas already been set and return if so.Fixes 974578017fc1 ("iavf: Add waiting so the port is initializedin
remove")Signed-off-by Slawomir Laba <[email protected]>Signed-off-by Mateusz Palczewski <[email protected]>---
v2: Fixed author--- drivers/net/ethernet/intel/iavf/iavf_main.c | 19 ++++++++++--------- 1 file changed, 10 insertions(+), 9 deletions(-)Quoted 6 lines
drivers/net/ethernet/intel/iavf/iavf_main.c static int __maybe_unused
Quoted 6 lines
drivers/net/ethernet/intel/iavf/iavf_main.c +0
-0@@ -4846,23 +4846,24 @@ static int __maybe_unusediavf_resume(struct
Quoted 32 lines
Quoted 12 lines
drivers/net/ethernet/intel/iavf/iavf_main.c +3
-3device *dev_d) static void iavf_remove(struct pci_dev *pdev) {struct iavf_adapter *adapter =iavf_pdev_to_adapter (pdev); -struct net_device *netdev = adapter->netdev;struct iavf_fdir_fltr *fdir, *fdirtmp;struct iavf_vlan_filter *vlf, *vlftmp; +struct iavf_cloud_filter *cf, *cftmp;struct iavf_adv_rss *rss, *rsstmp;struct iavf_mac_filter *f, *ftmp; -struct iavf_cloud_filter *cf, *cftmp; -struct iavf_hw *hw = &adapter->hw; +struct net_device *netdev; +struct iavf_hw *hw;drivers/net/ethernet/intel/iavf/iavf_main.c static int __maybe_unused · line 4859
Quoted 10 lines
drivers/net/ethernet/intel/iavf/iavf_main.c +2
-5int err; -/* When reboot/shutdown is in progress no need to do anything -* as the adapter is already REMOVE state that was set during -* iavf_shutdown() callback. -*/ -if (adapter->state == __IAVF_REMOVE) +if (!adapter) +return ;The check for !adapter cannot work.
iavf_pdev_to_adapter(pdev)will never return NULL. It does:return netdev_priv(pci_get_drvdata(pdev));Even if
pci_get_drvdata(pdev)somehow returns NULL (which I don't think it will, because the driver never resets the drvdata before freeing netdev),netdev_priv()would make a non-NULL value out of it (it adds the aligned size of structnet_deviceto the pointer).
Quoted 6 lines
Quoted 2 lines
CCing Ivan, who will have more comments about the adapter states. Michal
Yes, to make your patch working correctly you need to adjust
iavf_pdev_to_adapter()appropriately and also set pci drvdata to NULL priorfree_netdev():
Thanks Ivan for a nice fix of this problem. The only way that this check would work is when iavf_probe fails with no memory. We also came to the conclusion that this check is not really necessary and our update on this patch would be to simply remove the check on the adapter to NULL. What do you think about this?
Quoted 22 lines
drivers/net/ethernet/intel/iavf/iavf_main.c +4
-1@@ -164,7 +164,9 @@ int virtchnl_status_to_errno(enumvirtchnl_status_code v_status) */ static struct iavf_adapter *iavf_pdev_to_adapter (struct pci_dev *pdev) { -return netdev_priv (pci_get_drvdata (pdev)); +struct net_device *netdev =pci_get_drvdata (pdev); + +return netdev ?netdev_priv (netdev) :NULL ; }/** @@ -4899,6 +4901,7 @@ static void iavf_remove(struct pci_dev *pdev) }spin_unlock_bh (&adapter->adv_rss_lock); +pci_set_drvdata (pdev,NULL );free_netdev (netdev);pci_disable_pcie_error_reporting (pdev);
Quoted 4 lines
Regarding adapter states...
__IAVF_REMOVEcan be removed as redundant and instead only use__IAVF_IN_REMOVE_TASKbit.Ivan
I divided iavf_remove function into two logical pieces. The first piece helps the driver to survive races of watchdog init states and iavf_remove call. So when init fails the driver doesn't reinit if remove is triggered. Additionally the __IAVF_IN_REMOVE_TASK was introduced in order to fix race condition between register_netdevice in init and unregister_netdevice in remove. The second piece is the cleanup of resources after netdev gets unregistered. I had no better idea on how to deal with unregister_netdevice call without holding crit_lock. Unregister_netdevice function will call iavf_close which requires this lock in order to free traffic irqs and cleanup rx and tx resources.
The __IAVF_REMOVE state is used in different purpose, for example when the driver is ready to stop workqueues (after netdev gets unregistered) and iavf_remove already holds the crit_lock for final cleanups.
_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
Re: [Intel-wired-lan] [PATCH net v2 2/2] iavf: Fix race condition between iavf_shutdown and iavf_remove
Quoted 80 lines · SlawomirX
Quoted 80 lines
-----Original Message-----
From: Ivan Vecera <[email protected]>
Sent: Thursday, August 4, 2022 11:38 AM
To: mschmidt <[email protected]>
Cc: Palczewski, Mateusz <[email protected]>; intel-wired- [email protected]; Laba, SlawomirX <[email protected]>
Subject: Re: [Intel-wired-lan] [PATCH net v2 2/2] iavf: Fix race condition betweeniavf_shutdownandiavf_removeQuoted 26 lines · On Thu, 4 Aug 2022 11:14:58 +0200 Michal Schmidt <[email protected]> wrote:
Quoted 23 lines · Mateusz Palczewski
From: Slawomir Laba <[email protected]>
Fix a deadlock introduced by commit
974578017fc1("iavf: Add waiting so the port is initialized in remove") due to race condition betweeniavf_shutdownandiavf_remove, whereiavf_removestucks forever in while loop sinceiavf_shutdownalready set__IAVF_REMOVEadapter state.Fix this by checking if the
__IAVF_IN_REMOVE_TASKhas already been set and return if so.Fixes 974578017fc1 ("iavf: Add waiting so the port is initializedin
remove")Signed-off-by Slawomir Laba <[email protected]>Signed-off-by Mateusz Palczewski <[email protected]>---
v2: Fixed author--- drivers/net/ethernet/intel/iavf/iavf_main.c | 19 ++++++++++--------- 1 file changed, 10 insertions(+), 9 deletions(-)Quoted 6 lines
drivers/net/ethernet/intel/iavf/iavf_main.c static int __maybe_unused
Quoted 6 lines
drivers/net/ethernet/intel/iavf/iavf_main.c +0
-0@@ -4846,23 +4846,24 @@ static int __maybe_unusediavf_resume(struct
Quoted 34 lines
Quoted 12 lines
drivers/net/ethernet/intel/iavf/iavf_main.c +3
-3device *dev_d) static void iavf_remove(struct pci_dev *pdev) {struct iavf_adapter *adapter =iavf_pdev_to_adapter (pdev); -struct net_device *netdev = adapter->netdev;struct iavf_fdir_fltr *fdir, *fdirtmp;struct iavf_vlan_filter *vlf, *vlftmp; +struct iavf_cloud_filter *cf, *cftmp;struct iavf_adv_rss *rss, *rsstmp;struct iavf_mac_filter *f, *ftmp; -struct iavf_cloud_filter *cf, *cftmp; -struct iavf_hw *hw = &adapter->hw; +struct net_device *netdev; +struct iavf_hw *hw;drivers/net/ethernet/intel/iavf/iavf_main.c static int __maybe_unused · line 4859
Quoted 10 lines
drivers/net/ethernet/intel/iavf/iavf_main.c +2
-5int err; -/* When reboot/shutdown is in progress no need to do anything -* as the adapter is already REMOVE state that was set during -* iavf_shutdown() callback. -*/ -if (adapter->state == __IAVF_REMOVE) +if (!adapter) +return ;The check for !adapter cannot work.
iavf_pdev_to_adapter(pdev)will never return NULL. It does:return netdev_priv(pci_get_drvdata(pdev));Even if
pci_get_drvdata(pdev)somehow returns NULL (which I don't think it will, because the driver never resets the drvdata before freeing netdev),netdev_priv()would make a non-NULL value out of it (it adds the aligned size of structnet_deviceto the pointer).CCing Ivan, who will have more comments about the adapter states. Michal
Quoted 7 lines
Quoted 3 lines
Yes, to make your patch working correctly you need to adjust
iavf_pdev_to_adapter()appropriately and also set pci drvdata to NULL priorfree_netdev():Thanks Ivan for a nice fix of this problem. The only way that this check would work is when
iavf_probefails with no memory. We also came to the conclusion that this check is not really necessary and our update on this patch would be to simply remove the check on the adapter to NULL. What do you think about this?
OK, make sense.
Quoted 33 lines
Quoted 22 lines
drivers/net/ethernet/intel/iavf/iavf_main.c +4
-1@@ -164,7 +164,9 @@ int virtchnl_status_to_errno(enumvirtchnl_status_code v_status) */ static struct iavf_adapter *iavf_pdev_to_adapter (struct pci_dev *pdev) { -return netdev_priv (pci_get_drvdata (pdev)); +struct net_device *netdev =pci_get_drvdata (pdev); + +return netdev ?netdev_priv (netdev) :NULL ; }/** @@ -4899,6 +4901,7 @@ static void iavf_remove(struct pci_dev *pdev) }spin_unlock_bh (&adapter->adv_rss_lock); +pci_set_drvdata (pdev,NULL );free_netdev (netdev);pci_disable_pcie_error_reporting (pdev);Quoted 4 lines
Regarding adapter states...
__IAVF_REMOVEcan be removed as redundant and instead only use__IAVF_IN_REMOVE_TASKbit.Ivan
I divided
iavf_removefunction into two logical pieces. The first piece helps the driver to survive races of watchdog init states andiavf_removecall. So when init fails the driver doesn't reinit if remove is triggered. Additionally the__IAVF_IN_REMOVE_TASKwas introduced in order to fix race condition betweenregister_netdevicein init andunregister_netdevicein remove. The second piece is the cleanup of resources after netdev gets unregistered. I had no better idea on how to deal withunregister_netdevicecall without holdingcrit_lock.Unregister_netdevicefunction will calliavf_closewhich requires this lock in order to free traffic irqs and cleanup rx and tx resources.
Quoted 2 lines
The
__IAVF_REMOVEstate is used in different purpose, for example when the driver is ready to stop workqueues (after netdev gets unregistered) andiavf_removealready holds thecrit_lockfor final cleanups.
OK, got it.
Thanks,
Ivan
_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan