[PATCH for-next] RDMA/efa: Reset device on probe failure

[PATCH for-next] RDMA/efa: Reset device on probe failure

[Michael Margolin]

Make sure the device is being reset on driver exit whatever the reason
is, to keep the device aligned and allow it to close shared resources
(e.g. admin queue).

Reviewed-by: Firas Jahjah <firasj@amazon.com>
Reviewed-by: Yonatan Nachum <ynachum@amazon.com>
Signed-off-by: Michael Margolin <mrgolin@amazon.com>
---
 drivers/infiniband/hw/efa/efa_main.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/drivers/infiniband/hw/efa/efa_main.c b/drivers/infiniband/hw/efa/efa_main.c
index ad225823e6f2..0b102089e0ab 100644
--- a/drivers/infiniband/hw/efa/efa_main.c
+++ b/drivers/infiniband/hw/efa/efa_main.c
@@ -470,7 +470,6 @@ static void efa_ib_device_remove(struct efa_dev *dev)
 	ibdev_info(&dev->ibdev, "Unregister ib device\n");
 	ib_unregister_device(&dev->ibdev);
 	efa_destroy_eqs(dev);
-	efa_com_dev_reset(&dev->edev, EFA_REGS_RESET_NORMAL);
 	efa_release_doorbell_bar(dev);
 }
 
@@ -643,12 +642,13 @@ static struct efa_dev *efa_probe_device(struct pci_dev *pdev)
 	return ERR_PTR(err);
 }
 
-static void efa_remove_device(struct pci_dev *pdev)
+static void efa_remove_device(struct pci_dev *pdev, bool on_error)
 {
 	struct efa_dev *dev = pci_get_drvdata(pdev);
 	struct efa_com_dev *edev;
 
 	edev = &dev->edev;
+	efa_com_dev_reset(edev, on_error ? EFA_REGS_RESET_INIT_ERR : EFA_REGS_RESET_NORMAL);
 	efa_com_admin_destroy(edev);
 	efa_free_irq(dev, &dev->admin_irq);
 	efa_disable_msix(dev);
@@ -676,7 +676,7 @@ static int efa_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 	return 0;
 
 err_remove_device:
-	efa_remove_device(pdev);
+	efa_remove_device(pdev, true);
 	return err;
 }
 
@@ -685,7 +685,7 @@ static void efa_remove(struct pci_dev *pdev)
 	struct efa_dev *dev = pci_get_drvdata(pdev);
 
 	efa_ib_device_remove(dev);
-	efa_remove_device(pdev);
+	efa_remove_device(pdev, false);
 }
 
 static void efa_shutdown(struct pci_dev *pdev)
-- 
2.40.1

Re: [PATCH for-next] RDMA/efa: Reset device on probe failure

[Gal Pressman]

On 25/12/2024 15:15, Michael Margolin wrote:
> Make sure the device is being reset on driver exit whatever the reason
> is, to keep the device aligned and allow it to close shared resources
> (e.g. admin queue).
> 
> Reviewed-by: Firas Jahjah <firasj@amazon.com>
> Reviewed-by: Yonatan Nachum <ynachum@amazon.com>
> Signed-off-by: Michael Margolin <mrgolin@amazon.com>
> ---
> @@ -685,7 +685,7 @@ static void efa_remove(struct pci_dev *pdev)
>  	struct efa_dev *dev = pci_get_drvdata(pdev);
>  
>  	efa_ib_device_remove(dev);

This already calls efa_com_dev_reset(), you now perform double reset in
the normal remove flow.

> -	efa_remove_device(pdev);
> +	efa_remove_device(pdev, false);
>  }

Re: [PATCH for-next] RDMA/efa: Reset device on probe failure

[Gal Pressman]

On 30/12/2024 8:46, Gal Pressman wrote:
> On 25/12/2024 15:15, Michael Margolin wrote:
>> Make sure the device is being reset on driver exit whatever the reason
>> is, to keep the device aligned and allow it to close shared resources
>> (e.g. admin queue).
>>
>> Reviewed-by: Firas Jahjah <firasj@amazon.com>
>> Reviewed-by: Yonatan Nachum <ynachum@amazon.com>
>> Signed-off-by: Michael Margolin <mrgolin@amazon.com>
>> ---
>> @@ -685,7 +685,7 @@ static void efa_remove(struct pci_dev *pdev)
>>  	struct efa_dev *dev = pci_get_drvdata(pdev);
>>  
>>  	efa_ib_device_remove(dev);
> 
> This already calls efa_com_dev_reset(), you now perform double reset in
> the normal remove flow.

Sorry, I obviously missed the part that removed it from
efa_ib_device_remove().

Reviewed-by: Gal Pressman <gal.pressman@linux.dev>

> 
>> -	efa_remove_device(pdev);
>> +	efa_remove_device(pdev, false);
>>  }

Re: [PATCH for-next] RDMA/efa: Reset device on probe failure

[Leon Romanovsky]

On Wed, 25 Dec 2024 13:15:48 +0000, Michael Margolin wrote:
> Make sure the device is being reset on driver exit whatever the reason
> is, to keep the device aligned and allow it to close shared resources
> (e.g. admin queue).
> 
> 

Applied, thanks!

[1/1] RDMA/efa: Reset device on probe failure
      https://git.kernel.org/rdma/rdma/c/123c13f10ed362

Best regards,
-- 
Leon Romanovsky <leon@kernel.org>