diff mbox series

[v2,3/9] NTB: Change to use pci_aer_clear_uncorrect_error_status()

Message ID 20220927153524.49172-4-chenzhuo.1@bytedance.com (mailing list archive)
State Superseded
Headers show
Series PCI/AER: Fix and optimize usage of status clearing api | expand

Commit Message

Zhuo Chen Sept. 27, 2022, 3:35 p.m. UTC
Status bits for ERR_NONFATAL errors only are cleared in
pci_aer_clear_nonfatal_status(), but we want clear uncorrectable
error status in idt_init_pci(), so we change to use
pci_aer_clear_uncorrect_error_status().

Signed-off-by: Zhuo Chen <chenzhuo.1@bytedance.com>
---
 drivers/ntb/hw/idt/ntb_hw_idt.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

Comments

Kuppuswamy Sathyanarayanan Sept. 27, 2022, 7:39 p.m. UTC | #1
On 9/27/22 8:35 AM, Zhuo Chen wrote:
> Status bits for ERR_NONFATAL errors only are cleared in
> pci_aer_clear_nonfatal_status(), but we want clear uncorrectable
> error status in idt_init_pci(), so we change to use
> pci_aer_clear_uncorrect_error_status().

You mean currently driver does not clear fatal errors now, and it is
a problem? Any error reported?

Also, I am wondering why is it required to clear errors during init
code. Is it a norm?

> 
> Signed-off-by: Zhuo Chen <chenzhuo.1@bytedance.com>
> ---
>  drivers/ntb/hw/idt/ntb_hw_idt.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/ntb/hw/idt/ntb_hw_idt.c b/drivers/ntb/hw/idt/ntb_hw_idt.c
> index 0ed6f809ff2e..d5f0aa87f817 100644
> --- a/drivers/ntb/hw/idt/ntb_hw_idt.c
> +++ b/drivers/ntb/hw/idt/ntb_hw_idt.c
> @@ -2657,8 +2657,8 @@ static int idt_init_pci(struct idt_ntb_dev *ndev)
>  	ret = pci_enable_pcie_error_reporting(pdev);
>  	if (ret != 0)
>  		dev_warn(&pdev->dev, "PCIe AER capability disabled\n");
> -	else /* Cleanup nonfatal error status before getting to init */
> -		pci_aer_clear_nonfatal_status(pdev);
> +	else /* Cleanup uncorrectable error status before getting to init */
> +		pci_aer_clear_uncorrect_error_status(pdev);
>  
>  	/* First enable the PCI device */
>  	ret = pcim_enable_device(pdev);
Zhuo Chen Sept. 28, 2022, 4:20 a.m. UTC | #2
On 9/28/22 3:39 AM, Sathyanarayanan Kuppuswamy wrote:
> 
> 
> On 9/27/22 8:35 AM, Zhuo Chen wrote:
>> Status bits for ERR_NONFATAL errors only are cleared in
>> pci_aer_clear_nonfatal_status(), but we want clear uncorrectable
>> error status in idt_init_pci(), so we change to use
>> pci_aer_clear_uncorrect_error_status().
> 
> You mean currently driver does not clear fatal errors now, and it is
> a problem? Any error reported?
> 
Hi Sathyanarayanan,

No error reports yet, I just changes the behavior back to what it was 
before commit e7b0b847de6d ("PCI/AER: Clear only ERR_NONFATAL bits 
during non-fatal recovery"), because this commit change the original 
function in commit bf2a952d31d2 ("NTB: Add IDT 89HPESxNTx PCIe-switches 
support").

> Also, I am wondering why is it required to clear errors during init
> code. Is it a norm?
> 
I think there is no need to clear errors during init code.
>>
>> Signed-off-by: Zhuo Chen <chenzhuo.1@bytedance.com>
>> ---
>>   drivers/ntb/hw/idt/ntb_hw_idt.c | 4 ++--
>>   1 file changed, 2 insertions(+), 2 deletions(-)
>>
>> diff --git a/drivers/ntb/hw/idt/ntb_hw_idt.c b/drivers/ntb/hw/idt/ntb_hw_idt.c
>> index 0ed6f809ff2e..d5f0aa87f817 100644
>> --- a/drivers/ntb/hw/idt/ntb_hw_idt.c
>> +++ b/drivers/ntb/hw/idt/ntb_hw_idt.c
>> @@ -2657,8 +2657,8 @@ static int idt_init_pci(struct idt_ntb_dev *ndev)
>>   	ret = pci_enable_pcie_error_reporting(pdev);
>>   	if (ret != 0)
>>   		dev_warn(&pdev->dev, "PCIe AER capability disabled\n");
>> -	else /* Cleanup nonfatal error status before getting to init */
>> -		pci_aer_clear_nonfatal_status(pdev);
>> +	else /* Cleanup uncorrectable error status before getting to init */
>> +		pci_aer_clear_uncorrect_error_status(pdev);
>>   
>>   	/* First enable the PCI device */
>>   	ret = pcim_enable_device(pdev);
>
Kuppuswamy Sathyanarayanan Sept. 28, 2022, 4:30 a.m. UTC | #3
On 9/27/22 9:20 PM, Zhuo Chen wrote:
> 
> 
> On 9/28/22 3:39 AM, Sathyanarayanan Kuppuswamy wrote:
>>
>>
>> On 9/27/22 8:35 AM, Zhuo Chen wrote:
>>> Status bits for ERR_NONFATAL errors only are cleared in
>>> pci_aer_clear_nonfatal_status(), but we want clear uncorrectable
>>> error status in idt_init_pci(), so we change to use
>>> pci_aer_clear_uncorrect_error_status().
>>
>> You mean currently driver does not clear fatal errors now, and it is
>> a problem? Any error reported?
>>
> Hi Sathyanarayanan,
> 
> No error reports yet, I just changes the behavior back to what it was before commit e7b0b847de6d ("PCI/AER: Clear only ERR_NONFATAL bits during non-fatal recovery"), because this commit change the original function in commit bf2a952d31d2 ("NTB: Add IDT 89HPESxNTx PCIe-switches support").
> 

Ok. Thanks for clarifying.

>> Also, I am wondering why is it required to clear errors during init
>> code. Is it a norm?
>>
> I think there is no need to clear errors during init code.
>>>
>>> Signed-off-by: Zhuo Chen <chenzhuo.1@bytedance.com>
>>> ---
>>>   drivers/ntb/hw/idt/ntb_hw_idt.c | 4 ++--
>>>   1 file changed, 2 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/drivers/ntb/hw/idt/ntb_hw_idt.c b/drivers/ntb/hw/idt/ntb_hw_idt.c
>>> index 0ed6f809ff2e..d5f0aa87f817 100644
>>> --- a/drivers/ntb/hw/idt/ntb_hw_idt.c
>>> +++ b/drivers/ntb/hw/idt/ntb_hw_idt.c
>>> @@ -2657,8 +2657,8 @@ static int idt_init_pci(struct idt_ntb_dev *ndev)
>>>       ret = pci_enable_pcie_error_reporting(pdev);
>>>       if (ret != 0)
>>>           dev_warn(&pdev->dev, "PCIe AER capability disabled\n");
>>> -    else /* Cleanup nonfatal error status before getting to init */
>>> -        pci_aer_clear_nonfatal_status(pdev);
>>> +    else /* Cleanup uncorrectable error status before getting to init */
>>> +        pci_aer_clear_uncorrect_error_status(pdev);
>>>         /* First enable the PCI device */
>>>       ret = pcim_enable_device(pdev);
>>
>
Serge Semin Sept. 28, 2022, 11 a.m. UTC | #4
On Tue, Sep 27, 2022 at 11:35:18PM +0800, Zhuo Chen wrote:
> Status bits for ERR_NONFATAL errors only are cleared in
> pci_aer_clear_nonfatal_status(), but we want clear uncorrectable
> error status in idt_init_pci(), so we change to use
> pci_aer_clear_uncorrect_error_status().

Have the modification changed since:
https://lore.kernel.org/linux-pci/20220901181634.99591-2-chenzhuo.1@bytedance.com/
?
AFAICS it hasn't. Then my ab-tag should have been preserved on v2.
One more time:
Acked-by: Serge Semin <fancer.lancer@gmail.com>

Don't forget to add it should you have some more patchset re-spins.

-Sergey

> 
> Signed-off-by: Zhuo Chen <chenzhuo.1@bytedance.com>
> ---
>  drivers/ntb/hw/idt/ntb_hw_idt.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/ntb/hw/idt/ntb_hw_idt.c b/drivers/ntb/hw/idt/ntb_hw_idt.c
> index 0ed6f809ff2e..d5f0aa87f817 100644
> --- a/drivers/ntb/hw/idt/ntb_hw_idt.c
> +++ b/drivers/ntb/hw/idt/ntb_hw_idt.c
> @@ -2657,8 +2657,8 @@ static int idt_init_pci(struct idt_ntb_dev *ndev)
>  	ret = pci_enable_pcie_error_reporting(pdev);
>  	if (ret != 0)
>  		dev_warn(&pdev->dev, "PCIe AER capability disabled\n");
> -	else /* Cleanup nonfatal error status before getting to init */
> -		pci_aer_clear_nonfatal_status(pdev);
> +	else /* Cleanup uncorrectable error status before getting to init */
> +		pci_aer_clear_uncorrect_error_status(pdev);
>  
>  	/* First enable the PCI device */
>  	ret = pcim_enable_device(pdev);
> -- 
> 2.30.1 (Apple Git-130)
>
diff mbox series

Patch

diff --git a/drivers/ntb/hw/idt/ntb_hw_idt.c b/drivers/ntb/hw/idt/ntb_hw_idt.c
index 0ed6f809ff2e..d5f0aa87f817 100644
--- a/drivers/ntb/hw/idt/ntb_hw_idt.c
+++ b/drivers/ntb/hw/idt/ntb_hw_idt.c
@@ -2657,8 +2657,8 @@  static int idt_init_pci(struct idt_ntb_dev *ndev)
 	ret = pci_enable_pcie_error_reporting(pdev);
 	if (ret != 0)
 		dev_warn(&pdev->dev, "PCIe AER capability disabled\n");
-	else /* Cleanup nonfatal error status before getting to init */
-		pci_aer_clear_nonfatal_status(pdev);
+	else /* Cleanup uncorrectable error status before getting to init */
+		pci_aer_clear_uncorrect_error_status(pdev);
 
 	/* First enable the PCI device */
 	ret = pcim_enable_device(pdev);