[PATCH] [SCSI] qla2xxx: Handle the INTx not connected while passing through

[PATCH] [SCSI] qla2xxx: Handle the INTx not connected while passing through

[Shivaprasad G Bhat]

The PCI_INTERRUPT_PIN reports if the device supports the INTx.
However, when the device is assigned to a guest via vfio, the
PCI_INTERRUPT_PIN is set to 0(i.e none) if the line is not
connected and|or the platform cannot route the interrupt.

In such cases, the guest PCI_INTERRUPT_PIN is 0 and the port
number becomes -1(255, uint8_t underflow) for qla[25|27|28]xx and
qla2031 devices. The flt_region_nvram is never set, and subsequently
the lun detection fails. Below warnings show the NVRAM configuration
failure.

 []-0073:1: Inconsistent NVRAM checksum=0xffffffc0 id=HCAM version=0x100.
 []-0074:1: Falling back to functioning (yet invalid -- WWPN) defaults.
 []-0076:1: NVRAM configuration failed.

The patch handles the case, and sets the port_no to devfn like
its done everywhere else.

Reference: commit 2bd42b03ab6b ("vfio/pci: Virtualize zero INTx PIN if no pdev->irq")
Signed-off-by: Shivaprasad G Bhat <sbhat@linux.ibm.com>
---
 drivers/scsi/qla2xxx/qla_os.c |   15 ++++++++++-----
 1 file changed, 10 insertions(+), 5 deletions(-)

diff --git a/drivers/scsi/qla2xxx/qla_os.c b/drivers/scsi/qla2xxx/qla_os.c
index 72b1c28e4dae..a8d6a0a021f4 100644
--- a/drivers/scsi/qla2xxx/qla_os.c
+++ b/drivers/scsi/qla2xxx/qla_os.c
@@ -2803,11 +2803,16 @@ qla2x00_set_isp_flags(struct qla_hw_data *ha)
 	else {
 		/* Get adapter physical port no from interrupt pin register. */
 		pci_read_config_byte(ha->pdev, PCI_INTERRUPT_PIN, &ha->port_no);
-		if (IS_QLA25XX(ha) || IS_QLA2031(ha) ||
-		    IS_QLA27XX(ha) || IS_QLA28XX(ha))
-			ha->port_no--;
-		else
-			ha->port_no = !(ha->port_no & 1);
+		if (ha->port_no == 0) {
+			/* None of INT[A|B|C|D], may be virtualized by vfio */
+			ha->port_no = PCI_FUNC(ha->pdev->devfn);
+		} else {
+			if (IS_QLA25XX(ha) || IS_QLA2031(ha) ||
+			    IS_QLA27XX(ha) || IS_QLA28XX(ha))
+				ha->port_no--;
+			else
+				ha->port_no = !(ha->port_no & 1);
+		}
 	}
 
 	ql_dbg_pci(ql_dbg_init, ha->pdev, 0x000b,

Fwd: [PATCH] [SCSI] qla2xxx: Handle the INTx not connected while passing through

[priyama2]

Hi Shiva,

I have tested this patch and can confirm it resolves the issue. Could 
you please add the following tags to the patch in v2:

Reported-by: Priya A <priyama2@in.ibm.com>
Tested-by: Priya A <priyama2@in.ibm.com>


Thanks,
Priya A



-------- Forwarded Message --------
Subject: 	[PATCH] [SCSI] qla2xxx: Handle the INTx not connected while 
passing through
Date: 	Fri, 15 May 2026 13:45:18 +0000
From: 	Shivaprasad G Bhat <sbhat@linux.ibm.com>
To: 	njavali@marvell.com, GR-QLogic-Storage-Upstream@marvell.com, 
James.Bottomley@HansenPartnership.com, martin.petersen@oracle.com
CC: 	linux-scsi@vger.kernel.org, linux-kernel@vger.kernel.org, 
alex.williamson@nvidia.com, Kyle.Mahlkuch@ibm.com, sbhat@linux.ibm.com



The PCI_INTERRUPT_PIN reports if the device supports the INTx.
However, when the device is assigned to a guest via vfio, the
PCI_INTERRUPT_PIN is set to 0(i.e none) if the line is not
connected and|or the platform cannot route the interrupt.

In such cases, the guest PCI_INTERRUPT_PIN is 0 and the port
number becomes -1(255, uint8_t underflow) for qla[25|27|28]xx and
qla2031 devices. The flt_region_nvram is never set, and subsequently
the lun detection fails. Below warnings show the NVRAM configuration
failure.

[]-0073:1: Inconsistent NVRAM checksum=0xffffffc0 id=HCAM version=0x100.
[]-0074:1: Falling back to functioning (yet invalid -- WWPN) defaults.
[]-0076:1: NVRAM configuration failed.

The patch handles the case, and sets the port_no to devfn like
its done everywhere else.

Reference: commit 2bd42b03ab6b ("vfio/pci: Virtualize zero INTx PIN if 
no pdev->irq")
Signed-off-by: Shivaprasad G Bhat <sbhat@linux.ibm.com>
---
drivers/scsi/qla2xxx/qla_os.c | 15 ++++++++++-----
1 file changed, 10 insertions(+), 5 deletions(-)

diff --git a/drivers/scsi/qla2xxx/qla_os.c b/drivers/scsi/qla2xxx/qla_os.c
index 72b1c28e4dae..a8d6a0a021f4 100644
--- a/drivers/scsi/qla2xxx/qla_os.c
+++ b/drivers/scsi/qla2xxx/qla_os.c
@@ -2803,11 +2803,16 @@ qla2x00_set_isp_flags(struct qla_hw_data *ha)
else {
/* Get adapter physical port no from interrupt pin register. */
pci_read_config_byte(ha->pdev, PCI_INTERRUPT_PIN, &ha->port_no);
- if (IS_QLA25XX(ha) || IS_QLA2031(ha) ||
- IS_QLA27XX(ha) || IS_QLA28XX(ha))
- ha->port_no--;
- else
- ha->port_no = !(ha->port_no & 1);
+ if (ha->port_no == 0) {
+ /* None of INT[A|B|C|D], may be virtualized by vfio */
+ ha->port_no = PCI_FUNC(ha->pdev->devfn);
+ } else {
+ if (IS_QLA25XX(ha) || IS_QLA2031(ha) ||
+ IS_QLA27XX(ha) || IS_QLA28XX(ha))
+ ha->port_no--;
+ else
+ ha->port_no = !(ha->port_no & 1);
+ }
}
ql_dbg_pci(ql_dbg_init, ha->pdev, 0x000b,

Re: [PATCH] [SCSI] qla2xxx: Handle the INTx not connected while passing through

[Madhavan Srinivasan]

On 5/15/26 7:15 PM, Shivaprasad G Bhat wrote:
> The PCI_INTERRUPT_PIN reports if the device supports the INTx.
> However, when the device is assigned to a guest via vfio, the
> PCI_INTERRUPT_PIN is set to 0(i.e none) if the line is not
> connected and|or the platform cannot route the interrupt.
>
> In such cases, the guest PCI_INTERRUPT_PIN is 0 and the port
> number becomes -1(255, uint8_t underflow) for qla[25|27|28]xx and
> qla2031 devices. The flt_region_nvram is never set, and subsequently
> the lun detection fails. Below warnings show the NVRAM configuration
> failure.
>
>   []-0073:1: Inconsistent NVRAM checksum=0xffffffc0 id=HCAM version=0x100.
>   []-0074:1: Falling back to functioning (yet invalid -- WWPN) defaults.
>   []-0076:1: NVRAM configuration failed.
>
> The patch handles the case, and sets the port_no to devfn like
> its done everywhere else.

Any update on this? do you have any comments/concerns that should be 
addressed

Maddy

> Reference: commit 2bd42b03ab6b ("vfio/pci: Virtualize zero INTx PIN if no pdev->irq")
> Signed-off-by: Shivaprasad G Bhat <sbhat@linux.ibm.com>
> ---
>   drivers/scsi/qla2xxx/qla_os.c |   15 ++++++++++-----
>   1 file changed, 10 insertions(+), 5 deletions(-)
>
> diff --git a/drivers/scsi/qla2xxx/qla_os.c b/drivers/scsi/qla2xxx/qla_os.c
> index 72b1c28e4dae..a8d6a0a021f4 100644
> --- a/drivers/scsi/qla2xxx/qla_os.c
> +++ b/drivers/scsi/qla2xxx/qla_os.c
> @@ -2803,11 +2803,16 @@ qla2x00_set_isp_flags(struct qla_hw_data *ha)
>   	else {
>   		/* Get adapter physical port no from interrupt pin register. */
>   		pci_read_config_byte(ha->pdev, PCI_INTERRUPT_PIN, &ha->port_no);
> -		if (IS_QLA25XX(ha) || IS_QLA2031(ha) ||
> -		    IS_QLA27XX(ha) || IS_QLA28XX(ha))
> -			ha->port_no--;
> -		else
> -			ha->port_no = !(ha->port_no & 1);
> +		if (ha->port_no == 0) {
> +			/* None of INT[A|B|C|D], may be virtualized by vfio */
> +			ha->port_no = PCI_FUNC(ha->pdev->devfn);
> +		} else {
> +			if (IS_QLA25XX(ha) || IS_QLA2031(ha) ||
> +			    IS_QLA27XX(ha) || IS_QLA28XX(ha))
> +				ha->port_no--;
> +			else
> +				ha->port_no = !(ha->port_no & 1);
> +		}
>   	}
>   
>   	ql_dbg_pci(ql_dbg_init, ha->pdev, 0x000b,
>
>

Re: [PATCH] [SCSI] qla2xxx: Handle the INTx not connected while passing through

[Shivaprasad G Bhat]

On 6/3/26 1:42 PM, Madhavan Srinivasan wrote:
>
> On 5/15/26 7:15 PM, Shivaprasad G Bhat wrote:
>> The PCI_INTERRUPT_PIN reports if the device supports the INTx.
>> However, when the device is assigned to a guest via vfio, the
>> PCI_INTERRUPT_PIN is set to 0(i.e none) if the line is not
>> connected and|or the platform cannot route the interrupt.
>>
>> In such cases, the guest PCI_INTERRUPT_PIN is 0 and the port
>> number becomes -1(255, uint8_t underflow) for qla[25|27|28]xx and
>> qla2031 devices. The flt_region_nvram is never set, and subsequently
>> the lun detection fails. Below warnings show the NVRAM configuration
>> failure.
>>
>>   []-0073:1: Inconsistent NVRAM checksum=0xffffffc0 id=HCAM 
>> version=0x100.
>>   []-0074:1: Falling back to functioning (yet invalid -- WWPN) defaults.
>>   []-0076:1: NVRAM configuration failed.
>>
>> The patch handles the case, and sets the port_no to devfn like
>> its done everywhere else.
>
> Any update on this? do you have any comments/concerns that should be 
> addressed
>
Hi Maddy,


Priya has tested this already. I have requested Kyle to help with 
reviewing this patch.

Meanwhile, would request Nilesh, James or Martin to provide feedback, if 
any.


Thanks,

Shivaprasad



> Maddy
>
>> Reference: commit 2bd42b03ab6b ("vfio/pci: Virtualize zero INTx PIN 
>> if no pdev->irq")
>> Signed-off-by: Shivaprasad G Bhat <sbhat@linux.ibm.com>
>> ---
>>   drivers/scsi/qla2xxx/qla_os.c |   15 ++++++++++-----
>>   1 file changed, 10 insertions(+), 5 deletions(-)
>>
>> diff --git a/drivers/scsi/qla2xxx/qla_os.c 
>> b/drivers/scsi/qla2xxx/qla_os.c
>> index 72b1c28e4dae..a8d6a0a021f4 100644
>> --- a/drivers/scsi/qla2xxx/qla_os.c
>> +++ b/drivers/scsi/qla2xxx/qla_os.c
>> @@ -2803,11 +2803,16 @@ qla2x00_set_isp_flags(struct qla_hw_data *ha)
>>       else {
>>           /* Get adapter physical port no from interrupt pin 
>> register. */
>>           pci_read_config_byte(ha->pdev, PCI_INTERRUPT_PIN, 
>> &ha->port_no);
>> -        if (IS_QLA25XX(ha) || IS_QLA2031(ha) ||
>> -            IS_QLA27XX(ha) || IS_QLA28XX(ha))
>> -            ha->port_no--;
>> -        else
>> -            ha->port_no = !(ha->port_no & 1);
>> +        if (ha->port_no == 0) {
>> +            /* None of INT[A|B|C|D], may be virtualized by vfio */
>> +            ha->port_no = PCI_FUNC(ha->pdev->devfn);
>> +        } else {
>> +            if (IS_QLA25XX(ha) || IS_QLA2031(ha) ||
>> +                IS_QLA27XX(ha) || IS_QLA28XX(ha))
>> +                ha->port_no--;
>> +            else
>> +                ha->port_no = !(ha->port_no & 1);
>> +        }
>>       }
>>         ql_dbg_pci(ql_dbg_init, ha->pdev, 0x000b,
>>
>>

Re: [PATCH] [SCSI] qla2xxx: Handle the INTx not connected while passing through

[Kyle Mahlkuch]

> The PCI_INTERRUPT_PIN reports if the device supports the INTx.
> However, when the device is assigned to a guest via vfio, the
> PCI_INTERRUPT_PIN is set to 0(i.e none) if the line is not
> connected and|or the platform cannot route the interrupt.
> 
> In such cases, the guest PCI_INTERRUPT_PIN is 0 and the port
> number becomes -1(255, uint8_t underflow) for qla[25|27|28]xx and
> qla2031 devices. The flt_region_nvram is never set, and subsequently
> the lun detection fails. Below warnings show the NVRAM configuration
> failure.
> 
>   []-0073:1: Inconsistent NVRAM checksum=0xffffffc0 id=HCAM version=0x100.
>   []-0074:1: Falling back to functioning (yet invalid -- WWPN) defaults.
>   []-0076:1: NVRAM configuration failed.
> 
> The patch handles the case, and sets the port_no to devfn like
> its done everywhere else.
> 
> Reference: commit 2bd42b03ab6b ("vfio/pci: Virtualize zero INTx PIN if no pdev->irq")
> Signed-off-by: Shivaprasad G Bhat <sbhat@linux.ibm.com>

Hi Shiva,
Your changes look reasonable and fix a known bug.

Reviewed-by: Kyle Mahlkuch <kmahlkuc@linux.ibm.com>

Re: [PATCH] [SCSI] qla2xxx: Handle the INTx not connected while passing through

[Martin K. Petersen]

On Fri, 15 May 2026 13:45:18 +0000, Shivaprasad G Bhat wrote:

> The PCI_INTERRUPT_PIN reports if the device supports the INTx.
> However, when the device is assigned to a guest via vfio, the
> PCI_INTERRUPT_PIN is set to 0(i.e none) if the line is not
> connected and|or the platform cannot route the interrupt.
> 
> In such cases, the guest PCI_INTERRUPT_PIN is 0 and the port
> number becomes -1(255, uint8_t underflow) for qla[25|27|28]xx and
> qla2031 devices. The flt_region_nvram is never set, and subsequently
> the lun detection fails. Below warnings show the NVRAM configuration
> failure.
> 
> [...]

Applied to 7.3/scsi-queue, thanks!

[1/1] qla2xxx: Handle the INTx not connected while passing through
      https://git.kernel.org/mkp/scsi/c/7268e509b43a

-- 
Martin K. Petersen

Re: [PATCH] [SCSI] qla2xxx: Handle the INTx not connected while passing through

[Shivaprasad G Bhat]

Hi Martin,


On 7/29/26 8:25 AM, Martin K. Petersen wrote:
> On Fri, 15 May 2026 13:45:18 +0000, Shivaprasad G Bhat wrote:
> 
>> The PCI_INTERRUPT_PIN reports if the device supports the INTx.
>> However, when the device is assigned to a guest via vfio, the
>> PCI_INTERRUPT_PIN is set to 0(i.e none) if the line is not
>> connected and|or the platform cannot route the interrupt.
>>
>> In such cases, the guest PCI_INTERRUPT_PIN is 0 and the port
>> number becomes -1(255, uint8_t underflow) for qla[25|27|28]xx and
>> qla2031 devices. The flt_region_nvram is never set, and subsequently
>> the lun detection fails. Below warnings show the NVRAM configuration
>> failure.
>>
>> [...]
> 
> Applied to 7.3/scsi-queue, thanks!

Thanks for pulling the patch. Would it possible to get this fix into 7.2 
stream as well?

Regards,
Shivaprasad
</snip>

Re: [PATCH] [SCSI] qla2xxx: Handle the INTx not connected while passing through

[Martin K. Petersen]

Shivaprasad,

> Thanks for pulling the patch. Would it possible to get this fix into
> 7.2 stream as well?

Due to conflicting with the large qla2xxx update this cycle, I deemed it
best to leave it in 7.3. But you can post a stable backport request once
Linus has merged the SCSI tree.

-- 
Martin K. Petersen