From: Hrvoje Popovski <hrvoje@srce.hr>
Subject: Re: ixv(4) description in dmesg
To: Yuichiro NAITO <naito.yuichiro@gmail.com>, tech@openbsd.org
Date: Fri, 13 Mar 2026 14:08:42 +0100

On 13.3.2026. 11:17, Yuichiro NAITO wrote:
> Oh, I changed the wrong PCI-ID.
> The ixv(4) is a VF driver for the Intel 82599 with a PCI-ID of 0x10ed.
> I updated my patch. Please obsolete the previous one.
> 

Hi,

thank you. Now it's consistent with other VF devices.


ixv0 at pci4 dev 0 function 0 "Intel 82599 VF" rev 0x01, msix, 2 queues,
address 00:0c:29:4f:8c:3d


> diff --git a/sys/dev/pci/if_ixv.c b/sys/dev/pci/if_ixv.c
> index 0dd860ef022..1b3d9e5dd49 100644
> --- a/sys/dev/pci/if_ixv.c
> +++ b/sys/dev/pci/if_ixv.c
> @@ -1313,8 +1313,8 @@ ixv_allocate_msix(struct ix_softc *sc)
>  		pci_conf_write(pa->pa_pc, pa->pa_tag, off, reg | PCI_MSIX_MC_MSIXE);
>  	}
>  
> -	printf(", %s, %d queue%s\n", pci_intr_string(pa->pa_pc, ih),
> -	    i, (i > 1) ? "s" : "");
> +	printf(", %s, %d queue%s, address %s\n", pci_intr_string(pa->pa_pc, ih),
> +	    i, (i > 1) ? "s" : "", ether_sprintf(sc->hw.mac.addr));
>  
>  	return (0);
>  
> diff --git a/sys/dev/pci/pcidevs b/sys/dev/pci/pcidevs
> index 1bd33a74644..7e09ea92e3b 100644
> --- a/sys/dev/pci/pcidevs
> +++ b/sys/dev/pci/pcidevs
> @@ -4091,7 +4091,7 @@ product INTEL 82576_QUAD_COPPER	0x10e8	82576
>  product INTEL 82577LM		0x10ea	82577LM
>  product INTEL 82577LC		0x10eb	82577LC
>  product INTEL 82598EB_CX4_DUAL	0x10ec	82598EB
> -product INTEL 82599VF		0x10ed	82599
> +product INTEL 82599VF		0x10ed	82599 VF
>  product INTEL 82578DM		0x10ef	82578DM
>  product INTEL 82578DC		0x10f0	82578DC
>  product INTEL 82598_DA_DUAL	0x10f1	82598
> 
> 
> On 3/13/26 17:21, Yuichiro NAITO wrote:
>> Hi, 
>>
>> On 3/11/26 08:17, Hrvoje Popovski wrote:
>>> Hi,
>>>
>>> before this diff in dmesg it was
>>> ixv0 at pci4 dev 0 function 0 "Intel 82599" rev 0x01, msix, 2 queues
>>>
>>> after diff
>>> ixv0 at pci4 dev 0 function 0 "Intel 82599 VF" rev 0x01, msix, 2 queues
>>>
>>> while mcx vf and iavf are
>>> mcx0 at pci19 dev 0 function 0 "Mellanox ConnectX-4 Lx VF" rev 0x00: FW
>>> 14.32.2104, msix, 8 queues, address 00:0c:29:4f:8c:3b
>>>
>>> iavf0 at pci27 dev 0 function 0 "Intel XL710/X710 VF" rev 0x01, VF
>>> version 1.1, VF 0 VSI 26, msix, 4 queues, address 00:0c:29:4f:8c:3c
>>>
>>> it would be nice to have mac addess in dmesg for ixv?
>>
>> I feel it's OK to print "82599 VF" in the dmesg.
>> And also printing the MAC address in the ixv probe message is easy. 
>> The `pcidevs.h` and `pcidevs_data.h` are generated from the `pcidevs` file.
>> These two header files don't need to be reviewed.
>> See the `Makefile` in sys/dev/pci directory for details.
>>
>> The following diff merges the printing '82599 VF' and the MAC address.
>>
>> OK?
>>
>> diff --git a/sys/dev/pci/if_ixv.c b/sys/dev/pci/if_ixv.c
>> index 0dd860ef022..1b3d9e5dd49 100644
>> --- a/sys/dev/pci/if_ixv.c
>> +++ b/sys/dev/pci/if_ixv.c
>> @@ -1313,8 +1313,8 @@ ixv_allocate_msix(struct ix_softc *sc)
>>  		pci_conf_write(pa->pa_pc, pa->pa_tag, off, reg | PCI_MSIX_MC_MSIXE);
>>  	}
>>  
>> -	printf(", %s, %d queue%s\n", pci_intr_string(pa->pa_pc, ih),
>> -	    i, (i > 1) ? "s" : "");
>> +	printf(", %s, %d queue%s, address %s\n", pci_intr_string(pa->pa_pc, ih),
>> +	    i, (i > 1) ? "s" : "", ether_sprintf(sc->hw.mac.addr));
>>  
>>  	return (0);
>>  
>> diff --git a/sys/dev/pci/pcidevs b/sys/dev/pci/pcidevs
>> index 1bd33a74644..af836c3f51b 100644
>> --- a/sys/dev/pci/pcidevs
>> +++ b/sys/dev/pci/pcidevs
>> @@ -3968,7 +3968,7 @@ product INTEL 82545GM_COPPER	0x1026	82545GM
>>  product INTEL 82545GM_FIBER	0x1027	82545GM
>>  product INTEL 82545GM_SERDES	0x1028	82545GM
>>  product INTEL PRO_100		0x1029	PRO/100
>> -product INTEL 82559		0x1030	82559
>> +product INTEL 82559		0x1030	82559 VF
>>  product INTEL PRO_100_VE_0	0x1031	PRO/100 VE
>>  product INTEL PRO_100_VE_1	0x1032	PRO/100 VE
>>  product INTEL PRO_100_VM_0	0x1033	PRO/100 VM
>>
>>
>