Message-ID: <18111eee-3b3b-4cfe-98be-ab5f93b0aafb@intel.com>
Date: Wed, 24 Sep 2025 16:52:09 +0200
From: Alexander Lobakin <aleksander.lobakin@...el.com>
To: Pavan Kumar Linga <pavan.kumar.linga@...el.com>
CC: <intel-wired-lan@...ts.osuosl.org>, <madhu.chittim@...el.com>,
	<netdev@...r.kernel.org>, Sridhar Samudrala <sridhar.samudrala@...el.com>,
	Simon Horman <horms@...nel.org>
Subject: Re: [Intel-wired-lan] [PATCH net-next v4] idpf: add support for IDPF
 PCI programming interface

From: Pavan Kumar Linga <pavan.kumar.linga@...el.com>
Date: Tue,  2 Sep 2025 20:58:52 -0700

> At present IDPF supports only 0x1452 and 0x145C as PF and VF device IDs
> on our current generation hardware. Future hardware exposes a new set of
> device IDs for each generation. To avoid adding a new device ID for each
> generation and to make the driver forward and backward compatible,
> make use of the IDPF PCI programming interface to load the driver.
> 
> Write and read the VF_ARQBAL mailbox register to find if the current
> device is a PF or a VF.
> 
> PCI SIG allocated a new programming interface for the IDPF compliant
> ethernet network controller devices. It can be found at:
> https://members.pcisig.com/wg/PCI-SIG/document/20113
> with the document titled as 'PCI Code and ID Assignment Revision 1.16'
> or any latest revisions.
> 
> Tested this patch by doing a simple driver load/unload on Intel IPU E2000
> hardware which supports 0x1452 and 0x145C device IDs and new hardware
> which supports the IDPF PCI programming interface.
> 
> Reviewed-by: Sridhar Samudrala <sridhar.samudrala@...el.com>
> Signed-off-by: Madhu Chittim <madhu.chittim@...el.com>
> Reviewed-by: Simon Horman <horms@...nel.org>
> Signed-off-by: Pavan Kumar Linga <pavan.kumar.linga@...el.com>
> ---
> v4:
> - add testing info
> - use resource_size_t instead of long
> - add error statement for ioremap failure
> 
> v3:
> - reworked logic to avoid gotos
> 
> v2:
> - replace *u8 with *bool in idpf_is_vf_device function parameter
> - use ~0 instead of 0xffffff in PCI_DEVICE_CLASS parameter
> 
> ---
>  drivers/net/ethernet/intel/idpf/idpf.h        |  1 +
>  drivers/net/ethernet/intel/idpf/idpf_main.c   | 73 ++++++++++++++-----
>  drivers/net/ethernet/intel/idpf/idpf_vf_dev.c | 40 ++++++++++
>  3 files changed, 97 insertions(+), 17 deletions(-)
> 
> diff --git a/drivers/net/ethernet/intel/idpf/idpf.h b/drivers/net/ethernet/intel/idpf/idpf.h
> index c56abf8b4c92..4a16e481faf7 100644
> --- a/drivers/net/ethernet/intel/idpf/idpf.h
> +++ b/drivers/net/ethernet/intel/idpf/idpf.h
> @@ -1041,6 +1041,7 @@ void idpf_mbx_task(struct work_struct *work);
>  void idpf_vc_event_task(struct work_struct *work);
>  void idpf_dev_ops_init(struct idpf_adapter *adapter);
>  void idpf_vf_dev_ops_init(struct idpf_adapter *adapter);
> +int idpf_is_vf_device(struct pci_dev *pdev, bool *is_vf);
>  int idpf_intr_req(struct idpf_adapter *adapter);
>  void idpf_intr_rel(struct idpf_adapter *adapter);
>  u16 idpf_get_max_tx_hdr_size(struct idpf_adapter *adapter);
> diff --git a/drivers/net/ethernet/intel/idpf/idpf_main.c b/drivers/net/ethernet/intel/idpf/idpf_main.c
> index 8c46481d2e1f..493604d50143 100644
> --- a/drivers/net/ethernet/intel/idpf/idpf_main.c
> +++ b/drivers/net/ethernet/intel/idpf/idpf_main.c
> @@ -7,11 +7,57 @@
>  
>  #define DRV_SUMMARY	"Intel(R) Infrastructure Data Path Function Linux Driver"
>  
> +#define IDPF_NETWORK_ETHERNET_PROGIF				0x01
> +#define IDPF_CLASS_NETWORK_ETHERNET_PROGIF			\
> +	(PCI_CLASS_NETWORK_ETHERNET << 8 | IDPF_NETWORK_ETHERNET_PROGIF)
> +
>  MODULE_DESCRIPTION(DRV_SUMMARY);
>  MODULE_IMPORT_NS("LIBETH");
>  MODULE_IMPORT_NS("LIBETH_XDP");
>  MODULE_LICENSE("GPL");
>  
> +/**
> + * idpf_dev_init - Initialize device specific parameters
> + * @adapter: adapter to initialize
> + * @ent: entry in idpf_pci_tbl
> + *
> + * Return: %0 on success, -%errno on failure.
> + */
> +static int idpf_dev_init(struct idpf_adapter *adapter,
> +			 const struct pci_device_id *ent)
> +{
> +	bool is_vf = false;
> +	int err;
> +
> +	if (ent->class == IDPF_CLASS_NETWORK_ETHERNET_PROGIF) {
> +		err = idpf_is_vf_device(adapter->pdev, &is_vf);
> +		if (err)
> +			return err;
> +		if (is_vf) {
> +			idpf_vf_dev_ops_init(adapter);
> +			adapter->crc_enable = true;
> +		} else {
> +			idpf_dev_ops_init(adapter);
> +		}
> +
> +		return 0;
> +	}
> +
> +	switch (ent->device) {
> +	case IDPF_DEV_ID_PF:
> +		idpf_dev_ops_init(adapter);
> +		break;
> +	case IDPF_DEV_ID_VF:
> +		idpf_vf_dev_ops_init(adapter);
> +		adapter->crc_enable = true;
> +		break;
> +	default:
> +		return -ENODEV;
> +	}
> +
> +	return 0;
> +}
> +
>  /**
>   * idpf_remove - Device removal routine
>   * @pdev: PCI device information struct
> @@ -165,21 +211,6 @@ static int idpf_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
>  	adapter->req_tx_splitq = true;
>  	adapter->req_rx_splitq = true;
>  
> -	switch (ent->device) {
> -	case IDPF_DEV_ID_PF:
> -		idpf_dev_ops_init(adapter);
> -		break;
> -	case IDPF_DEV_ID_VF:
> -		idpf_vf_dev_ops_init(adapter);
> -		adapter->crc_enable = true;
> -		break;
> -	default:
> -		err = -ENODEV;
> -		dev_err(&pdev->dev, "Unexpected dev ID 0x%x in idpf probe\n",
> -			ent->device);
> -		goto err_free;
> -	}
> -
>  	adapter->pdev = pdev;
>  	err = pcim_enable_device(pdev);
>  	if (err)
> @@ -259,11 +290,18 @@ static int idpf_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
>  	/* setup msglvl */
>  	adapter->msg_enable = netif_msg_init(-1, IDPF_AVAIL_NETIF_M);
>  
> +	err = idpf_dev_init(adapter, ent);
> +	if (err) {
> +		dev_err(&pdev->dev, "Unexpected dev ID 0x%x in idpf probe\n",
> +			ent->device);
> +		goto destroy_vc_event_wq;
> +	}
> +
>  	err = idpf_cfg_hw(adapter);
>  	if (err) {
>  		dev_err(dev, "Failed to configure HW structure for adapter: %d\n",
>  			err);
> -		goto err_cfg_hw;
> +		goto destroy_vc_event_wq;
>  	}
>  
>  	mutex_init(&adapter->vport_ctrl_lock);
> @@ -284,7 +322,7 @@ static int idpf_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
>  
>  	return 0;
>  
> -err_cfg_hw:
> +destroy_vc_event_wq:
>  	destroy_workqueue(adapter->vc_event_wq);
>  err_vc_event_wq_alloc:
>  	destroy_workqueue(adapter->stats_wq);
> @@ -304,6 +342,7 @@ static int idpf_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
>  static const struct pci_device_id idpf_pci_tbl[] = {
>  	{ PCI_VDEVICE(INTEL, IDPF_DEV_ID_PF)},
>  	{ PCI_VDEVICE(INTEL, IDPF_DEV_ID_VF)},
> +	{ PCI_DEVICE_CLASS(IDPF_CLASS_NETWORK_ETHERNET_PROGIF, ~0)},
>  	{ /* Sentinel */ }
>  };
>  MODULE_DEVICE_TABLE(pci, idpf_pci_tbl);
> diff --git a/drivers/net/ethernet/intel/idpf/idpf_vf_dev.c b/drivers/net/ethernet/intel/idpf/idpf_vf_dev.c
> index 7527b967e2e7..4774b933ae50 100644
> --- a/drivers/net/ethernet/intel/idpf/idpf_vf_dev.c
> +++ b/drivers/net/ethernet/intel/idpf/idpf_vf_dev.c
> @@ -7,6 +7,46 @@
>  
>  #define IDPF_VF_ITR_IDX_SPACING		0x40
>  
> +#define IDPF_VF_TEST_VAL		0xFEED0000

0xfeed0000U

> +
> +/**
> + * idpf_is_vf_device - Helper to find if it is a VF device
> + * @pdev: PCI device information struct
> + * @is_vf: used to update VF device status
> + *
> + * Return: %0 on success, -%errno on failure.
> + */
> +int idpf_is_vf_device(struct pci_dev *pdev, bool *is_vf)
> +{
> +	struct resource mbx_region;
> +	resource_size_t mbx_start;
> +	void __iomem *mbx_addr;
> +	resource_size_t len;
> +
> +	resource_set_range(&mbx_region,	VF_BASE, IDPF_VF_MBX_REGION_SZ);

Why use `struct resource` at all here. You  have start and len already.
Encapsulating them in a resource only complicates the code.

	mbx_addr = ioremap(pci_resource_start(pdev, 0) + VF_BASE,
			   IDPF_VF_MBX_REGION_SZ);

> +
> +	mbx_start = pci_resource_start(pdev, 0) + mbx_region.start;
> +	len = resource_size(&mbx_region);
> +
> +	mbx_addr = ioremap(mbx_start, len);
> +	if (!mbx_addr) {
> +		pci_err(pdev, "Failed to allocate BAR0 mbx region\n");
> +
> +		return -EIO;

I'd remove this newline.

> +	}
> +
> +	writel(IDPF_VF_TEST_VAL, mbx_addr + VF_ARQBAL - VF_BASE);
> +
> +	/* Force memory write to complete before reading it back */
> +	wmb();

Make sure you know what you are doing. writel() is not writel_relaxed(),
it already has a barrier inside.

> +
> +	*is_vf = readl(mbx_addr + VF_ARQBAL - VF_BASE) == IDPF_VF_TEST_VAL;

Maybe put this `mbx_addr + VF_ARQBAL - VF_BASE` in a variable to not
spell it out two times...

Also weird logic. You don't map the whole BAR, you map only a piece of
it. Ok. But going this route, you need to read only one register. Why
not then

	addr = ioremap(pci_resource_start(pdev, 0) + VF_ARQBAL, 4);

	writel(IDPF_VF_TEST_VAL, addr);
	is_vf = readl(addr) == IDPF_VF_TEST_VAL;

?

> +
> +	iounmap(mbx_addr);
> +
> +	return 0;
> +}

Instead of returning only either 0 or -EIO and passing the result via a
pointer to bool, you could instead just return either:

* IDPF_DEV_ID_PF
* IDPF_DEV_ID_VF
* -EIO

> +
>  /**
>   * idpf_vf_ctlq_reg_init - initialize default mailbox registers
>   * @adapter: adapter structure

Thanks,
Olek

Powered by blists - more mailing lists