RE: [PATCH v6 2/8] net/rnp: add ethdev probe and remove

["11" <caowenbo@mucse.com>]

> -----Original Message-----
> From: Ferruh Yigit <ferruh.yigit@amd.com>
> Sent: 2023年9月5日 23:36
> To: Wenbo Cao <caowenbo@mucse.com>; Anatoly Burakov
> <anatoly.burakov@intel.com>
> Cc: dev@dpdk.org; thomas@monjalon.net; andrew.rybchenko@oktetlabs.ru;
> yaojun@mucse.com
> Subject: Re: [PATCH v6 2/8] net/rnp: add ethdev probe and remove
> 
> On 9/1/2023 3:30 AM, Wenbo Cao wrote:
> > Add basic PCIe ethdev probe and remove.
> >
> > Signed-off-by: Wenbo Cao <caowenbo@mucse.com>
> > ---
> >  drivers/net/rnp/rnp.h        | 13 ++++++
> >  drivers/net/rnp/rnp_ethdev.c | 83
> > ++++++++++++++++++++++++++++++++++++
> >  2 files changed, 96 insertions(+)
> >  create mode 100644 drivers/net/rnp/rnp.h
> >
> > diff --git a/drivers/net/rnp/rnp.h b/drivers/net/rnp/rnp.h new file
> > mode 100644 index 0000000000..76d281cc0a
> > --- /dev/null
> > +++ b/drivers/net/rnp/rnp.h
> > @@ -0,0 +1,13 @@
> > +/* SPDX-License-Identifier: BSD-3-Clause
> > + * Copyright(C) 2023 Mucse IC Design Ltd.
> > + */
> > +#ifndef __RNP_H__
> > +#define __RNP_H__
> > +
> > +#define PCI_VENDOR_ID_MUCSE	(0x8848)
> > +#define RNP_DEV_ID_N10G		(0x1000)
> > +
> > +struct rnp_eth_port {
> > +} __rte_cache_aligned;
> > +
> 
> Is the struct needs to be cache aligned?
> 
This struct may use frequently so I want  it to fast to load.
> > +#endif /* __RNP_H__ */
> > diff --git a/drivers/net/rnp/rnp_ethdev.c
> > b/drivers/net/rnp/rnp_ethdev.c index 9ce3c0b497..390f2e7743 100644
> > --- a/drivers/net/rnp/rnp_ethdev.c
> > +++ b/drivers/net/rnp/rnp_ethdev.c
> > @@ -1,3 +1,86 @@
> >  /* SPDX-License-Identifier: BSD-3-Clause
> >   * Copyright(C) 2023 Mucse IC Design Ltd.
> >   */
> > +
> > +#include <ethdev_pci.h>
> > +#include <rte_io.h>
> > +#include <rte_malloc.h>
> > +
> > +#include "rnp.h"
> > +
> > +static int
> > +rnp_eth_dev_init(struct rte_eth_dev *eth_dev) {
> > +	RTE_SET_USED(eth_dev);
> > +
> > +	return -ENODEV;
> > +}
> > +
> > +static int
> > +rnp_eth_dev_uninit(struct rte_eth_dev *eth_dev) {
> > +	RTE_SET_USED(eth_dev);
> > +
> > +	return -ENODEV;
> > +}
> > +
> > +static int
> > +rnp_pci_remove(struct rte_pci_device *pci_dev) {
> > +	struct rte_eth_dev *eth_dev;
> > +	int rc;
> > +
> > +	eth_dev = rte_eth_dev_allocated(pci_dev->device.name);
> > +
> > +	if (eth_dev) {
> > +		/* Cleanup eth dev */
> > +		rc = rte_eth_dev_pci_generic_remove(pci_dev,
> > +				rnp_eth_dev_uninit);
> > +		if (rc)
> > +			return rc;
> > +	}
> > +	/* Nothing to be done for secondary processes */
> > +	if (rte_eal_process_type() != RTE_PROC_PRIMARY)
> > +		return 0;
> > +
> 
> Primary also don't do anything after this stage, I suggest adding this code when
> primary & secondary diverges.
> 
Ok, I get it.
> > +	return 0;
> > +}
> > +
> > +static int
> > +rnp_pci_probe(struct rte_pci_driver *pci_drv, struct rte_pci_device
> > +*pci_dev) {
> > +	int rc;
> > +
> > +	RTE_SET_USED(pci_drv);
> > +
> > +	rc = rte_eth_dev_pci_generic_probe(pci_dev, sizeof(struct
> rnp_eth_port),
> > +					   rnp_eth_dev_init);
> > +
> > +	/* On error on secondary, recheck if port exists in primary or
> > +	 * in mid of detach state.
> > +	 */
> > +	if (rte_eal_process_type() != RTE_PROC_PRIMARY && rc)
> > +		if (!rte_eth_dev_allocated(pci_dev->device.name))
> > +			return 0;
> 
> Why this additional secondary check is required?
> 'rte_eth_dev_pci_generic_probe()' should be dealing with primary/secondary
> process case.
> 
Because of multiple port mode single pcie-bdf the other port
Alloc use xx:xx.0_0, xx:xx.0_1, xx:xx.0_2, xx:xx.0_0,  use rnp_alloc_eth_port api.