Message-ID: <abdae94d-10f6-4a31-4131-916606c60363@amd.com>
Date:   Wed, 1 Feb 2023 09:03:10 +0000
From:   "Lucero Palau, Alejandro" <alejandro.lucero-palau@....com>
To:     Jiri Pirko <jiri@...nulli.us>,
        "Lucero Palau, Alejandro" <alejandro.lucero-palau@....com>
CC:     "netdev@...r.kernel.org" <netdev@...r.kernel.org>,
        "linux-net-drivers (AMD-Xilinx)" <linux-net-drivers@....com>,
        "davem@...emloft.net" <davem@...emloft.net>,
        "kuba@...nel.org" <kuba@...nel.org>,
        "pabeni@...hat.com" <pabeni@...hat.com>,
        "edumazet@...gle.com" <edumazet@...gle.com>,
        "habetsm.xilinx@...il.com" <habetsm.xilinx@...il.com>,
        "ecree.xilinx@...il.com" <ecree.xilinx@...il.com>,
        "linux-doc@...r.kernel.org" <linux-doc@...r.kernel.org>,
        "corbet@....net" <corbet@....net>,
        "jiri@...dia.com" <jiri@...dia.com>
Subject: Re: [PATCH v4 net-next 5/8] sfc: add devlink port support for ef100



On 1/31/23 16:12, Jiri Pirko wrote:
> Tue, Jan 31, 2023 at 03:58:19PM CET, alejandro.lucero-palau@....com <mailto:alejandro.lucero-palau@....com> wrote:
>> From: Alejandro Lucero <alejandro.lucero-palau@....com <mailto:alejandro.lucero-palau@....com>>
>>
>> Using the data when enumerating mports, create devlink ports just before
>> netdevs are registered and remove those devlink ports after netdev has
>> been unregistered.
>>
>> Signed-off-by: Alejandro Lucero <alejandro.lucero-palau@....com <mailto:alejandro.lucero-palau@....com>>
>> ---
>> drivers/net/ethernet/sfc/ef100_netdev.c | 9 +++
>> drivers/net/ethernet/sfc/ef100_rep.c | 22 ++++++
>> drivers/net/ethernet/sfc/ef100_rep.h | 7 ++
>> drivers/net/ethernet/sfc/efx_devlink.c | 97 +++++++++++++++++++++++++
>> drivers/net/ethernet/sfc/efx_devlink.h | 6 ++
>> drivers/net/ethernet/sfc/mae.h | 2 +
>> drivers/net/ethernet/sfc/net_driver.h | 2 +
>> 7 files changed, 145 insertions(+)
>>
>> diff --git a/drivers/net/ethernet/sfc/ef100_netdev.c b/drivers/net/ethernet/sfc/ef100_netdev.c
>> index b10a226f4a07..36774b55d413 100644
>> --- a/drivers/net/ethernet/sfc/ef100_netdev.c
>> +++ b/drivers/net/ethernet/sfc/ef100_netdev.c
>> @@ -335,7 +335,9 @@ void ef100_remove_netdev(struct efx_probe_data *probe_data)
>>
>> /* devlink lock */
>> efx_fini_devlink_start(efx);
>> +
> Does not seem related to the patch.
>
>
>
>> ef100_unregister_netdev(efx);
>> + ef100_pf_unset_devlink_port(efx);
>>
>> #ifdef CONFIG_SFC_SRIOV
>> efx_fini_tc(efx);
>> @@ -423,6 +425,8 @@ int ef100_probe_netdev(struct efx_probe_data *probe_data)
>> rc = ef100_probe_netdev_pf(efx);
>> if (rc)
>> goto fail;
>> +
>> + ef100_pf_set_devlink_port(efx);
>> }
>>
>> efx->netdev_notifier.notifier_call = ef100_netdev_event;
>> @@ -433,7 +437,12 @@ int ef100_probe_netdev(struct efx_probe_data *probe_data)
>> goto fail;
>> }
>>
>> + /* devlink unlock */
>> + efx_probe_devlink_done(efx);
>> + return rc;
>> fail:
>> + /* remove devlink port if does exist */
>> + ef100_pf_unset_devlink_port(efx);
>> /* devlink unlock */
>> efx_probe_devlink_done(efx);
>> return rc;
>> diff --git a/drivers/net/ethernet/sfc/ef100_rep.c b/drivers/net/ethernet/sfc/ef100_rep.c
>> index 9cd1a3ac67e0..6b5bc5d6955d 100644
>> --- a/drivers/net/ethernet/sfc/ef100_rep.c
>> +++ b/drivers/net/ethernet/sfc/ef100_rep.c
>> @@ -16,6 +16,7 @@
>> #include "mae.h"
>> #include "rx_common.h"
>> #include "tc_bindings.h"
>> +#include "efx_devlink.h"
>>
>> #define EFX_EF100_REP_DRIVER "efx_ef100_rep"
>>
>> @@ -297,6 +298,7 @@ int efx_ef100_vfrep_create(struct efx_nic *efx, unsigned int i)
>> i, rc);
>> goto fail1;
>> }
>> + ef100_rep_set_devlink_port(efv);
>> rc = register_netdev(efv->net_dev);
>> if (rc) {
>> pci_err(efx->pci_dev,
>> @@ -308,6 +310,7 @@ int efx_ef100_vfrep_create(struct efx_nic *efx, unsigned int i)
>> efv->net_dev->name);
>> return 0;
>> fail2:
>> + ef100_rep_unset_devlink_port(efv);
>> efx_ef100_deconfigure_rep(efv);
>> fail1:
>> efx_ef100_rep_destroy_netdev(efv);
>> @@ -323,6 +326,7 @@ void efx_ef100_vfrep_destroy(struct efx_nic *efx, struct efx_rep *efv)
>> return;
>> netif_dbg(efx, drv, rep_dev, "Removing VF representor\n");
>> unregister_netdev(rep_dev);
>> + ef100_rep_unset_devlink_port(efv);
>> efx_ef100_deconfigure_rep(efv);
>> efx_ef100_rep_destroy_netdev(efv);
>> }
>> @@ -339,6 +343,24 @@ void efx_ef100_fini_vfreps(struct efx_nic *efx)
>> efx_ef100_vfrep_destroy(efx, efv);
>> }
>>
>> +static bool ef100_mport_is_pcie_vnic(struct mae_mport_desc *mport_desc)
>> +{
>> + return mport_desc->mport_type == MAE_MPORT_DESC_MPORT_TYPE_VNIC &&
>> + mport_desc->vnic_client_type == MAE_MPORT_DESC_VNIC_CLIENT_TYPE_FUNCTION;
>> +}
>> +
>> +bool ef100_mport_on_local_intf(struct efx_nic *efx,
>> + struct mae_mport_desc *mport_desc)
>> +{
>> + struct ef100_nic_data *nic_data = efx->nic_data;
>> + bool pcie_func;
>> +
>> + pcie_func = ef100_mport_is_pcie_vnic(mport_desc);
>> +
>> + return nic_data->have_local_intf && pcie_func &&
>> + mport_desc->interface_idx == nic_data->local_mae_intf;
>> +}
>> +
>> void efx_ef100_init_reps(struct efx_nic *efx)
>> {
>> struct ef100_nic_data *nic_data = efx->nic_data;
>> diff --git a/drivers/net/ethernet/sfc/ef100_rep.h b/drivers/net/ethernet/sfc/ef100_rep.h
>> index 328ac0cbb532..ae6add4b0855 100644
>> --- a/drivers/net/ethernet/sfc/ef100_rep.h
>> +++ b/drivers/net/ethernet/sfc/ef100_rep.h
>> @@ -22,6 +22,8 @@ struct efx_rep_sw_stats {
>> atomic64_t rx_dropped, tx_errors;
>> };
>>
>> +struct devlink_port;
>> +
>> /**
>> * struct efx_rep - Private data for an Efx representor
>> *
>> @@ -39,6 +41,7 @@ struct efx_rep_sw_stats {
>> * @rx_lock: protects @rx_list
>> * @napi: NAPI control structure
>> * @stats: software traffic counters for netdev stats
>> + * @dl_port: devlink port associated to this netdev representor
>> */
>> struct efx_rep {
>> struct efx_nic *parent;
>> @@ -54,6 +57,7 @@ struct efx_rep {
>> spinlock_t rx_lock;
>> struct napi_struct napi;
>> struct efx_rep_sw_stats stats;
>> + struct devlink_port *dl_port;
>> };
>>
>> int efx_ef100_vfrep_create(struct efx_nic *efx, unsigned int i);
>> @@ -69,4 +73,7 @@ struct efx_rep *efx_ef100_find_rep_by_mport(struct efx_nic *efx, u16 mport);
>> extern const struct net_device_ops efx_ef100_rep_netdev_ops;
>> void efx_ef100_init_reps(struct efx_nic *efx);
>> void efx_ef100_fini_reps(struct efx_nic *efx);
>> +struct mae_mport_desc;
>> +bool ef100_mport_on_local_intf(struct efx_nic *efx,
>> + struct mae_mport_desc *mport_desc);
>> #endif /* EF100_REP_H */
>> diff --git a/drivers/net/ethernet/sfc/efx_devlink.c b/drivers/net/ethernet/sfc/efx_devlink.c
>> index ff5adfe3905e..b1637eb372ad 100644
>> --- a/drivers/net/ethernet/sfc/efx_devlink.c
>> +++ b/drivers/net/ethernet/sfc/efx_devlink.c
>> @@ -16,11 +16,48 @@
>> #include "mcdi.h"
>> #include "mcdi_functions.h"
>> #include "mcdi_pcol.h"
>> +#include "mae.h"
>> +#include "ef100_rep.h"
>>
>> struct efx_devlink {
>> struct efx_nic *efx;
>> };
>>
>> +static void efx_devlink_del_port(struct devlink_port *dl_port)
>> +{
>> + if (!dl_port)
>> + return;
>> + devl_port_unregister(dl_port);
>> +}
>> +
>> +static int efx_devlink_add_port(struct efx_nic *efx,
>> + struct mae_mport_desc *mport)
>> +{
>> + bool external = false;
>> +
>> + if (!ef100_mport_on_local_intf(efx, mport))
>> + external = true;
>> +
>> + switch (mport->mport_type) {
>> + case MAE_MPORT_DESC_MPORT_TYPE_VNIC:
>> + if (mport->vf_idx != MAE_MPORT_DESC_VF_IDX_NULL)
>> + devlink_port_attrs_pci_vf_set(&mport->dl_port, 0, mport->pf_idx,
>> + mport->vf_idx,
>> + external);
>> + else
>> + devlink_port_attrs_pci_pf_set(&mport->dl_port, 0, mport->pf_idx,
>> + external);
>> + break;
>> + default:
>> + /* MAE_MPORT_DESC_MPORT_ALIAS and UNDEFINED */
>> + return 0;
>> + }
>> +
>> + mport->dl_port.index = mport->mport_id;
>> +
>> + return devl_port_register(efx->devlink, &mport->dl_port, mport->mport_id);
>> +}
>> +
>> static int efx_devlink_info_nvram_partition(struct efx_nic *efx,
>> struct devlink_info_req *req,
>> unsigned int partition_type,
>> @@ -428,6 +465,66 @@ static const struct devlink_ops sfc_devlink_ops = {
>> .info_get = efx_devlink_info_get,
>> };
>>
>> +static struct devlink_port *ef100_set_devlink_port(struct efx_nic *efx, u32 idx)
>> +{
>> + struct mae_mport_desc *mport;
>> + u32 id;
>> +
>> + if (efx_mae_lookup_mport(efx, idx, &id)) {
>> + /* This should not happen. */
>> + if (idx == MAE_MPORT_DESC_VF_IDX_NULL)
>> + pci_warn(efx->pci_dev, "No mport ID found for PF.\n");
>> + else
>> + pci_warn(efx->pci_dev, "No mport ID found for VF %u.\n",
>> + idx);
>> + return NULL;
>> + }
>> +
>> + mport = efx_mae_get_mport(efx, id);
>> + if (!mport) {
>> + /* This should not happen. */
> Should not happen? WARN_ON?

It makes sense.

I will change it.

>
>> + if (idx == MAE_MPORT_DESC_VF_IDX_NULL)
>> + pci_warn(efx->pci_dev, "No mport found for PF.\n");
>> + else
>> + pci_warn(efx->pci_dev, "No mport found for VF %u.\n",
>> + idx);
>> + return NULL;
>> + }
>> +
>> + if (efx_devlink_add_port(efx, mport)) {
> Store the return value into variable please.
>

OK

>> + if (idx == MAE_MPORT_DESC_VF_IDX_NULL)
>> + pci_warn(efx->pci_dev,
>> + "devlink port creation for PF failed.\n");
>> + else
>> + pci_warn(efx->pci_dev,
>> + "devlink_port creationg for VF %u failed.\n",
>> + idx);
>> + return NULL;
>> + }
>> +
>> + return &mport->dl_port;
>> +}
>> +
>> +void ef100_rep_set_devlink_port(struct efx_rep *efv)
>> +{
>> + efv->dl_port = ef100_set_devlink_port(efv->parent, efv->idx);
>> +}
>> +
>> +void ef100_pf_set_devlink_port(struct efx_nic *efx)
>> +{
>> + efx->dl_port = ef100_set_devlink_port(efx, MAE_MPORT_DESC_VF_IDX_NULL);
>> +}
>> +
>> +void ef100_rep_unset_devlink_port(struct efx_rep *efv)
>> +{
>> + efx_devlink_del_port(efv->dl_port);
>> +}
>> +
>> +void ef100_pf_unset_devlink_port(struct efx_nic *efx)
>> +{
>> + efx_devlink_del_port(efx->dl_port);
>> +}
>> +
>> void efx_fini_devlink_start(struct efx_nic *efx)
>> {
>> if (efx->devlink)
>> diff --git a/drivers/net/ethernet/sfc/efx_devlink.h b/drivers/net/ethernet/sfc/efx_devlink.h
>> index 8bcd077d8d8d..d453a180c44c 100644
>> --- a/drivers/net/ethernet/sfc/efx_devlink.h
>> +++ b/drivers/net/ethernet/sfc/efx_devlink.h
>> @@ -36,4 +36,10 @@ void efx_probe_devlink_done(struct efx_nic *efx);
>> void efx_fini_devlink_start(struct efx_nic *efx);
>> void efx_fini_devlink(struct efx_nic *efx);
>>
>> +struct efx_rep;
>> +
>> +void ef100_pf_set_devlink_port(struct efx_nic *efx);
>> +void ef100_rep_set_devlink_port(struct efx_rep *efv);
>> +void ef100_pf_unset_devlink_port(struct efx_nic *efx);
>> +void ef100_rep_unset_devlink_port(struct efx_rep *efv);
>> #endif /* _EFX_DEVLINK_H */
>> diff --git a/drivers/net/ethernet/sfc/mae.h b/drivers/net/ethernet/sfc/mae.h
>> index d9adeafc0654..e1b7967132ad 100644
>> --- a/drivers/net/ethernet/sfc/mae.h
>> +++ b/drivers/net/ethernet/sfc/mae.h
>> @@ -13,6 +13,7 @@
>> #define EF100_MAE_H
>> /* MCDI interface for the ef100 Match-Action Engine */
>>
>> +#include <net/devlink.h>
>> #include "net_driver.h"
>> #include "tc.h"
>> #include "mcdi_pcol.h" /* needed for various MC_CMD_MAE_*_NULL defines */
>> @@ -44,6 +45,7 @@ struct mae_mport_desc {
>> };
>> struct rhash_head linkage;
>> struct efx_rep *efv;
>> + struct devlink_port dl_port;
> This is quite odd to store the devlink port struct in something called
> "match action engine"... Why don't you put it to struct efx_nic
> directly? Much more suitable.
>

Well, it is not called just mae but mae_mport_desc. And mae ports are 
obviously related to devlink ports.

We need to keep mae ports, and not always a mae port/devlink port will 
be related to an efx_nic structure.

If I'm not wrong, devlink was created because not everything in today's 
networking devices can map smoothly to kernel networking structures, not 
because kernel implementation is lacking such representation but because 
it does not make sense semantically. Of course, efx_nic is not a generic kernel
one, but again, we could need devlink ports but not such driver's struct.

>> };
>>
>> int efx_mae_enumerate_mports(struct efx_nic *efx);
>> diff --git a/drivers/net/ethernet/sfc/net_driver.h b/drivers/net/ethernet/sfc/net_driver.h
>> index bc9efbfb3d6b..fcd51d3992fa 100644
>> --- a/drivers/net/ethernet/sfc/net_driver.h
>> +++ b/drivers/net/ethernet/sfc/net_driver.h
>> @@ -998,6 +998,7 @@ struct efx_mae;
>> * @netdev_notifier: Netdevice notifier.
>> * @tc: state for TC offload (EF100).
>> * @devlink: reference to devlink structure owned by this device
>> + * @dl_port: devlink port associated with the PF
>> * @mem_bar: The BAR that is mapped into membase.
>> * @reg_base: Offset from the start of the bar to the function control window.
>> * @monitor_work: Hardware monitor workitem
>> @@ -1185,6 +1186,7 @@ struct efx_nic {
>> struct efx_tc_state *tc;
>>
>> struct devlink *devlink;
>> + struct devlink_port *dl_port;
>> unsigned int mem_bar;
>> u32 reg_base;
>>
>> -- 
>> 2.17.1
>>

Powered by blists - more mailing lists