microsoft
/
WSL2-Linux-Kernel
зеркало из https://github.com/microsoft/WSL2-Linux-Kernel.git
1
0
Форкнуть 0
Код Задачи Пакеты Проекты Релизы Вики Активность
WSL2-Linux-Kernel/drivers/net/ethernet/sfc/ef10_sriov.c

784 строки
19 KiB
C
Исходник Обычный вид История

sfc: Enable VF's via a write to the sysfs file sriov_numvfs This patch adds support for the use of sriov_configure on EF10 to enable Virtual Functions while the driver is loaded. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:55:58 +03:00
/****************************************************************************
* Driver for Solarflare network controllers and boards
* Copyright 2015 Solarflare Communications Inc.
*
* This program is free software; you can redistribute it and/or modify it
* under the terms of the GNU General Public License version 2 as published
* by the Free Software Foundation, incorporated herein by reference.
*/
#include <linux/pci.h>
#include <linux/module.h>
#include "net_driver.h"
#include "ef10_sriov.h"
#include "efx.h"
#include "nic.h"
#include "mcdi_pcol.h"
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
static int efx_ef10_evb_port_assign(struct efx_nic *efx, unsigned int port_id,
unsigned int vf_fn)
sfc: Enable VF's via a write to the sysfs file sriov_numvfs This patch adds support for the use of sriov_configure on EF10 to enable Virtual Functions while the driver is loaded. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:55:58 +03:00
{
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
MCDI_DECLARE_BUF(inbuf, MC_CMD_EVB_PORT_ASSIGN_IN_LEN);
struct efx_ef10_nic_data *nic_data = efx->nic_data;
sfc: Enable VF's via a write to the sysfs file sriov_numvfs This patch adds support for the use of sriov_configure on EF10 to enable Virtual Functions while the driver is loaded. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:55:58 +03:00
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
MCDI_SET_DWORD(inbuf, EVB_PORT_ASSIGN_IN_PORT_ID, port_id);
MCDI_POPULATE_DWORD_2(inbuf, EVB_PORT_ASSIGN_IN_FUNCTION,
EVB_PORT_ASSIGN_IN_PF, nic_data->pf_index,
EVB_PORT_ASSIGN_IN_VF, vf_fn);
return efx_mcdi_rpc(efx, MC_CMD_EVB_PORT_ASSIGN, inbuf, sizeof(inbuf),
NULL, 0, NULL);
sfc: Enable VF's via a write to the sysfs file sriov_numvfs This patch adds support for the use of sriov_configure on EF10 to enable Virtual Functions while the driver is loaded. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:55:58 +03:00
}
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
static int efx_ef10_vport_add_mac(struct efx_nic *efx,
unsigned int port_id, u8 *mac)
sfc: Enable VF's via a write to the sysfs file sriov_numvfs This patch adds support for the use of sriov_configure on EF10 to enable Virtual Functions while the driver is loaded. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:55:58 +03:00
{
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
MCDI_DECLARE_BUF(inbuf, MC_CMD_VPORT_ADD_MAC_ADDRESS_IN_LEN);
sfc: Enable VF's via a write to the sysfs file sriov_numvfs This patch adds support for the use of sriov_configure on EF10 to enable Virtual Functions while the driver is loaded. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:55:58 +03:00
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
MCDI_SET_DWORD(inbuf, VPORT_ADD_MAC_ADDRESS_IN_VPORT_ID, port_id);
ether_addr_copy(MCDI_PTR(inbuf, VPORT_ADD_MAC_ADDRESS_IN_MACADDR), mac);
sfc: Enable VF's via a write to the sysfs file sriov_numvfs This patch adds support for the use of sriov_configure on EF10 to enable Virtual Functions while the driver is loaded. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:55:58 +03:00
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
return efx_mcdi_rpc(efx, MC_CMD_VPORT_ADD_MAC_ADDRESS, inbuf,
sizeof(inbuf), NULL, 0, NULL);
sfc: Enable VF's via a write to the sysfs file sriov_numvfs This patch adds support for the use of sriov_configure on EF10 to enable Virtual Functions while the driver is loaded. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:55:58 +03:00
}
sfc: create VEB vswitch and vport above default firmware setup Adds functions to allocate and free vswitches and vports; vadaptors are automatically allocated and freed when TX/RX queues are initialised and finalised. This vswitching structure is only created if the firmware supports it, so a check that full-featured firmware is running is performed first. If the MC resets, the vswitching infrastructure will need to be recreated, so mark the "must_probe_vswitching" flag when an MC reboot is detected. Don't try to create a vswitch if vf-count=0 This allocation of vswitches and vports does not currently support configuring VLAN tags, but that can be added in a future change. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:57:34 +03:00
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
static int efx_ef10_vport_del_mac(struct efx_nic *efx,
unsigned int port_id, u8 *mac)
sfc: Prepare to bind the sfc driver to the VF. Added efx_nic_type structure for VF. Mapped a different BAR for VF as it uses BAR 0 for memory. Added functions sriov_init and sriov_fini. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:14 +03:00
{
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
MCDI_DECLARE_BUF(inbuf, MC_CMD_VPORT_DEL_MAC_ADDRESS_IN_LEN);
sfc: Prepare to bind the sfc driver to the VF. Added efx_nic_type structure for VF. Mapped a different BAR for VF as it uses BAR 0 for memory. Added functions sriov_init and sriov_fini. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:14 +03:00
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
MCDI_SET_DWORD(inbuf, VPORT_DEL_MAC_ADDRESS_IN_VPORT_ID, port_id);
ether_addr_copy(MCDI_PTR(inbuf, VPORT_DEL_MAC_ADDRESS_IN_MACADDR), mac);
sfc: Prepare to bind the sfc driver to the VF. Added efx_nic_type structure for VF. Mapped a different BAR for VF as it uses BAR 0 for memory. Added functions sriov_init and sriov_fini. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:14 +03:00
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
return efx_mcdi_rpc(efx, MC_CMD_VPORT_DEL_MAC_ADDRESS, inbuf,
sizeof(inbuf), NULL, 0, NULL);
sfc: Prepare to bind the sfc driver to the VF. Added efx_nic_type structure for VF. Mapped a different BAR for VF as it uses BAR 0 for memory. Added functions sriov_init and sriov_fini. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:14 +03:00
}
sfc: create VEB vswitch and vport above default firmware setup Adds functions to allocate and free vswitches and vports; vadaptors are automatically allocated and freed when TX/RX queues are initialised and finalised. This vswitching structure is only created if the firmware supports it, so a check that full-featured firmware is running is performed first. If the MC resets, the vswitching infrastructure will need to be recreated, so mark the "must_probe_vswitching" flag when an MC reboot is detected. Don't try to create a vswitch if vf-count=0 This allocation of vswitches and vports does not currently support configuring VLAN tags, but that can be added in a future change. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:57:34 +03:00
static int efx_ef10_vswitch_alloc(struct efx_nic *efx, unsigned int port_id,
unsigned int vswitch_type)
{
MCDI_DECLARE_BUF(inbuf, MC_CMD_VSWITCH_ALLOC_IN_LEN);
sfc: add ndo_set_vf_vlan() function for EF10 The max vlan tags that can be offloaded is 2, including any upstream VLAN aggregator. Currently there is no way for the net driver to know whether the upstream vswitch (if any) is using vlan tags, so there is no way to know how many tags we can request. Along with the implementation for the ndo_set_vf_vlan callback, this patch also adds 2 VLAN tags for the driver created VEB switch if possible, that way it is possible to offload as many tags as are allowed. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-20 13:11:54 +03:00
int rc;
sfc: create VEB vswitch and vport above default firmware setup Adds functions to allocate and free vswitches and vports; vadaptors are automatically allocated and freed when TX/RX queues are initialised and finalised. This vswitching structure is only created if the firmware supports it, so a check that full-featured firmware is running is performed first. If the MC resets, the vswitching infrastructure will need to be recreated, so mark the "must_probe_vswitching" flag when an MC reboot is detected. Don't try to create a vswitch if vf-count=0 This allocation of vswitches and vports does not currently support configuring VLAN tags, but that can be added in a future change. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:57:34 +03:00
MCDI_SET_DWORD(inbuf, VSWITCH_ALLOC_IN_UPSTREAM_PORT_ID, port_id);
MCDI_SET_DWORD(inbuf, VSWITCH_ALLOC_IN_TYPE, vswitch_type);
sfc: add ndo_set_vf_vlan() function for EF10 The max vlan tags that can be offloaded is 2, including any upstream VLAN aggregator. Currently there is no way for the net driver to know whether the upstream vswitch (if any) is using vlan tags, so there is no way to know how many tags we can request. Along with the implementation for the ndo_set_vf_vlan callback, this patch also adds 2 VLAN tags for the driver created VEB switch if possible, that way it is possible to offload as many tags as are allowed. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-20 13:11:54 +03:00
MCDI_SET_DWORD(inbuf, VSWITCH_ALLOC_IN_NUM_VLAN_TAGS, 2);
sfc: create VEB vswitch and vport above default firmware setup Adds functions to allocate and free vswitches and vports; vadaptors are automatically allocated and freed when TX/RX queues are initialised and finalised. This vswitching structure is only created if the firmware supports it, so a check that full-featured firmware is running is performed first. If the MC resets, the vswitching infrastructure will need to be recreated, so mark the "must_probe_vswitching" flag when an MC reboot is detected. Don't try to create a vswitch if vf-count=0 This allocation of vswitches and vports does not currently support configuring VLAN tags, but that can be added in a future change. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:57:34 +03:00
MCDI_POPULATE_DWORD_1(inbuf, VSWITCH_ALLOC_IN_FLAGS,
VSWITCH_ALLOC_IN_FLAG_AUTO_PORT, 0);
sfc: add ndo_set_vf_vlan() function for EF10 The max vlan tags that can be offloaded is 2, including any upstream VLAN aggregator. Currently there is no way for the net driver to know whether the upstream vswitch (if any) is using vlan tags, so there is no way to know how many tags we can request. Along with the implementation for the ndo_set_vf_vlan callback, this patch also adds 2 VLAN tags for the driver created VEB switch if possible, that way it is possible to offload as many tags as are allowed. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-20 13:11:54 +03:00
/* Quietly try to allocate 2 VLAN tags */
rc = efx_mcdi_rpc_quiet(efx, MC_CMD_VSWITCH_ALLOC, inbuf, sizeof(inbuf),
NULL, 0, NULL);
/* If 2 VLAN tags is too many, revert to trying with 1 VLAN tags */
if (rc == -EPROTO) {
MCDI_SET_DWORD(inbuf, VSWITCH_ALLOC_IN_NUM_VLAN_TAGS, 1);
rc = efx_mcdi_rpc(efx, MC_CMD_VSWITCH_ALLOC, inbuf,
sizeof(inbuf), NULL, 0, NULL);
} else if (rc) {
efx_mcdi_display_error(efx, MC_CMD_VSWITCH_ALLOC,
MC_CMD_VSWITCH_ALLOC_IN_LEN,
NULL, 0, rc);
}
return rc;
sfc: create VEB vswitch and vport above default firmware setup Adds functions to allocate and free vswitches and vports; vadaptors are automatically allocated and freed when TX/RX queues are initialised and finalised. This vswitching structure is only created if the firmware supports it, so a check that full-featured firmware is running is performed first. If the MC resets, the vswitching infrastructure will need to be recreated, so mark the "must_probe_vswitching" flag when an MC reboot is detected. Don't try to create a vswitch if vf-count=0 This allocation of vswitches and vports does not currently support configuring VLAN tags, but that can be added in a future change. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:57:34 +03:00
}
static int efx_ef10_vswitch_free(struct efx_nic *efx, unsigned int port_id)
{
MCDI_DECLARE_BUF(inbuf, MC_CMD_VSWITCH_FREE_IN_LEN);
MCDI_SET_DWORD(inbuf, VSWITCH_FREE_IN_UPSTREAM_PORT_ID, port_id);
return efx_mcdi_rpc(efx, MC_CMD_VSWITCH_FREE, inbuf, sizeof(inbuf),
NULL, 0, NULL);
}
static int efx_ef10_vport_alloc(struct efx_nic *efx,
unsigned int port_id_in,
unsigned int vport_type,
sfc: add ndo_set_vf_vlan() function for EF10 The max vlan tags that can be offloaded is 2, including any upstream VLAN aggregator. Currently there is no way for the net driver to know whether the upstream vswitch (if any) is using vlan tags, so there is no way to know how many tags we can request. Along with the implementation for the ndo_set_vf_vlan callback, this patch also adds 2 VLAN tags for the driver created VEB switch if possible, that way it is possible to offload as many tags as are allowed. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-20 13:11:54 +03:00
u16 vlan,
sfc: create VEB vswitch and vport above default firmware setup Adds functions to allocate and free vswitches and vports; vadaptors are automatically allocated and freed when TX/RX queues are initialised and finalised. This vswitching structure is only created if the firmware supports it, so a check that full-featured firmware is running is performed first. If the MC resets, the vswitching infrastructure will need to be recreated, so mark the "must_probe_vswitching" flag when an MC reboot is detected. Don't try to create a vswitch if vf-count=0 This allocation of vswitches and vports does not currently support configuring VLAN tags, but that can be added in a future change. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:57:34 +03:00
unsigned int *port_id_out)
{
MCDI_DECLARE_BUF(inbuf, MC_CMD_VPORT_ALLOC_IN_LEN);
MCDI_DECLARE_BUF(outbuf, MC_CMD_VPORT_ALLOC_OUT_LEN);
size_t outlen;
int rc;
EFX_WARN_ON_PARANOID(!port_id_out);
MCDI_SET_DWORD(inbuf, VPORT_ALLOC_IN_UPSTREAM_PORT_ID, port_id_in);
MCDI_SET_DWORD(inbuf, VPORT_ALLOC_IN_TYPE, vport_type);
sfc: add ndo_set_vf_vlan() function for EF10 The max vlan tags that can be offloaded is 2, including any upstream VLAN aggregator. Currently there is no way for the net driver to know whether the upstream vswitch (if any) is using vlan tags, so there is no way to know how many tags we can request. Along with the implementation for the ndo_set_vf_vlan callback, this patch also adds 2 VLAN tags for the driver created VEB switch if possible, that way it is possible to offload as many tags as are allowed. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-20 13:11:54 +03:00
MCDI_SET_DWORD(inbuf, VPORT_ALLOC_IN_NUM_VLAN_TAGS,
(vlan != EFX_EF10_NO_VLAN));
sfc: create VEB vswitch and vport above default firmware setup Adds functions to allocate and free vswitches and vports; vadaptors are automatically allocated and freed when TX/RX queues are initialised and finalised. This vswitching structure is only created if the firmware supports it, so a check that full-featured firmware is running is performed first. If the MC resets, the vswitching infrastructure will need to be recreated, so mark the "must_probe_vswitching" flag when an MC reboot is detected. Don't try to create a vswitch if vf-count=0 This allocation of vswitches and vports does not currently support configuring VLAN tags, but that can be added in a future change. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:57:34 +03:00
MCDI_POPULATE_DWORD_1(inbuf, VPORT_ALLOC_IN_FLAGS,
VPORT_ALLOC_IN_FLAG_AUTO_PORT, 0);
sfc: add ndo_set_vf_vlan() function for EF10 The max vlan tags that can be offloaded is 2, including any upstream VLAN aggregator. Currently there is no way for the net driver to know whether the upstream vswitch (if any) is using vlan tags, so there is no way to know how many tags we can request. Along with the implementation for the ndo_set_vf_vlan callback, this patch also adds 2 VLAN tags for the driver created VEB switch if possible, that way it is possible to offload as many tags as are allowed. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-20 13:11:54 +03:00
if (vlan != EFX_EF10_NO_VLAN)
MCDI_POPULATE_DWORD_1(inbuf, VPORT_ALLOC_IN_VLAN_TAGS,
VPORT_ALLOC_IN_VLAN_TAG_0, vlan);
sfc: create VEB vswitch and vport above default firmware setup Adds functions to allocate and free vswitches and vports; vadaptors are automatically allocated and freed when TX/RX queues are initialised and finalised. This vswitching structure is only created if the firmware supports it, so a check that full-featured firmware is running is performed first. If the MC resets, the vswitching infrastructure will need to be recreated, so mark the "must_probe_vswitching" flag when an MC reboot is detected. Don't try to create a vswitch if vf-count=0 This allocation of vswitches and vports does not currently support configuring VLAN tags, but that can be added in a future change. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:57:34 +03:00
rc = efx_mcdi_rpc(efx, MC_CMD_VPORT_ALLOC, inbuf, sizeof(inbuf),
outbuf, sizeof(outbuf), &outlen);
if (rc)
return rc;
if (outlen < MC_CMD_VPORT_ALLOC_OUT_LEN)
return -EIO;
*port_id_out = MCDI_DWORD(outbuf, VPORT_ALLOC_OUT_VPORT_ID);
return 0;
}
static int efx_ef10_vport_free(struct efx_nic *efx, unsigned int port_id)
{
MCDI_DECLARE_BUF(inbuf, MC_CMD_VPORT_FREE_IN_LEN);
MCDI_SET_DWORD(inbuf, VPORT_FREE_IN_VPORT_ID, port_id);
return efx_mcdi_rpc(efx, MC_CMD_VPORT_FREE, inbuf, sizeof(inbuf),
NULL, 0, NULL);
}
sfc: manually allocate and free vadaptors To be able to use MC_CMD_VADAPTOR_SET_MAC, vadaptors must be manually allocated and freed as automatic vadaptors will disappear when their reference_count reaches zero, which must happen before the MAC address is changed. Vadaptors are allocated and freed in the vswitching_probe/remove functions for PFs and VFs, and this means that vadaptors are restored correctly following an MC reboot or other reset when required. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:54 +03:00
static int efx_ef10_vadaptor_alloc(struct efx_nic *efx, unsigned int port_id)
{
MCDI_DECLARE_BUF(inbuf, MC_CMD_VADAPTOR_ALLOC_IN_LEN);
MCDI_SET_DWORD(inbuf, VADAPTOR_ALLOC_IN_UPSTREAM_PORT_ID, port_id);
return efx_mcdi_rpc(efx, MC_CMD_VADAPTOR_ALLOC, inbuf, sizeof(inbuf),
NULL, 0, NULL);
}
static int efx_ef10_vadaptor_free(struct efx_nic *efx, unsigned int port_id)
{
MCDI_DECLARE_BUF(inbuf, MC_CMD_VADAPTOR_FREE_IN_LEN);
MCDI_SET_DWORD(inbuf, VADAPTOR_FREE_IN_UPSTREAM_PORT_ID, port_id);
return efx_mcdi_rpc(efx, MC_CMD_VADAPTOR_FREE, inbuf, sizeof(inbuf),
NULL, 0, NULL);
}
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
static void efx_ef10_sriov_free_vf_vports(struct efx_nic *efx)
{
struct efx_ef10_nic_data *nic_data = efx->nic_data;
int i;
if (!nic_data->vf)
return;
for (i = 0; i < efx->vf_count; i++) {
struct ef10_vf *vf = nic_data->vf + i;
sfc: leak vports if a VF is assigned during PF unload If any VF is assigned as the PF is unloaded, do not attempt to remove its vport or the vswitch. These will be removed if the driver binds to the PF again, as an entity reset occurs during probe. A 'force' flag is added to efx_ef10_pci_sriov_disable() to distinguish between disabling SR-IOV and driver unload. SR-IOV cannot be disabled if VFs are assigned to guests. If the PF driver is unloaded while VFs are assigned, the driver may try to bind to the VF again at a later point if the driver has been reloaded and the VF returns to the same domain as the PF. In this case, the PF will not have a VF data structure, so the VF can check this and drop out of probe early. In this case, efx->vf_count will be zero but VFs will be present. The user is advised to remove the VF and re-create it. The check at the beginning of efx_ef10_pci_sriov_disable() that efx->vf_count is non-zero is removed to allow SR-IOV to be disabled in this case. Also, if the PF driver is unloaded, it will disable SR-IOV to remove these unknown VFs. By not disabling bus-mastering if VFs are still assigned, the VF will continue to pass traffic after the PF has been removed. When using the max_vfs module parameter, if VFs are already present do not try to initialise any more. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-02 13:41:00 +03:00
/* If VF is assigned, do not free the vport */
if (vf->pci_dev &&
vf->pci_dev->dev_flags & PCI_DEV_FLAGS_ASSIGNED)
continue;
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
if (vf->vport_assigned) {
efx_ef10_evb_port_assign(efx, EVB_PORT_ID_NULL, i);
vf->vport_assigned = 0;
}
if (!is_zero_ether_addr(vf->mac)) {
efx_ef10_vport_del_mac(efx, vf->vport_id, vf->mac);
eth_zero_addr(vf->mac);
}
if (vf->vport_id) {
efx_ef10_vport_free(efx, vf->vport_id);
vf->vport_id = 0;
}
sfc: Store the efx_nic struct of the current VF in the VF data struct Initialised in efx_probe_vf and removal is dealt with in efx_ef10_remove. vf->efx is needed in future patches to change the MAC address of the VF via the parent PF, while the driver is bound to the VF. Example: ip link set dev vf NUM mac LLADDR Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-20 13:09:46 +03:00
vf->efx = NULL;
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
}
}
static void efx_ef10_sriov_free_vf_vswitching(struct efx_nic *efx)
{
struct efx_ef10_nic_data *nic_data = efx->nic_data;
efx_ef10_sriov_free_vf_vports(efx);
kfree(nic_data->vf);
nic_data->vf = NULL;
}
static int efx_ef10_sriov_assign_vf_vport(struct efx_nic *efx,
unsigned int vf_i)
{
struct efx_ef10_nic_data *nic_data = efx->nic_data;
struct ef10_vf *vf = nic_data->vf + vf_i;
int rc;
if (WARN_ON_ONCE(!nic_data->vf))
return -EOPNOTSUPP;
rc = efx_ef10_vport_alloc(efx, EVB_PORT_ID_ASSIGNED,
MC_CMD_VPORT_ALLOC_IN_VPORT_TYPE_NORMAL,
sfc: add ndo_set_vf_vlan() function for EF10 The max vlan tags that can be offloaded is 2, including any upstream VLAN aggregator. Currently there is no way for the net driver to know whether the upstream vswitch (if any) is using vlan tags, so there is no way to know how many tags we can request. Along with the implementation for the ndo_set_vf_vlan callback, this patch also adds 2 VLAN tags for the driver created VEB switch if possible, that way it is possible to offload as many tags as are allowed. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-20 13:11:54 +03:00
vf->vlan, &vf->vport_id);
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
if (rc)
return rc;
rc = efx_ef10_vport_add_mac(efx, vf->vport_id, vf->mac);
if (rc) {
eth_zero_addr(vf->mac);
return rc;
}
rc = efx_ef10_evb_port_assign(efx, vf->vport_id, vf_i);
if (rc)
return rc;
vf->vport_assigned = 1;
return 0;
}
static int efx_ef10_sriov_alloc_vf_vswitching(struct efx_nic *efx)
{
struct efx_ef10_nic_data *nic_data = efx->nic_data;
unsigned int i;
int rc;
nic_data->vf = kcalloc(efx->vf_count, sizeof(struct ef10_vf),
GFP_KERNEL);
if (!nic_data->vf)
return -ENOMEM;
for (i = 0; i < efx->vf_count; i++) {
random_ether_addr(nic_data->vf[i].mac);
sfc: Store the efx_nic struct of the current VF in the VF data struct Initialised in efx_probe_vf and removal is dealt with in efx_ef10_remove. vf->efx is needed in future patches to change the MAC address of the VF via the parent PF, while the driver is bound to the VF. Example: ip link set dev vf NUM mac LLADDR Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-20 13:09:46 +03:00
nic_data->vf[i].efx = NULL;
sfc: add ndo_set_vf_vlan() function for EF10 The max vlan tags that can be offloaded is 2, including any upstream VLAN aggregator. Currently there is no way for the net driver to know whether the upstream vswitch (if any) is using vlan tags, so there is no way to know how many tags we can request. Along with the implementation for the ndo_set_vf_vlan callback, this patch also adds 2 VLAN tags for the driver created VEB switch if possible, that way it is possible to offload as many tags as are allowed. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-20 13:11:54 +03:00
nic_data->vf[i].vlan = EFX_EF10_NO_VLAN;
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
rc = efx_ef10_sriov_assign_vf_vport(efx, i);
if (rc)
goto fail;
}
return 0;
fail:
efx_ef10_sriov_free_vf_vports(efx);
kfree(nic_data->vf);
nic_data->vf = NULL;
return rc;
}
static int efx_ef10_sriov_restore_vf_vswitching(struct efx_nic *efx)
{
unsigned int i;
int rc;
for (i = 0; i < efx->vf_count; i++) {
rc = efx_ef10_sriov_assign_vf_vport(efx, i);
if (rc)
goto fail;
}
return 0;
fail:
efx_ef10_sriov_free_vf_vswitching(efx);
return rc;
}
sfc: create VEB vswitch and vport above default firmware setup Adds functions to allocate and free vswitches and vports; vadaptors are automatically allocated and freed when TX/RX queues are initialised and finalised. This vswitching structure is only created if the firmware supports it, so a check that full-featured firmware is running is performed first. If the MC resets, the vswitching infrastructure will need to be recreated, so mark the "must_probe_vswitching" flag when an MC reboot is detected. Don't try to create a vswitch if vf-count=0 This allocation of vswitches and vports does not currently support configuring VLAN tags, but that can be added in a future change. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:57:34 +03:00
/* On top of the default firmware vswitch setup, create a VEB vswitch and
* expansion vport for use by this function.
*/
sfc: manually allocate and free vadaptors To be able to use MC_CMD_VADAPTOR_SET_MAC, vadaptors must be manually allocated and freed as automatic vadaptors will disappear when their reference_count reaches zero, which must happen before the MAC address is changed. Vadaptors are allocated and freed in the vswitching_probe/remove functions for PFs and VFs, and this means that vadaptors are restored correctly following an MC reboot or other reset when required. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:54 +03:00
int efx_ef10_vswitching_probe_pf(struct efx_nic *efx)
sfc: create VEB vswitch and vport above default firmware setup Adds functions to allocate and free vswitches and vports; vadaptors are automatically allocated and freed when TX/RX queues are initialised and finalised. This vswitching structure is only created if the firmware supports it, so a check that full-featured firmware is running is performed first. If the MC resets, the vswitching infrastructure will need to be recreated, so mark the "must_probe_vswitching" flag when an MC reboot is detected. Don't try to create a vswitch if vf-count=0 This allocation of vswitches and vports does not currently support configuring VLAN tags, but that can be added in a future change. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:57:34 +03:00
{
struct efx_ef10_nic_data *nic_data = efx->nic_data;
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
struct net_device *net_dev = efx->net_dev;
sfc: create VEB vswitch and vport above default firmware setup Adds functions to allocate and free vswitches and vports; vadaptors are automatically allocated and freed when TX/RX queues are initialised and finalised. This vswitching structure is only created if the firmware supports it, so a check that full-featured firmware is running is performed first. If the MC resets, the vswitching infrastructure will need to be recreated, so mark the "must_probe_vswitching" flag when an MC reboot is detected. Don't try to create a vswitch if vf-count=0 This allocation of vswitches and vports does not currently support configuring VLAN tags, but that can be added in a future change. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:57:34 +03:00
int rc;
sfc: manually allocate and free vadaptors To be able to use MC_CMD_VADAPTOR_SET_MAC, vadaptors must be manually allocated and freed as automatic vadaptors will disappear when their reference_count reaches zero, which must happen before the MAC address is changed. Vadaptors are allocated and freed in the vswitching_probe/remove functions for PFs and VFs, and this means that vadaptors are restored correctly following an MC reboot or other reset when required. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:54 +03:00
if (pci_sriov_get_totalvfs(efx->pci_dev) <= 0) {
/* vswitch not needed as we have no VFs */
efx_ef10_vadaptor_alloc(efx, nic_data->vport_id);
return 0;
}
sfc: create VEB vswitch and vport above default firmware setup Adds functions to allocate and free vswitches and vports; vadaptors are automatically allocated and freed when TX/RX queues are initialised and finalised. This vswitching structure is only created if the firmware supports it, so a check that full-featured firmware is running is performed first. If the MC resets, the vswitching infrastructure will need to be recreated, so mark the "must_probe_vswitching" flag when an MC reboot is detected. Don't try to create a vswitch if vf-count=0 This allocation of vswitches and vports does not currently support configuring VLAN tags, but that can be added in a future change. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:57:34 +03:00
rc = efx_ef10_vswitch_alloc(efx, EVB_PORT_ID_ASSIGNED,
MC_CMD_VSWITCH_ALLOC_IN_VSWITCH_TYPE_VEB);
if (rc)
goto fail1;
rc = efx_ef10_vport_alloc(efx, EVB_PORT_ID_ASSIGNED,
MC_CMD_VPORT_ALLOC_IN_VPORT_TYPE_NORMAL,
sfc: add ndo_set_vf_vlan() function for EF10 The max vlan tags that can be offloaded is 2, including any upstream VLAN aggregator. Currently there is no way for the net driver to know whether the upstream vswitch (if any) is using vlan tags, so there is no way to know how many tags we can request. Along with the implementation for the ndo_set_vf_vlan callback, this patch also adds 2 VLAN tags for the driver created VEB switch if possible, that way it is possible to offload as many tags as are allowed. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-20 13:11:54 +03:00
EFX_EF10_NO_VLAN, &nic_data->vport_id);
sfc: create VEB vswitch and vport above default firmware setup Adds functions to allocate and free vswitches and vports; vadaptors are automatically allocated and freed when TX/RX queues are initialised and finalised. This vswitching structure is only created if the firmware supports it, so a check that full-featured firmware is running is performed first. If the MC resets, the vswitching infrastructure will need to be recreated, so mark the "must_probe_vswitching" flag when an MC reboot is detected. Don't try to create a vswitch if vf-count=0 This allocation of vswitches and vports does not currently support configuring VLAN tags, but that can be added in a future change. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:57:34 +03:00
if (rc)
goto fail2;
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
rc = efx_ef10_vport_add_mac(efx, nic_data->vport_id, net_dev->dev_addr);
if (rc)
goto fail3;
ether_addr_copy(nic_data->vport_mac, net_dev->dev_addr);
sfc: manually allocate and free vadaptors To be able to use MC_CMD_VADAPTOR_SET_MAC, vadaptors must be manually allocated and freed as automatic vadaptors will disappear when their reference_count reaches zero, which must happen before the MAC address is changed. Vadaptors are allocated and freed in the vswitching_probe/remove functions for PFs and VFs, and this means that vadaptors are restored correctly following an MC reboot or other reset when required. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:54 +03:00
rc = efx_ef10_vadaptor_alloc(efx, nic_data->vport_id);
if (rc)
goto fail4;
sfc: create VEB vswitch and vport above default firmware setup Adds functions to allocate and free vswitches and vports; vadaptors are automatically allocated and freed when TX/RX queues are initialised and finalised. This vswitching structure is only created if the firmware supports it, so a check that full-featured firmware is running is performed first. If the MC resets, the vswitching infrastructure will need to be recreated, so mark the "must_probe_vswitching" flag when an MC reboot is detected. Don't try to create a vswitch if vf-count=0 This allocation of vswitches and vports does not currently support configuring VLAN tags, but that can be added in a future change. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:57:34 +03:00
return 0;
sfc: manually allocate and free vadaptors To be able to use MC_CMD_VADAPTOR_SET_MAC, vadaptors must be manually allocated and freed as automatic vadaptors will disappear when their reference_count reaches zero, which must happen before the MAC address is changed. Vadaptors are allocated and freed in the vswitching_probe/remove functions for PFs and VFs, and this means that vadaptors are restored correctly following an MC reboot or other reset when required. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:54 +03:00
fail4:
efx_ef10_vport_del_mac(efx, nic_data->vport_id, nic_data->vport_mac);
eth_zero_addr(nic_data->vport_mac);
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
fail3:
efx_ef10_vport_free(efx, nic_data->vport_id);
nic_data->vport_id = EVB_PORT_ID_ASSIGNED;
sfc: create VEB vswitch and vport above default firmware setup Adds functions to allocate and free vswitches and vports; vadaptors are automatically allocated and freed when TX/RX queues are initialised and finalised. This vswitching structure is only created if the firmware supports it, so a check that full-featured firmware is running is performed first. If the MC resets, the vswitching infrastructure will need to be recreated, so mark the "must_probe_vswitching" flag when an MC reboot is detected. Don't try to create a vswitch if vf-count=0 This allocation of vswitches and vports does not currently support configuring VLAN tags, but that can be added in a future change. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:57:34 +03:00
fail2:
efx_ef10_vswitch_free(efx, EVB_PORT_ID_ASSIGNED);
fail1:
return rc;
}
sfc: manually allocate and free vadaptors To be able to use MC_CMD_VADAPTOR_SET_MAC, vadaptors must be manually allocated and freed as automatic vadaptors will disappear when their reference_count reaches zero, which must happen before the MAC address is changed. Vadaptors are allocated and freed in the vswitching_probe/remove functions for PFs and VFs, and this means that vadaptors are restored correctly following an MC reboot or other reset when required. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:54 +03:00
int efx_ef10_vswitching_probe_vf(struct efx_nic *efx)
{
struct efx_ef10_nic_data *nic_data = efx->nic_data;
return efx_ef10_vadaptor_alloc(efx, nic_data->vport_id);
}
int efx_ef10_vswitching_restore_pf(struct efx_nic *efx)
sfc: create VEB vswitch and vport above default firmware setup Adds functions to allocate and free vswitches and vports; vadaptors are automatically allocated and freed when TX/RX queues are initialised and finalised. This vswitching structure is only created if the firmware supports it, so a check that full-featured firmware is running is performed first. If the MC resets, the vswitching infrastructure will need to be recreated, so mark the "must_probe_vswitching" flag when an MC reboot is detected. Don't try to create a vswitch if vf-count=0 This allocation of vswitches and vports does not currently support configuring VLAN tags, but that can be added in a future change. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:57:34 +03:00
{
struct efx_ef10_nic_data *nic_data = efx->nic_data;
int rc;
if (!nic_data->must_probe_vswitching)
return 0;
sfc: manually allocate and free vadaptors To be able to use MC_CMD_VADAPTOR_SET_MAC, vadaptors must be manually allocated and freed as automatic vadaptors will disappear when their reference_count reaches zero, which must happen before the MAC address is changed. Vadaptors are allocated and freed in the vswitching_probe/remove functions for PFs and VFs, and this means that vadaptors are restored correctly following an MC reboot or other reset when required. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:54 +03:00
rc = efx_ef10_vswitching_probe_pf(efx);
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
if (rc)
goto fail;
rc = efx_ef10_sriov_restore_vf_vswitching(efx);
if (rc)
goto fail;
sfc: create VEB vswitch and vport above default firmware setup Adds functions to allocate and free vswitches and vports; vadaptors are automatically allocated and freed when TX/RX queues are initialised and finalised. This vswitching structure is only created if the firmware supports it, so a check that full-featured firmware is running is performed first. If the MC resets, the vswitching infrastructure will need to be recreated, so mark the "must_probe_vswitching" flag when an MC reboot is detected. Don't try to create a vswitch if vf-count=0 This allocation of vswitches and vports does not currently support configuring VLAN tags, but that can be added in a future change. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:57:34 +03:00
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
nic_data->must_probe_vswitching = false;
fail:
sfc: create VEB vswitch and vport above default firmware setup Adds functions to allocate and free vswitches and vports; vadaptors are automatically allocated and freed when TX/RX queues are initialised and finalised. This vswitching structure is only created if the firmware supports it, so a check that full-featured firmware is running is performed first. If the MC resets, the vswitching infrastructure will need to be recreated, so mark the "must_probe_vswitching" flag when an MC reboot is detected. Don't try to create a vswitch if vf-count=0 This allocation of vswitches and vports does not currently support configuring VLAN tags, but that can be added in a future change. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:57:34 +03:00
return rc;
}
sfc: manually allocate and free vadaptors To be able to use MC_CMD_VADAPTOR_SET_MAC, vadaptors must be manually allocated and freed as automatic vadaptors will disappear when their reference_count reaches zero, which must happen before the MAC address is changed. Vadaptors are allocated and freed in the vswitching_probe/remove functions for PFs and VFs, and this means that vadaptors are restored correctly following an MC reboot or other reset when required. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:54 +03:00
int efx_ef10_vswitching_restore_vf(struct efx_nic *efx)
{
struct efx_ef10_nic_data *nic_data = efx->nic_data;
int rc;
if (!nic_data->must_probe_vswitching)
return 0;
rc = efx_ef10_vadaptor_free(efx, EVB_PORT_ID_ASSIGNED);
if (rc)
return rc;
nic_data->must_probe_vswitching = false;
return 0;
}
void efx_ef10_vswitching_remove_pf(struct efx_nic *efx)
sfc: create VEB vswitch and vport above default firmware setup Adds functions to allocate and free vswitches and vports; vadaptors are automatically allocated and freed when TX/RX queues are initialised and finalised. This vswitching structure is only created if the firmware supports it, so a check that full-featured firmware is running is performed first. If the MC resets, the vswitching infrastructure will need to be recreated, so mark the "must_probe_vswitching" flag when an MC reboot is detected. Don't try to create a vswitch if vf-count=0 This allocation of vswitches and vports does not currently support configuring VLAN tags, but that can be added in a future change. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:57:34 +03:00
{
struct efx_ef10_nic_data *nic_data = efx->nic_data;
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
efx_ef10_sriov_free_vf_vswitching(efx);
sfc: manually allocate and free vadaptors To be able to use MC_CMD_VADAPTOR_SET_MAC, vadaptors must be manually allocated and freed as automatic vadaptors will disappear when their reference_count reaches zero, which must happen before the MAC address is changed. Vadaptors are allocated and freed in the vswitching_probe/remove functions for PFs and VFs, and this means that vadaptors are restored correctly following an MC reboot or other reset when required. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:54 +03:00
efx_ef10_vadaptor_free(efx, nic_data->vport_id);
sfc: create VEB vswitch and vport above default firmware setup Adds functions to allocate and free vswitches and vports; vadaptors are automatically allocated and freed when TX/RX queues are initialised and finalised. This vswitching structure is only created if the firmware supports it, so a check that full-featured firmware is running is performed first. If the MC resets, the vswitching infrastructure will need to be recreated, so mark the "must_probe_vswitching" flag when an MC reboot is detected. Don't try to create a vswitch if vf-count=0 This allocation of vswitches and vports does not currently support configuring VLAN tags, but that can be added in a future change. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:57:34 +03:00
if (nic_data->vport_id == EVB_PORT_ID_ASSIGNED)
return; /* No vswitch was ever created */
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
if (!is_zero_ether_addr(nic_data->vport_mac)) {
efx_ef10_vport_del_mac(efx, nic_data->vport_id,
efx->net_dev->dev_addr);
eth_zero_addr(nic_data->vport_mac);
}
sfc: create VEB vswitch and vport above default firmware setup Adds functions to allocate and free vswitches and vports; vadaptors are automatically allocated and freed when TX/RX queues are initialised and finalised. This vswitching structure is only created if the firmware supports it, so a check that full-featured firmware is running is performed first. If the MC resets, the vswitching infrastructure will need to be recreated, so mark the "must_probe_vswitching" flag when an MC reboot is detected. Don't try to create a vswitch if vf-count=0 This allocation of vswitches and vports does not currently support configuring VLAN tags, but that can be added in a future change. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:57:34 +03:00
efx_ef10_vport_free(efx, nic_data->vport_id);
nic_data->vport_id = EVB_PORT_ID_ASSIGNED;
sfc: leak vports if a VF is assigned during PF unload If any VF is assigned as the PF is unloaded, do not attempt to remove its vport or the vswitch. These will be removed if the driver binds to the PF again, as an entity reset occurs during probe. A 'force' flag is added to efx_ef10_pci_sriov_disable() to distinguish between disabling SR-IOV and driver unload. SR-IOV cannot be disabled if VFs are assigned to guests. If the PF driver is unloaded while VFs are assigned, the driver may try to bind to the VF again at a later point if the driver has been reloaded and the VF returns to the same domain as the PF. In this case, the PF will not have a VF data structure, so the VF can check this and drop out of probe early. In this case, efx->vf_count will be zero but VFs will be present. The user is advised to remove the VF and re-create it. The check at the beginning of efx_ef10_pci_sriov_disable() that efx->vf_count is non-zero is removed to allow SR-IOV to be disabled in this case. Also, if the PF driver is unloaded, it will disable SR-IOV to remove these unknown VFs. By not disabling bus-mastering if VFs are still assigned, the VF will continue to pass traffic after the PF has been removed. When using the max_vfs module parameter, if VFs are already present do not try to initialise any more. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-02 13:41:00 +03:00
/* Only free the vswitch if no VFs are assigned */
if (!pci_vfs_assigned(efx->pci_dev))
efx_ef10_vswitch_free(efx, nic_data->vport_id);
sfc: create VEB vswitch and vport above default firmware setup Adds functions to allocate and free vswitches and vports; vadaptors are automatically allocated and freed when TX/RX queues are initialised and finalised. This vswitching structure is only created if the firmware supports it, so a check that full-featured firmware is running is performed first. If the MC resets, the vswitching infrastructure will need to be recreated, so mark the "must_probe_vswitching" flag when an MC reboot is detected. Don't try to create a vswitch if vf-count=0 This allocation of vswitches and vports does not currently support configuring VLAN tags, but that can be added in a future change. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:57:34 +03:00
}
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
sfc: manually allocate and free vadaptors To be able to use MC_CMD_VADAPTOR_SET_MAC, vadaptors must be manually allocated and freed as automatic vadaptors will disappear when their reference_count reaches zero, which must happen before the MAC address is changed. Vadaptors are allocated and freed in the vswitching_probe/remove functions for PFs and VFs, and this means that vadaptors are restored correctly following an MC reboot or other reset when required. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:54 +03:00
void efx_ef10_vswitching_remove_vf(struct efx_nic *efx)
{
efx_ef10_vadaptor_free(efx, EVB_PORT_ID_ASSIGNED);
}
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
static int efx_ef10_pci_sriov_enable(struct efx_nic *efx, int num_vfs)
{
int rc = 0;
struct pci_dev *dev = efx->pci_dev;
efx->vf_count = num_vfs;
rc = efx_ef10_sriov_alloc_vf_vswitching(efx);
if (rc)
goto fail1;
rc = pci_enable_sriov(dev, num_vfs);
if (rc)
goto fail2;
return 0;
fail2:
efx_ef10_sriov_free_vf_vswitching(efx);
fail1:
efx->vf_count = 0;
netif_err(efx, probe, efx->net_dev,
"Failed to enable SRIOV VFs\n");
return rc;
}
sfc: leak vports if a VF is assigned during PF unload If any VF is assigned as the PF is unloaded, do not attempt to remove its vport or the vswitch. These will be removed if the driver binds to the PF again, as an entity reset occurs during probe. A 'force' flag is added to efx_ef10_pci_sriov_disable() to distinguish between disabling SR-IOV and driver unload. SR-IOV cannot be disabled if VFs are assigned to guests. If the PF driver is unloaded while VFs are assigned, the driver may try to bind to the VF again at a later point if the driver has been reloaded and the VF returns to the same domain as the PF. In this case, the PF will not have a VF data structure, so the VF can check this and drop out of probe early. In this case, efx->vf_count will be zero but VFs will be present. The user is advised to remove the VF and re-create it. The check at the beginning of efx_ef10_pci_sriov_disable() that efx->vf_count is non-zero is removed to allow SR-IOV to be disabled in this case. Also, if the PF driver is unloaded, it will disable SR-IOV to remove these unknown VFs. By not disabling bus-mastering if VFs are still assigned, the VF will continue to pass traffic after the PF has been removed. When using the max_vfs module parameter, if VFs are already present do not try to initialise any more. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-02 13:41:00 +03:00
static int efx_ef10_pci_sriov_disable(struct efx_nic *efx, bool force)
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
{
struct pci_dev *dev = efx->pci_dev;
sfc: leak vports if a VF is assigned during PF unload If any VF is assigned as the PF is unloaded, do not attempt to remove its vport or the vswitch. These will be removed if the driver binds to the PF again, as an entity reset occurs during probe. A 'force' flag is added to efx_ef10_pci_sriov_disable() to distinguish between disabling SR-IOV and driver unload. SR-IOV cannot be disabled if VFs are assigned to guests. If the PF driver is unloaded while VFs are assigned, the driver may try to bind to the VF again at a later point if the driver has been reloaded and the VF returns to the same domain as the PF. In this case, the PF will not have a VF data structure, so the VF can check this and drop out of probe early. In this case, efx->vf_count will be zero but VFs will be present. The user is advised to remove the VF and re-create it. The check at the beginning of efx_ef10_pci_sriov_disable() that efx->vf_count is non-zero is removed to allow SR-IOV to be disabled in this case. Also, if the PF driver is unloaded, it will disable SR-IOV to remove these unknown VFs. By not disabling bus-mastering if VFs are still assigned, the VF will continue to pass traffic after the PF has been removed. When using the max_vfs module parameter, if VFs are already present do not try to initialise any more. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-02 13:41:00 +03:00
unsigned int vfs_assigned = 0;
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
sfc: leak vports if a VF is assigned during PF unload If any VF is assigned as the PF is unloaded, do not attempt to remove its vport or the vswitch. These will be removed if the driver binds to the PF again, as an entity reset occurs during probe. A 'force' flag is added to efx_ef10_pci_sriov_disable() to distinguish between disabling SR-IOV and driver unload. SR-IOV cannot be disabled if VFs are assigned to guests. If the PF driver is unloaded while VFs are assigned, the driver may try to bind to the VF again at a later point if the driver has been reloaded and the VF returns to the same domain as the PF. In this case, the PF will not have a VF data structure, so the VF can check this and drop out of probe early. In this case, efx->vf_count will be zero but VFs will be present. The user is advised to remove the VF and re-create it. The check at the beginning of efx_ef10_pci_sriov_disable() that efx->vf_count is non-zero is removed to allow SR-IOV to be disabled in this case. Also, if the PF driver is unloaded, it will disable SR-IOV to remove these unknown VFs. By not disabling bus-mastering if VFs are still assigned, the VF will continue to pass traffic after the PF has been removed. When using the max_vfs module parameter, if VFs are already present do not try to initialise any more. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-02 13:41:00 +03:00
vfs_assigned = pci_vfs_assigned(dev);
sfc: do not allow VFs to be destroyed if assigned to guests Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-02 13:40:31 +03:00
sfc: leak vports if a VF is assigned during PF unload If any VF is assigned as the PF is unloaded, do not attempt to remove its vport or the vswitch. These will be removed if the driver binds to the PF again, as an entity reset occurs during probe. A 'force' flag is added to efx_ef10_pci_sriov_disable() to distinguish between disabling SR-IOV and driver unload. SR-IOV cannot be disabled if VFs are assigned to guests. If the PF driver is unloaded while VFs are assigned, the driver may try to bind to the VF again at a later point if the driver has been reloaded and the VF returns to the same domain as the PF. In this case, the PF will not have a VF data structure, so the VF can check this and drop out of probe early. In this case, efx->vf_count will be zero but VFs will be present. The user is advised to remove the VF and re-create it. The check at the beginning of efx_ef10_pci_sriov_disable() that efx->vf_count is non-zero is removed to allow SR-IOV to be disabled in this case. Also, if the PF driver is unloaded, it will disable SR-IOV to remove these unknown VFs. By not disabling bus-mastering if VFs are still assigned, the VF will continue to pass traffic after the PF has been removed. When using the max_vfs module parameter, if VFs are already present do not try to initialise any more. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-02 13:41:00 +03:00
if (vfs_assigned && !force) {
netif_info(efx, drv, efx->net_dev, "VFs are assigned to guests; "
"please detach them before disabling SR-IOV\n");
sfc: do not allow VFs to be destroyed if assigned to guests Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-02 13:40:31 +03:00
return -EBUSY;
}
sfc: leak vports if a VF is assigned during PF unload If any VF is assigned as the PF is unloaded, do not attempt to remove its vport or the vswitch. These will be removed if the driver binds to the PF again, as an entity reset occurs during probe. A 'force' flag is added to efx_ef10_pci_sriov_disable() to distinguish between disabling SR-IOV and driver unload. SR-IOV cannot be disabled if VFs are assigned to guests. If the PF driver is unloaded while VFs are assigned, the driver may try to bind to the VF again at a later point if the driver has been reloaded and the VF returns to the same domain as the PF. In this case, the PF will not have a VF data structure, so the VF can check this and drop out of probe early. In this case, efx->vf_count will be zero but VFs will be present. The user is advised to remove the VF and re-create it. The check at the beginning of efx_ef10_pci_sriov_disable() that efx->vf_count is non-zero is removed to allow SR-IOV to be disabled in this case. Also, if the PF driver is unloaded, it will disable SR-IOV to remove these unknown VFs. By not disabling bus-mastering if VFs are still assigned, the VF will continue to pass traffic after the PF has been removed. When using the max_vfs module parameter, if VFs are already present do not try to initialise any more. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-02 13:41:00 +03:00
if (!vfs_assigned)
pci_disable_sriov(dev);
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
efx_ef10_sriov_free_vf_vswitching(efx);
efx->vf_count = 0;
return 0;
}
int efx_ef10_sriov_configure(struct efx_nic *efx, int num_vfs)
{
if (num_vfs == 0)
sfc: leak vports if a VF is assigned during PF unload If any VF is assigned as the PF is unloaded, do not attempt to remove its vport or the vswitch. These will be removed if the driver binds to the PF again, as an entity reset occurs during probe. A 'force' flag is added to efx_ef10_pci_sriov_disable() to distinguish between disabling SR-IOV and driver unload. SR-IOV cannot be disabled if VFs are assigned to guests. If the PF driver is unloaded while VFs are assigned, the driver may try to bind to the VF again at a later point if the driver has been reloaded and the VF returns to the same domain as the PF. In this case, the PF will not have a VF data structure, so the VF can check this and drop out of probe early. In this case, efx->vf_count will be zero but VFs will be present. The user is advised to remove the VF and re-create it. The check at the beginning of efx_ef10_pci_sriov_disable() that efx->vf_count is non-zero is removed to allow SR-IOV to be disabled in this case. Also, if the PF driver is unloaded, it will disable SR-IOV to remove these unknown VFs. By not disabling bus-mastering if VFs are still assigned, the VF will continue to pass traffic after the PF has been removed. When using the max_vfs module parameter, if VFs are already present do not try to initialise any more. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-02 13:41:00 +03:00
return efx_ef10_pci_sriov_disable(efx, false);
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
else
return efx_ef10_pci_sriov_enable(efx, num_vfs);
}
int efx_ef10_sriov_init(struct efx_nic *efx)
{
return 0;
}
void efx_ef10_sriov_fini(struct efx_nic *efx)
{
struct efx_ef10_nic_data *nic_data = efx->nic_data;
sfc: force removal of VF and vport on driver removal When the driver unloads, force the unbind and removal of any VFs in the host with the PF. The PF cannot remove vports and vswitches if they are still being used by a VF driver, and when unloading the sfc driver the removal order is not guaranteed, so the instruction from the PF to the VF to unbind enforces a suitable ordering so that vswitches and vports can be removed. As a result of this, manually unbinding the driver from a single PF will result in all of its VFs in the host also being removed. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-02 13:40:46 +03:00
unsigned int i;
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
int rc;
sfc: leak vports if a VF is assigned during PF unload If any VF is assigned as the PF is unloaded, do not attempt to remove its vport or the vswitch. These will be removed if the driver binds to the PF again, as an entity reset occurs during probe. A 'force' flag is added to efx_ef10_pci_sriov_disable() to distinguish between disabling SR-IOV and driver unload. SR-IOV cannot be disabled if VFs are assigned to guests. If the PF driver is unloaded while VFs are assigned, the driver may try to bind to the VF again at a later point if the driver has been reloaded and the VF returns to the same domain as the PF. In this case, the PF will not have a VF data structure, so the VF can check this and drop out of probe early. In this case, efx->vf_count will be zero but VFs will be present. The user is advised to remove the VF and re-create it. The check at the beginning of efx_ef10_pci_sriov_disable() that efx->vf_count is non-zero is removed to allow SR-IOV to be disabled in this case. Also, if the PF driver is unloaded, it will disable SR-IOV to remove these unknown VFs. By not disabling bus-mastering if VFs are still assigned, the VF will continue to pass traffic after the PF has been removed. When using the max_vfs module parameter, if VFs are already present do not try to initialise any more. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-02 13:41:00 +03:00
if (!nic_data->vf) {
/* Remove any un-assigned orphaned VFs */
if (pci_num_vf(efx->pci_dev) && !pci_vfs_assigned(efx->pci_dev))
pci_disable_sriov(efx->pci_dev);
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
return;
sfc: leak vports if a VF is assigned during PF unload If any VF is assigned as the PF is unloaded, do not attempt to remove its vport or the vswitch. These will be removed if the driver binds to the PF again, as an entity reset occurs during probe. A 'force' flag is added to efx_ef10_pci_sriov_disable() to distinguish between disabling SR-IOV and driver unload. SR-IOV cannot be disabled if VFs are assigned to guests. If the PF driver is unloaded while VFs are assigned, the driver may try to bind to the VF again at a later point if the driver has been reloaded and the VF returns to the same domain as the PF. In this case, the PF will not have a VF data structure, so the VF can check this and drop out of probe early. In this case, efx->vf_count will be zero but VFs will be present. The user is advised to remove the VF and re-create it. The check at the beginning of efx_ef10_pci_sriov_disable() that efx->vf_count is non-zero is removed to allow SR-IOV to be disabled in this case. Also, if the PF driver is unloaded, it will disable SR-IOV to remove these unknown VFs. By not disabling bus-mastering if VFs are still assigned, the VF will continue to pass traffic after the PF has been removed. When using the max_vfs module parameter, if VFs are already present do not try to initialise any more. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-02 13:41:00 +03:00
}
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
sfc: force removal of VF and vport on driver removal When the driver unloads, force the unbind and removal of any VFs in the host with the PF. The PF cannot remove vports and vswitches if they are still being used by a VF driver, and when unloading the sfc driver the removal order is not guaranteed, so the instruction from the PF to the VF to unbind enforces a suitable ordering so that vswitches and vports can be removed. As a result of this, manually unbinding the driver from a single PF will result in all of its VFs in the host also being removed. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-02 13:40:46 +03:00
/* Remove any VFs in the host */
for (i = 0; i < efx->vf_count; ++i) {
struct efx_nic *vf_efx = nic_data->vf[i].efx;
if (vf_efx)
vf_efx->pci_dev->driver->remove(vf_efx->pci_dev);
}
sfc: leak vports if a VF is assigned during PF unload If any VF is assigned as the PF is unloaded, do not attempt to remove its vport or the vswitch. These will be removed if the driver binds to the PF again, as an entity reset occurs during probe. A 'force' flag is added to efx_ef10_pci_sriov_disable() to distinguish between disabling SR-IOV and driver unload. SR-IOV cannot be disabled if VFs are assigned to guests. If the PF driver is unloaded while VFs are assigned, the driver may try to bind to the VF again at a later point if the driver has been reloaded and the VF returns to the same domain as the PF. In this case, the PF will not have a VF data structure, so the VF can check this and drop out of probe early. In this case, efx->vf_count will be zero but VFs will be present. The user is advised to remove the VF and re-create it. The check at the beginning of efx_ef10_pci_sriov_disable() that efx->vf_count is non-zero is removed to allow SR-IOV to be disabled in this case. Also, if the PF driver is unloaded, it will disable SR-IOV to remove these unknown VFs. By not disabling bus-mastering if VFs are still assigned, the VF will continue to pass traffic after the PF has been removed. When using the max_vfs module parameter, if VFs are already present do not try to initialise any more. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-02 13:41:00 +03:00
rc = efx_ef10_pci_sriov_disable(efx, true);
sfc: create vports for VFs and assign random MAC addresses The parent PF creates vports for all its child VFs and adds MAC addresses to these. When the VF driver loads, it can make an MCDI call to get the MAC address that the parent PF assigned it. The parent PF also assigns a mac address to its own vport because implicit creation of a vAdaptor will only work on evb ports with MAC addresses assigned. The vport MAC address needs to be stored in the PF's nic_data struct as it can later be changed on the vadaptor (and its net_dev struct). When removing a vport the original MAC address must be deleted. A new flag is needed in the VF data structure to identify whether a vport has been assigned to the VF. This is to determine whether it needs to be un-assigned before freeing the vport. Also, attempting to un-assign a vport which is not assigned will result in an EALREADY error. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-06 02:58:31 +03:00
if (rc)
netif_dbg(efx, drv, efx->net_dev,
"Disabling SRIOV was not successful rc=%d\n", rc);
else
netif_dbg(efx, drv, efx->net_dev, "SRIOV disabled\n");
}
sfc: add ndo_set_vf_mac() function for EF10 Implement a response to this entrypoint. The ndo_set_vf_mac() entrypoint is only exposed in the driver if CONFIG_SFC_SRIOV is defined. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-20 13:11:03 +03:00
static int efx_ef10_vport_del_vf_mac(struct efx_nic *efx, unsigned int port_id,
u8 *mac)
{
MCDI_DECLARE_BUF(inbuf, MC_CMD_VPORT_DEL_MAC_ADDRESS_IN_LEN);
MCDI_DECLARE_BUF_ERR(outbuf);
size_t outlen;
int rc;
MCDI_SET_DWORD(inbuf, VPORT_DEL_MAC_ADDRESS_IN_VPORT_ID, port_id);
ether_addr_copy(MCDI_PTR(inbuf, VPORT_DEL_MAC_ADDRESS_IN_MACADDR), mac);
rc = efx_mcdi_rpc(efx, MC_CMD_VPORT_DEL_MAC_ADDRESS, inbuf,
sizeof(inbuf), outbuf, sizeof(outbuf), &outlen);
return rc;
}
int efx_ef10_sriov_set_vf_mac(struct efx_nic *efx, int vf_i, u8 *mac)
{
struct efx_ef10_nic_data *nic_data = efx->nic_data;
struct ef10_vf *vf;
int rc;
if (!nic_data->vf)
return -EOPNOTSUPP;
if (vf_i >= efx->vf_count)
return -EINVAL;
vf = nic_data->vf + vf_i;
if (vf->efx) {
efx_device_detach_sync(vf->efx);
efx_net_stop(vf->efx->net_dev);
down_write(&vf->efx->filter_sem);
vf->efx->type->filter_table_remove(vf->efx);
rc = efx_ef10_vadaptor_free(vf->efx, EVB_PORT_ID_ASSIGNED);
if (rc) {
up_write(&vf->efx->filter_sem);
return rc;
}
}
rc = efx_ef10_evb_port_assign(efx, EVB_PORT_ID_NULL, vf_i);
if (rc)
return rc;
if (!is_zero_ether_addr(vf->mac)) {
rc = efx_ef10_vport_del_vf_mac(efx, vf->vport_id, vf->mac);
if (rc)
return rc;
}
if (!is_zero_ether_addr(mac)) {
rc = efx_ef10_vport_add_mac(efx, vf->vport_id, mac);
if (rc) {
eth_zero_addr(vf->mac);
goto fail;
}
if (vf->efx)
ether_addr_copy(vf->efx->net_dev->dev_addr, mac);
}
ether_addr_copy(vf->mac, mac);
rc = efx_ef10_evb_port_assign(efx, vf->vport_id, vf_i);
if (rc)
goto fail;
if (vf->efx) {
/* VF cannot use the vport_id that the PF created */
rc = efx_ef10_vadaptor_alloc(vf->efx, EVB_PORT_ID_ASSIGNED);
if (rc) {
up_write(&vf->efx->filter_sem);
return rc;
}
vf->efx->type->filter_table_probe(vf->efx);
up_write(&vf->efx->filter_sem);
efx_net_open(vf->efx->net_dev);
netif_device_attach(vf->efx->net_dev);
}
return 0;
fail:
memset(vf->mac, 0, ETH_ALEN);
return rc;
}
sfc: Add ndo_get_vf_config() function for EF10 rtnetlink calls ndo_get_vf_config when compiling information about a network interface, so that the VFs associated with a PF can be listed (eg: ip link show). Implement a response to this entry point and return PF-set MAC address for VF in ndo_get_vf_config Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-20 13:11:18 +03:00
sfc: add ndo_set_vf_vlan() function for EF10 The max vlan tags that can be offloaded is 2, including any upstream VLAN aggregator. Currently there is no way for the net driver to know whether the upstream vswitch (if any) is using vlan tags, so there is no way to know how many tags we can request. Along with the implementation for the ndo_set_vf_vlan callback, this patch also adds 2 VLAN tags for the driver created VEB switch if possible, that way it is possible to offload as many tags as are allowed. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-20 13:11:54 +03:00
int efx_ef10_sriov_set_vf_vlan(struct efx_nic *efx, int vf_i, u16 vlan,
u8 qos)
{
struct efx_ef10_nic_data *nic_data = efx->nic_data;
struct ef10_vf *vf;
u16 old_vlan, new_vlan;
int rc = 0, rc2 = 0;
if (vf_i >= efx->vf_count)
return -EINVAL;
if (qos != 0)
return -EINVAL;
vf = nic_data->vf + vf_i;
new_vlan = (vlan == 0) ? EFX_EF10_NO_VLAN : vlan;
if (new_vlan == vf->vlan)
return 0;
if (vf->efx) {
efx_device_detach_sync(vf->efx);
efx_net_stop(vf->efx->net_dev);
down_write(&vf->efx->filter_sem);
vf->efx->type->filter_table_remove(vf->efx);
rc = efx_ef10_vadaptor_free(vf->efx, EVB_PORT_ID_ASSIGNED);
if (rc)
goto restore_filters;
}
if (vf->vport_assigned) {
rc = efx_ef10_evb_port_assign(efx, EVB_PORT_ID_NULL, vf_i);
if (rc) {
netif_warn(efx, drv, efx->net_dev,
"Failed to change vlan on VF %d.\n", vf_i);
netif_warn(efx, drv, efx->net_dev,
"This is likely because the VF is bound to a driver in a VM.\n");
netif_warn(efx, drv, efx->net_dev,
"Please unload the driver in the VM.\n");
goto restore_vadaptor;
}
vf->vport_assigned = 0;
}
if (!is_zero_ether_addr(vf->mac)) {
rc = efx_ef10_vport_del_mac(efx, vf->vport_id, vf->mac);
if (rc)
goto restore_evb_port;
}
if (vf->vport_id) {
rc = efx_ef10_vport_free(efx, vf->vport_id);
if (rc)
goto restore_mac;
vf->vport_id = 0;
}
/* Do the actual vlan change */
old_vlan = vf->vlan;
vf->vlan = new_vlan;
/* Restore everything in reverse order */
rc = efx_ef10_vport_alloc(efx, EVB_PORT_ID_ASSIGNED,
MC_CMD_VPORT_ALLOC_IN_VPORT_TYPE_NORMAL,
vf->vlan, &vf->vport_id);
if (rc)
goto reset_nic;
restore_mac:
if (!is_zero_ether_addr(vf->mac)) {
rc2 = efx_ef10_vport_add_mac(efx, vf->vport_id, vf->mac);
if (rc2) {
eth_zero_addr(vf->mac);
goto reset_nic;
}
}
restore_evb_port:
rc2 = efx_ef10_evb_port_assign(efx, vf->vport_id, vf_i);
if (rc2)
goto reset_nic;
else
vf->vport_assigned = 1;
restore_vadaptor:
if (vf->efx) {
rc2 = efx_ef10_vadaptor_alloc(vf->efx, EVB_PORT_ID_ASSIGNED);
if (rc2)
goto reset_nic;
}
restore_filters:
if (vf->efx) {
rc2 = vf->efx->type->filter_table_probe(vf->efx);
if (rc2)
goto reset_nic;
up_write(&vf->efx->filter_sem);
rc2 = efx_net_open(vf->efx->net_dev);
if (rc2)
goto reset_nic;
netif_device_attach(vf->efx->net_dev);
}
return rc;
reset_nic:
if (vf->efx) {
up_write(&vf->efx->filter_sem);
netif_err(efx, drv, efx->net_dev,
"Failed to restore VF - scheduling reset.\n");
efx_schedule_reset(vf->efx, RESET_TYPE_DATAPATH);
} else {
netif_err(efx, drv, efx->net_dev,
"Failed to restore the VF and cannot reset the VF "
"- VF is not functional.\n");
netif_err(efx, drv, efx->net_dev,
"Please reload the driver attached to the VF.\n");
}
return rc ? rc : rc2;
}
sfc: Implement dummy disable of VF spoof check for EF10 Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-20 13:12:30 +03:00
int efx_ef10_sriov_set_vf_spoofchk(struct efx_nic *efx, int vf_i,
bool spoofchk)
{
return spoofchk ? -EOPNOTSUPP : 0;
}
sfc: add ndo_set_vf_link_state() function for EF10 Exercised with "ip link set <PF intf> vf <vf_i> state {auto|enable|disable}" Sets the reporting policy for VF link state to either - mirror physical link state - always up - always down get VF link state mode in efx_ef10_sriov_get_vf_config Exercised by "ip link show <PF intf>"; output will include a line like vf 0 MAC 12:34:56:78:9a:bc, link-state auto Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-20 13:12:13 +03:00
int efx_ef10_sriov_set_vf_link_state(struct efx_nic *efx, int vf_i,
int link_state)
{
MCDI_DECLARE_BUF(inbuf, MC_CMD_LINK_STATE_MODE_IN_LEN);
struct efx_ef10_nic_data *nic_data = efx->nic_data;
BUILD_BUG_ON(IFLA_VF_LINK_STATE_AUTO !=
MC_CMD_LINK_STATE_MODE_IN_LINK_STATE_AUTO);
BUILD_BUG_ON(IFLA_VF_LINK_STATE_ENABLE !=
MC_CMD_LINK_STATE_MODE_IN_LINK_STATE_UP);
BUILD_BUG_ON(IFLA_VF_LINK_STATE_DISABLE !=
MC_CMD_LINK_STATE_MODE_IN_LINK_STATE_DOWN);
MCDI_POPULATE_DWORD_2(inbuf, LINK_STATE_MODE_IN_FUNCTION,
LINK_STATE_MODE_IN_FUNCTION_PF,
nic_data->pf_index,
LINK_STATE_MODE_IN_FUNCTION_VF, vf_i);
MCDI_SET_DWORD(inbuf, LINK_STATE_MODE_IN_NEW_MODE, link_state);
return efx_mcdi_rpc(efx, MC_CMD_LINK_STATE_MODE, inbuf, sizeof(inbuf),
NULL, 0, NULL); /* don't care what old mode was */
}
sfc: Add ndo_get_vf_config() function for EF10 rtnetlink calls ndo_get_vf_config when compiling information about a network interface, so that the VFs associated with a PF can be listed (eg: ip link show). Implement a response to this entry point and return PF-set MAC address for VF in ndo_get_vf_config Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-20 13:11:18 +03:00
int efx_ef10_sriov_get_vf_config(struct efx_nic *efx, int vf_i,
struct ifla_vf_info *ivf)
{
sfc: add ndo_set_vf_link_state() function for EF10 Exercised with "ip link set <PF intf> vf <vf_i> state {auto|enable|disable}" Sets the reporting policy for VF link state to either - mirror physical link state - always up - always down get VF link state mode in efx_ef10_sriov_get_vf_config Exercised by "ip link show <PF intf>"; output will include a line like vf 0 MAC 12:34:56:78:9a:bc, link-state auto Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-20 13:12:13 +03:00
MCDI_DECLARE_BUF(inbuf, MC_CMD_LINK_STATE_MODE_IN_LEN);
MCDI_DECLARE_BUF(outbuf, MC_CMD_LINK_STATE_MODE_OUT_LEN);
sfc: Add ndo_get_vf_config() function for EF10 rtnetlink calls ndo_get_vf_config when compiling information about a network interface, so that the VFs associated with a PF can be listed (eg: ip link show). Implement a response to this entry point and return PF-set MAC address for VF in ndo_get_vf_config Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-20 13:11:18 +03:00
struct efx_ef10_nic_data *nic_data = efx->nic_data;
struct ef10_vf *vf;
sfc: add ndo_set_vf_link_state() function for EF10 Exercised with "ip link set <PF intf> vf <vf_i> state {auto|enable|disable}" Sets the reporting policy for VF link state to either - mirror physical link state - always up - always down get VF link state mode in efx_ef10_sriov_get_vf_config Exercised by "ip link show <PF intf>"; output will include a line like vf 0 MAC 12:34:56:78:9a:bc, link-state auto Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-20 13:12:13 +03:00
size_t outlen;
int rc;
sfc: Add ndo_get_vf_config() function for EF10 rtnetlink calls ndo_get_vf_config when compiling information about a network interface, so that the VFs associated with a PF can be listed (eg: ip link show). Implement a response to this entry point and return PF-set MAC address for VF in ndo_get_vf_config Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-20 13:11:18 +03:00
if (vf_i >= efx->vf_count)
return -EINVAL;
if (!nic_data->vf)
return -EOPNOTSUPP;
vf = nic_data->vf + vf_i;
ivf->vf = vf_i;
ivf->min_tx_rate = 0;
ivf->max_tx_rate = 0;
ether_addr_copy(ivf->mac, vf->mac);
sfc: add ndo_set_vf_vlan() function for EF10 The max vlan tags that can be offloaded is 2, including any upstream VLAN aggregator. Currently there is no way for the net driver to know whether the upstream vswitch (if any) is using vlan tags, so there is no way to know how many tags we can request. Along with the implementation for the ndo_set_vf_vlan callback, this patch also adds 2 VLAN tags for the driver created VEB switch if possible, that way it is possible to offload as many tags as are allowed. Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-20 13:11:54 +03:00
ivf->vlan = (vf->vlan == EFX_EF10_NO_VLAN) ? 0 : vf->vlan;
sfc: Add ndo_get_vf_config() function for EF10 rtnetlink calls ndo_get_vf_config when compiling information about a network interface, so that the VFs associated with a PF can be listed (eg: ip link show). Implement a response to this entry point and return PF-set MAC address for VF in ndo_get_vf_config Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-20 13:11:18 +03:00
ivf->qos = 0;
sfc: add ndo_set_vf_link_state() function for EF10 Exercised with "ip link set <PF intf> vf <vf_i> state {auto|enable|disable}" Sets the reporting policy for VF link state to either - mirror physical link state - always up - always down get VF link state mode in efx_ef10_sriov_get_vf_config Exercised by "ip link show <PF intf>"; output will include a line like vf 0 MAC 12:34:56:78:9a:bc, link-state auto Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-20 13:12:13 +03:00
MCDI_POPULATE_DWORD_2(inbuf, LINK_STATE_MODE_IN_FUNCTION,
LINK_STATE_MODE_IN_FUNCTION_PF,
nic_data->pf_index,
LINK_STATE_MODE_IN_FUNCTION_VF, vf_i);
MCDI_SET_DWORD(inbuf, LINK_STATE_MODE_IN_NEW_MODE,
MC_CMD_LINK_STATE_MODE_IN_DO_NOT_CHANGE);
rc = efx_mcdi_rpc(efx, MC_CMD_LINK_STATE_MODE, inbuf, sizeof(inbuf),
outbuf, sizeof(outbuf), &outlen);
if (rc)
return rc;
if (outlen < MC_CMD_LINK_STATE_MODE_OUT_LEN)
return -EIO;
ivf->linkstate = MCDI_DWORD(outbuf, LINK_STATE_MODE_OUT_OLD_MODE);
sfc: Add ndo_get_vf_config() function for EF10 rtnetlink calls ndo_get_vf_config when compiling information about a network interface, so that the VFs associated with a PF can be listed (eg: ip link show). Implement a response to this entry point and return PF-set MAC address for VF in ndo_get_vf_config Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-05-20 13:11:18 +03:00
return 0;
}
sfc: Implement ndo_gets_phys_port_id() for EF10 VFs Signed-off-by: Shradha Shah <sshah@solarflare.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2015-06-02 13:38:16 +03:00
int efx_ef10_sriov_get_phys_port_id(struct efx_nic *efx,
struct netdev_phys_item_id *ppid)
{
struct efx_ef10_nic_data *nic_data = efx->nic_data;
if (!is_valid_ether_addr(nic_data->port_id))
return -EOPNOTSUPP;
ppid->id_len = ETH_ALEN;
memcpy(ppid->id, nic_data->port_id, ppid->id_len);
return 0;
}