linux/drivers/net/ethernet/intel/ice/ice_lib.c

3050 lines
76 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0
/* Copyright (c) 2018, Intel Corporation. */
#include "ice.h"
#include "ice_base.h"
#include "ice_flow.h"
#include "ice_lib.h"
#include "ice_dcb_lib.h"
/**
* ice_vsi_type_str - maps VSI type enum to string equivalents
* @type: VSI type enum
*/
const char *ice_vsi_type_str(enum ice_vsi_type type)
{
switch (type) {
case ICE_VSI_PF:
return "ICE_VSI_PF";
case ICE_VSI_VF:
return "ICE_VSI_VF";
case ICE_VSI_LB:
return "ICE_VSI_LB";
default:
return "unknown";
}
}
/**
* ice_vsi_ctrl_rx_rings - Start or stop a VSI's Rx rings
* @vsi: the VSI being configured
* @ena: start or stop the Rx rings
*/
static int ice_vsi_ctrl_rx_rings(struct ice_vsi *vsi, bool ena)
{
int i, ret = 0;
for (i = 0; i < vsi->num_rxq; i++) {
ret = ice_vsi_ctrl_rx_ring(vsi, ena, i);
if (ret)
break;
}
return ret;
}
/**
* ice_vsi_alloc_arrays - Allocate queue and vector pointer arrays for the VSI
* @vsi: VSI pointer
*
* On error: returns error code (negative)
* On success: returns 0
*/
static int ice_vsi_alloc_arrays(struct ice_vsi *vsi)
{
struct ice_pf *pf = vsi->back;
struct device *dev;
dev = ice_pf_to_dev(pf);
/* allocate memory for both Tx and Rx ring pointers */
vsi->tx_rings = devm_kcalloc(dev, vsi->alloc_txq,
sizeof(*vsi->tx_rings), GFP_KERNEL);
if (!vsi->tx_rings)
ice: Alloc queue management bitmaps and arrays dynamically The total number of queues available on the device is divided between multiple physical functions (PF) in the firmware and provided to the driver when it gets function capabilities from the firmware. Thus each PF knows how many Tx/Rx queues it has. These queues are then doled out to different VSIs (for LAN traffic, SR-IOV VF traffic, etc.) To track usage of these queues at the PF level, the driver uses two bitmaps avail_txqs and avail_rxqs. At the VSI level (i.e. struct ice_vsi instances) the driver uses two arrays txq_map and rxq_map, to track ownership of VSIs' queues in avail_txqs and avail_rxqs respectively. The aforementioned bitmaps and arrays should be allocated dynamically, because the number of queues supported by a PF is only available once function capabilities have been queried. The current static allocation consumes way more memory than required. This patch removes the DECLARE_BITMAP for avail_txqs and avail_rxqs and instead uses bitmap_zalloc to allocate the bitmaps during init. Similarly txq_map and rxq_map are now allocated in ice_vsi_alloc_arrays. As a result ICE_MAX_TXQS and ICE_MAX_RXQS defines are no longer needed. Also as txq_map and rxq_map are now allocated and freed, some code reordering was required in ice_vsi_rebuild for correct functioning. Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-08-02 16:25:21 +08:00
return -ENOMEM;
vsi->rx_rings = devm_kcalloc(dev, vsi->alloc_rxq,
sizeof(*vsi->rx_rings), GFP_KERNEL);
if (!vsi->rx_rings)
ice: Alloc queue management bitmaps and arrays dynamically The total number of queues available on the device is divided between multiple physical functions (PF) in the firmware and provided to the driver when it gets function capabilities from the firmware. Thus each PF knows how many Tx/Rx queues it has. These queues are then doled out to different VSIs (for LAN traffic, SR-IOV VF traffic, etc.) To track usage of these queues at the PF level, the driver uses two bitmaps avail_txqs and avail_rxqs. At the VSI level (i.e. struct ice_vsi instances) the driver uses two arrays txq_map and rxq_map, to track ownership of VSIs' queues in avail_txqs and avail_rxqs respectively. The aforementioned bitmaps and arrays should be allocated dynamically, because the number of queues supported by a PF is only available once function capabilities have been queried. The current static allocation consumes way more memory than required. This patch removes the DECLARE_BITMAP for avail_txqs and avail_rxqs and instead uses bitmap_zalloc to allocate the bitmaps during init. Similarly txq_map and rxq_map are now allocated in ice_vsi_alloc_arrays. As a result ICE_MAX_TXQS and ICE_MAX_RXQS defines are no longer needed. Also as txq_map and rxq_map are now allocated and freed, some code reordering was required in ice_vsi_rebuild for correct functioning. Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-08-02 16:25:21 +08:00
goto err_rings;
/* XDP will have vsi->alloc_txq Tx queues as well, so double the size */
vsi->txq_map = devm_kcalloc(dev, (2 * vsi->alloc_txq),
ice: Alloc queue management bitmaps and arrays dynamically The total number of queues available on the device is divided between multiple physical functions (PF) in the firmware and provided to the driver when it gets function capabilities from the firmware. Thus each PF knows how many Tx/Rx queues it has. These queues are then doled out to different VSIs (for LAN traffic, SR-IOV VF traffic, etc.) To track usage of these queues at the PF level, the driver uses two bitmaps avail_txqs and avail_rxqs. At the VSI level (i.e. struct ice_vsi instances) the driver uses two arrays txq_map and rxq_map, to track ownership of VSIs' queues in avail_txqs and avail_rxqs respectively. The aforementioned bitmaps and arrays should be allocated dynamically, because the number of queues supported by a PF is only available once function capabilities have been queried. The current static allocation consumes way more memory than required. This patch removes the DECLARE_BITMAP for avail_txqs and avail_rxqs and instead uses bitmap_zalloc to allocate the bitmaps during init. Similarly txq_map and rxq_map are now allocated in ice_vsi_alloc_arrays. As a result ICE_MAX_TXQS and ICE_MAX_RXQS defines are no longer needed. Also as txq_map and rxq_map are now allocated and freed, some code reordering was required in ice_vsi_rebuild for correct functioning. Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-08-02 16:25:21 +08:00
sizeof(*vsi->txq_map), GFP_KERNEL);
if (!vsi->txq_map)
goto err_txq_map;
vsi->rxq_map = devm_kcalloc(dev, vsi->alloc_rxq,
ice: Alloc queue management bitmaps and arrays dynamically The total number of queues available on the device is divided between multiple physical functions (PF) in the firmware and provided to the driver when it gets function capabilities from the firmware. Thus each PF knows how many Tx/Rx queues it has. These queues are then doled out to different VSIs (for LAN traffic, SR-IOV VF traffic, etc.) To track usage of these queues at the PF level, the driver uses two bitmaps avail_txqs and avail_rxqs. At the VSI level (i.e. struct ice_vsi instances) the driver uses two arrays txq_map and rxq_map, to track ownership of VSIs' queues in avail_txqs and avail_rxqs respectively. The aforementioned bitmaps and arrays should be allocated dynamically, because the number of queues supported by a PF is only available once function capabilities have been queried. The current static allocation consumes way more memory than required. This patch removes the DECLARE_BITMAP for avail_txqs and avail_rxqs and instead uses bitmap_zalloc to allocate the bitmaps during init. Similarly txq_map and rxq_map are now allocated in ice_vsi_alloc_arrays. As a result ICE_MAX_TXQS and ICE_MAX_RXQS defines are no longer needed. Also as txq_map and rxq_map are now allocated and freed, some code reordering was required in ice_vsi_rebuild for correct functioning. Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-08-02 16:25:21 +08:00
sizeof(*vsi->rxq_map), GFP_KERNEL);
if (!vsi->rxq_map)
goto err_rxq_map;
/* There is no need to allocate q_vectors for a loopback VSI. */
if (vsi->type == ICE_VSI_LB)
return 0;
/* allocate memory for q_vector pointers */
vsi->q_vectors = devm_kcalloc(dev, vsi->num_q_vectors,
sizeof(*vsi->q_vectors), GFP_KERNEL);
if (!vsi->q_vectors)
goto err_vectors;
return 0;
err_vectors:
devm_kfree(dev, vsi->rxq_map);
ice: Alloc queue management bitmaps and arrays dynamically The total number of queues available on the device is divided between multiple physical functions (PF) in the firmware and provided to the driver when it gets function capabilities from the firmware. Thus each PF knows how many Tx/Rx queues it has. These queues are then doled out to different VSIs (for LAN traffic, SR-IOV VF traffic, etc.) To track usage of these queues at the PF level, the driver uses two bitmaps avail_txqs and avail_rxqs. At the VSI level (i.e. struct ice_vsi instances) the driver uses two arrays txq_map and rxq_map, to track ownership of VSIs' queues in avail_txqs and avail_rxqs respectively. The aforementioned bitmaps and arrays should be allocated dynamically, because the number of queues supported by a PF is only available once function capabilities have been queried. The current static allocation consumes way more memory than required. This patch removes the DECLARE_BITMAP for avail_txqs and avail_rxqs and instead uses bitmap_zalloc to allocate the bitmaps during init. Similarly txq_map and rxq_map are now allocated in ice_vsi_alloc_arrays. As a result ICE_MAX_TXQS and ICE_MAX_RXQS defines are no longer needed. Also as txq_map and rxq_map are now allocated and freed, some code reordering was required in ice_vsi_rebuild for correct functioning. Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-08-02 16:25:21 +08:00
err_rxq_map:
devm_kfree(dev, vsi->txq_map);
ice: Alloc queue management bitmaps and arrays dynamically The total number of queues available on the device is divided between multiple physical functions (PF) in the firmware and provided to the driver when it gets function capabilities from the firmware. Thus each PF knows how many Tx/Rx queues it has. These queues are then doled out to different VSIs (for LAN traffic, SR-IOV VF traffic, etc.) To track usage of these queues at the PF level, the driver uses two bitmaps avail_txqs and avail_rxqs. At the VSI level (i.e. struct ice_vsi instances) the driver uses two arrays txq_map and rxq_map, to track ownership of VSIs' queues in avail_txqs and avail_rxqs respectively. The aforementioned bitmaps and arrays should be allocated dynamically, because the number of queues supported by a PF is only available once function capabilities have been queried. The current static allocation consumes way more memory than required. This patch removes the DECLARE_BITMAP for avail_txqs and avail_rxqs and instead uses bitmap_zalloc to allocate the bitmaps during init. Similarly txq_map and rxq_map are now allocated in ice_vsi_alloc_arrays. As a result ICE_MAX_TXQS and ICE_MAX_RXQS defines are no longer needed. Also as txq_map and rxq_map are now allocated and freed, some code reordering was required in ice_vsi_rebuild for correct functioning. Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-08-02 16:25:21 +08:00
err_txq_map:
devm_kfree(dev, vsi->rx_rings);
ice: Alloc queue management bitmaps and arrays dynamically The total number of queues available on the device is divided between multiple physical functions (PF) in the firmware and provided to the driver when it gets function capabilities from the firmware. Thus each PF knows how many Tx/Rx queues it has. These queues are then doled out to different VSIs (for LAN traffic, SR-IOV VF traffic, etc.) To track usage of these queues at the PF level, the driver uses two bitmaps avail_txqs and avail_rxqs. At the VSI level (i.e. struct ice_vsi instances) the driver uses two arrays txq_map and rxq_map, to track ownership of VSIs' queues in avail_txqs and avail_rxqs respectively. The aforementioned bitmaps and arrays should be allocated dynamically, because the number of queues supported by a PF is only available once function capabilities have been queried. The current static allocation consumes way more memory than required. This patch removes the DECLARE_BITMAP for avail_txqs and avail_rxqs and instead uses bitmap_zalloc to allocate the bitmaps during init. Similarly txq_map and rxq_map are now allocated in ice_vsi_alloc_arrays. As a result ICE_MAX_TXQS and ICE_MAX_RXQS defines are no longer needed. Also as txq_map and rxq_map are now allocated and freed, some code reordering was required in ice_vsi_rebuild for correct functioning. Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-08-02 16:25:21 +08:00
err_rings:
devm_kfree(dev, vsi->tx_rings);
return -ENOMEM;
}
/**
* ice_vsi_set_num_desc - Set number of descriptors for queues on this VSI
* @vsi: the VSI being configured
*/
static void ice_vsi_set_num_desc(struct ice_vsi *vsi)
{
switch (vsi->type) {
case ICE_VSI_PF:
/* fall through */
case ICE_VSI_LB:
vsi->num_rx_desc = ICE_DFLT_NUM_RX_DESC;
vsi->num_tx_desc = ICE_DFLT_NUM_TX_DESC;
break;
default:
dev_dbg(&vsi->back->pdev->dev,
"Not setting number of Tx/Rx descriptors for VSI type %d\n",
vsi->type);
break;
}
}
/**
* ice_vsi_set_num_qs - Set number of queues, descriptors and vectors for a VSI
* @vsi: the VSI being configured
* @vf_id: ID of the VF being configured
*
* Return 0 on success and a negative value on error
*/
static void ice_vsi_set_num_qs(struct ice_vsi *vsi, u16 vf_id)
{
struct ice_pf *pf = vsi->back;
struct ice_vf *vf = NULL;
if (vsi->type == ICE_VSI_VF)
vsi->vf_id = vf_id;
switch (vsi->type) {
case ICE_VSI_PF:
vsi->alloc_txq = min_t(int, ice_get_avail_txq_count(pf),
num_online_cpus());
if (vsi->req_txq) {
vsi->alloc_txq = vsi->req_txq;
vsi->num_txq = vsi->req_txq;
}
pf->num_lan_tx = vsi->alloc_txq;
/* only 1 Rx queue unless RSS is enabled */
if (!test_bit(ICE_FLAG_RSS_ENA, pf->flags)) {
vsi->alloc_rxq = 1;
} else {
vsi->alloc_rxq = min_t(int, ice_get_avail_rxq_count(pf),
num_online_cpus());
if (vsi->req_rxq) {
vsi->alloc_rxq = vsi->req_rxq;
vsi->num_rxq = vsi->req_rxq;
}
}
pf->num_lan_rx = vsi->alloc_rxq;
vsi->num_q_vectors = max_t(int, vsi->alloc_rxq, vsi->alloc_txq);
break;
case ICE_VSI_VF:
vf = &pf->vf[vsi->vf_id];
vsi->alloc_txq = vf->num_vf_qs;
vsi->alloc_rxq = vf->num_vf_qs;
/* pf->num_vf_msix includes (VF miscellaneous vector +
* data queue interrupts). Since vsi->num_q_vectors is number
* of queues vectors, subtract 1 (ICE_NONQ_VECS_VF) from the
* original vector count
*/
vsi->num_q_vectors = pf->num_vf_msix - ICE_NONQ_VECS_VF;
break;
case ICE_VSI_LB:
vsi->alloc_txq = 1;
vsi->alloc_rxq = 1;
break;
default:
dev_warn(ice_pf_to_dev(pf), "Unknown VSI type %d\n", vsi->type);
break;
}
ice_vsi_set_num_desc(vsi);
}
/**
* ice_get_free_slot - get the next non-NULL location index in array
* @array: array to search
* @size: size of the array
* @curr: last known occupied index to be used as a search hint
*
* void * is being used to keep the functionality generic. This lets us use this
* function on any array of pointers.
*/
static int ice_get_free_slot(void *array, int size, int curr)
{
int **tmp_array = (int **)array;
int next;
if (curr < (size - 1) && !tmp_array[curr + 1]) {
next = curr + 1;
} else {
int i = 0;
while ((i < size) && (tmp_array[i]))
i++;
if (i == size)
next = ICE_NO_VSI;
else
next = i;
}
return next;
}
/**
* ice_vsi_delete - delete a VSI from the switch
* @vsi: pointer to VSI being removed
*/
void ice_vsi_delete(struct ice_vsi *vsi)
{
struct ice_pf *pf = vsi->back;
struct ice_vsi_ctx *ctxt;
enum ice_status status;
ctxt = kzalloc(sizeof(*ctxt), GFP_KERNEL);
if (!ctxt)
return;
if (vsi->type == ICE_VSI_VF)
ctxt->vf_num = vsi->vf_id;
ctxt->vsi_num = vsi->vsi_num;
memcpy(&ctxt->info, &vsi->info, sizeof(ctxt->info));
status = ice_free_vsi(&pf->hw, vsi->idx, ctxt, false, NULL);
if (status)
dev_err(ice_pf_to_dev(pf), "Failed to delete VSI %i in FW - error: %d\n",
vsi->vsi_num, status);
kfree(ctxt);
}
/**
* ice_vsi_free_arrays - De-allocate queue and vector pointer arrays for the VSI
* @vsi: pointer to VSI being cleared
*/
static void ice_vsi_free_arrays(struct ice_vsi *vsi)
{
struct ice_pf *pf = vsi->back;
struct device *dev;
dev = ice_pf_to_dev(pf);
/* free the ring and vector containers */
if (vsi->q_vectors) {
devm_kfree(dev, vsi->q_vectors);
vsi->q_vectors = NULL;
}
if (vsi->tx_rings) {
devm_kfree(dev, vsi->tx_rings);
vsi->tx_rings = NULL;
}
if (vsi->rx_rings) {
devm_kfree(dev, vsi->rx_rings);
vsi->rx_rings = NULL;
}
ice: Alloc queue management bitmaps and arrays dynamically The total number of queues available on the device is divided between multiple physical functions (PF) in the firmware and provided to the driver when it gets function capabilities from the firmware. Thus each PF knows how many Tx/Rx queues it has. These queues are then doled out to different VSIs (for LAN traffic, SR-IOV VF traffic, etc.) To track usage of these queues at the PF level, the driver uses two bitmaps avail_txqs and avail_rxqs. At the VSI level (i.e. struct ice_vsi instances) the driver uses two arrays txq_map and rxq_map, to track ownership of VSIs' queues in avail_txqs and avail_rxqs respectively. The aforementioned bitmaps and arrays should be allocated dynamically, because the number of queues supported by a PF is only available once function capabilities have been queried. The current static allocation consumes way more memory than required. This patch removes the DECLARE_BITMAP for avail_txqs and avail_rxqs and instead uses bitmap_zalloc to allocate the bitmaps during init. Similarly txq_map and rxq_map are now allocated in ice_vsi_alloc_arrays. As a result ICE_MAX_TXQS and ICE_MAX_RXQS defines are no longer needed. Also as txq_map and rxq_map are now allocated and freed, some code reordering was required in ice_vsi_rebuild for correct functioning. Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-08-02 16:25:21 +08:00
if (vsi->txq_map) {
devm_kfree(dev, vsi->txq_map);
ice: Alloc queue management bitmaps and arrays dynamically The total number of queues available on the device is divided between multiple physical functions (PF) in the firmware and provided to the driver when it gets function capabilities from the firmware. Thus each PF knows how many Tx/Rx queues it has. These queues are then doled out to different VSIs (for LAN traffic, SR-IOV VF traffic, etc.) To track usage of these queues at the PF level, the driver uses two bitmaps avail_txqs and avail_rxqs. At the VSI level (i.e. struct ice_vsi instances) the driver uses two arrays txq_map and rxq_map, to track ownership of VSIs' queues in avail_txqs and avail_rxqs respectively. The aforementioned bitmaps and arrays should be allocated dynamically, because the number of queues supported by a PF is only available once function capabilities have been queried. The current static allocation consumes way more memory than required. This patch removes the DECLARE_BITMAP for avail_txqs and avail_rxqs and instead uses bitmap_zalloc to allocate the bitmaps during init. Similarly txq_map and rxq_map are now allocated in ice_vsi_alloc_arrays. As a result ICE_MAX_TXQS and ICE_MAX_RXQS defines are no longer needed. Also as txq_map and rxq_map are now allocated and freed, some code reordering was required in ice_vsi_rebuild for correct functioning. Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-08-02 16:25:21 +08:00
vsi->txq_map = NULL;
}
if (vsi->rxq_map) {
devm_kfree(dev, vsi->rxq_map);
ice: Alloc queue management bitmaps and arrays dynamically The total number of queues available on the device is divided between multiple physical functions (PF) in the firmware and provided to the driver when it gets function capabilities from the firmware. Thus each PF knows how many Tx/Rx queues it has. These queues are then doled out to different VSIs (for LAN traffic, SR-IOV VF traffic, etc.) To track usage of these queues at the PF level, the driver uses two bitmaps avail_txqs and avail_rxqs. At the VSI level (i.e. struct ice_vsi instances) the driver uses two arrays txq_map and rxq_map, to track ownership of VSIs' queues in avail_txqs and avail_rxqs respectively. The aforementioned bitmaps and arrays should be allocated dynamically, because the number of queues supported by a PF is only available once function capabilities have been queried. The current static allocation consumes way more memory than required. This patch removes the DECLARE_BITMAP for avail_txqs and avail_rxqs and instead uses bitmap_zalloc to allocate the bitmaps during init. Similarly txq_map and rxq_map are now allocated in ice_vsi_alloc_arrays. As a result ICE_MAX_TXQS and ICE_MAX_RXQS defines are no longer needed. Also as txq_map and rxq_map are now allocated and freed, some code reordering was required in ice_vsi_rebuild for correct functioning. Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-08-02 16:25:21 +08:00
vsi->rxq_map = NULL;
}
}
/**
* ice_vsi_clear - clean up and deallocate the provided VSI
* @vsi: pointer to VSI being cleared
*
* This deallocates the VSI's queue resources, removes it from the PF's
* VSI array if necessary, and deallocates the VSI
*
* Returns 0 on success, negative on failure
*/
int ice_vsi_clear(struct ice_vsi *vsi)
{
struct ice_pf *pf = NULL;
struct device *dev;
if (!vsi)
return 0;
if (!vsi->back)
return -EINVAL;
pf = vsi->back;
dev = ice_pf_to_dev(pf);
if (!pf->vsi[vsi->idx] || pf->vsi[vsi->idx] != vsi) {
dev_dbg(dev, "vsi does not exist at pf->vsi[%d]\n", vsi->idx);
return -EINVAL;
}
mutex_lock(&pf->sw_mutex);
/* updates the PF for this cleared VSI */
pf->vsi[vsi->idx] = NULL;
if (vsi->idx < pf->next_vsi)
pf->next_vsi = vsi->idx;
ice_vsi_free_arrays(vsi);
mutex_unlock(&pf->sw_mutex);
devm_kfree(dev, vsi);
return 0;
}
/**
* ice_msix_clean_rings - MSIX mode Interrupt Handler
* @irq: interrupt number
* @data: pointer to a q_vector
*/
static irqreturn_t ice_msix_clean_rings(int __always_unused irq, void *data)
{
struct ice_q_vector *q_vector = (struct ice_q_vector *)data;
if (!q_vector->tx.ring && !q_vector->rx.ring)
return IRQ_HANDLED;
napi_schedule(&q_vector->napi);
return IRQ_HANDLED;
}
/**
* ice_vsi_alloc - Allocates the next available struct VSI in the PF
* @pf: board private structure
* @type: type of VSI
* @vf_id: ID of the VF being configured
*
* returns a pointer to a VSI on success, NULL on failure.
*/
static struct ice_vsi *
ice_vsi_alloc(struct ice_pf *pf, enum ice_vsi_type type, u16 vf_id)
{
struct device *dev = ice_pf_to_dev(pf);
struct ice_vsi *vsi = NULL;
/* Need to protect the allocation of the VSIs at the PF level */
mutex_lock(&pf->sw_mutex);
/* If we have already allocated our maximum number of VSIs,
* pf->next_vsi will be ICE_NO_VSI. If not, pf->next_vsi index
* is available to be populated
*/
if (pf->next_vsi == ICE_NO_VSI) {
dev_dbg(dev, "out of VSI slots!\n");
goto unlock_pf;
}
vsi = devm_kzalloc(dev, sizeof(*vsi), GFP_KERNEL);
if (!vsi)
goto unlock_pf;
vsi->type = type;
vsi->back = pf;
set_bit(__ICE_DOWN, vsi->state);
vsi->idx = pf->next_vsi;
if (type == ICE_VSI_VF)
ice_vsi_set_num_qs(vsi, vf_id);
else
ice_vsi_set_num_qs(vsi, ICE_INVAL_VFID);
switch (vsi->type) {
case ICE_VSI_PF:
if (ice_vsi_alloc_arrays(vsi))
goto err_rings;
/* Setup default MSIX irq handler for VSI */
vsi->irq_handler = ice_msix_clean_rings;
break;
case ICE_VSI_VF:
if (ice_vsi_alloc_arrays(vsi))
goto err_rings;
break;
case ICE_VSI_LB:
if (ice_vsi_alloc_arrays(vsi))
goto err_rings;
break;
default:
dev_warn(dev, "Unknown VSI type %d\n", vsi->type);
goto unlock_pf;
}
/* fill VSI slot in the PF struct */
pf->vsi[pf->next_vsi] = vsi;
/* prepare pf->next_vsi for next use */
pf->next_vsi = ice_get_free_slot(pf->vsi, pf->num_alloc_vsi,
pf->next_vsi);
goto unlock_pf;
err_rings:
devm_kfree(dev, vsi);
vsi = NULL;
unlock_pf:
mutex_unlock(&pf->sw_mutex);
return vsi;
}
/**
* ice_vsi_get_qs - Assign queues from PF to VSI
* @vsi: the VSI to assign queues to
*
* Returns 0 on success and a negative value on error
*/
static int ice_vsi_get_qs(struct ice_vsi *vsi)
{
struct ice_pf *pf = vsi->back;
struct ice_qs_cfg tx_qs_cfg = {
.qs_mutex = &pf->avail_q_mutex,
.pf_map = pf->avail_txqs,
ice: Alloc queue management bitmaps and arrays dynamically The total number of queues available on the device is divided between multiple physical functions (PF) in the firmware and provided to the driver when it gets function capabilities from the firmware. Thus each PF knows how many Tx/Rx queues it has. These queues are then doled out to different VSIs (for LAN traffic, SR-IOV VF traffic, etc.) To track usage of these queues at the PF level, the driver uses two bitmaps avail_txqs and avail_rxqs. At the VSI level (i.e. struct ice_vsi instances) the driver uses two arrays txq_map and rxq_map, to track ownership of VSIs' queues in avail_txqs and avail_rxqs respectively. The aforementioned bitmaps and arrays should be allocated dynamically, because the number of queues supported by a PF is only available once function capabilities have been queried. The current static allocation consumes way more memory than required. This patch removes the DECLARE_BITMAP for avail_txqs and avail_rxqs and instead uses bitmap_zalloc to allocate the bitmaps during init. Similarly txq_map and rxq_map are now allocated in ice_vsi_alloc_arrays. As a result ICE_MAX_TXQS and ICE_MAX_RXQS defines are no longer needed. Also as txq_map and rxq_map are now allocated and freed, some code reordering was required in ice_vsi_rebuild for correct functioning. Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-08-02 16:25:21 +08:00
.pf_map_size = pf->max_pf_txqs,
.q_count = vsi->alloc_txq,
.scatter_count = ICE_MAX_SCATTER_TXQS,
.vsi_map = vsi->txq_map,
.vsi_map_offset = 0,
.mapping_mode = vsi->tx_mapping_mode
};
struct ice_qs_cfg rx_qs_cfg = {
.qs_mutex = &pf->avail_q_mutex,
.pf_map = pf->avail_rxqs,
ice: Alloc queue management bitmaps and arrays dynamically The total number of queues available on the device is divided between multiple physical functions (PF) in the firmware and provided to the driver when it gets function capabilities from the firmware. Thus each PF knows how many Tx/Rx queues it has. These queues are then doled out to different VSIs (for LAN traffic, SR-IOV VF traffic, etc.) To track usage of these queues at the PF level, the driver uses two bitmaps avail_txqs and avail_rxqs. At the VSI level (i.e. struct ice_vsi instances) the driver uses two arrays txq_map and rxq_map, to track ownership of VSIs' queues in avail_txqs and avail_rxqs respectively. The aforementioned bitmaps and arrays should be allocated dynamically, because the number of queues supported by a PF is only available once function capabilities have been queried. The current static allocation consumes way more memory than required. This patch removes the DECLARE_BITMAP for avail_txqs and avail_rxqs and instead uses bitmap_zalloc to allocate the bitmaps during init. Similarly txq_map and rxq_map are now allocated in ice_vsi_alloc_arrays. As a result ICE_MAX_TXQS and ICE_MAX_RXQS defines are no longer needed. Also as txq_map and rxq_map are now allocated and freed, some code reordering was required in ice_vsi_rebuild for correct functioning. Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-08-02 16:25:21 +08:00
.pf_map_size = pf->max_pf_rxqs,
.q_count = vsi->alloc_rxq,
.scatter_count = ICE_MAX_SCATTER_RXQS,
.vsi_map = vsi->rxq_map,
.vsi_map_offset = 0,
.mapping_mode = vsi->rx_mapping_mode
};
int ret = 0;
vsi->tx_mapping_mode = ICE_VSI_MAP_CONTIG;
vsi->rx_mapping_mode = ICE_VSI_MAP_CONTIG;
ret = __ice_vsi_get_qs(&tx_qs_cfg);
if (!ret)
ret = __ice_vsi_get_qs(&rx_qs_cfg);
return ret;
}
/**
* ice_vsi_put_qs - Release queues from VSI to PF
* @vsi: the VSI that is going to release queues
*/
void ice_vsi_put_qs(struct ice_vsi *vsi)
{
struct ice_pf *pf = vsi->back;
int i;
mutex_lock(&pf->avail_q_mutex);
for (i = 0; i < vsi->alloc_txq; i++) {
clear_bit(vsi->txq_map[i], pf->avail_txqs);
vsi->txq_map[i] = ICE_INVAL_Q_INDEX;
}
for (i = 0; i < vsi->alloc_rxq; i++) {
clear_bit(vsi->rxq_map[i], pf->avail_rxqs);
vsi->rxq_map[i] = ICE_INVAL_Q_INDEX;
}
mutex_unlock(&pf->avail_q_mutex);
}
/**
* ice_is_safe_mode
* @pf: pointer to the PF struct
*
* returns true if driver is in safe mode, false otherwise
*/
bool ice_is_safe_mode(struct ice_pf *pf)
{
return !test_bit(ICE_FLAG_ADV_FEATURES, pf->flags);
}
/**
* ice_rss_clean - Delete RSS related VSI structures that hold user inputs
* @vsi: the VSI being removed
*/
static void ice_rss_clean(struct ice_vsi *vsi)
{
struct ice_pf *pf = vsi->back;
struct device *dev;
dev = ice_pf_to_dev(pf);
if (vsi->rss_hkey_user)
devm_kfree(dev, vsi->rss_hkey_user);
if (vsi->rss_lut_user)
devm_kfree(dev, vsi->rss_lut_user);
}
/**
* ice_vsi_set_rss_params - Setup RSS capabilities per VSI type
* @vsi: the VSI being configured
*/
static void ice_vsi_set_rss_params(struct ice_vsi *vsi)
{
struct ice_hw_common_caps *cap;
struct ice_pf *pf = vsi->back;
if (!test_bit(ICE_FLAG_RSS_ENA, pf->flags)) {
vsi->rss_size = 1;
return;
}
cap = &pf->hw.func_caps.common_cap;
switch (vsi->type) {
case ICE_VSI_PF:
/* PF VSI will inherit RSS instance of PF */
vsi->rss_table_size = cap->rss_table_size;
vsi->rss_size = min_t(int, num_online_cpus(),
BIT(cap->rss_table_entry_width));
vsi->rss_lut_type = ICE_AQC_GSET_RSS_LUT_TABLE_TYPE_PF;
break;
case ICE_VSI_VF:
/* VF VSI will gets a small RSS table
* For VSI_LUT, LUT size should be set to 64 bytes
*/
vsi->rss_table_size = ICE_VSIQF_HLUT_ARRAY_SIZE;
vsi->rss_size = min_t(int, num_online_cpus(),
BIT(cap->rss_table_entry_width));
vsi->rss_lut_type = ICE_AQC_GSET_RSS_LUT_TABLE_TYPE_VSI;
break;
case ICE_VSI_LB:
break;
default:
dev_warn(ice_pf_to_dev(pf), "Unknown VSI type %d\n",
vsi->type);
break;
}
}
/**
* ice_set_dflt_vsi_ctx - Set default VSI context before adding a VSI
* @ctxt: the VSI context being set
*
* This initializes a default VSI context for all sections except the Queues.
*/
static void ice_set_dflt_vsi_ctx(struct ice_vsi_ctx *ctxt)
{
u32 table = 0;
memset(&ctxt->info, 0, sizeof(ctxt->info));
/* VSI's should be allocated from shared pool */
ctxt->alloc_from_pool = true;
/* Src pruning enabled by default */
ctxt->info.sw_flags = ICE_AQ_VSI_SW_FLAG_SRC_PRUNE;
/* Traffic from VSI can be sent to LAN */
ctxt->info.sw_flags2 = ICE_AQ_VSI_SW_FLAG_LAN_ENA;
/* By default bits 3 and 4 in vlan_flags are 0's which results in legacy
* behavior (show VLAN, DEI, and UP) in descriptor. Also, allow all
* packets untagged/tagged.
*/
ctxt->info.vlan_flags = ((ICE_AQ_VSI_VLAN_MODE_ALL &
ICE_AQ_VSI_VLAN_MODE_M) >>
ICE_AQ_VSI_VLAN_MODE_S);
/* Have 1:1 UP mapping for both ingress/egress tables */
table |= ICE_UP_TABLE_TRANSLATE(0, 0);
table |= ICE_UP_TABLE_TRANSLATE(1, 1);
table |= ICE_UP_TABLE_TRANSLATE(2, 2);
table |= ICE_UP_TABLE_TRANSLATE(3, 3);
table |= ICE_UP_TABLE_TRANSLATE(4, 4);
table |= ICE_UP_TABLE_TRANSLATE(5, 5);
table |= ICE_UP_TABLE_TRANSLATE(6, 6);
table |= ICE_UP_TABLE_TRANSLATE(7, 7);
ctxt->info.ingress_table = cpu_to_le32(table);
ctxt->info.egress_table = cpu_to_le32(table);
/* Have 1:1 UP mapping for outer to inner UP table */
ctxt->info.outer_up_table = cpu_to_le32(table);
/* No Outer tag support outer_tag_flags remains to zero */
}
/**
* ice_vsi_setup_q_map - Setup a VSI queue map
* @vsi: the VSI being configured
* @ctxt: VSI context structure
*/
static void ice_vsi_setup_q_map(struct ice_vsi *vsi, struct ice_vsi_ctx *ctxt)
{
u16 offset = 0, qmap = 0, tx_count = 0;
u16 qcount_tx = vsi->alloc_txq;
u16 qcount_rx = vsi->alloc_rxq;
u16 tx_numq_tc, rx_numq_tc;
u16 pow = 0, max_rss = 0;
bool ena_tc0 = false;
u8 netdev_tc = 0;
int i;
/* at least TC0 should be enabled by default */
if (vsi->tc_cfg.numtc) {
if (!(vsi->tc_cfg.ena_tc & BIT(0)))
ena_tc0 = true;
} else {
ena_tc0 = true;
}
if (ena_tc0) {
vsi->tc_cfg.numtc++;
vsi->tc_cfg.ena_tc |= 1;
}
rx_numq_tc = qcount_rx / vsi->tc_cfg.numtc;
if (!rx_numq_tc)
rx_numq_tc = 1;
tx_numq_tc = qcount_tx / vsi->tc_cfg.numtc;
if (!tx_numq_tc)
tx_numq_tc = 1;
/* TC mapping is a function of the number of Rx queues assigned to the
* VSI for each traffic class and the offset of these queues.
* The first 10 bits are for queue offset for TC0, next 4 bits for no:of
* queues allocated to TC0. No:of queues is a power-of-2.
*
* If TC is not enabled, the queue offset is set to 0, and allocate one
* queue, this way, traffic for the given TC will be sent to the default
* queue.
*
* Setup number and offset of Rx queues for all TCs for the VSI
*/
qcount_rx = rx_numq_tc;
/* qcount will change if RSS is enabled */
if (test_bit(ICE_FLAG_RSS_ENA, vsi->back->flags)) {
if (vsi->type == ICE_VSI_PF || vsi->type == ICE_VSI_VF) {
if (vsi->type == ICE_VSI_PF)
max_rss = ICE_MAX_LG_RSS_QS;
else
max_rss = ICE_MAX_SMALL_RSS_QS;
qcount_rx = min_t(int, rx_numq_tc, max_rss);
if (!vsi->req_rxq)
qcount_rx = min_t(int, qcount_rx,
vsi->rss_size);
}
}
/* find the (rounded up) power-of-2 of qcount */
pow = order_base_2(qcount_rx);
ice_for_each_traffic_class(i) {
if (!(vsi->tc_cfg.ena_tc & BIT(i))) {
/* TC is not enabled */
vsi->tc_cfg.tc_info[i].qoffset = 0;
vsi->tc_cfg.tc_info[i].qcount_rx = 1;
vsi->tc_cfg.tc_info[i].qcount_tx = 1;
vsi->tc_cfg.tc_info[i].netdev_tc = 0;
ctxt->info.tc_mapping[i] = 0;
continue;
}
/* TC is enabled */
vsi->tc_cfg.tc_info[i].qoffset = offset;
vsi->tc_cfg.tc_info[i].qcount_rx = qcount_rx;
vsi->tc_cfg.tc_info[i].qcount_tx = tx_numq_tc;
vsi->tc_cfg.tc_info[i].netdev_tc = netdev_tc++;
qmap = ((offset << ICE_AQ_VSI_TC_Q_OFFSET_S) &
ICE_AQ_VSI_TC_Q_OFFSET_M) |
((pow << ICE_AQ_VSI_TC_Q_NUM_S) &
ICE_AQ_VSI_TC_Q_NUM_M);
offset += qcount_rx;
tx_count += tx_numq_tc;
ctxt->info.tc_mapping[i] = cpu_to_le16(qmap);
}
/* if offset is non-zero, means it is calculated correctly based on
* enabled TCs for a given VSI otherwise qcount_rx will always
* be correct and non-zero because it is based off - VSI's
* allocated Rx queues which is at least 1 (hence qcount_tx will be
* at least 1)
*/
if (offset)
vsi->num_rxq = offset;
else
vsi->num_rxq = qcount_rx;
vsi->num_txq = tx_count;
if (vsi->type == ICE_VSI_VF && vsi->num_txq != vsi->num_rxq) {
dev_dbg(&vsi->back->pdev->dev, "VF VSI should have same number of Tx and Rx queues. Hence making them equal\n");
/* since there is a chance that num_rxq could have been changed
* in the above for loop, make num_txq equal to num_rxq.
*/
vsi->num_txq = vsi->num_rxq;
}
/* Rx queue mapping */
ctxt->info.mapping_flags |= cpu_to_le16(ICE_AQ_VSI_Q_MAP_CONTIG);
/* q_mapping buffer holds the info for the first queue allocated for
* this VSI in the PF space and also the number of queues associated
* with this VSI.
*/
ctxt->info.q_mapping[0] = cpu_to_le16(vsi->rxq_map[0]);
ctxt->info.q_mapping[1] = cpu_to_le16(vsi->num_rxq);
}
/**
* ice_set_rss_vsi_ctx - Set RSS VSI context before adding a VSI
* @ctxt: the VSI context being set
* @vsi: the VSI being configured
*/
static void ice_set_rss_vsi_ctx(struct ice_vsi_ctx *ctxt, struct ice_vsi *vsi)
{
u8 lut_type, hash_type;
struct device *dev;
struct ice_pf *pf;
pf = vsi->back;
dev = ice_pf_to_dev(pf);
switch (vsi->type) {
case ICE_VSI_PF:
/* PF VSI will inherit RSS instance of PF */
lut_type = ICE_AQ_VSI_Q_OPT_RSS_LUT_PF;
hash_type = ICE_AQ_VSI_Q_OPT_RSS_TPLZ;
break;
case ICE_VSI_VF:
/* VF VSI will gets a small RSS table which is a VSI LUT type */
lut_type = ICE_AQ_VSI_Q_OPT_RSS_LUT_VSI;
hash_type = ICE_AQ_VSI_Q_OPT_RSS_TPLZ;
break;
case ICE_VSI_LB:
dev_dbg(dev, "Unsupported VSI type %s\n",
ice_vsi_type_str(vsi->type));
return;
default:
dev_warn(dev, "Unknown VSI type %d\n", vsi->type);
return;
}
ctxt->info.q_opt_rss = ((lut_type << ICE_AQ_VSI_Q_OPT_RSS_LUT_S) &
ICE_AQ_VSI_Q_OPT_RSS_LUT_M) |
((hash_type << ICE_AQ_VSI_Q_OPT_RSS_HASH_S) &
ICE_AQ_VSI_Q_OPT_RSS_HASH_M);
}
/**
* ice_vsi_init - Create and initialize a VSI
* @vsi: the VSI being configured
* @init_vsi: is this call creating a VSI
*
* This initializes a VSI context depending on the VSI type to be added and
* passes it down to the add_vsi aq command to create a new VSI.
*/
static int ice_vsi_init(struct ice_vsi *vsi, bool init_vsi)
{
struct ice_pf *pf = vsi->back;
struct ice_hw *hw = &pf->hw;
struct ice_vsi_ctx *ctxt;
struct device *dev;
int ret = 0;
dev = ice_pf_to_dev(pf);
ctxt = kzalloc(sizeof(*ctxt), GFP_KERNEL);
if (!ctxt)
return -ENOMEM;
ctxt->info = vsi->info;
switch (vsi->type) {
case ICE_VSI_LB:
/* fall through */
case ICE_VSI_PF:
ctxt->flags = ICE_AQ_VSI_TYPE_PF;
break;
case ICE_VSI_VF:
ctxt->flags = ICE_AQ_VSI_TYPE_VF;
/* VF number here is the absolute VF number (0-255) */
ctxt->vf_num = vsi->vf_id + hw->func_caps.vf_base_id;
break;
default:
ret = -ENODEV;
goto out;
}
ice_set_dflt_vsi_ctx(ctxt);
/* if the switch is in VEB mode, allow VSI loopback */
if (vsi->vsw->bridge_mode == BRIDGE_MODE_VEB)
ctxt->info.sw_flags |= ICE_AQ_VSI_SW_FLAG_ALLOW_LB;
/* Set LUT type and HASH type if RSS is enabled */
if (test_bit(ICE_FLAG_RSS_ENA, pf->flags)) {
ice_set_rss_vsi_ctx(ctxt, vsi);
/* if updating VSI context, make sure to set valid_section:
* to indicate which section of VSI context being updated
*/
if (!init_vsi)
ctxt->info.valid_sections |=
cpu_to_le16(ICE_AQ_VSI_PROP_Q_OPT_VALID);
}
ctxt->info.sw_id = vsi->port_info->sw_id;
ice_vsi_setup_q_map(vsi, ctxt);
if (!init_vsi) /* means VSI being updated */
/* must to indicate which section of VSI context are
* being modified
*/
ctxt->info.valid_sections |=
cpu_to_le16(ICE_AQ_VSI_PROP_RXQ_MAP_VALID);
ice: Fix VF spoofchk There are many things wrong with the function ice_set_vf_spoofchk(). 1. The VSI being modified is the PF VSI, not the VF VSI. 2. We are enabling Rx VLAN pruning instead of Tx VLAN anti-spoof. 3. The spoofchk setting for each VF is not initialized correctly or re-initialized correctly on reset. To fix [1] we need to make sure we are modifying the VF VSI. This is done by using the vf->lan_vsi_idx to index into the PF's VSI array. To fix [2] replace setting Rx VLAN pruning in ice_set_vf_spoofchk() with setting Tx VLAN anti-spoof. To Fix [3] we need to make sure the initial VSI settings match what is done in ice_set_vf_spoofchk() for spoofchk=on. Also make sure this also works for VF reset. This was done by modifying ice_vsi_init() to account for the current spoofchk state of the VF VSI. Because of these changes, Tx VLAN anti-spoof needs to be removed from ice_cfg_vlan_pruning(). This is okay for the VF because this is now controlled from the admin enabling/disabling spoofchk. For the PF, Tx VLAN anti-spoof should not be set. This change requires us to call ice_set_vf_spoofchk() when configuring promiscuous mode for the VF which requires ice_set_vf_spoofchk() to move in order to prevent a forward declaration prototype. Also, add VLAN 0 by default when allocating a VF since the PF is unaware if the guest OS is running the 8021q module. Without this, MDD events will trigger on untagged traffic because spoofcheck is enabled by default. Due to this change, ignore add/delete messages for VLAN 0 from VIRTCHNL since this is added/deleted during VF initialization/teardown respectively and should not be modified. Signed-off-by: Brett Creeley <brett.creeley@intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-12-12 19:12:54 +08:00
/* enable/disable MAC and VLAN anti-spoof when spoofchk is on/off
* respectively
*/
if (vsi->type == ICE_VSI_VF) {
ctxt->info.valid_sections |=
cpu_to_le16(ICE_AQ_VSI_PROP_SECURITY_VALID);
ice: Fix VF spoofchk There are many things wrong with the function ice_set_vf_spoofchk(). 1. The VSI being modified is the PF VSI, not the VF VSI. 2. We are enabling Rx VLAN pruning instead of Tx VLAN anti-spoof. 3. The spoofchk setting for each VF is not initialized correctly or re-initialized correctly on reset. To fix [1] we need to make sure we are modifying the VF VSI. This is done by using the vf->lan_vsi_idx to index into the PF's VSI array. To fix [2] replace setting Rx VLAN pruning in ice_set_vf_spoofchk() with setting Tx VLAN anti-spoof. To Fix [3] we need to make sure the initial VSI settings match what is done in ice_set_vf_spoofchk() for spoofchk=on. Also make sure this also works for VF reset. This was done by modifying ice_vsi_init() to account for the current spoofchk state of the VF VSI. Because of these changes, Tx VLAN anti-spoof needs to be removed from ice_cfg_vlan_pruning(). This is okay for the VF because this is now controlled from the admin enabling/disabling spoofchk. For the PF, Tx VLAN anti-spoof should not be set. This change requires us to call ice_set_vf_spoofchk() when configuring promiscuous mode for the VF which requires ice_set_vf_spoofchk() to move in order to prevent a forward declaration prototype. Also, add VLAN 0 by default when allocating a VF since the PF is unaware if the guest OS is running the 8021q module. Without this, MDD events will trigger on untagged traffic because spoofcheck is enabled by default. Due to this change, ignore add/delete messages for VLAN 0 from VIRTCHNL since this is added/deleted during VF initialization/teardown respectively and should not be modified. Signed-off-by: Brett Creeley <brett.creeley@intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-12-12 19:12:54 +08:00
if (pf->vf[vsi->vf_id].spoofchk) {
ctxt->info.sec_flags |=
ICE_AQ_VSI_SEC_FLAG_ENA_MAC_ANTI_SPOOF |
(ICE_AQ_VSI_SEC_TX_VLAN_PRUNE_ENA <<
ICE_AQ_VSI_SEC_TX_PRUNE_ENA_S);
} else {
ctxt->info.sec_flags &=
~(ICE_AQ_VSI_SEC_FLAG_ENA_MAC_ANTI_SPOOF |
(ICE_AQ_VSI_SEC_TX_VLAN_PRUNE_ENA <<
ICE_AQ_VSI_SEC_TX_PRUNE_ENA_S));
}
}
/* Allow control frames out of main VSI */
if (vsi->type == ICE_VSI_PF) {
ctxt->info.sec_flags |= ICE_AQ_VSI_SEC_FLAG_ALLOW_DEST_OVRD;
ctxt->info.valid_sections |=
cpu_to_le16(ICE_AQ_VSI_PROP_SECURITY_VALID);
}
if (init_vsi) {
ret = ice_add_vsi(hw, vsi->idx, ctxt, NULL);
if (ret) {
dev_err(dev, "Add VSI failed, err %d\n", ret);
ret = -EIO;
goto out;
}
} else {
ret = ice_update_vsi(hw, vsi->idx, ctxt, NULL);
if (ret) {
dev_err(dev, "Update VSI failed, err %d\n", ret);
ret = -EIO;
goto out;
}
}
/* keep context for update VSI operations */
vsi->info = ctxt->info;
/* record VSI number returned */
vsi->vsi_num = ctxt->vsi_num;
out:
kfree(ctxt);
return ret;
}
/**
* ice_vsi_setup_vector_base - Set up the base vector for the given VSI
* @vsi: ptr to the VSI
*
* This should only be called after ice_vsi_alloc() which allocates the
* corresponding SW VSI structure and initializes num_queue_pairs for the
* newly allocated VSI.
*
* Returns 0 on success or negative on failure
*/
static int ice_vsi_setup_vector_base(struct ice_vsi *vsi)
{
struct ice_pf *pf = vsi->back;
struct device *dev;
ice: Refactor interrupt tracking Currently we have two MSI-x (IRQ) trackers, one for OS requested MSI-x entries (sw_irq_tracker) and one for hardware MSI-x vectors (hw_irq_tracker). Generally the sw_irq_tracker has less entries than the hw_irq_tracker because the hw_irq_tracker has entries equal to the max allowed MSI-x per PF and the sw_irq_tracker is mainly the minimum (non SR-IOV portion of the vectors, kernel granted IRQs). All of the non SR-IOV portions of the driver (i.e. LAN queues, RDMA queues, OICR, etc.) take at least one of each type of tracker resource. SR-IOV only grabs entries from the hw_irq_tracker. There are a few issues with this approach that can be seen when doing any kind of device reconfiguration (i.e. ethtool -L, SR-IOV, etc.). One of them being, any time the driver creates an ice_q_vector and associates it to a LAN queue pair it will grab and use one entry from the hw_irq_tracker and one from the sw_irq_tracker. If the indices on these does not match it will cause a Tx timeout, which will cause a reset and then the indices will match up again and traffic will resume. The mismatched indices come from the trackers not being the same size and/or the search_hint in the two trackers not being equal. Another reason for the refactor is the co-existence of features with SR-IOV. If SR-IOV is enabled and the interrupts are taken from the end of the sw_irq_tracker then other features can no longer use this space because the hardware has now given the remaining interrupts to SR-IOV. This patch reworks how we track MSI-x vectors by removing the hw_irq_tracker completely and instead MSI-x resources needed for SR-IOV are determined all at once instead of per VF. This can be done because when creating VFs we know how many are wanted and how many MSI-x vectors each VF needs. This also allows us to start using MSI-x resources from the end of the PF's allowed MSI-x vectors so we are less likely to use entries needed for other features (i.e. RDMA, L2 Offload, etc). This patch also reworks the ice_res_tracker structure by removing the search_hint and adding a new member - "end". Instead of having a search_hint we will always search from 0. The new member, "end", will be used to manipulate the end of the ice_res_tracker (specifically sw_irq_tracker) during runtime based on MSI-x vectors needed by SR-IOV. In the normal case, the end of ice_res_tracker will be equal to the ice_res_tracker's num_entries. The sriov_base_vector member was added to the PF structure. It is used to represent the starting MSI-x index of all the needed MSI-x vectors for all SR-IOV VFs. Depending on how many MSI-x are needed, SR-IOV may have to take resources from the sw_irq_tracker. This is done by setting the sw_irq_tracker->end equal to the pf->sriov_base_vector. When all SR-IOV VFs are removed then the sw_irq_tracker->end is reset back to sw_irq_tracker->num_entries. The sriov_base_vector, along with the VF's number of MSI-x (pf->num_vf_msix), vf_id, and the base MSI-x index on the PF (pf->hw.func_caps.common_cap.msix_vector_first_id), is used to calculate the first HW absolute MSI-x index for each VF, which is used to write to the VPINT_ALLOC[_PCI] and GLINT_VECT2FUNC registers to program the VFs MSI-x PCI configuration bits. Also, the sriov_base_vector is used along with VF's num_vf_msix, vf_id, and q_vector->v_idx to determine the MSI-x register index (used for writing to GLINT_DYN_CTL) within the PF's space. Interrupt changes removed any references to hw_base_vector, hw_oicr_idx, and hw_irq_tracker. Only sw_base_vector, sw_oicr_idx, and sw_irq_tracker variables remain. Change all of these by removing the "sw_" prefix to help avoid confusion with these variables and their use. Signed-off-by: Brett Creeley <brett.creeley@intel.com> Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-04-17 01:30:44 +08:00
u16 num_q_vectors;
dev = ice_pf_to_dev(pf);
ice: Refactor interrupt tracking Currently we have two MSI-x (IRQ) trackers, one for OS requested MSI-x entries (sw_irq_tracker) and one for hardware MSI-x vectors (hw_irq_tracker). Generally the sw_irq_tracker has less entries than the hw_irq_tracker because the hw_irq_tracker has entries equal to the max allowed MSI-x per PF and the sw_irq_tracker is mainly the minimum (non SR-IOV portion of the vectors, kernel granted IRQs). All of the non SR-IOV portions of the driver (i.e. LAN queues, RDMA queues, OICR, etc.) take at least one of each type of tracker resource. SR-IOV only grabs entries from the hw_irq_tracker. There are a few issues with this approach that can be seen when doing any kind of device reconfiguration (i.e. ethtool -L, SR-IOV, etc.). One of them being, any time the driver creates an ice_q_vector and associates it to a LAN queue pair it will grab and use one entry from the hw_irq_tracker and one from the sw_irq_tracker. If the indices on these does not match it will cause a Tx timeout, which will cause a reset and then the indices will match up again and traffic will resume. The mismatched indices come from the trackers not being the same size and/or the search_hint in the two trackers not being equal. Another reason for the refactor is the co-existence of features with SR-IOV. If SR-IOV is enabled and the interrupts are taken from the end of the sw_irq_tracker then other features can no longer use this space because the hardware has now given the remaining interrupts to SR-IOV. This patch reworks how we track MSI-x vectors by removing the hw_irq_tracker completely and instead MSI-x resources needed for SR-IOV are determined all at once instead of per VF. This can be done because when creating VFs we know how many are wanted and how many MSI-x vectors each VF needs. This also allows us to start using MSI-x resources from the end of the PF's allowed MSI-x vectors so we are less likely to use entries needed for other features (i.e. RDMA, L2 Offload, etc). This patch also reworks the ice_res_tracker structure by removing the search_hint and adding a new member - "end". Instead of having a search_hint we will always search from 0. The new member, "end", will be used to manipulate the end of the ice_res_tracker (specifically sw_irq_tracker) during runtime based on MSI-x vectors needed by SR-IOV. In the normal case, the end of ice_res_tracker will be equal to the ice_res_tracker's num_entries. The sriov_base_vector member was added to the PF structure. It is used to represent the starting MSI-x index of all the needed MSI-x vectors for all SR-IOV VFs. Depending on how many MSI-x are needed, SR-IOV may have to take resources from the sw_irq_tracker. This is done by setting the sw_irq_tracker->end equal to the pf->sriov_base_vector. When all SR-IOV VFs are removed then the sw_irq_tracker->end is reset back to sw_irq_tracker->num_entries. The sriov_base_vector, along with the VF's number of MSI-x (pf->num_vf_msix), vf_id, and the base MSI-x index on the PF (pf->hw.func_caps.common_cap.msix_vector_first_id), is used to calculate the first HW absolute MSI-x index for each VF, which is used to write to the VPINT_ALLOC[_PCI] and GLINT_VECT2FUNC registers to program the VFs MSI-x PCI configuration bits. Also, the sriov_base_vector is used along with VF's num_vf_msix, vf_id, and q_vector->v_idx to determine the MSI-x register index (used for writing to GLINT_DYN_CTL) within the PF's space. Interrupt changes removed any references to hw_base_vector, hw_oicr_idx, and hw_irq_tracker. Only sw_base_vector, sw_oicr_idx, and sw_irq_tracker variables remain. Change all of these by removing the "sw_" prefix to help avoid confusion with these variables and their use. Signed-off-by: Brett Creeley <brett.creeley@intel.com> Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-04-17 01:30:44 +08:00
/* SRIOV doesn't grab irq_tracker entries for each VSI */
if (vsi->type == ICE_VSI_VF)
return 0;
if (vsi->base_vector) {
dev_dbg(dev, "VSI %d has non-zero base vector %d\n",
ice: Refactor interrupt tracking Currently we have two MSI-x (IRQ) trackers, one for OS requested MSI-x entries (sw_irq_tracker) and one for hardware MSI-x vectors (hw_irq_tracker). Generally the sw_irq_tracker has less entries than the hw_irq_tracker because the hw_irq_tracker has entries equal to the max allowed MSI-x per PF and the sw_irq_tracker is mainly the minimum (non SR-IOV portion of the vectors, kernel granted IRQs). All of the non SR-IOV portions of the driver (i.e. LAN queues, RDMA queues, OICR, etc.) take at least one of each type of tracker resource. SR-IOV only grabs entries from the hw_irq_tracker. There are a few issues with this approach that can be seen when doing any kind of device reconfiguration (i.e. ethtool -L, SR-IOV, etc.). One of them being, any time the driver creates an ice_q_vector and associates it to a LAN queue pair it will grab and use one entry from the hw_irq_tracker and one from the sw_irq_tracker. If the indices on these does not match it will cause a Tx timeout, which will cause a reset and then the indices will match up again and traffic will resume. The mismatched indices come from the trackers not being the same size and/or the search_hint in the two trackers not being equal. Another reason for the refactor is the co-existence of features with SR-IOV. If SR-IOV is enabled and the interrupts are taken from the end of the sw_irq_tracker then other features can no longer use this space because the hardware has now given the remaining interrupts to SR-IOV. This patch reworks how we track MSI-x vectors by removing the hw_irq_tracker completely and instead MSI-x resources needed for SR-IOV are determined all at once instead of per VF. This can be done because when creating VFs we know how many are wanted and how many MSI-x vectors each VF needs. This also allows us to start using MSI-x resources from the end of the PF's allowed MSI-x vectors so we are less likely to use entries needed for other features (i.e. RDMA, L2 Offload, etc). This patch also reworks the ice_res_tracker structure by removing the search_hint and adding a new member - "end". Instead of having a search_hint we will always search from 0. The new member, "end", will be used to manipulate the end of the ice_res_tracker (specifically sw_irq_tracker) during runtime based on MSI-x vectors needed by SR-IOV. In the normal case, the end of ice_res_tracker will be equal to the ice_res_tracker's num_entries. The sriov_base_vector member was added to the PF structure. It is used to represent the starting MSI-x index of all the needed MSI-x vectors for all SR-IOV VFs. Depending on how many MSI-x are needed, SR-IOV may have to take resources from the sw_irq_tracker. This is done by setting the sw_irq_tracker->end equal to the pf->sriov_base_vector. When all SR-IOV VFs are removed then the sw_irq_tracker->end is reset back to sw_irq_tracker->num_entries. The sriov_base_vector, along with the VF's number of MSI-x (pf->num_vf_msix), vf_id, and the base MSI-x index on the PF (pf->hw.func_caps.common_cap.msix_vector_first_id), is used to calculate the first HW absolute MSI-x index for each VF, which is used to write to the VPINT_ALLOC[_PCI] and GLINT_VECT2FUNC registers to program the VFs MSI-x PCI configuration bits. Also, the sriov_base_vector is used along with VF's num_vf_msix, vf_id, and q_vector->v_idx to determine the MSI-x register index (used for writing to GLINT_DYN_CTL) within the PF's space. Interrupt changes removed any references to hw_base_vector, hw_oicr_idx, and hw_irq_tracker. Only sw_base_vector, sw_oicr_idx, and sw_irq_tracker variables remain. Change all of these by removing the "sw_" prefix to help avoid confusion with these variables and their use. Signed-off-by: Brett Creeley <brett.creeley@intel.com> Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-04-17 01:30:44 +08:00
vsi->vsi_num, vsi->base_vector);
return -EEXIST;
}
ice: Refactor interrupt tracking Currently we have two MSI-x (IRQ) trackers, one for OS requested MSI-x entries (sw_irq_tracker) and one for hardware MSI-x vectors (hw_irq_tracker). Generally the sw_irq_tracker has less entries than the hw_irq_tracker because the hw_irq_tracker has entries equal to the max allowed MSI-x per PF and the sw_irq_tracker is mainly the minimum (non SR-IOV portion of the vectors, kernel granted IRQs). All of the non SR-IOV portions of the driver (i.e. LAN queues, RDMA queues, OICR, etc.) take at least one of each type of tracker resource. SR-IOV only grabs entries from the hw_irq_tracker. There are a few issues with this approach that can be seen when doing any kind of device reconfiguration (i.e. ethtool -L, SR-IOV, etc.). One of them being, any time the driver creates an ice_q_vector and associates it to a LAN queue pair it will grab and use one entry from the hw_irq_tracker and one from the sw_irq_tracker. If the indices on these does not match it will cause a Tx timeout, which will cause a reset and then the indices will match up again and traffic will resume. The mismatched indices come from the trackers not being the same size and/or the search_hint in the two trackers not being equal. Another reason for the refactor is the co-existence of features with SR-IOV. If SR-IOV is enabled and the interrupts are taken from the end of the sw_irq_tracker then other features can no longer use this space because the hardware has now given the remaining interrupts to SR-IOV. This patch reworks how we track MSI-x vectors by removing the hw_irq_tracker completely and instead MSI-x resources needed for SR-IOV are determined all at once instead of per VF. This can be done because when creating VFs we know how many are wanted and how many MSI-x vectors each VF needs. This also allows us to start using MSI-x resources from the end of the PF's allowed MSI-x vectors so we are less likely to use entries needed for other features (i.e. RDMA, L2 Offload, etc). This patch also reworks the ice_res_tracker structure by removing the search_hint and adding a new member - "end". Instead of having a search_hint we will always search from 0. The new member, "end", will be used to manipulate the end of the ice_res_tracker (specifically sw_irq_tracker) during runtime based on MSI-x vectors needed by SR-IOV. In the normal case, the end of ice_res_tracker will be equal to the ice_res_tracker's num_entries. The sriov_base_vector member was added to the PF structure. It is used to represent the starting MSI-x index of all the needed MSI-x vectors for all SR-IOV VFs. Depending on how many MSI-x are needed, SR-IOV may have to take resources from the sw_irq_tracker. This is done by setting the sw_irq_tracker->end equal to the pf->sriov_base_vector. When all SR-IOV VFs are removed then the sw_irq_tracker->end is reset back to sw_irq_tracker->num_entries. The sriov_base_vector, along with the VF's number of MSI-x (pf->num_vf_msix), vf_id, and the base MSI-x index on the PF (pf->hw.func_caps.common_cap.msix_vector_first_id), is used to calculate the first HW absolute MSI-x index for each VF, which is used to write to the VPINT_ALLOC[_PCI] and GLINT_VECT2FUNC registers to program the VFs MSI-x PCI configuration bits. Also, the sriov_base_vector is used along with VF's num_vf_msix, vf_id, and q_vector->v_idx to determine the MSI-x register index (used for writing to GLINT_DYN_CTL) within the PF's space. Interrupt changes removed any references to hw_base_vector, hw_oicr_idx, and hw_irq_tracker. Only sw_base_vector, sw_oicr_idx, and sw_irq_tracker variables remain. Change all of these by removing the "sw_" prefix to help avoid confusion with these variables and their use. Signed-off-by: Brett Creeley <brett.creeley@intel.com> Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-04-17 01:30:44 +08:00
num_q_vectors = vsi->num_q_vectors;
/* reserve slots from OS requested IRQs */
vsi->base_vector = ice_get_res(pf, pf->irq_tracker, num_q_vectors,
vsi->idx);
if (vsi->base_vector < 0) {
dev_err(dev,
ice: Refactor interrupt tracking Currently we have two MSI-x (IRQ) trackers, one for OS requested MSI-x entries (sw_irq_tracker) and one for hardware MSI-x vectors (hw_irq_tracker). Generally the sw_irq_tracker has less entries than the hw_irq_tracker because the hw_irq_tracker has entries equal to the max allowed MSI-x per PF and the sw_irq_tracker is mainly the minimum (non SR-IOV portion of the vectors, kernel granted IRQs). All of the non SR-IOV portions of the driver (i.e. LAN queues, RDMA queues, OICR, etc.) take at least one of each type of tracker resource. SR-IOV only grabs entries from the hw_irq_tracker. There are a few issues with this approach that can be seen when doing any kind of device reconfiguration (i.e. ethtool -L, SR-IOV, etc.). One of them being, any time the driver creates an ice_q_vector and associates it to a LAN queue pair it will grab and use one entry from the hw_irq_tracker and one from the sw_irq_tracker. If the indices on these does not match it will cause a Tx timeout, which will cause a reset and then the indices will match up again and traffic will resume. The mismatched indices come from the trackers not being the same size and/or the search_hint in the two trackers not being equal. Another reason for the refactor is the co-existence of features with SR-IOV. If SR-IOV is enabled and the interrupts are taken from the end of the sw_irq_tracker then other features can no longer use this space because the hardware has now given the remaining interrupts to SR-IOV. This patch reworks how we track MSI-x vectors by removing the hw_irq_tracker completely and instead MSI-x resources needed for SR-IOV are determined all at once instead of per VF. This can be done because when creating VFs we know how many are wanted and how many MSI-x vectors each VF needs. This also allows us to start using MSI-x resources from the end of the PF's allowed MSI-x vectors so we are less likely to use entries needed for other features (i.e. RDMA, L2 Offload, etc). This patch also reworks the ice_res_tracker structure by removing the search_hint and adding a new member - "end". Instead of having a search_hint we will always search from 0. The new member, "end", will be used to manipulate the end of the ice_res_tracker (specifically sw_irq_tracker) during runtime based on MSI-x vectors needed by SR-IOV. In the normal case, the end of ice_res_tracker will be equal to the ice_res_tracker's num_entries. The sriov_base_vector member was added to the PF structure. It is used to represent the starting MSI-x index of all the needed MSI-x vectors for all SR-IOV VFs. Depending on how many MSI-x are needed, SR-IOV may have to take resources from the sw_irq_tracker. This is done by setting the sw_irq_tracker->end equal to the pf->sriov_base_vector. When all SR-IOV VFs are removed then the sw_irq_tracker->end is reset back to sw_irq_tracker->num_entries. The sriov_base_vector, along with the VF's number of MSI-x (pf->num_vf_msix), vf_id, and the base MSI-x index on the PF (pf->hw.func_caps.common_cap.msix_vector_first_id), is used to calculate the first HW absolute MSI-x index for each VF, which is used to write to the VPINT_ALLOC[_PCI] and GLINT_VECT2FUNC registers to program the VFs MSI-x PCI configuration bits. Also, the sriov_base_vector is used along with VF's num_vf_msix, vf_id, and q_vector->v_idx to determine the MSI-x register index (used for writing to GLINT_DYN_CTL) within the PF's space. Interrupt changes removed any references to hw_base_vector, hw_oicr_idx, and hw_irq_tracker. Only sw_base_vector, sw_oicr_idx, and sw_irq_tracker variables remain. Change all of these by removing the "sw_" prefix to help avoid confusion with these variables and their use. Signed-off-by: Brett Creeley <brett.creeley@intel.com> Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-04-17 01:30:44 +08:00
"Failed to get tracking for %d vectors for VSI %d, err=%d\n",
num_q_vectors, vsi->vsi_num, vsi->base_vector);
return -ENOENT;
}
ice: Refactor interrupt tracking Currently we have two MSI-x (IRQ) trackers, one for OS requested MSI-x entries (sw_irq_tracker) and one for hardware MSI-x vectors (hw_irq_tracker). Generally the sw_irq_tracker has less entries than the hw_irq_tracker because the hw_irq_tracker has entries equal to the max allowed MSI-x per PF and the sw_irq_tracker is mainly the minimum (non SR-IOV portion of the vectors, kernel granted IRQs). All of the non SR-IOV portions of the driver (i.e. LAN queues, RDMA queues, OICR, etc.) take at least one of each type of tracker resource. SR-IOV only grabs entries from the hw_irq_tracker. There are a few issues with this approach that can be seen when doing any kind of device reconfiguration (i.e. ethtool -L, SR-IOV, etc.). One of them being, any time the driver creates an ice_q_vector and associates it to a LAN queue pair it will grab and use one entry from the hw_irq_tracker and one from the sw_irq_tracker. If the indices on these does not match it will cause a Tx timeout, which will cause a reset and then the indices will match up again and traffic will resume. The mismatched indices come from the trackers not being the same size and/or the search_hint in the two trackers not being equal. Another reason for the refactor is the co-existence of features with SR-IOV. If SR-IOV is enabled and the interrupts are taken from the end of the sw_irq_tracker then other features can no longer use this space because the hardware has now given the remaining interrupts to SR-IOV. This patch reworks how we track MSI-x vectors by removing the hw_irq_tracker completely and instead MSI-x resources needed for SR-IOV are determined all at once instead of per VF. This can be done because when creating VFs we know how many are wanted and how many MSI-x vectors each VF needs. This also allows us to start using MSI-x resources from the end of the PF's allowed MSI-x vectors so we are less likely to use entries needed for other features (i.e. RDMA, L2 Offload, etc). This patch also reworks the ice_res_tracker structure by removing the search_hint and adding a new member - "end". Instead of having a search_hint we will always search from 0. The new member, "end", will be used to manipulate the end of the ice_res_tracker (specifically sw_irq_tracker) during runtime based on MSI-x vectors needed by SR-IOV. In the normal case, the end of ice_res_tracker will be equal to the ice_res_tracker's num_entries. The sriov_base_vector member was added to the PF structure. It is used to represent the starting MSI-x index of all the needed MSI-x vectors for all SR-IOV VFs. Depending on how many MSI-x are needed, SR-IOV may have to take resources from the sw_irq_tracker. This is done by setting the sw_irq_tracker->end equal to the pf->sriov_base_vector. When all SR-IOV VFs are removed then the sw_irq_tracker->end is reset back to sw_irq_tracker->num_entries. The sriov_base_vector, along with the VF's number of MSI-x (pf->num_vf_msix), vf_id, and the base MSI-x index on the PF (pf->hw.func_caps.common_cap.msix_vector_first_id), is used to calculate the first HW absolute MSI-x index for each VF, which is used to write to the VPINT_ALLOC[_PCI] and GLINT_VECT2FUNC registers to program the VFs MSI-x PCI configuration bits. Also, the sriov_base_vector is used along with VF's num_vf_msix, vf_id, and q_vector->v_idx to determine the MSI-x register index (used for writing to GLINT_DYN_CTL) within the PF's space. Interrupt changes removed any references to hw_base_vector, hw_oicr_idx, and hw_irq_tracker. Only sw_base_vector, sw_oicr_idx, and sw_irq_tracker variables remain. Change all of these by removing the "sw_" prefix to help avoid confusion with these variables and their use. Signed-off-by: Brett Creeley <brett.creeley@intel.com> Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-04-17 01:30:44 +08:00
pf->num_avail_sw_msix -= num_q_vectors;
return 0;
}
/**
* ice_vsi_clear_rings - Deallocates the Tx and Rx rings for VSI
* @vsi: the VSI having rings deallocated
*/
static void ice_vsi_clear_rings(struct ice_vsi *vsi)
{
int i;
if (vsi->tx_rings) {
for (i = 0; i < vsi->alloc_txq; i++) {
if (vsi->tx_rings[i]) {
kfree_rcu(vsi->tx_rings[i], rcu);
vsi->tx_rings[i] = NULL;
}
}
}
if (vsi->rx_rings) {
for (i = 0; i < vsi->alloc_rxq; i++) {
if (vsi->rx_rings[i]) {
kfree_rcu(vsi->rx_rings[i], rcu);
vsi->rx_rings[i] = NULL;
}
}
}
}
/**
* ice_vsi_alloc_rings - Allocates Tx and Rx rings for the VSI
* @vsi: VSI which is having rings allocated
*/
static int ice_vsi_alloc_rings(struct ice_vsi *vsi)
{
struct ice_pf *pf = vsi->back;
struct device *dev;
int i;
dev = ice_pf_to_dev(pf);
/* Allocate Tx rings */
for (i = 0; i < vsi->alloc_txq; i++) {
struct ice_ring *ring;
/* allocate with kzalloc(), free with kfree_rcu() */
ring = kzalloc(sizeof(*ring), GFP_KERNEL);
if (!ring)
goto err_out;
ring->q_index = i;
ring->reg_idx = vsi->txq_map[i];
ring->ring_active = false;
ring->vsi = vsi;
ring->dev = dev;
ring->count = vsi->num_tx_desc;
vsi->tx_rings[i] = ring;
}
/* Allocate Rx rings */
for (i = 0; i < vsi->alloc_rxq; i++) {
struct ice_ring *ring;
/* allocate with kzalloc(), free with kfree_rcu() */
ring = kzalloc(sizeof(*ring), GFP_KERNEL);
if (!ring)
goto err_out;
ring->q_index = i;
ring->reg_idx = vsi->rxq_map[i];
ring->ring_active = false;
ring->vsi = vsi;
ring->netdev = vsi->netdev;
ring->dev = dev;
ring->count = vsi->num_rx_desc;
vsi->rx_rings[i] = ring;
}
return 0;
err_out:
ice_vsi_clear_rings(vsi);
return -ENOMEM;
}
/**
* ice_vsi_manage_rss_lut - disable/enable RSS
* @vsi: the VSI being changed
* @ena: boolean value indicating if this is an enable or disable request
*
* In the event of disable request for RSS, this function will zero out RSS
* LUT, while in the event of enable request for RSS, it will reconfigure RSS
* LUT.
*/
int ice_vsi_manage_rss_lut(struct ice_vsi *vsi, bool ena)
{
int err = 0;
u8 *lut;
lut = kzalloc(vsi->rss_table_size, GFP_KERNEL);
if (!lut)
return -ENOMEM;
if (ena) {
if (vsi->rss_lut_user)
memcpy(lut, vsi->rss_lut_user, vsi->rss_table_size);
else
ice_fill_rss_lut(lut, vsi->rss_table_size,
vsi->rss_size);
}
err = ice_set_rss(vsi, NULL, lut, vsi->rss_table_size);
kfree(lut);
return err;
}
/**
* ice_vsi_cfg_rss_lut_key - Configure RSS params for a VSI
* @vsi: VSI to be configured
*/
static int ice_vsi_cfg_rss_lut_key(struct ice_vsi *vsi)
{
struct ice_aqc_get_set_rss_keys *key;
struct ice_pf *pf = vsi->back;
enum ice_status status;
struct device *dev;
int err = 0;
u8 *lut;
dev = ice_pf_to_dev(pf);
vsi->rss_size = min_t(int, vsi->rss_size, vsi->num_rxq);
lut = kzalloc(vsi->rss_table_size, GFP_KERNEL);
if (!lut)
return -ENOMEM;
if (vsi->rss_lut_user)
memcpy(lut, vsi->rss_lut_user, vsi->rss_table_size);
else
ice_fill_rss_lut(lut, vsi->rss_table_size, vsi->rss_size);
status = ice_aq_set_rss_lut(&pf->hw, vsi->idx, vsi->rss_lut_type, lut,
vsi->rss_table_size);
if (status) {
dev_err(dev, "set_rss_lut failed, error %d\n", status);
err = -EIO;
goto ice_vsi_cfg_rss_exit;
}
key = kzalloc(sizeof(*key), GFP_KERNEL);
if (!key) {
err = -ENOMEM;
goto ice_vsi_cfg_rss_exit;
}
if (vsi->rss_hkey_user)
memcpy(key,
(struct ice_aqc_get_set_rss_keys *)vsi->rss_hkey_user,
ICE_GET_SET_RSS_KEY_EXTEND_KEY_SIZE);
else
netdev_rss_key_fill((void *)key,
ICE_GET_SET_RSS_KEY_EXTEND_KEY_SIZE);
status = ice_aq_set_rss_key(&pf->hw, vsi->idx, key);
if (status) {
dev_err(dev, "set_rss_key failed, error %d\n", status);
err = -EIO;
}
kfree(key);
ice_vsi_cfg_rss_exit:
kfree(lut);
return err;
}
/**
* ice_vsi_set_rss_flow_fld - Sets RSS input set for different flows
* @vsi: VSI to be configured
*
* This function will only be called after successful download package call
* during initialization of PF. Since the downloaded package will erase the
* RSS section, this function will configure RSS input sets for different
* flow types. The last profile added has the highest priority, therefore 2
* tuple profiles (i.e. IPv4 src/dst) are added before 4 tuple profiles
* (i.e. IPv4 src/dst TCP src/dst port).
*/
static void ice_vsi_set_rss_flow_fld(struct ice_vsi *vsi)
{
u16 vsi_handle = vsi->idx, vsi_num = vsi->vsi_num;
struct ice_pf *pf = vsi->back;
struct ice_hw *hw = &pf->hw;
enum ice_status status;
struct device *dev;
dev = ice_pf_to_dev(pf);
if (ice_is_safe_mode(pf)) {
dev_dbg(dev, "Advanced RSS disabled. Package download failed, vsi num = %d\n",
vsi_num);
return;
}
/* configure RSS for IPv4 with input set IP src/dst */
status = ice_add_rss_cfg(hw, vsi_handle, ICE_FLOW_HASH_IPV4,
ICE_FLOW_SEG_HDR_IPV4);
if (status)
dev_dbg(dev, "ice_add_rss_cfg failed for ipv4 flow, vsi = %d, error = %d\n",
vsi_num, status);
/* configure RSS for IPv6 with input set IPv6 src/dst */
status = ice_add_rss_cfg(hw, vsi_handle, ICE_FLOW_HASH_IPV6,
ICE_FLOW_SEG_HDR_IPV6);
if (status)
dev_dbg(dev, "ice_add_rss_cfg failed for ipv6 flow, vsi = %d, error = %d\n",
vsi_num, status);
/* configure RSS for tcp4 with input set IP src/dst, TCP src/dst */
status = ice_add_rss_cfg(hw, vsi_handle, ICE_HASH_TCP_IPV4,
ICE_FLOW_SEG_HDR_TCP | ICE_FLOW_SEG_HDR_IPV4);
if (status)
dev_dbg(dev, "ice_add_rss_cfg failed for tcp4 flow, vsi = %d, error = %d\n",
vsi_num, status);
/* configure RSS for udp4 with input set IP src/dst, UDP src/dst */
status = ice_add_rss_cfg(hw, vsi_handle, ICE_HASH_UDP_IPV4,
ICE_FLOW_SEG_HDR_UDP | ICE_FLOW_SEG_HDR_IPV4);
if (status)
dev_dbg(dev, "ice_add_rss_cfg failed for udp4 flow, vsi = %d, error = %d\n",
vsi_num, status);
/* configure RSS for sctp4 with input set IP src/dst */
status = ice_add_rss_cfg(hw, vsi_handle, ICE_FLOW_HASH_IPV4,
ICE_FLOW_SEG_HDR_SCTP | ICE_FLOW_SEG_HDR_IPV4);
if (status)
dev_dbg(dev, "ice_add_rss_cfg failed for sctp4 flow, vsi = %d, error = %d\n",
vsi_num, status);
/* configure RSS for tcp6 with input set IPv6 src/dst, TCP src/dst */
status = ice_add_rss_cfg(hw, vsi_handle, ICE_HASH_TCP_IPV6,
ICE_FLOW_SEG_HDR_TCP | ICE_FLOW_SEG_HDR_IPV6);
if (status)
dev_dbg(dev, "ice_add_rss_cfg failed for tcp6 flow, vsi = %d, error = %d\n",
vsi_num, status);
/* configure RSS for udp6 with input set IPv6 src/dst, UDP src/dst */
status = ice_add_rss_cfg(hw, vsi_handle, ICE_HASH_UDP_IPV6,
ICE_FLOW_SEG_HDR_UDP | ICE_FLOW_SEG_HDR_IPV6);
if (status)
dev_dbg(dev, "ice_add_rss_cfg failed for udp6 flow, vsi = %d, error = %d\n",
vsi_num, status);
/* configure RSS for sctp6 with input set IPv6 src/dst */
status = ice_add_rss_cfg(hw, vsi_handle, ICE_FLOW_HASH_IPV6,
ICE_FLOW_SEG_HDR_SCTP | ICE_FLOW_SEG_HDR_IPV6);
if (status)
dev_dbg(dev, "ice_add_rss_cfg failed for sctp6 flow, vsi = %d, error = %d\n",
vsi_num, status);
}
/**
* ice_add_mac_to_list - Add a MAC address filter entry to the list
* @vsi: the VSI to be forwarded to
* @add_list: pointer to the list which contains MAC filter entries
* @macaddr: the MAC address to be added.
*
* Adds MAC address filter entry to the temp list
*
* Returns 0 on success or ENOMEM on failure.
*/
int ice_add_mac_to_list(struct ice_vsi *vsi, struct list_head *add_list,
const u8 *macaddr)
{
struct ice_fltr_list_entry *tmp;
struct ice_pf *pf = vsi->back;
tmp = devm_kzalloc(ice_pf_to_dev(pf), sizeof(*tmp), GFP_ATOMIC);
if (!tmp)
return -ENOMEM;
tmp->fltr_info.flag = ICE_FLTR_TX;
tmp->fltr_info.src_id = ICE_SRC_ID_VSI;
tmp->fltr_info.lkup_type = ICE_SW_LKUP_MAC;
tmp->fltr_info.fltr_act = ICE_FWD_TO_VSI;
tmp->fltr_info.vsi_handle = vsi->idx;
ether_addr_copy(tmp->fltr_info.l_data.mac.mac_addr, macaddr);
INIT_LIST_HEAD(&tmp->list_entry);
list_add(&tmp->list_entry, add_list);
return 0;
}
/**
* ice_update_eth_stats - Update VSI-specific ethernet statistics counters
* @vsi: the VSI to be updated
*/
void ice_update_eth_stats(struct ice_vsi *vsi)
{
struct ice_eth_stats *prev_es, *cur_es;
struct ice_hw *hw = &vsi->back->hw;
u16 vsi_num = vsi->vsi_num; /* HW absolute index of a VSI */
prev_es = &vsi->eth_stats_prev;
cur_es = &vsi->eth_stats;
ice_stat_update40(hw, GLV_GORCL(vsi_num), vsi->stat_offsets_loaded,
&prev_es->rx_bytes, &cur_es->rx_bytes);
ice_stat_update40(hw, GLV_UPRCL(vsi_num), vsi->stat_offsets_loaded,
&prev_es->rx_unicast, &cur_es->rx_unicast);
ice_stat_update40(hw, GLV_MPRCL(vsi_num), vsi->stat_offsets_loaded,
&prev_es->rx_multicast, &cur_es->rx_multicast);
ice_stat_update40(hw, GLV_BPRCL(vsi_num), vsi->stat_offsets_loaded,
&prev_es->rx_broadcast, &cur_es->rx_broadcast);
ice_stat_update32(hw, GLV_RDPC(vsi_num), vsi->stat_offsets_loaded,
&prev_es->rx_discards, &cur_es->rx_discards);
ice_stat_update40(hw, GLV_GOTCL(vsi_num), vsi->stat_offsets_loaded,
&prev_es->tx_bytes, &cur_es->tx_bytes);
ice_stat_update40(hw, GLV_UPTCL(vsi_num), vsi->stat_offsets_loaded,
&prev_es->tx_unicast, &cur_es->tx_unicast);
ice_stat_update40(hw, GLV_MPTCL(vsi_num), vsi->stat_offsets_loaded,
&prev_es->tx_multicast, &cur_es->tx_multicast);
ice_stat_update40(hw, GLV_BPTCL(vsi_num), vsi->stat_offsets_loaded,
&prev_es->tx_broadcast, &cur_es->tx_broadcast);
ice_stat_update32(hw, GLV_TEPC(vsi_num), vsi->stat_offsets_loaded,
&prev_es->tx_errors, &cur_es->tx_errors);
vsi->stat_offsets_loaded = true;
}
/**
* ice_free_fltr_list - free filter lists helper
* @dev: pointer to the device struct
* @h: pointer to the list head to be freed
*
* Helper function to free filter lists previously created using
* ice_add_mac_to_list
*/
void ice_free_fltr_list(struct device *dev, struct list_head *h)
{
struct ice_fltr_list_entry *e, *tmp;
list_for_each_entry_safe(e, tmp, h, list_entry) {
list_del(&e->list_entry);
devm_kfree(dev, e);
}
}
/**
* ice_vsi_add_vlan - Add VSI membership for given VLAN
* @vsi: the VSI being configured
* @vid: VLAN ID to be added
*/
int ice_vsi_add_vlan(struct ice_vsi *vsi, u16 vid)
{
struct ice_fltr_list_entry *tmp;
struct ice_pf *pf = vsi->back;
LIST_HEAD(tmp_add_list);
enum ice_status status;
struct device *dev;
int err = 0;
dev = ice_pf_to_dev(pf);
tmp = devm_kzalloc(dev, sizeof(*tmp), GFP_KERNEL);
if (!tmp)
return -ENOMEM;
tmp->fltr_info.lkup_type = ICE_SW_LKUP_VLAN;
tmp->fltr_info.fltr_act = ICE_FWD_TO_VSI;
tmp->fltr_info.flag = ICE_FLTR_TX;
tmp->fltr_info.src_id = ICE_SRC_ID_VSI;
tmp->fltr_info.vsi_handle = vsi->idx;
tmp->fltr_info.l_data.vlan.vlan_id = vid;
INIT_LIST_HEAD(&tmp->list_entry);
list_add(&tmp->list_entry, &tmp_add_list);
status = ice_add_vlan(&pf->hw, &tmp_add_list);
if (status) {
err = -ENODEV;
dev_err(dev, "Failure Adding VLAN %d on VSI %i\n", vid,
vsi->vsi_num);
}
ice_free_fltr_list(dev, &tmp_add_list);
return err;
}
/**
* ice_vsi_kill_vlan - Remove VSI membership for a given VLAN
* @vsi: the VSI being configured
* @vid: VLAN ID to be removed
*
* Returns 0 on success and negative on failure
*/
int ice_vsi_kill_vlan(struct ice_vsi *vsi, u16 vid)
{
struct ice_fltr_list_entry *list;
struct ice_pf *pf = vsi->back;
LIST_HEAD(tmp_add_list);
enum ice_status status;
struct device *dev;
int err = 0;
dev = ice_pf_to_dev(pf);
list = devm_kzalloc(dev, sizeof(*list), GFP_KERNEL);
if (!list)
return -ENOMEM;
list->fltr_info.lkup_type = ICE_SW_LKUP_VLAN;
list->fltr_info.vsi_handle = vsi->idx;
list->fltr_info.fltr_act = ICE_FWD_TO_VSI;
list->fltr_info.l_data.vlan.vlan_id = vid;
list->fltr_info.flag = ICE_FLTR_TX;
list->fltr_info.src_id = ICE_SRC_ID_VSI;
INIT_LIST_HEAD(&list->list_entry);
list_add(&list->list_entry, &tmp_add_list);
status = ice_remove_vlan(&pf->hw, &tmp_add_list);
if (status == ICE_ERR_DOES_NOT_EXIST) {
dev_dbg(dev,
"Failed to remove VLAN %d on VSI %i, it does not exist, status: %d\n",
vid, vsi->vsi_num, status);
} else if (status) {
dev_err(dev,
"Error removing VLAN %d on vsi %i error: %d\n",
vid, vsi->vsi_num, status);
err = -EIO;
}
ice_free_fltr_list(dev, &tmp_add_list);
return err;
}
/**
* ice_vsi_cfg_frame_size - setup max frame size and Rx buffer length
* @vsi: VSI
*/
void ice_vsi_cfg_frame_size(struct ice_vsi *vsi)
{
ice: introduce legacy Rx flag Add an ethtool "legacy-rx" priv flag for toggling the Rx path. This control knob will be mainly used for build_skb usage as well as buffer size/MTU manipulation. In preparation for adding build_skb support in a way that it takes care of how we set the values of max_frame and rx_buf_len fields of struct ice_vsi. Specifically, in this patch mentioned fields are set to values that will allow us to provide headroom and tailroom in-place. This can be mostly broken down onto following: - for legacy-rx "on" ethtool control knob, old behaviour is kept; - for standard 1500 MTU size configure the buffer of size 1536, as network stack is expecting the NET_SKB_PAD to be provided and NET_IP_ALIGN can have a non-zero value (these can be typically equal to 32 and 2, respectively); - for larger MTUs go with max_frame set to 9k and configure the 3k buffer in case when PAGE_SIZE of underlying arch is less than 8k; 3k buffer is implying the need for order 1 page, so that our page recycling scheme can still be applied; With that said, substitute the hardcoded ICE_RXBUF_2048 and PAGE_SIZE values in DMA API that we're making use of with rx_ring->rx_buf_len and ice_rx_pg_size(rx_ring). The latter is an introduced helper for determining the page size based on its order (which was figured out via ice_rx_pg_order). Last but not least, take care of truesize calculation. In the followup patch the headroom/tailroom computation logic will be introduced. This change aligns the buffer and frame configuration with other Intel drivers, most importantly with iavf. Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-10-24 16:11:22 +08:00
if (!vsi->netdev || test_bit(ICE_FLAG_LEGACY_RX, vsi->back->flags)) {
vsi->max_frame = ICE_AQ_SET_MAC_FRAME_SIZE_MAX;
vsi->rx_buf_len = ICE_RXBUF_2048;
#if (PAGE_SIZE < 8192)
} else if (!ICE_2K_TOO_SMALL_WITH_PADDING &&
(vsi->netdev->mtu <= ETH_DATA_LEN)) {
ice: introduce legacy Rx flag Add an ethtool "legacy-rx" priv flag for toggling the Rx path. This control knob will be mainly used for build_skb usage as well as buffer size/MTU manipulation. In preparation for adding build_skb support in a way that it takes care of how we set the values of max_frame and rx_buf_len fields of struct ice_vsi. Specifically, in this patch mentioned fields are set to values that will allow us to provide headroom and tailroom in-place. This can be mostly broken down onto following: - for legacy-rx "on" ethtool control knob, old behaviour is kept; - for standard 1500 MTU size configure the buffer of size 1536, as network stack is expecting the NET_SKB_PAD to be provided and NET_IP_ALIGN can have a non-zero value (these can be typically equal to 32 and 2, respectively); - for larger MTUs go with max_frame set to 9k and configure the 3k buffer in case when PAGE_SIZE of underlying arch is less than 8k; 3k buffer is implying the need for order 1 page, so that our page recycling scheme can still be applied; With that said, substitute the hardcoded ICE_RXBUF_2048 and PAGE_SIZE values in DMA API that we're making use of with rx_ring->rx_buf_len and ice_rx_pg_size(rx_ring). The latter is an introduced helper for determining the page size based on its order (which was figured out via ice_rx_pg_order). Last but not least, take care of truesize calculation. In the followup patch the headroom/tailroom computation logic will be introduced. This change aligns the buffer and frame configuration with other Intel drivers, most importantly with iavf. Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-10-24 16:11:22 +08:00
vsi->max_frame = ICE_RXBUF_1536 - NET_IP_ALIGN;
vsi->rx_buf_len = ICE_RXBUF_1536 - NET_IP_ALIGN;
#endif
} else {
vsi->max_frame = ICE_AQ_SET_MAC_FRAME_SIZE_MAX;
#if (PAGE_SIZE < 8192)
vsi->rx_buf_len = ICE_RXBUF_3072;
#else
vsi->rx_buf_len = ICE_RXBUF_2048;
#endif
}
}
/**
* ice_vsi_cfg_rxqs - Configure the VSI for Rx
* @vsi: the VSI being configured
*
* Return 0 on success and a negative value on error
* Configure the Rx VSI for operation.
*/
int ice_vsi_cfg_rxqs(struct ice_vsi *vsi)
{
u16 i;
if (vsi->type == ICE_VSI_VF)
goto setup_rings;
ice_vsi_cfg_frame_size(vsi);
setup_rings:
/* set up individual rings */
for (i = 0; i < vsi->num_rxq; i++) {
int err;
err = ice_setup_rx_ctx(vsi->rx_rings[i]);
if (err) {
dev_err(&vsi->back->pdev->dev,
"ice_setup_rx_ctx failed for RxQ %d, err %d\n",
i, err);
return err;
}
}
return 0;
}
/**
* ice_vsi_cfg_txqs - Configure the VSI for Tx
* @vsi: the VSI being configured
* @rings: Tx ring array to be configured
*
* Return 0 on success and a negative value on error
* Configure the Tx VSI for operation.
*/
static int
ice_vsi_cfg_txqs(struct ice_vsi *vsi, struct ice_ring **rings)
{
struct ice_aqc_add_tx_qgrp *qg_buf;
u16 q_idx = 0;
int err = 0;
qg_buf = kzalloc(sizeof(*qg_buf), GFP_KERNEL);
if (!qg_buf)
return -ENOMEM;
qg_buf->num_txqs = 1;
for (q_idx = 0; q_idx < vsi->num_txq; q_idx++) {
err = ice_vsi_cfg_txq(vsi, rings[q_idx], qg_buf);
if (err)
goto err_cfg_txqs;
}
err_cfg_txqs:
kfree(qg_buf);
return err;
}
/**
* ice_vsi_cfg_lan_txqs - Configure the VSI for Tx
* @vsi: the VSI being configured
*
* Return 0 on success and a negative value on error
* Configure the Tx VSI for operation.
*/
int ice_vsi_cfg_lan_txqs(struct ice_vsi *vsi)
{
return ice_vsi_cfg_txqs(vsi, vsi->tx_rings);
}
/**
* ice_vsi_cfg_xdp_txqs - Configure Tx queues dedicated for XDP in given VSI
* @vsi: the VSI being configured
*
* Return 0 on success and a negative value on error
* Configure the Tx queues dedicated for XDP in given VSI for operation.
*/
int ice_vsi_cfg_xdp_txqs(struct ice_vsi *vsi)
{
int ret;
int i;
ret = ice_vsi_cfg_txqs(vsi, vsi->xdp_rings);
if (ret)
return ret;
for (i = 0; i < vsi->num_xdp_txq; i++)
vsi->xdp_rings[i]->xsk_umem = ice_xsk_umem(vsi->xdp_rings[i]);
return ret;
}
/**
* ice_intrl_usec_to_reg - convert interrupt rate limit to register value
* @intrl: interrupt rate limit in usecs
* @gran: interrupt rate limit granularity in usecs
*
* This function converts a decimal interrupt rate limit in usecs to the format
* expected by firmware.
*/
u32 ice_intrl_usec_to_reg(u8 intrl, u8 gran)
{
u32 val = intrl / gran;
if (val)
return val | GLINT_RATE_INTRL_ENA_M;
return 0;
}
/**
* ice_vsi_cfg_msix - MSIX mode Interrupt Config in the HW
* @vsi: the VSI being configured
*
* This configures MSIX mode interrupts for the PF VSI, and should not be used
* for the VF VSI.
*/
void ice_vsi_cfg_msix(struct ice_vsi *vsi)
{
struct ice_pf *pf = vsi->back;
struct ice_hw *hw = &pf->hw;
u32 txq = 0, rxq = 0;
int i, q;
for (i = 0; i < vsi->num_q_vectors; i++) {
struct ice_q_vector *q_vector = vsi->q_vectors[i];
u16 reg_idx = q_vector->reg_idx;
ice_cfg_itr(hw, q_vector);
wr32(hw, GLINT_RATE(reg_idx),
ice_intrl_usec_to_reg(q_vector->intrl, hw->intrl_gran));
/* Both Transmit Queue Interrupt Cause Control register
* and Receive Queue Interrupt Cause control register
* expects MSIX_INDX field to be the vector index
* within the function space and not the absolute
* vector index across PF or across device.
* For SR-IOV VF VSIs queue vector index always starts
* with 1 since first vector index(0) is used for OICR
* in VF space. Since VMDq and other PF VSIs are within
* the PF function space, use the vector index that is
* tracked for this PF.
*/
for (q = 0; q < q_vector->num_ring_tx; q++) {
ice_cfg_txq_interrupt(vsi, txq, reg_idx,
q_vector->tx.itr_idx);
txq++;
}
for (q = 0; q < q_vector->num_ring_rx; q++) {
ice_cfg_rxq_interrupt(vsi, rxq, reg_idx,
q_vector->rx.itr_idx);
rxq++;
}
}
}
/**
* ice_vsi_manage_vlan_insertion - Manage VLAN insertion for the VSI for Tx
* @vsi: the VSI being changed
*/
int ice_vsi_manage_vlan_insertion(struct ice_vsi *vsi)
{
struct ice_hw *hw = &vsi->back->hw;
struct ice_vsi_ctx *ctxt;
enum ice_status status;
int ret = 0;
ctxt = kzalloc(sizeof(*ctxt), GFP_KERNEL);
if (!ctxt)
return -ENOMEM;
/* Here we are configuring the VSI to let the driver add VLAN tags by
* setting vlan_flags to ICE_AQ_VSI_VLAN_MODE_ALL. The actual VLAN tag
* insertion happens in the Tx hot path, in ice_tx_map.
*/
ctxt->info.vlan_flags = ICE_AQ_VSI_VLAN_MODE_ALL;
/* Preserve existing VLAN strip setting */
ctxt->info.vlan_flags |= (vsi->info.vlan_flags &
ICE_AQ_VSI_VLAN_EMOD_M);
ctxt->info.valid_sections = cpu_to_le16(ICE_AQ_VSI_PROP_VLAN_VALID);
status = ice_update_vsi(hw, vsi->idx, ctxt, NULL);
if (status) {
dev_err(&vsi->back->pdev->dev, "update VSI for VLAN insert failed, err %d aq_err %d\n",
status, hw->adminq.sq_last_status);
ret = -EIO;
goto out;
}
vsi->info.vlan_flags = ctxt->info.vlan_flags;
out:
kfree(ctxt);
return ret;
}
/**
* ice_vsi_manage_vlan_stripping - Manage VLAN stripping for the VSI for Rx
* @vsi: the VSI being changed
* @ena: boolean value indicating if this is a enable or disable request
*/
int ice_vsi_manage_vlan_stripping(struct ice_vsi *vsi, bool ena)
{
struct ice_hw *hw = &vsi->back->hw;
struct ice_vsi_ctx *ctxt;
enum ice_status status;
int ret = 0;
ctxt = kzalloc(sizeof(*ctxt), GFP_KERNEL);
if (!ctxt)
return -ENOMEM;
/* Here we are configuring what the VSI should do with the VLAN tag in
* the Rx packet. We can either leave the tag in the packet or put it in
* the Rx descriptor.
*/
if (ena)
/* Strip VLAN tag from Rx packet and put it in the desc */
ctxt->info.vlan_flags = ICE_AQ_VSI_VLAN_EMOD_STR_BOTH;
else
/* Disable stripping. Leave tag in packet */
ctxt->info.vlan_flags = ICE_AQ_VSI_VLAN_EMOD_NOTHING;
/* Allow all packets untagged/tagged */
ctxt->info.vlan_flags |= ICE_AQ_VSI_VLAN_MODE_ALL;
ctxt->info.valid_sections = cpu_to_le16(ICE_AQ_VSI_PROP_VLAN_VALID);
status = ice_update_vsi(hw, vsi->idx, ctxt, NULL);
if (status) {
dev_err(&vsi->back->pdev->dev, "update VSI for VLAN strip failed, ena = %d err %d aq_err %d\n",
ena, status, hw->adminq.sq_last_status);
ret = -EIO;
goto out;
}
vsi->info.vlan_flags = ctxt->info.vlan_flags;
out:
kfree(ctxt);
return ret;
}
/**
* ice_vsi_start_rx_rings - start VSI's Rx rings
* @vsi: the VSI whose rings are to be started
*
* Returns 0 on success and a negative value on error
*/
int ice_vsi_start_rx_rings(struct ice_vsi *vsi)
{
return ice_vsi_ctrl_rx_rings(vsi, true);
}
/**
* ice_vsi_stop_rx_rings - stop VSI's Rx rings
* @vsi: the VSI
*
* Returns 0 on success and a negative value on error
*/
int ice_vsi_stop_rx_rings(struct ice_vsi *vsi)
{
return ice_vsi_ctrl_rx_rings(vsi, false);
}
/**
* ice_vsi_stop_tx_rings - Disable Tx rings
* @vsi: the VSI being configured
* @rst_src: reset source
* @rel_vmvf_num: Relative ID of VF/VM
* @rings: Tx ring array to be stopped
*/
static int
ice_vsi_stop_tx_rings(struct ice_vsi *vsi, enum ice_disq_rst_src rst_src,
u16 rel_vmvf_num, struct ice_ring **rings)
{
u16 q_idx;
if (vsi->num_txq > ICE_LAN_TXQ_MAX_QDIS)
return -EINVAL;
for (q_idx = 0; q_idx < vsi->num_txq; q_idx++) {
struct ice_txq_meta txq_meta = { };
int status;
if (!rings || !rings[q_idx])
return -EINVAL;
ice_fill_txq_meta(vsi, rings[q_idx], &txq_meta);
status = ice_vsi_stop_tx_ring(vsi, rst_src, rel_vmvf_num,
rings[q_idx], &txq_meta);
if (status)
return status;
}
return 0;
}
/**
* ice_vsi_stop_lan_tx_rings - Disable LAN Tx rings
* @vsi: the VSI being configured
* @rst_src: reset source
* @rel_vmvf_num: Relative ID of VF/VM
*/
int
ice_vsi_stop_lan_tx_rings(struct ice_vsi *vsi, enum ice_disq_rst_src rst_src,
u16 rel_vmvf_num)
{
return ice_vsi_stop_tx_rings(vsi, rst_src, rel_vmvf_num, vsi->tx_rings);
}
/**
* ice_vsi_stop_xdp_tx_rings - Disable XDP Tx rings
* @vsi: the VSI being configured
*/
int ice_vsi_stop_xdp_tx_rings(struct ice_vsi *vsi)
{
return ice_vsi_stop_tx_rings(vsi, ICE_NO_RESET, 0, vsi->xdp_rings);
}
/**
* ice_cfg_vlan_pruning - enable or disable VLAN pruning on the VSI
* @vsi: VSI to enable or disable VLAN pruning on
* @ena: set to true to enable VLAN pruning and false to disable it
* @vlan_promisc: enable valid security flags if not in VLAN promiscuous mode
*
* returns 0 if VSI is updated, negative otherwise
*/
int ice_cfg_vlan_pruning(struct ice_vsi *vsi, bool ena, bool vlan_promisc)
{
struct ice_vsi_ctx *ctxt;
struct ice_pf *pf;
int status;
if (!vsi)
return -EINVAL;
pf = vsi->back;
ctxt = kzalloc(sizeof(*ctxt), GFP_KERNEL);
if (!ctxt)
return -ENOMEM;
ctxt->info = vsi->info;
ice: Fix VF spoofchk There are many things wrong with the function ice_set_vf_spoofchk(). 1. The VSI being modified is the PF VSI, not the VF VSI. 2. We are enabling Rx VLAN pruning instead of Tx VLAN anti-spoof. 3. The spoofchk setting for each VF is not initialized correctly or re-initialized correctly on reset. To fix [1] we need to make sure we are modifying the VF VSI. This is done by using the vf->lan_vsi_idx to index into the PF's VSI array. To fix [2] replace setting Rx VLAN pruning in ice_set_vf_spoofchk() with setting Tx VLAN anti-spoof. To Fix [3] we need to make sure the initial VSI settings match what is done in ice_set_vf_spoofchk() for spoofchk=on. Also make sure this also works for VF reset. This was done by modifying ice_vsi_init() to account for the current spoofchk state of the VF VSI. Because of these changes, Tx VLAN anti-spoof needs to be removed from ice_cfg_vlan_pruning(). This is okay for the VF because this is now controlled from the admin enabling/disabling spoofchk. For the PF, Tx VLAN anti-spoof should not be set. This change requires us to call ice_set_vf_spoofchk() when configuring promiscuous mode for the VF which requires ice_set_vf_spoofchk() to move in order to prevent a forward declaration prototype. Also, add VLAN 0 by default when allocating a VF since the PF is unaware if the guest OS is running the 8021q module. Without this, MDD events will trigger on untagged traffic because spoofcheck is enabled by default. Due to this change, ignore add/delete messages for VLAN 0 from VIRTCHNL since this is added/deleted during VF initialization/teardown respectively and should not be modified. Signed-off-by: Brett Creeley <brett.creeley@intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-12-12 19:12:54 +08:00
if (ena)
ctxt->info.sw_flags2 |= ICE_AQ_VSI_SW_FLAG_RX_VLAN_PRUNE_ENA;
ice: Fix VF spoofchk There are many things wrong with the function ice_set_vf_spoofchk(). 1. The VSI being modified is the PF VSI, not the VF VSI. 2. We are enabling Rx VLAN pruning instead of Tx VLAN anti-spoof. 3. The spoofchk setting for each VF is not initialized correctly or re-initialized correctly on reset. To fix [1] we need to make sure we are modifying the VF VSI. This is done by using the vf->lan_vsi_idx to index into the PF's VSI array. To fix [2] replace setting Rx VLAN pruning in ice_set_vf_spoofchk() with setting Tx VLAN anti-spoof. To Fix [3] we need to make sure the initial VSI settings match what is done in ice_set_vf_spoofchk() for spoofchk=on. Also make sure this also works for VF reset. This was done by modifying ice_vsi_init() to account for the current spoofchk state of the VF VSI. Because of these changes, Tx VLAN anti-spoof needs to be removed from ice_cfg_vlan_pruning(). This is okay for the VF because this is now controlled from the admin enabling/disabling spoofchk. For the PF, Tx VLAN anti-spoof should not be set. This change requires us to call ice_set_vf_spoofchk() when configuring promiscuous mode for the VF which requires ice_set_vf_spoofchk() to move in order to prevent a forward declaration prototype. Also, add VLAN 0 by default when allocating a VF since the PF is unaware if the guest OS is running the 8021q module. Without this, MDD events will trigger on untagged traffic because spoofcheck is enabled by default. Due to this change, ignore add/delete messages for VLAN 0 from VIRTCHNL since this is added/deleted during VF initialization/teardown respectively and should not be modified. Signed-off-by: Brett Creeley <brett.creeley@intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-12-12 19:12:54 +08:00
else
ctxt->info.sw_flags2 &= ~ICE_AQ_VSI_SW_FLAG_RX_VLAN_PRUNE_ENA;
if (!vlan_promisc)
ctxt->info.valid_sections =
ice: Fix VF spoofchk There are many things wrong with the function ice_set_vf_spoofchk(). 1. The VSI being modified is the PF VSI, not the VF VSI. 2. We are enabling Rx VLAN pruning instead of Tx VLAN anti-spoof. 3. The spoofchk setting for each VF is not initialized correctly or re-initialized correctly on reset. To fix [1] we need to make sure we are modifying the VF VSI. This is done by using the vf->lan_vsi_idx to index into the PF's VSI array. To fix [2] replace setting Rx VLAN pruning in ice_set_vf_spoofchk() with setting Tx VLAN anti-spoof. To Fix [3] we need to make sure the initial VSI settings match what is done in ice_set_vf_spoofchk() for spoofchk=on. Also make sure this also works for VF reset. This was done by modifying ice_vsi_init() to account for the current spoofchk state of the VF VSI. Because of these changes, Tx VLAN anti-spoof needs to be removed from ice_cfg_vlan_pruning(). This is okay for the VF because this is now controlled from the admin enabling/disabling spoofchk. For the PF, Tx VLAN anti-spoof should not be set. This change requires us to call ice_set_vf_spoofchk() when configuring promiscuous mode for the VF which requires ice_set_vf_spoofchk() to move in order to prevent a forward declaration prototype. Also, add VLAN 0 by default when allocating a VF since the PF is unaware if the guest OS is running the 8021q module. Without this, MDD events will trigger on untagged traffic because spoofcheck is enabled by default. Due to this change, ignore add/delete messages for VLAN 0 from VIRTCHNL since this is added/deleted during VF initialization/teardown respectively and should not be modified. Signed-off-by: Brett Creeley <brett.creeley@intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-12-12 19:12:54 +08:00
cpu_to_le16(ICE_AQ_VSI_PROP_SW_VALID);
status = ice_update_vsi(&pf->hw, vsi->idx, ctxt, NULL);
if (status) {
netdev_err(vsi->netdev, "%sabling VLAN pruning on VSI handle: %d, VSI HW ID: %d failed, err = %d, aq_err = %d\n",
ena ? "En" : "Dis", vsi->idx, vsi->vsi_num, status,
pf->hw.adminq.sq_last_status);
goto err_out;
}
vsi->info.sw_flags2 = ctxt->info.sw_flags2;
kfree(ctxt);
return 0;
err_out:
kfree(ctxt);
return -EIO;
}
static void ice_vsi_set_tc_cfg(struct ice_vsi *vsi)
{
struct ice_dcbx_cfg *cfg = &vsi->port_info->local_dcbx_cfg;
vsi->tc_cfg.ena_tc = ice_dcb_get_ena_tc(cfg);
vsi->tc_cfg.numtc = ice_dcb_get_num_tc(cfg);
}
/**
* ice_vsi_set_q_vectors_reg_idx - set the HW register index for all q_vectors
* @vsi: VSI to set the q_vectors register index on
*/
static int
ice_vsi_set_q_vectors_reg_idx(struct ice_vsi *vsi)
{
u16 i;
if (!vsi || !vsi->q_vectors)
return -EINVAL;
ice_for_each_q_vector(vsi, i) {
struct ice_q_vector *q_vector = vsi->q_vectors[i];
if (!q_vector) {
dev_err(&vsi->back->pdev->dev,
"Failed to set reg_idx on q_vector %d VSI %d\n",
i, vsi->vsi_num);
goto clear_reg_idx;
}
ice: Refactor interrupt tracking Currently we have two MSI-x (IRQ) trackers, one for OS requested MSI-x entries (sw_irq_tracker) and one for hardware MSI-x vectors (hw_irq_tracker). Generally the sw_irq_tracker has less entries than the hw_irq_tracker because the hw_irq_tracker has entries equal to the max allowed MSI-x per PF and the sw_irq_tracker is mainly the minimum (non SR-IOV portion of the vectors, kernel granted IRQs). All of the non SR-IOV portions of the driver (i.e. LAN queues, RDMA queues, OICR, etc.) take at least one of each type of tracker resource. SR-IOV only grabs entries from the hw_irq_tracker. There are a few issues with this approach that can be seen when doing any kind of device reconfiguration (i.e. ethtool -L, SR-IOV, etc.). One of them being, any time the driver creates an ice_q_vector and associates it to a LAN queue pair it will grab and use one entry from the hw_irq_tracker and one from the sw_irq_tracker. If the indices on these does not match it will cause a Tx timeout, which will cause a reset and then the indices will match up again and traffic will resume. The mismatched indices come from the trackers not being the same size and/or the search_hint in the two trackers not being equal. Another reason for the refactor is the co-existence of features with SR-IOV. If SR-IOV is enabled and the interrupts are taken from the end of the sw_irq_tracker then other features can no longer use this space because the hardware has now given the remaining interrupts to SR-IOV. This patch reworks how we track MSI-x vectors by removing the hw_irq_tracker completely and instead MSI-x resources needed for SR-IOV are determined all at once instead of per VF. This can be done because when creating VFs we know how many are wanted and how many MSI-x vectors each VF needs. This also allows us to start using MSI-x resources from the end of the PF's allowed MSI-x vectors so we are less likely to use entries needed for other features (i.e. RDMA, L2 Offload, etc). This patch also reworks the ice_res_tracker structure by removing the search_hint and adding a new member - "end". Instead of having a search_hint we will always search from 0. The new member, "end", will be used to manipulate the end of the ice_res_tracker (specifically sw_irq_tracker) during runtime based on MSI-x vectors needed by SR-IOV. In the normal case, the end of ice_res_tracker will be equal to the ice_res_tracker's num_entries. The sriov_base_vector member was added to the PF structure. It is used to represent the starting MSI-x index of all the needed MSI-x vectors for all SR-IOV VFs. Depending on how many MSI-x are needed, SR-IOV may have to take resources from the sw_irq_tracker. This is done by setting the sw_irq_tracker->end equal to the pf->sriov_base_vector. When all SR-IOV VFs are removed then the sw_irq_tracker->end is reset back to sw_irq_tracker->num_entries. The sriov_base_vector, along with the VF's number of MSI-x (pf->num_vf_msix), vf_id, and the base MSI-x index on the PF (pf->hw.func_caps.common_cap.msix_vector_first_id), is used to calculate the first HW absolute MSI-x index for each VF, which is used to write to the VPINT_ALLOC[_PCI] and GLINT_VECT2FUNC registers to program the VFs MSI-x PCI configuration bits. Also, the sriov_base_vector is used along with VF's num_vf_msix, vf_id, and q_vector->v_idx to determine the MSI-x register index (used for writing to GLINT_DYN_CTL) within the PF's space. Interrupt changes removed any references to hw_base_vector, hw_oicr_idx, and hw_irq_tracker. Only sw_base_vector, sw_oicr_idx, and sw_irq_tracker variables remain. Change all of these by removing the "sw_" prefix to help avoid confusion with these variables and their use. Signed-off-by: Brett Creeley <brett.creeley@intel.com> Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-04-17 01:30:44 +08:00
if (vsi->type == ICE_VSI_VF) {
struct ice_vf *vf = &vsi->back->vf[vsi->vf_id];
q_vector->reg_idx = ice_calc_vf_reg_idx(vf, q_vector);
} else {
q_vector->reg_idx =
q_vector->v_idx + vsi->base_vector;
}
}
return 0;
clear_reg_idx:
ice_for_each_q_vector(vsi, i) {
struct ice_q_vector *q_vector = vsi->q_vectors[i];
if (q_vector)
q_vector->reg_idx = 0;
}
return -EINVAL;
}
/**
* ice_vsi_add_rem_eth_mac - Program VSI ethertype based filter with rule
* @vsi: the VSI being configured
* @add_rule: boolean value to add or remove ethertype filter rule
*/
static void
ice_vsi_add_rem_eth_mac(struct ice_vsi *vsi, bool add_rule)
{
struct ice_fltr_list_entry *list;
struct ice_pf *pf = vsi->back;
LIST_HEAD(tmp_add_list);
enum ice_status status;
struct device *dev;
dev = ice_pf_to_dev(pf);
list = devm_kzalloc(dev, sizeof(*list), GFP_KERNEL);
if (!list)
return;
list->fltr_info.lkup_type = ICE_SW_LKUP_ETHERTYPE;
list->fltr_info.fltr_act = ICE_DROP_PACKET;
list->fltr_info.flag = ICE_FLTR_TX;
list->fltr_info.src_id = ICE_SRC_ID_VSI;
list->fltr_info.vsi_handle = vsi->idx;
list->fltr_info.l_data.ethertype_mac.ethertype = vsi->ethtype;
INIT_LIST_HEAD(&list->list_entry);
list_add(&list->list_entry, &tmp_add_list);
if (add_rule)
status = ice_add_eth_mac(&pf->hw, &tmp_add_list);
else
status = ice_remove_eth_mac(&pf->hw, &tmp_add_list);
if (status)
dev_err(dev,
"Failure Adding or Removing Ethertype on VSI %i error: %d\n",
vsi->vsi_num, status);
ice_free_fltr_list(dev, &tmp_add_list);
}
/**
* ice_cfg_sw_lldp - Config switch rules for LLDP packet handling
* @vsi: the VSI being configured
* @tx: bool to determine Tx or Rx rule
* @create: bool to determine create or remove Rule
*/
void ice_cfg_sw_lldp(struct ice_vsi *vsi, bool tx, bool create)
{
struct ice_fltr_list_entry *list;
struct ice_pf *pf = vsi->back;
LIST_HEAD(tmp_add_list);
enum ice_status status;
struct device *dev;
dev = ice_pf_to_dev(pf);
list = devm_kzalloc(dev, sizeof(*list), GFP_KERNEL);
if (!list)
return;
list->fltr_info.lkup_type = ICE_SW_LKUP_ETHERTYPE;
list->fltr_info.vsi_handle = vsi->idx;
list->fltr_info.l_data.ethertype_mac.ethertype = ETH_P_LLDP;
if (tx) {
list->fltr_info.fltr_act = ICE_DROP_PACKET;
list->fltr_info.flag = ICE_FLTR_TX;
list->fltr_info.src_id = ICE_SRC_ID_VSI;
} else {
list->fltr_info.fltr_act = ICE_FWD_TO_VSI;
list->fltr_info.flag = ICE_FLTR_RX;
list->fltr_info.src_id = ICE_SRC_ID_LPORT;
}
INIT_LIST_HEAD(&list->list_entry);
list_add(&list->list_entry, &tmp_add_list);
if (create)
status = ice_add_eth_mac(&pf->hw, &tmp_add_list);
else
status = ice_remove_eth_mac(&pf->hw, &tmp_add_list);
if (status)
dev_err(dev, "Fail %s %s LLDP rule on VSI %i error: %d\n",
create ? "adding" : "removing", tx ? "TX" : "RX",
vsi->vsi_num, status);
ice_free_fltr_list(dev, &tmp_add_list);
}
/**
* ice_vsi_setup - Set up a VSI by a given type
* @pf: board private structure
* @pi: pointer to the port_info instance
* @type: VSI type
* @vf_id: defines VF ID to which this VSI connects. This field is meant to be
* used only for ICE_VSI_VF VSI type. For other VSI types, should
* fill-in ICE_INVAL_VFID as input.
*
* This allocates the sw VSI structure and its queue resources.
*
* Returns pointer to the successfully allocated and configured VSI sw struct on
* success, NULL on failure.
*/
struct ice_vsi *
ice_vsi_setup(struct ice_pf *pf, struct ice_port_info *pi,
enum ice_vsi_type type, u16 vf_id)
{
u16 max_txqs[ICE_MAX_TRAFFIC_CLASS] = { 0 };
struct device *dev = ice_pf_to_dev(pf);
enum ice_status status;
struct ice_vsi *vsi;
int ret, i;
if (type == ICE_VSI_VF)
vsi = ice_vsi_alloc(pf, type, vf_id);
else
vsi = ice_vsi_alloc(pf, type, ICE_INVAL_VFID);
if (!vsi) {
dev_err(dev, "could not allocate VSI\n");
return NULL;
}
vsi->port_info = pi;
vsi->vsw = pf->first_sw;
if (vsi->type == ICE_VSI_PF)
vsi->ethtype = ETH_P_PAUSE;
if (vsi->type == ICE_VSI_VF)
vsi->vf_id = vf_id;
if (ice_vsi_get_qs(vsi)) {
dev_err(dev, "Failed to allocate queues. vsi->idx = %d\n",
vsi->idx);
goto unroll_get_qs;
}
/* set RSS capabilities */
ice_vsi_set_rss_params(vsi);
/* set TC configuration */
ice_vsi_set_tc_cfg(vsi);
/* create the VSI */
ret = ice_vsi_init(vsi, true);
if (ret)
goto unroll_get_qs;
switch (vsi->type) {
case ICE_VSI_PF:
ret = ice_vsi_alloc_q_vectors(vsi);
if (ret)
goto unroll_vsi_init;
ret = ice_vsi_setup_vector_base(vsi);
if (ret)
goto unroll_alloc_q_vector;
ret = ice_vsi_set_q_vectors_reg_idx(vsi);
if (ret)
goto unroll_vector_base;
ret = ice_vsi_alloc_rings(vsi);
if (ret)
goto unroll_vector_base;
ice_vsi_map_rings_to_vectors(vsi);
/* Do not exit if configuring RSS had an issue, at least
* receive traffic on first queue. Hence no need to capture
* return value
*/
if (test_bit(ICE_FLAG_RSS_ENA, pf->flags)) {
ice_vsi_cfg_rss_lut_key(vsi);
ice_vsi_set_rss_flow_fld(vsi);
}
break;
case ICE_VSI_VF:
/* VF driver will take care of creating netdev for this type and
* map queues to vectors through Virtchnl, PF driver only
* creates a VSI and corresponding structures for bookkeeping
* purpose
*/
ret = ice_vsi_alloc_q_vectors(vsi);
if (ret)
goto unroll_vsi_init;
ret = ice_vsi_alloc_rings(vsi);
if (ret)
goto unroll_alloc_q_vector;
ret = ice_vsi_set_q_vectors_reg_idx(vsi);
if (ret)
goto unroll_vector_base;
/* Do not exit if configuring RSS had an issue, at least
* receive traffic on first queue. Hence no need to capture
* return value
*/
if (test_bit(ICE_FLAG_RSS_ENA, pf->flags))
ice_vsi_cfg_rss_lut_key(vsi);
break;
case ICE_VSI_LB:
ret = ice_vsi_alloc_rings(vsi);
if (ret)
goto unroll_vsi_init;
break;
default:
/* clean up the resources and exit */
goto unroll_vsi_init;
}
/* configure VSI nodes based on number of queues and TC's */
for (i = 0; i < vsi->tc_cfg.numtc; i++)
max_txqs[i] = vsi->alloc_txq;
status = ice_cfg_vsi_lan(vsi->port_info, vsi->idx, vsi->tc_cfg.ena_tc,
max_txqs);
if (status) {
dev_err(dev, "VSI %d failed lan queue config, error %d\n",
vsi->vsi_num, status);
goto unroll_vector_base;
}
/* Add switch rule to drop all Tx Flow Control Frames, of look up
* type ETHERTYPE from VSIs, and restrict malicious VF from sending
* out PAUSE or PFC frames. If enabled, FW can still send FC frames.
* The rule is added once for PF VSI in order to create appropriate
* recipe, since VSI/VSI list is ignored with drop action...
* Also add rules to handle LLDP Tx packets. Tx LLDP packets need to
* be dropped so that VFs cannot send LLDP packets to reconfig DCB
* settings in the HW.
*/
if (!ice_is_safe_mode(pf))
if (vsi->type == ICE_VSI_PF) {
ice_vsi_add_rem_eth_mac(vsi, true);
/* Tx LLDP packets */
ice_cfg_sw_lldp(vsi, true, true);
}
return vsi;
unroll_vector_base:
/* reclaim SW interrupts back to the common pool */
ice: Refactor interrupt tracking Currently we have two MSI-x (IRQ) trackers, one for OS requested MSI-x entries (sw_irq_tracker) and one for hardware MSI-x vectors (hw_irq_tracker). Generally the sw_irq_tracker has less entries than the hw_irq_tracker because the hw_irq_tracker has entries equal to the max allowed MSI-x per PF and the sw_irq_tracker is mainly the minimum (non SR-IOV portion of the vectors, kernel granted IRQs). All of the non SR-IOV portions of the driver (i.e. LAN queues, RDMA queues, OICR, etc.) take at least one of each type of tracker resource. SR-IOV only grabs entries from the hw_irq_tracker. There are a few issues with this approach that can be seen when doing any kind of device reconfiguration (i.e. ethtool -L, SR-IOV, etc.). One of them being, any time the driver creates an ice_q_vector and associates it to a LAN queue pair it will grab and use one entry from the hw_irq_tracker and one from the sw_irq_tracker. If the indices on these does not match it will cause a Tx timeout, which will cause a reset and then the indices will match up again and traffic will resume. The mismatched indices come from the trackers not being the same size and/or the search_hint in the two trackers not being equal. Another reason for the refactor is the co-existence of features with SR-IOV. If SR-IOV is enabled and the interrupts are taken from the end of the sw_irq_tracker then other features can no longer use this space because the hardware has now given the remaining interrupts to SR-IOV. This patch reworks how we track MSI-x vectors by removing the hw_irq_tracker completely and instead MSI-x resources needed for SR-IOV are determined all at once instead of per VF. This can be done because when creating VFs we know how many are wanted and how many MSI-x vectors each VF needs. This also allows us to start using MSI-x resources from the end of the PF's allowed MSI-x vectors so we are less likely to use entries needed for other features (i.e. RDMA, L2 Offload, etc). This patch also reworks the ice_res_tracker structure by removing the search_hint and adding a new member - "end". Instead of having a search_hint we will always search from 0. The new member, "end", will be used to manipulate the end of the ice_res_tracker (specifically sw_irq_tracker) during runtime based on MSI-x vectors needed by SR-IOV. In the normal case, the end of ice_res_tracker will be equal to the ice_res_tracker's num_entries. The sriov_base_vector member was added to the PF structure. It is used to represent the starting MSI-x index of all the needed MSI-x vectors for all SR-IOV VFs. Depending on how many MSI-x are needed, SR-IOV may have to take resources from the sw_irq_tracker. This is done by setting the sw_irq_tracker->end equal to the pf->sriov_base_vector. When all SR-IOV VFs are removed then the sw_irq_tracker->end is reset back to sw_irq_tracker->num_entries. The sriov_base_vector, along with the VF's number of MSI-x (pf->num_vf_msix), vf_id, and the base MSI-x index on the PF (pf->hw.func_caps.common_cap.msix_vector_first_id), is used to calculate the first HW absolute MSI-x index for each VF, which is used to write to the VPINT_ALLOC[_PCI] and GLINT_VECT2FUNC registers to program the VFs MSI-x PCI configuration bits. Also, the sriov_base_vector is used along with VF's num_vf_msix, vf_id, and q_vector->v_idx to determine the MSI-x register index (used for writing to GLINT_DYN_CTL) within the PF's space. Interrupt changes removed any references to hw_base_vector, hw_oicr_idx, and hw_irq_tracker. Only sw_base_vector, sw_oicr_idx, and sw_irq_tracker variables remain. Change all of these by removing the "sw_" prefix to help avoid confusion with these variables and their use. Signed-off-by: Brett Creeley <brett.creeley@intel.com> Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-04-17 01:30:44 +08:00
ice_free_res(pf->irq_tracker, vsi->base_vector, vsi->idx);
pf->num_avail_sw_msix += vsi->num_q_vectors;
unroll_alloc_q_vector:
ice_vsi_free_q_vectors(vsi);
unroll_vsi_init:
ice_vsi_delete(vsi);
unroll_get_qs:
ice_vsi_put_qs(vsi);
ice_vsi_clear(vsi);
return NULL;
}
/**
* ice_vsi_release_msix - Clear the queue to Interrupt mapping in HW
* @vsi: the VSI being cleaned up
*/
static void ice_vsi_release_msix(struct ice_vsi *vsi)
{
struct ice_pf *pf = vsi->back;
struct ice_hw *hw = &pf->hw;
u32 txq = 0;
u32 rxq = 0;
int i, q;
ice: Refactor interrupt tracking Currently we have two MSI-x (IRQ) trackers, one for OS requested MSI-x entries (sw_irq_tracker) and one for hardware MSI-x vectors (hw_irq_tracker). Generally the sw_irq_tracker has less entries than the hw_irq_tracker because the hw_irq_tracker has entries equal to the max allowed MSI-x per PF and the sw_irq_tracker is mainly the minimum (non SR-IOV portion of the vectors, kernel granted IRQs). All of the non SR-IOV portions of the driver (i.e. LAN queues, RDMA queues, OICR, etc.) take at least one of each type of tracker resource. SR-IOV only grabs entries from the hw_irq_tracker. There are a few issues with this approach that can be seen when doing any kind of device reconfiguration (i.e. ethtool -L, SR-IOV, etc.). One of them being, any time the driver creates an ice_q_vector and associates it to a LAN queue pair it will grab and use one entry from the hw_irq_tracker and one from the sw_irq_tracker. If the indices on these does not match it will cause a Tx timeout, which will cause a reset and then the indices will match up again and traffic will resume. The mismatched indices come from the trackers not being the same size and/or the search_hint in the two trackers not being equal. Another reason for the refactor is the co-existence of features with SR-IOV. If SR-IOV is enabled and the interrupts are taken from the end of the sw_irq_tracker then other features can no longer use this space because the hardware has now given the remaining interrupts to SR-IOV. This patch reworks how we track MSI-x vectors by removing the hw_irq_tracker completely and instead MSI-x resources needed for SR-IOV are determined all at once instead of per VF. This can be done because when creating VFs we know how many are wanted and how many MSI-x vectors each VF needs. This also allows us to start using MSI-x resources from the end of the PF's allowed MSI-x vectors so we are less likely to use entries needed for other features (i.e. RDMA, L2 Offload, etc). This patch also reworks the ice_res_tracker structure by removing the search_hint and adding a new member - "end". Instead of having a search_hint we will always search from 0. The new member, "end", will be used to manipulate the end of the ice_res_tracker (specifically sw_irq_tracker) during runtime based on MSI-x vectors needed by SR-IOV. In the normal case, the end of ice_res_tracker will be equal to the ice_res_tracker's num_entries. The sriov_base_vector member was added to the PF structure. It is used to represent the starting MSI-x index of all the needed MSI-x vectors for all SR-IOV VFs. Depending on how many MSI-x are needed, SR-IOV may have to take resources from the sw_irq_tracker. This is done by setting the sw_irq_tracker->end equal to the pf->sriov_base_vector. When all SR-IOV VFs are removed then the sw_irq_tracker->end is reset back to sw_irq_tracker->num_entries. The sriov_base_vector, along with the VF's number of MSI-x (pf->num_vf_msix), vf_id, and the base MSI-x index on the PF (pf->hw.func_caps.common_cap.msix_vector_first_id), is used to calculate the first HW absolute MSI-x index for each VF, which is used to write to the VPINT_ALLOC[_PCI] and GLINT_VECT2FUNC registers to program the VFs MSI-x PCI configuration bits. Also, the sriov_base_vector is used along with VF's num_vf_msix, vf_id, and q_vector->v_idx to determine the MSI-x register index (used for writing to GLINT_DYN_CTL) within the PF's space. Interrupt changes removed any references to hw_base_vector, hw_oicr_idx, and hw_irq_tracker. Only sw_base_vector, sw_oicr_idx, and sw_irq_tracker variables remain. Change all of these by removing the "sw_" prefix to help avoid confusion with these variables and their use. Signed-off-by: Brett Creeley <brett.creeley@intel.com> Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-04-17 01:30:44 +08:00
for (i = 0; i < vsi->num_q_vectors; i++) {
struct ice_q_vector *q_vector = vsi->q_vectors[i];
ice: Refactor interrupt tracking Currently we have two MSI-x (IRQ) trackers, one for OS requested MSI-x entries (sw_irq_tracker) and one for hardware MSI-x vectors (hw_irq_tracker). Generally the sw_irq_tracker has less entries than the hw_irq_tracker because the hw_irq_tracker has entries equal to the max allowed MSI-x per PF and the sw_irq_tracker is mainly the minimum (non SR-IOV portion of the vectors, kernel granted IRQs). All of the non SR-IOV portions of the driver (i.e. LAN queues, RDMA queues, OICR, etc.) take at least one of each type of tracker resource. SR-IOV only grabs entries from the hw_irq_tracker. There are a few issues with this approach that can be seen when doing any kind of device reconfiguration (i.e. ethtool -L, SR-IOV, etc.). One of them being, any time the driver creates an ice_q_vector and associates it to a LAN queue pair it will grab and use one entry from the hw_irq_tracker and one from the sw_irq_tracker. If the indices on these does not match it will cause a Tx timeout, which will cause a reset and then the indices will match up again and traffic will resume. The mismatched indices come from the trackers not being the same size and/or the search_hint in the two trackers not being equal. Another reason for the refactor is the co-existence of features with SR-IOV. If SR-IOV is enabled and the interrupts are taken from the end of the sw_irq_tracker then other features can no longer use this space because the hardware has now given the remaining interrupts to SR-IOV. This patch reworks how we track MSI-x vectors by removing the hw_irq_tracker completely and instead MSI-x resources needed for SR-IOV are determined all at once instead of per VF. This can be done because when creating VFs we know how many are wanted and how many MSI-x vectors each VF needs. This also allows us to start using MSI-x resources from the end of the PF's allowed MSI-x vectors so we are less likely to use entries needed for other features (i.e. RDMA, L2 Offload, etc). This patch also reworks the ice_res_tracker structure by removing the search_hint and adding a new member - "end". Instead of having a search_hint we will always search from 0. The new member, "end", will be used to manipulate the end of the ice_res_tracker (specifically sw_irq_tracker) during runtime based on MSI-x vectors needed by SR-IOV. In the normal case, the end of ice_res_tracker will be equal to the ice_res_tracker's num_entries. The sriov_base_vector member was added to the PF structure. It is used to represent the starting MSI-x index of all the needed MSI-x vectors for all SR-IOV VFs. Depending on how many MSI-x are needed, SR-IOV may have to take resources from the sw_irq_tracker. This is done by setting the sw_irq_tracker->end equal to the pf->sriov_base_vector. When all SR-IOV VFs are removed then the sw_irq_tracker->end is reset back to sw_irq_tracker->num_entries. The sriov_base_vector, along with the VF's number of MSI-x (pf->num_vf_msix), vf_id, and the base MSI-x index on the PF (pf->hw.func_caps.common_cap.msix_vector_first_id), is used to calculate the first HW absolute MSI-x index for each VF, which is used to write to the VPINT_ALLOC[_PCI] and GLINT_VECT2FUNC registers to program the VFs MSI-x PCI configuration bits. Also, the sriov_base_vector is used along with VF's num_vf_msix, vf_id, and q_vector->v_idx to determine the MSI-x register index (used for writing to GLINT_DYN_CTL) within the PF's space. Interrupt changes removed any references to hw_base_vector, hw_oicr_idx, and hw_irq_tracker. Only sw_base_vector, sw_oicr_idx, and sw_irq_tracker variables remain. Change all of these by removing the "sw_" prefix to help avoid confusion with these variables and their use. Signed-off-by: Brett Creeley <brett.creeley@intel.com> Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-04-17 01:30:44 +08:00
u16 reg_idx = q_vector->reg_idx;
ice: Refactor interrupt tracking Currently we have two MSI-x (IRQ) trackers, one for OS requested MSI-x entries (sw_irq_tracker) and one for hardware MSI-x vectors (hw_irq_tracker). Generally the sw_irq_tracker has less entries than the hw_irq_tracker because the hw_irq_tracker has entries equal to the max allowed MSI-x per PF and the sw_irq_tracker is mainly the minimum (non SR-IOV portion of the vectors, kernel granted IRQs). All of the non SR-IOV portions of the driver (i.e. LAN queues, RDMA queues, OICR, etc.) take at least one of each type of tracker resource. SR-IOV only grabs entries from the hw_irq_tracker. There are a few issues with this approach that can be seen when doing any kind of device reconfiguration (i.e. ethtool -L, SR-IOV, etc.). One of them being, any time the driver creates an ice_q_vector and associates it to a LAN queue pair it will grab and use one entry from the hw_irq_tracker and one from the sw_irq_tracker. If the indices on these does not match it will cause a Tx timeout, which will cause a reset and then the indices will match up again and traffic will resume. The mismatched indices come from the trackers not being the same size and/or the search_hint in the two trackers not being equal. Another reason for the refactor is the co-existence of features with SR-IOV. If SR-IOV is enabled and the interrupts are taken from the end of the sw_irq_tracker then other features can no longer use this space because the hardware has now given the remaining interrupts to SR-IOV. This patch reworks how we track MSI-x vectors by removing the hw_irq_tracker completely and instead MSI-x resources needed for SR-IOV are determined all at once instead of per VF. This can be done because when creating VFs we know how many are wanted and how many MSI-x vectors each VF needs. This also allows us to start using MSI-x resources from the end of the PF's allowed MSI-x vectors so we are less likely to use entries needed for other features (i.e. RDMA, L2 Offload, etc). This patch also reworks the ice_res_tracker structure by removing the search_hint and adding a new member - "end". Instead of having a search_hint we will always search from 0. The new member, "end", will be used to manipulate the end of the ice_res_tracker (specifically sw_irq_tracker) during runtime based on MSI-x vectors needed by SR-IOV. In the normal case, the end of ice_res_tracker will be equal to the ice_res_tracker's num_entries. The sriov_base_vector member was added to the PF structure. It is used to represent the starting MSI-x index of all the needed MSI-x vectors for all SR-IOV VFs. Depending on how many MSI-x are needed, SR-IOV may have to take resources from the sw_irq_tracker. This is done by setting the sw_irq_tracker->end equal to the pf->sriov_base_vector. When all SR-IOV VFs are removed then the sw_irq_tracker->end is reset back to sw_irq_tracker->num_entries. The sriov_base_vector, along with the VF's number of MSI-x (pf->num_vf_msix), vf_id, and the base MSI-x index on the PF (pf->hw.func_caps.common_cap.msix_vector_first_id), is used to calculate the first HW absolute MSI-x index for each VF, which is used to write to the VPINT_ALLOC[_PCI] and GLINT_VECT2FUNC registers to program the VFs MSI-x PCI configuration bits. Also, the sriov_base_vector is used along with VF's num_vf_msix, vf_id, and q_vector->v_idx to determine the MSI-x register index (used for writing to GLINT_DYN_CTL) within the PF's space. Interrupt changes removed any references to hw_base_vector, hw_oicr_idx, and hw_irq_tracker. Only sw_base_vector, sw_oicr_idx, and sw_irq_tracker variables remain. Change all of these by removing the "sw_" prefix to help avoid confusion with these variables and their use. Signed-off-by: Brett Creeley <brett.creeley@intel.com> Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-04-17 01:30:44 +08:00
wr32(hw, GLINT_ITR(ICE_IDX_ITR0, reg_idx), 0);
wr32(hw, GLINT_ITR(ICE_IDX_ITR1, reg_idx), 0);
for (q = 0; q < q_vector->num_ring_tx; q++) {
wr32(hw, QINT_TQCTL(vsi->txq_map[txq]), 0);
if (ice_is_xdp_ena_vsi(vsi)) {
u32 xdp_txq = txq + vsi->num_xdp_txq;
wr32(hw, QINT_TQCTL(vsi->txq_map[xdp_txq]), 0);
}
txq++;
}
for (q = 0; q < q_vector->num_ring_rx; q++) {
wr32(hw, QINT_RQCTL(vsi->rxq_map[rxq]), 0);
rxq++;
}
}
ice_flush(hw);
}
/**
* ice_vsi_free_irq - Free the IRQ association with the OS
* @vsi: the VSI being configured
*/
void ice_vsi_free_irq(struct ice_vsi *vsi)
{
struct ice_pf *pf = vsi->back;
ice: Refactor interrupt tracking Currently we have two MSI-x (IRQ) trackers, one for OS requested MSI-x entries (sw_irq_tracker) and one for hardware MSI-x vectors (hw_irq_tracker). Generally the sw_irq_tracker has less entries than the hw_irq_tracker because the hw_irq_tracker has entries equal to the max allowed MSI-x per PF and the sw_irq_tracker is mainly the minimum (non SR-IOV portion of the vectors, kernel granted IRQs). All of the non SR-IOV portions of the driver (i.e. LAN queues, RDMA queues, OICR, etc.) take at least one of each type of tracker resource. SR-IOV only grabs entries from the hw_irq_tracker. There are a few issues with this approach that can be seen when doing any kind of device reconfiguration (i.e. ethtool -L, SR-IOV, etc.). One of them being, any time the driver creates an ice_q_vector and associates it to a LAN queue pair it will grab and use one entry from the hw_irq_tracker and one from the sw_irq_tracker. If the indices on these does not match it will cause a Tx timeout, which will cause a reset and then the indices will match up again and traffic will resume. The mismatched indices come from the trackers not being the same size and/or the search_hint in the two trackers not being equal. Another reason for the refactor is the co-existence of features with SR-IOV. If SR-IOV is enabled and the interrupts are taken from the end of the sw_irq_tracker then other features can no longer use this space because the hardware has now given the remaining interrupts to SR-IOV. This patch reworks how we track MSI-x vectors by removing the hw_irq_tracker completely and instead MSI-x resources needed for SR-IOV are determined all at once instead of per VF. This can be done because when creating VFs we know how many are wanted and how many MSI-x vectors each VF needs. This also allows us to start using MSI-x resources from the end of the PF's allowed MSI-x vectors so we are less likely to use entries needed for other features (i.e. RDMA, L2 Offload, etc). This patch also reworks the ice_res_tracker structure by removing the search_hint and adding a new member - "end". Instead of having a search_hint we will always search from 0. The new member, "end", will be used to manipulate the end of the ice_res_tracker (specifically sw_irq_tracker) during runtime based on MSI-x vectors needed by SR-IOV. In the normal case, the end of ice_res_tracker will be equal to the ice_res_tracker's num_entries. The sriov_base_vector member was added to the PF structure. It is used to represent the starting MSI-x index of all the needed MSI-x vectors for all SR-IOV VFs. Depending on how many MSI-x are needed, SR-IOV may have to take resources from the sw_irq_tracker. This is done by setting the sw_irq_tracker->end equal to the pf->sriov_base_vector. When all SR-IOV VFs are removed then the sw_irq_tracker->end is reset back to sw_irq_tracker->num_entries. The sriov_base_vector, along with the VF's number of MSI-x (pf->num_vf_msix), vf_id, and the base MSI-x index on the PF (pf->hw.func_caps.common_cap.msix_vector_first_id), is used to calculate the first HW absolute MSI-x index for each VF, which is used to write to the VPINT_ALLOC[_PCI] and GLINT_VECT2FUNC registers to program the VFs MSI-x PCI configuration bits. Also, the sriov_base_vector is used along with VF's num_vf_msix, vf_id, and q_vector->v_idx to determine the MSI-x register index (used for writing to GLINT_DYN_CTL) within the PF's space. Interrupt changes removed any references to hw_base_vector, hw_oicr_idx, and hw_irq_tracker. Only sw_base_vector, sw_oicr_idx, and sw_irq_tracker variables remain. Change all of these by removing the "sw_" prefix to help avoid confusion with these variables and their use. Signed-off-by: Brett Creeley <brett.creeley@intel.com> Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-04-17 01:30:44 +08:00
int base = vsi->base_vector;
int i;
if (!vsi->q_vectors || !vsi->irqs_ready)
return;
ice_vsi_release_msix(vsi);
if (vsi->type == ICE_VSI_VF)
return;
vsi->irqs_ready = false;
ice_for_each_q_vector(vsi, i) {
u16 vector = i + base;
int irq_num;
irq_num = pf->msix_entries[vector].vector;
/* free only the irqs that were actually requested */
if (!vsi->q_vectors[i] ||
!(vsi->q_vectors[i]->num_ring_tx ||
vsi->q_vectors[i]->num_ring_rx))
continue;
/* clear the affinity notifier in the IRQ descriptor */
irq_set_affinity_notifier(irq_num, NULL);
/* clear the affinity_mask in the IRQ descriptor */
irq_set_affinity_hint(irq_num, NULL);
synchronize_irq(irq_num);
devm_free_irq(ice_pf_to_dev(pf), irq_num, vsi->q_vectors[i]);
}
}
/**
* ice_vsi_free_tx_rings - Free Tx resources for VSI queues
* @vsi: the VSI having resources freed
*/
void ice_vsi_free_tx_rings(struct ice_vsi *vsi)
{
int i;
if (!vsi->tx_rings)
return;
ice_for_each_txq(vsi, i)
if (vsi->tx_rings[i] && vsi->tx_rings[i]->desc)
ice_free_tx_ring(vsi->tx_rings[i]);
}
/**
* ice_vsi_free_rx_rings - Free Rx resources for VSI queues
* @vsi: the VSI having resources freed
*/
void ice_vsi_free_rx_rings(struct ice_vsi *vsi)
{
int i;
if (!vsi->rx_rings)
return;
ice_for_each_rxq(vsi, i)
if (vsi->rx_rings[i] && vsi->rx_rings[i]->desc)
ice_free_rx_ring(vsi->rx_rings[i]);
}
/**
* ice_vsi_close - Shut down a VSI
* @vsi: the VSI being shut down
*/
void ice_vsi_close(struct ice_vsi *vsi)
{
if (!test_and_set_bit(__ICE_DOWN, vsi->state))
ice_down(vsi);
ice_vsi_free_irq(vsi);
ice_vsi_free_tx_rings(vsi);
ice_vsi_free_rx_rings(vsi);
}
/**
* ice_ena_vsi - resume a VSI
* @vsi: the VSI being resume
* @locked: is the rtnl_lock already held
*/
int ice_ena_vsi(struct ice_vsi *vsi, bool locked)
{
int err = 0;
if (!test_bit(__ICE_NEEDS_RESTART, vsi->state))
return 0;
clear_bit(__ICE_NEEDS_RESTART, vsi->state);
if (vsi->netdev && vsi->type == ICE_VSI_PF) {
if (netif_running(vsi->netdev)) {
if (!locked)
rtnl_lock();
err = ice_open(vsi->netdev);
if (!locked)
rtnl_unlock();
}
}
return err;
}
/**
* ice_dis_vsi - pause a VSI
* @vsi: the VSI being paused
* @locked: is the rtnl_lock already held
*/
void ice_dis_vsi(struct ice_vsi *vsi, bool locked)
{
if (test_bit(__ICE_DOWN, vsi->state))
return;
set_bit(__ICE_NEEDS_RESTART, vsi->state);
if (vsi->type == ICE_VSI_PF && vsi->netdev) {
if (netif_running(vsi->netdev)) {
if (!locked)
rtnl_lock();
ice_stop(vsi->netdev);
if (!locked)
rtnl_unlock();
} else {
ice_vsi_close(vsi);
}
}
}
/**
* ice_free_res - free a block of resources
* @res: pointer to the resource
* @index: starting index previously returned by ice_get_res
* @id: identifier to track owner
*
* Returns number of resources freed
*/
int ice_free_res(struct ice_res_tracker *res, u16 index, u16 id)
{
int count = 0;
int i;
ice: Refactor interrupt tracking Currently we have two MSI-x (IRQ) trackers, one for OS requested MSI-x entries (sw_irq_tracker) and one for hardware MSI-x vectors (hw_irq_tracker). Generally the sw_irq_tracker has less entries than the hw_irq_tracker because the hw_irq_tracker has entries equal to the max allowed MSI-x per PF and the sw_irq_tracker is mainly the minimum (non SR-IOV portion of the vectors, kernel granted IRQs). All of the non SR-IOV portions of the driver (i.e. LAN queues, RDMA queues, OICR, etc.) take at least one of each type of tracker resource. SR-IOV only grabs entries from the hw_irq_tracker. There are a few issues with this approach that can be seen when doing any kind of device reconfiguration (i.e. ethtool -L, SR-IOV, etc.). One of them being, any time the driver creates an ice_q_vector and associates it to a LAN queue pair it will grab and use one entry from the hw_irq_tracker and one from the sw_irq_tracker. If the indices on these does not match it will cause a Tx timeout, which will cause a reset and then the indices will match up again and traffic will resume. The mismatched indices come from the trackers not being the same size and/or the search_hint in the two trackers not being equal. Another reason for the refactor is the co-existence of features with SR-IOV. If SR-IOV is enabled and the interrupts are taken from the end of the sw_irq_tracker then other features can no longer use this space because the hardware has now given the remaining interrupts to SR-IOV. This patch reworks how we track MSI-x vectors by removing the hw_irq_tracker completely and instead MSI-x resources needed for SR-IOV are determined all at once instead of per VF. This can be done because when creating VFs we know how many are wanted and how many MSI-x vectors each VF needs. This also allows us to start using MSI-x resources from the end of the PF's allowed MSI-x vectors so we are less likely to use entries needed for other features (i.e. RDMA, L2 Offload, etc). This patch also reworks the ice_res_tracker structure by removing the search_hint and adding a new member - "end". Instead of having a search_hint we will always search from 0. The new member, "end", will be used to manipulate the end of the ice_res_tracker (specifically sw_irq_tracker) during runtime based on MSI-x vectors needed by SR-IOV. In the normal case, the end of ice_res_tracker will be equal to the ice_res_tracker's num_entries. The sriov_base_vector member was added to the PF structure. It is used to represent the starting MSI-x index of all the needed MSI-x vectors for all SR-IOV VFs. Depending on how many MSI-x are needed, SR-IOV may have to take resources from the sw_irq_tracker. This is done by setting the sw_irq_tracker->end equal to the pf->sriov_base_vector. When all SR-IOV VFs are removed then the sw_irq_tracker->end is reset back to sw_irq_tracker->num_entries. The sriov_base_vector, along with the VF's number of MSI-x (pf->num_vf_msix), vf_id, and the base MSI-x index on the PF (pf->hw.func_caps.common_cap.msix_vector_first_id), is used to calculate the first HW absolute MSI-x index for each VF, which is used to write to the VPINT_ALLOC[_PCI] and GLINT_VECT2FUNC registers to program the VFs MSI-x PCI configuration bits. Also, the sriov_base_vector is used along with VF's num_vf_msix, vf_id, and q_vector->v_idx to determine the MSI-x register index (used for writing to GLINT_DYN_CTL) within the PF's space. Interrupt changes removed any references to hw_base_vector, hw_oicr_idx, and hw_irq_tracker. Only sw_base_vector, sw_oicr_idx, and sw_irq_tracker variables remain. Change all of these by removing the "sw_" prefix to help avoid confusion with these variables and their use. Signed-off-by: Brett Creeley <brett.creeley@intel.com> Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-04-17 01:30:44 +08:00
if (!res || index >= res->end)
return -EINVAL;
id |= ICE_RES_VALID_BIT;
ice: Refactor interrupt tracking Currently we have two MSI-x (IRQ) trackers, one for OS requested MSI-x entries (sw_irq_tracker) and one for hardware MSI-x vectors (hw_irq_tracker). Generally the sw_irq_tracker has less entries than the hw_irq_tracker because the hw_irq_tracker has entries equal to the max allowed MSI-x per PF and the sw_irq_tracker is mainly the minimum (non SR-IOV portion of the vectors, kernel granted IRQs). All of the non SR-IOV portions of the driver (i.e. LAN queues, RDMA queues, OICR, etc.) take at least one of each type of tracker resource. SR-IOV only grabs entries from the hw_irq_tracker. There are a few issues with this approach that can be seen when doing any kind of device reconfiguration (i.e. ethtool -L, SR-IOV, etc.). One of them being, any time the driver creates an ice_q_vector and associates it to a LAN queue pair it will grab and use one entry from the hw_irq_tracker and one from the sw_irq_tracker. If the indices on these does not match it will cause a Tx timeout, which will cause a reset and then the indices will match up again and traffic will resume. The mismatched indices come from the trackers not being the same size and/or the search_hint in the two trackers not being equal. Another reason for the refactor is the co-existence of features with SR-IOV. If SR-IOV is enabled and the interrupts are taken from the end of the sw_irq_tracker then other features can no longer use this space because the hardware has now given the remaining interrupts to SR-IOV. This patch reworks how we track MSI-x vectors by removing the hw_irq_tracker completely and instead MSI-x resources needed for SR-IOV are determined all at once instead of per VF. This can be done because when creating VFs we know how many are wanted and how many MSI-x vectors each VF needs. This also allows us to start using MSI-x resources from the end of the PF's allowed MSI-x vectors so we are less likely to use entries needed for other features (i.e. RDMA, L2 Offload, etc). This patch also reworks the ice_res_tracker structure by removing the search_hint and adding a new member - "end". Instead of having a search_hint we will always search from 0. The new member, "end", will be used to manipulate the end of the ice_res_tracker (specifically sw_irq_tracker) during runtime based on MSI-x vectors needed by SR-IOV. In the normal case, the end of ice_res_tracker will be equal to the ice_res_tracker's num_entries. The sriov_base_vector member was added to the PF structure. It is used to represent the starting MSI-x index of all the needed MSI-x vectors for all SR-IOV VFs. Depending on how many MSI-x are needed, SR-IOV may have to take resources from the sw_irq_tracker. This is done by setting the sw_irq_tracker->end equal to the pf->sriov_base_vector. When all SR-IOV VFs are removed then the sw_irq_tracker->end is reset back to sw_irq_tracker->num_entries. The sriov_base_vector, along with the VF's number of MSI-x (pf->num_vf_msix), vf_id, and the base MSI-x index on the PF (pf->hw.func_caps.common_cap.msix_vector_first_id), is used to calculate the first HW absolute MSI-x index for each VF, which is used to write to the VPINT_ALLOC[_PCI] and GLINT_VECT2FUNC registers to program the VFs MSI-x PCI configuration bits. Also, the sriov_base_vector is used along with VF's num_vf_msix, vf_id, and q_vector->v_idx to determine the MSI-x register index (used for writing to GLINT_DYN_CTL) within the PF's space. Interrupt changes removed any references to hw_base_vector, hw_oicr_idx, and hw_irq_tracker. Only sw_base_vector, sw_oicr_idx, and sw_irq_tracker variables remain. Change all of these by removing the "sw_" prefix to help avoid confusion with these variables and their use. Signed-off-by: Brett Creeley <brett.creeley@intel.com> Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-04-17 01:30:44 +08:00
for (i = index; i < res->end && res->list[i] == id; i++) {
res->list[i] = 0;
count++;
}
return count;
}
/**
* ice_search_res - Search the tracker for a block of resources
* @res: pointer to the resource
* @needed: size of the block needed
* @id: identifier to track owner
*
* Returns the base item index of the block, or -ENOMEM for error
*/
static int ice_search_res(struct ice_res_tracker *res, u16 needed, u16 id)
{
ice: Refactor interrupt tracking Currently we have two MSI-x (IRQ) trackers, one for OS requested MSI-x entries (sw_irq_tracker) and one for hardware MSI-x vectors (hw_irq_tracker). Generally the sw_irq_tracker has less entries than the hw_irq_tracker because the hw_irq_tracker has entries equal to the max allowed MSI-x per PF and the sw_irq_tracker is mainly the minimum (non SR-IOV portion of the vectors, kernel granted IRQs). All of the non SR-IOV portions of the driver (i.e. LAN queues, RDMA queues, OICR, etc.) take at least one of each type of tracker resource. SR-IOV only grabs entries from the hw_irq_tracker. There are a few issues with this approach that can be seen when doing any kind of device reconfiguration (i.e. ethtool -L, SR-IOV, etc.). One of them being, any time the driver creates an ice_q_vector and associates it to a LAN queue pair it will grab and use one entry from the hw_irq_tracker and one from the sw_irq_tracker. If the indices on these does not match it will cause a Tx timeout, which will cause a reset and then the indices will match up again and traffic will resume. The mismatched indices come from the trackers not being the same size and/or the search_hint in the two trackers not being equal. Another reason for the refactor is the co-existence of features with SR-IOV. If SR-IOV is enabled and the interrupts are taken from the end of the sw_irq_tracker then other features can no longer use this space because the hardware has now given the remaining interrupts to SR-IOV. This patch reworks how we track MSI-x vectors by removing the hw_irq_tracker completely and instead MSI-x resources needed for SR-IOV are determined all at once instead of per VF. This can be done because when creating VFs we know how many are wanted and how many MSI-x vectors each VF needs. This also allows us to start using MSI-x resources from the end of the PF's allowed MSI-x vectors so we are less likely to use entries needed for other features (i.e. RDMA, L2 Offload, etc). This patch also reworks the ice_res_tracker structure by removing the search_hint and adding a new member - "end". Instead of having a search_hint we will always search from 0. The new member, "end", will be used to manipulate the end of the ice_res_tracker (specifically sw_irq_tracker) during runtime based on MSI-x vectors needed by SR-IOV. In the normal case, the end of ice_res_tracker will be equal to the ice_res_tracker's num_entries. The sriov_base_vector member was added to the PF structure. It is used to represent the starting MSI-x index of all the needed MSI-x vectors for all SR-IOV VFs. Depending on how many MSI-x are needed, SR-IOV may have to take resources from the sw_irq_tracker. This is done by setting the sw_irq_tracker->end equal to the pf->sriov_base_vector. When all SR-IOV VFs are removed then the sw_irq_tracker->end is reset back to sw_irq_tracker->num_entries. The sriov_base_vector, along with the VF's number of MSI-x (pf->num_vf_msix), vf_id, and the base MSI-x index on the PF (pf->hw.func_caps.common_cap.msix_vector_first_id), is used to calculate the first HW absolute MSI-x index for each VF, which is used to write to the VPINT_ALLOC[_PCI] and GLINT_VECT2FUNC registers to program the VFs MSI-x PCI configuration bits. Also, the sriov_base_vector is used along with VF's num_vf_msix, vf_id, and q_vector->v_idx to determine the MSI-x register index (used for writing to GLINT_DYN_CTL) within the PF's space. Interrupt changes removed any references to hw_base_vector, hw_oicr_idx, and hw_irq_tracker. Only sw_base_vector, sw_oicr_idx, and sw_irq_tracker variables remain. Change all of these by removing the "sw_" prefix to help avoid confusion with these variables and their use. Signed-off-by: Brett Creeley <brett.creeley@intel.com> Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-04-17 01:30:44 +08:00
int start = 0, end = 0;
ice: Refactor interrupt tracking Currently we have two MSI-x (IRQ) trackers, one for OS requested MSI-x entries (sw_irq_tracker) and one for hardware MSI-x vectors (hw_irq_tracker). Generally the sw_irq_tracker has less entries than the hw_irq_tracker because the hw_irq_tracker has entries equal to the max allowed MSI-x per PF and the sw_irq_tracker is mainly the minimum (non SR-IOV portion of the vectors, kernel granted IRQs). All of the non SR-IOV portions of the driver (i.e. LAN queues, RDMA queues, OICR, etc.) take at least one of each type of tracker resource. SR-IOV only grabs entries from the hw_irq_tracker. There are a few issues with this approach that can be seen when doing any kind of device reconfiguration (i.e. ethtool -L, SR-IOV, etc.). One of them being, any time the driver creates an ice_q_vector and associates it to a LAN queue pair it will grab and use one entry from the hw_irq_tracker and one from the sw_irq_tracker. If the indices on these does not match it will cause a Tx timeout, which will cause a reset and then the indices will match up again and traffic will resume. The mismatched indices come from the trackers not being the same size and/or the search_hint in the two trackers not being equal. Another reason for the refactor is the co-existence of features with SR-IOV. If SR-IOV is enabled and the interrupts are taken from the end of the sw_irq_tracker then other features can no longer use this space because the hardware has now given the remaining interrupts to SR-IOV. This patch reworks how we track MSI-x vectors by removing the hw_irq_tracker completely and instead MSI-x resources needed for SR-IOV are determined all at once instead of per VF. This can be done because when creating VFs we know how many are wanted and how many MSI-x vectors each VF needs. This also allows us to start using MSI-x resources from the end of the PF's allowed MSI-x vectors so we are less likely to use entries needed for other features (i.e. RDMA, L2 Offload, etc). This patch also reworks the ice_res_tracker structure by removing the search_hint and adding a new member - "end". Instead of having a search_hint we will always search from 0. The new member, "end", will be used to manipulate the end of the ice_res_tracker (specifically sw_irq_tracker) during runtime based on MSI-x vectors needed by SR-IOV. In the normal case, the end of ice_res_tracker will be equal to the ice_res_tracker's num_entries. The sriov_base_vector member was added to the PF structure. It is used to represent the starting MSI-x index of all the needed MSI-x vectors for all SR-IOV VFs. Depending on how many MSI-x are needed, SR-IOV may have to take resources from the sw_irq_tracker. This is done by setting the sw_irq_tracker->end equal to the pf->sriov_base_vector. When all SR-IOV VFs are removed then the sw_irq_tracker->end is reset back to sw_irq_tracker->num_entries. The sriov_base_vector, along with the VF's number of MSI-x (pf->num_vf_msix), vf_id, and the base MSI-x index on the PF (pf->hw.func_caps.common_cap.msix_vector_first_id), is used to calculate the first HW absolute MSI-x index for each VF, which is used to write to the VPINT_ALLOC[_PCI] and GLINT_VECT2FUNC registers to program the VFs MSI-x PCI configuration bits. Also, the sriov_base_vector is used along with VF's num_vf_msix, vf_id, and q_vector->v_idx to determine the MSI-x register index (used for writing to GLINT_DYN_CTL) within the PF's space. Interrupt changes removed any references to hw_base_vector, hw_oicr_idx, and hw_irq_tracker. Only sw_base_vector, sw_oicr_idx, and sw_irq_tracker variables remain. Change all of these by removing the "sw_" prefix to help avoid confusion with these variables and their use. Signed-off-by: Brett Creeley <brett.creeley@intel.com> Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-04-17 01:30:44 +08:00
if (needed > res->end)
return -ENOMEM;
id |= ICE_RES_VALID_BIT;
do {
/* skip already allocated entries */
if (res->list[end++] & ICE_RES_VALID_BIT) {
start = end;
ice: Refactor interrupt tracking Currently we have two MSI-x (IRQ) trackers, one for OS requested MSI-x entries (sw_irq_tracker) and one for hardware MSI-x vectors (hw_irq_tracker). Generally the sw_irq_tracker has less entries than the hw_irq_tracker because the hw_irq_tracker has entries equal to the max allowed MSI-x per PF and the sw_irq_tracker is mainly the minimum (non SR-IOV portion of the vectors, kernel granted IRQs). All of the non SR-IOV portions of the driver (i.e. LAN queues, RDMA queues, OICR, etc.) take at least one of each type of tracker resource. SR-IOV only grabs entries from the hw_irq_tracker. There are a few issues with this approach that can be seen when doing any kind of device reconfiguration (i.e. ethtool -L, SR-IOV, etc.). One of them being, any time the driver creates an ice_q_vector and associates it to a LAN queue pair it will grab and use one entry from the hw_irq_tracker and one from the sw_irq_tracker. If the indices on these does not match it will cause a Tx timeout, which will cause a reset and then the indices will match up again and traffic will resume. The mismatched indices come from the trackers not being the same size and/or the search_hint in the two trackers not being equal. Another reason for the refactor is the co-existence of features with SR-IOV. If SR-IOV is enabled and the interrupts are taken from the end of the sw_irq_tracker then other features can no longer use this space because the hardware has now given the remaining interrupts to SR-IOV. This patch reworks how we track MSI-x vectors by removing the hw_irq_tracker completely and instead MSI-x resources needed for SR-IOV are determined all at once instead of per VF. This can be done because when creating VFs we know how many are wanted and how many MSI-x vectors each VF needs. This also allows us to start using MSI-x resources from the end of the PF's allowed MSI-x vectors so we are less likely to use entries needed for other features (i.e. RDMA, L2 Offload, etc). This patch also reworks the ice_res_tracker structure by removing the search_hint and adding a new member - "end". Instead of having a search_hint we will always search from 0. The new member, "end", will be used to manipulate the end of the ice_res_tracker (specifically sw_irq_tracker) during runtime based on MSI-x vectors needed by SR-IOV. In the normal case, the end of ice_res_tracker will be equal to the ice_res_tracker's num_entries. The sriov_base_vector member was added to the PF structure. It is used to represent the starting MSI-x index of all the needed MSI-x vectors for all SR-IOV VFs. Depending on how many MSI-x are needed, SR-IOV may have to take resources from the sw_irq_tracker. This is done by setting the sw_irq_tracker->end equal to the pf->sriov_base_vector. When all SR-IOV VFs are removed then the sw_irq_tracker->end is reset back to sw_irq_tracker->num_entries. The sriov_base_vector, along with the VF's number of MSI-x (pf->num_vf_msix), vf_id, and the base MSI-x index on the PF (pf->hw.func_caps.common_cap.msix_vector_first_id), is used to calculate the first HW absolute MSI-x index for each VF, which is used to write to the VPINT_ALLOC[_PCI] and GLINT_VECT2FUNC registers to program the VFs MSI-x PCI configuration bits. Also, the sriov_base_vector is used along with VF's num_vf_msix, vf_id, and q_vector->v_idx to determine the MSI-x register index (used for writing to GLINT_DYN_CTL) within the PF's space. Interrupt changes removed any references to hw_base_vector, hw_oicr_idx, and hw_irq_tracker. Only sw_base_vector, sw_oicr_idx, and sw_irq_tracker variables remain. Change all of these by removing the "sw_" prefix to help avoid confusion with these variables and their use. Signed-off-by: Brett Creeley <brett.creeley@intel.com> Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-04-17 01:30:44 +08:00
if ((start + needed) > res->end)
break;
}
if (end == (start + needed)) {
int i = start;
/* there was enough, so assign it to the requestor */
while (i != end)
res->list[i++] = id;
return start;
}
ice: Refactor interrupt tracking Currently we have two MSI-x (IRQ) trackers, one for OS requested MSI-x entries (sw_irq_tracker) and one for hardware MSI-x vectors (hw_irq_tracker). Generally the sw_irq_tracker has less entries than the hw_irq_tracker because the hw_irq_tracker has entries equal to the max allowed MSI-x per PF and the sw_irq_tracker is mainly the minimum (non SR-IOV portion of the vectors, kernel granted IRQs). All of the non SR-IOV portions of the driver (i.e. LAN queues, RDMA queues, OICR, etc.) take at least one of each type of tracker resource. SR-IOV only grabs entries from the hw_irq_tracker. There are a few issues with this approach that can be seen when doing any kind of device reconfiguration (i.e. ethtool -L, SR-IOV, etc.). One of them being, any time the driver creates an ice_q_vector and associates it to a LAN queue pair it will grab and use one entry from the hw_irq_tracker and one from the sw_irq_tracker. If the indices on these does not match it will cause a Tx timeout, which will cause a reset and then the indices will match up again and traffic will resume. The mismatched indices come from the trackers not being the same size and/or the search_hint in the two trackers not being equal. Another reason for the refactor is the co-existence of features with SR-IOV. If SR-IOV is enabled and the interrupts are taken from the end of the sw_irq_tracker then other features can no longer use this space because the hardware has now given the remaining interrupts to SR-IOV. This patch reworks how we track MSI-x vectors by removing the hw_irq_tracker completely and instead MSI-x resources needed for SR-IOV are determined all at once instead of per VF. This can be done because when creating VFs we know how many are wanted and how many MSI-x vectors each VF needs. This also allows us to start using MSI-x resources from the end of the PF's allowed MSI-x vectors so we are less likely to use entries needed for other features (i.e. RDMA, L2 Offload, etc). This patch also reworks the ice_res_tracker structure by removing the search_hint and adding a new member - "end". Instead of having a search_hint we will always search from 0. The new member, "end", will be used to manipulate the end of the ice_res_tracker (specifically sw_irq_tracker) during runtime based on MSI-x vectors needed by SR-IOV. In the normal case, the end of ice_res_tracker will be equal to the ice_res_tracker's num_entries. The sriov_base_vector member was added to the PF structure. It is used to represent the starting MSI-x index of all the needed MSI-x vectors for all SR-IOV VFs. Depending on how many MSI-x are needed, SR-IOV may have to take resources from the sw_irq_tracker. This is done by setting the sw_irq_tracker->end equal to the pf->sriov_base_vector. When all SR-IOV VFs are removed then the sw_irq_tracker->end is reset back to sw_irq_tracker->num_entries. The sriov_base_vector, along with the VF's number of MSI-x (pf->num_vf_msix), vf_id, and the base MSI-x index on the PF (pf->hw.func_caps.common_cap.msix_vector_first_id), is used to calculate the first HW absolute MSI-x index for each VF, which is used to write to the VPINT_ALLOC[_PCI] and GLINT_VECT2FUNC registers to program the VFs MSI-x PCI configuration bits. Also, the sriov_base_vector is used along with VF's num_vf_msix, vf_id, and q_vector->v_idx to determine the MSI-x register index (used for writing to GLINT_DYN_CTL) within the PF's space. Interrupt changes removed any references to hw_base_vector, hw_oicr_idx, and hw_irq_tracker. Only sw_base_vector, sw_oicr_idx, and sw_irq_tracker variables remain. Change all of these by removing the "sw_" prefix to help avoid confusion with these variables and their use. Signed-off-by: Brett Creeley <brett.creeley@intel.com> Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-04-17 01:30:44 +08:00
} while (end < res->end);
return -ENOMEM;
}
/**
* ice_get_res - get a block of resources
* @pf: board private structure
* @res: pointer to the resource
* @needed: size of the block needed
* @id: identifier to track owner
*
ice: Refactor interrupt tracking Currently we have two MSI-x (IRQ) trackers, one for OS requested MSI-x entries (sw_irq_tracker) and one for hardware MSI-x vectors (hw_irq_tracker). Generally the sw_irq_tracker has less entries than the hw_irq_tracker because the hw_irq_tracker has entries equal to the max allowed MSI-x per PF and the sw_irq_tracker is mainly the minimum (non SR-IOV portion of the vectors, kernel granted IRQs). All of the non SR-IOV portions of the driver (i.e. LAN queues, RDMA queues, OICR, etc.) take at least one of each type of tracker resource. SR-IOV only grabs entries from the hw_irq_tracker. There are a few issues with this approach that can be seen when doing any kind of device reconfiguration (i.e. ethtool -L, SR-IOV, etc.). One of them being, any time the driver creates an ice_q_vector and associates it to a LAN queue pair it will grab and use one entry from the hw_irq_tracker and one from the sw_irq_tracker. If the indices on these does not match it will cause a Tx timeout, which will cause a reset and then the indices will match up again and traffic will resume. The mismatched indices come from the trackers not being the same size and/or the search_hint in the two trackers not being equal. Another reason for the refactor is the co-existence of features with SR-IOV. If SR-IOV is enabled and the interrupts are taken from the end of the sw_irq_tracker then other features can no longer use this space because the hardware has now given the remaining interrupts to SR-IOV. This patch reworks how we track MSI-x vectors by removing the hw_irq_tracker completely and instead MSI-x resources needed for SR-IOV are determined all at once instead of per VF. This can be done because when creating VFs we know how many are wanted and how many MSI-x vectors each VF needs. This also allows us to start using MSI-x resources from the end of the PF's allowed MSI-x vectors so we are less likely to use entries needed for other features (i.e. RDMA, L2 Offload, etc). This patch also reworks the ice_res_tracker structure by removing the search_hint and adding a new member - "end". Instead of having a search_hint we will always search from 0. The new member, "end", will be used to manipulate the end of the ice_res_tracker (specifically sw_irq_tracker) during runtime based on MSI-x vectors needed by SR-IOV. In the normal case, the end of ice_res_tracker will be equal to the ice_res_tracker's num_entries. The sriov_base_vector member was added to the PF structure. It is used to represent the starting MSI-x index of all the needed MSI-x vectors for all SR-IOV VFs. Depending on how many MSI-x are needed, SR-IOV may have to take resources from the sw_irq_tracker. This is done by setting the sw_irq_tracker->end equal to the pf->sriov_base_vector. When all SR-IOV VFs are removed then the sw_irq_tracker->end is reset back to sw_irq_tracker->num_entries. The sriov_base_vector, along with the VF's number of MSI-x (pf->num_vf_msix), vf_id, and the base MSI-x index on the PF (pf->hw.func_caps.common_cap.msix_vector_first_id), is used to calculate the first HW absolute MSI-x index for each VF, which is used to write to the VPINT_ALLOC[_PCI] and GLINT_VECT2FUNC registers to program the VFs MSI-x PCI configuration bits. Also, the sriov_base_vector is used along with VF's num_vf_msix, vf_id, and q_vector->v_idx to determine the MSI-x register index (used for writing to GLINT_DYN_CTL) within the PF's space. Interrupt changes removed any references to hw_base_vector, hw_oicr_idx, and hw_irq_tracker. Only sw_base_vector, sw_oicr_idx, and sw_irq_tracker variables remain. Change all of these by removing the "sw_" prefix to help avoid confusion with these variables and their use. Signed-off-by: Brett Creeley <brett.creeley@intel.com> Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-04-17 01:30:44 +08:00
* Returns the base item index of the block, or negative for error
*/
int
ice_get_res(struct ice_pf *pf, struct ice_res_tracker *res, u16 needed, u16 id)
{
if (!res || !pf)
return -EINVAL;
if (!needed || needed > res->num_entries || id >= ICE_RES_VALID_BIT) {
dev_err(ice_pf_to_dev(pf),
"param err: needed=%d, num_entries = %d id=0x%04x\n",
needed, res->num_entries, id);
return -EINVAL;
}
ice: Refactor interrupt tracking Currently we have two MSI-x (IRQ) trackers, one for OS requested MSI-x entries (sw_irq_tracker) and one for hardware MSI-x vectors (hw_irq_tracker). Generally the sw_irq_tracker has less entries than the hw_irq_tracker because the hw_irq_tracker has entries equal to the max allowed MSI-x per PF and the sw_irq_tracker is mainly the minimum (non SR-IOV portion of the vectors, kernel granted IRQs). All of the non SR-IOV portions of the driver (i.e. LAN queues, RDMA queues, OICR, etc.) take at least one of each type of tracker resource. SR-IOV only grabs entries from the hw_irq_tracker. There are a few issues with this approach that can be seen when doing any kind of device reconfiguration (i.e. ethtool -L, SR-IOV, etc.). One of them being, any time the driver creates an ice_q_vector and associates it to a LAN queue pair it will grab and use one entry from the hw_irq_tracker and one from the sw_irq_tracker. If the indices on these does not match it will cause a Tx timeout, which will cause a reset and then the indices will match up again and traffic will resume. The mismatched indices come from the trackers not being the same size and/or the search_hint in the two trackers not being equal. Another reason for the refactor is the co-existence of features with SR-IOV. If SR-IOV is enabled and the interrupts are taken from the end of the sw_irq_tracker then other features can no longer use this space because the hardware has now given the remaining interrupts to SR-IOV. This patch reworks how we track MSI-x vectors by removing the hw_irq_tracker completely and instead MSI-x resources needed for SR-IOV are determined all at once instead of per VF. This can be done because when creating VFs we know how many are wanted and how many MSI-x vectors each VF needs. This also allows us to start using MSI-x resources from the end of the PF's allowed MSI-x vectors so we are less likely to use entries needed for other features (i.e. RDMA, L2 Offload, etc). This patch also reworks the ice_res_tracker structure by removing the search_hint and adding a new member - "end". Instead of having a search_hint we will always search from 0. The new member, "end", will be used to manipulate the end of the ice_res_tracker (specifically sw_irq_tracker) during runtime based on MSI-x vectors needed by SR-IOV. In the normal case, the end of ice_res_tracker will be equal to the ice_res_tracker's num_entries. The sriov_base_vector member was added to the PF structure. It is used to represent the starting MSI-x index of all the needed MSI-x vectors for all SR-IOV VFs. Depending on how many MSI-x are needed, SR-IOV may have to take resources from the sw_irq_tracker. This is done by setting the sw_irq_tracker->end equal to the pf->sriov_base_vector. When all SR-IOV VFs are removed then the sw_irq_tracker->end is reset back to sw_irq_tracker->num_entries. The sriov_base_vector, along with the VF's number of MSI-x (pf->num_vf_msix), vf_id, and the base MSI-x index on the PF (pf->hw.func_caps.common_cap.msix_vector_first_id), is used to calculate the first HW absolute MSI-x index for each VF, which is used to write to the VPINT_ALLOC[_PCI] and GLINT_VECT2FUNC registers to program the VFs MSI-x PCI configuration bits. Also, the sriov_base_vector is used along with VF's num_vf_msix, vf_id, and q_vector->v_idx to determine the MSI-x register index (used for writing to GLINT_DYN_CTL) within the PF's space. Interrupt changes removed any references to hw_base_vector, hw_oicr_idx, and hw_irq_tracker. Only sw_base_vector, sw_oicr_idx, and sw_irq_tracker variables remain. Change all of these by removing the "sw_" prefix to help avoid confusion with these variables and their use. Signed-off-by: Brett Creeley <brett.creeley@intel.com> Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-04-17 01:30:44 +08:00
return ice_search_res(res, needed, id);
}
/**
* ice_vsi_dis_irq - Mask off queue interrupt generation on the VSI
* @vsi: the VSI being un-configured
*/
void ice_vsi_dis_irq(struct ice_vsi *vsi)
{
ice: Refactor interrupt tracking Currently we have two MSI-x (IRQ) trackers, one for OS requested MSI-x entries (sw_irq_tracker) and one for hardware MSI-x vectors (hw_irq_tracker). Generally the sw_irq_tracker has less entries than the hw_irq_tracker because the hw_irq_tracker has entries equal to the max allowed MSI-x per PF and the sw_irq_tracker is mainly the minimum (non SR-IOV portion of the vectors, kernel granted IRQs). All of the non SR-IOV portions of the driver (i.e. LAN queues, RDMA queues, OICR, etc.) take at least one of each type of tracker resource. SR-IOV only grabs entries from the hw_irq_tracker. There are a few issues with this approach that can be seen when doing any kind of device reconfiguration (i.e. ethtool -L, SR-IOV, etc.). One of them being, any time the driver creates an ice_q_vector and associates it to a LAN queue pair it will grab and use one entry from the hw_irq_tracker and one from the sw_irq_tracker. If the indices on these does not match it will cause a Tx timeout, which will cause a reset and then the indices will match up again and traffic will resume. The mismatched indices come from the trackers not being the same size and/or the search_hint in the two trackers not being equal. Another reason for the refactor is the co-existence of features with SR-IOV. If SR-IOV is enabled and the interrupts are taken from the end of the sw_irq_tracker then other features can no longer use this space because the hardware has now given the remaining interrupts to SR-IOV. This patch reworks how we track MSI-x vectors by removing the hw_irq_tracker completely and instead MSI-x resources needed for SR-IOV are determined all at once instead of per VF. This can be done because when creating VFs we know how many are wanted and how many MSI-x vectors each VF needs. This also allows us to start using MSI-x resources from the end of the PF's allowed MSI-x vectors so we are less likely to use entries needed for other features (i.e. RDMA, L2 Offload, etc). This patch also reworks the ice_res_tracker structure by removing the search_hint and adding a new member - "end". Instead of having a search_hint we will always search from 0. The new member, "end", will be used to manipulate the end of the ice_res_tracker (specifically sw_irq_tracker) during runtime based on MSI-x vectors needed by SR-IOV. In the normal case, the end of ice_res_tracker will be equal to the ice_res_tracker's num_entries. The sriov_base_vector member was added to the PF structure. It is used to represent the starting MSI-x index of all the needed MSI-x vectors for all SR-IOV VFs. Depending on how many MSI-x are needed, SR-IOV may have to take resources from the sw_irq_tracker. This is done by setting the sw_irq_tracker->end equal to the pf->sriov_base_vector. When all SR-IOV VFs are removed then the sw_irq_tracker->end is reset back to sw_irq_tracker->num_entries. The sriov_base_vector, along with the VF's number of MSI-x (pf->num_vf_msix), vf_id, and the base MSI-x index on the PF (pf->hw.func_caps.common_cap.msix_vector_first_id), is used to calculate the first HW absolute MSI-x index for each VF, which is used to write to the VPINT_ALLOC[_PCI] and GLINT_VECT2FUNC registers to program the VFs MSI-x PCI configuration bits. Also, the sriov_base_vector is used along with VF's num_vf_msix, vf_id, and q_vector->v_idx to determine the MSI-x register index (used for writing to GLINT_DYN_CTL) within the PF's space. Interrupt changes removed any references to hw_base_vector, hw_oicr_idx, and hw_irq_tracker. Only sw_base_vector, sw_oicr_idx, and sw_irq_tracker variables remain. Change all of these by removing the "sw_" prefix to help avoid confusion with these variables and their use. Signed-off-by: Brett Creeley <brett.creeley@intel.com> Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-04-17 01:30:44 +08:00
int base = vsi->base_vector;
struct ice_pf *pf = vsi->back;
struct ice_hw *hw = &pf->hw;
u32 val;
int i;
/* disable interrupt causation from each queue */
if (vsi->tx_rings) {
ice_for_each_txq(vsi, i) {
if (vsi->tx_rings[i]) {
u16 reg;
reg = vsi->tx_rings[i]->reg_idx;
val = rd32(hw, QINT_TQCTL(reg));
val &= ~QINT_TQCTL_CAUSE_ENA_M;
wr32(hw, QINT_TQCTL(reg), val);
}
}
}
if (vsi->rx_rings) {
ice_for_each_rxq(vsi, i) {
if (vsi->rx_rings[i]) {
u16 reg;
reg = vsi->rx_rings[i]->reg_idx;
val = rd32(hw, QINT_RQCTL(reg));
val &= ~QINT_RQCTL_CAUSE_ENA_M;
wr32(hw, QINT_RQCTL(reg), val);
}
}
}
/* disable each interrupt */
ice_for_each_q_vector(vsi, i) {
if (!vsi->q_vectors[i])
continue;
wr32(hw, GLINT_DYN_CTL(vsi->q_vectors[i]->reg_idx), 0);
}
ice_flush(hw);
/* don't call synchronize_irq() for VF's from the host */
if (vsi->type == ICE_VSI_VF)
return;
ice_for_each_q_vector(vsi, i)
synchronize_irq(pf->msix_entries[i + base].vector);
}
/**
* ice_napi_del - Remove NAPI handler for the VSI
* @vsi: VSI for which NAPI handler is to be removed
*/
void ice_napi_del(struct ice_vsi *vsi)
{
int v_idx;
if (!vsi->netdev)
return;
ice_for_each_q_vector(vsi, v_idx)
netif_napi_del(&vsi->q_vectors[v_idx]->napi);
}
/**
* ice_vsi_release - Delete a VSI and free its resources
* @vsi: the VSI being removed
*
* Returns 0 on success or < 0 on error
*/
int ice_vsi_release(struct ice_vsi *vsi)
{
struct ice_pf *pf;
if (!vsi->back)
return -ENODEV;
pf = vsi->back;
/* do not unregister while driver is in the reset recovery pending
* state. Since reset/rebuild happens through PF service task workqueue,
* it's not a good idea to unregister netdev that is associated to the
* PF that is running the work queue items currently. This is done to
* avoid check_flush_dependency() warning on this wq
*/
if (vsi->netdev && !ice_is_reset_in_progress(pf->state))
unregister_netdev(vsi->netdev);
if (test_bit(ICE_FLAG_RSS_ENA, pf->flags))
ice_rss_clean(vsi);
/* Disable VSI and free resources */
if (vsi->type != ICE_VSI_LB)
ice_vsi_dis_irq(vsi);
ice_vsi_close(vsi);
ice: Refactor interrupt tracking Currently we have two MSI-x (IRQ) trackers, one for OS requested MSI-x entries (sw_irq_tracker) and one for hardware MSI-x vectors (hw_irq_tracker). Generally the sw_irq_tracker has less entries than the hw_irq_tracker because the hw_irq_tracker has entries equal to the max allowed MSI-x per PF and the sw_irq_tracker is mainly the minimum (non SR-IOV portion of the vectors, kernel granted IRQs). All of the non SR-IOV portions of the driver (i.e. LAN queues, RDMA queues, OICR, etc.) take at least one of each type of tracker resource. SR-IOV only grabs entries from the hw_irq_tracker. There are a few issues with this approach that can be seen when doing any kind of device reconfiguration (i.e. ethtool -L, SR-IOV, etc.). One of them being, any time the driver creates an ice_q_vector and associates it to a LAN queue pair it will grab and use one entry from the hw_irq_tracker and one from the sw_irq_tracker. If the indices on these does not match it will cause a Tx timeout, which will cause a reset and then the indices will match up again and traffic will resume. The mismatched indices come from the trackers not being the same size and/or the search_hint in the two trackers not being equal. Another reason for the refactor is the co-existence of features with SR-IOV. If SR-IOV is enabled and the interrupts are taken from the end of the sw_irq_tracker then other features can no longer use this space because the hardware has now given the remaining interrupts to SR-IOV. This patch reworks how we track MSI-x vectors by removing the hw_irq_tracker completely and instead MSI-x resources needed for SR-IOV are determined all at once instead of per VF. This can be done because when creating VFs we know how many are wanted and how many MSI-x vectors each VF needs. This also allows us to start using MSI-x resources from the end of the PF's allowed MSI-x vectors so we are less likely to use entries needed for other features (i.e. RDMA, L2 Offload, etc). This patch also reworks the ice_res_tracker structure by removing the search_hint and adding a new member - "end". Instead of having a search_hint we will always search from 0. The new member, "end", will be used to manipulate the end of the ice_res_tracker (specifically sw_irq_tracker) during runtime based on MSI-x vectors needed by SR-IOV. In the normal case, the end of ice_res_tracker will be equal to the ice_res_tracker's num_entries. The sriov_base_vector member was added to the PF structure. It is used to represent the starting MSI-x index of all the needed MSI-x vectors for all SR-IOV VFs. Depending on how many MSI-x are needed, SR-IOV may have to take resources from the sw_irq_tracker. This is done by setting the sw_irq_tracker->end equal to the pf->sriov_base_vector. When all SR-IOV VFs are removed then the sw_irq_tracker->end is reset back to sw_irq_tracker->num_entries. The sriov_base_vector, along with the VF's number of MSI-x (pf->num_vf_msix), vf_id, and the base MSI-x index on the PF (pf->hw.func_caps.common_cap.msix_vector_first_id), is used to calculate the first HW absolute MSI-x index for each VF, which is used to write to the VPINT_ALLOC[_PCI] and GLINT_VECT2FUNC registers to program the VFs MSI-x PCI configuration bits. Also, the sriov_base_vector is used along with VF's num_vf_msix, vf_id, and q_vector->v_idx to determine the MSI-x register index (used for writing to GLINT_DYN_CTL) within the PF's space. Interrupt changes removed any references to hw_base_vector, hw_oicr_idx, and hw_irq_tracker. Only sw_base_vector, sw_oicr_idx, and sw_irq_tracker variables remain. Change all of these by removing the "sw_" prefix to help avoid confusion with these variables and their use. Signed-off-by: Brett Creeley <brett.creeley@intel.com> Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-04-17 01:30:44 +08:00
/* SR-IOV determines needed MSIX resources all at once instead of per
* VSI since when VFs are spawned we know how many VFs there are and how
* many interrupts each VF needs. SR-IOV MSIX resources are also
* cleared in the same manner.
*/
if (vsi->type != ICE_VSI_VF) {
/* reclaim SW interrupts back to the common pool */
ice: Refactor interrupt tracking Currently we have two MSI-x (IRQ) trackers, one for OS requested MSI-x entries (sw_irq_tracker) and one for hardware MSI-x vectors (hw_irq_tracker). Generally the sw_irq_tracker has less entries than the hw_irq_tracker because the hw_irq_tracker has entries equal to the max allowed MSI-x per PF and the sw_irq_tracker is mainly the minimum (non SR-IOV portion of the vectors, kernel granted IRQs). All of the non SR-IOV portions of the driver (i.e. LAN queues, RDMA queues, OICR, etc.) take at least one of each type of tracker resource. SR-IOV only grabs entries from the hw_irq_tracker. There are a few issues with this approach that can be seen when doing any kind of device reconfiguration (i.e. ethtool -L, SR-IOV, etc.). One of them being, any time the driver creates an ice_q_vector and associates it to a LAN queue pair it will grab and use one entry from the hw_irq_tracker and one from the sw_irq_tracker. If the indices on these does not match it will cause a Tx timeout, which will cause a reset and then the indices will match up again and traffic will resume. The mismatched indices come from the trackers not being the same size and/or the search_hint in the two trackers not being equal. Another reason for the refactor is the co-existence of features with SR-IOV. If SR-IOV is enabled and the interrupts are taken from the end of the sw_irq_tracker then other features can no longer use this space because the hardware has now given the remaining interrupts to SR-IOV. This patch reworks how we track MSI-x vectors by removing the hw_irq_tracker completely and instead MSI-x resources needed for SR-IOV are determined all at once instead of per VF. This can be done because when creating VFs we know how many are wanted and how many MSI-x vectors each VF needs. This also allows us to start using MSI-x resources from the end of the PF's allowed MSI-x vectors so we are less likely to use entries needed for other features (i.e. RDMA, L2 Offload, etc). This patch also reworks the ice_res_tracker structure by removing the search_hint and adding a new member - "end". Instead of having a search_hint we will always search from 0. The new member, "end", will be used to manipulate the end of the ice_res_tracker (specifically sw_irq_tracker) during runtime based on MSI-x vectors needed by SR-IOV. In the normal case, the end of ice_res_tracker will be equal to the ice_res_tracker's num_entries. The sriov_base_vector member was added to the PF structure. It is used to represent the starting MSI-x index of all the needed MSI-x vectors for all SR-IOV VFs. Depending on how many MSI-x are needed, SR-IOV may have to take resources from the sw_irq_tracker. This is done by setting the sw_irq_tracker->end equal to the pf->sriov_base_vector. When all SR-IOV VFs are removed then the sw_irq_tracker->end is reset back to sw_irq_tracker->num_entries. The sriov_base_vector, along with the VF's number of MSI-x (pf->num_vf_msix), vf_id, and the base MSI-x index on the PF (pf->hw.func_caps.common_cap.msix_vector_first_id), is used to calculate the first HW absolute MSI-x index for each VF, which is used to write to the VPINT_ALLOC[_PCI] and GLINT_VECT2FUNC registers to program the VFs MSI-x PCI configuration bits. Also, the sriov_base_vector is used along with VF's num_vf_msix, vf_id, and q_vector->v_idx to determine the MSI-x register index (used for writing to GLINT_DYN_CTL) within the PF's space. Interrupt changes removed any references to hw_base_vector, hw_oicr_idx, and hw_irq_tracker. Only sw_base_vector, sw_oicr_idx, and sw_irq_tracker variables remain. Change all of these by removing the "sw_" prefix to help avoid confusion with these variables and their use. Signed-off-by: Brett Creeley <brett.creeley@intel.com> Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-04-17 01:30:44 +08:00
ice_free_res(pf->irq_tracker, vsi->base_vector, vsi->idx);
pf->num_avail_sw_msix += vsi->num_q_vectors;
}
if (!ice_is_safe_mode(pf)) {
if (vsi->type == ICE_VSI_PF) {
ice_vsi_add_rem_eth_mac(vsi, false);
ice_cfg_sw_lldp(vsi, true, false);
/* The Rx rule will only exist to remove if the LLDP FW
* engine is currently stopped
*/
if (!test_bit(ICE_FLAG_FW_LLDP_AGENT, pf->flags))
ice_cfg_sw_lldp(vsi, false, false);
}
}
ice_remove_vsi_fltr(&pf->hw, vsi->idx);
ice_rm_vsi_lan_cfg(vsi->port_info, vsi->idx);
ice_vsi_delete(vsi);
ice_vsi_free_q_vectors(vsi);
/* make sure unregister_netdev() was called by checking __ICE_DOWN */
if (vsi->netdev && test_bit(__ICE_DOWN, vsi->state)) {
free_netdev(vsi->netdev);
vsi->netdev = NULL;
}
ice_vsi_clear_rings(vsi);
ice_vsi_put_qs(vsi);
/* retain SW VSI data structure since it is needed to unregister and
* free VSI netdev when PF is not in reset recovery pending state,\
* for ex: during rmmod.
*/
if (!ice_is_reset_in_progress(pf->state))
ice_vsi_clear(vsi);
return 0;
}
/**
* ice_vsi_rebuild_update_coalesce - set coalesce for a q_vector
* @q_vector: pointer to q_vector which is being updated
* @coalesce: pointer to array of struct with stored coalesce
*
* Set coalesce param in q_vector and update these parameters in HW.
*/
static void
ice_vsi_rebuild_update_coalesce(struct ice_q_vector *q_vector,
struct ice_coalesce_stored *coalesce)
{
struct ice_ring_container *rx_rc = &q_vector->rx;
struct ice_ring_container *tx_rc = &q_vector->tx;
struct ice_hw *hw = &q_vector->vsi->back->hw;
tx_rc->itr_setting = coalesce->itr_tx;
rx_rc->itr_setting = coalesce->itr_rx;
/* dynamic ITR values will be updated during Tx/Rx */
if (!ITR_IS_DYNAMIC(tx_rc->itr_setting))
wr32(hw, GLINT_ITR(tx_rc->itr_idx, q_vector->reg_idx),
ITR_REG_ALIGN(tx_rc->itr_setting) >>
ICE_ITR_GRAN_S);
if (!ITR_IS_DYNAMIC(rx_rc->itr_setting))
wr32(hw, GLINT_ITR(rx_rc->itr_idx, q_vector->reg_idx),
ITR_REG_ALIGN(rx_rc->itr_setting) >>
ICE_ITR_GRAN_S);
q_vector->intrl = coalesce->intrl;
wr32(hw, GLINT_RATE(q_vector->reg_idx),
ice_intrl_usec_to_reg(q_vector->intrl, hw->intrl_gran));
}
/**
* ice_vsi_rebuild_get_coalesce - get coalesce from all q_vectors
* @vsi: VSI connected with q_vectors
* @coalesce: array of struct with stored coalesce
*
* Returns array size.
*/
static int
ice_vsi_rebuild_get_coalesce(struct ice_vsi *vsi,
struct ice_coalesce_stored *coalesce)
{
int i;
ice_for_each_q_vector(vsi, i) {
struct ice_q_vector *q_vector = vsi->q_vectors[i];
coalesce[i].itr_tx = q_vector->tx.itr_setting;
coalesce[i].itr_rx = q_vector->rx.itr_setting;
coalesce[i].intrl = q_vector->intrl;
}
return vsi->num_q_vectors;
}
/**
* ice_vsi_rebuild_set_coalesce - set coalesce from earlier saved arrays
* @vsi: VSI connected with q_vectors
* @coalesce: pointer to array of struct with stored coalesce
* @size: size of coalesce array
*
* Before this function, ice_vsi_rebuild_get_coalesce should be called to save
* ITR params in arrays. If size is 0 or coalesce wasn't stored set coalesce
* to default value.
*/
static void
ice_vsi_rebuild_set_coalesce(struct ice_vsi *vsi,
struct ice_coalesce_stored *coalesce, int size)
{
int i;
if ((size && !coalesce) || !vsi)
return;
for (i = 0; i < size && i < vsi->num_q_vectors; i++)
ice_vsi_rebuild_update_coalesce(vsi->q_vectors[i],
&coalesce[i]);
for (; i < vsi->num_q_vectors; i++) {
struct ice_coalesce_stored coalesce_dflt = {
.itr_tx = ICE_DFLT_TX_ITR,
.itr_rx = ICE_DFLT_RX_ITR,
.intrl = 0
};
ice_vsi_rebuild_update_coalesce(vsi->q_vectors[i],
&coalesce_dflt);
}
}
/**
* ice_vsi_rebuild - Rebuild VSI after reset
* @vsi: VSI to be rebuild
* @init_vsi: is this an initialization or a reconfigure of the VSI
*
* Returns 0 on success and negative value on failure
*/
int ice_vsi_rebuild(struct ice_vsi *vsi, bool init_vsi)
{
u16 max_txqs[ICE_MAX_TRAFFIC_CLASS] = { 0 };
struct ice_coalesce_stored *coalesce;
int prev_num_q_vectors = 0;
struct ice_vf *vf = NULL;
enum ice_status status;
struct ice_pf *pf;
int ret, i;
if (!vsi)
return -EINVAL;
pf = vsi->back;
if (vsi->type == ICE_VSI_VF)
vf = &pf->vf[vsi->vf_id];
coalesce = kcalloc(vsi->num_q_vectors,
sizeof(struct ice_coalesce_stored), GFP_KERNEL);
if (coalesce)
prev_num_q_vectors = ice_vsi_rebuild_get_coalesce(vsi,
coalesce);
ice_rm_vsi_lan_cfg(vsi->port_info, vsi->idx);
ice_vsi_free_q_vectors(vsi);
ice: Refactor interrupt tracking Currently we have two MSI-x (IRQ) trackers, one for OS requested MSI-x entries (sw_irq_tracker) and one for hardware MSI-x vectors (hw_irq_tracker). Generally the sw_irq_tracker has less entries than the hw_irq_tracker because the hw_irq_tracker has entries equal to the max allowed MSI-x per PF and the sw_irq_tracker is mainly the minimum (non SR-IOV portion of the vectors, kernel granted IRQs). All of the non SR-IOV portions of the driver (i.e. LAN queues, RDMA queues, OICR, etc.) take at least one of each type of tracker resource. SR-IOV only grabs entries from the hw_irq_tracker. There are a few issues with this approach that can be seen when doing any kind of device reconfiguration (i.e. ethtool -L, SR-IOV, etc.). One of them being, any time the driver creates an ice_q_vector and associates it to a LAN queue pair it will grab and use one entry from the hw_irq_tracker and one from the sw_irq_tracker. If the indices on these does not match it will cause a Tx timeout, which will cause a reset and then the indices will match up again and traffic will resume. The mismatched indices come from the trackers not being the same size and/or the search_hint in the two trackers not being equal. Another reason for the refactor is the co-existence of features with SR-IOV. If SR-IOV is enabled and the interrupts are taken from the end of the sw_irq_tracker then other features can no longer use this space because the hardware has now given the remaining interrupts to SR-IOV. This patch reworks how we track MSI-x vectors by removing the hw_irq_tracker completely and instead MSI-x resources needed for SR-IOV are determined all at once instead of per VF. This can be done because when creating VFs we know how many are wanted and how many MSI-x vectors each VF needs. This also allows us to start using MSI-x resources from the end of the PF's allowed MSI-x vectors so we are less likely to use entries needed for other features (i.e. RDMA, L2 Offload, etc). This patch also reworks the ice_res_tracker structure by removing the search_hint and adding a new member - "end". Instead of having a search_hint we will always search from 0. The new member, "end", will be used to manipulate the end of the ice_res_tracker (specifically sw_irq_tracker) during runtime based on MSI-x vectors needed by SR-IOV. In the normal case, the end of ice_res_tracker will be equal to the ice_res_tracker's num_entries. The sriov_base_vector member was added to the PF structure. It is used to represent the starting MSI-x index of all the needed MSI-x vectors for all SR-IOV VFs. Depending on how many MSI-x are needed, SR-IOV may have to take resources from the sw_irq_tracker. This is done by setting the sw_irq_tracker->end equal to the pf->sriov_base_vector. When all SR-IOV VFs are removed then the sw_irq_tracker->end is reset back to sw_irq_tracker->num_entries. The sriov_base_vector, along with the VF's number of MSI-x (pf->num_vf_msix), vf_id, and the base MSI-x index on the PF (pf->hw.func_caps.common_cap.msix_vector_first_id), is used to calculate the first HW absolute MSI-x index for each VF, which is used to write to the VPINT_ALLOC[_PCI] and GLINT_VECT2FUNC registers to program the VFs MSI-x PCI configuration bits. Also, the sriov_base_vector is used along with VF's num_vf_msix, vf_id, and q_vector->v_idx to determine the MSI-x register index (used for writing to GLINT_DYN_CTL) within the PF's space. Interrupt changes removed any references to hw_base_vector, hw_oicr_idx, and hw_irq_tracker. Only sw_base_vector, sw_oicr_idx, and sw_irq_tracker variables remain. Change all of these by removing the "sw_" prefix to help avoid confusion with these variables and their use. Signed-off-by: Brett Creeley <brett.creeley@intel.com> Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-04-17 01:30:44 +08:00
/* SR-IOV determines needed MSIX resources all at once instead of per
* VSI since when VFs are spawned we know how many VFs there are and how
* many interrupts each VF needs. SR-IOV MSIX resources are also
* cleared in the same manner.
*/
if (vsi->type != ICE_VSI_VF) {
/* reclaim SW interrupts back to the common pool */
ice: Refactor interrupt tracking Currently we have two MSI-x (IRQ) trackers, one for OS requested MSI-x entries (sw_irq_tracker) and one for hardware MSI-x vectors (hw_irq_tracker). Generally the sw_irq_tracker has less entries than the hw_irq_tracker because the hw_irq_tracker has entries equal to the max allowed MSI-x per PF and the sw_irq_tracker is mainly the minimum (non SR-IOV portion of the vectors, kernel granted IRQs). All of the non SR-IOV portions of the driver (i.e. LAN queues, RDMA queues, OICR, etc.) take at least one of each type of tracker resource. SR-IOV only grabs entries from the hw_irq_tracker. There are a few issues with this approach that can be seen when doing any kind of device reconfiguration (i.e. ethtool -L, SR-IOV, etc.). One of them being, any time the driver creates an ice_q_vector and associates it to a LAN queue pair it will grab and use one entry from the hw_irq_tracker and one from the sw_irq_tracker. If the indices on these does not match it will cause a Tx timeout, which will cause a reset and then the indices will match up again and traffic will resume. The mismatched indices come from the trackers not being the same size and/or the search_hint in the two trackers not being equal. Another reason for the refactor is the co-existence of features with SR-IOV. If SR-IOV is enabled and the interrupts are taken from the end of the sw_irq_tracker then other features can no longer use this space because the hardware has now given the remaining interrupts to SR-IOV. This patch reworks how we track MSI-x vectors by removing the hw_irq_tracker completely and instead MSI-x resources needed for SR-IOV are determined all at once instead of per VF. This can be done because when creating VFs we know how many are wanted and how many MSI-x vectors each VF needs. This also allows us to start using MSI-x resources from the end of the PF's allowed MSI-x vectors so we are less likely to use entries needed for other features (i.e. RDMA, L2 Offload, etc). This patch also reworks the ice_res_tracker structure by removing the search_hint and adding a new member - "end". Instead of having a search_hint we will always search from 0. The new member, "end", will be used to manipulate the end of the ice_res_tracker (specifically sw_irq_tracker) during runtime based on MSI-x vectors needed by SR-IOV. In the normal case, the end of ice_res_tracker will be equal to the ice_res_tracker's num_entries. The sriov_base_vector member was added to the PF structure. It is used to represent the starting MSI-x index of all the needed MSI-x vectors for all SR-IOV VFs. Depending on how many MSI-x are needed, SR-IOV may have to take resources from the sw_irq_tracker. This is done by setting the sw_irq_tracker->end equal to the pf->sriov_base_vector. When all SR-IOV VFs are removed then the sw_irq_tracker->end is reset back to sw_irq_tracker->num_entries. The sriov_base_vector, along with the VF's number of MSI-x (pf->num_vf_msix), vf_id, and the base MSI-x index on the PF (pf->hw.func_caps.common_cap.msix_vector_first_id), is used to calculate the first HW absolute MSI-x index for each VF, which is used to write to the VPINT_ALLOC[_PCI] and GLINT_VECT2FUNC registers to program the VFs MSI-x PCI configuration bits. Also, the sriov_base_vector is used along with VF's num_vf_msix, vf_id, and q_vector->v_idx to determine the MSI-x register index (used for writing to GLINT_DYN_CTL) within the PF's space. Interrupt changes removed any references to hw_base_vector, hw_oicr_idx, and hw_irq_tracker. Only sw_base_vector, sw_oicr_idx, and sw_irq_tracker variables remain. Change all of these by removing the "sw_" prefix to help avoid confusion with these variables and their use. Signed-off-by: Brett Creeley <brett.creeley@intel.com> Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-04-17 01:30:44 +08:00
ice_free_res(pf->irq_tracker, vsi->base_vector, vsi->idx);
pf->num_avail_sw_msix += vsi->num_q_vectors;
ice: Refactor interrupt tracking Currently we have two MSI-x (IRQ) trackers, one for OS requested MSI-x entries (sw_irq_tracker) and one for hardware MSI-x vectors (hw_irq_tracker). Generally the sw_irq_tracker has less entries than the hw_irq_tracker because the hw_irq_tracker has entries equal to the max allowed MSI-x per PF and the sw_irq_tracker is mainly the minimum (non SR-IOV portion of the vectors, kernel granted IRQs). All of the non SR-IOV portions of the driver (i.e. LAN queues, RDMA queues, OICR, etc.) take at least one of each type of tracker resource. SR-IOV only grabs entries from the hw_irq_tracker. There are a few issues with this approach that can be seen when doing any kind of device reconfiguration (i.e. ethtool -L, SR-IOV, etc.). One of them being, any time the driver creates an ice_q_vector and associates it to a LAN queue pair it will grab and use one entry from the hw_irq_tracker and one from the sw_irq_tracker. If the indices on these does not match it will cause a Tx timeout, which will cause a reset and then the indices will match up again and traffic will resume. The mismatched indices come from the trackers not being the same size and/or the search_hint in the two trackers not being equal. Another reason for the refactor is the co-existence of features with SR-IOV. If SR-IOV is enabled and the interrupts are taken from the end of the sw_irq_tracker then other features can no longer use this space because the hardware has now given the remaining interrupts to SR-IOV. This patch reworks how we track MSI-x vectors by removing the hw_irq_tracker completely and instead MSI-x resources needed for SR-IOV are determined all at once instead of per VF. This can be done because when creating VFs we know how many are wanted and how many MSI-x vectors each VF needs. This also allows us to start using MSI-x resources from the end of the PF's allowed MSI-x vectors so we are less likely to use entries needed for other features (i.e. RDMA, L2 Offload, etc). This patch also reworks the ice_res_tracker structure by removing the search_hint and adding a new member - "end". Instead of having a search_hint we will always search from 0. The new member, "end", will be used to manipulate the end of the ice_res_tracker (specifically sw_irq_tracker) during runtime based on MSI-x vectors needed by SR-IOV. In the normal case, the end of ice_res_tracker will be equal to the ice_res_tracker's num_entries. The sriov_base_vector member was added to the PF structure. It is used to represent the starting MSI-x index of all the needed MSI-x vectors for all SR-IOV VFs. Depending on how many MSI-x are needed, SR-IOV may have to take resources from the sw_irq_tracker. This is done by setting the sw_irq_tracker->end equal to the pf->sriov_base_vector. When all SR-IOV VFs are removed then the sw_irq_tracker->end is reset back to sw_irq_tracker->num_entries. The sriov_base_vector, along with the VF's number of MSI-x (pf->num_vf_msix), vf_id, and the base MSI-x index on the PF (pf->hw.func_caps.common_cap.msix_vector_first_id), is used to calculate the first HW absolute MSI-x index for each VF, which is used to write to the VPINT_ALLOC[_PCI] and GLINT_VECT2FUNC registers to program the VFs MSI-x PCI configuration bits. Also, the sriov_base_vector is used along with VF's num_vf_msix, vf_id, and q_vector->v_idx to determine the MSI-x register index (used for writing to GLINT_DYN_CTL) within the PF's space. Interrupt changes removed any references to hw_base_vector, hw_oicr_idx, and hw_irq_tracker. Only sw_base_vector, sw_oicr_idx, and sw_irq_tracker variables remain. Change all of these by removing the "sw_" prefix to help avoid confusion with these variables and their use. Signed-off-by: Brett Creeley <brett.creeley@intel.com> Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-04-17 01:30:44 +08:00
vsi->base_vector = 0;
}
if (ice_is_xdp_ena_vsi(vsi))
/* return value check can be skipped here, it always returns
* 0 if reset is in progress
*/
ice_destroy_xdp_rings(vsi);
ice: Alloc queue management bitmaps and arrays dynamically The total number of queues available on the device is divided between multiple physical functions (PF) in the firmware and provided to the driver when it gets function capabilities from the firmware. Thus each PF knows how many Tx/Rx queues it has. These queues are then doled out to different VSIs (for LAN traffic, SR-IOV VF traffic, etc.) To track usage of these queues at the PF level, the driver uses two bitmaps avail_txqs and avail_rxqs. At the VSI level (i.e. struct ice_vsi instances) the driver uses two arrays txq_map and rxq_map, to track ownership of VSIs' queues in avail_txqs and avail_rxqs respectively. The aforementioned bitmaps and arrays should be allocated dynamically, because the number of queues supported by a PF is only available once function capabilities have been queried. The current static allocation consumes way more memory than required. This patch removes the DECLARE_BITMAP for avail_txqs and avail_rxqs and instead uses bitmap_zalloc to allocate the bitmaps during init. Similarly txq_map and rxq_map are now allocated in ice_vsi_alloc_arrays. As a result ICE_MAX_TXQS and ICE_MAX_RXQS defines are no longer needed. Also as txq_map and rxq_map are now allocated and freed, some code reordering was required in ice_vsi_rebuild for correct functioning. Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-08-02 16:25:21 +08:00
ice_vsi_put_qs(vsi);
ice_vsi_clear_rings(vsi);
ice_vsi_free_arrays(vsi);
ice_dev_onetime_setup(&pf->hw);
if (vsi->type == ICE_VSI_VF)
ice_vsi_set_num_qs(vsi, vf->vf_id);
else
ice_vsi_set_num_qs(vsi, ICE_INVAL_VFID);
ice: Alloc queue management bitmaps and arrays dynamically The total number of queues available on the device is divided between multiple physical functions (PF) in the firmware and provided to the driver when it gets function capabilities from the firmware. Thus each PF knows how many Tx/Rx queues it has. These queues are then doled out to different VSIs (for LAN traffic, SR-IOV VF traffic, etc.) To track usage of these queues at the PF level, the driver uses two bitmaps avail_txqs and avail_rxqs. At the VSI level (i.e. struct ice_vsi instances) the driver uses two arrays txq_map and rxq_map, to track ownership of VSIs' queues in avail_txqs and avail_rxqs respectively. The aforementioned bitmaps and arrays should be allocated dynamically, because the number of queues supported by a PF is only available once function capabilities have been queried. The current static allocation consumes way more memory than required. This patch removes the DECLARE_BITMAP for avail_txqs and avail_rxqs and instead uses bitmap_zalloc to allocate the bitmaps during init. Similarly txq_map and rxq_map are now allocated in ice_vsi_alloc_arrays. As a result ICE_MAX_TXQS and ICE_MAX_RXQS defines are no longer needed. Also as txq_map and rxq_map are now allocated and freed, some code reordering was required in ice_vsi_rebuild for correct functioning. Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Tested-by: Andrew Bowers <andrewx.bowers@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
2019-08-02 16:25:21 +08:00
ret = ice_vsi_alloc_arrays(vsi);
if (ret < 0)
goto err_vsi;
ice_vsi_get_qs(vsi);
ice_vsi_set_tc_cfg(vsi);
/* Initialize VSI struct elements and create VSI in FW */
ret = ice_vsi_init(vsi, init_vsi);
if (ret < 0)
goto err_vsi;
switch (vsi->type) {
case ICE_VSI_PF:
ret = ice_vsi_alloc_q_vectors(vsi);
if (ret)
goto err_rings;
ret = ice_vsi_setup_vector_base(vsi);
if (ret)
goto err_vectors;
ret = ice_vsi_set_q_vectors_reg_idx(vsi);
if (ret)
goto err_vectors;
ret = ice_vsi_alloc_rings(vsi);
if (ret)
goto err_vectors;
ice_vsi_map_rings_to_vectors(vsi);
if (ice_is_xdp_ena_vsi(vsi)) {
vsi->num_xdp_txq = vsi->alloc_txq;
ret = ice_prepare_xdp_rings(vsi, vsi->xdp_prog);
if (ret)
goto err_vectors;
}
/* Do not exit if configuring RSS had an issue, at least
* receive traffic on first queue. Hence no need to capture
* return value
*/
if (test_bit(ICE_FLAG_RSS_ENA, pf->flags))
ice_vsi_cfg_rss_lut_key(vsi);
break;
case ICE_VSI_VF:
ret = ice_vsi_alloc_q_vectors(vsi);
if (ret)
goto err_rings;
ret = ice_vsi_set_q_vectors_reg_idx(vsi);
if (ret)
goto err_vectors;
ret = ice_vsi_alloc_rings(vsi);
if (ret)
goto err_vectors;
break;
default:
break;
}
/* configure VSI nodes based on number of queues and TC's */
for (i = 0; i < vsi->tc_cfg.numtc; i++) {
max_txqs[i] = vsi->alloc_txq;
if (ice_is_xdp_ena_vsi(vsi))
max_txqs[i] += vsi->num_xdp_txq;
}
status = ice_cfg_vsi_lan(vsi->port_info, vsi->idx, vsi->tc_cfg.ena_tc,
max_txqs);
if (status) {
dev_err(ice_pf_to_dev(pf),
"VSI %d failed lan queue config, error %d\n",
vsi->vsi_num, status);
if (init_vsi) {
ret = -EIO;
goto err_vectors;
} else {
return ice_schedule_reset(pf, ICE_RESET_PFR);
}
}
ice_vsi_rebuild_set_coalesce(vsi, coalesce, prev_num_q_vectors);
kfree(coalesce);
return 0;
err_vectors:
ice_vsi_free_q_vectors(vsi);
err_rings:
if (vsi->netdev) {
vsi->current_netdev_flags = 0;
unregister_netdev(vsi->netdev);
free_netdev(vsi->netdev);
vsi->netdev = NULL;
}
err_vsi:
ice_vsi_clear(vsi);
set_bit(__ICE_RESET_FAILED, pf->state);
kfree(coalesce);
return ret;
}
/**
* ice_is_reset_in_progress - check for a reset in progress
* @state: PF state field
*/
bool ice_is_reset_in_progress(unsigned long *state)
{
return test_bit(__ICE_RESET_OICR_RECV, state) ||
test_bit(__ICE_DCBNL_DEVRESET, state) ||
test_bit(__ICE_PFR_REQ, state) ||
test_bit(__ICE_CORER_REQ, state) ||
test_bit(__ICE_GLOBR_REQ, state);
}
#ifdef CONFIG_DCB
/**
* ice_vsi_update_q_map - update our copy of the VSI info with new queue map
* @vsi: VSI being configured
* @ctx: the context buffer returned from AQ VSI update command
*/
static void ice_vsi_update_q_map(struct ice_vsi *vsi, struct ice_vsi_ctx *ctx)
{
vsi->info.mapping_flags = ctx->info.mapping_flags;
memcpy(&vsi->info.q_mapping, &ctx->info.q_mapping,
sizeof(vsi->info.q_mapping));
memcpy(&vsi->info.tc_mapping, ctx->info.tc_mapping,
sizeof(vsi->info.tc_mapping));
}
/**
* ice_vsi_cfg_tc - Configure VSI Tx Sched for given TC map
* @vsi: VSI to be configured
* @ena_tc: TC bitmap
*
* VSI queues expected to be quiesced before calling this function
*/
int ice_vsi_cfg_tc(struct ice_vsi *vsi, u8 ena_tc)
{
u16 max_txqs[ICE_MAX_TRAFFIC_CLASS] = { 0 };
struct ice_vsi_ctx *ctx;
struct ice_pf *pf = vsi->back;
enum ice_status status;
struct device *dev;
int i, ret = 0;
u8 num_tc = 0;
dev = ice_pf_to_dev(pf);
ice_for_each_traffic_class(i) {
/* build bitmap of enabled TCs */
if (ena_tc & BIT(i))
num_tc++;
/* populate max_txqs per TC */
max_txqs[i] = vsi->alloc_txq;
}
vsi->tc_cfg.ena_tc = ena_tc;
vsi->tc_cfg.numtc = num_tc;
ctx = kzalloc(sizeof(*ctx), GFP_KERNEL);
if (!ctx)
return -ENOMEM;
ctx->vf_num = 0;
ctx->info = vsi->info;
ice_vsi_setup_q_map(vsi, ctx);
/* must to indicate which section of VSI context are being modified */
ctx->info.valid_sections = cpu_to_le16(ICE_AQ_VSI_PROP_RXQ_MAP_VALID);
status = ice_update_vsi(&pf->hw, vsi->idx, ctx, NULL);
if (status) {
dev_info(dev, "Failed VSI Update\n");
ret = -EIO;
goto out;
}
status = ice_cfg_vsi_lan(vsi->port_info, vsi->idx, vsi->tc_cfg.ena_tc,
max_txqs);
if (status) {
dev_err(dev, "VSI %d failed TC config, error %d\n",
vsi->vsi_num, status);
ret = -EIO;
goto out;
}
ice_vsi_update_q_map(vsi, ctx);
vsi->info.valid_sections = 0;
ice_vsi_cfg_netdev_tc(vsi, ena_tc);
out:
kfree(ctx);
return ret;
}
#endif /* CONFIG_DCB */
/**
* ice_nvm_version_str - format the NVM version strings
* @hw: ptr to the hardware info
*/
char *ice_nvm_version_str(struct ice_hw *hw)
{
u8 oem_ver, oem_patch, ver_hi, ver_lo;
static char buf[ICE_NVM_VER_LEN];
u16 oem_build;
ice_get_nvm_version(hw, &oem_ver, &oem_build, &oem_patch, &ver_hi,
&ver_lo);
snprintf(buf, sizeof(buf), "%x.%02x 0x%x %d.%d.%d", ver_hi, ver_lo,
hw->nvm.eetrack, oem_ver, oem_build, oem_patch);
return buf;
}
/**
* ice_update_ring_stats - Update ring statistics
* @ring: ring to update
* @cont: used to increment per-vector counters
* @pkts: number of processed packets
* @bytes: number of processed bytes
*
* This function assumes that caller has acquired a u64_stats_sync lock.
*/
static void
ice_update_ring_stats(struct ice_ring *ring, struct ice_ring_container *cont,
u64 pkts, u64 bytes)
{
ring->stats.bytes += bytes;
ring->stats.pkts += pkts;
cont->total_bytes += bytes;
cont->total_pkts += pkts;
}
/**
* ice_update_tx_ring_stats - Update Tx ring specific counters
* @tx_ring: ring to update
* @pkts: number of processed packets
* @bytes: number of processed bytes
*/
void ice_update_tx_ring_stats(struct ice_ring *tx_ring, u64 pkts, u64 bytes)
{
u64_stats_update_begin(&tx_ring->syncp);
ice_update_ring_stats(tx_ring, &tx_ring->q_vector->tx, pkts, bytes);
u64_stats_update_end(&tx_ring->syncp);
}
/**
* ice_update_rx_ring_stats - Update Rx ring specific counters
* @rx_ring: ring to update
* @pkts: number of processed packets
* @bytes: number of processed bytes
*/
void ice_update_rx_ring_stats(struct ice_ring *rx_ring, u64 pkts, u64 bytes)
{
u64_stats_update_begin(&rx_ring->syncp);
ice_update_ring_stats(rx_ring, &rx_ring->q_vector->rx, pkts, bytes);
u64_stats_update_end(&rx_ring->syncp);
}
/**
* ice_vsi_cfg_mac_fltr - Add or remove a MAC address filter for a VSI
* @vsi: the VSI being configured MAC filter
* @macaddr: the MAC address to be added.
* @set: Add or delete a MAC filter
*
* Adds or removes MAC address filter entry for VF VSI
*/
enum ice_status
ice_vsi_cfg_mac_fltr(struct ice_vsi *vsi, const u8 *macaddr, bool set)
{
LIST_HEAD(tmp_add_list);
enum ice_status status;
/* Update MAC filter list to be added or removed for a VSI */
if (ice_add_mac_to_list(vsi, &tmp_add_list, macaddr)) {
status = ICE_ERR_NO_MEMORY;
goto cfg_mac_fltr_exit;
}
if (set)
status = ice_add_mac(&vsi->back->hw, &tmp_add_list);
else
status = ice_remove_mac(&vsi->back->hw, &tmp_add_list);
cfg_mac_fltr_exit:
ice_free_fltr_list(&vsi->back->pdev->dev, &tmp_add_list);
return status;
}
/**
* ice_is_dflt_vsi_in_use - check if the default forwarding VSI is being used
* @sw: switch to check if its default forwarding VSI is free
*
* Return true if the default forwarding VSI is already being used, else returns
* false signalling that it's available to use.
*/
bool ice_is_dflt_vsi_in_use(struct ice_sw *sw)
{
return (sw->dflt_vsi && sw->dflt_vsi_ena);
}
/**
* ice_is_vsi_dflt_vsi - check if the VSI passed in is the default VSI
* @sw: switch for the default forwarding VSI to compare against
* @vsi: VSI to compare against default forwarding VSI
*
* If this VSI passed in is the default forwarding VSI then return true, else
* return false
*/
bool ice_is_vsi_dflt_vsi(struct ice_sw *sw, struct ice_vsi *vsi)
{
return (sw->dflt_vsi == vsi && sw->dflt_vsi_ena);
}
/**
* ice_set_dflt_vsi - set the default forwarding VSI
* @sw: switch used to assign the default forwarding VSI
* @vsi: VSI getting set as the default forwarding VSI on the switch
*
* If the VSI passed in is already the default VSI and it's enabled just return
* success.
*
* If there is already a default VSI on the switch and it's enabled then return
* -EEXIST since there can only be one default VSI per switch.
*
* Otherwise try to set the VSI passed in as the switch's default VSI and
* return the result.
*/
int ice_set_dflt_vsi(struct ice_sw *sw, struct ice_vsi *vsi)
{
enum ice_status status;
struct device *dev;
if (!sw || !vsi)
return -EINVAL;
dev = ice_pf_to_dev(vsi->back);
/* the VSI passed in is already the default VSI */
if (ice_is_vsi_dflt_vsi(sw, vsi)) {
dev_dbg(dev, "VSI %d passed in is already the default forwarding VSI, nothing to do\n",
vsi->vsi_num);
return 0;
}
/* another VSI is already the default VSI for this switch */
if (ice_is_dflt_vsi_in_use(sw)) {
dev_err(dev,
"Default forwarding VSI %d already in use, disable it and try again\n",
sw->dflt_vsi->vsi_num);
return -EEXIST;
}
status = ice_cfg_dflt_vsi(&vsi->back->hw, vsi->idx, true, ICE_FLTR_RX);
if (status) {
dev_err(dev,
"Failed to set VSI %d as the default forwarding VSI, error %d\n",
vsi->vsi_num, status);
return -EIO;
}
sw->dflt_vsi = vsi;
sw->dflt_vsi_ena = true;
return 0;
}
/**
* ice_clear_dflt_vsi - clear the default forwarding VSI
* @sw: switch used to clear the default VSI
*
* If the switch has no default VSI or it's not enabled then return error.
*
* Otherwise try to clear the default VSI and return the result.
*/
int ice_clear_dflt_vsi(struct ice_sw *sw)
{
struct ice_vsi *dflt_vsi;
enum ice_status status;
struct device *dev;
if (!sw)
return -EINVAL;
dev = ice_pf_to_dev(sw->pf);
dflt_vsi = sw->dflt_vsi;
/* there is no default VSI configured */
if (!ice_is_dflt_vsi_in_use(sw))
return -ENODEV;
status = ice_cfg_dflt_vsi(&dflt_vsi->back->hw, dflt_vsi->idx, false,
ICE_FLTR_RX);
if (status) {
dev_err(dev,
"Failed to clear the default forwarding VSI %d, error %d\n",
dflt_vsi->vsi_num, status);
return -EIO;
}
sw->dflt_vsi = NULL;
sw->dflt_vsi_ena = false;
return 0;
}