Commit c83ae452 authored by David S. Miller's avatar David S. Miller

Merge branch '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/next-queue

Tony Nguyen says:

====================
ice: support dynamic interrupt allocation

Piotr Raczynski says:

This patchset reimplements MSIX interrupt allocation logic to allow dynamic
interrupt allocation after MSIX has been initially enabled. This allows
current and future features to allocate and free interrupts as needed and
will help to drastically decrease number of initially preallocated
interrupts (even down to the API hard limit of 1). Although this patchset
does not change behavior in terms of actual number of allocated interrupts
during probe, it will be subject to change.

First few patches prepares to introduce dynamic allocation by moving
interrupt allocation code to separate file and update allocation API used
in the driver to the currently preferred one.

Due to the current contract between ice and irdma driver which is directly
accessing msix entries allocated by ice driver, even after moving away from
older pci_enable_msix_range function, still keep msix_entries array for
irdma use.

Next patches refactors and removes redundant code from SRIOV related logic
as it also make it easier to move away from static allocation scheme.

Last patches actually enables dynamic allocation of MSIX interrupts. First,
introduce functions to allocate and free interrupts individually. This sets
ground for the rest of the changes even if that patch still allocates the
interrupts from the preallocated pool. Since this patch starts to keep
interrupt details in ice_q_vector structure we can get rid of functions
that calculates base vector number and register offset for the interrupt
as it is equal to the interrupt index. Only keep separate register offset
functions for the VF VSIs.

Next, replace homegrown interrupt tracker with much simpler xarray based
approach. As new API always allocate interrupts one by one, also track
interrupts in the same manner.

Lastly, extend the interrupt tracker to deal both with preallocated and
dynamically allocated vectors and use pci_msix_alloc_irq_at and
pci_msix_free_irq functions. Since not all architecture supports dynamic
allocation, check it before trying to allocate a new interrupt.

As previously mentioned, this patchset does not change number of initially
allocated interrupts during init phase but now it can and will likely be
changed.

Patch 1-3 -> move code around and use newer API
Patch 4-5 -> refactor and remove redundant SRIOV code
Patch 6   -> allocate every interrupt individually
Patch 7   -> replace homegrown interrupt tracker with xarray
Patch 8   -> allow dynamic interrupt allocation
---
v2:
Patch 4
 - simplify ice_vsi_setup_vector_base and account for num_avail_sw_msix
Patch 8
 - prevent q_vector leak in case vf ctrl VSI error

v1: https://lore.kernel.org/netdev/20230509170048.2235678-1-anthony.l.nguyen@intel.com/
====================
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
parents 030d71fd 011670cc
......@@ -18,6 +18,7 @@ ice-y := ice_main.o \
ice_txrx_lib.o \
ice_txrx.o \
ice_fltr.o \
ice_irq.o \
ice_pf_vsi_vlan_ops.o \
ice_vsi_vlan_ops.o \
ice_vsi_vlan_lib.o \
......
......@@ -74,6 +74,7 @@
#include "ice_lag.h"
#include "ice_vsi_vlan_ops.h"
#include "ice_gnss.h"
#include "ice_irq.h"
#define ICE_BAR0 0
#define ICE_REQ_DESC_MULTIPLE 32
......@@ -103,11 +104,6 @@
#define ICE_Q_WAIT_RETRY_LIMIT 10
#define ICE_Q_WAIT_MAX_RETRY (5 * ICE_Q_WAIT_RETRY_LIMIT)
#define ICE_MAX_LG_RSS_QS 256
#define ICE_RES_VALID_BIT 0x8000
#define ICE_RES_MISC_VEC_ID (ICE_RES_VALID_BIT - 1)
#define ICE_RES_RDMA_VEC_ID (ICE_RES_MISC_VEC_ID - 1)
/* All VF control VSIs share the same IRQ, so assign a unique ID for them */
#define ICE_RES_VF_CTRL_VEC_ID (ICE_RES_RDMA_VEC_ID - 1)
#define ICE_INVAL_Q_INDEX 0xffff
#define ICE_MAX_RXQS_PER_TC 256 /* Used when setting VSI context per TC Rx queues */
......@@ -245,12 +241,6 @@ struct ice_tc_cfg {
struct ice_tc_info tc_info[ICE_MAX_TRAFFIC_CLASS];
};
struct ice_res_tracker {
u16 num_entries;
u16 end;
u16 list[];
};
struct ice_qs_cfg {
struct mutex *qs_mutex; /* will be assigned to &pf->avail_q_mutex */
unsigned long *pf_map;
......@@ -348,7 +338,9 @@ struct ice_vsi {
u32 rx_buf_failed;
u32 rx_page_failed;
u16 num_q_vectors;
u16 base_vector; /* IRQ base for OS reserved vectors */
/* tell if only dynamic irq allocation is allowed */
bool irq_dyn_alloc;
enum ice_vsi_type type;
u16 vsi_num; /* HW (absolute) index of this VSI */
u16 idx; /* software index in pf->vsi[] */
......@@ -479,6 +471,7 @@ struct ice_q_vector {
char name[ICE_INT_NAME_STR_LEN];
u16 total_events; /* net_dim(): number of interrupts processed */
struct msi_map irq;
} ____cacheline_internodealigned_in_smp;
enum ice_pf_flags {
......@@ -539,7 +532,7 @@ struct ice_pf {
/* OS reserved IRQ details */
struct msix_entry *msix_entries;
struct ice_res_tracker *irq_tracker;
struct ice_irq_tracker irq_tracker;
/* First MSIX vector used by SR-IOV VFs. Calculated by subtracting the
* number of MSIX vectors needed for all SR-IOV VFs from the number of
* MSIX vectors allowed on this PF.
......@@ -583,8 +576,7 @@ struct ice_pf {
u32 hw_csum_rx_error;
u32 oicr_err_reg;
u16 oicr_idx; /* Other interrupt cause MSIX vector index */
u16 num_avail_sw_msix; /* remaining MSIX SW vectors left unclaimed */
struct msi_map oicr_irq; /* Other interrupt cause MSIX vector */
u16 max_pf_txqs; /* Total Tx queues PF wide */
u16 max_pf_rxqs; /* Total Rx queues PF wide */
u16 num_lan_msix; /* Total MSIX vectors for base driver */
......@@ -670,7 +662,7 @@ ice_irq_dynamic_ena(struct ice_hw *hw, struct ice_vsi *vsi,
struct ice_q_vector *q_vector)
{
u32 vector = (vsi && q_vector) ? q_vector->reg_idx :
((struct ice_pf *)hw->back)->oicr_idx;
((struct ice_pf *)hw->back)->oicr_irq.index;
int itr = ICE_ITR_NONE;
u32 val;
......
......@@ -596,7 +596,7 @@ int ice_set_cpu_rx_rmap(struct ice_vsi *vsi)
{
struct net_device *netdev;
struct ice_pf *pf;
int base_idx, i;
int i;
if (!vsi || vsi->type != ICE_VSI_PF)
return 0;
......@@ -613,10 +613,9 @@ int ice_set_cpu_rx_rmap(struct ice_vsi *vsi)
if (unlikely(!netdev->rx_cpu_rmap))
return -EINVAL;
base_idx = vsi->base_vector;
ice_for_each_q_vector(vsi, i)
if (irq_cpu_rmap_add(netdev->rx_cpu_rmap,
pf->msix_entries[base_idx + i].vector)) {
vsi->q_vectors[i]->irq.virq)) {
ice_free_cpu_rx_rmap(vsi);
return -EINVAL;
}
......
......@@ -103,10 +103,10 @@ static int ice_vsi_alloc_q_vector(struct ice_vsi *vsi, u16 v_idx)
{
struct ice_pf *pf = vsi->back;
struct ice_q_vector *q_vector;
int err;
/* allocate q_vector */
q_vector = devm_kzalloc(ice_pf_to_dev(pf), sizeof(*q_vector),
GFP_KERNEL);
q_vector = kzalloc(sizeof(*q_vector), GFP_KERNEL);
if (!q_vector)
return -ENOMEM;
......@@ -118,9 +118,34 @@ static int ice_vsi_alloc_q_vector(struct ice_vsi *vsi, u16 v_idx)
q_vector->rx.itr_mode = ITR_DYNAMIC;
q_vector->tx.type = ICE_TX_CONTAINER;
q_vector->rx.type = ICE_RX_CONTAINER;
q_vector->irq.index = -ENOENT;
if (vsi->type == ICE_VSI_VF)
if (vsi->type == ICE_VSI_VF) {
q_vector->reg_idx = ice_calc_vf_reg_idx(vsi->vf, q_vector);
goto out;
} else if (vsi->type == ICE_VSI_CTRL && vsi->vf) {
struct ice_vsi *ctrl_vsi = ice_get_vf_ctrl_vsi(pf, vsi);
if (ctrl_vsi) {
if (unlikely(!ctrl_vsi->q_vectors)) {
err = -ENOENT;
goto err_free_q_vector;
}
q_vector->irq = ctrl_vsi->q_vectors[0]->irq;
goto skip_alloc;
}
}
q_vector->irq = ice_alloc_irq(pf, vsi->irq_dyn_alloc);
if (q_vector->irq.index < 0) {
err = -ENOMEM;
goto err_free_q_vector;
}
skip_alloc:
q_vector->reg_idx = q_vector->irq.index;
/* only set affinity_mask if the CPU is online */
if (cpu_online(v_idx))
cpumask_set_cpu(v_idx, &q_vector->affinity_mask);
......@@ -137,6 +162,11 @@ static int ice_vsi_alloc_q_vector(struct ice_vsi *vsi, u16 v_idx)
vsi->q_vectors[v_idx] = q_vector;
return 0;
err_free_q_vector:
kfree(q_vector);
return err;
}
/**
......@@ -168,7 +198,19 @@ static void ice_free_q_vector(struct ice_vsi *vsi, int v_idx)
if (vsi->netdev)
netif_napi_del(&q_vector->napi);
devm_kfree(dev, q_vector);
/* release MSIX interrupt if q_vector had interrupt allocated */
if (q_vector->irq.index < 0)
goto free_q_vector;
/* only free last VF ctrl vsi interrupt */
if (vsi->type == ICE_VSI_CTRL && vsi->vf &&
ice_get_vf_ctrl_vsi(pf, vsi))
goto free_q_vector;
ice_free_irq(pf, q_vector->irq);
free_q_vector:
kfree(q_vector);
vsi->q_vectors[v_idx] = NULL;
}
......
......@@ -956,7 +956,7 @@ static u64 ice_intr_test(struct net_device *netdev)
netdev_info(netdev, "interrupt test\n");
wr32(&pf->hw, GLINT_DYN_CTL(pf->oicr_idx),
wr32(&pf->hw, GLINT_DYN_CTL(pf->oicr_irq.index),
GLINT_DYN_CTL_SW_ITR_INDX_M |
GLINT_DYN_CTL_INTENA_MSK_M |
GLINT_DYN_CTL_SWINT_TRIG_M);
......
......@@ -229,20 +229,34 @@ void ice_get_qos_params(struct ice_pf *pf, struct iidc_qos_params *qos)
EXPORT_SYMBOL_GPL(ice_get_qos_params);
/**
* ice_reserve_rdma_qvector - Reserve vector resources for RDMA driver
* ice_alloc_rdma_qvectors - Allocate vector resources for RDMA driver
* @pf: board private structure to initialize
*/
static int ice_reserve_rdma_qvector(struct ice_pf *pf)
static int ice_alloc_rdma_qvectors(struct ice_pf *pf)
{
if (ice_is_rdma_ena(pf)) {
int index;
index = ice_get_res(pf, pf->irq_tracker, pf->num_rdma_msix,
ICE_RES_RDMA_VEC_ID);
if (index < 0)
return index;
pf->num_avail_sw_msix -= pf->num_rdma_msix;
pf->rdma_base_vector = (u16)index;
int i;
pf->msix_entries = kcalloc(pf->num_rdma_msix,
sizeof(*pf->msix_entries),
GFP_KERNEL);
if (!pf->msix_entries)
return -ENOMEM;
/* RDMA is the only user of pf->msix_entries array */
pf->rdma_base_vector = 0;
for (i = 0; i < pf->num_rdma_msix; i++) {
struct msix_entry *entry = &pf->msix_entries[i];
struct msi_map map;
map = ice_alloc_irq(pf, false);
if (map.index < 0)
break;
entry->entry = map.index;
entry->vector = map.virq;
}
}
return 0;
}
......@@ -253,9 +267,21 @@ static int ice_reserve_rdma_qvector(struct ice_pf *pf)
*/
static void ice_free_rdma_qvector(struct ice_pf *pf)
{
pf->num_avail_sw_msix -= pf->num_rdma_msix;
ice_free_res(pf->irq_tracker, pf->rdma_base_vector,
ICE_RES_RDMA_VEC_ID);
int i;
if (!pf->msix_entries)
return;
for (i = 0; i < pf->num_rdma_msix; i++) {
struct msi_map map;
map.index = pf->msix_entries[i].entry;
map.virq = pf->msix_entries[i].vector;
ice_free_irq(pf, map);
}
kfree(pf->msix_entries);
pf->msix_entries = NULL;
}
/**
......@@ -357,7 +383,7 @@ int ice_init_rdma(struct ice_pf *pf)
}
/* Reserve vector resources */
ret = ice_reserve_rdma_qvector(pf);
ret = ice_alloc_rdma_qvectors(pf);
if (ret < 0) {
dev_err(dev, "failed to reserve vectors for RDMA\n");
goto err_reserve_rdma_qvector;
......
This diff is collapsed.
/* SPDX-License-Identifier: GPL-2.0 */
/* Copyright (C) 2023, Intel Corporation. */
#ifndef _ICE_IRQ_H_
#define _ICE_IRQ_H_
struct ice_irq_entry {
unsigned int index;
bool dynamic; /* allocation type flag */
};
struct ice_irq_tracker {
struct xarray entries;
u16 num_entries; /* total vectors available */
u16 num_static; /* preallocated entries */
};
int ice_init_interrupt_scheme(struct ice_pf *pf);
void ice_clear_interrupt_scheme(struct ice_pf *pf);
struct msi_map ice_alloc_irq(struct ice_pf *pf, bool dyn_only);
void ice_free_irq(struct ice_pf *pf, struct msi_map map);
int ice_get_max_used_msix_vector(struct ice_pf *pf);
#endif
......@@ -1370,190 +1370,6 @@ static int ice_vsi_init(struct ice_vsi *vsi, u32 vsi_flags)
return ret;
}
/**
* 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;
if (!res || index >= res->end)
return -EINVAL;
id |= ICE_RES_VALID_BIT;
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)
{
u16 start = 0, end = 0;
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;
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;
}
} while (end < res->end);
return -ENOMEM;
}
/**
* ice_get_free_res_count - Get free count from a resource tracker
* @res: Resource tracker instance
*/
static u16 ice_get_free_res_count(struct ice_res_tracker *res)
{
u16 i, count = 0;
for (i = 0; i < res->end; i++)
if (!(res->list[i] & ICE_RES_VALID_BIT))
count++;
return count;
}
/**
* 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
*
* 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;
}
return ice_search_res(res, needed, id);
}
/**
* ice_get_vf_ctrl_res - Get VF control VSI resource
* @pf: pointer to the PF structure
* @vsi: the VSI to allocate a resource for
*
* Look up whether another VF has already allocated the control VSI resource.
* If so, re-use this resource so that we share it among all VFs.
*
* Otherwise, allocate the resource and return it.
*/
static int ice_get_vf_ctrl_res(struct ice_pf *pf, struct ice_vsi *vsi)
{
struct ice_vf *vf;
unsigned int bkt;
int base;
rcu_read_lock();
ice_for_each_vf_rcu(pf, bkt, vf) {
if (vf != vsi->vf && vf->ctrl_vsi_idx != ICE_NO_VSI) {
base = pf->vsi[vf->ctrl_vsi_idx]->base_vector;
rcu_read_unlock();
return base;
}
}
rcu_read_unlock();
return ice_get_res(pf, pf->irq_tracker, vsi->num_q_vectors,
ICE_RES_VF_CTRL_VEC_ID);
}
/**
* 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_def() 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;
u16 num_q_vectors;
int base;
dev = ice_pf_to_dev(pf);
/* SRIOV doesn't grab irq_tracker entries for each VSI */
if (vsi->type == ICE_VSI_VF)
return 0;
if (vsi->type == ICE_VSI_CHNL)
return 0;
if (vsi->base_vector) {
dev_dbg(dev, "VSI %d has non-zero base vector %d\n",
vsi->vsi_num, vsi->base_vector);
return -EEXIST;
}
num_q_vectors = vsi->num_q_vectors;
/* reserve slots from OS requested IRQs */
if (vsi->type == ICE_VSI_CTRL && vsi->vf) {
base = ice_get_vf_ctrl_res(pf, vsi);
} else {
base = ice_get_res(pf, pf->irq_tracker, num_q_vectors,
vsi->idx);
}
if (base < 0) {
dev_err(dev, "%d MSI-X interrupts available. %s %d failed to get %d MSI-X vectors\n",
ice_get_free_res_count(pf->irq_tracker),
ice_vsi_type_str(vsi->type), vsi->idx, num_q_vectors);
return -ENOENT;
}
vsi->base_vector = (u16)base;
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
......@@ -2409,50 +2225,6 @@ static void ice_vsi_set_tc_cfg(struct ice_vsi *vsi)
ice_vsi_set_dcb_tc_cfg(vsi);
}
/**
* 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(ice_pf_to_dev(vsi->back), "Failed to set reg_idx on q_vector %d VSI %d\n",
i, vsi->vsi_num);
goto clear_reg_idx;
}
if (vsi->type == ICE_VSI_VF) {
struct ice_vf *vf = vsi->vf;
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_cfg_sw_lldp - Config switch rules for LLDP packet handling
* @vsi: the VSI being configured
......@@ -2611,37 +2383,6 @@ static void ice_set_agg_vsi(struct ice_vsi *vsi)
vsi->agg_node->num_vsis);
}
/**
* ice_free_vf_ctrl_res - Free the VF control VSI resource
* @pf: pointer to PF structure
* @vsi: the VSI to free resources for
*
* Check if the VF control VSI resource is still in use. If no VF is using it
* any more, release the VSI resource. Otherwise, leave it to be cleaned up
* once no other VF uses it.
*/
static void ice_free_vf_ctrl_res(struct ice_pf *pf, struct ice_vsi *vsi)
{
struct ice_vf *vf;
unsigned int bkt;
rcu_read_lock();
ice_for_each_vf_rcu(pf, bkt, vf) {
if (vf != vsi->vf && vf->ctrl_vsi_idx != ICE_NO_VSI) {
rcu_read_unlock();
return;
}
}
rcu_read_unlock();
/* No other VFs left that have control VSI. It is now safe to reclaim
* SW interrupts back to the common pool.
*/
ice_free_res(pf->irq_tracker, vsi->base_vector,
ICE_RES_VF_CTRL_VEC_ID);
pf->num_avail_sw_msix += vsi->num_q_vectors;
}
static int ice_vsi_cfg_tc_lan(struct ice_pf *pf, struct ice_vsi *vsi)
{
u16 max_txqs[ICE_MAX_TRAFFIC_CLASS] = { 0 };
......@@ -2728,14 +2469,6 @@ ice_vsi_cfg_def(struct ice_vsi *vsi, struct ice_vsi_cfg_params *params)
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;
......@@ -2786,10 +2519,6 @@ ice_vsi_cfg_def(struct ice_vsi *vsi, struct ice_vsi_cfg_params *params)
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_ring_stats(vsi);
if (ret)
goto unroll_vector_base;
......@@ -2822,8 +2551,6 @@ ice_vsi_cfg_def(struct ice_vsi *vsi, struct ice_vsi_cfg_params *params)
unroll_vector_base:
/* reclaim SW interrupts back to the common pool */
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:
......@@ -2915,14 +2642,6 @@ void ice_vsi_decfg(struct ice_vsi *vsi)
* many interrupts each VF needs. SR-IOV MSIX resources are also
* cleared in the same manner.
*/
if (vsi->type == ICE_VSI_CTRL && vsi->vf) {
ice_free_vf_ctrl_res(pf, vsi);
} else if (vsi->type != ICE_VSI_VF) {
/* reclaim SW interrupts back to the common pool */
ice_free_res(pf->irq_tracker, vsi->base_vector, vsi->idx);
pf->num_avail_sw_msix += vsi->num_q_vectors;
vsi->base_vector = 0;
}
if (vsi->type == ICE_VSI_VF &&
vsi->agg_node && vsi->agg_node->valid)
......@@ -3039,7 +2758,6 @@ static void ice_vsi_release_msix(struct ice_vsi *vsi)
void ice_vsi_free_irq(struct ice_vsi *vsi)
{
struct ice_pf *pf = vsi->back;
int base = vsi->base_vector;
int i;
if (!vsi->q_vectors || !vsi->irqs_ready)
......@@ -3053,10 +2771,9 @@ void ice_vsi_free_irq(struct ice_vsi *vsi)
ice_free_cpu_rx_rmap(vsi);
ice_for_each_q_vector(vsi, i) {
u16 vector = i + base;
int irq_num;
irq_num = pf->msix_entries[vector].vector;
irq_num = vsi->q_vectors[i]->irq.virq;
/* free only the irqs that were actually requested */
if (!vsi->q_vectors[i] ||
......@@ -3188,7 +2905,6 @@ void ice_dis_vsi(struct ice_vsi *vsi, bool locked)
*/
void ice_vsi_dis_irq(struct ice_vsi *vsi)
{
int base = vsi->base_vector;
struct ice_pf *pf = vsi->back;
struct ice_hw *hw = &pf->hw;
u32 val;
......@@ -3235,7 +2951,7 @@ void ice_vsi_dis_irq(struct ice_vsi *vsi)
return;
ice_for_each_q_vector(vsi, i)
synchronize_irq(pf->msix_entries[i + base].vector);
synchronize_irq(vsi->q_vectors[i]->irq.virq);
}
/**
......
......@@ -104,11 +104,6 @@ int ice_ena_vsi(struct ice_vsi *vsi, bool locked);
void ice_vsi_decfg(struct ice_vsi *vsi);
void ice_dis_vsi(struct ice_vsi *vsi, bool locked);
int ice_free_res(struct ice_res_tracker *res, u16 index, u16 id);
int
ice_get_res(struct ice_pf *pf, struct ice_res_tracker *res, u16 needed, u16 id);
int ice_vsi_rebuild(struct ice_vsi *vsi, u32 vsi_flags);
int ice_vsi_cfg(struct ice_vsi *vsi, struct ice_vsi_cfg_params *params);
......
......@@ -2490,7 +2490,6 @@ static int ice_vsi_req_irq_msix(struct ice_vsi *vsi, char *basename)
{
int q_vectors = vsi->num_q_vectors;
struct ice_pf *pf = vsi->back;
int base = vsi->base_vector;
struct device *dev;
int rx_int_idx = 0;
int tx_int_idx = 0;
......@@ -2501,7 +2500,7 @@ static int ice_vsi_req_irq_msix(struct ice_vsi *vsi, char *basename)
for (vector = 0; vector < q_vectors; vector++) {
struct ice_q_vector *q_vector = vsi->q_vectors[vector];
irq_num = pf->msix_entries[base + vector].vector;
irq_num = q_vector->irq.virq;
if (q_vector->tx.tx_ring && q_vector->rx.rx_ring) {
snprintf(q_vector->name, sizeof(q_vector->name) - 1,
......@@ -2555,9 +2554,8 @@ static int ice_vsi_req_irq_msix(struct ice_vsi *vsi, char *basename)
return 0;
free_q_irqs:
while (vector) {
vector--;
irq_num = pf->msix_entries[base + vector].vector;
while (vector--) {
irq_num = vsi->q_vectors[vector]->irq.virq;
if (!IS_ENABLED(CONFIG_RFS_ACCEL))
irq_set_affinity_notifier(irq_num, NULL);
irq_set_affinity_hint(irq_num, NULL);
......@@ -3047,7 +3045,7 @@ static void ice_ena_misc_vector(struct ice_pf *pf)
wr32(hw, PFINT_OICR_ENA, val);
/* SW_ITR_IDX = 0, but don't change INTENA */
wr32(hw, GLINT_DYN_CTL(pf->oicr_idx),
wr32(hw, GLINT_DYN_CTL(pf->oicr_irq.index),
GLINT_DYN_CTL_SW_ITR_INDX_M | GLINT_DYN_CTL_INTENA_MSK_M);
}
......@@ -3234,6 +3232,7 @@ static void ice_dis_ctrlq_interrupts(struct ice_hw *hw)
*/
static void ice_free_irq_msix_misc(struct ice_pf *pf)
{
int misc_irq_num = pf->oicr_irq.virq;
struct ice_hw *hw = &pf->hw;
ice_dis_ctrlq_interrupts(hw);
......@@ -3242,14 +3241,10 @@ static void ice_free_irq_msix_misc(struct ice_pf *pf)
wr32(hw, PFINT_OICR_ENA, 0);
ice_flush(hw);
if (pf->msix_entries) {
synchronize_irq(pf->msix_entries[pf->oicr_idx].vector);
devm_free_irq(ice_pf_to_dev(pf),
pf->msix_entries[pf->oicr_idx].vector, pf);
}
synchronize_irq(misc_irq_num);
devm_free_irq(ice_pf_to_dev(pf), misc_irq_num, pf);
pf->num_avail_sw_msix += 1;
ice_free_res(pf->irq_tracker, pf->oicr_idx, ICE_RES_MISC_VEC_ID);
ice_free_irq(pf, pf->oicr_irq);
}
/**
......@@ -3295,7 +3290,8 @@ static int ice_req_irq_msix_misc(struct ice_pf *pf)
{
struct device *dev = ice_pf_to_dev(pf);
struct ice_hw *hw = &pf->hw;
int oicr_idx, err = 0;
struct msi_map oicr_irq;
int err = 0;
if (!pf->int_name[0])
snprintf(pf->int_name, sizeof(pf->int_name) - 1, "%s-%s:misc",
......@@ -3309,30 +3305,26 @@ static int ice_req_irq_msix_misc(struct ice_pf *pf)
goto skip_req_irq;
/* reserve one vector in irq_tracker for misc interrupts */
oicr_idx = ice_get_res(pf, pf->irq_tracker, 1, ICE_RES_MISC_VEC_ID);
if (oicr_idx < 0)
return oicr_idx;
pf->num_avail_sw_msix -= 1;
pf->oicr_idx = (u16)oicr_idx;
err = devm_request_threaded_irq(dev,
pf->msix_entries[pf->oicr_idx].vector,
ice_misc_intr, ice_misc_intr_thread_fn,
0, pf->int_name, pf);
oicr_irq = ice_alloc_irq(pf, false);
if (oicr_irq.index < 0)
return oicr_irq.index;
pf->oicr_irq = oicr_irq;
err = devm_request_threaded_irq(dev, pf->oicr_irq.virq, ice_misc_intr,
ice_misc_intr_thread_fn, 0,
pf->int_name, pf);
if (err) {
dev_err(dev, "devm_request_threaded_irq for %s failed: %d\n",
pf->int_name, err);
ice_free_res(pf->irq_tracker, 1, ICE_RES_MISC_VEC_ID);
pf->num_avail_sw_msix += 1;
ice_free_irq(pf, pf->oicr_irq);
return err;
}
skip_req_irq:
ice_ena_misc_vector(pf);
ice_ena_ctrlq_interrupts(hw, pf->oicr_idx);
wr32(hw, GLINT_ITR(ICE_RX_ITR, pf->oicr_idx),
ice_ena_ctrlq_interrupts(hw, pf->oicr_irq.index);
wr32(hw, GLINT_ITR(ICE_RX_ITR, pf->oicr_irq.index),
ITR_REG_ALIGN(ICE_ITR_8K) >> ICE_ITR_GRAN_S);
ice_flush(hw);
......@@ -3900,224 +3892,6 @@ static int ice_init_pf(struct ice_pf *pf)
return 0;
}
/**
* ice_reduce_msix_usage - Reduce usage of MSI-X vectors
* @pf: board private structure
* @v_remain: number of remaining MSI-X vectors to be distributed
*
* Reduce the usage of MSI-X vectors when entire request cannot be fulfilled.
* pf->num_lan_msix and pf->num_rdma_msix values are set based on number of
* remaining vectors.
*/
static void ice_reduce_msix_usage(struct ice_pf *pf, int v_remain)
{
int v_rdma;
if (!ice_is_rdma_ena(pf)) {
pf->num_lan_msix = v_remain;
return;
}
/* RDMA needs at least 1 interrupt in addition to AEQ MSIX */
v_rdma = ICE_RDMA_NUM_AEQ_MSIX + 1;
if (v_remain < ICE_MIN_LAN_TXRX_MSIX + ICE_MIN_RDMA_MSIX) {
dev_warn(ice_pf_to_dev(pf), "Not enough MSI-X vectors to support RDMA.\n");
clear_bit(ICE_FLAG_RDMA_ENA, pf->flags);
pf->num_rdma_msix = 0;
pf->num_lan_msix = ICE_MIN_LAN_TXRX_MSIX;
} else if ((v_remain < ICE_MIN_LAN_TXRX_MSIX + v_rdma) ||
(v_remain - v_rdma < v_rdma)) {
/* Support minimum RDMA and give remaining vectors to LAN MSIX */
pf->num_rdma_msix = ICE_MIN_RDMA_MSIX;
pf->num_lan_msix = v_remain - ICE_MIN_RDMA_MSIX;
} else {
/* Split remaining MSIX with RDMA after accounting for AEQ MSIX
*/
pf->num_rdma_msix = (v_remain - ICE_RDMA_NUM_AEQ_MSIX) / 2 +
ICE_RDMA_NUM_AEQ_MSIX;
pf->num_lan_msix = v_remain - pf->num_rdma_msix;
}
}
/**
* ice_ena_msix_range - Request a range of MSIX vectors from the OS
* @pf: board private structure
*
* Compute the number of MSIX vectors wanted and request from the OS. Adjust
* device usage if there are not enough vectors. Return the number of vectors
* reserved or negative on failure.
*/
static int ice_ena_msix_range(struct ice_pf *pf)
{
int num_cpus, hw_num_msix, v_other, v_wanted, v_actual;
struct device *dev = ice_pf_to_dev(pf);
int err, i;
hw_num_msix = pf->hw.func_caps.common_cap.num_msix_vectors;
num_cpus = num_online_cpus();
/* LAN miscellaneous handler */
v_other = ICE_MIN_LAN_OICR_MSIX;
/* Flow Director */
if (test_bit(ICE_FLAG_FD_ENA, pf->flags))
v_other += ICE_FDIR_MSIX;
/* switchdev */
v_other += ICE_ESWITCH_MSIX;
v_wanted = v_other;
/* LAN traffic */
pf->num_lan_msix = num_cpus;
v_wanted += pf->num_lan_msix;
/* RDMA auxiliary driver */
if (ice_is_rdma_ena(pf)) {
pf->num_rdma_msix = num_cpus + ICE_RDMA_NUM_AEQ_MSIX;
v_wanted += pf->num_rdma_msix;
}
if (v_wanted > hw_num_msix) {
int v_remain;
dev_warn(dev, "not enough device MSI-X vectors. wanted = %d, available = %d\n",
v_wanted, hw_num_msix);
if (hw_num_msix < ICE_MIN_MSIX) {
err = -ERANGE;
goto exit_err;
}
v_remain = hw_num_msix - v_other;
if (v_remain < ICE_MIN_LAN_TXRX_MSIX) {
v_other = ICE_MIN_MSIX - ICE_MIN_LAN_TXRX_MSIX;
v_remain = ICE_MIN_LAN_TXRX_MSIX;
}
ice_reduce_msix_usage(pf, v_remain);
v_wanted = pf->num_lan_msix + pf->num_rdma_msix + v_other;
dev_notice(dev, "Reducing request to %d MSI-X vectors for LAN traffic.\n",
pf->num_lan_msix);
if (ice_is_rdma_ena(pf))
dev_notice(dev, "Reducing request to %d MSI-X vectors for RDMA.\n",
pf->num_rdma_msix);
}
pf->msix_entries = devm_kcalloc(dev, v_wanted,
sizeof(*pf->msix_entries), GFP_KERNEL);
if (!pf->msix_entries) {
err = -ENOMEM;
goto exit_err;
}
for (i = 0; i < v_wanted; i++)
pf->msix_entries[i].entry = i;
/* actually reserve the vectors */
v_actual = pci_enable_msix_range(pf->pdev, pf->msix_entries,
ICE_MIN_MSIX, v_wanted);
if (v_actual < 0) {
dev_err(dev, "unable to reserve MSI-X vectors\n");
err = v_actual;
goto msix_err;
}
if (v_actual < v_wanted) {
dev_warn(dev, "not enough OS MSI-X vectors. requested = %d, obtained = %d\n",
v_wanted, v_actual);
if (v_actual < ICE_MIN_MSIX) {
/* error if we can't get minimum vectors */
pci_disable_msix(pf->pdev);
err = -ERANGE;
goto msix_err;
} else {
int v_remain = v_actual - v_other;
if (v_remain < ICE_MIN_LAN_TXRX_MSIX)
v_remain = ICE_MIN_LAN_TXRX_MSIX;
ice_reduce_msix_usage(pf, v_remain);
dev_notice(dev, "Enabled %d MSI-X vectors for LAN traffic.\n",
pf->num_lan_msix);
if (ice_is_rdma_ena(pf))
dev_notice(dev, "Enabled %d MSI-X vectors for RDMA.\n",
pf->num_rdma_msix);
}
}
return v_actual;
msix_err:
devm_kfree(dev, pf->msix_entries);
exit_err:
pf->num_rdma_msix = 0;
pf->num_lan_msix = 0;
return err;
}
/**
* ice_dis_msix - Disable MSI-X interrupt setup in OS
* @pf: board private structure
*/
static void ice_dis_msix(struct ice_pf *pf)
{
pci_disable_msix(pf->pdev);
devm_kfree(ice_pf_to_dev(pf), pf->msix_entries);
pf->msix_entries = NULL;
}
/**
* ice_clear_interrupt_scheme - Undo things done by ice_init_interrupt_scheme
* @pf: board private structure
*/
static void ice_clear_interrupt_scheme(struct ice_pf *pf)
{
ice_dis_msix(pf);
if (pf->irq_tracker) {
devm_kfree(ice_pf_to_dev(pf), pf->irq_tracker);
pf->irq_tracker = NULL;
}
}
/**
* ice_init_interrupt_scheme - Determine proper interrupt scheme
* @pf: board private structure to initialize
*/
static int ice_init_interrupt_scheme(struct ice_pf *pf)
{
int vectors;
vectors = ice_ena_msix_range(pf);
if (vectors < 0)
return vectors;
/* set up vector assignment tracking */
pf->irq_tracker = devm_kzalloc(ice_pf_to_dev(pf),
struct_size(pf->irq_tracker, list, vectors),
GFP_KERNEL);
if (!pf->irq_tracker) {
ice_dis_msix(pf);
return -ENOMEM;
}
/* populate SW interrupts pool with number of OS granted IRQs. */
pf->num_avail_sw_msix = (u16)vectors;
pf->irq_tracker->num_entries = (u16)vectors;
pf->irq_tracker->end = pf->irq_tracker->num_entries;
return 0;
}
/**
* ice_is_wol_supported - check if WoL is supported
* @hw: pointer to hardware info
......
......@@ -911,7 +911,7 @@ ice_ptp_release_tx_tracker(struct ice_pf *pf, struct ice_ptp_tx *tx)
spin_unlock(&tx->lock);
/* wait for potentially outstanding interrupt to complete */
synchronize_irq(pf->msix_entries[pf->oicr_idx].vector);
synchronize_irq(pf->oicr_irq.virq);
ice_ptp_flush_tx_tracker(pf, tx);
......
......@@ -135,18 +135,9 @@ static void ice_dis_vf_mappings(struct ice_vf *vf)
*/
static int ice_sriov_free_msix_res(struct ice_pf *pf)
{
struct ice_res_tracker *res;
if (!pf)
return -EINVAL;
res = pf->irq_tracker;
if (!res)
return -EINVAL;
/* give back irq_tracker resources used */
WARN_ON(pf->sriov_base_vector < res->num_entries);
pf->sriov_base_vector = 0;
return 0;
......@@ -409,29 +400,6 @@ int ice_calc_vf_reg_idx(struct ice_vf *vf, struct ice_q_vector *q_vector)
q_vector->v_idx + 1;
}
/**
* ice_get_max_valid_res_idx - Get the max valid resource index
* @res: pointer to the resource to find the max valid index for
*
* Start from the end of the ice_res_tracker and return right when we find the
* first res->list entry with the ICE_RES_VALID_BIT set. This function is only
* valid for SR-IOV because it is the only consumer that manipulates the
* res->end and this is always called when res->end is set to res->num_entries.
*/
static int ice_get_max_valid_res_idx(struct ice_res_tracker *res)
{
int i;
if (!res)
return -EINVAL;
for (i = res->num_entries - 1; i >= 0; i--)
if (res->list[i] & ICE_RES_VALID_BIT)
return i;
return 0;
}
/**
* ice_sriov_set_msix_res - Set any used MSIX resources
* @pf: pointer to PF structure
......@@ -450,7 +418,7 @@ static int ice_get_max_valid_res_idx(struct ice_res_tracker *res)
static int ice_sriov_set_msix_res(struct ice_pf *pf, u16 num_msix_needed)
{
u16 total_vectors = pf->hw.func_caps.common_cap.num_msix_vectors;
int vectors_used = pf->irq_tracker->num_entries;
int vectors_used = ice_get_max_used_msix_vector(pf);
int sriov_base_vector;
sriov_base_vector = total_vectors - num_msix_needed;
......@@ -490,7 +458,7 @@ static int ice_sriov_set_msix_res(struct ice_pf *pf, u16 num_msix_needed)
*/
static int ice_set_per_vf_res(struct ice_pf *pf, u16 num_vfs)
{
int max_valid_res_idx = ice_get_max_valid_res_idx(pf->irq_tracker);
int vectors_used = ice_get_max_used_msix_vector(pf);
u16 num_msix_per_vf, num_txq, num_rxq, avail_qs;
int msix_avail_per_vf, msix_avail_for_sriov;
struct device *dev = ice_pf_to_dev(pf);
......@@ -501,12 +469,9 @@ static int ice_set_per_vf_res(struct ice_pf *pf, u16 num_vfs)
if (!num_vfs)
return -EINVAL;
if (max_valid_res_idx < 0)
return -ENOSPC;
/* determine MSI-X resources per VF */
msix_avail_for_sriov = pf->hw.func_caps.common_cap.num_msix_vectors -
pf->irq_tracker->num_entries;
vectors_used;
msix_avail_per_vf = msix_avail_for_sriov / num_vfs;
if (msix_avail_per_vf >= ICE_NUM_VF_MSIX_MED) {
num_msix_per_vf = ICE_NUM_VF_MSIX_MED;
......@@ -871,7 +836,7 @@ static int ice_ena_vfs(struct ice_pf *pf, u16 num_vfs)
int ret;
/* Disable global interrupt 0 so we don't try to handle the VFLR. */
wr32(hw, GLINT_DYN_CTL(pf->oicr_idx),
wr32(hw, GLINT_DYN_CTL(pf->oicr_irq.index),
ICE_ITR_NONE << GLINT_DYN_CTL_ITR_INDX_S);
set_bit(ICE_OICR_INTR_DIS, pf->state);
ice_flush(hw);
......
......@@ -1310,3 +1310,35 @@ void ice_vf_set_initialized(struct ice_vf *vf)
set_bit(ICE_VF_STATE_INIT, vf->vf_states);
memset(&vf->vlan_v2_caps, 0, sizeof(vf->vlan_v2_caps));
}
/**
* ice_get_vf_ctrl_vsi - Get first VF control VSI pointer
* @pf: the PF private structure
* @vsi: pointer to the VSI
*
* Return first found VF control VSI other than the vsi
* passed by parameter. This function is used to determine
* whether new resources have to be allocated for control VSI
* or they can be shared with existing one.
*
* Return found VF control VSI pointer other itself. Return
* NULL Otherwise.
*
*/
struct ice_vsi *ice_get_vf_ctrl_vsi(struct ice_pf *pf, struct ice_vsi *vsi)
{
struct ice_vsi *ctrl_vsi = NULL;
struct ice_vf *vf;
unsigned int bkt;
rcu_read_lock();
ice_for_each_vf_rcu(pf, bkt, vf) {
if (vf != vsi->vf && vf->ctrl_vsi_idx != ICE_NO_VSI) {
ctrl_vsi = pf->vsi[vf->ctrl_vsi_idx];
break;
}
}
rcu_read_unlock();
return ctrl_vsi;
}
......@@ -226,6 +226,7 @@ int
ice_vf_clear_vsi_promisc(struct ice_vf *vf, struct ice_vsi *vsi, u8 promisc_m);
int ice_reset_vf(struct ice_vf *vf, u32 flags);
void ice_reset_all_vfs(struct ice_pf *pf);
struct ice_vsi *ice_get_vf_ctrl_vsi(struct ice_pf *pf, struct ice_vsi *vsi);
#else /* CONFIG_PCI_IOV */
static inline struct ice_vf *ice_get_vf_by_id(struct ice_pf *pf, u16 vf_id)
{
......@@ -290,6 +291,12 @@ static inline int ice_reset_vf(struct ice_vf *vf, u32 flags)
static inline void ice_reset_all_vfs(struct ice_pf *pf)
{
}
static inline struct ice_vsi *
ice_get_vf_ctrl_vsi(struct ice_pf *pf, struct ice_vsi *vsi)
{
return NULL;
}
#endif /* !CONFIG_PCI_IOV */
#endif /* _ICE_VF_LIB_H_ */
......@@ -90,7 +90,6 @@ ice_qvec_dis_irq(struct ice_vsi *vsi, struct ice_rx_ring *rx_ring,
{
struct ice_pf *pf = vsi->back;
struct ice_hw *hw = &pf->hw;
int base = vsi->base_vector;
u16 reg;
u32 val;
......@@ -103,11 +102,9 @@ ice_qvec_dis_irq(struct ice_vsi *vsi, struct ice_rx_ring *rx_ring,
wr32(hw, QINT_RQCTL(reg), val);
if (q_vector) {
u16 v_idx = q_vector->v_idx;
wr32(hw, GLINT_DYN_CTL(q_vector->reg_idx), 0);
ice_flush(hw);
synchronize_irq(pf->msix_entries[v_idx + base].vector);
synchronize_irq(q_vector->irq.virq);
}
}
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment