iwlwifi: pcie: properly implement NAPI

Instead of pretending to have NAPI and then relying entirely on
interrupts anyway, properly implement NAPI and schedule the poll
when we get an interrupt, re-enabling the interrupt only after
the poll completed.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>
Link: https://lore.kernel.org/r/iwlwifi.20210117130510.a5951ac4fc06.I9c84a147288fcfb1b019572c6758f2d92949f5d7@changeid
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>
This commit is contained in:
Johannes Berg 2021-01-17 13:10:34 +02:00 committed by Luca Coelho
parent d4e3a341b8
commit 25edc8f259
5 changed files with 171 additions and 83 deletions

View File

@ -418,8 +418,7 @@ IWL_TRANS_GET_PCIE_TRANS(struct iwl_trans *trans)
return (void *)trans->trans_specific;
}
static inline void iwl_pcie_clear_irq(struct iwl_trans *trans,
struct msix_entry *entry)
static inline void iwl_pcie_clear_irq(struct iwl_trans *trans, int queue)
{
/*
* Before sending the interrupt the HW disables it to prevent
@ -429,7 +428,7 @@ static inline void iwl_pcie_clear_irq(struct iwl_trans *trans,
* write 1 clear (W1C) register, meaning that it's being clear
* by writing 1 to the bit.
*/
iwl_write32(trans, CSR_MSIX_AUTOMASK_ST_AD, BIT(entry->entry));
iwl_write32(trans, CSR_MSIX_AUTOMASK_ST_AD, BIT(queue));
}
static inline struct iwl_trans *
@ -462,7 +461,6 @@ int iwl_pcie_rx_stop(struct iwl_trans *trans);
void iwl_pcie_rx_free(struct iwl_trans *trans);
void iwl_pcie_free_rbs_pool(struct iwl_trans *trans);
void iwl_pcie_rx_init_rxb_lists(struct iwl_rxq *rxq);
int iwl_pcie_dummy_napi_poll(struct napi_struct *napi, int budget);
void iwl_pcie_rxq_alloc_rbs(struct iwl_trans *trans, gfp_t priority,
struct iwl_rxq *rxq);
@ -569,9 +567,9 @@ static inline void iwl_disable_interrupts(struct iwl_trans *trans)
{
struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
spin_lock(&trans_pcie->irq_lock);
spin_lock_bh(&trans_pcie->irq_lock);
_iwl_disable_interrupts(trans);
spin_unlock(&trans_pcie->irq_lock);
spin_unlock_bh(&trans_pcie->irq_lock);
}
static inline void _iwl_enable_interrupts(struct iwl_trans *trans)
@ -601,9 +599,9 @@ static inline void iwl_enable_interrupts(struct iwl_trans *trans)
{
struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
spin_lock(&trans_pcie->irq_lock);
spin_lock_bh(&trans_pcie->irq_lock);
_iwl_enable_interrupts(trans);
spin_unlock(&trans_pcie->irq_lock);
spin_unlock_bh(&trans_pcie->irq_lock);
}
static inline void iwl_enable_hw_int_msk_msix(struct iwl_trans *trans, u32 msk)
{

View File

@ -207,10 +207,10 @@ static void iwl_pcie_rxq_check_wrptr(struct iwl_trans *trans)
if (!rxq->need_update)
continue;
spin_lock(&rxq->lock);
spin_lock_bh(&rxq->lock);
iwl_pcie_rxq_inc_wr_ptr(trans, rxq);
rxq->need_update = false;
spin_unlock(&rxq->lock);
spin_unlock_bh(&rxq->lock);
}
}
@ -255,7 +255,7 @@ static void iwl_pcie_rxmq_restock(struct iwl_trans *trans,
if (!test_bit(STATUS_DEVICE_ENABLED, &trans->status))
return;
spin_lock(&rxq->lock);
spin_lock_bh(&rxq->lock);
while (rxq->free_count) {
/* Get next free Rx buffer, remove from free list */
rxb = list_first_entry(&rxq->rx_free, struct iwl_rx_mem_buffer,
@ -269,16 +269,16 @@ static void iwl_pcie_rxmq_restock(struct iwl_trans *trans,
rxq->write = (rxq->write + 1) & (rxq->queue_size - 1);
rxq->free_count--;
}
spin_unlock(&rxq->lock);
spin_unlock_bh(&rxq->lock);
/*
* If we've added more space for the firmware to place data, tell it.
* Increment device's write pointer in multiples of 8.
*/
if (rxq->write_actual != (rxq->write & ~0x7)) {
spin_lock(&rxq->lock);
spin_lock_bh(&rxq->lock);
iwl_pcie_rxq_inc_wr_ptr(trans, rxq);
spin_unlock(&rxq->lock);
spin_unlock_bh(&rxq->lock);
}
}
@ -514,10 +514,10 @@ static void iwl_pcie_rx_allocator(struct iwl_trans *trans)
IWL_DEBUG_TPT(trans, "Pending allocation requests = %d\n", pending);
/* If we were scheduled - there is at least one request */
spin_lock(&rba->lock);
spin_lock_bh(&rba->lock);
/* swap out the rba->rbd_empty to a local list */
list_replace_init(&rba->rbd_empty, &local_empty);
spin_unlock(&rba->lock);
spin_unlock_bh(&rba->lock);
while (pending) {
int i;
@ -577,21 +577,21 @@ static void iwl_pcie_rx_allocator(struct iwl_trans *trans)
pending);
}
spin_lock(&rba->lock);
spin_lock_bh(&rba->lock);
/* add the allocated rbds to the allocator allocated list */
list_splice_tail(&local_allocated, &rba->rbd_allocated);
/* get more empty RBDs for current pending requests */
list_splice_tail_init(&rba->rbd_empty, &local_empty);
spin_unlock(&rba->lock);
spin_unlock_bh(&rba->lock);
atomic_inc(&rba->req_ready);
}
spin_lock(&rba->lock);
spin_lock_bh(&rba->lock);
/* return unused rbds to the allocator empty list */
list_splice_tail(&local_empty, &rba->rbd_empty);
spin_unlock(&rba->lock);
spin_unlock_bh(&rba->lock);
IWL_DEBUG_TPT(trans, "%s, exit.\n", __func__);
}
@ -1008,10 +1008,76 @@ void iwl_pcie_rx_init_rxb_lists(struct iwl_rxq *rxq)
rxq->used_count = 0;
}
int iwl_pcie_dummy_napi_poll(struct napi_struct *napi, int budget)
static int iwl_pcie_rx_handle(struct iwl_trans *trans, int queue, int budget);
static int iwl_pcie_napi_poll(struct napi_struct *napi, int budget)
{
WARN_ON(1);
return 0;
struct iwl_rxq *rxq = container_of(napi, struct iwl_rxq, napi);
struct iwl_trans_pcie *trans_pcie;
struct iwl_trans *trans;
int ret;
trans_pcie = container_of(napi->dev, struct iwl_trans_pcie, napi_dev);
trans = trans_pcie->trans;
ret = iwl_pcie_rx_handle(trans, rxq->id, budget);
if (ret < budget) {
spin_lock(&trans_pcie->irq_lock);
if (test_bit(STATUS_INT_ENABLED, &trans->status))
_iwl_enable_interrupts(trans);
spin_unlock(&trans_pcie->irq_lock);
napi_complete_done(&rxq->napi, ret);
}
return ret;
}
static int iwl_pcie_napi_poll_msix(struct napi_struct *napi, int budget)
{
struct iwl_rxq *rxq = container_of(napi, struct iwl_rxq, napi);
struct iwl_trans_pcie *trans_pcie;
struct iwl_trans *trans;
int ret;
trans_pcie = container_of(napi->dev, struct iwl_trans_pcie, napi_dev);
trans = trans_pcie->trans;
ret = iwl_pcie_rx_handle(trans, rxq->id, budget);
if (ret < budget) {
spin_lock(&trans_pcie->irq_lock);
iwl_pcie_clear_irq(trans, rxq->id);
spin_unlock(&trans_pcie->irq_lock);
napi_complete_done(&rxq->napi, ret);
}
return ret;
}
static int iwl_pcie_napi_poll_msix_shared(struct napi_struct *napi, int budget)
{
struct iwl_rxq *rxq = container_of(napi, struct iwl_rxq, napi);
struct iwl_trans_pcie *trans_pcie;
struct iwl_trans *trans;
int ret;
trans_pcie = container_of(napi->dev, struct iwl_trans_pcie, napi_dev);
trans = trans_pcie->trans;
ret = iwl_pcie_rx_handle(trans, rxq->id, budget);
if (ret < budget) {
spin_lock(&trans_pcie->irq_lock);
iwl_pcie_clear_irq(trans, 0);
spin_unlock(&trans_pcie->irq_lock);
napi_complete_done(&rxq->napi, ret);
}
return ret;
}
static int _iwl_pcie_rx_init(struct iwl_trans *trans)
@ -1030,12 +1096,12 @@ static int _iwl_pcie_rx_init(struct iwl_trans *trans)
cancel_work_sync(&rba->rx_alloc);
spin_lock(&rba->lock);
spin_lock_bh(&rba->lock);
atomic_set(&rba->req_pending, 0);
atomic_set(&rba->req_ready, 0);
INIT_LIST_HEAD(&rba->rbd_allocated);
INIT_LIST_HEAD(&rba->rbd_empty);
spin_unlock(&rba->lock);
spin_unlock_bh(&rba->lock);
/* free all first - we might be reconfigured for a different size */
iwl_pcie_free_rbs_pool(trans);
@ -1062,9 +1128,25 @@ static int _iwl_pcie_rx_init(struct iwl_trans *trans)
iwl_pcie_rx_init_rxb_lists(rxq);
if (!rxq->napi.poll)
if (!rxq->napi.poll) {
int (*poll)(struct napi_struct *, int) = iwl_pcie_napi_poll;
if (trans_pcie->msix_enabled) {
poll = iwl_pcie_napi_poll_msix;
if (trans_pcie->shared_vec_mask & IWL_SHARED_IRQ_NON_RX &&
i == 0)
poll = iwl_pcie_napi_poll_msix_shared;
if (trans_pcie->shared_vec_mask & IWL_SHARED_IRQ_FIRST_RSS &&
i == 1)
poll = iwl_pcie_napi_poll_msix_shared;
}
netif_napi_add(&trans_pcie->napi_dev, &rxq->napi,
iwl_pcie_dummy_napi_poll, 64);
poll, NAPI_POLL_WEIGHT);
napi_enable(&rxq->napi);
}
spin_unlock(&rxq->lock);
}
@ -1163,8 +1245,10 @@ void iwl_pcie_rx_free(struct iwl_trans *trans)
iwl_pcie_free_rxq_dma(trans, rxq);
if (rxq->napi.poll)
if (rxq->napi.poll) {
napi_disable(&rxq->napi);
netif_napi_del(&rxq->napi);
}
}
kfree(trans_pcie->rx_pool);
kfree(trans_pcie->global_table);
@ -1417,16 +1501,15 @@ out_err:
/*
* iwl_pcie_rx_handle - Main entry function for receiving responses from fw
*/
static void iwl_pcie_rx_handle(struct iwl_trans *trans, int queue)
static int iwl_pcie_rx_handle(struct iwl_trans *trans, int queue, int budget)
{
struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
struct napi_struct *napi;
struct iwl_rxq *rxq;
u32 r, i, count = 0;
u32 r, i, count = 0, handled = 0;
bool emergency = false;
if (WARN_ON_ONCE(!trans_pcie->rxq || !trans_pcie->rxq[queue].bd))
return;
return budget;
rxq = &trans_pcie->rxq[queue];
@ -1444,7 +1527,7 @@ restart:
if (i == r)
IWL_DEBUG_RX(trans, "Q %d: HW = SW = %d\n", rxq->id, r);
while (i != r) {
while (i != r && ++handled < budget) {
struct iwl_rb_allocator *rba = &trans_pcie->rba;
struct iwl_rx_mem_buffer *rxb;
/* number of RBDs still waiting for page allocation */
@ -1545,18 +1628,9 @@ out:
if (unlikely(emergency && count))
iwl_pcie_rxq_alloc_rbs(trans, GFP_ATOMIC, rxq);
napi = &rxq->napi;
if (napi->poll) {
napi_gro_flush(napi, false);
if (napi->rx_count) {
netif_receive_skb_list(&napi->rx_list);
INIT_LIST_HEAD(&napi->rx_list);
napi->rx_count = 0;
}
}
iwl_pcie_rxq_restock(trans, rxq);
return handled;
}
static struct iwl_trans_pcie *iwl_pcie_get_trans_pcie(struct msix_entry *entry)
@ -1576,6 +1650,7 @@ irqreturn_t iwl_pcie_irq_rx_msix_handler(int irq, void *dev_id)
struct msix_entry *entry = dev_id;
struct iwl_trans_pcie *trans_pcie = iwl_pcie_get_trans_pcie(entry);
struct iwl_trans *trans = trans_pcie->trans;
struct iwl_rxq *rxq = &trans_pcie->rxq[entry->entry];
trace_iwlwifi_dev_irq_msix(trans->dev, entry, false, 0, 0);
@ -1585,11 +1660,12 @@ irqreturn_t iwl_pcie_irq_rx_msix_handler(int irq, void *dev_id)
lock_map_acquire(&trans->sync_cmd_lockdep_map);
local_bh_disable();
iwl_pcie_rx_handle(trans, entry->entry);
if (napi_schedule_prep(&rxq->napi))
__napi_schedule(&rxq->napi);
else
iwl_pcie_clear_irq(trans, entry->entry);
local_bh_enable();
iwl_pcie_clear_irq(trans, entry);
lock_map_release(&trans->sync_cmd_lockdep_map);
return IRQ_HANDLED;
@ -1757,10 +1833,11 @@ irqreturn_t iwl_pcie_irq_handler(int irq, void *dev_id)
struct isr_statistics *isr_stats = &trans_pcie->isr_stats;
u32 inta = 0;
u32 handled = 0;
bool polling = false;
lock_map_acquire(&trans->sync_cmd_lockdep_map);
spin_lock(&trans_pcie->irq_lock);
spin_lock_bh(&trans_pcie->irq_lock);
/* dram interrupt table not set yet,
* use legacy interrupt.
@ -1797,7 +1874,7 @@ irqreturn_t iwl_pcie_irq_handler(int irq, void *dev_id)
*/
if (test_bit(STATUS_INT_ENABLED, &trans->status))
_iwl_enable_interrupts(trans);
spin_unlock(&trans_pcie->irq_lock);
spin_unlock_bh(&trans_pcie->irq_lock);
lock_map_release(&trans->sync_cmd_lockdep_map);
return IRQ_NONE;
}
@ -1808,7 +1885,7 @@ irqreturn_t iwl_pcie_irq_handler(int irq, void *dev_id)
* already raised an interrupt.
*/
IWL_WARN(trans, "HARDWARE GONE?? INTA == 0x%08x\n", inta);
spin_unlock(&trans_pcie->irq_lock);
spin_unlock_bh(&trans_pcie->irq_lock);
goto out;
}
@ -1829,7 +1906,7 @@ irqreturn_t iwl_pcie_irq_handler(int irq, void *dev_id)
IWL_DEBUG_ISR(trans, "inta 0x%08x, enabled 0x%08x\n",
inta, iwl_read32(trans, CSR_INT_MASK));
spin_unlock(&trans_pcie->irq_lock);
spin_unlock_bh(&trans_pcie->irq_lock);
/* Now service all interrupt bits discovered above. */
if (inta & CSR_INT_BIT_HW_ERR) {
@ -1949,7 +2026,10 @@ irqreturn_t iwl_pcie_irq_handler(int irq, void *dev_id)
isr_stats->rx++;
local_bh_disable();
iwl_pcie_rx_handle(trans, 0);
if (napi_schedule_prep(&trans_pcie->rxq[0].napi)) {
polling = true;
__napi_schedule(&trans_pcie->rxq[0].napi);
}
local_bh_enable();
}
@ -1974,20 +2054,22 @@ irqreturn_t iwl_pcie_irq_handler(int irq, void *dev_id)
inta & ~trans_pcie->inta_mask);
}
spin_lock(&trans_pcie->irq_lock);
/* only Re-enable all interrupt if disabled by irq */
if (test_bit(STATUS_INT_ENABLED, &trans->status))
_iwl_enable_interrupts(trans);
/* we are loading the firmware, enable FH_TX interrupt only */
else if (handled & CSR_INT_BIT_FH_TX)
iwl_enable_fw_load_int(trans);
/* Re-enable RF_KILL if it occurred */
else if (handled & CSR_INT_BIT_RF_KILL)
iwl_enable_rfkill_int(trans);
/* Re-enable the ALIVE / Rx interrupt if it occurred */
else if (handled & (CSR_INT_BIT_ALIVE | CSR_INT_BIT_FH_RX))
iwl_enable_fw_load_int_ctx_info(trans);
spin_unlock(&trans_pcie->irq_lock);
if (!polling) {
spin_lock_bh(&trans_pcie->irq_lock);
/* only Re-enable all interrupt if disabled by irq */
if (test_bit(STATUS_INT_ENABLED, &trans->status))
_iwl_enable_interrupts(trans);
/* we are loading the firmware, enable FH_TX interrupt only */
else if (handled & CSR_INT_BIT_FH_TX)
iwl_enable_fw_load_int(trans);
/* Re-enable RF_KILL if it occurred */
else if (handled & CSR_INT_BIT_RF_KILL)
iwl_enable_rfkill_int(trans);
/* Re-enable the ALIVE / Rx interrupt if it occurred */
else if (handled & (CSR_INT_BIT_ALIVE | CSR_INT_BIT_FH_RX))
iwl_enable_fw_load_int_ctx_info(trans);
spin_unlock_bh(&trans_pcie->irq_lock);
}
out:
lock_map_release(&trans->sync_cmd_lockdep_map);
@ -2049,7 +2131,7 @@ void iwl_pcie_reset_ict(struct iwl_trans *trans)
if (!trans_pcie->ict_tbl)
return;
spin_lock(&trans_pcie->irq_lock);
spin_lock_bh(&trans_pcie->irq_lock);
_iwl_disable_interrupts(trans);
memset(trans_pcie->ict_tbl, 0, ICT_SIZE);
@ -2067,7 +2149,7 @@ void iwl_pcie_reset_ict(struct iwl_trans *trans)
trans_pcie->ict_index = 0;
iwl_write32(trans, CSR_INT, trans_pcie->inta_mask);
_iwl_enable_interrupts(trans);
spin_unlock(&trans_pcie->irq_lock);
spin_unlock_bh(&trans_pcie->irq_lock);
}
/* Device is going down disable ict interrupt usage */
@ -2075,9 +2157,9 @@ void iwl_pcie_disable_ict(struct iwl_trans *trans)
{
struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
spin_lock(&trans_pcie->irq_lock);
spin_lock_bh(&trans_pcie->irq_lock);
trans_pcie->use_ict = false;
spin_unlock(&trans_pcie->irq_lock);
spin_unlock_bh(&trans_pcie->irq_lock);
}
irqreturn_t iwl_pcie_isr(int irq, void *data)
@ -2109,10 +2191,11 @@ irqreturn_t iwl_pcie_irq_msix_handler(int irq, void *dev_id)
struct iwl_trans *trans = trans_pcie->trans;
struct isr_statistics *isr_stats = &trans_pcie->isr_stats;
u32 inta_fh, inta_hw;
bool polling = false;
lock_map_acquire(&trans->sync_cmd_lockdep_map);
spin_lock(&trans_pcie->irq_lock);
spin_lock_bh(&trans_pcie->irq_lock);
inta_fh = iwl_read32(trans, CSR_MSIX_FH_INT_CAUSES_AD);
inta_hw = iwl_read32(trans, CSR_MSIX_HW_INT_CAUSES_AD);
/*
@ -2120,7 +2203,7 @@ irqreturn_t iwl_pcie_irq_msix_handler(int irq, void *dev_id)
*/
iwl_write32(trans, CSR_MSIX_FH_INT_CAUSES_AD, inta_fh);
iwl_write32(trans, CSR_MSIX_HW_INT_CAUSES_AD, inta_hw);
spin_unlock(&trans_pcie->irq_lock);
spin_unlock_bh(&trans_pcie->irq_lock);
trace_iwlwifi_dev_irq_msix(trans->dev, entry, true, inta_fh, inta_hw);
@ -2146,14 +2229,20 @@ irqreturn_t iwl_pcie_irq_msix_handler(int irq, void *dev_id)
if ((trans_pcie->shared_vec_mask & IWL_SHARED_IRQ_NON_RX) &&
inta_fh & MSIX_FH_INT_CAUSES_Q0) {
local_bh_disable();
iwl_pcie_rx_handle(trans, 0);
if (napi_schedule_prep(&trans_pcie->rxq[0].napi)) {
polling = true;
__napi_schedule(&trans_pcie->rxq[0].napi);
}
local_bh_enable();
}
if ((trans_pcie->shared_vec_mask & IWL_SHARED_IRQ_FIRST_RSS) &&
inta_fh & MSIX_FH_INT_CAUSES_Q1) {
local_bh_disable();
iwl_pcie_rx_handle(trans, 1);
if (napi_schedule_prep(&trans_pcie->rxq[1].napi)) {
polling = true;
__napi_schedule(&trans_pcie->rxq[1].napi);
}
local_bh_enable();
}
@ -2248,7 +2337,8 @@ irqreturn_t iwl_pcie_irq_msix_handler(int irq, void *dev_id)
wake_up(&trans_pcie->fw_reset_waitq);
}
iwl_pcie_clear_irq(trans, entry);
if (!polling)
iwl_pcie_clear_irq(trans, entry->entry);
lock_map_release(&trans->sync_cmd_lockdep_map);

View File

@ -213,9 +213,9 @@ static int iwl_pcie_gen2_nic_init(struct iwl_trans *trans)
trans->cfg->min_txq_size);
/* TODO: most of the logic can be removed in A0 - but not in Z0 */
spin_lock(&trans_pcie->irq_lock);
spin_lock_bh(&trans_pcie->irq_lock);
iwl_pcie_gen2_apm_init(trans);
spin_unlock(&trans_pcie->irq_lock);
spin_unlock_bh(&trans_pcie->irq_lock);
iwl_op_mode_nic_config(trans->op_mode);

View File

@ -511,9 +511,9 @@ static int iwl_pcie_nic_init(struct iwl_trans *trans)
int ret;
/* nic_init */
spin_lock(&trans_pcie->irq_lock);
spin_lock_bh(&trans_pcie->irq_lock);
ret = iwl_pcie_apm_init(trans);
spin_unlock(&trans_pcie->irq_lock);
spin_unlock_bh(&trans_pcie->irq_lock);
if (ret)
return ret;

View File

@ -393,7 +393,7 @@ static void iwl_pcie_tx_stop_fh(struct iwl_trans *trans)
int ch, ret;
u32 mask = 0;
spin_lock(&trans_pcie->irq_lock);
spin_lock_bh(&trans_pcie->irq_lock);
if (!iwl_trans_grab_nic_access(trans, &flags))
goto out;
@ -414,7 +414,7 @@ static void iwl_pcie_tx_stop_fh(struct iwl_trans *trans)
iwl_trans_release_nic_access(trans, &flags);
out:
spin_unlock(&trans_pcie->irq_lock);
spin_unlock_bh(&trans_pcie->irq_lock);
}
/*
@ -571,7 +571,7 @@ int iwl_pcie_tx_init(struct iwl_trans *trans)
alloc = true;
}
spin_lock(&trans_pcie->irq_lock);
spin_lock_bh(&trans_pcie->irq_lock);
/* Turn off all Tx DMA fifos */
iwl_scd_deactivate_fifos(trans);
@ -580,7 +580,7 @@ int iwl_pcie_tx_init(struct iwl_trans *trans)
iwl_write_direct32(trans, FH_KW_MEM_ADDR_REG,
trans_pcie->kw.dma >> 4);
spin_unlock(&trans_pcie->irq_lock);
spin_unlock_bh(&trans_pcie->irq_lock);
/* Alloc and init all Tx queues, including the command queue (#4/#9) */
for (txq_id = 0; txq_id < trans->trans_cfg->base_params->num_of_queues;