1 files changed, 277 insertions, 83 deletions

diff --git a/drivers/net/wireless/intel/iwlwifi/pcie/rx.c b/drivers/net/wireless/intel/iwlwifi/pcie/rx.c
index 07973ef826c1..7f8a2322cda2 100644
--- a/drivers/net/wireless/intel/iwlwifi/pcie/rx.c
+++ b/drivers/net/wireless/intel/iwlwifi/pcie/rx.c
@@ -210,8 +210,12 @@ static void iwl_pcie_rxq_inc_wr_ptr(struct iwl_trans *trans,
 	if (trans->cfg->mq_rx_supported)
 		iwl_write_prph(trans, RFH_Q_FRBDCB_WIDX(rxq->id),
 			       rxq->write_actual);
-	else
-		iwl_write32(trans, FH_RSCSR_CHNL0_WPTR, rxq->write_actual);
+	/*
+	 * write to FH_RSCSR_CHNL0_WPTR register even in MQ as a W/A to
+	 * hardware shadow registers bug - writing to RFH_Q_FRBDCB_WIDX will
+	 * not wake the NIC.
+	 */
+	iwl_write32(trans, FH_RSCSR_CHNL0_WPTR, rxq->write_actual);
 }
 
 static void iwl_pcie_rxq_check_wrptr(struct iwl_trans *trans)
@@ -231,6 +235,9 @@ static void iwl_pcie_rxq_check_wrptr(struct iwl_trans *trans)
 	}
 }
 
+/*
+ * iwl_pcie_rxq_mq_restock - restock implementation for multi-queue rx
+ */
 static void iwl_pcie_rxq_mq_restock(struct iwl_trans *trans,
 				    struct iwl_rxq *rxq)
 {
@@ -277,17 +284,10 @@ static void iwl_pcie_rxq_mq_restock(struct iwl_trans *trans,
 }
 
 /*
- * iwl_pcie_rxq_restock - refill RX queue from pre-allocated pool
- *
- * If there are slots in the RX queue that need to be restocked,
- * and we have free pre-allocated buffers, fill the ranks as much
- * as we can, pulling from rx_free.
- *
- * This moves the 'write' index forward to catch up with 'processed', and
- * also updates the memory address in the firmware to reference the new
- * target buffer.
+ * iwl_pcie_rxq_sq_restock - restock implementation for single queue rx
  */
-static void iwl_pcie_rxq_restock(struct iwl_trans *trans, struct iwl_rxq *rxq)
+static void iwl_pcie_rxq_sq_restock(struct iwl_trans *trans,
+				    struct iwl_rxq *rxq)
 {
 	struct iwl_rx_mem_buffer *rxb;
 
@@ -332,6 +332,26 @@ static void iwl_pcie_rxq_restock(struct iwl_trans *trans, struct iwl_rxq *rxq)
 }
 
 /*
+ * iwl_pcie_rxq_restock - refill RX queue from pre-allocated pool
+ *
+ * If there are slots in the RX queue that need to be restocked,
+ * and we have free pre-allocated buffers, fill the ranks as much
+ * as we can, pulling from rx_free.
+ *
+ * This moves the 'write' index forward to catch up with 'processed', and
+ * also updates the memory address in the firmware to reference the new
+ * target buffer.
+ */
+static
+void iwl_pcie_rxq_restock(struct iwl_trans *trans, struct iwl_rxq *rxq)
+{
+	if (trans->cfg->mq_rx_supported)
+		iwl_pcie_rxq_mq_restock(trans, rxq);
+	else
+		iwl_pcie_rxq_sq_restock(trans, rxq);
+}
+
+/*
  * iwl_pcie_rx_alloc_page - allocates and returns a page.
  *
  */
@@ -434,7 +454,7 @@ static void iwl_pcie_free_rbs_pool(struct iwl_trans *trans)
 	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
 	int i;
 
-	for (i = 0; i < MQ_RX_POOL_SIZE; i++) {
+	for (i = 0; i < RX_POOL_SIZE; i++) {
 		if (!trans_pcie->rx_pool[i].page)
 			continue;
 		dma_unmap_page(trans->dev, trans_pcie->rx_pool[i].page_dma,
@@ -539,40 +559,46 @@ static void iwl_pcie_rx_allocator(struct iwl_trans *trans)
 }
 
 /*
- * iwl_pcie_rx_allocator_get - Returns the pre-allocated pages
+ * iwl_pcie_rx_allocator_get - returns the pre-allocated pages
 .*
 .* Called by queue when the queue posted allocation request and
  * has freed 8 RBDs in order to restock itself.
+ * This function directly moves the allocated RBs to the queue's ownership
+ * and updates the relevant counters.
  */
-static int iwl_pcie_rx_allocator_get(struct iwl_trans *trans,
-				     struct iwl_rx_mem_buffer
-				     *out[RX_CLAIM_REQ_ALLOC])
+static void iwl_pcie_rx_allocator_get(struct iwl_trans *trans,
+				      struct iwl_rxq *rxq)
 {
 	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
 	struct iwl_rb_allocator *rba = &trans_pcie->rba;
 	int i;
 
+	lockdep_assert_held(&rxq->lock);
+
 	/*
 	 * atomic_dec_if_positive returns req_ready - 1 for any scenario.
 	 * If req_ready is 0 atomic_dec_if_positive will return -1 and this
-	 * function will return -ENOMEM, as there are no ready requests.
+	 * function will return early, as there are no ready requests.
 	 * atomic_dec_if_positive will perofrm the *actual* decrement only if
 	 * req_ready > 0, i.e. - there are ready requests and the function
 	 * hands one request to the caller.
 	 */
 	if (atomic_dec_if_positive(&rba->req_ready) < 0)
-		return -ENOMEM;
+		return;
 
 	spin_lock(&rba->lock);
 	for (i = 0; i < RX_CLAIM_REQ_ALLOC; i++) {
 		/* Get next free Rx buffer, remove it from free list */
-		out[i] = list_first_entry(&rba->rbd_allocated,
-			       struct iwl_rx_mem_buffer, list);
-		list_del(&out[i]->list);
+		struct iwl_rx_mem_buffer *rxb =
+			list_first_entry(&rba->rbd_allocated,
+					 struct iwl_rx_mem_buffer, list);
+
+		list_move(&rxb->list, &rxq->rx_free);
 	}
 	spin_unlock(&rba->lock);
 
-	return 0;
+	rxq->used_count -= RX_CLAIM_REQ_ALLOC;
+	rxq->free_count += RX_CLAIM_REQ_ALLOC;
 }
 
 static void iwl_pcie_rx_allocator_work(struct work_struct *data)
@@ -783,16 +809,25 @@ static void iwl_pcie_rx_mq_hw_init(struct iwl_trans *trans)
 	 * Single frame mode
 	 * Rx buffer size 4 or 8k or 12k
 	 * Min RB size 4 or 8
+	 * Drop frames that exceed RB size
 	 * 512 RBDs
 	 */
 	iwl_write_prph(trans, RFH_RXF_DMA_CFG,
 		       RFH_DMA_EN_ENABLE_VAL |
 		       rb_size | RFH_RXF_DMA_SINGLE_FRAME_MASK |
 		       RFH_RXF_DMA_MIN_RB_4_8 |
+		       RFH_RXF_DMA_DROP_TOO_LARGE_MASK |
 		       RFH_RXF_DMA_RBDCB_SIZE_512);
 
+	/*
+	 * Activate DMA snooping.
+	 * Set RX DMA chunk size to 64B
+	 * Default queue is 0
+	 */
 	iwl_write_prph(trans, RFH_GEN_CFG, RFH_GEN_CFG_RFH_DMA_SNOOP |
-					  RFH_GEN_CFG_SERVICE_DMA_SNOOP);
+		       (DEFAULT_RXQ_NUM << RFH_GEN_CFG_DEFAULT_RXQ_NUM_POS) |
+		       RFH_GEN_CFG_SERVICE_DMA_SNOOP);
+	/* Enable the relevant rx queues */
 	iwl_write_prph(trans, RFH_RXF_RXQ_ACTIVE, enabled);
 
 	/* Set interrupt coalescing timer to default (2048 usecs) */
@@ -820,7 +855,7 @@ int iwl_pcie_rx_init(struct iwl_trans *trans)
 	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
 	struct iwl_rxq *def_rxq;
 	struct iwl_rb_allocator *rba = &trans_pcie->rba;
-	int i, err, num_rbds, allocator_pool_size;
+	int i, err, queue_size, allocator_pool_size, num_alloc;
 
 	if (!trans_pcie->rxq) {
 		err = iwl_pcie_rx_alloc(trans);
@@ -872,11 +907,14 @@ int iwl_pcie_rx_init(struct iwl_trans *trans)
 	}
 
 	/* move the pool to the default queue and allocator ownerships */
-	num_rbds = trans->cfg->mq_rx_supported ?
-		     MQ_RX_POOL_SIZE : RX_QUEUE_SIZE;
+	queue_size = trans->cfg->mq_rx_supported ?
+		     MQ_RX_NUM_RBDS : RX_QUEUE_SIZE;
 	allocator_pool_size = trans->num_rx_queues *
 		(RX_CLAIM_REQ_ALLOC - RX_POST_REQ_ALLOC);
-	for (i = 0; i < num_rbds; i++) {
+	num_alloc = queue_size + allocator_pool_size;
+	BUILD_BUG_ON(ARRAY_SIZE(trans_pcie->global_table) !=
+		     ARRAY_SIZE(trans_pcie->rx_pool));
+	for (i = 0; i < num_alloc; i++) {
 		struct iwl_rx_mem_buffer *rxb = &trans_pcie->rx_pool[i];
 
 		if (i < allocator_pool_size)
@@ -891,7 +929,7 @@ int iwl_pcie_rx_init(struct iwl_trans *trans)
 	if (trans->cfg->mq_rx_supported) {
 		iwl_pcie_rx_mq_hw_init(trans);
 	} else {
-		iwl_pcie_rxq_restock(trans, def_rxq);
+		iwl_pcie_rxq_sq_restock(trans, def_rxq);
 		iwl_pcie_rx_hw_init(trans, def_rxq);
 	}
 
@@ -1135,11 +1173,11 @@ static void iwl_pcie_rx_handle_rb(struct iwl_trans *trans,
 /*
  * iwl_pcie_rx_handle - Main entry function for receiving responses from fw
  */
-static void iwl_pcie_rx_handle(struct iwl_trans *trans)
+static void iwl_pcie_rx_handle(struct iwl_trans *trans, int queue)
 {
 	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
-	struct iwl_rxq *rxq = &trans_pcie->rxq[0];
-	u32 r, i, j, count = 0;
+	struct iwl_rxq *rxq = &trans_pcie->rxq[queue];
+	u32 r, i, count = 0;
 	bool emergency = false;
 
 restart:
@@ -1149,9 +1187,12 @@ restart:
 	r = le16_to_cpu(ACCESS_ONCE(rxq->rb_stts->closed_rb_num)) & 0x0FFF;
 	i = rxq->read;
 
+	/* W/A 9000 device step A0 wrap-around bug */
+	r &= (rxq->queue_size - 1);
+
 	/* Rx interrupt, but nothing sent from uCode */
 	if (i == r)
-		IWL_DEBUG_RX(trans, "HW = SW = %d\n", r);
+		IWL_DEBUG_RX(trans, "Q %d: HW = SW = %d\n", rxq->id, r);
 
 	while (i != r) {
 		struct iwl_rx_mem_buffer *rxb;
@@ -1164,78 +1205,55 @@ restart:
 			 * used_bd is a 32 bit but only 12 are used to retrieve
 			 * the vid
 			 */
-			u16 vid = (u16)le32_to_cpu(rxq->used_bd[i]);
+			u16 vid = le32_to_cpu(rxq->used_bd[i]) & 0x0FFF;
 
+			if (WARN(vid >= ARRAY_SIZE(trans_pcie->global_table),
+				 "Invalid rxb index from HW %u\n", (u32)vid))
+				goto out;
 			rxb = trans_pcie->global_table[vid];
 		} else {
 			rxb = rxq->queue[i];
 			rxq->queue[i] = NULL;
 		}
 
-		IWL_DEBUG_RX(trans, "rxbuf: HW = %d, SW = %d\n", r, i);
+		IWL_DEBUG_RX(trans, "Q %d: HW = %d, SW = %d\n", rxq->id, r, i);
 		iwl_pcie_rx_handle_rb(trans, rxq, rxb, emergency);
 
 		i = (i + 1) & (rxq->queue_size - 1);
 
-		/* If we have RX_CLAIM_REQ_ALLOC released rx buffers -
-		 * try to claim the pre-allocated buffers from the allocator */
-		if (rxq->used_count >= RX_CLAIM_REQ_ALLOC) {
+		/*
+		 * If we have RX_CLAIM_REQ_ALLOC released rx buffers -
+		 * try to claim the pre-allocated buffers from the allocator.
+		 * If not ready - will try to reclaim next time.
+		 * There is no need to reschedule work - allocator exits only
+		 * on success
+		 */
+		if (rxq->used_count >= RX_CLAIM_REQ_ALLOC)
+			iwl_pcie_rx_allocator_get(trans, rxq);
+
+		if (rxq->used_count % RX_CLAIM_REQ_ALLOC == 0 && !emergency) {
 			struct iwl_rb_allocator *rba = &trans_pcie->rba;
-			struct iwl_rx_mem_buffer *out[RX_CLAIM_REQ_ALLOC];
-
-			if (rxq->used_count % RX_CLAIM_REQ_ALLOC == 0 &&
-			    !emergency) {
-				/* Add the remaining 6 empty RBDs
-				* for allocator use
-				 */
-				spin_lock(&rba->lock);
-				list_splice_tail_init(&rxq->rx_used,
-						      &rba->rbd_empty);
-				spin_unlock(&rba->lock);
-			}
 
-			/* If not ready - continue, will try to reclaim later.
-			* No need to reschedule work - allocator exits only on
-			* success */
-			if (!iwl_pcie_rx_allocator_get(trans, out)) {
-				/* If success - then RX_CLAIM_REQ_ALLOC
-				 * buffers were retrieved and should be added
-				 * to free list */
-				rxq->used_count -= RX_CLAIM_REQ_ALLOC;
-				for (j = 0; j < RX_CLAIM_REQ_ALLOC; j++) {
-					list_add_tail(&out[j]->list,
-						      &rxq->rx_free);
-					rxq->free_count++;
-				}
-			}
-		}
-		if (emergency) {
+			/* Add the remaining empty RBDs for allocator use */
+			spin_lock(&rba->lock);
+			list_splice_tail_init(&rxq->rx_used, &rba->rbd_empty);
+			spin_unlock(&rba->lock);
+		} else if (emergency) {
 			count++;
 			if (count == 8) {
 				count = 0;
 				if (rxq->used_count < rxq->queue_size / 3)
 					emergency = false;
+
+				rxq->read = i;
 				spin_unlock(&rxq->lock);
 				iwl_pcie_rxq_alloc_rbs(trans, GFP_ATOMIC, rxq);
-				spin_lock(&rxq->lock);
-			}
-		}
-		/* handle restock for three cases, can be all of them at once:
-		* - we just pulled buffers from the allocator
-		* - we have 8+ unstolen pages accumulated
-		* - we are in emergency and allocated buffers
-		 */
-		if (rxq->free_count >=  RX_CLAIM_REQ_ALLOC) {
-			rxq->read = i;
-			spin_unlock(&rxq->lock);
-			if (trans->cfg->mq_rx_supported)
-				iwl_pcie_rxq_mq_restock(trans, rxq);
-			else
 				iwl_pcie_rxq_restock(trans, rxq);
-			goto restart;
+				goto restart;
+			}
 		}
 	}
-
+out:
 	/* Backtrack one entry */
 	rxq->read = i;
 	spin_unlock(&rxq->lock);
@@ -1257,6 +1275,56 @@ restart:
 
 	if (rxq->napi.poll)
 		napi_gro_flush(&rxq->napi, false);
+
+	iwl_pcie_rxq_restock(trans, rxq);
+}
+
+static struct iwl_trans_pcie *iwl_pcie_get_trans_pcie(struct msix_entry *entry)
+{
+	u8 queue = entry->entry;
+	struct msix_entry *entries = entry - queue;
+
+	return container_of(entries, struct iwl_trans_pcie, msix_entries[0]);
+}
+
+static inline void iwl_pcie_clear_irq(struct iwl_trans *trans,
+				      struct msix_entry *entry)
+{
+	/*
+	 * Before sending the interrupt the HW disables it to prevent
+	 * a nested interrupt. This is done by writing 1 to the corresponding
+	 * bit in the mask register. After handling the interrupt, it should be
+	 * re-enabled by clearing this bit. This register is defined as
+	 * write 1 clear (W1C) register, meaning that it's being clear
+	 * by writing 1 to the bit.
+	 */
+	iwl_write_direct32(trans, CSR_MSIX_AUTOMASK_ST_AD, BIT(entry->entry));
+}
+
+/*
+ * iwl_pcie_rx_msix_handle - Main entry function for receiving responses from fw
+ * This interrupt handler should be used with RSS queue only.
+ */
+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;
+
+	if (WARN_ON(entry->entry >= trans->num_rx_queues))
+		return IRQ_NONE;
+
+	lock_map_acquire(&trans->sync_cmd_lockdep_map);
+
+	local_bh_disable();
+	iwl_pcie_rx_handle(trans, entry->entry);
+	local_bh_enable();
+
+	iwl_pcie_clear_irq(trans, entry);
+
+	lock_map_release(&trans->sync_cmd_lockdep_map);
+
+	return IRQ_HANDLED;
 }
 
 /*
@@ -1589,7 +1657,7 @@ irqreturn_t iwl_pcie_irq_handler(int irq, void *dev_id)
 		isr_stats->rx++;
 
 		local_bh_disable();
-		iwl_pcie_rx_handle(trans);
+		iwl_pcie_rx_handle(trans, 0);
 		local_bh_enable();
 	}
 
@@ -1732,3 +1800,129 @@ irqreturn_t iwl_pcie_isr(int irq, void *data)
 
 	return IRQ_WAKE_THREAD;
 }
+
+irqreturn_t iwl_pcie_msix_isr(int irq, void *data)
+{
+	return IRQ_WAKE_THREAD;
+}
+
+irqreturn_t iwl_pcie_irq_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 isr_statistics *isr_stats = &trans_pcie->isr_stats;
+	u32 inta_fh, inta_hw;
+
+	lock_map_acquire(&trans->sync_cmd_lockdep_map);
+
+	spin_lock(&trans_pcie->irq_lock);
+	inta_fh = iwl_read_direct32(trans, CSR_MSIX_FH_INT_CAUSES_AD);
+	inta_hw = iwl_read_direct32(trans, CSR_MSIX_HW_INT_CAUSES_AD);
+	/*
+	 * Clear causes registers to avoid being handling the same cause.
+	 */
+	iwl_write_direct32(trans, CSR_MSIX_FH_INT_CAUSES_AD, inta_fh);
+	iwl_write_direct32(trans, CSR_MSIX_HW_INT_CAUSES_AD, inta_hw);
+	spin_unlock(&trans_pcie->irq_lock);
+
+	if (unlikely(!(inta_fh | inta_hw))) {
+		IWL_DEBUG_ISR(trans, "Ignore interrupt, inta == 0\n");
+		lock_map_release(&trans->sync_cmd_lockdep_map);
+		return IRQ_NONE;
+	}
+
+	if (iwl_have_debug_level(IWL_DL_ISR))
+		IWL_DEBUG_ISR(trans, "ISR inta_fh 0x%08x, enabled 0x%08x\n",
+			      inta_fh,
+			      iwl_read32(trans, CSR_MSIX_FH_INT_MASK_AD));
+
+	/* This "Tx" DMA channel is used only for loading uCode */
+	if (inta_fh & MSIX_FH_INT_CAUSES_D2S_CH0_NUM) {
+		IWL_DEBUG_ISR(trans, "uCode load interrupt\n");
+		isr_stats->tx++;
+		/*
+		 * Wake up uCode load routine,
+		 * now that load is complete
+		 */
+		trans_pcie->ucode_write_complete = true;
+		wake_up(&trans_pcie->ucode_write_waitq);
+	}
+
+	/* Error detected by uCode */
+	if ((inta_fh & MSIX_FH_INT_CAUSES_FH_ERR) ||
+	    (inta_hw & MSIX_HW_INT_CAUSES_REG_SW_ERR)) {
+		IWL_ERR(trans,
+			"Microcode SW error detected. Restarting 0x%X.\n",
+			inta_fh);
+		isr_stats->sw++;
+		iwl_pcie_irq_handle_error(trans);
+	}
+
+	/* After checking FH register check HW register */
+	if (iwl_have_debug_level(IWL_DL_ISR))
+		IWL_DEBUG_ISR(trans,
+			      "ISR inta_hw 0x%08x, enabled 0x%08x\n",
+			      inta_hw,
+			      iwl_read32(trans, CSR_MSIX_HW_INT_MASK_AD));
+
+	/* Alive notification via Rx interrupt will do the real work */
+	if (inta_hw & MSIX_HW_INT_CAUSES_REG_ALIVE) {
+		IWL_DEBUG_ISR(trans, "Alive interrupt\n");
+		isr_stats->alive++;
+	}
+
+	/* uCode wakes up after power-down sleep */
+	if (inta_hw & MSIX_HW_INT_CAUSES_REG_WAKEUP) {
+		IWL_DEBUG_ISR(trans, "Wakeup interrupt\n");
+		iwl_pcie_rxq_check_wrptr(trans);
+		iwl_pcie_txq_check_wrptrs(trans);
+
+		isr_stats->wakeup++;
+	}
+
+	/* Chip got too hot and stopped itself */
+	if (inta_hw & MSIX_HW_INT_CAUSES_REG_CT_KILL) {
+		IWL_ERR(trans, "Microcode CT kill error detected.\n");
+		isr_stats->ctkill++;
+	}
+
+	/* HW RF KILL switch toggled */
+	if (inta_hw & MSIX_HW_INT_CAUSES_REG_RF_KILL) {
+		bool hw_rfkill;
+
+		hw_rfkill = iwl_is_rfkill_set(trans);
+		IWL_WARN(trans, "RF_KILL bit toggled to %s.\n",
+			 hw_rfkill ? "disable radio" : "enable radio");
+
+		isr_stats->rfkill++;
+
+		mutex_lock(&trans_pcie->mutex);
+		iwl_trans_pcie_rf_kill(trans, hw_rfkill);
+		mutex_unlock(&trans_pcie->mutex);
+		if (hw_rfkill) {
+			set_bit(STATUS_RFKILL, &trans->status);
+			if (test_and_clear_bit(STATUS_SYNC_HCMD_ACTIVE,
+					       &trans->status))
+				IWL_DEBUG_RF_KILL(trans,
+						  "Rfkill while SYNC HCMD in flight\n");
+			wake_up(&trans_pcie->wait_command_queue);
+		} else {
+			clear_bit(STATUS_RFKILL, &trans->status);
+		}
+	}
+
+	if (inta_hw & MSIX_HW_INT_CAUSES_REG_HW_ERR) {
+		IWL_ERR(trans,
+			"Hardware error detected. Restarting.\n");
+
+		isr_stats->hw++;
+		iwl_pcie_irq_handle_error(trans);
+	}
+
+	iwl_pcie_clear_irq(trans, entry);
+
+	lock_map_release(&trans->sync_cmd_lockdep_map);
+
+	return IRQ_HANDLED;
+}