Skip to contentSkip to conversation

Add an i40e_napi_poll tracepoint

PATCH In review Joe Damato 2 participants 9 messages Oct 7, 2022 – Oct 14, 2022

Subsystems (3)
lore
Share & tools
lore

Joe Damato

author
#0 lore raw

v4 cover Add an i40e_napi_poll tracepoint

Greetings:

Welcome to v4.

Debugging and tuning the NAPI and i40e NIC parameters can be a bit tricky as there are many different options to test.

This change adds a tracepoint to i40e_napi_poll which exposes a lot of helpful debug information for users who'd like to get a better understanding of how their NIC is performing as they adjust various parameters and tuning knobs.

Note: this series does not touch any XDP related code paths. This tracepoint will only work when not using XDP. Care has been taken to avoid changing control flow in i40e_napi_poll with this change.

With this series applied, you can use the tracepoint with perf (assuming XDP is not being used) by running:

$ sudo perf trace -e i40e:i40e_napi_poll -a --call-graph=fp --libtraceevent_print

388.258 :0/0 i40e:i40e_napi_poll(i40e_napi_poll on dev eth2 q i40e-eth2-TxRx-9 irq 346 irq_mask 00000000,00000000,00000000,00000000,00000000,00800000 curr_cpu 23 budget 64 bpr 64 rx_cleaned 28 tx_cleaned 0 rx_clean_complete 1 tx_clean_complete 1)

	i40e_napi_poll ([i40e])
	i40e_napi_poll ([i40e])
	__napi_poll ([kernel.kallsyms])
	net_rx_action ([kernel.kallsyms])
	__do_softirq ([kernel.kallsyms])
	common_interrupt ([kernel.kallsyms])
	asm_common_interrupt ([kernel.kallsyms])
	intel_idle_irq ([kernel.kallsyms])
	cpuidle_enter_state ([kernel.kallsyms])
	cpuidle_enter ([kernel.kallsyms])
	do_idle ([kernel.kallsyms])
	cpu_startup_entry ([kernel.kallsyms])
	[0x243fd8] ([kernel.kallsyms])
	secondary_startup_64_no_verify ([kernel.kallsyms])

The output is verbose, but is helpful when trying to determine the impact of various turning parameters.

Thanks,
Joe

v3 -> v4:
- All XDP related changes were dropped from both the TX and RX paths (patches 2/4 and 3/4 were updated).
- A conditional was added to only fire the tracepoint if XDP is NOT enabled (patch 4/4 updated).
- Fixed format specifier for rx_cleaned and tx_cleaned (from %lu to %u in patch 4/4).

v2 -> v3:
- Add an rx_clean_complete to the RX patch so that it can be output in tracepoint instead of the valued of 'clean_complete' which can be ambiguous (patch 3/4 was updated).
- Update the tracepoint to swap 'clean_complete' with 'rx_clean_complete' (patch 4/4 was updated).

v1 -> v2:
- TX path modified to push an out parameter through the function call chain instead of modifying control flow.
- RX path modified to also use an out parameter to track the number of packets processed.
- Naming of tracepoint struct members and format string modified to be more readable.

Thanks,
Joe

Joe Damato (4):
i40e: Store the irq number in i40e_q_vector
i40e: Record number TXes cleaned during NAPI
i40e: Record number of RXes cleaned during NAPI
i40e: Add i40e_napi_poll tracepoint

 drivers/net/ethernet/intel/i40e/i40e.h       |  1 +
 drivers/net/ethernet/intel/i40e/i40e_main.c  |  1 +
 drivers/net/ethernet/intel/i40e/i40e_trace.h | 49 ++++++++++++++++++++++++++++
 drivers/net/ethernet/intel/i40e/i40e_txrx.c  | 27 +++++++++++----
 4 files changed, 72 insertions(+), 6 deletions(-)
Signature · 6 lines
2.7.4

_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
author
#1 lore raw

PATCH v4 1/4 i40e: Store the irq number in i40e_q_vector

Make it easy to figure out the IRQ number for a particular i40e_q_vector by storing the assigned IRQ in the structure itself.

---
 drivers/net/ethernet/intel/i40e/i40e.h      | 1 +
 drivers/net/ethernet/intel/i40e/i40e_main.c | 1 +
 2 files changed, 2 insertions(+)
drivers/net/ethernet/intel/i40e/i40e.h +1 -0
index 9926c4e..8e1f395 100644
@@ -992,6 +992,7 @@ struct i40e_q_vector {
 	struct rcu_head rcu;	/* to avoid race with update stats on free */
 	char name[I40E_INT_NAME_STR_LEN];
 	bool arm_wb_state;
+	int irq_num;		/* IRQ assigned to this q_vector */
 } ____cacheline_internodealigned_in_smp;
 
 /* lan device */
drivers/net/ethernet/intel/i40e/i40e_main.c +1 -0
index 6b7535a..6efe130 100644
@@ -4123,6 +4123,7 @@ static int i40e_vsi_request_irq_msix(struct i40e_vsi *vsi, char *basename)
 		}
 
 		/* register for affinity change notifications */
+		q_vector->irq_num = irq_num;
 		q_vector->affinity_notify.notify = i40e_irq_affinity_notify;
 		q_vector->affinity_notify.release = i40e_irq_affinity_release;
 		irq_set_affinity_notifier(irq_num, &q_vector->affinity_notify);
Signature · 6 lines
2.7.4

_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
Tested-by on v4 1/4
#2 lore raw

RE: [Intel-wired-lan] [next-queue v4 1/4] i40e: Store the irq number in i40e_q_vector

Quoted 15 lines

-----Original Message-----
From: Intel-wired-lan <[email protected]> On Behalf Of Joe Damato
Sent: Saturday, October 8, 2022 3:09 AM
To: [email protected]
Cc: [email protected]; Joe Damato <[email protected]>; [email protected]; [email protected]
Subject: [Intel-wired-lan] [next-queue v4 1/4] i40e: Store the irq number in i40e_q_vector

Make it easy to figure out the IRQ number for a particular i40e_q_vector by storing the assigned IRQ in the structure itself.

Quoted 8 lines
---
 drivers/net/ethernet/intel/i40e/i40e.h      | 1 +
 drivers/net/ethernet/intel/i40e/i40e_main.c | 1 +
 2 files changed, 2 insertions(+)

_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan

author
#3 lore raw

PATCH v4 2/4 i40e: Record number TXes cleaned during NAPI

In reply to Joe Damato

Update i40e_clean_tx_irq to take an out parameter (tx_cleaned) which stores the number TXs cleaned.

No XDP related TX code is touched. Care has been taken to avoid changing the control flow of i40e_clean_tx_irq and i40e_napi_poll.

---
 drivers/net/ethernet/intel/i40e/i40e_txrx.c | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)
drivers/net/ethernet/intel/i40e/i40e_txrx.c +8 -3
index b97c95f..274de1c 100644
@@ -923,11 +923,13 @@ void i40e_detect_recover_hung(struct i40e_vsi *vsi)
  * @vsi: the VSI we care about
  * @tx_ring: Tx ring to clean
  * @napi_budget: Used to determine if we are in netpoll
+ * @tx_cleaned: Out parameter set to the number of TXes cleaned
  *
  * Returns true if there's any budget left (e.g. the clean is finished)
  **/
 static bool i40e_clean_tx_irq(struct i40e_vsi *vsi,
-			      struct i40e_ring *tx_ring, int napi_budget)
+			      struct i40e_ring *tx_ring, int napi_budget,
+			      unsigned int *tx_cleaned)
 {
 	int i = tx_ring->next_to_clean;
 	struct i40e_tx_buffer *tx_buf;
@@ -1048,6 +1050,7 @@ static bool i40e_clean_tx_irq(struct i40e_vsi *vsi,
 		}
 	}
 
+	*tx_cleaned = total_packets;
 	return !!budget;
 }
 
@@ -2689,10 +2692,12 @@ int i40e_napi_poll(struct napi_struct *napi, int budget)
 			       container_of(napi, struct i40e_q_vector, napi);
 	struct i40e_vsi *vsi = q_vector->vsi;
 	struct i40e_ring *ring;
+	bool tx_clean_complete = true;
 	bool clean_complete = true;
 	bool arm_wb = false;
 	int budget_per_ring;
 	int work_done = 0;
+	unsigned int tx_cleaned = 0;
 
 	if (test_bit(__I40E_VSI_DOWN, vsi->state)) {
 		napi_complete(napi);
@@ -2705,10 +2710,10 @@ int i40e_napi_poll(struct napi_struct *napi, int budget)
 	i40e_for_each_ring(ring, q_vector->tx) {
 		bool wd = ring->xsk_pool ?
 			  i40e_clean_xdp_tx_irq(vsi, ring) :
-			  i40e_clean_tx_irq(vsi, ring, budget);
+			  i40e_clean_tx_irq(vsi, ring, budget, &tx_cleaned);
 
 		if (!wd) {
-			clean_complete = false;
+			clean_complete = tx_clean_complete = false;
 			continue;
 		}
 		arm_wb |= ring->arm_wb;
Signature · 6 lines
2.7.4

_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
Tested-by on v4 2/4
#4 lore raw

RE: [Intel-wired-lan] [next-queue v4 2/4] i40e: Record number TXes cleaned during NAPI

Quoted 18 lines

-----Original Message-----
From: Intel-wired-lan <[email protected]> On Behalf Of Joe Damato
Sent: Saturday, October 8, 2022 3:09 AM
To: [email protected]
Cc: [email protected]; Joe Damato <[email protected]>; [email protected]; [email protected]
Subject: [Intel-wired-lan] [next-queue v4 2/4] i40e: Record number TXes cleaned during NAPI

Update i40e_clean_tx_irq to take an out parameter (tx_cleaned) which stores the number TXs cleaned.

No XDP related TX code is touched. Care has been taken to avoid changing the control flow of i40e_clean_tx_irq and i40e_napi_poll.

Quoted 7 lines
---
 drivers/net/ethernet/intel/i40e/i40e_txrx.c | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)

_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan

author
#5 lore raw

PATCH v4 3/4 i40e: Record number of RXes cleaned during NAPI

In reply to Joe Damato

Adjust i40e_clean_rx_irq to accept an out parameter which records the number of RX packets cleaned.

No XDP related code is modified and care has been taken to avoid changing control flow.

---
 drivers/net/ethernet/intel/i40e/i40e_txrx.c | 12 +++++++++---
 1 file changed, 9 insertions(+), 3 deletions(-)
drivers/net/ethernet/intel/i40e/i40e_txrx.c +9 -3
index 274de1c..5901e58 100644
@@ -2425,6 +2425,7 @@ static void i40e_inc_ntc(struct i40e_ring *rx_ring)
  * i40e_clean_rx_irq - Clean completed descriptors from Rx ring - bounce buf
  * @rx_ring: rx descriptor ring to transact packets on
  * @budget: Total limit on number of packets to process
+ * @rx_cleaned: Out parameter of the number of packets processed
  *
  * This function provides a "bounce buffer" approach to Rx interrupt
  * processing.  The advantage to this is that on systems that have
@@ -2433,7 +2434,8 @@ static void i40e_inc_ntc(struct i40e_ring *rx_ring)
  *
  * Returns amount of work completed
  **/
-static int i40e_clean_rx_irq(struct i40e_ring *rx_ring, int budget)
+static int i40e_clean_rx_irq(struct i40e_ring *rx_ring, int budget,
+			     unsigned int *rx_cleaned)
 {
 	unsigned int total_rx_bytes = 0, total_rx_packets = 0, frame_sz = 0;
 	u16 cleaned_count = I40E_DESC_UNUSED(rx_ring);
@@ -2570,6 +2572,8 @@ static int i40e_clean_rx_irq(struct i40e_ring *rx_ring, int budget)
 
 	i40e_update_rx_stats(rx_ring, total_rx_bytes, total_rx_packets);
 
+	*rx_cleaned = total_rx_packets;
+
 	/* guarantee a trip back through this routine if there was a failure */
 	return failure ? budget : (int)total_rx_packets;
 }
@@ -2693,11 +2697,13 @@ int i40e_napi_poll(struct napi_struct *napi, int budget)
 	struct i40e_vsi *vsi = q_vector->vsi;
 	struct i40e_ring *ring;
 	bool tx_clean_complete = true;
+	bool rx_clean_complete = true;
 	bool clean_complete = true;
 	bool arm_wb = false;
 	int budget_per_ring;
 	int work_done = 0;
 	unsigned int tx_cleaned = 0;
+	unsigned int rx_cleaned = 0;
 
 	if (test_bit(__I40E_VSI_DOWN, vsi->state)) {
 		napi_complete(napi);
@@ -2738,12 +2744,12 @@ int i40e_napi_poll(struct napi_struct *napi, int budget)
 	i40e_for_each_ring(ring, q_vector->rx) {
 		int cleaned = ring->xsk_pool ?
 			      i40e_clean_rx_irq_zc(ring, budget_per_ring) :
-			      i40e_clean_rx_irq(ring, budget_per_ring);
+			      i40e_clean_rx_irq(ring, budget_per_ring, &rx_cleaned);
 
 		work_done += cleaned;
 		/* if we clean as many as budgeted, we must not be done */
 		if (cleaned >= budget_per_ring)
-			clean_complete = false;
+			clean_complete = rx_clean_complete = false;
 	}
 
 	/* If work not completed, return budget and polling will return */
Signature · 1 line
2.7.4
Tested-by on v4 3/4
#6 lore raw

RE: [Intel-wired-lan] [next-queue v4 3/4] i40e: Record number of RXes cleaned during NAPI

Quoted 18 lines

-----Original Message-----
From: Intel-wired-lan <[email protected]> On Behalf Of Joe Damato
Sent: Saturday, October 8, 2022 3:09 AM
To: [email protected]
Cc: [email protected]; Joe Damato <[email protected]>; [email protected]; [email protected]
Subject: [Intel-wired-lan] [next-queue v4 3/4] i40e: Record number of RXes cleaned during NAPI

Adjust i40e_clean_rx_irq to accept an out parameter which records the number of RX packets cleaned.

No XDP related code is modified and care has been taken to avoid changing control flow.

Quoted 7 lines
---
 drivers/net/ethernet/intel/i40e/i40e_txrx.c | 12 +++++++++---
 1 file changed, 9 insertions(+), 3 deletions(-)
author
#7 lore raw

PATCH v4 4/4 i40e: Add i40e_napi_poll tracepoint

In reply to Joe Damato

Add a tracepoint for i40e_napi_poll that allows users to get detailed information about the amount of work done. This information can help users better tune the correct NAPI parameters (like weight and budget), as well as debug NIC settings like rx-usecs and tx-usecs, etc.

When perf is attached, this tracepoint only fires when not in XDP mode.

An example of the output from this tracepoint:

$ sudo perf trace -e i40e:i40e_napi_poll -a --call-graph=fp --libtraceevent_print

[..snip..]

388.258 :0/0 i40e:i40e_napi_poll(i40e_napi_poll on dev eth2 q i40e-eth2-TxRx-9 irq 346 irq_mask 00000000,00000000,00000000,00000000,00000000,00800000 curr_cpu 23 budget 64 bpr 64 rx_cleaned 28 tx_cleaned 0 rx_clean_complete 1 tx_clean_complete 1)

	i40e_napi_poll ([i40e])
	i40e_napi_poll ([i40e])
	__napi_poll ([kernel.kallsyms])
	net_rx_action ([kernel.kallsyms])
	__do_softirq ([kernel.kallsyms])
	common_interrupt ([kernel.kallsyms])
	asm_common_interrupt ([kernel.kallsyms])
	intel_idle_irq ([kernel.kallsyms])
	cpuidle_enter_state ([kernel.kallsyms])
	cpuidle_enter ([kernel.kallsyms])
	do_idle ([kernel.kallsyms])
	cpu_startup_entry ([kernel.kallsyms])
	[0x243fd8] ([kernel.kallsyms])
	secondary_startup_64_no_verify ([kernel.kallsyms])
---
 drivers/net/ethernet/intel/i40e/i40e_trace.h | 49 ++++++++++++++++++++++++++++
 drivers/net/ethernet/intel/i40e/i40e_txrx.c  |  4 +++
 2 files changed, 53 insertions(+)
drivers/net/ethernet/intel/i40e/i40e_trace.h +49 -0
index b5b1229..79d587a 100644
@@ -55,6 +55,55 @@
  * being built from shared code.
  */
 
+#define NO_DEV "(i40e no_device)"
+
+TRACE_EVENT(i40e_napi_poll,
+
+	TP_PROTO(struct napi_struct *napi, struct i40e_q_vector *q, int budget,
+		 int budget_per_ring, unsigned int rx_cleaned, unsigned int tx_cleaned,
+		 bool rx_clean_complete, bool tx_clean_complete),
+
+	TP_ARGS(napi, q, budget, budget_per_ring, rx_cleaned, tx_cleaned,
+		rx_clean_complete, tx_clean_complete),
+
+	TP_STRUCT__entry(
+		__field(int, budget)
+		__field(int, budget_per_ring)
+		__field(unsigned int, rx_cleaned)
+		__field(unsigned int, tx_cleaned)
+		__field(int, rx_clean_complete)
+		__field(int, tx_clean_complete)
+		__field(int, irq_num)
+		__field(int, curr_cpu)
+		__string(qname, q->name)
+		__string(dev_name, napi->dev ? napi->dev->name : NO_DEV)
+		__bitmask(irq_affinity,	nr_cpumask_bits)
+	),
+
+	TP_fast_assign(
+		__entry->budget = budget;
+		__entry->budget_per_ring = budget_per_ring;
+		__entry->rx_cleaned = rx_cleaned;
+		__entry->tx_cleaned = tx_cleaned;
+		__entry->rx_clean_complete = rx_clean_complete;
+		__entry->tx_clean_complete = tx_clean_complete;
+		__entry->irq_num = q->irq_num;
+		__entry->curr_cpu = get_cpu();
+		__assign_str(qname, q->name);
+		__assign_str(dev_name, napi->dev ? napi->dev->name : NO_DEV);
+		__assign_bitmask(irq_affinity, cpumask_bits(&q->affinity_mask),
+				 nr_cpumask_bits);
+	),
+
+	TP_printk("i40e_napi_poll on dev %s q %s irq %d irq_mask %s curr_cpu %d "
+		  "budget %d bpr %d rx_cleaned %u tx_cleaned %u "
+		  "rx_clean_complete %d tx_clean_complete %d",
+		__get_str(dev_name), __get_str(qname), __entry->irq_num,
+		__get_bitmask(irq_affinity), __entry->curr_cpu, __entry->budget,
+		__entry->budget_per_ring, __entry->rx_cleaned, __entry->tx_cleaned,
+		__entry->rx_clean_complete, __entry->tx_clean_complete)
+);
+
 /* Events related to a vsi & ring */
 DECLARE_EVENT_CLASS(
 	i40e_tx_template,
drivers/net/ethernet/intel/i40e/i40e_txrx.c +4 -0
index 5901e58..f2b1b94 100644
@@ -2752,6 +2752,10 @@ int i40e_napi_poll(struct napi_struct *napi, int budget)
 			clean_complete = rx_clean_complete = false;
 	}
 
+	if (!i40e_enabled_xdp_vsi(vsi))
+		trace_i40e_napi_poll(napi, q_vector, budget, budget_per_ring, rx_cleaned,
+				     tx_cleaned, rx_clean_complete, tx_clean_complete);
+
 	/* If work not completed, return budget and polling will return */
 	if (!clean_complete) {
 		int cpu_id = smp_processor_id();
Signature · 1 line
2.7.4
Tested-by on v4 4/4
#8 lore raw

RE: [Intel-wired-lan] [next-queue v4 4/4] i40e: Add i40e_napi_poll tracepoint

Quoted 46 lines

-----Original Message-----
From: Intel-wired-lan <[email protected]> On Behalf Of Joe Damato
Sent: Saturday, October 8, 2022 3:09 AM
To: [email protected]
Cc: [email protected]; Joe Damato <[email protected]>; [email protected]; [email protected]
Subject: [Intel-wired-lan] [next-queue v4 4/4] i40e: Add i40e_napi_poll tracepoint

Add a tracepoint for i40e_napi_poll that allows users to get detailed information about the amount of work done. This information can help users better tune the correct NAPI parameters (like weight and budget), as well as debug NIC settings like rx-usecs and tx-usecs, etc.

When perf is attached, this tracepoint only fires when not in XDP mode.

An example of the output from this tracepoint:

$ sudo perf trace -e i40e:i40e_napi_poll -a --call-graph=fp --
libtraceevent_print

[..snip..]

388.258 :0/0 i40e:i40e_napi_poll(i40e_napi_poll on dev eth2 q i40e-eth2- TxRx-9 irq 346 irq_mask 00000000,00000000,00000000,00000000,00000000,00800000 curr_cpu 23 budget 64 bpr 64 rx_cleaned 28 tx_cleaned 0 rx_clean_complete 1 tx_clean_complete 1)

	i40e_napi_poll ([i40e])
	i40e_napi_poll ([i40e])
	__napi_poll ([kernel.kallsyms])
	net_rx_action ([kernel.kallsyms])
	__do_softirq ([kernel.kallsyms])
	common_interrupt ([kernel.kallsyms])
	asm_common_interrupt ([kernel.kallsyms])
	intel_idle_irq ([kernel.kallsyms])
	cpuidle_enter_state ([kernel.kallsyms])
	cpuidle_enter ([kernel.kallsyms])
	do_idle ([kernel.kallsyms])
	cpu_startup_entry ([kernel.kallsyms])
	[0x243fd8] ([kernel.kallsyms])
	secondary_startup_64_no_verify ([kernel.kallsyms])
Quoted 9 lines
---
 drivers/net/ethernet/intel/i40e/i40e_trace.h | 49
++++++++++++++++++++++++++++
drivers/net/ethernet/intel/i40e/i40e_txrx.c  |  4 +++
 2 files changed, 53 insertions(+)

_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan