[iwl-net,v3] igb: Fix not clearing TimeSync interrupts for 82580

Commit Message

Daiwei Li Aug. 14, 2024, 4:55 a.m. UTC

82580 NICs have a hardware bug that makes it
necessary to write into the TSICR (TimeSync Interrupt Cause) register
to clear it:
https://lore.kernel.org/all/[e-mail]/

Add a conditional so only for 82580 we write into the TSICR register,
so we don't risk losing events for other models.

Without this change, when running ptp4l with an Intel 82580 card,
I get the following output:

> timed out while polling for tx timestamp increasing tx_timestamp_timeout or
> increasing kworker priority may correct this issue, but a driver bug likely
> causes it

This goes away with this change.

This (partially) reverts commit ee14cc9ea19b ("igb: Fix missing time sync events").

Fixes: ee14cc9ea19b ("igb: Fix missing time sync events")
Closes: https://lore.kernel.org/intel-wired-lan/CAN0jFd1kO0MMtOh8N2Ztxn6f7vvDKp2h507sMryobkBKe=xk=[e-mail]/
Tested-by: Daiwei Li <[e-mail]>
Suggested-by: Vinicius Costa Gomes <[e-mail]>
Signed-off-by: Daiwei Li <[e-mail]>
---
 drivers/net/ethernet/intel/igb/igb_main.c | 10 ++++++++++
 1 file changed, 10 insertions(+)

Comments

Simon Horman Aug. 14, 2024, 1:50 p.m. UTC | #1

On Tue, Aug 13, 2024 at 09:55:53PM -0700, Daiwei Li wrote:
> 82580 NICs have a hardware bug that makes it
> necessary to write into the TSICR (TimeSync Interrupt Cause) register
> to clear it:
> https://lore.kernel.org/all/[e-mail]/
> 
> Add a conditional so only for 82580 we write into the TSICR register,
> so we don't risk losing events for other models.
> 
> Without this change, when running ptp4l with an Intel 82580 card,
> I get the following output:
> 
> > timed out while polling for tx timestamp increasing tx_timestamp_timeout or
> > increasing kworker priority may correct this issue, but a driver bug likely
> > causes it
> 
> This goes away with this change.
> 
> This (partially) reverts commit ee14cc9ea19b ("igb: Fix missing time sync events").
> 
> Fixes: ee14cc9ea19b ("igb: Fix missing time sync events")
> Closes: https://lore.kernel.org/intel-wired-lan/CAN0jFd1kO0MMtOh8N2Ztxn6f7vvDKp2h507sMryobkBKe=xk=[e-mail]/
> Tested-by: Daiwei Li <[e-mail]>
> Suggested-by: Vinicius Costa Gomes <[e-mail]>
> Signed-off-by: Daiwei Li <[e-mail]>

Reviewed-by: Simon Horman <[e-mail]>

Vinicius Costa Gomes Aug. 16, 2024, 1:07 a.m. UTC | #2

Daiwei Li <[e-mail]> writes:

> 82580 NICs have a hardware bug that makes it
> necessary to write into the TSICR (TimeSync Interrupt Cause) register
> to clear it:
> https://lore.kernel.org/all/[e-mail]/
>
> Add a conditional so only for 82580 we write into the TSICR register,
> so we don't risk losing events for other models.
>
> Without this change, when running ptp4l with an Intel 82580 card,
> I get the following output:
>
>> timed out while polling for tx timestamp increasing tx_timestamp_timeout or
>> increasing kworker priority may correct this issue, but a driver bug likely
>> causes it
>
> This goes away with this change.
>
> This (partially) reverts commit ee14cc9ea19b ("igb: Fix missing time sync events").
>
> Fixes: ee14cc9ea19b ("igb: Fix missing time sync events")
> Closes: https://lore.kernel.org/intel-wired-lan/CAN0jFd1kO0MMtOh8N2Ztxn6f7vvDKp2h507sMryobkBKe=xk=[e-mail]/
> Tested-by: Daiwei Li <[e-mail]>
> Suggested-by: Vinicius Costa Gomes <[e-mail]>
> Signed-off-by: Daiwei Li <[e-mail]>

Acked-by: Vinicius Costa Gomes <[e-mail]>


Cheers,

Kurt Kanzenbach Aug. 16, 2024, 9:39 a.m. UTC | #3

On Tue Aug 13 2024, Daiwei Li wrote:
> 82580 NICs have a hardware bug that makes it
> necessary to write into the TSICR (TimeSync Interrupt Cause) register
> to clear it:
> https://lore.kernel.org/all/[e-mail]/
>
> Add a conditional so only for 82580 we write into the TSICR register,
> so we don't risk losing events for other models.
>
> Without this change, when running ptp4l with an Intel 82580 card,
> I get the following output:
>
>> timed out while polling for tx timestamp increasing tx_timestamp_timeout or
>> increasing kworker priority may correct this issue, but a driver bug likely
>> causes it
>
> This goes away with this change.
>
> This (partially) reverts commit ee14cc9ea19b ("igb: Fix missing time sync events").
>
> Fixes: ee14cc9ea19b ("igb: Fix missing time sync events")
> Closes: https://lore.kernel.org/intel-wired-lan/CAN0jFd1kO0MMtOh8N2Ztxn6f7vvDKp2h507sMryobkBKe=xk=[e-mail]/
> Tested-by: Daiwei Li <[e-mail]>
> Suggested-by: Vinicius Costa Gomes <[e-mail]>
> Signed-off-by: Daiwei Li <[e-mail]>

Reviewed-by: Kurt Kanzenbach <[e-mail]>

Pucha, HimasekharX Reddy Aug. 22, 2024, 1:05 p.m. UTC | #4

> -----Original Message-----
> From: Intel-wired-lan <[e-mail]> On Behalf Of Daiwei Li
> Sent: Wednesday, August 14, 2024 10:26 AM
> To: [e-mail]
> Cc: Daiwei Li <[e-mail]>; Neftin, Sasha <[e-mail]>; Gomes, Vinicius <[e-mail]>; [e-mail]; [e-mail]; [e-mail]; [e-mail]; [e-mail]; Nguyen, Anthony L <[e-mail]>; Kitszel, Przemyslaw <[e-mail]>; [e-mail]; [e-mail]; [e-mail]
> Subject: [Intel-wired-lan] [PATCH iwl-net v3] igb: Fix not clearing TimeSync interrupts for 82580
>
> 82580 NICs have a hardware bug that makes it necessary to write into the TSICR (TimeSync Interrupt Cause) register to clear it:
> https://lore.kernel.org/all/[e-mail]/
>
> Add a conditional so only for 82580 we write into the TSICR register, so we don't risk losing events for other models.
>
> Without this change, when running ptp4l with an Intel 82580 card, I get the following output:
>
> > timed out while polling for tx timestamp increasing 
> > tx_timestamp_timeout or increasing kworker priority may correct this 
> > issue, but a driver bug likely causes it
>
> This goes away with this change.
>
> This (partially) reverts commit ee14cc9ea19b ("igb: Fix missing time sync events").
>
> Fixes: ee14cc9ea19b ("igb: Fix missing time sync events")
> Closes: https://lore.kernel.org/intel-wired-lan/CAN0jFd1kO0MMtOh8N2Ztxn6f7vvDKp2h507sMryobkBKe=xk=[e-mail]/
> Tested-by: Daiwei Li <[e-mail]>
> Suggested-by: Vinicius Costa Gomes <[e-mail]>
> Signed-off-by: Daiwei Li <[e-mail]>
> ---
>  drivers/net/ethernet/intel/igb/igb_main.c | 10 ++++++++++
>  1 file changed, 10 insertions(+)
>

Tested-by: Pucha Himasekhar Reddy <[e-mail]> (A Contingent worker at Intel)

Patch

diff --git a/drivers/net/ethernet/intel/igb/igb_main.c b/drivers/net/ethernet/intel/igb/igb_main.c
index ada42ba63549..69d7e8b16437 100644
--- a/drivers/net/ethernet/intel/igb/igb_main.c
+++ b/drivers/net/ethernet/intel/igb/igb_main.c
@@ -6984,9 +6984,19 @@  static void igb_extts(struct igb_adapter *adapter, int tsintr_tt)
 static void igb_tsync_interrupt(struct igb_adapter *adapter)
 {
 	struct e1000_hw *hw = &adapter->hw;
+	const u32 mask = (TSINTR_SYS_WRAP | E1000_TSICR_TXTS |
+			  TSINTR_TT0 | TSINTR_TT1 |
+			  TSINTR_AUTT0 | TSINTR_AUTT1);
 	u32 tsicr = rd32(E1000_TSICR);
 	struct ptp_clock_event event;
 
+	if (hw->mac.type == e1000_82580) {
+		/* 82580 has a hardware bug that requires an explicit
+		 * write to clear the TimeSync interrupt cause.
+		 */
+		wr32(E1000_TSICR, tsicr & mask);
+	}
+
 	if (tsicr & TSINTR_SYS_WRAP) {
 		event.type = PTP_CLOCK_PPS;
 		if (adapter->ptp_caps.pps)