Skip to contentSkip to conversation

net: mvpp2: disable link IRQ waiting for IDLE frames

PATCH In review Marcin Wojtas 3 participants 5 messages Feb 27, 2019 – Apr 23, 2019

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 net: mvpp2: disable link IRQ waiting for IDLE frames

Current version of the driver was configuring XLG MAC in a way to wait 3 IDLE frames before allowing for the link-up interrupt to be triggered. This resulted in an issue, preventing to detect the link change during RX traffic on the interface. Fix that.

Fixes 4bb043262878 ("net: mvpp2: phylink support")
---
 drivers/net/ethernet/marvell/mvpp2/mvpp2_main.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
drivers/net/ethernet/marvell/mvpp2/mvpp2_main.c +1 -2
index 16066c2..f1378f9 100644
@@ -4532,8 +4532,7 @@ static void mvpp2_xlg_config(struct mvpp2_port *port, unsigned int mode,
 		ctrl0 |= MVPP22_XLG_CTRL0_RX_FLOW_CTRL_EN;
 
 	ctrl4 &= ~MVPP22_XLG_CTRL4_MACMODSELECT_GMAC;
-	ctrl4 |= MVPP22_XLG_CTRL4_FWD_FC | MVPP22_XLG_CTRL4_FWD_PFC |
-		 MVPP22_XLG_CTRL4_EN_IDLE_CHECK;
+	ctrl4 |= MVPP22_XLG_CTRL4_FWD_FC | MVPP22_XLG_CTRL4_FWD_PFC;
 
 	writel(ctrl0, port->base + MVPP22_XLG_CTRL0_REG);
 	writel(ctrl4, port->base + MVPP22_XLG_CTRL4_REG);
Signature · 1 line
2.7.4
#1 lore raw
Quoted 5 lines · Marcin Wojtas

Current version of the driver was configuring XLG MAC in a way to wait 3 IDLE frames before allowing for the link-up interrupt to be triggered. This resulted in an issue, preventing to detect the link change during RX traffic on the interface. Fix that.

What about noise sensitivity? Does this now mean that on boards such as Macchiatobin Single Shot or Clearfog GT9k where the serdes is connected directly to a SFP cage, that noise on the serdes lines triggers a link indication?

Quoted 30 lines
Fixes 4bb043262878 ("net: mvpp2: phylink support")
---
 drivers/net/ethernet/marvell/mvpp2/mvpp2_main.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
drivers/net/ethernet/marvell/mvpp2/mvpp2_main.c +1 -2
index 16066c2..f1378f9 100644
@@ -4532,8 +4532,7 @@ static void mvpp2_xlg_config(struct mvpp2_port *port, unsigned int mode,
 		ctrl0 |= MVPP22_XLG_CTRL0_RX_FLOW_CTRL_EN;
 
 	ctrl4 &= ~MVPP22_XLG_CTRL4_MACMODSELECT_GMAC;
-	ctrl4 |= MVPP22_XLG_CTRL4_FWD_FC | MVPP22_XLG_CTRL4_FWD_PFC |
-		 MVPP22_XLG_CTRL4_EN_IDLE_CHECK;
+	ctrl4 |= MVPP22_XLG_CTRL4_FWD_FC | MVPP22_XLG_CTRL4_FWD_PFC;
 
 	writel(ctrl0, port->base + MVPP22_XLG_CTRL0_REG);
 	writel(ctrl4, port->base + MVPP22_XLG_CTRL4_REG);
Signature · 7 lines
2.7.4


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
Signature · 3 lines
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTC broadband for 0.8mile line in suburbia: sync at 12.1Mbps down 622kbps up
According to speedtest.net: 11.9Mbps down 500kbps up
#2 lore raw

Hi Russell,

czw., 28 lut 2019 o 10:36 Russell King - ARM Linux admin <[email protected]> napisał(a):

Quoted 8 lines · Marcin Wojtas
Quoted 5 lines

Current version of the driver was configuring XLG MAC in a way to wait 3 IDLE frames before allowing for the link-up interrupt to be triggered. This resulted in an issue, preventing to detect the link change during RX traffic on the interface. Fix that.

Quoted 5 lines

What about noise sensitivity? Does this now mean that on boards such as Macchiatobin Single Shot or Clearfog GT9k where the serdes is connected directly to a SFP cage, that noise on the serdes lines triggers a link indication?

Afaik waiting for idle frames is not required by any standard. In fact this feature of the XLG MAC caused a risk that after dropping the link during data transfer it would never go up again until remote peer stops it.

I tested it on 3 different boards with SerDes directly connected to the SFP cage and I'm also not aware of any reporte about noise-triggered false link interrupts in XLG MAC (proprietary Marvell mvpp2x driver did not use waiting 3 idle frames feature). I'd rather assume the MAC is not vulnerable to it, and only properly established link can trigger an event.

Best regards,
Marcin

Quoted 35 lines
Quoted 25 lines
Fixes 4bb043262878 ("net: mvpp2: phylink support")
---
 drivers/net/ethernet/marvell/mvpp2/mvpp2_main.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
drivers/net/ethernet/marvell/mvpp2/mvpp2_main.c +1 -3
index 16066c2..f1378f9 100644
@@ -4532,8 +4532,7 @@ static void mvpp2_xlg_config(struct mvpp2_port *port, unsigned int mode,
              ctrl0 |= MVPP22_XLG_CTRL0_RX_FLOW_CTRL_EN;
 
      ctrl4 &= ~MVPP22_XLG_CTRL4_MACMODSELECT_GMAC;
-     ctrl4 |= MVPP22_XLG_CTRL4_FWD_FC | MVPP22_XLG_CTRL4_FWD_PFC |
-              MVPP22_XLG_CTRL4_EN_IDLE_CHECK;
+     ctrl4 |= MVPP22_XLG_CTRL4_FWD_FC | MVPP22_XLG_CTRL4_FWD_PFC;
 
      writel(ctrl0, port->base + MVPP22_XLG_CTRL0_REG);
      writel(ctrl4, port->base + MVPP22_XLG_CTRL4_REG);
--

2.7.4

Quoted 5 lines

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

--
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTC broadband for 0.8mile line in suburbia: sync at 12.1Mbps down 622kbps up According to speedtest.net: 11.9Mbps down 500kbps up

#3 lore raw

Hi David, Rusell,

czw., 28 lut 2019 o 14:58 Marcin Wojtas <[email protected]> napisał(a):

Quoted 24 lines

Hi Russell,

czw., 28 lut 2019 o 10:36 Russell King - ARM Linux admin <[email protected]> napisał(a):

Quoted 8 lines · Marcin Wojtas
Quoted 5 lines

Current version of the driver was configuring XLG MAC in a way to wait 3 IDLE frames before allowing for the link-up interrupt to be triggered. This resulted in an issue, preventing to detect the link change during RX traffic on the interface. Fix that.

Quoted 5 lines

What about noise sensitivity? Does this now mean that on boards such as Macchiatobin Single Shot or Clearfog GT9k where the serdes is connected directly to a SFP cage, that noise on the serdes lines triggers a link indication?

Afaik waiting for idle frames is not required by any standard. In fact this feature of the XLG MAC caused a risk that after dropping the link during data transfer it would never go up again until remote peer stops it.

Quoted 7 lines

I tested it on 3 different boards with SerDes directly connected to the SFP cage and I'm also not aware of any reporte about noise-triggered false link interrupts in XLG MAC (proprietary Marvell mvpp2x driver did not use waiting 3 idle frames feature). I'd rather assume the MAC is not vulnerable to it, and only properly established link can trigger an event.

This patch seems to get forgotten - any objections to merge it?

Best regards,
Marcin

Quoted 38 lines

Best regards,
Marcin

Quoted 35 lines
Quoted 25 lines
Fixes 4bb043262878 ("net: mvpp2: phylink support")
---
 drivers/net/ethernet/marvell/mvpp2/mvpp2_main.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
drivers/net/ethernet/marvell/mvpp2/mvpp2_main.c +1 -3
index 16066c2..f1378f9 100644
@@ -4532,8 +4532,7 @@ static void mvpp2_xlg_config(struct mvpp2_port *port, unsigned int mode,
              ctrl0 |= MVPP22_XLG_CTRL0_RX_FLOW_CTRL_EN;
 
      ctrl4 &= ~MVPP22_XLG_CTRL4_MACMODSELECT_GMAC;
-     ctrl4 |= MVPP22_XLG_CTRL4_FWD_FC | MVPP22_XLG_CTRL4_FWD_PFC |
-              MVPP22_XLG_CTRL4_EN_IDLE_CHECK;
+     ctrl4 |= MVPP22_XLG_CTRL4_FWD_FC | MVPP22_XLG_CTRL4_FWD_PFC;
 
      writel(ctrl0, port->base + MVPP22_XLG_CTRL0_REG);
      writel(ctrl4, port->base + MVPP22_XLG_CTRL4_REG);
--

2.7.4

Quoted 5 lines

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

--
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTC broadband for 0.8mile line in suburbia: sync at 12.1Mbps down 622kbps up According to speedtest.net: 11.9Mbps down 500kbps up

maintainer
#4 lore raw

From: Marcin Wojtas <[email protected]>
Date: Tue, 23 Apr 2019 18:08:00 +0200

Quoted 35 lines

Hi David, Rusell,

czw., 28 lut 2019 o 14:58 Marcin Wojtas <[email protected]> napisał(a):

Quoted 24 lines

Hi Russell,

czw., 28 lut 2019 o 10:36 Russell King - ARM Linux admin <[email protected]> napisał(a):

Quoted 8 lines · Marcin Wojtas
Quoted 5 lines

Current version of the driver was configuring XLG MAC in a way to wait 3 IDLE frames before allowing for the link-up interrupt to be triggered. This resulted in an issue, preventing to detect the link change during RX traffic on the interface. Fix that.

Quoted 5 lines

What about noise sensitivity? Does this now mean that on boards such as Macchiatobin Single Shot or Clearfog GT9k where the serdes is connected directly to a SFP cage, that noise on the serdes lines triggers a link indication?

Afaik waiting for idle frames is not required by any standard. In fact this feature of the XLG MAC caused a risk that after dropping the link during data transfer it would never go up again until remote peer stops it.

Quoted 7 lines

I tested it on 3 different boards with SerDes directly connected to the SFP cage and I'm also not aware of any reporte about noise-triggered false link interrupts in XLG MAC (proprietary Marvell mvpp2x driver did not use waiting 3 idle frames feature). I'd rather assume the MAC is not vulnerable to it, and only properly established link can trigger an event.

Quoted 1 line

This patch seems to get forgotten - any objections to merge it?

If there is agreement, please repost it.

Thank you.