Skip to contentSkip to conversation

enic: add ethtool get_channel support

PATCH Superseded Jon Kohler 3 participants 5 messages Jun 24, 2024 – Jun 26, 2024

Subsystems (3)

v2 of 4 Superseded Latest is v4 →

lore
Share & tools
lore

Jon Kohler

author
#0 lore raw

PATCH v2 1/1 enic: add ethtool get_channel support

Add .get_channel to enic_ethtool_ops to enable basic ethtool -l support to get the current channel configuration.

Note that the driver does not support dynamically changing queue configuration, so .set_channel is intentionally unused. Instead, users should use Cisco's hardware management tools (UCSM/IMC) to modify virtual interface card configuration out of band.

---
v1
- https://lore.kernel.org/netdev/[email protected]/T/#u
v1 -> v2:
- Addressed comments from Przemek and Jakub

---
 .../net/ethernet/cisco/enic/enic_ethtool.c    | 27 +++++++++++++++++++
 1 file changed, 27 insertions(+)
drivers/net/ethernet/cisco/enic/enic_ethtool.c +27 -0
index 241906697019..54f542238b4e 100644
@@ -608,6 +608,32 @@ static int enic_get_ts_info(struct net_device *netdev,
 	return 0;
 }
 
+static void enic_get_channels(struct net_device *netdev,
+			      struct ethtool_channels *channels)
+{
+	struct enic *enic = netdev_priv(netdev);
+
+	switch (vnic_dev_get_intr_mode(enic->vdev)) {
+	case VNIC_DEV_INTR_MODE_MSIX:
+		channels->max_rx = ENIC_RQ_MAX;
+		channels->max_tx = ENIC_WQ_MAX;
+		channels->rx_count = enic->rq_count;
+		channels->tx_count = enic->wq_count;
+		break;
+	case VNIC_DEV_INTR_MODE_MSI:
+		channels->max_rx = 1;
+		channels->max_tx = 1;
+		channels->rx_count = 1;
+		channels->tx_count = 1;
+		break;
+	case VNIC_DEV_INTR_MODE_INTX:
+		channels->max_combined = 1;
+		channels->combined_count = 1;
+	default:
+		break;
+	}
+}
+
 static const struct ethtool_ops enic_ethtool_ops = {
 	.supported_coalesce_params = ETHTOOL_COALESCE_USECS |
 				     ETHTOOL_COALESCE_USE_ADAPTIVE_RX |
@@ -632,6 +658,7 @@ static const struct ethtool_ops enic_ethtool_ops = {
 	.set_rxfh = enic_set_rxfh,
 	.get_link_ksettings = enic_get_ksettings,
 	.get_ts_info = enic_get_ts_info,
+	.get_channels = enic_get_channels,
 };
 
 void enic_set_ethtool_ops(struct net_device *netdev)
Signature · 1 line
2.43.0
Reviewed-by on v2 1/1
#1 lore raw
Quoted 10 lines

-----Original Message-----
From: Jon Kohler <[email protected]>
Sent: Tuesday, June 25, 2024 12:19 AM
To: Christian Benvenuti <[email protected]>; Satish Kharat <[email protected]>; David S. Miller <[email protected]>; Eric Dumazet <[email protected]>; Jakub Kicinski <[email protected]>; Paolo Abeni <[email protected]>; [email protected]; linux- [email protected]
Cc: Jon Kohler <[email protected]>
Subject: [PATCH v2] enic: add ethtool get_channel support

It is better to indicate the patch is for NET or NET-NEXT ex: [ PATCH net-next v2].

Quoted 75 lines

Add .get_channel to enic_ethtool_ops to enable basic ethtool -l support to get the current channel configuration. Note that the driver does not support dynamically changing queue configuration, so .set_channel is intentionally unused. Instead,
Add .get_channel to enic_ethtool_ops to enable basic ethtool -l support to get the current channel configuration.

Note that the driver does not support dynamically changing queue configuration, so .set_channel is intentionally unused. Instead, users should use Cisco's hardware management tools (UCSM/IMC) to modify virtual interface card configuration out of band.

---
v1
- https://urldefense.proofpoint.com/v2/url?u=https-
3A__lore.kernel.org_netdev_20240618160146.3900470-2D1-2Djon- 40nutanix.com_T_- 23u&d=DwIDAg&c=nKjWec2b6R0mOyPaz7xtfQ&r=c3MsgrR-U- HFhmFd6R4MWRZG-8QeikJn5PkjqMTpBSg&m=IIWcqk3E- aJ8g4EtpjIC9Jg_2T37OpnceY8mwAIuWEBmyGG9tHKaQp1rAgD5__2K&s=hTRY fTAQB9Tli-3DbKoTQkrJ2OxTBab-RqcKIJUjQTc&e=
v1 -> v2:
- Addressed comments from Przemek and Jakub

---
 .../net/ethernet/cisco/enic/enic_ethtool.c    | 27 +++++++++++++++++++
 1 file changed, 27 insertions(+)
drivers/net/ethernet/cisco/enic/enic_ethtool.c +26 -1
index 241906697019..54f542238b4e 100644
@@ -608,6 +608,32 @@ static int enic_get_ts_info(struct net_device *netdev,
 	return 0;
 }
 
+static void enic_get_channels(struct net_device *netdev,
+			      struct ethtool_channels *channels) {
+	struct enic *enic = netdev_priv(netdev);
+
+	switch (vnic_dev_get_intr_mode(enic->vdev)) {
+	case VNIC_DEV_INTR_MODE_MSIX:
+		channels->max_rx = ENIC_RQ_MAX;
+		channels->max_tx = ENIC_WQ_MAX;
+		channels->rx_count = enic->rq_count;
+		channels->tx_count = enic->wq_count;
+		break;
+	case VNIC_DEV_INTR_MODE_MSI:
+		channels->max_rx = 1;
+		channels->max_tx = 1;
+		channels->rx_count = 1;
+		channels->tx_count = 1;
+		break;
+	case VNIC_DEV_INTR_MODE_INTX:
+		channels->max_combined = 1;
+		channels->combined_count = 1;
+	default:
+		break;
+	}
+}
+
 static const struct ethtool_ops enic_ethtool_ops = {
 	.supported_coalesce_params = ETHTOOL_COALESCE_USECS |
 				     ETHTOOL_COALESCE_USE_ADAPTIVE_RX |
@@ -632,6 +658,7 @@ static const struct ethtool_ops enic_ethtool_ops = {
 	.set_rxfh = enic_set_rxfh,
 	.get_link_ksettings = enic_get_ksettings,
 	.get_ts_info = enic_get_ts_info,
+	.get_channels = enic_get_channels,
 };
 
 void enic_set_ethtool_ops(struct net_device *netdev)
--
Quoted 2 lines

2.43.0

author
#2 lore raw
Quoted 20 lines · Sai Krishna Gajula

!-------------------------------------------------------------------|
CAUTION: External Email

|-------------------------------------------------------------------!

Quoted 10 lines

-----Original Message-----
From: Jon Kohler <[email protected]>
Sent: Tuesday, June 25, 2024 12:19 AM
To: Christian Benvenuti <[email protected]>; Satish Kharat <[email protected]>; David S. Miller <[email protected]>; Eric Dumazet <[email protected]>; Jakub Kicinski <[email protected]>; Paolo Abeni <[email protected]>; [email protected]; linux- [email protected]
Cc: Jon Kohler <[email protected]>
Subject: [PATCH v2] enic: add ethtool get_channel support

Quoted 1 line

It is better to indicate the patch is for NET or NET-NEXT ex: [ PATCH net-next v2].

Thanks for the Reviewed-By and the tip here. Do you suggest I send out a v3 with an updated tag? My apologies for the confusion.

Quoted 79 lines
Quoted 76 lines

Add .get_channel to enic_ethtool_ops to enable basic ethtool -l support to get the current channel configuration. Note that the driver does not support dynamically changing queue configuration, so .set_channel is intentionally unused. Instead,
Add .get_channel to enic_ethtool_ops to enable basic ethtool -l support to get the current channel configuration.

Note that the driver does not support dynamically changing queue configuration, so .set_channel is intentionally unused. Instead, users should use Cisco's hardware management tools (UCSM/IMC) to modify virtual interface card configuration out of band.

---
v1
- https://urldefense.proofpoint.com/v2/url?u=https-
3A__lore.kernel.org_netdev_20240618160146.3900470-2D1-2Djon- 40nutanix.com_T_- 23u&d=DwIDAg&c=nKjWec2b6R0mOyPaz7xtfQ&r=c3MsgrR-U- HFhmFd6R4MWRZG-8QeikJn5PkjqMTpBSg&m=IIWcqk3E- aJ8g4EtpjIC9Jg_2T37OpnceY8mwAIuWEBmyGG9tHKaQp1rAgD5__2K&s=hTRY fTAQB9Tli-3DbKoTQkrJ2OxTBab-RqcKIJUjQTc&e=
v1 -> v2:
- Addressed comments from Przemek and Jakub
---
.../net/ethernet/cisco/enic/enic_ethtool.c | 27 +++++++++++++++++++

1 file changed, 27 insertions(+)
drivers/net/ethernet/cisco/enic/enic_ethtool.c +25 -0
index 241906697019..54f542238b4e 100644
@@ -608,6 +608,32 @@ static int enic_get_ts_info(struct net_device *netdev,
 return 0;
 }
 
+static void enic_get_channels(struct net_device *netdev,
+       struct ethtool_channels *channels) {
+ struct enic *enic = netdev_priv(netdev);
+
+ switch (vnic_dev_get_intr_mode(enic->vdev)) {
+ case VNIC_DEV_INTR_MODE_MSIX:
+ channels->max_rx = ENIC_RQ_MAX;
+ channels->max_tx = ENIC_WQ_MAX;
+ channels->rx_count = enic->rq_count;
+ channels->tx_count = enic->wq_count;
+ break;
+ case VNIC_DEV_INTR_MODE_MSI:
+ channels->max_rx = 1;
+ channels->max_tx = 1;
+ channels->rx_count = 1;
+ channels->tx_count = 1;
+ break;
+ case VNIC_DEV_INTR_MODE_INTX:
+ channels->max_combined = 1;
+ channels->combined_count = 1;
+ default:
+ break;
+ }
+}
+
 static const struct ethtool_ops enic_ethtool_ops = {

.supported_coalesce_params = ETHTOOL_COALESCE_USECS |
ETHTOOL_COALESCE_USE_ADAPTIVE_RX |

drivers/net/ethernet/cisco/enic/enic_ethtool.c +1 -1
@@ -632,6 +658,7 @@ static const struct ethtool_ops enic_ethtool_ops = {
 .set_rxfh = enic_set_rxfh,
 .get_link_ksettings = enic_get_ksettings,
 .get_ts_info = enic_get_ts_info,
+ .get_channels = enic_get_channels,
 };
 
 void enic_set_ethtool_ops(struct net_device *netdev)
--
Quoted 2 lines

2.43.0

maintainer
#3 lore raw
In reply to Jon Kohler
Quoted 9 lines · On Mon, 24 Jun 2024 11:49:00 -0700 Jon Kohler wrote:
drivers/net/ethernet/cisco/enic/enic_ethtool.c +9 -0
+	switch (vnic_dev_get_intr_mode(enic->vdev)) {
+	case VNIC_DEV_INTR_MODE_MSIX:
+		channels->max_rx = ENIC_RQ_MAX;
+		channels->max_tx = ENIC_WQ_MAX;
+		channels->rx_count = enic->rq_count;
+		channels->tx_count = enic->wq_count;
+		break;
+	case VNIC_DEV_INTR_MODE_MSI:
+		channels->max_rx = 1;

drivers/net/ethernet/cisco/enic/enic_ethtool.c static int enic_get_ts_info(struct net_device *netdev, · line 634

Quoted 10 lines
drivers/net/ethernet/cisco/enic/enic_ethtool.c +10 -0
+		channels->max_tx = 1;
+		channels->rx_count = 1;
+		channels->tx_count = 1;
+		break;
+	case VNIC_DEV_INTR_MODE_INTX:
+		channels->max_combined = 1;
+		channels->combined_count = 1;
+	default:
+		break;
+	}

sorry for not responding properly to your earlier email, but I think MSI should also be combined. What matters is whether the IRQ serves just one of {Rx, Tx} or both.

For MSI, I see:

1 . enic_dev_init() does:

	netif_napi_add(netdev, &enic->napi[0], enic_poll);
                                               ^^^^^^^^^

2. enic_request_intr() does

	request_irq(enic->pdev->irq, enic_isr_msi, ...
                                     ^^^^^^^^^^^^

3. enic_isr_msi() does

	napi_schedule_irqoff(&enic->napi[0]); 

thus matching the NAPI from step #1.

4. enic_poll() calls both enic_wq_service, and enic_rq_service

So it's combined, AFAICT, similar to INTX in the relevant parts.

Signature · 1 line
pw-bot: cr
author
#4 lore raw
Quoted 39 lines · Jakub Kicinski
Quoted 11 lines · On Mon, 24 Jun 2024 11:49:00 -0700 Jon Kohler wrote:

+ switch (vnic_dev_get_intr_mode(enic->vdev)) {
+ case VNIC_DEV_INTR_MODE_MSIX:
+ channels->max_rx = ENIC_RQ_MAX;
+ channels->max_tx = ENIC_WQ_MAX;
+ channels->rx_count = enic->rq_count;
+ channels->tx_count = enic->wq_count;
+ break;
+ case VNIC_DEV_INTR_MODE_MSI:
+ channels->max_rx = 1;
+ channels->max_tx = 1;
+ channels->rx_count = 1;

Quoted 8 lines

+ channels->tx_count = 1;
+ break;
+ case VNIC_DEV_INTR_MODE_INTX:
+ channels->max_combined = 1;
+ channels->combined_count = 1;
+ default:
+ break;
+ }

sorry for not responding properly to your earlier email, but I think MSI should also be combined. What matters is whether the IRQ serves just one of {Rx, Tx} or both.

For MSI, I see:

1 . enic_dev_init() does:

netif_napi_add(netdev, &enic->napi[0], enic_poll);
                                              ^^^^^^^^^

2. enic_request_intr() does
request_irq(enic->pdev->irq, enic_isr_msi, ...

                                    ^^^^^^^^^^^^
Quoted 7 lines

3. enic_isr_msi() does
napi_schedule_irqoff(&enic->napi[0]);
thus matching the NAPI from step #1.

4. enic_poll() calls both enic_wq_service, and enic_rq_service

So it's combined, AFAICT, similar to INTX in the relevant parts.

Ok, thanks for the tip, I appreciate it. I’ll send out a v3 shortly

Quoted 2 lines
Signature · 1 line
pw-bot: cr