diff mbox series

[net-next,v6,2/3] net: dsa: microchip: ksz8: Add function to configure ports with integrated PHYs

Message ID 20231123112051.713142-3-o.rempel@pengutronix.de (mailing list archive)
State Superseded
Delegated to: Netdev Maintainers
Headers show
Series Fine-Tune Flow Control and Speed Configurations in Microchip KSZ8xxx DSA Driver | expand

Checks

Context Check Description
netdev/series_format success Posting correctly formatted
netdev/codegen success Generated files up to date
netdev/tree_selection success Clearly marked for net-next
netdev/fixes_present success Fixes tag not required for -next series
netdev/header_inline success No static functions without inline keyword in header files
netdev/build_32bit success Errors and warnings before: 1127 this patch: 1127
netdev/cc_maintainers success CCed 9 of 9 maintainers
netdev/build_clang success Errors and warnings before: 1154 this patch: 1154
netdev/verify_signedoff success Signed-off-by tag matches author and committer
netdev/deprecated_api success None detected
netdev/check_selftest success No net selftest shell script
netdev/verify_fixes success No Fixes tag
netdev/build_allmodconfig_warn success Errors and warnings before: 1154 this patch: 1154
netdev/checkpatch success total: 0 errors, 0 warnings, 0 checks, 86 lines checked
netdev/build_clang_rust success No Rust files in patch. Skipping build
netdev/kdoc success Errors and warnings before: 0 this patch: 0
netdev/source_inline success Was 0 now: 0

Commit Message

Oleksij Rempel Nov. 23, 2023, 11:20 a.m. UTC
This patch introduces the function 'ksz8_phy_port_link_up' to the
Microchip KSZ8xxx driver. This function is responsible for setting up
flow control and duplex settings for the ports that are integrated with
PHYs.

The KSZ8795 switch supports asymmetric pause control, which can't be
fully utilized since a single bit controls both RX and TX pause. Despite
this, the flow control can be adjusted based on the auto-negotiation
process, taking into account the capabilities of both link partners.

On the other hand, the KSZ8873's PORT_FORCE_FLOW_CTRL bit can be set by
the hardware bootstrap, which ignores the auto-negotiation result.
Therefore, even in auto-negotiation mode, we need to ensure that this
bit is correctly set.

When auto-negotiation isn't in use, we enforce symmetric pause control
for the KSZ8795 switch.

Please note, forcing flow control disable on a port while still
advertising pause support isn't possible. While this scenario
might not be practical or desired, it's important to be aware of this
limitation when working with the KSZ8873 and similar devices.

Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
---
 drivers/net/dsa/microchip/ksz8795.c | 74 +++++++++++++++++++++++++++++
 1 file changed, 74 insertions(+)

Comments

Russell King (Oracle) Nov. 23, 2023, 11:55 a.m. UTC | #1
On Thu, Nov 23, 2023 at 12:20:50PM +0100, Oleksij Rempel wrote:
> This patch introduces the function 'ksz8_phy_port_link_up' to the
> Microchip KSZ8xxx driver. This function is responsible for setting up
> flow control and duplex settings for the ports that are integrated with
> PHYs.
> 
> The KSZ8795 switch supports asymmetric pause control, which can't be
> fully utilized since a single bit controls both RX and TX pause. Despite
> this, the flow control can be adjusted based on the auto-negotiation
> process, taking into account the capabilities of both link partners.
> 
> On the other hand, the KSZ8873's PORT_FORCE_FLOW_CTRL bit can be set by
> the hardware bootstrap, which ignores the auto-negotiation result.
> Therefore, even in auto-negotiation mode, we need to ensure that this
> bit is correctly set.
> 
> When auto-negotiation isn't in use, we enforce symmetric pause control
> for the KSZ8795 switch.

This doesn't sound right.

I would suggest that if PORT_FORCE_FLOW_CTRL is cleared, and autoneg is
disabled, the result will be that flow control will be disabled.

Also, I think PORT_FORCE_FLOW_CTRL should be used in combination with
permit_pause_to_mac, which you can get at in mac_config() using
state->pause & MLO_PAUSE_AN.

So, what I would suggest is that in mac_config(), the driver stores
in its private data something like:

	dev->manual_flow[port] = !(state->pause & MLO_PAUSE_AN);

And then in mac_link_up(), you can examine this, and if manual_flow
for the port is set _and_ tx_pause is also set, then you can set
PORT_FORCE_FLOW_CTRL.

I think this will give a better approximation to the intent of the
user API (assuming symmetric pause only):

Autoneg		Pause Autoneg	rx,tx	PORT_FORCE_FLOW_CTRL
1		1		x	0
0		1		x	0 (flow control probably disabled)
x		0		1	1 (flow control force enabled)
1		0		0	0 (flow control still depends on
					   aneg result due to hardware)
0		0		0	0 (flow control probably disabled)

Whereas, what I think you're proposing is:

Autoneg		Pause Autoneg	rx,tx	PORT_FORCE_FLOW_CTRL
1		1		x	0
1		0		x	0 (flow control still depends on
					   aneg result)
0		x		1	1 (flow control force enabled)
0		x		0	0 (flow control probably disabled)

The difference is that flow control can be forced with my suggestion
when Autoneg is still enabled, but the pause autoneg bit is cleared
and we want flow control enabled.

Thanks.
diff mbox series

Patch

diff --git a/drivers/net/dsa/microchip/ksz8795.c b/drivers/net/dsa/microchip/ksz8795.c
index 3ed3ba69c3ee..a1aa69afd750 100644
--- a/drivers/net/dsa/microchip/ksz8795.c
+++ b/drivers/net/dsa/microchip/ksz8795.c
@@ -1528,6 +1528,78 @@  void ksz8_config_cpu_port(struct dsa_switch *ds)
 	}
 }
 
+/**
+ * ksz8_phy_port_link_up - Configures ports with integrated PHYs
+ * @dev: The KSZ device instance.
+ * @port: The port number to configure.
+ * @duplex: The desired duplex mode.
+ * @tx_pause: If true, enables transmit pause.
+ * @rx_pause: If true, enables receive pause.
+ *
+ * Description:
+ * The function configures flow control settings for a given port based on the
+ * desired settings and current duplex mode.
+ *
+ * According to the KSZ8873 datasheet, the PORT_FORCE_FLOW_CTRL bit in the
+ * Port Control 2 register (0x1A for Port 1, 0x22 for Port 2, 0x32 for Port 3)
+ * determines how flow control is handled on the port:
+ *    "1 = will always enable full-duplex flow control on the port, regardless
+ *         of AN result.
+ *     0 = full-duplex flow control is enabled based on AN result."
+ *
+ * This means that the flow control behavior depends on the state of this bit:
+ * - If PORT_FORCE_FLOW_CTRL is set to 1, the switch will ignore AN results and
+ *   force flow control on the port.
+ * - If PORT_FORCE_FLOW_CTRL is set to 0, the switch will enable or disable
+ *   flow control based on the AN results.
+ *
+ * However, there is a potential limitation in this configuration. It is
+ * currently not possible to force disable flow control on a port if we still
+ * advertise pause support. While such a configuration is not currently
+ * supported by Linux, and may not make practical sense, it's important to be
+ * aware of this limitation when working with the KSZ8873 and similar devices.
+ */
+static void ksz8_phy_port_link_up(struct ksz_device *dev, int port, int duplex,
+				  bool tx_pause, bool rx_pause)
+{
+	const u16 *regs = dev->info->regs;
+	bool aneg_en;
+	u8 sctrl = 0;
+	u8 fctrl;
+	int ret;
+
+	/* The KSZ8795 switch differs from the KSZ8873 by supporting
+	 * asymmetric pause control. However, since a single bit is used to
+	 * control both RX and TX pause, we can't enforce asymmetric pause
+	 * control - both TX and RX pause will be either enabled or disabled
+	 * together.
+	 *
+	 * If auto-negotiation is enabled, we usually allow the flow control to
+	 * be determined by the auto-negotiation process based on the
+	 * capabilities of both link partners. However, for KSZ8873, the
+	 * PORT_FORCE_FLOW_CTRL bit may be set by the hardware bootstrap,
+	 * ignoring the auto-negotiation result. Thus, even in auto-negotiation
+	 * mode, we need to ensure that the PORT_FORCE_FLOW_CTRL bit is
+	 * properly cleared.
+	 *
+	 * In the absence of auto-negotiation, we will enforce symmetric
+	 * pause control for both variants of switches - KSZ8873 and KSZ8795.
+	 */
+	ret = ksz_pread8(dev, port, regs[P_FORCE_CTRL], &fctrl);
+	if (ret)
+		return;
+
+	if (ksz_is_ksz88x3(dev))
+		aneg_en = fctrl & PORT_AUTO_NEG_ENABLE;
+	else
+		aneg_en = !(fctrl & PORT_AUTO_NEG_DISABLE);
+
+	if (!aneg_en && (tx_pause || rx_pause))
+		sctrl |= PORT_FORCE_FLOW_CTRL;
+
+	ksz_prmw8(dev, port, regs[P_STP_CTRL], PORT_FORCE_FLOW_CTRL, sctrl);
+}
+
 /**
  * ksz8_cpu_port_link_up - Configures the CPU port of the switch.
  * @dev: The KSZ device instance.
@@ -1577,6 +1649,8 @@  void ksz8_phylink_mac_link_up(struct ksz_device *dev, int port,
 	 */
 	if (dev->cpu_port == port)
 		ksz8_cpu_port_link_up(dev, speed, duplex, tx_pause, rx_pause);
+	else if (dev->info->internal_phy[port])
+		ksz8_phy_port_link_up(dev, port, duplex, tx_pause, rx_pause);
 }
 
 static int ksz8_handle_global_errata(struct dsa_switch *ds)