[PATCH 0/5] Introduce DSA Ethernet switch class and Felix driver

classic Classic list List threaded Threaded
16 messages Options
Reply | Threaded
Open this post in threaded view
|

[PATCH 0/5] Introduce DSA Ethernet switch class and Felix driver

Claudiu Manoil-2
DSA stands for Distributed Switch Architecture and it is a subsystem
introduced in the Linux kernel to support switches that:
- have an Ethernet link up to the CPU
- use some form of tagging to identify the source/destination port for
  Rx/Tx
- may be cascaded in tree-like structures.

DSA is described in depth here:
https://www.kernel.org/doc/Documentation/networking/dsa/dsa.txt

This patch set introduces a DSA class in U-Boot to support drivers of DSA
switches.  DSA drivers have to implement the following ops:
- enable/disable of switch ports,
- insert a tag in frames being transmitted, used by the switch to select
  the egress port,
- parse a tag in frames being received, used for Rx traffic.

DSA class code deals with presentation of switch ports as Ethernet
interfaces, deals with the master Ethernet device for I/O and helps with
parsing of the DT assuming the structure follows the DSA kernel binding.

Support for switch cascading is not included yet.

In the sandbox environment, the DSA sandbox driver, the switch ports and
master eth interface look like this:
=> dm tree
 Class     Index  Probed  Driver                Name
-----------------------------------------------------------
[...]
 eth           4  [ + ]   eth_sandbox           |-- dsa-test-eth
 dsa           0  [ + ]   dsa_sandbox           |-- dsa-test
 eth           5  [ + ]   dsa-port              |   |-- lan0
 eth           6  [ + ]   dsa-port              |   `-- lan1

=> setenv ethact lan1
=> ping 1.2.3.5
Using lan1 device
host 1.2.3.5 is alive
=>

This patch set also introduces a driver for the Ethernet switch integrated
into NXP LS1028A, called Felix.  The switch has 4 front panel ports, I/O
to/from it is done though an ENETC Ethernet interface and meta-data is
carried between the switch and the driver though an additional header
pre-pended to the original frame.
Network commands like tftp can be used on these front panel ports.  The
ports are disabled unless used so they do not cause issues on network
topologies that include loops.

Felix as seen on LS1028A RDB:
=> dm tree
 Class     Index  Probed  Driver                Name
-----------------------------------------------------------
[...]
 dsa           0  [ + ]   felix-switch          |   |-- felix-switch
 eth           4  [ + ]   dsa-port              |   |   |-- swp0
 eth           5  [ + ]   dsa-port              |   |   |-- swp1
 eth           6  [ + ]   dsa-port              |   |   |-- swp2
 eth           7  [ + ]   dsa-port              |   |   `-- swp3

=> mdio list
[...]
10 - Vitesse VSC8514 <--> swp0
11 - Vitesse VSC8514 <--> swp1
12 - Vitesse VSC8514 <--> swp2
13 - Vitesse VSC8514 <--> swp3

NOTE:
This patchset is a major rework of the dsa-class code since the last
submission from May 5th:
https://patchwork.ozlabs.org/project/uboot/cover/1588700588-8587-1-git-send-email-claudiu.manoil@.../
The basic concepts and data path operation (tagging) in the DSA class
code remain the same as in the initial patchset from Alex, however the
external API has been changed significantly (simplified), the driver
model integration has been improved to the point that the DSA class
code no longer needs to allocate extra memory internally (via malloc),
reduced memory footprint, internal state data moved from the external
API and internalized, cleaner external API, internal code reworked,
completly reworked DSA sandbox driver and unit tests for better coverage
and to integrate better with the eth sandbox driver and tests, etc.
The DSA class code is now ported on top of the latest u-boot master
branch (obviously), and I tested it thoroughly with the sandbox driver
and unit tests.  The felix driver has been tested with an older uboot,
since there are some issues booting the latest upstream uboot on a
LS1028A RDB board.  Note that the felix driver itself didn't change
since the last submission, except for the DSA API function calls.

Alex Marginean (3):
  drivers: net: Add Felix DSA switch driver
  arm: dts: ls1028a: Add Ethernet switch node and dependencies
  configs: ls1028a: Enable the Ethernet switch driver in defconfig

Claudiu Manoil (2):
  net: Introduce DSA class for Ethernet switches
  sandbox: Add a DSA sandbox driver and unit test

 arch/Kconfig                                 |   1 +
 arch/arm/dts/fsl-ls1028a-rdb.dts             |  36 ++
 arch/arm/dts/fsl-ls1028a.dtsi                |  55 +-
 arch/sandbox/dts/test.dts                    |  40 ++
 configs/ls1028aqds_tfa_SECURE_BOOT_defconfig |   2 +
 configs/ls1028aqds_tfa_defconfig             |   2 +
 configs/ls1028ardb_tfa_SECURE_BOOT_defconfig |   2 +
 configs/ls1028ardb_tfa_defconfig             |   2 +
 drivers/net/Kconfig                          |  23 +
 drivers/net/Makefile                         |   1 +
 drivers/net/dsa_sandbox.c                    | 181 +++++++
 drivers/net/fsl_enetc.h                      |   5 +
 drivers/net/mscc_eswitch/Kconfig             |   8 +
 drivers/net/mscc_eswitch/Makefile            |   1 +
 drivers/net/mscc_eswitch/felix_switch.c      | 432 ++++++++++++++++
 include/configs/sandbox.h                    |   2 +
 include/dm/uclass-id.h                       |   1 +
 include/net.h                                |   6 +
 include/net/dsa.h                            | 206 ++++++++
 net/Makefile                                 |   1 +
 net/dsa-uclass.c                             | 517 +++++++++++++++++++
 test/dm/Makefile                             |   1 +
 test/dm/dsa.c                                |  91 ++++
 test/dm/eth.c                                |  10 +-
 24 files changed, 1620 insertions(+), 6 deletions(-)
 create mode 100644 drivers/net/dsa_sandbox.c
 create mode 100644 drivers/net/mscc_eswitch/felix_switch.c
 create mode 100644 include/net/dsa.h
 create mode 100644 net/dsa-uclass.c
 create mode 100644 test/dm/dsa.c

--
2.17.1

Reply | Threaded
Open this post in threaded view
|

[PATCH 1/5] net: Introduce DSA class for Ethernet switches

Claudiu Manoil-2
DSA stands for Distributed Switch Architecture and it covers switches that
are connected to the CPU through an Ethernet link and generally use frame
tags to pass information about the source/destination ports to/from CPU.
Front panel ports are presented as regular ethernet devices in U-Boot and
they are expected to support the typical networking commands.
DSA switches may be cascaded, DSA class code does not currently support
this.

Signed-off-by: Alex Marginean <[hidden email]>
Signed-off-by: Claudiu Manoil <[hidden email]>
---
 drivers/net/Kconfig    |  14 ++
 include/dm/uclass-id.h |   1 +
 include/net.h          |   6 +
 include/net/dsa.h      | 206 ++++++++++++++++
 net/Makefile           |   1 +
 net/dsa-uclass.c       | 517 +++++++++++++++++++++++++++++++++++++++++
 6 files changed, 745 insertions(+)
 create mode 100644 include/net/dsa.h
 create mode 100644 net/dsa-uclass.c

diff --git a/drivers/net/Kconfig b/drivers/net/Kconfig
index 3a5e036880..382224d04c 100644
--- a/drivers/net/Kconfig
+++ b/drivers/net/Kconfig
@@ -37,6 +37,20 @@ config DM_MDIO_MUX
   This is currently implemented in net/mdio-mux-uclass.c
   Look in include/miiphy.h for details.
 
+config DM_DSA
+ bool "Enable Driver Model for DSA switches"
+ depends on DM_ETH && DM_MDIO
+ help
+  Enable driver model for DSA switches
+
+  Adds UCLASS_DSA class supporting switches that follow the Distributed
+  Switch Architecture (DSA).  These switches rely on the presence of a
+  management switch port connected to an Ethernet controller capable of
+  receiving frames from the switch.  This host Ethernet controller is
+  called the "master" Ethernet interface in DSA terminology.
+  This is currently implemented in net/dsa-uclass.c, refer to
+  include/net/dsa.h for API details.
+
 config MDIO_SANDBOX
  depends on DM_MDIO && SANDBOX
  default y
diff --git a/include/dm/uclass-id.h b/include/dm/uclass-id.h
index ae4425d7a5..d75de368c5 100644
--- a/include/dm/uclass-id.h
+++ b/include/dm/uclass-id.h
@@ -46,6 +46,7 @@ enum uclass_id {
  UCLASS_DISPLAY, /* Display (e.g. DisplayPort, HDMI) */
  UCLASS_DSI_HOST, /* Display Serial Interface host */
  UCLASS_DMA, /* Direct Memory Access */
+ UCLASS_DSA, /* Distributed (Ethernet) Switch Architecture */
  UCLASS_EFI, /* EFI managed devices */
  UCLASS_ETH, /* Ethernet device */
  UCLASS_ETH_PHY, /* Ethernet PHY device */
diff --git a/include/net.h b/include/net.h
index 13da69b7c1..b95d6a6f60 100644
--- a/include/net.h
+++ b/include/net.h
@@ -499,7 +499,13 @@ struct icmp_hdr {
  * maximum packet size and multiple of 32 bytes =  1536
  */
 #define PKTSIZE 1522
+#ifndef CONFIG_DM_DSA
 #define PKTSIZE_ALIGN 1536
+#else
+/* Maximum DSA tagging overhead (headroom and/or tailroom) */
+#define DSA_MAX_OVR 256
+#define PKTSIZE_ALIGN (1536 + DSA_MAX_OVR)
+#endif
 
 /*
  * Maximum receive ring size; that is, the number of packets
diff --git a/include/net/dsa.h b/include/net/dsa.h
new file mode 100644
index 0000000000..a2bf4851df
--- /dev/null
+++ b/include/net/dsa.h
@@ -0,0 +1,206 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * Copyright 2019-2021 NXP
+ */
+
+#ifndef __DSA_H__
+#define __DSA_H__
+
+#include <phy.h>
+#include <net.h>
+
+/**
+ * DSA stands for Distributed Switch Architecture and it is infrastructure
+ * intended to support drivers for Switches that rely on an intermediary
+ * Ethernet device for I/O.  These switches may support cascading allowing
+ * them to be arranged as a tree.
+ * DSA is documented in detail in the Linux kernel documentation under
+ * Documentation/networking/dsa/dsa.txt
+ * The network layout of such a switch is shown below:
+ *
+ * |------|
+ * | eth0 | <--- master eth device (regular eth driver)
+ * |------|
+ *  ^  |
+ * tag added by switch -->|  |
+ *  |  |
+ *  |  |<-- tag added by DSA driver
+ *  |  v
+ * |--------------------------------------|
+ * |      | CPU port |       | <-- DSA (switch) device
+ * |      ------------             |     (DSA driver)
+ * | _________  _________       _________ |
+ * | | port0 |  | port1 |  ...  | portn | | <-- ports as eth devices
+ * |-+-------+--+-------+-------+-------+-|     ('dsa-port' eth driver)
+ *
+ * In U-Boot the intent is to allow access to front panel ports (shown at the
+ * bottom of the picture) through the master Ethernet dev (eth0 in the picture).
+ * Front panel ports are presented as regular Ethernet devices in U-Boot and
+ * they are expected to support the typical networking commands.
+ * In general DSA switches require the use of tags, extra headers added both by
+ * software on Tx and by the switch on Rx.  These tags carry at a minimum port
+ * information and switch information for cascaded set-ups.
+ * In U-Boot these tags are inserted and parsed by the DSA switch driver, the
+ * class code helps with headroom/tailroom for the extra headers.
+ *
+ * TODO:
+ * - handle switch cascading, for now U-Boot only supports stand-alone switches.
+ * - Add support to probe DSA switches connected to a MDIO bus, this is needed
+ * to convert switch drivers that are now under drivers/net/phy.
+ */
+
+#define DSA_PORT_NAME_LENGTH 16
+
+/* Maximum number of ports each DSA device can have */
+#define DSA_MAX_PORTS 12
+
+/**
+ * struct dsa_ops - DSA operations
+ *
+ * @port_enable:  Initialize a switch port for I/O.
+ * @port_disable: Disable I/O for a port.
+ * @xmit:         Insert the DSA tag for transmission.
+ *                DSA drivers receive a copy of the packet with headroom and
+ *                tailroom reserved and set to 0. 'packet' points to headroom
+ *                and 'length' is updated to include both head and tailroom.
+ * @rcv:          Process the DSA tag on reception and return the port index
+ *  from the h/w provided tag. Return the index via 'portp'.
+ *                'packet' and 'length' describe the frame as received from
+ *                master including any additional headers.
+ */
+struct dsa_ops {
+ int (*port_enable)(struct udevice *dev, int port,
+   struct phy_device *phy);
+ void (*port_disable)(struct udevice *dev, int port,
+     struct phy_device *phy);
+ int (*xmit)(struct udevice *dev, int port, void *packet, int length);
+ int (*rcv)(struct udevice *dev, int *portp, void *packet, int length);
+};
+
+#define dsa_get_ops(dev) ((struct dsa_ops *)(dev)->driver->ops)
+
+/**
+ * struct dsa_port_pdata - DSA port platform data
+ *
+ * @phy:   PHY device associated with this port.
+ *         The uclass code attempts to set this field for all ports except CPU
+ *         port, based on DT information.  It may be NULL.
+ * @index: Port index in the DSA switch, set by the uclass code.
+ * @name:  Name of the port Eth device.  If a label property is present in the
+ *         port DT node, it is used as name.
+ */
+struct dsa_port_pdata {
+ struct phy_device *phy;
+ u32 index;
+ char name[DSA_PORT_NAME_LENGTH];
+};
+
+/**
+ * struct dsa_pdata - Per-device platform data for DSA DM
+ *
+ * @num_ports:   Number of ports the device has, must be <= DSA_MAX_PORTS.
+ * This number is extracted from the DT 'ports' node of this
+ * DSA device, and it counts the CPU port and all the other
+ * port subnodes including the disabled ones.
+ * @cpu_port:    Index of the switch port linked to the master Ethernet.
+ * The uclass code sets this based on DT information.
+ * @master_node: DT node of the master Ethernet.
+ */
+struct dsa_pdata {
+ int num_ports;
+ u32 cpu_port;
+ ofnode master_node;
+};
+
+/**
+ * dsa_set_tagging() - Configure the headroom and/or tailroom sizes
+ *
+ * The DSA class code allocates headroom and tailroom on Tx before
+ * calling the DSA driver's xmit function.
+ * All drivers must call this at probe time.
+ *
+ * @dev: DSA device pointer
+ * @headroom: Size, in bytes, of headroom needed for the DSA tag.
+ * @tailroom: Size, in bytes, of tailroom needed for the DSA tag.
+ * Total headroom and tailroom size should not exceed
+ * DSA_MAX_OVR.
+ * @return 0 if OK, -ve on error
+ */
+int dsa_set_tagging(struct udevice *dev, ushort headroom, ushort tailroom);
+
+/* DSA helpers */
+
+/**
+ * dsa_get_master() - Return a reference to the master Ethernet device
+ *
+ * Can be called at driver probe time or later.
+ *
+ * @dev: DSA device pointer
+ * @return Master Eth 'udevice' pointer if OK, NULL on error
+ */
+struct udevice *dsa_get_master(struct udevice *dev);
+
+/**
+ * dsa_get_port_by_index() - Return a reference to an active DSA port device
+ *
+ * Can be called at driver probe time or later.
+ *
+ * @dev: DSA device pointer
+ * @index: Port index in the DSA device, must be an active
+ * non-CPU port
+ * @return DSA port 'udevice' pointer if OK, NULL on error
+ */
+struct udevice *dsa_get_port_by_index(struct udevice *dev, int index);
+
+/**
+ * dsa_first_port() - Return a reference to the first active DSA port device
+ *
+ * Can be called at driver probe time or later.
+ *
+ * @dev: DSA device pointer
+ * @return DSA port 'udevice' pointer if OK, NULL on error
+ */
+struct udevice *dsa_first_port(struct udevice *dev);
+
+/**
+ * dsa_next_port() - Return a reference to the next active DSA port device
+ *
+ * Can be called at driver probe time or later.
+ *
+ * @pdev: DSA port device pointer
+ * @return DSA port 'udevice' pointer if OK, NULL on error
+ */
+struct udevice *dsa_next_port(struct udevice *pdev);
+
+/**
+ * dsa_foreach_port() - iterate over all active ports of a DSA device
+ *
+ * @port_dev: DSA port device (non-CPU) reference ('udevice' pointer, lvalue)
+ * @dev: DSA device pointer ('udevice' pointer)
+ */
+#define dsa_foreach_port(port_dev, dev) \
+ for (port_dev = dsa_first_port(dev); \
+     port_dev; \
+     port_dev = dsa_next_port(port_dev))
+
+/**
+ * dsa_port_get_pdata() - Helper that returns the platdata of an active
+ * (non-CPU) DSA port device.
+ *
+ * Can be called at driver probe time or later.
+ *
+ * @pdev: DSA port device pointer
+ * @return 'dsa_port_pdata' pointer if OK, NULL on error
+ */
+static inline struct dsa_port_pdata *
+ dsa_port_get_pdata(struct udevice *pdev)
+{
+ struct eth_pdata *eth = dev_get_plat(pdev);
+
+ if (!eth)
+ return NULL;
+
+ return eth->priv_pdata;
+}
+
+#endif /* __DSA_H__ */
diff --git a/net/Makefile b/net/Makefile
index 76527f704c..fb3eba840f 100644
--- a/net/Makefile
+++ b/net/Makefile
@@ -9,6 +9,7 @@ obj-$(CONFIG_NET)      += arp.o
 obj-$(CONFIG_CMD_BOOTP) += bootp.o
 obj-$(CONFIG_CMD_CDP)  += cdp.o
 obj-$(CONFIG_CMD_DNS)  += dns.o
+obj-$(CONFIG_DM_DSA)   += dsa-uclass.o
 ifdef CONFIG_DM_ETH
 obj-$(CONFIG_NET)      += eth-uclass.o
 else
diff --git a/net/dsa-uclass.c b/net/dsa-uclass.c
new file mode 100644
index 0000000000..85b0d761e5
--- /dev/null
+++ b/net/dsa-uclass.c
@@ -0,0 +1,517 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright 2019-2021 NXP
+ */
+
+#include <net/dsa.h>
+#include <dm/lists.h>
+#include <dm/device_compat.h>
+#include <dm/device-internal.h>
+#include <dm/uclass-internal.h>
+#include <linux/bitmap.h>
+#include <miiphy.h>
+
+#define DSA_PORT_CHILD_DRV_NAME "dsa-port"
+
+/* per-device internal state structure */
+struct dsa_priv {
+ int num_ports;
+ u32 cpu_port;
+ int headroom;
+ int tailroom;
+
+ DECLARE_BITMAP(active_ports, DSA_MAX_PORTS);
+ struct udevice *master_dev;
+};
+
+/* external API */
+int dsa_set_tagging(struct udevice *dev, ushort headroom, ushort tailroom)
+{
+ struct dsa_priv *priv;
+
+ if (!dev || !dev_get_uclass_priv(dev))
+ return -ENODEV;
+
+ if (headroom + tailroom > DSA_MAX_OVR)
+ return -EINVAL;
+
+ priv = dev_get_uclass_priv(dev);
+
+ if (headroom > 0)
+ priv->headroom = headroom;
+ if (tailroom > 0)
+ priv->tailroom = tailroom;
+
+ return 0;
+}
+
+struct udevice *dsa_get_port_by_index(struct udevice *dev, int index)
+{
+ struct dsa_port_pdata *port_pdata;
+ struct udevice *pdev;
+
+ if (!dev)
+ return NULL;
+
+ for (device_find_first_child(dev, &pdev); pdev;
+     device_find_next_child(&pdev)) {
+ port_pdata = dev_get_parent_plat(pdev);
+ if (port_pdata->index == index)
+ break;
+ }
+
+ return pdev;
+}
+
+struct udevice *dsa_first_port(struct udevice *dev)
+{
+ struct dsa_priv *priv = dev_get_uclass_priv(dev);
+ int index;
+
+ if (!priv)
+ return NULL;
+
+ index = find_first_bit(priv->active_ports, DSA_MAX_PORTS);
+
+ return dsa_get_port_by_index(dev, index);
+}
+
+struct udevice *dsa_next_port(struct udevice *pdev)
+{
+ struct dsa_port_pdata *port_pdata;
+ struct dsa_priv *priv;
+ struct udevice *dev;
+ int index;
+
+ if (!pdev)
+ return NULL;
+
+ dev = dev_get_parent(pdev);
+ priv = dev_get_uclass_priv(dev);
+ if (!priv)
+ return NULL;
+
+ port_pdata = dev_get_parent_plat(pdev);
+ index = find_next_bit(priv->active_ports, DSA_MAX_PORTS,
+      port_pdata->index + 1);
+
+ return dsa_get_port_by_index(dev, index);
+}
+
+/* returns the DSA master Ethernet device */
+struct udevice *dsa_get_master(struct udevice *dev)
+{
+ struct dsa_priv *priv = dev_get_uclass_priv(dev);
+
+ if (!priv)
+ return NULL;
+
+ return priv->master_dev;
+}
+
+/*
+ * Start the desired port, the CPU port and the master Eth interface.
+ * TODO: if cascaded we may need to _start ports in other switches too
+ */
+static int dsa_port_start(struct udevice *pdev)
+{
+ struct udevice *dev = dev_get_parent(pdev);
+ struct dsa_priv *priv = dev_get_uclass_priv(dev);
+ struct udevice *master = dsa_get_master(dev);
+ struct dsa_ops *ops = dsa_get_ops(dev);
+ int err;
+
+ if (!priv)
+ return -ENODEV;
+
+ if (!master) {
+ dev_err(pdev, "DSA master Ethernet device not found!\n");
+ return -EINVAL;
+ }
+
+ if (ops->port_enable) {
+ struct dsa_port_pdata *port_pdata;
+
+ port_pdata = dev_get_parent_plat(pdev);
+ err = ops->port_enable(dev, port_pdata->index,
+       port_pdata->phy);
+ if (err)
+ return err;
+
+ err = ops->port_enable(dev, priv->cpu_port, NULL);
+ if (err)
+ return err;
+ }
+
+ return eth_get_ops(master)->start(master);
+}
+
+/* Stop the desired port, the CPU port and the master Eth interface */
+static void dsa_port_stop(struct udevice *pdev)
+{
+ struct udevice *dev = dev_get_parent(pdev);
+ struct dsa_priv *priv = dev_get_uclass_priv(dev);
+ struct udevice *master = dsa_get_master(dev);
+ struct dsa_ops *ops = dsa_get_ops(dev);
+
+ if (!priv)
+ return;
+
+ if (ops->port_disable) {
+ struct dsa_port_pdata *port_pdata;
+
+ port_pdata = dev_get_parent_plat(pdev);
+ ops->port_disable(dev, port_pdata->index, port_pdata->phy);
+ ops->port_disable(dev, priv->cpu_port, NULL);
+ }
+
+ /*
+ * stop master only if it's active, don't probe it otherwise.
+ * Under normal usage it would be active because we're using it, but
+ * during tear-down it may have been removed ahead of us.
+ */
+ if (master && device_active(master))
+ eth_get_ops(master)->stop(master);
+}
+
+/*
+ * Insert a DSA tag and call master Ethernet send on the resulting packet
+ * We copy the frame to a stack buffer where we have reserved headroom and
+ * tailroom space.  Headroom and tailroom are set to 0.
+ */
+static int dsa_port_send(struct udevice *pdev, void *packet, int length)
+{
+ struct udevice *dev = dev_get_parent(pdev);
+ struct dsa_priv *priv = dev_get_uclass_priv(dev);
+ int head = priv->headroom, tail = priv->tailroom;
+ struct udevice *master = dsa_get_master(dev);
+ struct dsa_ops *ops = dsa_get_ops(dev);
+ uchar dsa_packet_tmp[PKTSIZE_ALIGN];
+ struct dsa_port_pdata *port_pdata;
+ int err;
+
+ if (!master)
+ return -EINVAL;
+
+ if (length + head + tail > PKTSIZE_ALIGN)
+ return -EINVAL;
+
+ memset(dsa_packet_tmp, 0, head);
+ memset(dsa_packet_tmp + head + length, 0, tail);
+ memcpy(dsa_packet_tmp + head, packet, length);
+ length += head + tail;
+ /* copy back to preserve original buffer alignment */
+ memcpy(packet, dsa_packet_tmp, length);
+
+ port_pdata = dev_get_parent_plat(pdev);
+ err = ops->xmit(dev, port_pdata->index, packet, length);
+ if (err)
+ return err;
+
+ return eth_get_ops(master)->send(master, packet, length);
+}
+
+/* Receive a frame from master Ethernet, process it and pass it on */
+static int dsa_port_recv(struct udevice *pdev, int flags, uchar **packetp)
+{
+ struct udevice *dev = dev_get_parent(pdev);
+ struct dsa_priv *priv = dev_get_uclass_priv(dev);
+ int head = priv->headroom, tail = priv->tailroom;
+ struct udevice *master = dsa_get_master(dev);
+ struct dsa_ops *ops = dsa_get_ops(dev);
+ struct dsa_port_pdata *port_pdata;
+ int length, port_index, err;
+
+ if (!master)
+ return -EINVAL;
+
+ length = eth_get_ops(master)->recv(master, flags, packetp);
+ if (length <= 0)
+ return length;
+
+ /*
+ * If we receive frames from a different port or frames that DSA driver
+ * doesn't like we discard them here.
+ * In case of discard we return with no frame and expect to be called
+ * again instead of looping here, so upper layer can deal with timeouts.
+ */
+ port_pdata = dev_get_parent_plat(pdev);
+ err = ops->rcv(dev, &port_index, *packetp, length);
+ dev_dbg(pdev, "index: %d, len: %d\n", port_index, length);
+ if (err || port_index != port_pdata->index || (length <= head + tail)) {
+ if (eth_get_ops(master)->free_pkt)
+ eth_get_ops(master)->free_pkt(master, *packetp, length);
+ return -EAGAIN;
+ }
+
+ /*
+ * We move the pointer over headroom here to avoid a copy.  If free_pkt
+ * gets called we move the pointer back before calling master free_pkt.
+ */
+ *packetp += head;
+
+ return length - head - tail;
+}
+
+static int dsa_port_free_pkt(struct udevice *pdev, uchar *packet, int length)
+{
+ struct udevice *dev = dev_get_parent(pdev);
+ struct udevice *master = dsa_get_master(dev);
+ struct dsa_priv *priv;
+
+ if (!master)
+ return -EINVAL;
+
+ priv = dev_get_uclass_priv(dev);
+ if (eth_get_ops(master)->free_pkt) {
+ /* return the original pointer and length to master Eth */
+ packet -= priv->headroom;
+ length += priv->headroom - priv->tailroom;
+
+ return eth_get_ops(master)->free_pkt(master, packet, length);
+ }
+
+ return 0;
+}
+
+static int dsa_port_of_to_pdata(struct udevice *pdev)
+{
+ struct dsa_port_pdata *port_pdata;
+ struct dsa_pdata *dsa_pdata;
+ struct eth_pdata *eth_pdata;
+ struct udevice *dev;
+ const char *label;
+ u32 index;
+ int err;
+
+ if (!pdev)
+ return -ENODEV;
+
+ err = ofnode_read_u32(dev_ofnode(pdev), "reg", &index);
+ if (err)
+ return err;
+
+ dev = dev_get_parent(pdev);
+ dsa_pdata = dev_get_uclass_plat(dev);
+
+ port_pdata = dev_get_parent_plat(pdev);
+ port_pdata->index = index;
+
+ label = ofnode_read_string(dev_ofnode(pdev), "label");
+ if (label)
+ strncpy(port_pdata->name, label, DSA_PORT_NAME_LENGTH);
+
+ eth_pdata = dev_get_plat(pdev);
+ eth_pdata->priv_pdata = port_pdata;
+
+ dev_dbg(pdev, "port %d node %s\n", port_pdata->index,
+ ofnode_get_name(dev_ofnode(pdev)));
+
+ return 0;
+}
+
+static const struct eth_ops dsa_port_ops = {
+ .start = dsa_port_start,
+ .send = dsa_port_send,
+ .recv = dsa_port_recv,
+ .stop = dsa_port_stop,
+ .free_pkt = dsa_port_free_pkt,
+};
+
+static int dsa_port_probe(struct udevice *pdev)
+{
+ struct udevice *dev = dev_get_parent(pdev);
+ struct eth_pdata *eth_pdata, *master_pdata;
+ struct dsa_port_pdata *port_pdata;
+ struct dsa_priv *dsa_priv;
+
+ port_pdata = dev_get_parent_plat(pdev);
+ dsa_priv = dev_get_uclass_priv(dev);
+
+ __set_bit(port_pdata->index, dsa_priv->active_ports);
+
+ port_pdata->phy = dm_eth_phy_connect(pdev);
+
+ dev = dsa_get_master(dev);
+ if (!dev)
+ return 0;
+
+ /* inherit port's hwaddr from master */
+ master_pdata = dev_get_plat(dev);
+ eth_pdata = dev_get_plat(pdev);
+ memcpy(eth_pdata->enetaddr, master_pdata->enetaddr, ARP_HLEN);
+
+ return 0;
+}
+
+static int dsa_port_remove(struct udevice *pdev)
+{
+ struct udevice *dev = dev_get_parent(pdev);
+ struct dsa_port_pdata *port_pdata;
+ struct dsa_priv *dsa_priv;
+
+ port_pdata = dev_get_parent_plat(pdev);
+ dsa_priv = dev_get_uclass_priv(dev);
+
+ __clear_bit(port_pdata->index, dsa_priv->active_ports);
+
+ port_pdata->phy = NULL;
+
+ return 0;
+}
+
+U_BOOT_DRIVER(dsa_port) = {
+ .name = DSA_PORT_CHILD_DRV_NAME,
+ .id = UCLASS_ETH,
+ .ops = &dsa_port_ops,
+ .probe = dsa_port_probe,
+ .remove = dsa_port_remove,
+ .of_to_plat = dsa_port_of_to_pdata,
+ .plat_auto = sizeof(struct eth_pdata),
+};
+
+/*
+ * This function mostly deals with pulling information out of the device tree
+ * into the pdata structure.
+ * It goes through the list of switch ports, registers an eth device for each
+ * front panel port and identifies the cpu port connected to master eth device.
+ * TODO: support cascaded switches
+ */
+static int dsa_post_bind(struct udevice *dev)
+{
+ struct dsa_pdata *pdata = dev_get_uclass_plat(dev);
+ ofnode node = dev_ofnode(dev), pnode;
+ int i, err, first_err = 0;
+
+ if (!pdata || !ofnode_valid(node))
+ return -ENODEV;
+
+ pdata->master_node = ofnode_null();
+
+ node = ofnode_find_subnode(node, "ports");
+ if (!ofnode_valid(node)) {
+ dev_err(dev, "ports node is missing under DSA device!\n");
+ return -EINVAL;
+ }
+
+ pdata->num_ports = ofnode_get_child_count(node);
+ if (pdata->num_ports <= 0 || pdata->num_ports > DSA_MAX_PORTS) {
+ dev_err(dev, "invalid number of ports (%d)\n",
+ pdata->num_ports);
+ return -EINVAL;
+ }
+
+ /* look for the CPU port */
+ ofnode_for_each_subnode(pnode, node) {
+ u32 ethernet;
+
+ if (ofnode_read_u32(pnode, "ethernet", &ethernet))
+ continue;
+
+ pdata->master_node = ofnode_get_by_phandle(ethernet);
+ break;
+ }
+
+ if (!ofnode_valid(pdata->master_node)) {
+ dev_err(dev, "master eth node missing!\n");
+ return -EINVAL;
+ }
+
+ if (ofnode_read_u32(pnode, "reg", &pdata->cpu_port)) {
+ dev_err(dev, "CPU port node not valid!\n");
+ return -EINVAL;
+ }
+
+ dev_dbg(dev, "master node %s on port %d\n",
+ ofnode_get_name(pdata->master_node), pdata->cpu_port);
+
+ for (i = 0; i < pdata->num_ports; i++) {
+ char name[DSA_PORT_NAME_LENGTH];
+ struct udevice *pdev;
+
+ /*
+ * If this is the CPU port don't register it as an ETH device,
+ * we skip it on purpose since I/O to/from it from the CPU
+ * isn't useful
+ * TODO: cpu port may have a PHY and we don't handle that yet.
+ */
+ if (i == pdata->cpu_port)
+ continue;
+
+ /*
+ * Set up default port names.  If present, DT port labels
+ * will override the default port names.
+ */
+ snprintf(name, DSA_PORT_NAME_LENGTH, "%s@%d", dev->name, i);
+
+ ofnode_for_each_subnode(pnode, node) {
+ u32 reg;
+
+ if (ofnode_read_u32(pnode, "reg", &reg))
+ continue;
+
+ if (reg == i)
+ break;
+ }
+
+ /*
+ * skip registration if port id not found or if the port
+ * is explicitly disabled in DT
+ */
+ if (!ofnode_valid(pnode) || !ofnode_is_available(pnode))
+ continue;
+
+ err = device_bind_driver_to_node(dev, DSA_PORT_CHILD_DRV_NAME,
+ name, pnode, &pdev);
+ if (pdev) {
+ struct dsa_port_pdata *port_pdata;
+
+ port_pdata = dev_get_parent_plat(pdev);
+ strncpy(port_pdata->name, name, DSA_PORT_NAME_LENGTH);
+ pdev->name = port_pdata->name;
+ }
+
+ /* try to bind all ports but keep 1st error */
+ if (err && !first_err)
+ first_err = err;
+ }
+
+ if (first_err)
+ return first_err;
+
+ dev_dbg(dev, "DSA ports successfully bound\n");
+
+ return 0;
+}
+
+/**
+ * Initialize the uclass per device internal state structure (priv).
+ * TODO: pick up references to other switch devices here, if we're cascaded.
+ */
+static int dsa_pre_probe(struct udevice *dev)
+{
+ struct dsa_pdata *pdata = dev_get_uclass_plat(dev);
+ struct dsa_priv *priv = dev_get_uclass_priv(dev);
+
+ if (!pdata || !priv)
+ return -ENODEV;
+
+ priv->num_ports = pdata->num_ports;
+ priv->cpu_port = pdata->cpu_port;
+
+ if (ofnode_valid(pdata->master_node))
+ uclass_find_device_by_ofnode(UCLASS_ETH, pdata->master_node,
+     &priv->master_dev);
+ return 0;
+}
+
+UCLASS_DRIVER(dsa) = {
+ .id = UCLASS_DSA,
+ .name = "dsa",
+ .post_bind = dsa_post_bind,
+ .pre_probe = dsa_pre_probe,
+ .per_device_auto = sizeof(struct dsa_priv),
+ .per_device_plat_auto = sizeof(struct dsa_pdata),
+ .per_child_plat_auto = sizeof(struct dsa_port_pdata),
+ .flags = DM_UC_FLAG_SEQ_ALIAS,
+};
--
2.17.1

Reply | Threaded
Open this post in threaded view
|

[PATCH 2/5] sandbox: Add a DSA sandbox driver and unit test

Claudiu Manoil-2
In reply to this post by Claudiu Manoil-2
The DSA sandbox driver is used for unit testing the DSA class code.
It implements a simple 2 port switch plus 1 CPU port, and uses a very
simple tag to identify the ports.
The DSA sandbox device is connected via CPU port to a regular Ethernet
sandbox device, called 'dsa-test-eth, managed by the existing eth sandbox
driver.  The 'dsa-test-eth' is not intended for testing the eth class
code however, but it is used to emulate traffic through the 'lan0' and
'lan1' front pannel switch ports.  To achieve this the dsa sandbox driver
registers a tx handler for the 'dsa-test-eth' device.  The switch ports,
labeled as 'lan0' and 'lan1', are also registered as eth devices by the
dsa class code this time.  So pinging through these switch ports is as
easy as:
=> setenv ethact lan0
=> ping 1.2.3.5

Unit tests for the dsa class code were also added.  The 'dsa_probe'
test exercises most API functions from dsa.h.  The 'dsa' unit test
simply exercises ARP/ICMP traffic through the two switch ports,
including tag injection and extraction, with the help of the dsa
sandbox driver.
I took care to minimize the impact on the existing eth unit tests,
though some adjustments needed to be made with the addition of
extra eth interfaces used by the dsa unit tests. The additional eth
interfaces also require MAC addresses, these have been added to the
sandbox default environment.

Signed-off-by: Alex Marginean <[hidden email]>
Signed-off-by: Claudiu Manoil <[hidden email]>
---
 arch/Kconfig              |   1 +
 arch/sandbox/dts/test.dts |  40 +++++++++
 drivers/net/Kconfig       |   9 ++
 drivers/net/Makefile      |   1 +
 drivers/net/dsa_sandbox.c | 181 ++++++++++++++++++++++++++++++++++++++
 include/configs/sandbox.h |   2 +
 test/dm/Makefile          |   1 +
 test/dm/dsa.c             |  91 +++++++++++++++++++
 test/dm/eth.c             |  10 +--
 9 files changed, 331 insertions(+), 5 deletions(-)
 create mode 100644 drivers/net/dsa_sandbox.c
 create mode 100644 test/dm/dsa.c

diff --git a/arch/Kconfig b/arch/Kconfig
index 27843cd79c..8a8130f0c8 100644
--- a/arch/Kconfig
+++ b/arch/Kconfig
@@ -160,6 +160,7 @@ config SANDBOX
  imply CMD_CLONE
  imply SILENT_CONSOLE
  imply BOOTARGS_SUBST
+ imply DM_DSA
 
 config SH
  bool "SuperH architecture"
diff --git a/arch/sandbox/dts/test.dts b/arch/sandbox/dts/test.dts
index c9b9b7b75e..8e85b343db 100644
--- a/arch/sandbox/dts/test.dts
+++ b/arch/sandbox/dts/test.dts
@@ -14,7 +14,9 @@
  aliases {
  console = &uart0;
  eth0 = "/eth@10002000";
+ eth2 = &swp_0;
  eth3 = &eth_3;
+ eth4 = &dsa_eth0;
  eth5 = &eth_5;
  gpio1 = &gpio_a;
  gpio2 = &gpio_b;
@@ -375,6 +377,44 @@
  fake-host-hwaddr = [00 00 66 44 22 22];
  };
 
+ dsa_eth0: dsa-test-eth {
+ compatible = "sandbox,eth";
+ reg = <0x10006000 0x1000>;
+ fake-host-hwaddr = [00 00 66 44 22 66];
+ };
+
+ dsa-test {
+ compatible = "sandbox,dsa";
+
+ ports {
+ #address-cells = <1>;
+ #size-cells = <0>;
+ swp_0: port@0 {
+ reg = <0>;
+ label = "lan0";
+ };
+
+ swp_1: port@1 {
+ reg = <1>;
+ label = "lan1";
+ phy-mode = "rgmii-txid";
+ fixed-link {
+ speed = <100>;
+ full-duplex;
+ };
+ };
+
+ port@2 {
+ reg = <2>;
+ ethernet = <&dsa_eth0>;
+ fixed-link {
+ speed = <1000>;
+ full-duplex;
+ };
+ };
+ };
+ };
+
  firmware {
  sandbox_firmware: sandbox-firmware {
  compatible = "sandbox,firmware";
diff --git a/drivers/net/Kconfig b/drivers/net/Kconfig
index 382224d04c..214aebf14a 100644
--- a/drivers/net/Kconfig
+++ b/drivers/net/Kconfig
@@ -77,6 +77,15 @@ config DM_ETH_PHY
  help
   Enable driver model for Ethernet Generic PHY .
 
+config DSA_SANDBOX
+ depends on DM_DSA && SANDBOX
+ default y
+ bool "Sandbox: Mocked DSA driver"
+ help
+  This driver implements a dummy DSA switch connected to a dummy sandbox
+  Ethernet device used as DSA master, to test DSA class code, including
+  exported DSA API and datapath processing of Ethernet traffic.
+
 menuconfig NETDEVICES
  bool "Network device support"
  depends on NET
diff --git a/drivers/net/Makefile b/drivers/net/Makefile
index e3bdda359d..545fbbc0ce 100644
--- a/drivers/net/Makefile
+++ b/drivers/net/Makefile
@@ -21,6 +21,7 @@ obj-$(CONFIG_ETH_DESIGNWARE_SOCFPGA) += dwmac_socfpga.o
 obj-$(CONFIG_ETH_DESIGNWARE_S700) += dwmac_s700.o
 obj-$(CONFIG_DRIVER_DM9000) += dm9000x.o
 obj-$(CONFIG_DNET) += dnet.o
+obj-$(CONFIG_DSA_SANDBOX) += dsa_sandbox.o
 obj-$(CONFIG_DM_ETH_PHY) += eth-phy-uclass.o
 obj-$(CONFIG_E1000) += e1000.o
 obj-$(CONFIG_E1000_SPI) += e1000_spi.o
diff --git a/drivers/net/dsa_sandbox.c b/drivers/net/dsa_sandbox.c
new file mode 100644
index 0000000000..44548df86a
--- /dev/null
+++ b/drivers/net/dsa_sandbox.c
@@ -0,0 +1,181 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright 2019-2021 NXP
+ */
+
+#include <asm/eth.h>
+#include <net/dsa.h>
+#include <net.h>
+
+#define DSA_SANDBOX_MAGIC 0x00415344
+#define DSA_SANDBOX_TAG_LEN sizeof(struct dsa_sandbox_tag)
+
+struct dsa_sandbox_priv {
+ struct eth_sandbox_priv *master_priv;
+ int port_en_mask;
+};
+
+struct dsa_sandbox_tag {
+ u32 magic;
+ u32 port;
+};
+
+static bool sb_dsa_port_enabled(struct udevice *dev, int port)
+{
+ struct dsa_sandbox_priv *priv = dev_get_priv(dev);
+
+ return priv->port_en_mask & BIT(port);
+}
+
+static bool sb_dsa_master_enabled(struct udevice *dev)
+{
+ struct dsa_sandbox_priv *priv = dev_get_priv(dev);
+
+ return !priv->master_priv->disabled;
+}
+
+static int dsa_sandbox_port_enable(struct udevice *dev, int port,
+   struct phy_device *phy)
+{
+ struct dsa_sandbox_priv *priv = dev_get_priv(dev);
+
+ if (!sb_dsa_master_enabled(dev))
+ return -EFAULT;
+
+ priv->port_en_mask |= BIT(port);
+
+ return 0;
+}
+
+static void dsa_sandbox_port_disable(struct udevice *dev, int port,
+     struct phy_device *phy)
+{
+ struct dsa_sandbox_priv *priv = dev_get_priv(dev);
+
+ priv->port_en_mask &= ~BIT(port);
+}
+
+static int dsa_sandbox_xmit(struct udevice *dev, int port, void *packet,
+    int length)
+{
+ struct dsa_sandbox_tag *tag = packet;
+
+ if (!sb_dsa_master_enabled(dev))
+ return -EFAULT;
+
+ if (!sb_dsa_port_enabled(dev, port))
+ return -EFAULT;
+
+ tag->magic = DSA_SANDBOX_MAGIC;
+ tag->port = port;
+
+ return 0;
+}
+
+static int dsa_sandbox_rcv(struct udevice *dev, int *port, void *packet,
+   int length)
+{
+ struct dsa_sandbox_tag *tag = packet;
+
+ if (!sb_dsa_master_enabled(dev))
+ return -EFAULT;
+
+ if (tag->magic != DSA_SANDBOX_MAGIC)
+ return -EFAULT;
+
+ *port = tag->port;
+ if (!sb_dsa_port_enabled(dev, tag->port))
+ return -EFAULT;
+
+ return 0;
+}
+
+static const struct dsa_ops dsa_sandbox_ops = {
+ .port_enable = dsa_sandbox_port_enable,
+ .port_disable = dsa_sandbox_port_disable,
+ .xmit = dsa_sandbox_xmit,
+ .rcv = dsa_sandbox_rcv,
+};
+
+static int sb_dsa_handler(struct udevice *dev, void *packet,
+  unsigned int len)
+{
+ struct eth_sandbox_priv *master_priv;
+ struct dsa_sandbox_tag *tag = packet;
+ struct udevice *pdev, *dsa_dev;
+ u32 port_index;
+ void *rx_buf;
+ int i;
+
+ /* this emulates the switch hw and the network side */
+ if (tag->magic != DSA_SANDBOX_MAGIC)
+ return -EFAULT;
+
+ port_index = tag->port;
+ master_priv = dev_get_priv(dev);
+ dsa_dev = master_priv->priv;
+ if (!sb_dsa_port_enabled(dsa_dev, port_index))
+ return -EFAULT;
+
+ pdev = dsa_get_port_by_index(dsa_dev, port_index);
+
+ packet += DSA_SANDBOX_TAG_LEN;
+ len -= DSA_SANDBOX_TAG_LEN;
+
+ if (!sandbox_eth_arp_req_to_reply(dev, packet, len))
+ goto dsa_tagging;
+ if (!sandbox_eth_ping_req_to_reply(dev, packet, len))
+ goto dsa_tagging;
+
+ return 0;
+
+dsa_tagging:
+ master_priv->recv_packets--;
+ i = master_priv->recv_packets;
+ rx_buf = master_priv->recv_packet_buffer[i];
+ len = master_priv->recv_packet_length[i];
+ memmove(rx_buf + DSA_SANDBOX_TAG_LEN, rx_buf, len);
+
+ tag = rx_buf;
+ tag->magic = DSA_SANDBOX_MAGIC;
+ tag->port = port_index;
+ len += DSA_SANDBOX_TAG_LEN;
+ master_priv->recv_packet_length[i] = len;
+ master_priv->recv_packets++;
+
+ return 0;
+}
+
+static int dsa_sandbox_probe(struct udevice *dev)
+{
+ struct dsa_sandbox_priv *priv = dev_get_priv(dev);
+ struct udevice *master = dsa_get_master(dev);
+ struct eth_sandbox_priv *master_priv;
+
+ if (!master)
+ return -ENODEV;
+
+ dsa_set_tagging(dev, DSA_SANDBOX_TAG_LEN, 0);
+
+ master_priv = dev_get_priv(master);
+ master_priv->priv = dev;
+ master_priv->tx_handler = sb_dsa_handler;
+
+ priv->master_priv = master_priv;
+
+ return 0;
+}
+
+static const struct udevice_id dsa_sandbox_ids[] = {
+ { .compatible = "sandbox,dsa" },
+ { }
+};
+
+U_BOOT_DRIVER(dsa_sandbox) = {
+ .name = "dsa_sandbox",
+ .id = UCLASS_DSA,
+ .of_match = dsa_sandbox_ids,
+ .probe = dsa_sandbox_probe,
+ .ops = &dsa_sandbox_ops,
+ .priv_auto = sizeof(struct dsa_sandbox_priv),
+};
diff --git a/include/configs/sandbox.h b/include/configs/sandbox.h
index e0708fe573..91f636b2c1 100644
--- a/include/configs/sandbox.h
+++ b/include/configs/sandbox.h
@@ -93,7 +93,9 @@
 #endif
 
 #define SANDBOX_ETH_SETTINGS "ethaddr=00:00:11:22:33:44\0" \
+ "eth2addr=00:00:11:22:33:48\0" \
  "eth3addr=00:00:11:22:33:45\0" \
+ "eth4addr=00:00:11:22:33:48\0" \
  "eth5addr=00:00:11:22:33:46\0" \
  "eth6addr=00:00:11:22:33:47\0" \
  "ipaddr=1.2.3.4\0"
diff --git a/test/dm/Makefile b/test/dm/Makefile
index 46e076ed09..c14363cf9d 100644
--- a/test/dm/Makefile
+++ b/test/dm/Makefile
@@ -26,6 +26,7 @@ obj-$(CONFIG_DM_BOOTCOUNT) += bootcount.o
 obj-$(CONFIG_CLK) += clk.o clk_ccf.o
 obj-$(CONFIG_DEVRES) += devres.o
 obj-$(CONFIG_VIDEO_MIPI_DSI) += dsi_host.o
+obj-$(CONFIG_DM_DSA) += dsa.o
 obj-$(CONFIG_DM_ETH) += eth.o
 obj-$(CONFIG_FIRMWARE) += firmware.o
 obj-$(CONFIG_DM_GPIO) += gpio.o
diff --git a/test/dm/dsa.c b/test/dm/dsa.c
new file mode 100644
index 0000000000..e2cd4b7d3b
--- /dev/null
+++ b/test/dm/dsa.c
@@ -0,0 +1,91 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright 2020-2021 NXP
+ */
+
+#include <net/dsa.h>
+#include <dm/test.h>
+#include <test/ut.h>
+#include <net.h>
+#include <dm/uclass-internal.h>
+#include <dm/device-internal.h>
+
+/* This test exercises the major dsa.h API functions, after making sure
+ * that the DSA ports and the master Eth are correctly probed.
+ */
+static int dm_test_dsa_probe(struct unit_test_state *uts)
+{
+ struct udevice *dev_dsa, *dev_port, *dev_master;
+ struct dsa_port_pdata *port_pdata;
+ struct dsa_pdata *dsa_pdata;
+ enum uclass_id id;
+ int n;
+
+ id = uclass_get_by_name("dsa");
+ ut_assert(id == UCLASS_DSA);
+
+ ut_assertok(uclass_find_device_by_name(UCLASS_DSA, "dsa-test",
+       &dev_dsa));
+ ut_assertok(uclass_find_device_by_name(UCLASS_ETH, "dsa-test-eth",
+       &dev_master));
+ ut_assertok(device_probe(dev_master));
+
+ ut_assertok(uclass_find_device_by_name(UCLASS_ETH, "dsa-test@0",
+       &dev_port));
+ ut_assertok(device_probe(dev_port));
+
+ ut_assertok(uclass_find_device_by_name(UCLASS_ETH, "dsa-test@1",
+       &dev_port));
+ ut_assertok(device_probe(dev_port));
+
+ /* exercise DSA API */
+ dsa_pdata = dev_get_uclass_plat(dev_dsa);
+ ut_assertnonnull(dsa_pdata);
+ /* includes CPU port */
+ ut_assert(dsa_pdata->num_ports == 3);
+
+ n = 0;
+ dsa_foreach_port(dev_port, dev_dsa) {
+ char name[DSA_PORT_NAME_LENGTH];
+
+ port_pdata = dsa_port_get_pdata(dev_port);
+ ut_assertnonnull(port_pdata);
+
+ snprintf(name, DSA_PORT_NAME_LENGTH, "lan%d", n);
+ ut_asserteq_str(name, port_pdata->name);
+ n++;
+ }
+
+ debug("active non-CPU ports: %d\n", n);
+ ut_assert(n == 2);
+
+ dev_master = dsa_get_master(dev_dsa);
+ ut_assertnonnull(dev_master);
+ ut_asserteq_str("dsa-test-eth", dev_master->name);
+
+ return 0;
+}
+
+DM_TEST(dm_test_dsa_probe, UT_TESTF_SCAN_FDT);
+
+/* This test sends ping requests with the local address through each DSA port
+ * via the sandbox DSA master Eth.
+ */
+static int dm_test_dsa(struct unit_test_state *uts)
+{
+ net_ping_ip = string_to_ip("1.2.3.5");
+
+ env_set("ethact", "eth2");
+ ut_assertok(net_loop(PING));
+
+ env_set("ethact", "lan0");
+ ut_assertok(net_loop(PING));
+ env_set("ethact", "lan1");
+ ut_assertok(net_loop(PING));
+
+ env_set("ethact", "");
+
+ return 0;
+}
+
+DM_TEST(dm_test_dsa, UT_TESTF_SCAN_FDT);
diff --git a/test/dm/eth.c b/test/dm/eth.c
index fa8a69da70..e4ee695610 100644
--- a/test/dm/eth.c
+++ b/test/dm/eth.c
@@ -53,8 +53,8 @@ static int dm_test_eth_alias(struct unit_test_state *uts)
  ut_assertok(net_loop(PING));
  ut_asserteq_str("eth@10004000", env_get("ethact"));
 
- /* Expected to fail since eth2 is not defined in the device tree */
- env_set("ethact", "eth2");
+ /* Expected to fail since eth1 is not defined in the device tree */
+ env_set("ethact", "eth1");
  ut_assertok(net_loop(PING));
  ut_asserteq_str("eth@10002000", env_get("ethact"));
 
@@ -227,7 +227,7 @@ static int _dm_test_net_retry(struct unit_test_state *uts)
  * the active device should be eth0
  */
  sandbox_eth_disable_response(1, true);
- env_set("ethact", "eth@10004000");
+ env_set("ethact", "lan1");
  env_set("netretry", "yes");
  sandbox_eth_skip_timeout();
  ut_assertok(net_loop(PING));
@@ -237,11 +237,11 @@ static int _dm_test_net_retry(struct unit_test_state *uts)
  * eth1 is disabled and netretry is no, so the ping should fail and the
  * active device should be eth1
  */
- env_set("ethact", "eth@10004000");
+ env_set("ethact", "lan1");
  env_set("netretry", "no");
  sandbox_eth_skip_timeout();
  ut_asserteq(-ENONET, net_loop(PING));
- ut_asserteq_str("eth@10004000", env_get("ethact"));
+ ut_asserteq_str("lan1", env_get("ethact"));
 
  return 0;
 }
--
2.17.1

Reply | Threaded
Open this post in threaded view
|

[PATCH 3/5] drivers: net: Add Felix DSA switch driver

Claudiu Manoil-2
In reply to this post by Claudiu Manoil-2
From: Alex Marginean <[hidden email]>

This driver is used for the Ethernet switch integrated into LS1028A NXP.
Felix on LS1028A has 4 front panel ports and two internal ports, I/O
to/from the switch is done through an ENETC Ethernet interface.
The 4 front panel ports are available as Ethernet interfaces and can be
used with the typical network commands like tftp.

Signed-off-by: Alex Marginean <[hidden email]>
Signed-off-by: Vladimir Oltean <[hidden email]>
Signed-off-by: Claudiu Manoil <[hidden email]>
---
 drivers/net/fsl_enetc.h                 |   5 +
 drivers/net/mscc_eswitch/Kconfig        |   8 +
 drivers/net/mscc_eswitch/Makefile       |   1 +
 drivers/net/mscc_eswitch/felix_switch.c | 432 ++++++++++++++++++++++++
 4 files changed, 446 insertions(+)
 create mode 100644 drivers/net/mscc_eswitch/felix_switch.c

diff --git a/drivers/net/fsl_enetc.h b/drivers/net/fsl_enetc.h
index 37e7e85843..110c1d78fb 100644
--- a/drivers/net/fsl_enetc.h
+++ b/drivers/net/fsl_enetc.h
@@ -201,6 +201,11 @@ struct enetc_priv {
 /* PCS replicator block for USXGMII */
 #define ENETC_PCS_DEVAD_REPL 0x1f
 
+#define ENETC_PCS_REPL_LINK_TIMER_1 0x12
+#define  ENETC_PCS_REPL_LINK_TIMER_1_DEF 0x0003
+#define ENETC_PCS_REPL_LINK_TIMER_2 0x13
+#define  ENETC_PCS_REPL_LINK_TIMER_2_DEF 0x06a0
+
 /* ENETC external MDIO registers */
 #define ENETC_MDIO_BASE 0x1c00
 #define ENETC_MDIO_CFG 0x00
diff --git a/drivers/net/mscc_eswitch/Kconfig b/drivers/net/mscc_eswitch/Kconfig
index 80dd22f98b..11fb08edaa 100644
--- a/drivers/net/mscc_eswitch/Kconfig
+++ b/drivers/net/mscc_eswitch/Kconfig
@@ -36,3 +36,11 @@ config MSCC_SERVAL_SWITCH
  select PHYLIB
  help
   This driver supports the Serval network switch device.
+
+config MSCC_FELIX_SWITCH
+ bool "Felix switch driver"
+ depends on DM_DSA && DM_PCI
+ select FSL_ENETC
+ help
+  This driver supports the Ethernet switch integrated in LS1028A NXP
+  SoC.
diff --git a/drivers/net/mscc_eswitch/Makefile b/drivers/net/mscc_eswitch/Makefile
index d583fe9fc4..22342ed114 100644
--- a/drivers/net/mscc_eswitch/Makefile
+++ b/drivers/net/mscc_eswitch/Makefile
@@ -4,3 +4,4 @@ obj-$(CONFIG_MSCC_LUTON_SWITCH) += luton_switch.o mscc_xfer.o mscc_mac_table.o m
 obj-$(CONFIG_MSCC_JR2_SWITCH) += jr2_switch.o mscc_xfer.o mscc_miim.o
 obj-$(CONFIG_MSCC_SERVALT_SWITCH) += servalt_switch.o mscc_xfer.o mscc_miim.o
 obj-$(CONFIG_MSCC_SERVAL_SWITCH) += serval_switch.o mscc_xfer.o mscc_mac_table.o mscc_miim.o
+obj-$(CONFIG_MSCC_FELIX_SWITCH) += felix_switch.o
diff --git a/drivers/net/mscc_eswitch/felix_switch.c b/drivers/net/mscc_eswitch/felix_switch.c
new file mode 100644
index 0000000000..ae8b690e76
--- /dev/null
+++ b/drivers/net/mscc_eswitch/felix_switch.c
@@ -0,0 +1,432 @@
+// SPDX-License-Identifier: GPL-2.0+ OR BSD-3-Clause
+/*
+ * Felix Ethernet switch driver
+ * Copyright 2018-2021 NXP
+ */
+
+/*
+ * This driver is used for the Ethernet switch integrated into LS1028A NXP.
+ * Felix switch is derived from Microsemi Ocelot but there are several NXP
+ * adaptations that makes the two U-Boot drivers largely incompatible.
+ *
+ * Felix on LS1028A has 4 front panel ports and two internal ports, connected
+ * to ENETC interfaces.  We're using one of the ENETC interfaces to push traffic
+ * into the switch.  Injection/extraction headers are used to identify
+ * egress/ingress ports in the switch for Tx/Rx.
+ */
+
+#include <dm/device_compat.h>
+#include <linux/delay.h>
+#include <net/dsa.h>
+#include <asm/io.h>
+#include <miiphy.h>
+#include <pci.h>
+
+/* defines especially around PCS are reused from enetc */
+#include "../fsl_enetc.h"
+
+#define PCI_DEVICE_ID_FELIX_ETHSW 0xEEF0
+
+/* Felix has in fact 6 ports, but we don't use the last internal one */
+#define FELIX_PORT_COUNT 5
+/* Front panel port mask */
+#define FELIX_FP_PORT_MASK 0xf
+
+/* Register map for BAR4 */
+#define FELIX_SYS 0x010000
+#define FELIX_ES0 0x040000
+#define FELIX_IS1 0x050000
+#define FELIX_IS2 0x060000
+#define FELIX_GMII(port) (0x100000 + (port) * 0x10000)
+#define FELIX_QSYS 0x200000
+
+#define FELIX_SYS_SYSTEM (FELIX_SYS + 0x00000E00)
+#define  FELIX_SYS_SYSTEM_EN BIT(0)
+#define FELIX_SYS_RAM_CTRL (FELIX_SYS + 0x00000F24)
+#define  FELIX_SYS_RAM_CTRL_INIT BIT(1)
+#define FELIX_SYS_SYSTEM_PORT_MODE(a) (FELIX_SYS_SYSTEM + 0xC + (a) * 4)
+#define  FELIX_SYS_SYSTEM_PORT_MODE_CPU 0x0000001e
+
+#define FELIX_ES0_TCAM_CTRL (FELIX_ES0 + 0x000003C0)
+#define  FELIX_ES0_TCAM_CTRL_EN BIT(0)
+#define FELIX_IS1_TCAM_CTRL (FELIX_IS1 + 0x000003C0)
+#define  FELIX_IS1_TCAM_CTRL_EN BIT(0)
+#define FELIX_IS2_TCAM_CTRL (FELIX_IS2 + 0x000003C0)
+#define  FELIX_IS2_TCAM_CTRL_EN BIT(0)
+
+#define FELIX_GMII_CLOCK_CFG(port) (FELIX_GMII(port) + 0x00000000)
+#define  FELIX_GMII_CLOCK_CFG_LINK_1G 1
+#define  FELIX_GMII_CLOCK_CFG_LINK_100M 2
+#define  FELIX_GMII_CLOCK_CFG_LINK_10M 3
+#define FELIX_GMII_MAC_ENA_CFG(port) (FELIX_GMII(port) + 0x0000001C)
+#define  FELIX_GMII_MAX_ENA_CFG_TX BIT(0)
+#define  FELIX_GMII_MAX_ENA_CFG_RX BIT(4)
+#define FELIX_GMII_MAC_IFG_CFG(port) (FELIX_GMII(port) + 0x0000001C + 0x14)
+#define  FELIX_GMII_MAC_IFG_CFG_DEF 0x515
+
+#define FELIX_QSYS_SYSTEM (FELIX_QSYS + 0x0000F460)
+#define FELIX_QSYS_SYSTEM_SW_PORT_MODE(a) \
+ (FELIX_QSYS_SYSTEM + 0x20 + (a) * 4)
+#define  FELIX_QSYS_SYSTEM_SW_PORT_ENA BIT(14)
+#define  FELIX_QSYS_SYSTEM_SW_PORT_LOSSY BIT(9)
+#define  FELIX_QSYS_SYSTEM_SW_PORT_SCH(a) (((a) & 0x3800) << 11)
+#define FELIX_QSYS_SYSTEM_EXT_CPU_CFG (FELIX_QSYS_SYSTEM + 0x80)
+#define  FELIX_QSYS_SYSTEM_EXT_CPU_PORT(a) (((a) & 0xf) << 8 | 0xff)
+
+/* internal MDIO in BAR0 */
+#define FELIX_PM_IMDIO_BASE 0x8030
+
+/* Serdes block on LS1028A */
+#define FELIX_SERDES_BASE 0x1ea0000L
+#define FELIX_SERDES_LNATECR0(lane) (FELIX_SERDES_BASE + 0x818 + \
+ (lane) * 0x40)
+#define  FELIX_SERDES_LNATECR0_ADPT_EQ 0x00003000
+#define FELIX_SERDES_SGMIICR1(lane) (FELIX_SERDES_BASE + 0x1804 + \
+ (lane) * 0x10)
+#define  FELIX_SERDES_SGMIICR1_SGPCS BIT(11)
+#define  FELIX_SERDES_SGMIICR1_MDEV(a) (((a) & 0x1f) << 27)
+
+#define FELIX_PCS_CTRL 0
+#define  FELIX_PCS_CTRL_RST BIT(15)
+
+/*
+ * The long prefix format used here contains two dummy MAC addresses, a magic
+ * value in place of a VLAN tag followed by the extraction/injection header and
+ * the original L2 frame.  Out of all this we only use the port ID.
+ */
+#define FELIX_DSA_TAG_LEN sizeof(struct felix_dsa_tag)
+#define FELIX_DSA_TAG_MAGIC 0x0a008088
+#define FELIX_DSA_TAG_INJ_PORT 7
+#define  FELIX_DSA_TAG_INJ_PORT_SET(a) (0x1 << ((a) & FELIX_FP_PORT_MASK))
+#define FELIX_DSA_TAG_EXT_PORT 10
+#define  FELIX_DSA_TAG_EXT_PORT_GET(a) ((a) >> 3)
+
+struct felix_dsa_tag {
+ uchar d_mac[6];
+ uchar s_mac[6];
+ u32   magic;
+ uchar meta[16];
+};
+
+struct felix_priv {
+ void *regs_base;
+ void *imdio_base;
+ struct mii_dev imdio;
+};
+
+/* MDIO wrappers, we're using these to drive internal MDIO to get to serdes */
+static int felix_mdio_read(struct mii_dev *bus, int addr, int devad, int reg)
+{
+ struct enetc_mdio_priv priv;
+
+ priv.regs_base = bus->priv;
+ return enetc_mdio_read_priv(&priv, addr, devad, reg);
+}
+
+static int felix_mdio_write(struct mii_dev *bus, int addr, int devad, int reg,
+    u16 val)
+{
+ struct enetc_mdio_priv priv;
+
+ priv.regs_base = bus->priv;
+ return enetc_mdio_write_priv(&priv, addr, devad, reg, val);
+}
+
+/* set up serdes for SGMII */
+static void felix_init_sgmii(struct mii_dev *imdio, int pidx, bool an)
+{
+ u16 reg;
+
+ /* set up PCS lane address */
+ out_le32(FELIX_SERDES_SGMIICR1(pidx), FELIX_SERDES_SGMIICR1_SGPCS |
+ FELIX_SERDES_SGMIICR1_MDEV(pidx));
+
+ /*
+ * Set to SGMII mode, for 1Gbps enable AN, for 2.5Gbps set fixed speed.
+ * Although fixed speed is 1Gbps, we could be running at 2.5Gbps based
+ * on PLL configuration.  Setting 1G for 2.5G here is counter intuitive
+ * but intentional.
+ */
+ reg = ENETC_PCS_IF_MODE_SGMII;
+ reg |= an ? ENETC_PCS_IF_MODE_SGMII_AN : ENETC_PCS_IF_MODE_SPEED_1G;
+ felix_mdio_write(imdio, pidx, MDIO_DEVAD_NONE,
+ ENETC_PCS_IF_MODE, reg);
+
+ /* Dev ability - SGMII */
+ felix_mdio_write(imdio, pidx, MDIO_DEVAD_NONE,
+ ENETC_PCS_DEV_ABILITY, ENETC_PCS_DEV_ABILITY_SGMII);
+
+ /* Adjust link timer for SGMII */
+ felix_mdio_write(imdio, pidx, MDIO_DEVAD_NONE,
+ ENETC_PCS_LINK_TIMER1, ENETC_PCS_LINK_TIMER1_VAL);
+ felix_mdio_write(imdio, pidx, MDIO_DEVAD_NONE,
+ ENETC_PCS_LINK_TIMER2, ENETC_PCS_LINK_TIMER2_VAL);
+
+ reg = ENETC_PCS_CR_DEF_VAL;
+ reg |= an ? ENETC_PCS_CR_RESET_AN : ENETC_PCS_CR_RST;
+ /* restart PCS AN */
+ felix_mdio_write(imdio, pidx, MDIO_DEVAD_NONE,
+ ENETC_PCS_CR, reg);
+}
+
+/* set up MAC and serdes for (Q)SXGMII */
+static int felix_init_sxgmii(struct mii_dev *imdio, int pidx)
+{
+ int to = 1000;
+
+ /* set up transit equalization control on serdes lane */
+ out_le32(FELIX_SERDES_LNATECR0(1), FELIX_SERDES_LNATECR0_ADPT_EQ);
+
+ /*reset lane */
+ felix_mdio_write(imdio, pidx, MDIO_MMD_PCS, FELIX_PCS_CTRL,
+ FELIX_PCS_CTRL_RST);
+ while (felix_mdio_read(imdio, pidx, MDIO_MMD_PCS,
+       FELIX_PCS_CTRL) & FELIX_PCS_CTRL_RST &&
+ --to) {
+ mdelay(10);
+ }
+ if (felix_mdio_read(imdio, pidx, MDIO_MMD_PCS,
+    FELIX_PCS_CTRL) & FELIX_PCS_CTRL_RST)
+ return -ETIME;
+
+ /* Dev ability - SXGMII */
+ felix_mdio_write(imdio, pidx, ENETC_PCS_DEVAD_REPL,
+ ENETC_PCS_DEV_ABILITY, ENETC_PCS_DEV_ABILITY_SXGMII);
+
+ /* Restart PCS AN */
+ felix_mdio_write(imdio, pidx, ENETC_PCS_DEVAD_REPL, ENETC_PCS_CR,
+ ENETC_PCS_CR_RST | ENETC_PCS_CR_RESET_AN);
+ felix_mdio_write(imdio, pidx, ENETC_PCS_DEVAD_REPL,
+ ENETC_PCS_REPL_LINK_TIMER_1,
+ ENETC_PCS_REPL_LINK_TIMER_1_DEF);
+ felix_mdio_write(imdio, pidx, ENETC_PCS_DEVAD_REPL,
+ ENETC_PCS_REPL_LINK_TIMER_2,
+ ENETC_PCS_REPL_LINK_TIMER_2_DEF);
+
+ return 0;
+}
+
+/* Apply protocol specific configuration to MAC, serdes as needed */
+static void felix_start_pcs(struct udevice *pdev, struct mii_dev *imdio)
+{
+ struct dsa_port_pdata *port_pdata = dsa_port_get_pdata(pdev);
+ int if_type = PHY_INTERFACE_MODE_NONE;
+ bool autoneg = true;
+ const char *if_str;
+
+ if_str = ofnode_read_string(dev_ofnode(pdev), "phy-mode");
+ if (if_str)
+ if_type = phy_get_interface_by_name(if_str);
+ else
+ dev_dbg(pdev,
+ "phy-mode property not found, defaulting to NONE\n");
+ if (if_type < 0)
+ if_type = PHY_INTERFACE_MODE_NONE;
+
+ if (!port_pdata->phy || port_pdata->phy->phy_id == PHY_FIXED_ID ||
+    if_type == PHY_INTERFACE_MODE_SGMII_2500)
+ autoneg = false;
+
+ switch (if_type) {
+ case PHY_INTERFACE_MODE_SGMII:
+ case PHY_INTERFACE_MODE_SGMII_2500:
+ case PHY_INTERFACE_MODE_QSGMII:
+ felix_init_sgmii(imdio, port_pdata->index, autoneg);
+ break;
+ case PHY_INTERFACE_MODE_XGMII:
+ case PHY_INTERFACE_MODE_XFI:
+ case PHY_INTERFACE_MODE_USXGMII:
+ if (felix_init_sxgmii(imdio, port_pdata->index))
+ dev_err(pdev, "PCS reset time-out\n");
+ break;
+ }
+}
+
+void felix_init(struct udevice *dev)
+{
+ struct dsa_pdata *pdata = dev_get_uclass_plat(dev);
+ struct felix_priv *priv = dev_get_priv(dev);
+ void *base = priv->regs_base;
+ int supported, to = 100;
+ struct udevice *pdev;
+
+ /* Init core memories */
+ out_le32(base + FELIX_SYS_RAM_CTRL, FELIX_SYS_RAM_CTRL_INIT);
+ while (in_le32(base + FELIX_SYS_RAM_CTRL) & FELIX_SYS_RAM_CTRL_INIT &&
+       --to)
+ udelay(10);
+ if (in_le32(base + FELIX_SYS_RAM_CTRL) & FELIX_SYS_RAM_CTRL_INIT)
+ dev_dbg(dev, "Time-out waiting for switch memories\n");
+
+ /* Start switch core, set up ES0, IS1, IS2 */
+ out_le32(base + FELIX_SYS_SYSTEM, FELIX_SYS_SYSTEM_EN);
+ out_le32(base + FELIX_ES0_TCAM_CTRL, FELIX_ES0_TCAM_CTRL_EN);
+ out_le32(base + FELIX_IS1_TCAM_CTRL, FELIX_IS1_TCAM_CTRL_EN);
+ out_le32(base + FELIX_IS2_TCAM_CTRL, FELIX_IS2_TCAM_CTRL_EN);
+ udelay(20);
+
+ supported = PHY_GBIT_FEATURES | SUPPORTED_2500baseX_Full;
+ priv->imdio.read = felix_mdio_read;
+ priv->imdio.write = felix_mdio_write;
+ priv->imdio.priv = priv->imdio_base + FELIX_PM_IMDIO_BASE;
+ strncpy(priv->imdio.name, dev->name, MDIO_NAME_LEN);
+
+ dsa_foreach_port(pdev, dev) {
+ struct dsa_port_pdata *port_pdata = dsa_port_get_pdata(pdev);
+ struct phy_device *phy = port_pdata->phy;
+
+ /* Set up MAC registers */
+ out_le32(base + FELIX_GMII_CLOCK_CFG(port_pdata->index),
+ FELIX_GMII_CLOCK_CFG_LINK_1G);
+
+ out_le32(base + FELIX_GMII_MAC_IFG_CFG(port_pdata->index),
+ FELIX_GMII_MAC_IFG_CFG_DEF);
+
+ felix_start_pcs(pdev, &priv->imdio);
+
+ if (phy) {
+ phy->supported &= supported;
+ phy->advertising &= supported;
+ phy_config(phy);
+ }
+ }
+
+ /* set up CPU port */
+ out_le32(base + FELIX_QSYS_SYSTEM_EXT_CPU_CFG,
+ FELIX_QSYS_SYSTEM_EXT_CPU_PORT(pdata->cpu_port));
+ out_le32(base + FELIX_SYS_SYSTEM_PORT_MODE(pdata->cpu_port),
+ FELIX_SYS_SYSTEM_PORT_MODE_CPU);
+}
+
+/*
+ * Probe Felix:
+ * - enable the PCI function
+ * - map BAR 4
+ * - init switch core and port registers
+ */
+static int felix_probe(struct udevice *dev)
+{
+ struct felix_priv *priv = dev_get_priv(dev);
+
+ if (ofnode_valid(dev_ofnode(dev)) &&
+    !ofnode_is_available(dev_ofnode(dev))) {
+ dev_dbg(dev, "switch disabled\n");
+ return -ENODEV;
+ }
+
+ priv->imdio_base = dm_pci_map_bar(dev, PCI_BASE_ADDRESS_0, 0);
+ if (!priv->imdio_base) {
+ dev_dbg(dev, "failed to map BAR0\n");
+ return -EINVAL;
+ }
+
+ priv->regs_base = dm_pci_map_bar(dev, PCI_BASE_ADDRESS_4, 0);
+ if (!priv->regs_base) {
+ dev_dbg(dev, "failed to map BAR4\n");
+ return -EINVAL;
+ }
+
+ /* register internal MDIO for debug */
+ if (!miiphy_get_dev_by_name(dev->name)) {
+ struct mii_dev *mii_bus;
+
+ mii_bus = mdio_alloc();
+ mii_bus->read = felix_mdio_read;
+ mii_bus->write = felix_mdio_write;
+ mii_bus->priv = priv->imdio_base + FELIX_PM_IMDIO_BASE;
+ strncpy(mii_bus->name, dev->name, MDIO_NAME_LEN);
+ mdio_register(mii_bus);
+ }
+
+ dm_pci_clrset_config16(dev, PCI_COMMAND, 0, PCI_COMMAND_MEMORY);
+
+ dsa_set_tagging(dev, FELIX_DSA_TAG_LEN, 0);
+
+ /* set up registers */
+ felix_init(dev);
+
+ return 0;
+}
+
+static int felix_port_enable(struct udevice *dev, int pidx,
+     struct phy_device *phy)
+{
+ struct felix_priv *priv = dev_get_priv(dev);
+ void *base = priv->regs_base;
+
+ out_le32(base + FELIX_GMII_MAC_ENA_CFG(pidx),
+ FELIX_GMII_MAX_ENA_CFG_TX | FELIX_GMII_MAX_ENA_CFG_RX);
+
+ out_le32(base + FELIX_QSYS_SYSTEM_SW_PORT_MODE(pidx),
+ FELIX_QSYS_SYSTEM_SW_PORT_ENA |
+ FELIX_QSYS_SYSTEM_SW_PORT_LOSSY |
+ FELIX_QSYS_SYSTEM_SW_PORT_SCH(1));
+
+ if (phy)
+ phy_startup(phy);
+
+ return 0;
+}
+
+static void felix_port_disable(struct udevice *dev, int pidx,
+       struct phy_device *phy)
+{
+ struct felix_priv *priv = dev_get_priv(dev);
+ void *base = priv->regs_base;
+
+ out_le32(base + FELIX_GMII_MAC_ENA_CFG(pidx), 0);
+
+ out_le32(base + FELIX_QSYS_SYSTEM_SW_PORT_MODE(pidx),
+ FELIX_QSYS_SYSTEM_SW_PORT_LOSSY |
+ FELIX_QSYS_SYSTEM_SW_PORT_SCH(1));
+
+ /*
+ * we don't call phy_shutdown here to avoind waiting next time we use
+ * the port, but the downside is that remote side will think we're
+ * actively processing traffic although we are not.
+ */
+}
+
+static int felix_xmit(struct udevice *dev, int pidx, void *packet, int length)
+{
+ struct felix_dsa_tag *tag = packet;
+
+ tag->magic = FELIX_DSA_TAG_MAGIC;
+ tag->meta[FELIX_DSA_TAG_INJ_PORT] = FELIX_DSA_TAG_INJ_PORT_SET(pidx);
+
+ return 0;
+}
+
+static int felix_rcv(struct udevice *dev, int *pidx, void *packet, int length)
+{
+ struct felix_dsa_tag *tag = packet;
+
+ if (tag->magic != FELIX_DSA_TAG_MAGIC)
+ return -EINVAL;
+
+ *pidx = FELIX_DSA_TAG_EXT_PORT_GET(tag->meta[FELIX_DSA_TAG_EXT_PORT]);
+
+ return 0;
+}
+
+static const struct dsa_ops felix_dsa_ops = {
+ .port_enable  = felix_port_enable,
+ .port_disable = felix_port_disable,
+ .xmit         = felix_xmit,
+ .rcv          = felix_rcv,
+};
+
+U_BOOT_DRIVER(felix_ethsw) = {
+ .name = "felix-switch",
+ .id = UCLASS_DSA,
+ .probe = felix_probe,
+ .ops    = &felix_dsa_ops,
+ .priv_auto = sizeof(struct felix_priv),
+};
+
+static struct pci_device_id felix_ethsw_ids[] = {
+ { PCI_DEVICE(PCI_VENDOR_ID_FREESCALE, PCI_DEVICE_ID_FELIX_ETHSW) },
+ {}
+};
+
+U_BOOT_PCI_DEVICE(felix_ethsw, felix_ethsw_ids);
--
2.17.1

Reply | Threaded
Open this post in threaded view
|

[PATCH 4/5] arm: dts: ls1028a: Add Ethernet switch node and dependencies

Claudiu Manoil-2
In reply to this post by Claudiu Manoil-2
From: Alex Marginean <[hidden email]>

The definition follows the DSA binding in kernel and describes the switch,
its ports and PHYs.
ENETC PF6 is the 2nd Eth controller linked to the switch on LS1028A, it is
not used in U-Boot and was disabled.  Ethernet port aliases were also
added to better manage the multitude of ports available now, and to
enforce the order in which master and slave ports are probed.

Signed-off-by: Alex Marginean <[hidden email]>
Signed-off-by: Claudiu Manoil <[hidden email]>
---
 arch/arm/dts/fsl-ls1028a-rdb.dts | 36 +++++++++++++++++++++
 arch/arm/dts/fsl-ls1028a.dtsi    | 55 +++++++++++++++++++++++++++++++-
 2 files changed, 90 insertions(+), 1 deletion(-)

diff --git a/arch/arm/dts/fsl-ls1028a-rdb.dts b/arch/arm/dts/fsl-ls1028a-rdb.dts
index 85b4815b2e..92d83a5c0c 100644
--- a/arch/arm/dts/fsl-ls1028a-rdb.dts
+++ b/arch/arm/dts/fsl-ls1028a-rdb.dts
@@ -131,9 +131,45 @@
  phy-handle = <&rdb_phy0>;
 };
 
+&ethsw_ports {
+ port@0 {
+ status = "okay";
+ phy-mode = "qsgmii";
+ phy-handle = <&sw_phy0>;
+ };
+ port@1 {
+ status = "okay";
+ phy-mode = "qsgmii";
+ phy-handle = <&sw_phy1>;
+ };
+ port@2 {
+ status = "okay";
+ phy-mode = "qsgmii";
+ phy-handle = <&sw_phy2>;
+ };
+ port@3 {
+ status = "okay";
+ phy-mode = "qsgmii";
+ phy-handle = <&sw_phy3>;
+ };
+};
+
 &mdio0 {
  status = "okay";
  rdb_phy0: phy@2 {
  reg = <2>;
  };
+
+ sw_phy0: phy@10 {
+ reg = <0x10>;
+ };
+ sw_phy1: phy@11 {
+ reg = <0x11>;
+ };
+ sw_phy2: phy@12 {
+ reg = <0x12>;
+ };
+ sw_phy3: phy@13 {
+ reg = <0x13>;
+ };
 };
diff --git a/arch/arm/dts/fsl-ls1028a.dtsi b/arch/arm/dts/fsl-ls1028a.dtsi
index d0850237c7..e73769392f 100644
--- a/arch/arm/dts/fsl-ls1028a.dtsi
+++ b/arch/arm/dts/fsl-ls1028a.dtsi
@@ -14,6 +14,17 @@
  #address-cells = <2>;
  #size-cells = <2>;
 
+ aliases {
+ eth0 = &enetc0;
+ eth1 = &enetc1;
+ eth2 = &enetc2;
+ eth3 = &enetc6;
+ eth4 = &felix0;
+ eth5 = &felix1;
+ eth6 = &felix2;
+ eth7 = &felix3;
+ };
+
  sysclk: sysclk {
  compatible = "fixed-clock";
  #clock-cells = <0>;
@@ -151,9 +162,51 @@
  reg = <0x000300 0 0 0 0>;
  status = "disabled";
  };
+ ethsw: pci@0,5 {
+ #address-cells=<0>;
+ #size-cells=<1>;
+ reg = <0x000500 0 0 0 0>;
+
+ ethsw_ports: ports {
+ #address-cells = <1>;
+ #size-cells = <0>;
+
+ felix0: port@0 {
+ reg = <0>;
+ status = "disabled";
+ label = "swp0";
+ };
+ felix1: port@1 {
+ reg = <1>;
+ status = "disabled";
+ label = "swp1";
+ };
+ felix2: port@2 {
+ reg = <2>;
+ status = "disabled";
+ label = "swp2";
+ };
+ felix3: port@3 {
+ reg = <3>;
+ status = "disabled";
+ label = "swp3";
+ };
+ port@4 {
+ reg = <4>;
+ phy-mode = "internal";
+ status = "okay";
+ ethernet = <&enetc2>;
+ };
+ port@5 {
+ reg = <5>;
+ phy-mode = "internal";
+ status = "disabled";
+ };
+ };
+ };
  enetc6: pci@0,6 {
  reg = <0x000600 0 0 0 0>;
- status = "okay";
+ status = "disabled";
  phy-mode = "internal";
  };
  };
--
2.17.1

Reply | Threaded
Open this post in threaded view
|

[PATCH 5/5] configs: ls1028a: Enable the Ethernet switch driver in defconfig

Claudiu Manoil-2
In reply to this post by Claudiu Manoil-2
From: Alex Marginean <[hidden email]>

The switch driver for LS1028A Ethernet switch is now compiled in for
both LS1028A boards.

Signed-off-by: Alex Marginean <[hidden email]>
Signed-off-by: Claudiu Manoil <[hidden email]>
---
 configs/ls1028aqds_tfa_SECURE_BOOT_defconfig | 2 ++
 configs/ls1028aqds_tfa_defconfig             | 2 ++
 configs/ls1028ardb_tfa_SECURE_BOOT_defconfig | 2 ++
 configs/ls1028ardb_tfa_defconfig             | 2 ++
 4 files changed, 8 insertions(+)

diff --git a/configs/ls1028aqds_tfa_SECURE_BOOT_defconfig b/configs/ls1028aqds_tfa_SECURE_BOOT_defconfig
index 14c49cd0d6..43ddcf9166 100644
--- a/configs/ls1028aqds_tfa_SECURE_BOOT_defconfig
+++ b/configs/ls1028aqds_tfa_SECURE_BOOT_defconfig
@@ -61,6 +61,8 @@ CONFIG_DM_MDIO_MUX=y
 CONFIG_E1000=y
 CONFIG_FSL_ENETC=y
 CONFIG_MDIO_MUX_I2CREG=y
+CONFIG_DM_DSA=y
+CONFIG_MSCC_FELIX_SWITCH=y
 CONFIG_NVME=y
 CONFIG_PCI=y
 CONFIG_DM_PCI=y
diff --git a/configs/ls1028aqds_tfa_defconfig b/configs/ls1028aqds_tfa_defconfig
index 09a6923708..c790e99485 100644
--- a/configs/ls1028aqds_tfa_defconfig
+++ b/configs/ls1028aqds_tfa_defconfig
@@ -67,6 +67,8 @@ CONFIG_DM_MDIO_MUX=y
 CONFIG_E1000=y
 CONFIG_FSL_ENETC=y
 CONFIG_MDIO_MUX_I2CREG=y
+CONFIG_DM_DSA=y
+CONFIG_MSCC_FELIX_SWITCH=y
 CONFIG_NVME=y
 CONFIG_PCI=y
 CONFIG_DM_PCI=y
diff --git a/configs/ls1028ardb_tfa_SECURE_BOOT_defconfig b/configs/ls1028ardb_tfa_SECURE_BOOT_defconfig
index b034580aef..8b9e217c61 100644
--- a/configs/ls1028ardb_tfa_SECURE_BOOT_defconfig
+++ b/configs/ls1028ardb_tfa_SECURE_BOOT_defconfig
@@ -58,6 +58,8 @@ CONFIG_DM_MDIO=y
 CONFIG_PHY_GIGE=y
 CONFIG_E1000=y
 CONFIG_FSL_ENETC=y
+CONFIG_DM_DSA=y
+CONFIG_MSCC_FELIX_SWITCH=y
 CONFIG_NVME=y
 CONFIG_PCI=y
 CONFIG_DM_PCI=y
diff --git a/configs/ls1028ardb_tfa_defconfig b/configs/ls1028ardb_tfa_defconfig
index 4bed352420..aa82719e5e 100644
--- a/configs/ls1028ardb_tfa_defconfig
+++ b/configs/ls1028ardb_tfa_defconfig
@@ -64,6 +64,8 @@ CONFIG_DM_MDIO=y
 CONFIG_PHY_GIGE=y
 CONFIG_E1000=y
 CONFIG_FSL_ENETC=y
+CONFIG_DM_DSA=y
+CONFIG_MSCC_FELIX_SWITCH=y
 CONFIG_NVME=y
 CONFIG_PCI=y
 CONFIG_DM_PCI=y
--
2.17.1

Reply | Threaded
Open this post in threaded view
|

Re: [PATCH 4/5] arm: dts: ls1028a: Add Ethernet switch node and dependencies

Michael Walle-2
In reply to this post by Claudiu Manoil-2
Am 2021-01-13 19:05, schrieb Claudiu Manoil:

> From: Alex Marginean <[hidden email]>
>
> The definition follows the DSA binding in kernel and describes the
> switch,
> its ports and PHYs.
> ENETC PF6 is the 2nd Eth controller linked to the switch on LS1028A, it
> is
> not used in U-Boot and was disabled.  Ethernet port aliases were also
> added to better manage the multitude of ports available now, and to
> enforce the order in which master and slave ports are probed.
>
> Signed-off-by: Alex Marginean <[hidden email]>
> Signed-off-by: Claudiu Manoil <[hidden email]>
> ---
>  arch/arm/dts/fsl-ls1028a-rdb.dts | 36 +++++++++++++++++++++
>  arch/arm/dts/fsl-ls1028a.dtsi    | 55 +++++++++++++++++++++++++++++++-
>  2 files changed, 90 insertions(+), 1 deletion(-)
>
> diff --git a/arch/arm/dts/fsl-ls1028a-rdb.dts
> b/arch/arm/dts/fsl-ls1028a-rdb.dts
> index 85b4815b2e..92d83a5c0c 100644
> --- a/arch/arm/dts/fsl-ls1028a-rdb.dts
> +++ b/arch/arm/dts/fsl-ls1028a-rdb.dts
> @@ -131,9 +131,45 @@
>   phy-handle = <&rdb_phy0>;
>  };
>
> +&ethsw_ports {
> + port@0 {
> + status = "okay";
> + phy-mode = "qsgmii";
> + phy-handle = <&sw_phy0>;
> + };
> + port@1 {
> + status = "okay";
> + phy-mode = "qsgmii";
> + phy-handle = <&sw_phy1>;
> + };
> + port@2 {
> + status = "okay";
> + phy-mode = "qsgmii";
> + phy-handle = <&sw_phy2>;
> + };
> + port@3 {
> + status = "okay";
> + phy-mode = "qsgmii";
> + phy-handle = <&sw_phy3>;
> + };
> +};
> +
>  &mdio0 {
>   status = "okay";
>   rdb_phy0: phy@2 {
>   reg = <2>;
>   };
> +
> + sw_phy0: phy@10 {
> + reg = <0x10>;
> + };
> + sw_phy1: phy@11 {
> + reg = <0x11>;
> + };
> + sw_phy2: phy@12 {
> + reg = <0x12>;
> + };
> + sw_phy3: phy@13 {
> + reg = <0x13>;
> + };
>  };
> diff --git a/arch/arm/dts/fsl-ls1028a.dtsi
> b/arch/arm/dts/fsl-ls1028a.dtsi
> index d0850237c7..e73769392f 100644
> --- a/arch/arm/dts/fsl-ls1028a.dtsi
> +++ b/arch/arm/dts/fsl-ls1028a.dtsi
> @@ -14,6 +14,17 @@
>   #address-cells = <2>;
>   #size-cells = <2>;
>
> + aliases {
> + eth0 = &enetc0;
> + eth1 = &enetc1;
> + eth2 = &enetc2;
> + eth3 = &enetc6;
> + eth4 = &felix0;
> + eth5 = &felix1;
> + eth6 = &felix2;
> + eth7 = &felix3;
> + };

Don't include the aliases in the common dtsi. There are serveral
reasons for that:
  (1) it is really board dependent. not every board has all these
      ports.
  (2) it will mess up the device numbering for boards which use
      this dtsi. And with this it will also mess up the ethNaddr
      environment variable logic.

Please move them into the corresponding boards.

> +
>   sysclk: sysclk {
>   compatible = "fixed-clock";
>   #clock-cells = <0>;
> @@ -151,9 +162,51 @@
>   reg = <0x000300 0 0 0 0>;
>   status = "disabled";
>   };
> + ethsw: pci@0,5 {
> + #address-cells=<0>;
> + #size-cells=<1>;
> + reg = <0x000500 0 0 0 0>;
> +
> + ethsw_ports: ports {
> + #address-cells = <1>;
> + #size-cells = <0>;
> +
> + felix0: port@0 {
> + reg = <0>;
> + status = "disabled";
> + label = "swp0";
> + };
> + felix1: port@1 {
> + reg = <1>;
> + status = "disabled";
> + label = "swp1";
> + };
> + felix2: port@2 {
> + reg = <2>;
> + status = "disabled";
> + label = "swp2";
> + };
> + felix3: port@3 {
> + reg = <3>;
> + status = "disabled";
> + label = "swp3";
> + };
> + port@4 {
> + reg = <4>;
> + phy-mode = "internal";
> + status = "okay";
> + ethernet = <&enetc2>;
> + };

status = "disabled".

Why would you enable just this port if all the switch ports
are disabled.

> + port@5 {
> + reg = <5>;
> + phy-mode = "internal";
> + status = "disabled";
> + };
> + };
> + };
>   enetc6: pci@0,6 {
>   reg = <0x000600 0 0 0 0>;
> - status = "okay";
> + status = "disabled";
>   phy-mode = "internal";
>   };
>   };

-michael
Reply | Threaded
Open this post in threaded view
|

RE: [PATCH 4/5] arm: dts: ls1028a: Add Ethernet switch node and dependencies

Claudiu Manoil-2

>-----Original Message-----
>From: Michael Walle <[hidden email]>
>Sent: Wednesday, January 13, 2021 10:11 PM
[...]

>> --- a/arch/arm/dts/fsl-ls1028a.dtsi
>> +++ b/arch/arm/dts/fsl-ls1028a.dtsi
>> @@ -14,6 +14,17 @@
>>   #address-cells = <2>;
>>   #size-cells = <2>;
>>
>> + aliases {
>> + eth0 = &enetc0;
>> + eth1 = &enetc1;
>> + eth2 = &enetc2;
>> + eth3 = &enetc6;
>> + eth4 = &felix0;
>> + eth5 = &felix1;
>> + eth6 = &felix2;
>> + eth7 = &felix3;
>> + };
>
>Don't include the aliases in the common dtsi. There are serveral
>reasons for that:
>  (1) it is really board dependent. not every board has all these
>      ports.
>  (2) it will mess up the device numbering for boards which use
>      this dtsi. And with this it will also mess up the ethNaddr
>      environment variable logic.
>
>Please move them into the corresponding boards.
>

Hi Michael,

I think the point of this patch was to enable the felix switch for the LS1028A RDB only for now,
for simplicity, to make the patch set smaller. Follow-up patches would enable it for the remaining
boards.
But I understand that keeping the aliases in the fsl-ls1028a.dtsi can mess up other boards
that include it, including the Kontron boards.  Would you agree to update only the ls1028 RDB
board DT for now?

Alternatively you could test these patches on your boards and maybe provide the DT updates for
the Kontron boards as part of this patch set.

>> +
>>   sysclk: sysclk {
>>   compatible = "fixed-clock";
>>   #clock-cells = <0>;
>> @@ -151,9 +162,51 @@
>>   reg = <0x000300 0 0 0 0>;
>>   status = "disabled";
>>   };
>> + ethsw: pci@0,5 {
>> + #address-cells=<0>;
>> + #size-cells=<1>;
>> + reg = <0x000500 0 0 0 0>;
>> +
>> + ethsw_ports: ports {
>> + #address-cells = <1>;
>> + #size-cells = <0>;
>> +
>> + felix0: port@0 {
>> + reg = <0>;
>> + status = "disabled";
>> + label = "swp0";
>> + };
>> + felix1: port@1 {
>> + reg = <1>;
>> + status = "disabled";
>> + label = "swp1";
>> + };
>> + felix2: port@2 {
>> + reg = <2>;
>> + status = "disabled";
>> + label = "swp2";
>> + };
>> + felix3: port@3 {
>> + reg = <3>;
>> + status = "disabled";
>> + label = "swp3";
>> + };
>> + port@4 {
>> + reg = <4>;
>> + phy-mode = "internal";
>> + status = "okay";
>> + ethernet = <&enetc2>;
>> + };
>
>status = "disabled".
>
>Why would you enable just this port if all the switch ports
>are disabled.
>

The number of active front panel ports may vary from board to board.
These ports are supposed to be enabled in each board DT, depending
on setup. But a DSA switch always needs a CPU port regardless of how
many front side ports are active in each particular setup, and port@4 is
the CPU port.

Thanks,
Claudiu
Reply | Threaded
Open this post in threaded view
|

Re: [PATCH 4/5] arm: dts: ls1028a: Add Ethernet switch node and dependencies

Michael Walle-2
Hi Claudiu,

Am 2021-01-14 16:20, schrieb Claudiu Manoil:

>> -----Original Message-----
>> From: Michael Walle <[hidden email]>
>> Sent: Wednesday, January 13, 2021 10:11 PM
> [...]
>>> --- a/arch/arm/dts/fsl-ls1028a.dtsi
>>> +++ b/arch/arm/dts/fsl-ls1028a.dtsi
>>> @@ -14,6 +14,17 @@
>>>   #address-cells = <2>;
>>>   #size-cells = <2>;
>>>
>>> + aliases {
>>> + eth0 = &enetc0;
>>> + eth1 = &enetc1;
>>> + eth2 = &enetc2;
>>> + eth3 = &enetc6;
>>> + eth4 = &felix0;
>>> + eth5 = &felix1;
>>> + eth6 = &felix2;
>>> + eth7 = &felix3;
>>> + };
>>
>> Don't include the aliases in the common dtsi. There are serveral
>> reasons for that:
>>  (1) it is really board dependent. not every board has all these
>>      ports.
>>  (2) it will mess up the device numbering for boards which use
>>      this dtsi. And with this it will also mess up the ethNaddr
>>      environment variable logic.
>>
>> Please move them into the corresponding boards.
>>
> I think the point of this patch was to enable the felix switch for the
> LS1028A RDB only for now,
> for simplicity, to make the patch set smaller. Follow-up patches would
> enable it for the remaining
> boards.
> But I understand that keeping the aliases in the fsl-ls1028a.dtsi can
> mess up other boards
> that include it, including the Kontron boards.  Would you agree to
> update only the ls1028 RDB
> board DT for now?

Sure, once accepted, I'll post a follow-up for our board.

> Alternatively you could test these patches on your boards and maybe
> provide the DT updates for
> the Kontron boards as part of this patch set.

I'm going to test this anyways and add a Tested-by to this set, once it
tested successfully.

At the moment the only board variant which this would apply to is not
upstream yet. Patches are pending. But sure, if it will be in upstream
you could pick it up for this set.

>>> +
>>>   sysclk: sysclk {
>>>   compatible = "fixed-clock";
>>>   #clock-cells = <0>;
>>> @@ -151,9 +162,51 @@
>>>   reg = <0x000300 0 0 0 0>;
>>>   status = "disabled";
>>>   };
>>> + ethsw: pci@0,5 {
>>> + #address-cells=<0>;
>>> + #size-cells=<1>;
>>> + reg = <0x000500 0 0 0 0>;

This should also have
  status = "disabled"

>>> +
>>> + ethsw_ports: ports {
>>> + #address-cells = <1>;
>>> + #size-cells = <0>;
>>> +
>>> + felix0: port@0 {
>>> + reg = <0>;
>>> + status = "disabled";
>>> + label = "swp0";
>>> + };
>>> + felix1: port@1 {
>>> + reg = <1>;
>>> + status = "disabled";
>>> + label = "swp1";
>>> + };
>>> + felix2: port@2 {
>>> + reg = <2>;
>>> + status = "disabled";
>>> + label = "swp2";
>>> + };
>>> + felix3: port@3 {
>>> + reg = <3>;
>>> + status = "disabled";
>>> + label = "swp3";
>>> + };
>>> + port@4 {
>>> + reg = <4>;
>>> + phy-mode = "internal";
>>> + status = "okay";
>>> + ethernet = <&enetc2>;
>>> + };
>>
>> status = "disabled".
>>
>> Why would you enable just this port if all the switch ports
>> are disabled.
>>
>
> The number of active front panel ports may vary from board to board.
> These ports are supposed to be enabled in each board DT, depending
> on setup. But a DSA switch always needs a CPU port regardless of how
> many front side ports are active in each particular setup, and port@4
> is
> the CPU port.

Sure, but my point was that the default should be disabled - as it
should
be for any peripheral in the dtsi - and it should be enabled per board.
What
if I'd rather use the other internal port? The linux dtsi also has all
ports disabled by default. Speaking of the linux device tree. The u-boot
one and the linux one are out-of-sync. Is there a reason why you
couldn't
take the "ethernet-switch@0,5" fragment from the linux tree?

-michael
Reply | Threaded
Open this post in threaded view
|

Re: [PATCH 1/5] net: Introduce DSA class for Ethernet switches

Simon Glass-3
In reply to this post by Claudiu Manoil-2
Hi Claudiu,

On Wed, 13 Jan 2021 at 11:05, Claudiu Manoil <[hidden email]> wrote:

>
> DSA stands for Distributed Switch Architecture and it covers switches that
> are connected to the CPU through an Ethernet link and generally use frame
> tags to pass information about the source/destination ports to/from CPU.
> Front panel ports are presented as regular ethernet devices in U-Boot and
> they are expected to support the typical networking commands.
> DSA switches may be cascaded, DSA class code does not currently support
> this.
>
> Signed-off-by: Alex Marginean <[hidden email]>
> Signed-off-by: Claudiu Manoil <[hidden email]>
> ---
>  drivers/net/Kconfig    |  14 ++
>  include/dm/uclass-id.h |   1 +
>  include/net.h          |   6 +
>  include/net/dsa.h      | 206 ++++++++++++++++
>  net/Makefile           |   1 +
>  net/dsa-uclass.c       | 517 +++++++++++++++++++++++++++++++++++++++++
>  6 files changed, 745 insertions(+)
>  create mode 100644 include/net/dsa.h
>  create mode 100644 net/dsa-uclass.c

Reviewed-by: Simon Glass <[hidden email]>

I don't think it is necessary to have the 'if (!pdev)' checks around
the place. We need a way in U-Boot to have checks like that to catch
programming errors  but to be able to turn them off in production code
to reduce size.

I suppose a Kconfig would do it, with:

if (CONFIG_IS_ENABLED(SAFETY) && !pdev)
    return log_,msg_ref("safety", -ENODEV)

Also note that -ENODEV is used by drive rmodel so it generally isn't
safe to return it as a logic error. I think in this case because it
never happens, it should be OK.

Regards,
Simon
Reply | Threaded
Open this post in threaded view
|

Re: [PATCH 2/5] sandbox: Add a DSA sandbox driver and unit test

Simon Glass-3
In reply to this post by Claudiu Manoil-2
On Wed, 13 Jan 2021 at 11:05, Claudiu Manoil <[hidden email]> wrote:

>
> The DSA sandbox driver is used for unit testing the DSA class code.
> It implements a simple 2 port switch plus 1 CPU port, and uses a very
> simple tag to identify the ports.
> The DSA sandbox device is connected via CPU port to a regular Ethernet
> sandbox device, called 'dsa-test-eth, managed by the existing eth sandbox
> driver.  The 'dsa-test-eth' is not intended for testing the eth class
> code however, but it is used to emulate traffic through the 'lan0' and
> 'lan1' front pannel switch ports.  To achieve this the dsa sandbox driver
> registers a tx handler for the 'dsa-test-eth' device.  The switch ports,
> labeled as 'lan0' and 'lan1', are also registered as eth devices by the
> dsa class code this time.  So pinging through these switch ports is as
> easy as:
> => setenv ethact lan0
> => ping 1.2.3.5
>
> Unit tests for the dsa class code were also added.  The 'dsa_probe'
> test exercises most API functions from dsa.h.  The 'dsa' unit test
> simply exercises ARP/ICMP traffic through the two switch ports,
> including tag injection and extraction, with the help of the dsa
> sandbox driver.
> I took care to minimize the impact on the existing eth unit tests,
> though some adjustments needed to be made with the addition of
> extra eth interfaces used by the dsa unit tests. The additional eth
> interfaces also require MAC addresses, these have been added to the
> sandbox default environment.
>
> Signed-off-by: Alex Marginean <[hidden email]>
> Signed-off-by: Claudiu Manoil <[hidden email]>
> ---
>  arch/Kconfig              |   1 +
>  arch/sandbox/dts/test.dts |  40 +++++++++
>  drivers/net/Kconfig       |   9 ++
>  drivers/net/Makefile      |   1 +
>  drivers/net/dsa_sandbox.c | 181 ++++++++++++++++++++++++++++++++++++++
>  include/configs/sandbox.h |   2 +
>  test/dm/Makefile          |   1 +
>  test/dm/dsa.c             |  91 +++++++++++++++++++
>  test/dm/eth.c             |  10 +--
>  9 files changed, 331 insertions(+), 5 deletions(-)
>  create mode 100644 drivers/net/dsa_sandbox.c
>  create mode 100644 test/dm/dsa.c
>

Reviewed-by: Simon Glass <[hidden email]>
Reply | Threaded
Open this post in threaded view
|

RE: [PATCH 1/5] net: Introduce DSA class for Ethernet switches

Claudiu Manoil-2
In reply to this post by Simon Glass-3
>-----Original Message-----
>From: Simon Glass <[hidden email]>
>Sent: Thursday, January 14, 2021 5:42 PM
>To: Claudiu Manoil <[hidden email]>
>Cc: Joe Hershberger <[hidden email]>; Bin Meng
><[hidden email]>; Michael Walle <[hidden email]>; U-Boot Mailing
>List <[hidden email]>; Vladimir Oltean <[hidden email]>;
>Alexandru Marginean <[hidden email]>
>Subject: Re: [PATCH 1/5] net: Introduce DSA class for Ethernet switches
>
[...]

>
>Reviewed-by: Simon Glass <[hidden email]>
>
>I don't think it is necessary to have the 'if (!pdev)' checks around
>the place. We need a way in U-Boot to have checks like that to catch
>programming errors  but to be able to turn them off in production code
>to reduce size.
>
>I suppose a Kconfig would do it, with:
>
>if (CONFIG_IS_ENABLED(SAFETY) && !pdev)
>    return log_,msg_ref("safety", -ENODEV)
>
>Also note that -ENODEV is used by drive rmodel so it generally isn't
>safe to return it as a logic error. I think in this case because it
>never happens, it should be OK.
>

Thanks for the review, Simon.
I thought about using assert(pdev) checks, but during development the
simple "if (!pdev)..." proved more friendly.  I like your idea about enabling
the checks at compile time and disabling them in production.
For now, since this SAFETY flag is not implemented, my understanding is
that you’re ok with leaving the pdev checks in the code as they are right now
and sometime in the future these will be converted to the "SAFETY" construct
you mention.


Reply | Threaded
Open this post in threaded view
|

RE: [PATCH 4/5] arm: dts: ls1028a: Add Ethernet switch node and dependencies

Claudiu Manoil-2
In reply to this post by Michael Walle-2
>-----Original Message-----
>From: Michael Walle <[hidden email]>
>Sent: Thursday, January 14, 2021 5:40 PM
>To: Claudiu Manoil <[hidden email]>
>Cc: Joe Hershberger <[hidden email]>; Simon Glass
><[hidden email]>; Bin Meng <[hidden email]>; u-
>[hidden email]; Vladimir Oltean <[hidden email]>; Alexandru
>Marginean <[hidden email]>
>Subject: Re: [PATCH 4/5] arm: dts: ls1028a: Add Ethernet switch node and
>dependencies
>
>Hi Claudiu,
>
>Am 2021-01-14 16:20, schrieb Claudiu Manoil:
>>> -----Original Message-----
>>> From: Michael Walle <[hidden email]>
>>> Sent: Wednesday, January 13, 2021 10:11 PM
>> [...]
>>>> --- a/arch/arm/dts/fsl-ls1028a.dtsi
>>>> +++ b/arch/arm/dts/fsl-ls1028a.dtsi
>>>> @@ -14,6 +14,17 @@
>>>>   #address-cells = <2>;
>>>>   #size-cells = <2>;
>>>>
>>>> + aliases {
>>>> + eth0 = &enetc0;
>>>> + eth1 = &enetc1;
>>>> + eth2 = &enetc2;
>>>> + eth3 = &enetc6;
>>>> + eth4 = &felix0;
>>>> + eth5 = &felix1;
>>>> + eth6 = &felix2;
>>>> + eth7 = &felix3;
>>>> + };
>>>
>>> Don't include the aliases in the common dtsi. There are serveral
>>> reasons for that:
>>>  (1) it is really board dependent. not every board has all these
>>>      ports.
>>>  (2) it will mess up the device numbering for boards which use
>>>      this dtsi. And with this it will also mess up the ethNaddr
>>>      environment variable logic.
>>>
>>> Please move them into the corresponding boards.
>>>
>> I think the point of this patch was to enable the felix switch for the
>> LS1028A RDB only for now,
>> for simplicity, to make the patch set smaller. Follow-up patches would
>> enable it for the remaining
>> boards.
>> But I understand that keeping the aliases in the fsl-ls1028a.dtsi can
>> mess up other boards
>> that include it, including the Kontron boards.  Would you agree to
>> update only the ls1028 RDB
>> board DT for now?
>
>Sure, once accepted, I'll post a follow-up for our board.
>
>> Alternatively you could test these patches on your boards and maybe
>> provide the DT updates for
>> the Kontron boards as part of this patch set.
>
>I'm going to test this anyways and add a Tested-by to this set, once it
>tested successfully.
>
>At the moment the only board variant which this would apply to is not
>upstream yet. Patches are pending. But sure, if it will be in upstream
>you could pick it up for this set.
>
>>>> +
>>>>   sysclk: sysclk {
>>>>   compatible = "fixed-clock";
>>>>   #clock-cells = <0>;
>>>> @@ -151,9 +162,51 @@
>>>>   reg = <0x000300 0 0 0 0>;
>>>>   status = "disabled";
>>>>   };
>>>> + ethsw: pci@0,5 {
>>>> + #address-cells=<0>;
>>>> + #size-cells=<1>;
>>>> + reg = <0x000500 0 0 0 0>;
>
>This should also have
>  status = "disabled"
>
>>>> +
>>>> + ethsw_ports: ports {
>>>> + #address-cells = <1>;
>>>> + #size-cells = <0>;
>>>> +
>>>> + felix0: port@0 {
>>>> + reg = <0>;
>>>> + status = "disabled";
>>>> + label = "swp0";
>>>> + };
>>>> + felix1: port@1 {
>>>> + reg = <1>;
>>>> + status = "disabled";
>>>> + label = "swp1";
>>>> + };
>>>> + felix2: port@2 {
>>>> + reg = <2>;
>>>> + status = "disabled";
>>>> + label = "swp2";
>>>> + };
>>>> + felix3: port@3 {
>>>> + reg = <3>;
>>>> + status = "disabled";
>>>> + label = "swp3";
>>>> + };
>>>> + port@4 {
>>>> + reg = <4>;
>>>> + phy-mode = "internal";
>>>> + status = "okay";
>>>> + ethernet = <&enetc2>;
>>>> + };
>>>
>>> status = "disabled".
>>>
>>> Why would you enable just this port if all the switch ports
>>> are disabled.
>>>
>>
>> The number of active front panel ports may vary from board to board.
>> These ports are supposed to be enabled in each board DT, depending
>> on setup. But a DSA switch always needs a CPU port regardless of how
>> many front side ports are active in each particular setup, and port@4
>> is
>> the CPU port.
>
>Sure, but my point was that the default should be disabled - as it
>should be for any peripheral in the dtsi - and it should be enabled per
>board.
>What if I'd rather use the other internal port? The linux dtsi also has all
>ports disabled by default. Speaking of the linux device tree. The u-boot
>one and the linux one are out-of-sync. Is there a reason why you
>couldn't take the "ethernet-switch@0,5" fragment from the linux tree?
>

Michel, I agree with keeping the device tree nodes between linux and
uboot as consistent as possible, so port 5 can look like this in the dtsi :
mscc_felix_port5:  port@5 {
        reg = <5>;
        phy-mode = "internal";
        status = "disabled";

        fixed-link {
                speed = <1000>;
                full-duplex;
        };
};
as it is currently in the upstream kernel:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/arch/arm64/boot/dts/freescale/fsl-ls1028a.dtsi?h=v5.11-rc3

I'd drop however properties that are not used, like
managed = "in-band-status" for the external ports, since this is
phylink specific, and I don't expect we'll have phylink in uboot
anytime soon.
Reply | Threaded
Open this post in threaded view
|

Re: [PATCH 1/5] net: Introduce DSA class for Ethernet switches

Simon Glass-3
In reply to this post by Claudiu Manoil-2
Hi Claudiu,

On Fri, 15 Jan 2021 at 09:47, Claudiu Manoil <[hidden email]> wrote:

>
> >-----Original Message-----
> >From: Simon Glass <[hidden email]>
> >Sent: Thursday, January 14, 2021 5:42 PM
> >To: Claudiu Manoil <[hidden email]>
> >Cc: Joe Hershberger <[hidden email]>; Bin Meng
> ><[hidden email]>; Michael Walle <[hidden email]>; U-Boot Mailing
> >List <[hidden email]>; Vladimir Oltean <[hidden email]>;
> >Alexandru Marginean <[hidden email]>
> >Subject: Re: [PATCH 1/5] net: Introduce DSA class for Ethernet switches
> >
> [...]
> >
> >Reviewed-by: Simon Glass <[hidden email]>
> >
> >I don't think it is necessary to have the 'if (!pdev)' checks around
> >the place. We need a way in U-Boot to have checks like that to catch
> >programming errors  but to be able to turn them off in production code
> >to reduce size.
> >
> >I suppose a Kconfig would do it, with:
> >
> >if (CONFIG_IS_ENABLED(SAFETY) && !pdev)
> >    return log_,msg_ref("safety", -ENODEV)
> >
> >Also note that -ENODEV is used by drive rmodel so it generally isn't
> >safe to return it as a logic error. I think in this case because it
> >never happens, it should be OK.
> >
>
> Thanks for the review, Simon.
> I thought about using assert(pdev) checks, but during development the
> simple "if (!pdev)..." proved more friendly.  I like your idea about enabling
> the checks at compile time and disabling them in production.
> For now, since this SAFETY flag is not implemented, my understanding is
> that you’re ok with leaving the pdev checks in the code as they are right now
> and sometime in the future these will be converted to the "SAFETY" construct
> you mention.
>

Yes that's fine, you have my review tag.

+Tom Rini what do you think about CONFIG_SAFETY or similar to allow
these bug checks to be disabled for code-size reasons?

Regards,
Simon
Reply | Threaded
Open this post in threaded view
|

Re: [PATCH 1/5] net: Introduce DSA class for Ethernet switches

Tom Rini-4
On Tue, Jan 19, 2021 at 11:06:10AM -0700, Simon Glass wrote:

> Hi Claudiu,
>
> On Fri, 15 Jan 2021 at 09:47, Claudiu Manoil <[hidden email]> wrote:
> >
> > >-----Original Message-----
> > >From: Simon Glass <[hidden email]>
> > >Sent: Thursday, January 14, 2021 5:42 PM
> > >To: Claudiu Manoil <[hidden email]>
> > >Cc: Joe Hershberger <[hidden email]>; Bin Meng
> > ><[hidden email]>; Michael Walle <[hidden email]>; U-Boot Mailing
> > >List <[hidden email]>; Vladimir Oltean <[hidden email]>;
> > >Alexandru Marginean <[hidden email]>
> > >Subject: Re: [PATCH 1/5] net: Introduce DSA class for Ethernet switches
> > >
> > [...]
> > >
> > >Reviewed-by: Simon Glass <[hidden email]>
> > >
> > >I don't think it is necessary to have the 'if (!pdev)' checks around
> > >the place. We need a way in U-Boot to have checks like that to catch
> > >programming errors  but to be able to turn them off in production code
> > >to reduce size.
> > >
> > >I suppose a Kconfig would do it, with:
> > >
> > >if (CONFIG_IS_ENABLED(SAFETY) && !pdev)
> > >    return log_,msg_ref("safety", -ENODEV)
> > >
> > >Also note that -ENODEV is used by drive rmodel so it generally isn't
> > >safe to return it as a logic error. I think in this case because it
> > >never happens, it should be OK.
> > >
> >
> > Thanks for the review, Simon.
> > I thought about using assert(pdev) checks, but during development the
> > simple "if (!pdev)..." proved more friendly.  I like your idea about enabling
> > the checks at compile time and disabling them in production.
> > For now, since this SAFETY flag is not implemented, my understanding is
> > that you’re ok with leaving the pdev checks in the code as they are right now
> > and sometime in the future these will be converted to the "SAFETY" construct
> > you mention.
> >
>
> Yes that's fine, you have my review tag.
>
> +Tom Rini what do you think about CONFIG_SAFETY or similar to allow
> these bug checks to be disabled for code-size reasons?
I don't know.  Setting aside the name, my first concern is "so we
disable certain forms of sanity checks, now assuming a malicious entity
somewhere, what's now able to be exploited?"

--
Tom

signature.asc (673 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: [PATCH 1/5] net: Introduce DSA class for Ethernet switches

Simon Glass-3
Hi Tom,

On Tue, 19 Jan 2021 at 14:00, Tom Rini <[hidden email]> wrote:

>
> On Tue, Jan 19, 2021 at 11:06:10AM -0700, Simon Glass wrote:
> > Hi Claudiu,
> >
> > On Fri, 15 Jan 2021 at 09:47, Claudiu Manoil <[hidden email]> wrote:
> > >
> > > >-----Original Message-----
> > > >From: Simon Glass <[hidden email]>
> > > >Sent: Thursday, January 14, 2021 5:42 PM
> > > >To: Claudiu Manoil <[hidden email]>
> > > >Cc: Joe Hershberger <[hidden email]>; Bin Meng
> > > ><[hidden email]>; Michael Walle <[hidden email]>; U-Boot Mailing
> > > >List <[hidden email]>; Vladimir Oltean <[hidden email]>;
> > > >Alexandru Marginean <[hidden email]>
> > > >Subject: Re: [PATCH 1/5] net: Introduce DSA class for Ethernet switches
> > > >
> > > [...]
> > > >
> > > >Reviewed-by: Simon Glass <[hidden email]>
> > > >
> > > >I don't think it is necessary to have the 'if (!pdev)' checks around
> > > >the place. We need a way in U-Boot to have checks like that to catch
> > > >programming errors  but to be able to turn them off in production code
> > > >to reduce size.
> > > >
> > > >I suppose a Kconfig would do it, with:
> > > >
> > > >if (CONFIG_IS_ENABLED(SAFETY) && !pdev)
> > > >    return log_,msg_ref("safety", -ENODEV)
> > > >
> > > >Also note that -ENODEV is used by drive rmodel so it generally isn't
> > > >safe to return it as a logic error. I think in this case because it
> > > >never happens, it should be OK.
> > > >
> > >
> > > Thanks for the review, Simon.
> > > I thought about using assert(pdev) checks, but during development the
> > > simple "if (!pdev)..." proved more friendly.  I like your idea about enabling
> > > the checks at compile time and disabling them in production.
> > > For now, since this SAFETY flag is not implemented, my understanding is
> > > that you’re ok with leaving the pdev checks in the code as they are right now
> > > and sometime in the future these will be converted to the "SAFETY" construct
> > > you mention.
> > >
> >
> > Yes that's fine, you have my review tag.
> >
> > +Tom Rini what do you think about CONFIG_SAFETY or similar to allow
> > these bug checks to be disabled for code-size reasons?
>
> I don't know.  Setting aside the name, my first concern is "so we
> disable certain forms of sanity checks, now assuming a malicious entity
> somewhere, what's now able to be exploited?"

Well if you are able to inject code then I suppose you can do
anything. You don't need to worry about existing parameter checking.

The difference in my mind is whether there is user/input data
involved. If so, then we need lots of checks. If it is just telling
the clock to go a 2GHz, then the checks are probably just bloating the
code. So rather than leave this to individual preference, I am
wondering whether a Kconfig option might be best.

Regards,
Simon