Laguna: bridge and management port fixes

board/aarch64/microchip-ev23x71a/Config.in

@@ -1,6 +1,7 @@
 config BR2_PACKAGE_MICROCHIP_EV23X71A
 	bool "Microchip EV23X71A (Laguna)"
 	depends on BR2_aarch64
+	select BR2_PACKAGE_SYMREG
 	help
 	  Support for the Microchip EV23X71A evaluation board, based on
 	  the LAN969x (Laguna) family of TSN capable switches.  24 GbE

board/aarch64/microchip-ev23x71a/dts/microchip/infix.dtsi

@@ -13,3 +13,15 @@
 		};
 	};
 };
+
+/* Register access by name for the symreg tool, debugfs only */
+&axi {
+	symreg@e00c0000 {
+		compatible = "microchip,sparx5-symreg";
+		reg = <0xe00c0000 0x10000>,
+		      <0xe2000000 0x2000000>;
+		reg-names = "cpu", "dev";
+		microchip,symreg = <0xe00c0000 0x10000>,
+				   <0xe2000000 0x2000000>;
+	};
+};

board/aarch64/microchip-lan969x/README.md

@@ -0,0 +1,72 @@
+# Microchip LAN969x (Laguna)
+
+Files shared by the boards built on the LAN969x switch family:
+
+| Board                                           | Package                            |
+|-------------------------------------------------|------------------------------------|
+| [Microchip EV23X71A](../microchip-ev23x71a)     | `BR2_PACKAGE_MICROCHIP_EV23X71A`   |
+| [Novarq Tactical 1000](../novarq-tactical-1000) | `BR2_PACKAGE_NOVARQ_TACTICAL_1000` |
+
+`uboot/` holds the U-Boot config fragment and environment used by the
+`*_boot_defconfig` of each board.  Kernel device tree and configuration
+fixups stay with each board, see its `dts/` directory and `.mk` file.
+
+A new Laguna board should select `BR2_PACKAGE_SYMREG` in its `Config.in`
+and carry the `microchip,sparx5-symreg` node in its `dts/microchip/infix.dtsi`,
+copied from one of the boards above, so the tool below works on it too.
+
+## Debugging
+
+### Register access by name
+
+Every Laguna board ships `symreg`, Microchip's tool for reading and
+writing switch registers by their datasheet names.  It talks to the
+`symreg` debugfs driver, which maps the register space the device tree
+node describes and exposes it as `/sys/kernel/debug/symreg/mem`.  Root
+only, since debugfs is.
+
+```
+symreg --help                   # options and the register syntax
+symreg -m                       # MAC table
+symreg -v                       # VLAN table
+symreg -c 0                     # VCAP instance 0
+symreg -s                       # stream table
+
+symreg 'DEV2G5[29]*'            # all registers of port 29's device block
+symreg 'ASM*[29]*'              # ingress side of the same port
+symreg -a 'DSM*[29]*'           # same, with the physical addresses
+
+symreg REG[idx]                 # read one register
+symreg REG[idx].FIELD           # read one field
+symreg REG[idx].FIELD 1         # write one field, add -f for wildcards
+```
+
+Register and field names follow the LAN969x datasheet.  Quote the
+wildcards or the shell expands them.  The dispatcher `symreg` picks the
+SoC from `/proc/device-tree/compatible` and runs `symreg_lan969x`, which
+can also be called directly.
+
+The tool reads live hardware, so the useful pattern for an intermittent
+fault is a snapshot while the port works, another once it has failed,
+and a diff of the two.  For the management port, which is switch port 29
+on both boards:
+
+```bash
+for blk in 'DEV2G5[29]*' 'ASM*[29]*' 'DSM*[29]*' 'QSYS*[29]*'; do
+	symreg "$blk"
+done >/tmp/port29-$(date +%s).txt
+```
+
+Check the PHY as well.  The management PHY sits on the GCB MDIO bus and
+shares one level triggered interrupt line with the copper PHYs, so a
+stale link state can come from a missed PHY interrupt as well as from
+the switch.  Read it with `mdio` from mdio-tools, see the
+[EV23X71A README](../microchip-ev23x71a/README.md#debug-and-analysis),
+which also covers the VCAP debugfs and `ethtool` counters; both apply to
+either board.
+
+### Writing registers
+
+Writes take effect immediately and bypass the driver, which keeps no
+record of them.  Expect the driver's own view to disagree afterwards,
+and reboot before trusting any behaviour again.

board/aarch64/novarq-tactical-1000/Config.in

@@ -1,6 +1,7 @@
 config BR2_PACKAGE_NOVARQ_TACTICAL_1000
 	bool "Novarq Tactical 1000 (Laguna)"
 	depends on BR2_aarch64
+	select BR2_PACKAGE_SYMREG
 	help
 	  Support for the Novarq Tactical 1000, based on the LAN969x
 	  (Laguna) family of TSN capable switches.  24 GbE copper ports,

board/aarch64/novarq-tactical-1000/dts/microchip/infix.dtsi

@@ -13,3 +13,15 @@
 		};
 	};
 };
+
+/* Register access by name for the symreg tool, debugfs only */
+&axi {
+	symreg@e00c0000 {
+		compatible = "microchip,sparx5-symreg";
+		reg = <0xe00c0000 0x10000>,
+		      <0xe2000000 0x2000000>;
+		reg-names = "cpu", "dev";
+		microchip,symreg = <0xe00c0000 0x10000>,
+				   <0xe2000000 0x2000000>;
+	};
+};

board/aarch64/novarq-tactical-1000/rootfs/usr/libexec/infix/prod/provision-tactical

@@ -114,9 +114,17 @@ if ask "Create aux, primary ${imgsize}, secondary ${imgsize}, cfg ${cfgsize}, an
 	mkfs.ext4 -qF -L aux -m 0 -i 4096 -O ^metadata_csum,^metadata_csum_seed,uninit_bg "${disk}p4"
 	mkfs.ext4 -qF -L cfg -m 0 -i 4096 -O uninit_bg "${disk}p7"
 	mkfs.ext4 -qF -L var -m 0 -i 4096 -O uninit_bg "${disk}p8"
+	#
+	# RAUC tracks slot boot state in uboot.env, seed it as prod/provision does.
 	mnt=$(mktemp -d)
 	mount -t ext4 "${disk}p4" "$mnt"
 	echo 1 >"$mnt"/resized
+	mkenvimage -s 0x4000 -o "$mnt"/uboot.env - <<EOF
+BOOT_ORDER=primary secondary net
+BOOT_primary_LEFT=1
+BOOT_secondary_LEFT=1
+BOOT_net_LEFT=1
+EOF
 	umount "$mnt"
 	rmdir "$mnt"
 fi

board/common/rootfs/etc/bash.bashrc

@@ -33,21 +33,7 @@ bind "set show-all-if-ambiguous on"
 
 export LANG=C.UTF-8
 
-log()
-{
-    local fn="/var/log/syslog"
-    [ -n "$1" ] && fn="/var/log/$1"
-    less +G -r "$fn"
-}
-
-follow ()
-{
-    local fn="/var/log/syslog"
-    [ -n "$1" ] && fn="/var/log/$1"
-
-    tail -F -n +1 "$fn"
-}
-
+# log and follow are defined in /etc/profile.d/log.sh
 _logfile_completions()
 {
     local cur=${COMP_WORDS[COMP_CWORD]}

board/common/rootfs/etc/profile.d/log.sh

@@ -0,0 +1,15 @@
+# Page or follow a log file, default syslog.  Plain sh, so BusyBox ash
+# and bash both get it; the bash completions live in /etc/bash.bashrc.
+log()
+{
+	local fn="/var/log/syslog"
+	[ -n "$1" ] && fn="/var/log/$1"
+	less +G -r "$fn"
+}
+
+follow()
+{
+	local fn="/var/log/syslog"
+	[ -n "$1" ] && fn="/var/log/$1"
+	tail -F -n +1 "$fn"
+}

configs/aarch64_minimal_defconfig

@@ -111,13 +111,16 @@ BR2_PACKAGE_HOST_GENEXT2FS=y
 BR2_PACKAGE_HOST_UBOOT_TOOLS_FIT_SUPPORT=y
 BR2_PACKAGE_HOST_UBOOT_TOOLS_FIT_SIGNATURE_SUPPORT=y
 BR2_PACKAGE_HOST_UBOOT_TOOLS_FDT_ADD_PUBKEY=y
+BR2_PACKAGE_ACER_CONNECT_VERO_W6M=y
 BR2_PACKAGE_ALDER_ALDER=y
 BR2_PACKAGE_BANANAPI_BPI_R3=y
 BR2_PACKAGE_BANANAPI_BPI_R4=y
 BR2_PACKAGE_BANANAPI_BPI_R64=y
 BR2_PACKAGE_FRIENDLYARM_NANOPI_R2S=y
 BR2_PACKAGE_MARVELL_CN9130_CRB=y
 BR2_PACKAGE_MARVELL_ESPRESSOBIN=y
+BR2_PACKAGE_MICROCHIP_EV23X71A=y
+BR2_PACKAGE_NOVARQ_TACTICAL_1000=y
 BR2_PACKAGE_RASPBERRYPI_RPI64=y
 BR2_PACKAGE_STYX_DCP_SC_28P=y
 IX_VENDOR_HOME="https://www.kernelkit.org"

doc/ChangeLog.md

@@ -38,6 +38,9 @@ All notable changes to the project are documented in this file.
   the keystore.  Its default route and DNS servers are used like those from
   a DHCP server, and the TCP MSS of forwarded connections is clamped to the
   session MTU, see [PPPoE Client][pppoe]
+- LAN969x boards ship `symreg`, which reads and writes switch registers by
+  name and dumps the MAC, VLAN and VCAP tables, for debugging the switch
+  core, see `board/aarch64/microchip-lan969x/README.md`
 
 ### Fixes
 
@@ -59,6 +62,31 @@ All notable changes to the project are documented in this file.
 - Fix #1423: the default route from a DHCPv6 client, learned from router
   advertisements, is now a static route with the DHCPv6 route preference,
   like a DHCPv4 route.  Before, the route preference setting was ignored
+- LAN969x: a port outside any bridge stopped receiving multicast, including
+  IPv6 neighbour discovery and mDNS, when another port outside a bridge went
+  down or joined a bridge.  The port looked dead on the LAN until it was
+  taken down and up again, which is how the Tactical 1000 management port
+  kept dropping out of the test rig
+- LAN969x: kernel BUG "scheduling while atomic" when a multicast address was
+  added to or removed from a port outside any bridge
+- LAN969x: the IP address of a bridge was unreachable from its ports, while
+  forwarding between the ports worked.  Broadcast from a bridge port never
+  reached the CPU, so ARP went unanswered, and the bridge's own MAC address
+  was missing from the switch MAC table, so unicast to the bridge was
+  flooded to the ports instead of delivered
+- LAN969x: disabling the first port of a bridge, whose MAC address the
+  bridge uses, cut the bridge IP address off until the port was enabled
+  again
+- LAN969x: taking down one port of a link aggregate stopped unicast to the
+  aggregate on the remaining ports, and removing the aggregate cut off the
+  port whose address it had used until that port was disabled and enabled
+  again
+- PTP ports with PHY timestamping, e.g. the copper ports on LAN969x boards,
+  kept dropping to the faulty state because the transmit timestamp arrived
+  later than the 10 ms the PTP daemon waited for it, or never at all
+- PTP now time stamps with the clock of the switch core when the kernel
+  would have picked a PHY's, so all ports of a switch share one clock, which
+  boundary and transparent clocks and TSN schedules depend on
 
 [pppoe]: https://www.kernelkit.org/infix/latest/pppoe/
 

package/Config.in

@@ -40,6 +40,7 @@ source "$BR2_EXTERNAL_INFIX_PATH/package/python-spy/Config.in"
 source "$BR2_EXTERNAL_INFIX_PATH/package/python-libyang/Config.in"
 source "$BR2_EXTERNAL_INFIX_PATH/package/python-yangdoc/Config.in"
 source "$BR2_EXTERNAL_INFIX_PATH/package/skeleton-init-finit/Config.in"
+source "$BR2_EXTERNAL_INFIX_PATH/package/symreg/Config.in"
 source "$BR2_EXTERNAL_INFIX_PATH/package/tetris/Config.in"
 source "$BR2_EXTERNAL_INFIX_PATH/package/libyang-cpp/Config.in"
 source "$BR2_EXTERNAL_INFIX_PATH/package/sysrepo-cpp/Config.in"

package/symreg/Config.in

@@ -0,0 +1,10 @@
+config BR2_PACKAGE_SYMREG
+	bool "symreg - Microchip switch registers by name"
+	depends on BR2_aarch64
+	help
+	  Read and write Microchip LAN969x switch registers by symbolic
+	  name, and dump the MAC, VLAN, VCAP and stream tables, through
+	  the symreg debugfs driver.  Debug aid, root only.  Selected by
+	  the Laguna boards, see board/aarch64/microchip-lan969x/README.md
+
+	  https://github.com/microchip-ung/symreg-tool

package/symreg/symreg.hash

@@ -0,0 +1,3 @@
+# Locally calculated
+sha256  0d7eee2dbdd8e79bb91103d06ee057cb477764a046c5d6d8045e978166a70490  symreg-cb96993eeb34558926a8f5851a945ad4cd674690.tar.gz
+sha256  d620d728cc1c0efebe173f8840dd957a56d9ff40867c062d0571b4110ea15f1e  LICENSE

package/symreg/symreg.mk

@@ -0,0 +1,11 @@
+SYMREG_VERSION = cb96993eeb34558926a8f5851a945ad4cd674690
+SYMREG_SITE = $(call github,microchip-ung,symreg-tool,$(SYMREG_VERSION))
+SYMREG_LICENSE = MIT
+SYMREG_LICENSE_FILES = LICENSE
+SYMREG_CONF_OPTS = -DBUILD_LAN969X=ON
+
+define SYMREG_LINUX_CONFIG_FIXUPS
+	$(call KCONFIG_ENABLE_OPT,CONFIG_SPARX5_SYMREG)
+endef
+
+$(eval $(cmake-package))

patches/linux/6.18.55/0001-net-phy-marvell10g-Support-firmware-loading-on-88X33.patch

@@ -1,7 +1,7 @@
 From 1bdc682ca8860bb8299b6c25cc91ebf032a23cfa Mon Sep 17 00:00:00 2001
 From: Tobias Waldekranz <tobias@waldekranz.com>
 Date: Tue, 19 Sep 2023 18:38:10 +0200
-Subject: [PATCH 01/78] net: phy: marvell10g: Support firmware loading on
+Subject: [PATCH 01/85] net: phy: marvell10g: Support firmware loading on
  88X3310
 
 When probing, if a device is waiting for firmware to be loaded into
@@ -15,7 +15,7 @@ fall back.
  1 file changed, 161 insertions(+)
 
 diff --git a/drivers/net/phy/marvell10g.c b/drivers/net/phy/marvell10g.c
-index 8fd42131cdbf..44294519231e 100644
+index 8fd42131cdbf9..44294519231ed 100644
 --- a/drivers/net/phy/marvell10g.c
 +++ b/drivers/net/phy/marvell10g.c
 @@ -25,6 +25,7 @@

patches/linux/6.18.55/0002-net-phy-marvell10g-Fix-power-up-when-strapped-to-sta.patch

@@ -1,7 +1,7 @@
 From 920227e8923cf970730a4bed22bd2aa55dc9387e Mon Sep 17 00:00:00 2001
 From: Tobias Waldekranz <tobias@waldekranz.com>
 Date: Tue, 21 Nov 2023 20:15:24 +0100
-Subject: [PATCH 02/78] net: phy: marvell10g: Fix power-up when strapped to
+Subject: [PATCH 02/85] net: phy: marvell10g: Fix power-up when strapped to
  start powered down
 
 On devices which are hardware strapped to start powered down (PDSTATE
@@ -12,7 +12,7 @@ affected by this setting.
  1 file changed, 15 insertions(+), 3 deletions(-)
 
 diff --git a/drivers/net/phy/marvell10g.c b/drivers/net/phy/marvell10g.c
-index 44294519231e..40439db49601 100644
+index 44294519231ed..40439db49601d 100644
 --- a/drivers/net/phy/marvell10g.c
 +++ b/drivers/net/phy/marvell10g.c
 @@ -322,11 +322,23 @@ static int mv3310_power_down(struct phy_device *phydev)

patches/linux/6.18.55/0003-net-phy-marvell10g-Add-LED-support-for-88X3310.patch

@@ -1,7 +1,7 @@
 From c473017acbed8a778f07f9f1ce55028e85a9ad62 Mon Sep 17 00:00:00 2001
 From: Tobias Waldekranz <tobias@waldekranz.com>
 Date: Wed, 15 Nov 2023 20:58:42 +0100
-Subject: [PATCH 03/78] net: phy: marvell10g: Add LED support for 88X3310
+Subject: [PATCH 03/85] net: phy: marvell10g: Add LED support for 88X3310
 
 Pickup the LEDs from the state in which the hardware reset or
 bootloader left them, but also support further configuration via
@@ -21,7 +21,7 @@ Trigger support:
  1 file changed, 422 insertions(+)
 
 diff --git a/drivers/net/phy/marvell10g.c b/drivers/net/phy/marvell10g.c
-index 40439db49601..7ae4744f147b 100644
+index 40439db49601d..7ae4744f147bb 100644
 --- a/drivers/net/phy/marvell10g.c
 +++ b/drivers/net/phy/marvell10g.c
 @@ -28,6 +28,7 @@

patches/linux/6.18.55/0004-net-phy-marvell10g-Support-LEDs-tied-to-a-single-med.patch

@@ -1,7 +1,7 @@
 From 7ad37cfa715d86cc20f29432f106cedd457e4f26 Mon Sep 17 00:00:00 2001
 From: Tobias Waldekranz <tobias@waldekranz.com>
 Date: Tue, 12 Dec 2023 09:51:05 +0100
-Subject: [PATCH 04/78] net: phy: marvell10g: Support LEDs tied to a single
+Subject: [PATCH 04/85] net: phy: marvell10g: Support LEDs tied to a single
  media side
 
 In a combo-port setup, i.e. where both the copper and fiber interface
@@ -18,7 +18,7 @@ versa for the SFP cage.
  1 file changed, 22 insertions(+), 1 deletion(-)
 
 diff --git a/drivers/net/phy/marvell10g.c b/drivers/net/phy/marvell10g.c
-index 7ae4744f147b..5e48355ffcef 100644
+index 7ae4744f147bb..5e48355ffcef4 100644
 --- a/drivers/net/phy/marvell10g.c
 +++ b/drivers/net/phy/marvell10g.c
 @@ -192,6 +192,9 @@ struct mv3310_chip {

patches/linux/6.18.55/0005-net-phy-Do-not-resume-PHY-when-attaching.patch

@@ -1,7 +1,7 @@
 From 5f95f31c7e48087019db7d673ff32a31f4e06619 Mon Sep 17 00:00:00 2001
 From: Tobias Waldekranz <tobias@waldekranz.com>
 Date: Wed, 27 Mar 2024 10:10:19 +0100
-Subject: [PATCH 05/78] net: phy: Do not resume PHY when attaching
+Subject: [PATCH 05/85] net: phy: Do not resume PHY when attaching
 
 The PHY should not start negotiating with its link-partner until
 explicitly instructed to do so.
@@ -19,7 +19,7 @@ administratively down.
  1 file changed, 1 deletion(-)
 
 diff --git a/drivers/net/phy/phy_device.c b/drivers/net/phy/phy_device.c
-index 26b08e3dbd1d..7597308534a0 100644
+index 26b08e3dbd1de..7597308534a0a 100644
 --- a/drivers/net/phy/phy_device.c
 +++ b/drivers/net/phy/phy_device.c
 @@ -1753,7 +1753,6 @@ int phy_attach_direct(struct net_device *dev, struct phy_device *phydev,