linux-stable/drivers/net/dsa/rtl8366.c

535 lines
12 KiB
C
Raw Normal View History

net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
// SPDX-License-Identifier: GPL-2.0
/* Realtek SMI library helpers for the RTL8366x variants
* RTL8366RB and RTL8366S
*
* Copyright (C) 2017 Linus Walleij <linus.walleij@linaro.org>
* Copyright (C) 2009-2010 Gabor Juhos <juhosg@openwrt.org>
* Copyright (C) 2010 Antti Seppälä <a.seppala@gmail.com>
* Copyright (C) 2010 Roman Yeryomin <roman@advem.lv>
* Copyright (C) 2011 Colin Leitner <colin.leitner@googlemail.com>
*/
#include <linux/if_bridge.h>
#include <net/dsa.h>
#include "realtek-smi-core.h"
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
int rtl8366_mc_is_used(struct realtek_smi *smi, int mc_index, int *used)
{
int ret;
int i;
*used = 0;
for (i = 0; i < smi->num_ports; i++) {
int index = 0;
ret = smi->ops->get_mc_index(smi, i, &index);
if (ret)
return ret;
if (mc_index == index) {
*used = 1;
break;
}
}
return 0;
}
EXPORT_SYMBOL_GPL(rtl8366_mc_is_used);
/**
* rtl8366_obtain_mc() - retrieve or allocate a VLAN member configuration
* @smi: the Realtek SMI device instance
* @vid: the VLAN ID to look up or allocate
* @vlanmc: the pointer will be assigned to a pointer to a valid member config
* if successful
* @return: index of a new member config or negative error number
*/
static int rtl8366_obtain_mc(struct realtek_smi *smi, int vid,
struct rtl8366_vlan_mc *vlanmc)
{
struct rtl8366_vlan_4k vlan4k;
int ret;
int i;
/* Try to find an existing member config entry for this VID */
for (i = 0; i < smi->num_vlan_mc; i++) {
ret = smi->ops->get_vlan_mc(smi, i, vlanmc);
if (ret) {
dev_err(smi->dev, "error searching for VLAN MC %d for VID %d\n",
i, vid);
return ret;
}
if (vid == vlanmc->vid)
return i;
}
/* We have no MC entry for this VID, try to find an empty one */
for (i = 0; i < smi->num_vlan_mc; i++) {
ret = smi->ops->get_vlan_mc(smi, i, vlanmc);
if (ret) {
dev_err(smi->dev, "error searching for VLAN MC %d for VID %d\n",
i, vid);
return ret;
}
if (vlanmc->vid == 0 && vlanmc->member == 0) {
/* Update the entry from the 4K table */
ret = smi->ops->get_vlan_4k(smi, vid, &vlan4k);
if (ret) {
dev_err(smi->dev, "error looking for 4K VLAN MC %d for VID %d\n",
i, vid);
return ret;
}
vlanmc->vid = vid;
vlanmc->member = vlan4k.member;
vlanmc->untag = vlan4k.untag;
vlanmc->fid = vlan4k.fid;
ret = smi->ops->set_vlan_mc(smi, i, vlanmc);
if (ret) {
dev_err(smi->dev, "unable to set/update VLAN MC %d for VID %d\n",
i, vid);
return ret;
}
dev_dbg(smi->dev, "created new MC at index %d for VID %d\n",
i, vid);
return i;
}
}
/* MC table is full, try to find an unused entry and replace it */
for (i = 0; i < smi->num_vlan_mc; i++) {
int used;
ret = rtl8366_mc_is_used(smi, i, &used);
if (ret)
return ret;
if (!used) {
/* Update the entry from the 4K table */
ret = smi->ops->get_vlan_4k(smi, vid, &vlan4k);
if (ret)
return ret;
vlanmc->vid = vid;
vlanmc->member = vlan4k.member;
vlanmc->untag = vlan4k.untag;
vlanmc->fid = vlan4k.fid;
ret = smi->ops->set_vlan_mc(smi, i, vlanmc);
if (ret) {
dev_err(smi->dev, "unable to set/update VLAN MC %d for VID %d\n",
i, vid);
return ret;
}
dev_dbg(smi->dev, "recycled MC at index %i for VID %d\n",
i, vid);
return i;
}
}
dev_err(smi->dev, "all VLAN member configurations are in use\n");
return -ENOSPC;
}
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
int rtl8366_set_vlan(struct realtek_smi *smi, int vid, u32 member,
u32 untag, u32 fid)
{
struct rtl8366_vlan_mc vlanmc;
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
struct rtl8366_vlan_4k vlan4k;
int mc;
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
int ret;
if (!smi->ops->is_vlan_valid(smi, vid))
return -EINVAL;
dev_dbg(smi->dev,
"setting VLAN%d 4k members: 0x%02x, untagged: 0x%02x\n",
vid, member, untag);
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
/* Update the 4K table */
ret = smi->ops->get_vlan_4k(smi, vid, &vlan4k);
if (ret)
return ret;
vlan4k.member |= member;
vlan4k.untag |= untag;
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
vlan4k.fid = fid;
ret = smi->ops->set_vlan_4k(smi, &vlan4k);
if (ret)
return ret;
dev_dbg(smi->dev,
"resulting VLAN%d 4k members: 0x%02x, untagged: 0x%02x\n",
vid, vlan4k.member, vlan4k.untag);
/* Find or allocate a member config for this VID */
ret = rtl8366_obtain_mc(smi, vid, &vlanmc);
if (ret < 0)
return ret;
mc = ret;
/* Update the MC entry */
vlanmc.member |= member;
vlanmc.untag |= untag;
vlanmc.fid = fid;
/* Commit updates to the MC entry */
ret = smi->ops->set_vlan_mc(smi, mc, &vlanmc);
if (ret)
dev_err(smi->dev, "failed to commit changes to VLAN MC index %d for VID %d\n",
mc, vid);
else
dev_dbg(smi->dev,
"resulting VLAN%d MC members: 0x%02x, untagged: 0x%02x\n",
vid, vlanmc.member, vlanmc.untag);
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
return ret;
}
EXPORT_SYMBOL_GPL(rtl8366_set_vlan);
int rtl8366_set_pvid(struct realtek_smi *smi, unsigned int port,
unsigned int vid)
{
struct rtl8366_vlan_mc vlanmc;
int mc;
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
int ret;
if (!smi->ops->is_vlan_valid(smi, vid))
return -EINVAL;
/* Find or allocate a member config for this VID */
ret = rtl8366_obtain_mc(smi, vid, &vlanmc);
if (ret < 0)
return ret;
mc = ret;
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
ret = smi->ops->set_mc_index(smi, port, mc);
if (ret) {
dev_err(smi->dev, "set PVID: failed to set MC index %d for port %d\n",
mc, port);
return ret;
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
}
dev_dbg(smi->dev, "set PVID: the PVID for port %d set to %d using existing MC index %d\n",
port, vid, mc);
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
return 0;
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
}
EXPORT_SYMBOL_GPL(rtl8366_set_pvid);
int rtl8366_enable_vlan4k(struct realtek_smi *smi, bool enable)
{
int ret;
/* To enable 4k VLAN, ordinary VLAN must be enabled first,
* but if we disable 4k VLAN it is fine to leave ordinary
* VLAN enabled.
*/
if (enable) {
/* Make sure VLAN is ON */
ret = smi->ops->enable_vlan(smi, true);
if (ret)
return ret;
smi->vlan_enabled = true;
}
ret = smi->ops->enable_vlan4k(smi, enable);
if (ret)
return ret;
smi->vlan4k_enabled = enable;
return 0;
}
EXPORT_SYMBOL_GPL(rtl8366_enable_vlan4k);
int rtl8366_enable_vlan(struct realtek_smi *smi, bool enable)
{
int ret;
ret = smi->ops->enable_vlan(smi, enable);
if (ret)
return ret;
smi->vlan_enabled = enable;
/* If we turn VLAN off, make sure that we turn off
* 4k VLAN as well, if that happened to be on.
*/
if (!enable) {
smi->vlan4k_enabled = false;
ret = smi->ops->enable_vlan4k(smi, false);
}
return ret;
}
EXPORT_SYMBOL_GPL(rtl8366_enable_vlan);
int rtl8366_reset_vlan(struct realtek_smi *smi)
{
struct rtl8366_vlan_mc vlanmc;
int ret;
int i;
rtl8366_enable_vlan(smi, false);
rtl8366_enable_vlan4k(smi, false);
/* Clear the 16 VLAN member configurations */
vlanmc.vid = 0;
vlanmc.priority = 0;
vlanmc.member = 0;
vlanmc.untag = 0;
vlanmc.fid = 0;
for (i = 0; i < smi->num_vlan_mc; i++) {
ret = smi->ops->set_vlan_mc(smi, i, &vlanmc);
if (ret)
return ret;
}
return 0;
}
EXPORT_SYMBOL_GPL(rtl8366_reset_vlan);
int rtl8366_init_vlan(struct realtek_smi *smi)
{
int port;
int ret;
ret = rtl8366_reset_vlan(smi);
if (ret)
return ret;
/* Loop over the available ports, for each port, associate
* it with the VLAN (port+1)
*/
for (port = 0; port < smi->num_ports; port++) {
u32 mask;
if (port == smi->cpu_port)
/* For the CPU port, make all ports members of this
* VLAN.
*/
mask = GENMASK((int)smi->num_ports - 1, 0);
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
else
/* For all other ports, enable itself plus the
* CPU port.
*/
mask = BIT(port) | BIT(smi->cpu_port);
/* For each port, set the port as member of VLAN (port+1)
* and untagged, except for the CPU port: the CPU port (5) is
* member of VLAN 6 and so are ALL the other ports as well.
* Use filter 0 (no filter).
*/
dev_info(smi->dev, "VLAN%d port mask for port %d, %08x\n",
(port + 1), port, mask);
ret = rtl8366_set_vlan(smi, (port + 1), mask, mask, 0);
if (ret)
return ret;
dev_info(smi->dev, "VLAN%d port %d, PVID set to %d\n",
(port + 1), port, (port + 1));
ret = rtl8366_set_pvid(smi, port, (port + 1));
if (ret)
return ret;
}
return rtl8366_enable_vlan(smi, true);
}
EXPORT_SYMBOL_GPL(rtl8366_init_vlan);
int rtl8366_vlan_filtering(struct dsa_switch *ds, int port, bool vlan_filtering,
struct netlink_ext_ack *extack)
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
{
struct realtek_smi *smi = ds->priv;
struct rtl8366_vlan_4k vlan4k;
int ret;
/* Use VLAN nr port + 1 since VLAN0 is not valid */
net: switchdev: remove the transaction structure from port attributes Since the introduction of the switchdev API, port attributes were transmitted to drivers for offloading using a two-step transactional model, with a prepare phase that was supposed to catch all errors, and a commit phase that was supposed to never fail. Some classes of failures can never be avoided, like hardware access, or memory allocation. In the latter case, merely attempting to move the memory allocation to the preparation phase makes it impossible to avoid memory leaks, since commit 91cf8eceffc1 ("switchdev: Remove unused transaction item queue") which has removed the unused mechanism of passing on the allocated memory between one phase and another. It is time we admit that separating the preparation from the commit phase is something that is best left for the driver to decide, and not something that should be baked into the API, especially since there are no switchdev callers that depend on this. This patch removes the struct switchdev_trans member from switchdev port attribute notifier structures, and converts drivers to not look at this member. In part, this patch contains a revert of my previous commit 2e554a7a5d8a ("net: dsa: propagate switchdev vlan_filtering prepare phase to drivers"). For the most part, the conversion was trivial except for: - Rocker's world implementation based on Broadcom OF-DPA had an odd implementation of ofdpa_port_attr_bridge_flags_set. The conversion was done mechanically, by pasting the implementation twice, then only keeping the code that would get executed during prepare phase on top, then only keeping the code that gets executed during the commit phase on bottom, then simplifying the resulting code until this was obtained. - DSA's offloading of STP state, bridge flags, VLAN filtering and multicast router could be converted right away. But the ageing time could not, so a shim was introduced and this was left for a further commit. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Acked-by: Linus Walleij <linus.walleij@linaro.org> Acked-by: Jiri Pirko <jiri@nvidia.com> Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de> # hellcreek Reviewed-by: Linus Walleij <linus.walleij@linaro.org> # RTL8366RB Reviewed-by: Ido Schimmel <idosch@nvidia.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 00:01:50 +00:00
if (!smi->ops->is_vlan_valid(smi, port + 1))
return -EINVAL;
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
dev_info(smi->dev, "%s filtering on port %d\n",
vlan_filtering ? "enable" : "disable",
port);
/* TODO:
* The hardware support filter ID (FID) 0..7, I have no clue how to
* support this in the driver when the callback only says on/off.
*/
ret = smi->ops->get_vlan_4k(smi, port + 1, &vlan4k);
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
if (ret)
return ret;
/* Just set the filter to FID 1 for now then */
ret = rtl8366_set_vlan(smi, port + 1,
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
vlan4k.member,
vlan4k.untag,
1);
if (ret)
return ret;
return 0;
}
EXPORT_SYMBOL_GPL(rtl8366_vlan_filtering);
net: dsa: remove the transactional logic from VLAN objects It should be the driver's business to logically separate its VLAN offloading into a preparation and a commit phase, and some drivers don't need / can't do this. So remove the transactional shim from DSA and let drivers propagate errors directly from the .port_vlan_add callback. It would appear that the code has worse error handling now than it had before. DSA is the only in-kernel user of switchdev that offloads one switchdev object to more than one port: for every VLAN object offloaded to a user port, that VLAN is also offloaded to the CPU port. So the "prepare for user port -> check for errors -> prepare for CPU port -> check for errors -> commit for user port -> commit for CPU port" sequence appears to make more sense than the one we are using now: "offload to user port -> check for errors -> offload to CPU port -> check for errors", but it is really a compromise. In the new way, we can catch errors from the commit phase that we previously had to ignore. But we have our hands tied and cannot do any rollback now: if we add a VLAN on the CPU port and it fails, we can't do the rollback by simply deleting it from the user port, because the switchdev API is not so nice with us: it could have simply been there already, even with the same flags. So we don't even attempt to rollback anything on addition error, just leave whatever VLANs managed to get offloaded right where they are. This should not be a problem at all in practice. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Acked-by: Linus Walleij <linus.walleij@linaro.org> Acked-by: Jiri Pirko <jiri@nvidia.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 00:01:53 +00:00
int rtl8366_vlan_add(struct dsa_switch *ds, int port,
const struct switchdev_obj_port_vlan *vlan,
struct netlink_ext_ack *extack)
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
{
net: dsa: remove the transactional logic from VLAN objects It should be the driver's business to logically separate its VLAN offloading into a preparation and a commit phase, and some drivers don't need / can't do this. So remove the transactional shim from DSA and let drivers propagate errors directly from the .port_vlan_add callback. It would appear that the code has worse error handling now than it had before. DSA is the only in-kernel user of switchdev that offloads one switchdev object to more than one port: for every VLAN object offloaded to a user port, that VLAN is also offloaded to the CPU port. So the "prepare for user port -> check for errors -> prepare for CPU port -> check for errors -> commit for user port -> commit for CPU port" sequence appears to make more sense than the one we are using now: "offload to user port -> check for errors -> offload to CPU port -> check for errors", but it is really a compromise. In the new way, we can catch errors from the commit phase that we previously had to ignore. But we have our hands tied and cannot do any rollback now: if we add a VLAN on the CPU port and it fails, we can't do the rollback by simply deleting it from the user port, because the switchdev API is not so nice with us: it could have simply been there already, even with the same flags. So we don't even attempt to rollback anything on addition error, just leave whatever VLANs managed to get offloaded right where they are. This should not be a problem at all in practice. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Acked-by: Linus Walleij <linus.walleij@linaro.org> Acked-by: Jiri Pirko <jiri@nvidia.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 00:01:53 +00:00
bool untagged = !!(vlan->flags & BRIDGE_VLAN_INFO_UNTAGGED);
bool pvid = !!(vlan->flags & BRIDGE_VLAN_INFO_PVID);
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
struct realtek_smi *smi = ds->priv;
net: dsa: remove the transactional logic from VLAN objects It should be the driver's business to logically separate its VLAN offloading into a preparation and a commit phase, and some drivers don't need / can't do this. So remove the transactional shim from DSA and let drivers propagate errors directly from the .port_vlan_add callback. It would appear that the code has worse error handling now than it had before. DSA is the only in-kernel user of switchdev that offloads one switchdev object to more than one port: for every VLAN object offloaded to a user port, that VLAN is also offloaded to the CPU port. So the "prepare for user port -> check for errors -> prepare for CPU port -> check for errors -> commit for user port -> commit for CPU port" sequence appears to make more sense than the one we are using now: "offload to user port -> check for errors -> offload to CPU port -> check for errors", but it is really a compromise. In the new way, we can catch errors from the commit phase that we previously had to ignore. But we have our hands tied and cannot do any rollback now: if we add a VLAN on the CPU port and it fails, we can't do the rollback by simply deleting it from the user port, because the switchdev API is not so nice with us: it could have simply been there already, even with the same flags. So we don't even attempt to rollback anything on addition error, just leave whatever VLANs managed to get offloaded right where they are. This should not be a problem at all in practice. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Acked-by: Linus Walleij <linus.walleij@linaro.org> Acked-by: Jiri Pirko <jiri@nvidia.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 00:01:53 +00:00
u32 member = 0;
u32 untag = 0;
int ret;
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
if (!smi->ops->is_vlan_valid(smi, vlan->vid)) {
NL_SET_ERR_MSG_MOD(extack, "VLAN ID not valid");
net: switchdev: remove vid_begin -> vid_end range from VLAN objects The call path of a switchdev VLAN addition to the bridge looks something like this today: nbp_vlan_init | __br_vlan_set_default_pvid | | | | | br_afspec | | | | | | | v | | | br_process_vlan_info | | | | | | | v | | | br_vlan_info | | | / \ / | | / \ / | | / \ / | | / \ / v v v v v nbp_vlan_add br_vlan_add ------+ | ^ ^ | | | / | | | | / / / | \ br_vlan_get_master/ / v \ ^ / / br_vlan_add_existing \ | / / | \ | / / / \ | / / / \ | / / / \ | / / / v | | v / __vlan_add / / | / / | / v | / __vlan_vid_add | / \ | / v v v br_switchdev_port_vlan_add The ranges UAPI was introduced to the bridge in commit bdced7ef7838 ("bridge: support for multiple vlans and vlan ranges in setlink and dellink requests") (Jan 10 2015). But the VLAN ranges (parsed in br_afspec) have always been passed one by one, through struct bridge_vlan_info tmp_vinfo, to br_vlan_info. So the range never went too far in depth. Then Scott Feldman introduced the switchdev_port_bridge_setlink function in commit 47f8328bb1a4 ("switchdev: add new switchdev bridge setlink"). That marked the introduction of the SWITCHDEV_OBJ_PORT_VLAN, which made full use of the range. But switchdev_port_bridge_setlink was called like this: br_setlink -> br_afspec -> switchdev_port_bridge_setlink Basically, the switchdev and the bridge code were not tightly integrated. Then commit 41c498b9359e ("bridge: restore br_setlink back to original") came, and switchdev drivers were required to implement .ndo_bridge_setlink = switchdev_port_bridge_setlink for a while. In the meantime, commits such as 0944d6b5a2fa ("bridge: try switchdev op first in __vlan_vid_add/del") finally made switchdev penetrate the br_vlan_info() barrier and start to develop the call path we have today. But remember, br_vlan_info() still receives VLANs one by one. Then Arkadi Sharshevsky refactored the switchdev API in 2017 in commit 29ab586c3d83 ("net: switchdev: Remove bridge bypass support from switchdev") so that drivers would not implement .ndo_bridge_setlink any longer. The switchdev_port_bridge_setlink also got deleted. This refactoring removed the parallel bridge_setlink implementation from switchdev, and left the only switchdev VLAN objects to be the ones offloaded from __vlan_vid_add (basically RX filtering) and __vlan_add (the latter coming from commit 9c86ce2c1ae3 ("net: bridge: Notify about bridge VLANs")). That is to say, today the switchdev VLAN object ranges are not used in the kernel. Refactoring the above call path is a bit complicated, when the bridge VLAN call path is already a bit complicated. Let's go off and finish the job of commit 29ab586c3d83 by deleting the bogus iteration through the VLAN ranges from the drivers. Some aspects of this feature never made too much sense in the first place. For example, what is a range of VLANs all having the BRIDGE_VLAN_INFO_PVID flag supposed to mean, when a port can obviously have a single pvid? This particular configuration _is_ denied as of commit 6623c60dc28e ("bridge: vlan: enforce no pvid flag in vlan ranges"), but from an API perspective, the driver still has to play pretend, and only offload the vlan->vid_end as pvid. And the addition of a switchdev VLAN object can modify the flags of another, completely unrelated, switchdev VLAN object! (a VLAN that is PVID will invalidate the PVID flag from whatever other VLAN had previously been offloaded with switchdev and had that flag. Yet switchdev never notifies about that change, drivers are supposed to guess). Nonetheless, having a VLAN range in the API makes error handling look scarier than it really is - unwinding on errors and all of that. When in reality, no one really calls this API with more than one VLAN. It is all unnecessary complexity. And despite appearing pretentious (two-phase transactional model and all), the switchdev API is really sloppy because the VLAN addition and removal operations are not paired with one another (you can add a VLAN 100 times and delete it just once). The bridge notifies through switchdev of a VLAN addition not only when the flags of an existing VLAN change, but also when nothing changes. There are switchdev drivers out there who don't like adding a VLAN that has already been added, and those checks don't really belong at driver level. But the fact that the API contains ranges is yet another factor that prevents this from being addressed in the future. Of the existing switchdev pieces of hardware, it appears that only Mellanox Spectrum supports offloading more than one VLAN at a time, through mlxsw_sp_port_vlan_set. I have kept that code internal to the driver, because there is some more bookkeeping that makes use of it, but I deleted it from the switchdev API. But since the switchdev support for ranges has already been de facto deleted by a Mellanox employee and nobody noticed for 4 years, I'm going to assume it's not a biggie. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Ido Schimmel <idosch@nvidia.com> # switchdev and mlxsw Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de> # hellcreek Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 00:01:46 +00:00
return -EINVAL;
}
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
/* Enable VLAN in the hardware
* FIXME: what's with this 4k business?
* Just rtl8366_enable_vlan() seems inconclusive.
*/
net: dsa: remove the transactional logic from VLAN objects It should be the driver's business to logically separate its VLAN offloading into a preparation and a commit phase, and some drivers don't need / can't do this. So remove the transactional shim from DSA and let drivers propagate errors directly from the .port_vlan_add callback. It would appear that the code has worse error handling now than it had before. DSA is the only in-kernel user of switchdev that offloads one switchdev object to more than one port: for every VLAN object offloaded to a user port, that VLAN is also offloaded to the CPU port. So the "prepare for user port -> check for errors -> prepare for CPU port -> check for errors -> commit for user port -> commit for CPU port" sequence appears to make more sense than the one we are using now: "offload to user port -> check for errors -> offload to CPU port -> check for errors", but it is really a compromise. In the new way, we can catch errors from the commit phase that we previously had to ignore. But we have our hands tied and cannot do any rollback now: if we add a VLAN on the CPU port and it fails, we can't do the rollback by simply deleting it from the user port, because the switchdev API is not so nice with us: it could have simply been there already, even with the same flags. So we don't even attempt to rollback anything on addition error, just leave whatever VLANs managed to get offloaded right where they are. This should not be a problem at all in practice. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Acked-by: Linus Walleij <linus.walleij@linaro.org> Acked-by: Jiri Pirko <jiri@nvidia.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 00:01:53 +00:00
ret = rtl8366_enable_vlan4k(smi, true);
if (ret) {
NL_SET_ERR_MSG_MOD(extack, "Failed to enable VLAN 4K");
net: dsa: remove the transactional logic from VLAN objects It should be the driver's business to logically separate its VLAN offloading into a preparation and a commit phase, and some drivers don't need / can't do this. So remove the transactional shim from DSA and let drivers propagate errors directly from the .port_vlan_add callback. It would appear that the code has worse error handling now than it had before. DSA is the only in-kernel user of switchdev that offloads one switchdev object to more than one port: for every VLAN object offloaded to a user port, that VLAN is also offloaded to the CPU port. So the "prepare for user port -> check for errors -> prepare for CPU port -> check for errors -> commit for user port -> commit for CPU port" sequence appears to make more sense than the one we are using now: "offload to user port -> check for errors -> offload to CPU port -> check for errors", but it is really a compromise. In the new way, we can catch errors from the commit phase that we previously had to ignore. But we have our hands tied and cannot do any rollback now: if we add a VLAN on the CPU port and it fails, we can't do the rollback by simply deleting it from the user port, because the switchdev API is not so nice with us: it could have simply been there already, even with the same flags. So we don't even attempt to rollback anything on addition error, just leave whatever VLANs managed to get offloaded right where they are. This should not be a problem at all in practice. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Acked-by: Linus Walleij <linus.walleij@linaro.org> Acked-by: Jiri Pirko <jiri@nvidia.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 00:01:53 +00:00
return ret;
}
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
dev_info(smi->dev, "add VLAN %d on port %d, %s, %s\n",
net: switchdev: remove vid_begin -> vid_end range from VLAN objects The call path of a switchdev VLAN addition to the bridge looks something like this today: nbp_vlan_init | __br_vlan_set_default_pvid | | | | | br_afspec | | | | | | | v | | | br_process_vlan_info | | | | | | | v | | | br_vlan_info | | | / \ / | | / \ / | | / \ / | | / \ / v v v v v nbp_vlan_add br_vlan_add ------+ | ^ ^ | | | / | | | | / / / | \ br_vlan_get_master/ / v \ ^ / / br_vlan_add_existing \ | / / | \ | / / / \ | / / / \ | / / / \ | / / / v | | v / __vlan_add / / | / / | / v | / __vlan_vid_add | / \ | / v v v br_switchdev_port_vlan_add The ranges UAPI was introduced to the bridge in commit bdced7ef7838 ("bridge: support for multiple vlans and vlan ranges in setlink and dellink requests") (Jan 10 2015). But the VLAN ranges (parsed in br_afspec) have always been passed one by one, through struct bridge_vlan_info tmp_vinfo, to br_vlan_info. So the range never went too far in depth. Then Scott Feldman introduced the switchdev_port_bridge_setlink function in commit 47f8328bb1a4 ("switchdev: add new switchdev bridge setlink"). That marked the introduction of the SWITCHDEV_OBJ_PORT_VLAN, which made full use of the range. But switchdev_port_bridge_setlink was called like this: br_setlink -> br_afspec -> switchdev_port_bridge_setlink Basically, the switchdev and the bridge code were not tightly integrated. Then commit 41c498b9359e ("bridge: restore br_setlink back to original") came, and switchdev drivers were required to implement .ndo_bridge_setlink = switchdev_port_bridge_setlink for a while. In the meantime, commits such as 0944d6b5a2fa ("bridge: try switchdev op first in __vlan_vid_add/del") finally made switchdev penetrate the br_vlan_info() barrier and start to develop the call path we have today. But remember, br_vlan_info() still receives VLANs one by one. Then Arkadi Sharshevsky refactored the switchdev API in 2017 in commit 29ab586c3d83 ("net: switchdev: Remove bridge bypass support from switchdev") so that drivers would not implement .ndo_bridge_setlink any longer. The switchdev_port_bridge_setlink also got deleted. This refactoring removed the parallel bridge_setlink implementation from switchdev, and left the only switchdev VLAN objects to be the ones offloaded from __vlan_vid_add (basically RX filtering) and __vlan_add (the latter coming from commit 9c86ce2c1ae3 ("net: bridge: Notify about bridge VLANs")). That is to say, today the switchdev VLAN object ranges are not used in the kernel. Refactoring the above call path is a bit complicated, when the bridge VLAN call path is already a bit complicated. Let's go off and finish the job of commit 29ab586c3d83 by deleting the bogus iteration through the VLAN ranges from the drivers. Some aspects of this feature never made too much sense in the first place. For example, what is a range of VLANs all having the BRIDGE_VLAN_INFO_PVID flag supposed to mean, when a port can obviously have a single pvid? This particular configuration _is_ denied as of commit 6623c60dc28e ("bridge: vlan: enforce no pvid flag in vlan ranges"), but from an API perspective, the driver still has to play pretend, and only offload the vlan->vid_end as pvid. And the addition of a switchdev VLAN object can modify the flags of another, completely unrelated, switchdev VLAN object! (a VLAN that is PVID will invalidate the PVID flag from whatever other VLAN had previously been offloaded with switchdev and had that flag. Yet switchdev never notifies about that change, drivers are supposed to guess). Nonetheless, having a VLAN range in the API makes error handling look scarier than it really is - unwinding on errors and all of that. When in reality, no one really calls this API with more than one VLAN. It is all unnecessary complexity. And despite appearing pretentious (two-phase transactional model and all), the switchdev API is really sloppy because the VLAN addition and removal operations are not paired with one another (you can add a VLAN 100 times and delete it just once). The bridge notifies through switchdev of a VLAN addition not only when the flags of an existing VLAN change, but also when nothing changes. There are switchdev drivers out there who don't like adding a VLAN that has already been added, and those checks don't really belong at driver level. But the fact that the API contains ranges is yet another factor that prevents this from being addressed in the future. Of the existing switchdev pieces of hardware, it appears that only Mellanox Spectrum supports offloading more than one VLAN at a time, through mlxsw_sp_port_vlan_set. I have kept that code internal to the driver, because there is some more bookkeeping that makes use of it, but I deleted it from the switchdev API. But since the switchdev support for ranges has already been de facto deleted by a Mellanox employee and nobody noticed for 4 years, I'm going to assume it's not a biggie. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Ido Schimmel <idosch@nvidia.com> # switchdev and mlxsw Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de> # hellcreek Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 00:01:46 +00:00
vlan->vid, port, untagged ? "untagged" : "tagged",
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
pvid ? " PVID" : "no PVID");
if (dsa_is_dsa_port(ds, port) || dsa_is_cpu_port(ds, port))
dev_err(smi->dev, "port is DSA or CPU port\n");
net: switchdev: remove vid_begin -> vid_end range from VLAN objects The call path of a switchdev VLAN addition to the bridge looks something like this today: nbp_vlan_init | __br_vlan_set_default_pvid | | | | | br_afspec | | | | | | | v | | | br_process_vlan_info | | | | | | | v | | | br_vlan_info | | | / \ / | | / \ / | | / \ / | | / \ / v v v v v nbp_vlan_add br_vlan_add ------+ | ^ ^ | | | / | | | | / / / | \ br_vlan_get_master/ / v \ ^ / / br_vlan_add_existing \ | / / | \ | / / / \ | / / / \ | / / / \ | / / / v | | v / __vlan_add / / | / / | / v | / __vlan_vid_add | / \ | / v v v br_switchdev_port_vlan_add The ranges UAPI was introduced to the bridge in commit bdced7ef7838 ("bridge: support for multiple vlans and vlan ranges in setlink and dellink requests") (Jan 10 2015). But the VLAN ranges (parsed in br_afspec) have always been passed one by one, through struct bridge_vlan_info tmp_vinfo, to br_vlan_info. So the range never went too far in depth. Then Scott Feldman introduced the switchdev_port_bridge_setlink function in commit 47f8328bb1a4 ("switchdev: add new switchdev bridge setlink"). That marked the introduction of the SWITCHDEV_OBJ_PORT_VLAN, which made full use of the range. But switchdev_port_bridge_setlink was called like this: br_setlink -> br_afspec -> switchdev_port_bridge_setlink Basically, the switchdev and the bridge code were not tightly integrated. Then commit 41c498b9359e ("bridge: restore br_setlink back to original") came, and switchdev drivers were required to implement .ndo_bridge_setlink = switchdev_port_bridge_setlink for a while. In the meantime, commits such as 0944d6b5a2fa ("bridge: try switchdev op first in __vlan_vid_add/del") finally made switchdev penetrate the br_vlan_info() barrier and start to develop the call path we have today. But remember, br_vlan_info() still receives VLANs one by one. Then Arkadi Sharshevsky refactored the switchdev API in 2017 in commit 29ab586c3d83 ("net: switchdev: Remove bridge bypass support from switchdev") so that drivers would not implement .ndo_bridge_setlink any longer. The switchdev_port_bridge_setlink also got deleted. This refactoring removed the parallel bridge_setlink implementation from switchdev, and left the only switchdev VLAN objects to be the ones offloaded from __vlan_vid_add (basically RX filtering) and __vlan_add (the latter coming from commit 9c86ce2c1ae3 ("net: bridge: Notify about bridge VLANs")). That is to say, today the switchdev VLAN object ranges are not used in the kernel. Refactoring the above call path is a bit complicated, when the bridge VLAN call path is already a bit complicated. Let's go off and finish the job of commit 29ab586c3d83 by deleting the bogus iteration through the VLAN ranges from the drivers. Some aspects of this feature never made too much sense in the first place. For example, what is a range of VLANs all having the BRIDGE_VLAN_INFO_PVID flag supposed to mean, when a port can obviously have a single pvid? This particular configuration _is_ denied as of commit 6623c60dc28e ("bridge: vlan: enforce no pvid flag in vlan ranges"), but from an API perspective, the driver still has to play pretend, and only offload the vlan->vid_end as pvid. And the addition of a switchdev VLAN object can modify the flags of another, completely unrelated, switchdev VLAN object! (a VLAN that is PVID will invalidate the PVID flag from whatever other VLAN had previously been offloaded with switchdev and had that flag. Yet switchdev never notifies about that change, drivers are supposed to guess). Nonetheless, having a VLAN range in the API makes error handling look scarier than it really is - unwinding on errors and all of that. When in reality, no one really calls this API with more than one VLAN. It is all unnecessary complexity. And despite appearing pretentious (two-phase transactional model and all), the switchdev API is really sloppy because the VLAN addition and removal operations are not paired with one another (you can add a VLAN 100 times and delete it just once). The bridge notifies through switchdev of a VLAN addition not only when the flags of an existing VLAN change, but also when nothing changes. There are switchdev drivers out there who don't like adding a VLAN that has already been added, and those checks don't really belong at driver level. But the fact that the API contains ranges is yet another factor that prevents this from being addressed in the future. Of the existing switchdev pieces of hardware, it appears that only Mellanox Spectrum supports offloading more than one VLAN at a time, through mlxsw_sp_port_vlan_set. I have kept that code internal to the driver, because there is some more bookkeeping that makes use of it, but I deleted it from the switchdev API. But since the switchdev support for ranges has already been de facto deleted by a Mellanox employee and nobody noticed for 4 years, I'm going to assume it's not a biggie. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Ido Schimmel <idosch@nvidia.com> # switchdev and mlxsw Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de> # hellcreek Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 00:01:46 +00:00
member |= BIT(port);
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
net: switchdev: remove vid_begin -> vid_end range from VLAN objects The call path of a switchdev VLAN addition to the bridge looks something like this today: nbp_vlan_init | __br_vlan_set_default_pvid | | | | | br_afspec | | | | | | | v | | | br_process_vlan_info | | | | | | | v | | | br_vlan_info | | | / \ / | | / \ / | | / \ / | | / \ / v v v v v nbp_vlan_add br_vlan_add ------+ | ^ ^ | | | / | | | | / / / | \ br_vlan_get_master/ / v \ ^ / / br_vlan_add_existing \ | / / | \ | / / / \ | / / / \ | / / / \ | / / / v | | v / __vlan_add / / | / / | / v | / __vlan_vid_add | / \ | / v v v br_switchdev_port_vlan_add The ranges UAPI was introduced to the bridge in commit bdced7ef7838 ("bridge: support for multiple vlans and vlan ranges in setlink and dellink requests") (Jan 10 2015). But the VLAN ranges (parsed in br_afspec) have always been passed one by one, through struct bridge_vlan_info tmp_vinfo, to br_vlan_info. So the range never went too far in depth. Then Scott Feldman introduced the switchdev_port_bridge_setlink function in commit 47f8328bb1a4 ("switchdev: add new switchdev bridge setlink"). That marked the introduction of the SWITCHDEV_OBJ_PORT_VLAN, which made full use of the range. But switchdev_port_bridge_setlink was called like this: br_setlink -> br_afspec -> switchdev_port_bridge_setlink Basically, the switchdev and the bridge code were not tightly integrated. Then commit 41c498b9359e ("bridge: restore br_setlink back to original") came, and switchdev drivers were required to implement .ndo_bridge_setlink = switchdev_port_bridge_setlink for a while. In the meantime, commits such as 0944d6b5a2fa ("bridge: try switchdev op first in __vlan_vid_add/del") finally made switchdev penetrate the br_vlan_info() barrier and start to develop the call path we have today. But remember, br_vlan_info() still receives VLANs one by one. Then Arkadi Sharshevsky refactored the switchdev API in 2017 in commit 29ab586c3d83 ("net: switchdev: Remove bridge bypass support from switchdev") so that drivers would not implement .ndo_bridge_setlink any longer. The switchdev_port_bridge_setlink also got deleted. This refactoring removed the parallel bridge_setlink implementation from switchdev, and left the only switchdev VLAN objects to be the ones offloaded from __vlan_vid_add (basically RX filtering) and __vlan_add (the latter coming from commit 9c86ce2c1ae3 ("net: bridge: Notify about bridge VLANs")). That is to say, today the switchdev VLAN object ranges are not used in the kernel. Refactoring the above call path is a bit complicated, when the bridge VLAN call path is already a bit complicated. Let's go off and finish the job of commit 29ab586c3d83 by deleting the bogus iteration through the VLAN ranges from the drivers. Some aspects of this feature never made too much sense in the first place. For example, what is a range of VLANs all having the BRIDGE_VLAN_INFO_PVID flag supposed to mean, when a port can obviously have a single pvid? This particular configuration _is_ denied as of commit 6623c60dc28e ("bridge: vlan: enforce no pvid flag in vlan ranges"), but from an API perspective, the driver still has to play pretend, and only offload the vlan->vid_end as pvid. And the addition of a switchdev VLAN object can modify the flags of another, completely unrelated, switchdev VLAN object! (a VLAN that is PVID will invalidate the PVID flag from whatever other VLAN had previously been offloaded with switchdev and had that flag. Yet switchdev never notifies about that change, drivers are supposed to guess). Nonetheless, having a VLAN range in the API makes error handling look scarier than it really is - unwinding on errors and all of that. When in reality, no one really calls this API with more than one VLAN. It is all unnecessary complexity. And despite appearing pretentious (two-phase transactional model and all), the switchdev API is really sloppy because the VLAN addition and removal operations are not paired with one another (you can add a VLAN 100 times and delete it just once). The bridge notifies through switchdev of a VLAN addition not only when the flags of an existing VLAN change, but also when nothing changes. There are switchdev drivers out there who don't like adding a VLAN that has already been added, and those checks don't really belong at driver level. But the fact that the API contains ranges is yet another factor that prevents this from being addressed in the future. Of the existing switchdev pieces of hardware, it appears that only Mellanox Spectrum supports offloading more than one VLAN at a time, through mlxsw_sp_port_vlan_set. I have kept that code internal to the driver, because there is some more bookkeeping that makes use of it, but I deleted it from the switchdev API. But since the switchdev support for ranges has already been de facto deleted by a Mellanox employee and nobody noticed for 4 years, I'm going to assume it's not a biggie. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Ido Schimmel <idosch@nvidia.com> # switchdev and mlxsw Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de> # hellcreek Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 00:01:46 +00:00
if (untagged)
untag |= BIT(port);
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
net: switchdev: remove vid_begin -> vid_end range from VLAN objects The call path of a switchdev VLAN addition to the bridge looks something like this today: nbp_vlan_init | __br_vlan_set_default_pvid | | | | | br_afspec | | | | | | | v | | | br_process_vlan_info | | | | | | | v | | | br_vlan_info | | | / \ / | | / \ / | | / \ / | | / \ / v v v v v nbp_vlan_add br_vlan_add ------+ | ^ ^ | | | / | | | | / / / | \ br_vlan_get_master/ / v \ ^ / / br_vlan_add_existing \ | / / | \ | / / / \ | / / / \ | / / / \ | / / / v | | v / __vlan_add / / | / / | / v | / __vlan_vid_add | / \ | / v v v br_switchdev_port_vlan_add The ranges UAPI was introduced to the bridge in commit bdced7ef7838 ("bridge: support for multiple vlans and vlan ranges in setlink and dellink requests") (Jan 10 2015). But the VLAN ranges (parsed in br_afspec) have always been passed one by one, through struct bridge_vlan_info tmp_vinfo, to br_vlan_info. So the range never went too far in depth. Then Scott Feldman introduced the switchdev_port_bridge_setlink function in commit 47f8328bb1a4 ("switchdev: add new switchdev bridge setlink"). That marked the introduction of the SWITCHDEV_OBJ_PORT_VLAN, which made full use of the range. But switchdev_port_bridge_setlink was called like this: br_setlink -> br_afspec -> switchdev_port_bridge_setlink Basically, the switchdev and the bridge code were not tightly integrated. Then commit 41c498b9359e ("bridge: restore br_setlink back to original") came, and switchdev drivers were required to implement .ndo_bridge_setlink = switchdev_port_bridge_setlink for a while. In the meantime, commits such as 0944d6b5a2fa ("bridge: try switchdev op first in __vlan_vid_add/del") finally made switchdev penetrate the br_vlan_info() barrier and start to develop the call path we have today. But remember, br_vlan_info() still receives VLANs one by one. Then Arkadi Sharshevsky refactored the switchdev API in 2017 in commit 29ab586c3d83 ("net: switchdev: Remove bridge bypass support from switchdev") so that drivers would not implement .ndo_bridge_setlink any longer. The switchdev_port_bridge_setlink also got deleted. This refactoring removed the parallel bridge_setlink implementation from switchdev, and left the only switchdev VLAN objects to be the ones offloaded from __vlan_vid_add (basically RX filtering) and __vlan_add (the latter coming from commit 9c86ce2c1ae3 ("net: bridge: Notify about bridge VLANs")). That is to say, today the switchdev VLAN object ranges are not used in the kernel. Refactoring the above call path is a bit complicated, when the bridge VLAN call path is already a bit complicated. Let's go off and finish the job of commit 29ab586c3d83 by deleting the bogus iteration through the VLAN ranges from the drivers. Some aspects of this feature never made too much sense in the first place. For example, what is a range of VLANs all having the BRIDGE_VLAN_INFO_PVID flag supposed to mean, when a port can obviously have a single pvid? This particular configuration _is_ denied as of commit 6623c60dc28e ("bridge: vlan: enforce no pvid flag in vlan ranges"), but from an API perspective, the driver still has to play pretend, and only offload the vlan->vid_end as pvid. And the addition of a switchdev VLAN object can modify the flags of another, completely unrelated, switchdev VLAN object! (a VLAN that is PVID will invalidate the PVID flag from whatever other VLAN had previously been offloaded with switchdev and had that flag. Yet switchdev never notifies about that change, drivers are supposed to guess). Nonetheless, having a VLAN range in the API makes error handling look scarier than it really is - unwinding on errors and all of that. When in reality, no one really calls this API with more than one VLAN. It is all unnecessary complexity. And despite appearing pretentious (two-phase transactional model and all), the switchdev API is really sloppy because the VLAN addition and removal operations are not paired with one another (you can add a VLAN 100 times and delete it just once). The bridge notifies through switchdev of a VLAN addition not only when the flags of an existing VLAN change, but also when nothing changes. There are switchdev drivers out there who don't like adding a VLAN that has already been added, and those checks don't really belong at driver level. But the fact that the API contains ranges is yet another factor that prevents this from being addressed in the future. Of the existing switchdev pieces of hardware, it appears that only Mellanox Spectrum supports offloading more than one VLAN at a time, through mlxsw_sp_port_vlan_set. I have kept that code internal to the driver, because there is some more bookkeeping that makes use of it, but I deleted it from the switchdev API. But since the switchdev support for ranges has already been de facto deleted by a Mellanox employee and nobody noticed for 4 years, I'm going to assume it's not a biggie. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Ido Schimmel <idosch@nvidia.com> # switchdev and mlxsw Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de> # hellcreek Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 00:01:46 +00:00
ret = rtl8366_set_vlan(smi, vlan->vid, member, untag, 0);
net: dsa: remove the transactional logic from VLAN objects It should be the driver's business to logically separate its VLAN offloading into a preparation and a commit phase, and some drivers don't need / can't do this. So remove the transactional shim from DSA and let drivers propagate errors directly from the .port_vlan_add callback. It would appear that the code has worse error handling now than it had before. DSA is the only in-kernel user of switchdev that offloads one switchdev object to more than one port: for every VLAN object offloaded to a user port, that VLAN is also offloaded to the CPU port. So the "prepare for user port -> check for errors -> prepare for CPU port -> check for errors -> commit for user port -> commit for CPU port" sequence appears to make more sense than the one we are using now: "offload to user port -> check for errors -> offload to CPU port -> check for errors", but it is really a compromise. In the new way, we can catch errors from the commit phase that we previously had to ignore. But we have our hands tied and cannot do any rollback now: if we add a VLAN on the CPU port and it fails, we can't do the rollback by simply deleting it from the user port, because the switchdev API is not so nice with us: it could have simply been there already, even with the same flags. So we don't even attempt to rollback anything on addition error, just leave whatever VLANs managed to get offloaded right where they are. This should not be a problem at all in practice. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Acked-by: Linus Walleij <linus.walleij@linaro.org> Acked-by: Jiri Pirko <jiri@nvidia.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 00:01:53 +00:00
if (ret) {
net: switchdev: remove vid_begin -> vid_end range from VLAN objects The call path of a switchdev VLAN addition to the bridge looks something like this today: nbp_vlan_init | __br_vlan_set_default_pvid | | | | | br_afspec | | | | | | | v | | | br_process_vlan_info | | | | | | | v | | | br_vlan_info | | | / \ / | | / \ / | | / \ / | | / \ / v v v v v nbp_vlan_add br_vlan_add ------+ | ^ ^ | | | / | | | | / / / | \ br_vlan_get_master/ / v \ ^ / / br_vlan_add_existing \ | / / | \ | / / / \ | / / / \ | / / / \ | / / / v | | v / __vlan_add / / | / / | / v | / __vlan_vid_add | / \ | / v v v br_switchdev_port_vlan_add The ranges UAPI was introduced to the bridge in commit bdced7ef7838 ("bridge: support for multiple vlans and vlan ranges in setlink and dellink requests") (Jan 10 2015). But the VLAN ranges (parsed in br_afspec) have always been passed one by one, through struct bridge_vlan_info tmp_vinfo, to br_vlan_info. So the range never went too far in depth. Then Scott Feldman introduced the switchdev_port_bridge_setlink function in commit 47f8328bb1a4 ("switchdev: add new switchdev bridge setlink"). That marked the introduction of the SWITCHDEV_OBJ_PORT_VLAN, which made full use of the range. But switchdev_port_bridge_setlink was called like this: br_setlink -> br_afspec -> switchdev_port_bridge_setlink Basically, the switchdev and the bridge code were not tightly integrated. Then commit 41c498b9359e ("bridge: restore br_setlink back to original") came, and switchdev drivers were required to implement .ndo_bridge_setlink = switchdev_port_bridge_setlink for a while. In the meantime, commits such as 0944d6b5a2fa ("bridge: try switchdev op first in __vlan_vid_add/del") finally made switchdev penetrate the br_vlan_info() barrier and start to develop the call path we have today. But remember, br_vlan_info() still receives VLANs one by one. Then Arkadi Sharshevsky refactored the switchdev API in 2017 in commit 29ab586c3d83 ("net: switchdev: Remove bridge bypass support from switchdev") so that drivers would not implement .ndo_bridge_setlink any longer. The switchdev_port_bridge_setlink also got deleted. This refactoring removed the parallel bridge_setlink implementation from switchdev, and left the only switchdev VLAN objects to be the ones offloaded from __vlan_vid_add (basically RX filtering) and __vlan_add (the latter coming from commit 9c86ce2c1ae3 ("net: bridge: Notify about bridge VLANs")). That is to say, today the switchdev VLAN object ranges are not used in the kernel. Refactoring the above call path is a bit complicated, when the bridge VLAN call path is already a bit complicated. Let's go off and finish the job of commit 29ab586c3d83 by deleting the bogus iteration through the VLAN ranges from the drivers. Some aspects of this feature never made too much sense in the first place. For example, what is a range of VLANs all having the BRIDGE_VLAN_INFO_PVID flag supposed to mean, when a port can obviously have a single pvid? This particular configuration _is_ denied as of commit 6623c60dc28e ("bridge: vlan: enforce no pvid flag in vlan ranges"), but from an API perspective, the driver still has to play pretend, and only offload the vlan->vid_end as pvid. And the addition of a switchdev VLAN object can modify the flags of another, completely unrelated, switchdev VLAN object! (a VLAN that is PVID will invalidate the PVID flag from whatever other VLAN had previously been offloaded with switchdev and had that flag. Yet switchdev never notifies about that change, drivers are supposed to guess). Nonetheless, having a VLAN range in the API makes error handling look scarier than it really is - unwinding on errors and all of that. When in reality, no one really calls this API with more than one VLAN. It is all unnecessary complexity. And despite appearing pretentious (two-phase transactional model and all), the switchdev API is really sloppy because the VLAN addition and removal operations are not paired with one another (you can add a VLAN 100 times and delete it just once). The bridge notifies through switchdev of a VLAN addition not only when the flags of an existing VLAN change, but also when nothing changes. There are switchdev drivers out there who don't like adding a VLAN that has already been added, and those checks don't really belong at driver level. But the fact that the API contains ranges is yet another factor that prevents this from being addressed in the future. Of the existing switchdev pieces of hardware, it appears that only Mellanox Spectrum supports offloading more than one VLAN at a time, through mlxsw_sp_port_vlan_set. I have kept that code internal to the driver, because there is some more bookkeeping that makes use of it, but I deleted it from the switchdev API. But since the switchdev support for ranges has already been de facto deleted by a Mellanox employee and nobody noticed for 4 years, I'm going to assume it's not a biggie. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Ido Schimmel <idosch@nvidia.com> # switchdev and mlxsw Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de> # hellcreek Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 00:01:46 +00:00
dev_err(smi->dev, "failed to set up VLAN %04x", vlan->vid);
net: dsa: remove the transactional logic from VLAN objects It should be the driver's business to logically separate its VLAN offloading into a preparation and a commit phase, and some drivers don't need / can't do this. So remove the transactional shim from DSA and let drivers propagate errors directly from the .port_vlan_add callback. It would appear that the code has worse error handling now than it had before. DSA is the only in-kernel user of switchdev that offloads one switchdev object to more than one port: for every VLAN object offloaded to a user port, that VLAN is also offloaded to the CPU port. So the "prepare for user port -> check for errors -> prepare for CPU port -> check for errors -> commit for user port -> commit for CPU port" sequence appears to make more sense than the one we are using now: "offload to user port -> check for errors -> offload to CPU port -> check for errors", but it is really a compromise. In the new way, we can catch errors from the commit phase that we previously had to ignore. But we have our hands tied and cannot do any rollback now: if we add a VLAN on the CPU port and it fails, we can't do the rollback by simply deleting it from the user port, because the switchdev API is not so nice with us: it could have simply been there already, even with the same flags. So we don't even attempt to rollback anything on addition error, just leave whatever VLANs managed to get offloaded right where they are. This should not be a problem at all in practice. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Acked-by: Linus Walleij <linus.walleij@linaro.org> Acked-by: Jiri Pirko <jiri@nvidia.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 00:01:53 +00:00
return ret;
}
net: switchdev: remove vid_begin -> vid_end range from VLAN objects The call path of a switchdev VLAN addition to the bridge looks something like this today: nbp_vlan_init | __br_vlan_set_default_pvid | | | | | br_afspec | | | | | | | v | | | br_process_vlan_info | | | | | | | v | | | br_vlan_info | | | / \ / | | / \ / | | / \ / | | / \ / v v v v v nbp_vlan_add br_vlan_add ------+ | ^ ^ | | | / | | | | / / / | \ br_vlan_get_master/ / v \ ^ / / br_vlan_add_existing \ | / / | \ | / / / \ | / / / \ | / / / \ | / / / v | | v / __vlan_add / / | / / | / v | / __vlan_vid_add | / \ | / v v v br_switchdev_port_vlan_add The ranges UAPI was introduced to the bridge in commit bdced7ef7838 ("bridge: support for multiple vlans and vlan ranges in setlink and dellink requests") (Jan 10 2015). But the VLAN ranges (parsed in br_afspec) have always been passed one by one, through struct bridge_vlan_info tmp_vinfo, to br_vlan_info. So the range never went too far in depth. Then Scott Feldman introduced the switchdev_port_bridge_setlink function in commit 47f8328bb1a4 ("switchdev: add new switchdev bridge setlink"). That marked the introduction of the SWITCHDEV_OBJ_PORT_VLAN, which made full use of the range. But switchdev_port_bridge_setlink was called like this: br_setlink -> br_afspec -> switchdev_port_bridge_setlink Basically, the switchdev and the bridge code were not tightly integrated. Then commit 41c498b9359e ("bridge: restore br_setlink back to original") came, and switchdev drivers were required to implement .ndo_bridge_setlink = switchdev_port_bridge_setlink for a while. In the meantime, commits such as 0944d6b5a2fa ("bridge: try switchdev op first in __vlan_vid_add/del") finally made switchdev penetrate the br_vlan_info() barrier and start to develop the call path we have today. But remember, br_vlan_info() still receives VLANs one by one. Then Arkadi Sharshevsky refactored the switchdev API in 2017 in commit 29ab586c3d83 ("net: switchdev: Remove bridge bypass support from switchdev") so that drivers would not implement .ndo_bridge_setlink any longer. The switchdev_port_bridge_setlink also got deleted. This refactoring removed the parallel bridge_setlink implementation from switchdev, and left the only switchdev VLAN objects to be the ones offloaded from __vlan_vid_add (basically RX filtering) and __vlan_add (the latter coming from commit 9c86ce2c1ae3 ("net: bridge: Notify about bridge VLANs")). That is to say, today the switchdev VLAN object ranges are not used in the kernel. Refactoring the above call path is a bit complicated, when the bridge VLAN call path is already a bit complicated. Let's go off and finish the job of commit 29ab586c3d83 by deleting the bogus iteration through the VLAN ranges from the drivers. Some aspects of this feature never made too much sense in the first place. For example, what is a range of VLANs all having the BRIDGE_VLAN_INFO_PVID flag supposed to mean, when a port can obviously have a single pvid? This particular configuration _is_ denied as of commit 6623c60dc28e ("bridge: vlan: enforce no pvid flag in vlan ranges"), but from an API perspective, the driver still has to play pretend, and only offload the vlan->vid_end as pvid. And the addition of a switchdev VLAN object can modify the flags of another, completely unrelated, switchdev VLAN object! (a VLAN that is PVID will invalidate the PVID flag from whatever other VLAN had previously been offloaded with switchdev and had that flag. Yet switchdev never notifies about that change, drivers are supposed to guess). Nonetheless, having a VLAN range in the API makes error handling look scarier than it really is - unwinding on errors and all of that. When in reality, no one really calls this API with more than one VLAN. It is all unnecessary complexity. And despite appearing pretentious (two-phase transactional model and all), the switchdev API is really sloppy because the VLAN addition and removal operations are not paired with one another (you can add a VLAN 100 times and delete it just once). The bridge notifies through switchdev of a VLAN addition not only when the flags of an existing VLAN change, but also when nothing changes. There are switchdev drivers out there who don't like adding a VLAN that has already been added, and those checks don't really belong at driver level. But the fact that the API contains ranges is yet another factor that prevents this from being addressed in the future. Of the existing switchdev pieces of hardware, it appears that only Mellanox Spectrum supports offloading more than one VLAN at a time, through mlxsw_sp_port_vlan_set. I have kept that code internal to the driver, because there is some more bookkeeping that makes use of it, but I deleted it from the switchdev API. But since the switchdev support for ranges has already been de facto deleted by a Mellanox employee and nobody noticed for 4 years, I'm going to assume it's not a biggie. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Ido Schimmel <idosch@nvidia.com> # switchdev and mlxsw Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de> # hellcreek Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 00:01:46 +00:00
if (!pvid)
net: dsa: remove the transactional logic from VLAN objects It should be the driver's business to logically separate its VLAN offloading into a preparation and a commit phase, and some drivers don't need / can't do this. So remove the transactional shim from DSA and let drivers propagate errors directly from the .port_vlan_add callback. It would appear that the code has worse error handling now than it had before. DSA is the only in-kernel user of switchdev that offloads one switchdev object to more than one port: for every VLAN object offloaded to a user port, that VLAN is also offloaded to the CPU port. So the "prepare for user port -> check for errors -> prepare for CPU port -> check for errors -> commit for user port -> commit for CPU port" sequence appears to make more sense than the one we are using now: "offload to user port -> check for errors -> offload to CPU port -> check for errors", but it is really a compromise. In the new way, we can catch errors from the commit phase that we previously had to ignore. But we have our hands tied and cannot do any rollback now: if we add a VLAN on the CPU port and it fails, we can't do the rollback by simply deleting it from the user port, because the switchdev API is not so nice with us: it could have simply been there already, even with the same flags. So we don't even attempt to rollback anything on addition error, just leave whatever VLANs managed to get offloaded right where they are. This should not be a problem at all in practice. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Acked-by: Linus Walleij <linus.walleij@linaro.org> Acked-by: Jiri Pirko <jiri@nvidia.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 00:01:53 +00:00
return 0;
net: switchdev: remove vid_begin -> vid_end range from VLAN objects The call path of a switchdev VLAN addition to the bridge looks something like this today: nbp_vlan_init | __br_vlan_set_default_pvid | | | | | br_afspec | | | | | | | v | | | br_process_vlan_info | | | | | | | v | | | br_vlan_info | | | / \ / | | / \ / | | / \ / | | / \ / v v v v v nbp_vlan_add br_vlan_add ------+ | ^ ^ | | | / | | | | / / / | \ br_vlan_get_master/ / v \ ^ / / br_vlan_add_existing \ | / / | \ | / / / \ | / / / \ | / / / \ | / / / v | | v / __vlan_add / / | / / | / v | / __vlan_vid_add | / \ | / v v v br_switchdev_port_vlan_add The ranges UAPI was introduced to the bridge in commit bdced7ef7838 ("bridge: support for multiple vlans and vlan ranges in setlink and dellink requests") (Jan 10 2015). But the VLAN ranges (parsed in br_afspec) have always been passed one by one, through struct bridge_vlan_info tmp_vinfo, to br_vlan_info. So the range never went too far in depth. Then Scott Feldman introduced the switchdev_port_bridge_setlink function in commit 47f8328bb1a4 ("switchdev: add new switchdev bridge setlink"). That marked the introduction of the SWITCHDEV_OBJ_PORT_VLAN, which made full use of the range. But switchdev_port_bridge_setlink was called like this: br_setlink -> br_afspec -> switchdev_port_bridge_setlink Basically, the switchdev and the bridge code were not tightly integrated. Then commit 41c498b9359e ("bridge: restore br_setlink back to original") came, and switchdev drivers were required to implement .ndo_bridge_setlink = switchdev_port_bridge_setlink for a while. In the meantime, commits such as 0944d6b5a2fa ("bridge: try switchdev op first in __vlan_vid_add/del") finally made switchdev penetrate the br_vlan_info() barrier and start to develop the call path we have today. But remember, br_vlan_info() still receives VLANs one by one. Then Arkadi Sharshevsky refactored the switchdev API in 2017 in commit 29ab586c3d83 ("net: switchdev: Remove bridge bypass support from switchdev") so that drivers would not implement .ndo_bridge_setlink any longer. The switchdev_port_bridge_setlink also got deleted. This refactoring removed the parallel bridge_setlink implementation from switchdev, and left the only switchdev VLAN objects to be the ones offloaded from __vlan_vid_add (basically RX filtering) and __vlan_add (the latter coming from commit 9c86ce2c1ae3 ("net: bridge: Notify about bridge VLANs")). That is to say, today the switchdev VLAN object ranges are not used in the kernel. Refactoring the above call path is a bit complicated, when the bridge VLAN call path is already a bit complicated. Let's go off and finish the job of commit 29ab586c3d83 by deleting the bogus iteration through the VLAN ranges from the drivers. Some aspects of this feature never made too much sense in the first place. For example, what is a range of VLANs all having the BRIDGE_VLAN_INFO_PVID flag supposed to mean, when a port can obviously have a single pvid? This particular configuration _is_ denied as of commit 6623c60dc28e ("bridge: vlan: enforce no pvid flag in vlan ranges"), but from an API perspective, the driver still has to play pretend, and only offload the vlan->vid_end as pvid. And the addition of a switchdev VLAN object can modify the flags of another, completely unrelated, switchdev VLAN object! (a VLAN that is PVID will invalidate the PVID flag from whatever other VLAN had previously been offloaded with switchdev and had that flag. Yet switchdev never notifies about that change, drivers are supposed to guess). Nonetheless, having a VLAN range in the API makes error handling look scarier than it really is - unwinding on errors and all of that. When in reality, no one really calls this API with more than one VLAN. It is all unnecessary complexity. And despite appearing pretentious (two-phase transactional model and all), the switchdev API is really sloppy because the VLAN addition and removal operations are not paired with one another (you can add a VLAN 100 times and delete it just once). The bridge notifies through switchdev of a VLAN addition not only when the flags of an existing VLAN change, but also when nothing changes. There are switchdev drivers out there who don't like adding a VLAN that has already been added, and those checks don't really belong at driver level. But the fact that the API contains ranges is yet another factor that prevents this from being addressed in the future. Of the existing switchdev pieces of hardware, it appears that only Mellanox Spectrum supports offloading more than one VLAN at a time, through mlxsw_sp_port_vlan_set. I have kept that code internal to the driver, because there is some more bookkeeping that makes use of it, but I deleted it from the switchdev API. But since the switchdev support for ranges has already been de facto deleted by a Mellanox employee and nobody noticed for 4 years, I'm going to assume it's not a biggie. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Ido Schimmel <idosch@nvidia.com> # switchdev and mlxsw Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de> # hellcreek Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 00:01:46 +00:00
ret = rtl8366_set_pvid(smi, port, vlan->vid);
net: dsa: remove the transactional logic from VLAN objects It should be the driver's business to logically separate its VLAN offloading into a preparation and a commit phase, and some drivers don't need / can't do this. So remove the transactional shim from DSA and let drivers propagate errors directly from the .port_vlan_add callback. It would appear that the code has worse error handling now than it had before. DSA is the only in-kernel user of switchdev that offloads one switchdev object to more than one port: for every VLAN object offloaded to a user port, that VLAN is also offloaded to the CPU port. So the "prepare for user port -> check for errors -> prepare for CPU port -> check for errors -> commit for user port -> commit for CPU port" sequence appears to make more sense than the one we are using now: "offload to user port -> check for errors -> offload to CPU port -> check for errors", but it is really a compromise. In the new way, we can catch errors from the commit phase that we previously had to ignore. But we have our hands tied and cannot do any rollback now: if we add a VLAN on the CPU port and it fails, we can't do the rollback by simply deleting it from the user port, because the switchdev API is not so nice with us: it could have simply been there already, even with the same flags. So we don't even attempt to rollback anything on addition error, just leave whatever VLANs managed to get offloaded right where they are. This should not be a problem at all in practice. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Acked-by: Linus Walleij <linus.walleij@linaro.org> Acked-by: Jiri Pirko <jiri@nvidia.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 00:01:53 +00:00
if (ret) {
net: switchdev: remove vid_begin -> vid_end range from VLAN objects The call path of a switchdev VLAN addition to the bridge looks something like this today: nbp_vlan_init | __br_vlan_set_default_pvid | | | | | br_afspec | | | | | | | v | | | br_process_vlan_info | | | | | | | v | | | br_vlan_info | | | / \ / | | / \ / | | / \ / | | / \ / v v v v v nbp_vlan_add br_vlan_add ------+ | ^ ^ | | | / | | | | / / / | \ br_vlan_get_master/ / v \ ^ / / br_vlan_add_existing \ | / / | \ | / / / \ | / / / \ | / / / \ | / / / v | | v / __vlan_add / / | / / | / v | / __vlan_vid_add | / \ | / v v v br_switchdev_port_vlan_add The ranges UAPI was introduced to the bridge in commit bdced7ef7838 ("bridge: support for multiple vlans and vlan ranges in setlink and dellink requests") (Jan 10 2015). But the VLAN ranges (parsed in br_afspec) have always been passed one by one, through struct bridge_vlan_info tmp_vinfo, to br_vlan_info. So the range never went too far in depth. Then Scott Feldman introduced the switchdev_port_bridge_setlink function in commit 47f8328bb1a4 ("switchdev: add new switchdev bridge setlink"). That marked the introduction of the SWITCHDEV_OBJ_PORT_VLAN, which made full use of the range. But switchdev_port_bridge_setlink was called like this: br_setlink -> br_afspec -> switchdev_port_bridge_setlink Basically, the switchdev and the bridge code were not tightly integrated. Then commit 41c498b9359e ("bridge: restore br_setlink back to original") came, and switchdev drivers were required to implement .ndo_bridge_setlink = switchdev_port_bridge_setlink for a while. In the meantime, commits such as 0944d6b5a2fa ("bridge: try switchdev op first in __vlan_vid_add/del") finally made switchdev penetrate the br_vlan_info() barrier and start to develop the call path we have today. But remember, br_vlan_info() still receives VLANs one by one. Then Arkadi Sharshevsky refactored the switchdev API in 2017 in commit 29ab586c3d83 ("net: switchdev: Remove bridge bypass support from switchdev") so that drivers would not implement .ndo_bridge_setlink any longer. The switchdev_port_bridge_setlink also got deleted. This refactoring removed the parallel bridge_setlink implementation from switchdev, and left the only switchdev VLAN objects to be the ones offloaded from __vlan_vid_add (basically RX filtering) and __vlan_add (the latter coming from commit 9c86ce2c1ae3 ("net: bridge: Notify about bridge VLANs")). That is to say, today the switchdev VLAN object ranges are not used in the kernel. Refactoring the above call path is a bit complicated, when the bridge VLAN call path is already a bit complicated. Let's go off and finish the job of commit 29ab586c3d83 by deleting the bogus iteration through the VLAN ranges from the drivers. Some aspects of this feature never made too much sense in the first place. For example, what is a range of VLANs all having the BRIDGE_VLAN_INFO_PVID flag supposed to mean, when a port can obviously have a single pvid? This particular configuration _is_ denied as of commit 6623c60dc28e ("bridge: vlan: enforce no pvid flag in vlan ranges"), but from an API perspective, the driver still has to play pretend, and only offload the vlan->vid_end as pvid. And the addition of a switchdev VLAN object can modify the flags of another, completely unrelated, switchdev VLAN object! (a VLAN that is PVID will invalidate the PVID flag from whatever other VLAN had previously been offloaded with switchdev and had that flag. Yet switchdev never notifies about that change, drivers are supposed to guess). Nonetheless, having a VLAN range in the API makes error handling look scarier than it really is - unwinding on errors and all of that. When in reality, no one really calls this API with more than one VLAN. It is all unnecessary complexity. And despite appearing pretentious (two-phase transactional model and all), the switchdev API is really sloppy because the VLAN addition and removal operations are not paired with one another (you can add a VLAN 100 times and delete it just once). The bridge notifies through switchdev of a VLAN addition not only when the flags of an existing VLAN change, but also when nothing changes. There are switchdev drivers out there who don't like adding a VLAN that has already been added, and those checks don't really belong at driver level. But the fact that the API contains ranges is yet another factor that prevents this from being addressed in the future. Of the existing switchdev pieces of hardware, it appears that only Mellanox Spectrum supports offloading more than one VLAN at a time, through mlxsw_sp_port_vlan_set. I have kept that code internal to the driver, because there is some more bookkeeping that makes use of it, but I deleted it from the switchdev API. But since the switchdev support for ranges has already been de facto deleted by a Mellanox employee and nobody noticed for 4 years, I'm going to assume it's not a biggie. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Ido Schimmel <idosch@nvidia.com> # switchdev and mlxsw Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de> # hellcreek Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 00:01:46 +00:00
dev_err(smi->dev, "failed to set PVID on port %d to VLAN %04x",
port, vlan->vid);
net: dsa: remove the transactional logic from VLAN objects It should be the driver's business to logically separate its VLAN offloading into a preparation and a commit phase, and some drivers don't need / can't do this. So remove the transactional shim from DSA and let drivers propagate errors directly from the .port_vlan_add callback. It would appear that the code has worse error handling now than it had before. DSA is the only in-kernel user of switchdev that offloads one switchdev object to more than one port: for every VLAN object offloaded to a user port, that VLAN is also offloaded to the CPU port. So the "prepare for user port -> check for errors -> prepare for CPU port -> check for errors -> commit for user port -> commit for CPU port" sequence appears to make more sense than the one we are using now: "offload to user port -> check for errors -> offload to CPU port -> check for errors", but it is really a compromise. In the new way, we can catch errors from the commit phase that we previously had to ignore. But we have our hands tied and cannot do any rollback now: if we add a VLAN on the CPU port and it fails, we can't do the rollback by simply deleting it from the user port, because the switchdev API is not so nice with us: it could have simply been there already, even with the same flags. So we don't even attempt to rollback anything on addition error, just leave whatever VLANs managed to get offloaded right where they are. This should not be a problem at all in practice. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Acked-by: Linus Walleij <linus.walleij@linaro.org> Acked-by: Jiri Pirko <jiri@nvidia.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 00:01:53 +00:00
return ret;
}
net: dsa: remove the transactional logic from VLAN objects It should be the driver's business to logically separate its VLAN offloading into a preparation and a commit phase, and some drivers don't need / can't do this. So remove the transactional shim from DSA and let drivers propagate errors directly from the .port_vlan_add callback. It would appear that the code has worse error handling now than it had before. DSA is the only in-kernel user of switchdev that offloads one switchdev object to more than one port: for every VLAN object offloaded to a user port, that VLAN is also offloaded to the CPU port. So the "prepare for user port -> check for errors -> prepare for CPU port -> check for errors -> commit for user port -> commit for CPU port" sequence appears to make more sense than the one we are using now: "offload to user port -> check for errors -> offload to CPU port -> check for errors", but it is really a compromise. In the new way, we can catch errors from the commit phase that we previously had to ignore. But we have our hands tied and cannot do any rollback now: if we add a VLAN on the CPU port and it fails, we can't do the rollback by simply deleting it from the user port, because the switchdev API is not so nice with us: it could have simply been there already, even with the same flags. So we don't even attempt to rollback anything on addition error, just leave whatever VLANs managed to get offloaded right where they are. This should not be a problem at all in practice. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Acked-by: Linus Walleij <linus.walleij@linaro.org> Acked-by: Jiri Pirko <jiri@nvidia.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 00:01:53 +00:00
return 0;
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
}
EXPORT_SYMBOL_GPL(rtl8366_vlan_add);
int rtl8366_vlan_del(struct dsa_switch *ds, int port,
const struct switchdev_obj_port_vlan *vlan)
{
struct realtek_smi *smi = ds->priv;
net: switchdev: remove vid_begin -> vid_end range from VLAN objects The call path of a switchdev VLAN addition to the bridge looks something like this today: nbp_vlan_init | __br_vlan_set_default_pvid | | | | | br_afspec | | | | | | | v | | | br_process_vlan_info | | | | | | | v | | | br_vlan_info | | | / \ / | | / \ / | | / \ / | | / \ / v v v v v nbp_vlan_add br_vlan_add ------+ | ^ ^ | | | / | | | | / / / | \ br_vlan_get_master/ / v \ ^ / / br_vlan_add_existing \ | / / | \ | / / / \ | / / / \ | / / / \ | / / / v | | v / __vlan_add / / | / / | / v | / __vlan_vid_add | / \ | / v v v br_switchdev_port_vlan_add The ranges UAPI was introduced to the bridge in commit bdced7ef7838 ("bridge: support for multiple vlans and vlan ranges in setlink and dellink requests") (Jan 10 2015). But the VLAN ranges (parsed in br_afspec) have always been passed one by one, through struct bridge_vlan_info tmp_vinfo, to br_vlan_info. So the range never went too far in depth. Then Scott Feldman introduced the switchdev_port_bridge_setlink function in commit 47f8328bb1a4 ("switchdev: add new switchdev bridge setlink"). That marked the introduction of the SWITCHDEV_OBJ_PORT_VLAN, which made full use of the range. But switchdev_port_bridge_setlink was called like this: br_setlink -> br_afspec -> switchdev_port_bridge_setlink Basically, the switchdev and the bridge code were not tightly integrated. Then commit 41c498b9359e ("bridge: restore br_setlink back to original") came, and switchdev drivers were required to implement .ndo_bridge_setlink = switchdev_port_bridge_setlink for a while. In the meantime, commits such as 0944d6b5a2fa ("bridge: try switchdev op first in __vlan_vid_add/del") finally made switchdev penetrate the br_vlan_info() barrier and start to develop the call path we have today. But remember, br_vlan_info() still receives VLANs one by one. Then Arkadi Sharshevsky refactored the switchdev API in 2017 in commit 29ab586c3d83 ("net: switchdev: Remove bridge bypass support from switchdev") so that drivers would not implement .ndo_bridge_setlink any longer. The switchdev_port_bridge_setlink also got deleted. This refactoring removed the parallel bridge_setlink implementation from switchdev, and left the only switchdev VLAN objects to be the ones offloaded from __vlan_vid_add (basically RX filtering) and __vlan_add (the latter coming from commit 9c86ce2c1ae3 ("net: bridge: Notify about bridge VLANs")). That is to say, today the switchdev VLAN object ranges are not used in the kernel. Refactoring the above call path is a bit complicated, when the bridge VLAN call path is already a bit complicated. Let's go off and finish the job of commit 29ab586c3d83 by deleting the bogus iteration through the VLAN ranges from the drivers. Some aspects of this feature never made too much sense in the first place. For example, what is a range of VLANs all having the BRIDGE_VLAN_INFO_PVID flag supposed to mean, when a port can obviously have a single pvid? This particular configuration _is_ denied as of commit 6623c60dc28e ("bridge: vlan: enforce no pvid flag in vlan ranges"), but from an API perspective, the driver still has to play pretend, and only offload the vlan->vid_end as pvid. And the addition of a switchdev VLAN object can modify the flags of another, completely unrelated, switchdev VLAN object! (a VLAN that is PVID will invalidate the PVID flag from whatever other VLAN had previously been offloaded with switchdev and had that flag. Yet switchdev never notifies about that change, drivers are supposed to guess). Nonetheless, having a VLAN range in the API makes error handling look scarier than it really is - unwinding on errors and all of that. When in reality, no one really calls this API with more than one VLAN. It is all unnecessary complexity. And despite appearing pretentious (two-phase transactional model and all), the switchdev API is really sloppy because the VLAN addition and removal operations are not paired with one another (you can add a VLAN 100 times and delete it just once). The bridge notifies through switchdev of a VLAN addition not only when the flags of an existing VLAN change, but also when nothing changes. There are switchdev drivers out there who don't like adding a VLAN that has already been added, and those checks don't really belong at driver level. But the fact that the API contains ranges is yet another factor that prevents this from being addressed in the future. Of the existing switchdev pieces of hardware, it appears that only Mellanox Spectrum supports offloading more than one VLAN at a time, through mlxsw_sp_port_vlan_set. I have kept that code internal to the driver, because there is some more bookkeeping that makes use of it, but I deleted it from the switchdev API. But since the switchdev support for ranges has already been de facto deleted by a Mellanox employee and nobody noticed for 4 years, I'm going to assume it's not a biggie. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Ido Schimmel <idosch@nvidia.com> # switchdev and mlxsw Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de> # hellcreek Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 00:01:46 +00:00
int ret, i;
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
net: switchdev: remove vid_begin -> vid_end range from VLAN objects The call path of a switchdev VLAN addition to the bridge looks something like this today: nbp_vlan_init | __br_vlan_set_default_pvid | | | | | br_afspec | | | | | | | v | | | br_process_vlan_info | | | | | | | v | | | br_vlan_info | | | / \ / | | / \ / | | / \ / | | / \ / v v v v v nbp_vlan_add br_vlan_add ------+ | ^ ^ | | | / | | | | / / / | \ br_vlan_get_master/ / v \ ^ / / br_vlan_add_existing \ | / / | \ | / / / \ | / / / \ | / / / \ | / / / v | | v / __vlan_add / / | / / | / v | / __vlan_vid_add | / \ | / v v v br_switchdev_port_vlan_add The ranges UAPI was introduced to the bridge in commit bdced7ef7838 ("bridge: support for multiple vlans and vlan ranges in setlink and dellink requests") (Jan 10 2015). But the VLAN ranges (parsed in br_afspec) have always been passed one by one, through struct bridge_vlan_info tmp_vinfo, to br_vlan_info. So the range never went too far in depth. Then Scott Feldman introduced the switchdev_port_bridge_setlink function in commit 47f8328bb1a4 ("switchdev: add new switchdev bridge setlink"). That marked the introduction of the SWITCHDEV_OBJ_PORT_VLAN, which made full use of the range. But switchdev_port_bridge_setlink was called like this: br_setlink -> br_afspec -> switchdev_port_bridge_setlink Basically, the switchdev and the bridge code were not tightly integrated. Then commit 41c498b9359e ("bridge: restore br_setlink back to original") came, and switchdev drivers were required to implement .ndo_bridge_setlink = switchdev_port_bridge_setlink for a while. In the meantime, commits such as 0944d6b5a2fa ("bridge: try switchdev op first in __vlan_vid_add/del") finally made switchdev penetrate the br_vlan_info() barrier and start to develop the call path we have today. But remember, br_vlan_info() still receives VLANs one by one. Then Arkadi Sharshevsky refactored the switchdev API in 2017 in commit 29ab586c3d83 ("net: switchdev: Remove bridge bypass support from switchdev") so that drivers would not implement .ndo_bridge_setlink any longer. The switchdev_port_bridge_setlink also got deleted. This refactoring removed the parallel bridge_setlink implementation from switchdev, and left the only switchdev VLAN objects to be the ones offloaded from __vlan_vid_add (basically RX filtering) and __vlan_add (the latter coming from commit 9c86ce2c1ae3 ("net: bridge: Notify about bridge VLANs")). That is to say, today the switchdev VLAN object ranges are not used in the kernel. Refactoring the above call path is a bit complicated, when the bridge VLAN call path is already a bit complicated. Let's go off and finish the job of commit 29ab586c3d83 by deleting the bogus iteration through the VLAN ranges from the drivers. Some aspects of this feature never made too much sense in the first place. For example, what is a range of VLANs all having the BRIDGE_VLAN_INFO_PVID flag supposed to mean, when a port can obviously have a single pvid? This particular configuration _is_ denied as of commit 6623c60dc28e ("bridge: vlan: enforce no pvid flag in vlan ranges"), but from an API perspective, the driver still has to play pretend, and only offload the vlan->vid_end as pvid. And the addition of a switchdev VLAN object can modify the flags of another, completely unrelated, switchdev VLAN object! (a VLAN that is PVID will invalidate the PVID flag from whatever other VLAN had previously been offloaded with switchdev and had that flag. Yet switchdev never notifies about that change, drivers are supposed to guess). Nonetheless, having a VLAN range in the API makes error handling look scarier than it really is - unwinding on errors and all of that. When in reality, no one really calls this API with more than one VLAN. It is all unnecessary complexity. And despite appearing pretentious (two-phase transactional model and all), the switchdev API is really sloppy because the VLAN addition and removal operations are not paired with one another (you can add a VLAN 100 times and delete it just once). The bridge notifies through switchdev of a VLAN addition not only when the flags of an existing VLAN change, but also when nothing changes. There are switchdev drivers out there who don't like adding a VLAN that has already been added, and those checks don't really belong at driver level. But the fact that the API contains ranges is yet another factor that prevents this from being addressed in the future. Of the existing switchdev pieces of hardware, it appears that only Mellanox Spectrum supports offloading more than one VLAN at a time, through mlxsw_sp_port_vlan_set. I have kept that code internal to the driver, because there is some more bookkeeping that makes use of it, but I deleted it from the switchdev API. But since the switchdev support for ranges has already been de facto deleted by a Mellanox employee and nobody noticed for 4 years, I'm going to assume it's not a biggie. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Ido Schimmel <idosch@nvidia.com> # switchdev and mlxsw Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de> # hellcreek Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 00:01:46 +00:00
dev_info(smi->dev, "del VLAN %04x on port %d\n", vlan->vid, port);
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
net: switchdev: remove vid_begin -> vid_end range from VLAN objects The call path of a switchdev VLAN addition to the bridge looks something like this today: nbp_vlan_init | __br_vlan_set_default_pvid | | | | | br_afspec | | | | | | | v | | | br_process_vlan_info | | | | | | | v | | | br_vlan_info | | | / \ / | | / \ / | | / \ / | | / \ / v v v v v nbp_vlan_add br_vlan_add ------+ | ^ ^ | | | / | | | | / / / | \ br_vlan_get_master/ / v \ ^ / / br_vlan_add_existing \ | / / | \ | / / / \ | / / / \ | / / / \ | / / / v | | v / __vlan_add / / | / / | / v | / __vlan_vid_add | / \ | / v v v br_switchdev_port_vlan_add The ranges UAPI was introduced to the bridge in commit bdced7ef7838 ("bridge: support for multiple vlans and vlan ranges in setlink and dellink requests") (Jan 10 2015). But the VLAN ranges (parsed in br_afspec) have always been passed one by one, through struct bridge_vlan_info tmp_vinfo, to br_vlan_info. So the range never went too far in depth. Then Scott Feldman introduced the switchdev_port_bridge_setlink function in commit 47f8328bb1a4 ("switchdev: add new switchdev bridge setlink"). That marked the introduction of the SWITCHDEV_OBJ_PORT_VLAN, which made full use of the range. But switchdev_port_bridge_setlink was called like this: br_setlink -> br_afspec -> switchdev_port_bridge_setlink Basically, the switchdev and the bridge code were not tightly integrated. Then commit 41c498b9359e ("bridge: restore br_setlink back to original") came, and switchdev drivers were required to implement .ndo_bridge_setlink = switchdev_port_bridge_setlink for a while. In the meantime, commits such as 0944d6b5a2fa ("bridge: try switchdev op first in __vlan_vid_add/del") finally made switchdev penetrate the br_vlan_info() barrier and start to develop the call path we have today. But remember, br_vlan_info() still receives VLANs one by one. Then Arkadi Sharshevsky refactored the switchdev API in 2017 in commit 29ab586c3d83 ("net: switchdev: Remove bridge bypass support from switchdev") so that drivers would not implement .ndo_bridge_setlink any longer. The switchdev_port_bridge_setlink also got deleted. This refactoring removed the parallel bridge_setlink implementation from switchdev, and left the only switchdev VLAN objects to be the ones offloaded from __vlan_vid_add (basically RX filtering) and __vlan_add (the latter coming from commit 9c86ce2c1ae3 ("net: bridge: Notify about bridge VLANs")). That is to say, today the switchdev VLAN object ranges are not used in the kernel. Refactoring the above call path is a bit complicated, when the bridge VLAN call path is already a bit complicated. Let's go off and finish the job of commit 29ab586c3d83 by deleting the bogus iteration through the VLAN ranges from the drivers. Some aspects of this feature never made too much sense in the first place. For example, what is a range of VLANs all having the BRIDGE_VLAN_INFO_PVID flag supposed to mean, when a port can obviously have a single pvid? This particular configuration _is_ denied as of commit 6623c60dc28e ("bridge: vlan: enforce no pvid flag in vlan ranges"), but from an API perspective, the driver still has to play pretend, and only offload the vlan->vid_end as pvid. And the addition of a switchdev VLAN object can modify the flags of another, completely unrelated, switchdev VLAN object! (a VLAN that is PVID will invalidate the PVID flag from whatever other VLAN had previously been offloaded with switchdev and had that flag. Yet switchdev never notifies about that change, drivers are supposed to guess). Nonetheless, having a VLAN range in the API makes error handling look scarier than it really is - unwinding on errors and all of that. When in reality, no one really calls this API with more than one VLAN. It is all unnecessary complexity. And despite appearing pretentious (two-phase transactional model and all), the switchdev API is really sloppy because the VLAN addition and removal operations are not paired with one another (you can add a VLAN 100 times and delete it just once). The bridge notifies through switchdev of a VLAN addition not only when the flags of an existing VLAN change, but also when nothing changes. There are switchdev drivers out there who don't like adding a VLAN that has already been added, and those checks don't really belong at driver level. But the fact that the API contains ranges is yet another factor that prevents this from being addressed in the future. Of the existing switchdev pieces of hardware, it appears that only Mellanox Spectrum supports offloading more than one VLAN at a time, through mlxsw_sp_port_vlan_set. I have kept that code internal to the driver, because there is some more bookkeeping that makes use of it, but I deleted it from the switchdev API. But since the switchdev support for ranges has already been de facto deleted by a Mellanox employee and nobody noticed for 4 years, I'm going to assume it's not a biggie. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Ido Schimmel <idosch@nvidia.com> # switchdev and mlxsw Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de> # hellcreek Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 00:01:46 +00:00
for (i = 0; i < smi->num_vlan_mc; i++) {
struct rtl8366_vlan_mc vlanmc;
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
net: switchdev: remove vid_begin -> vid_end range from VLAN objects The call path of a switchdev VLAN addition to the bridge looks something like this today: nbp_vlan_init | __br_vlan_set_default_pvid | | | | | br_afspec | | | | | | | v | | | br_process_vlan_info | | | | | | | v | | | br_vlan_info | | | / \ / | | / \ / | | / \ / | | / \ / v v v v v nbp_vlan_add br_vlan_add ------+ | ^ ^ | | | / | | | | / / / | \ br_vlan_get_master/ / v \ ^ / / br_vlan_add_existing \ | / / | \ | / / / \ | / / / \ | / / / \ | / / / v | | v / __vlan_add / / | / / | / v | / __vlan_vid_add | / \ | / v v v br_switchdev_port_vlan_add The ranges UAPI was introduced to the bridge in commit bdced7ef7838 ("bridge: support for multiple vlans and vlan ranges in setlink and dellink requests") (Jan 10 2015). But the VLAN ranges (parsed in br_afspec) have always been passed one by one, through struct bridge_vlan_info tmp_vinfo, to br_vlan_info. So the range never went too far in depth. Then Scott Feldman introduced the switchdev_port_bridge_setlink function in commit 47f8328bb1a4 ("switchdev: add new switchdev bridge setlink"). That marked the introduction of the SWITCHDEV_OBJ_PORT_VLAN, which made full use of the range. But switchdev_port_bridge_setlink was called like this: br_setlink -> br_afspec -> switchdev_port_bridge_setlink Basically, the switchdev and the bridge code were not tightly integrated. Then commit 41c498b9359e ("bridge: restore br_setlink back to original") came, and switchdev drivers were required to implement .ndo_bridge_setlink = switchdev_port_bridge_setlink for a while. In the meantime, commits such as 0944d6b5a2fa ("bridge: try switchdev op first in __vlan_vid_add/del") finally made switchdev penetrate the br_vlan_info() barrier and start to develop the call path we have today. But remember, br_vlan_info() still receives VLANs one by one. Then Arkadi Sharshevsky refactored the switchdev API in 2017 in commit 29ab586c3d83 ("net: switchdev: Remove bridge bypass support from switchdev") so that drivers would not implement .ndo_bridge_setlink any longer. The switchdev_port_bridge_setlink also got deleted. This refactoring removed the parallel bridge_setlink implementation from switchdev, and left the only switchdev VLAN objects to be the ones offloaded from __vlan_vid_add (basically RX filtering) and __vlan_add (the latter coming from commit 9c86ce2c1ae3 ("net: bridge: Notify about bridge VLANs")). That is to say, today the switchdev VLAN object ranges are not used in the kernel. Refactoring the above call path is a bit complicated, when the bridge VLAN call path is already a bit complicated. Let's go off and finish the job of commit 29ab586c3d83 by deleting the bogus iteration through the VLAN ranges from the drivers. Some aspects of this feature never made too much sense in the first place. For example, what is a range of VLANs all having the BRIDGE_VLAN_INFO_PVID flag supposed to mean, when a port can obviously have a single pvid? This particular configuration _is_ denied as of commit 6623c60dc28e ("bridge: vlan: enforce no pvid flag in vlan ranges"), but from an API perspective, the driver still has to play pretend, and only offload the vlan->vid_end as pvid. And the addition of a switchdev VLAN object can modify the flags of another, completely unrelated, switchdev VLAN object! (a VLAN that is PVID will invalidate the PVID flag from whatever other VLAN had previously been offloaded with switchdev and had that flag. Yet switchdev never notifies about that change, drivers are supposed to guess). Nonetheless, having a VLAN range in the API makes error handling look scarier than it really is - unwinding on errors and all of that. When in reality, no one really calls this API with more than one VLAN. It is all unnecessary complexity. And despite appearing pretentious (two-phase transactional model and all), the switchdev API is really sloppy because the VLAN addition and removal operations are not paired with one another (you can add a VLAN 100 times and delete it just once). The bridge notifies through switchdev of a VLAN addition not only when the flags of an existing VLAN change, but also when nothing changes. There are switchdev drivers out there who don't like adding a VLAN that has already been added, and those checks don't really belong at driver level. But the fact that the API contains ranges is yet another factor that prevents this from being addressed in the future. Of the existing switchdev pieces of hardware, it appears that only Mellanox Spectrum supports offloading more than one VLAN at a time, through mlxsw_sp_port_vlan_set. I have kept that code internal to the driver, because there is some more bookkeeping that makes use of it, but I deleted it from the switchdev API. But since the switchdev support for ranges has already been de facto deleted by a Mellanox employee and nobody noticed for 4 years, I'm going to assume it's not a biggie. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Ido Schimmel <idosch@nvidia.com> # switchdev and mlxsw Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de> # hellcreek Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 00:01:46 +00:00
ret = smi->ops->get_vlan_mc(smi, i, &vlanmc);
if (ret)
return ret;
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
net: switchdev: remove vid_begin -> vid_end range from VLAN objects The call path of a switchdev VLAN addition to the bridge looks something like this today: nbp_vlan_init | __br_vlan_set_default_pvid | | | | | br_afspec | | | | | | | v | | | br_process_vlan_info | | | | | | | v | | | br_vlan_info | | | / \ / | | / \ / | | / \ / | | / \ / v v v v v nbp_vlan_add br_vlan_add ------+ | ^ ^ | | | / | | | | / / / | \ br_vlan_get_master/ / v \ ^ / / br_vlan_add_existing \ | / / | \ | / / / \ | / / / \ | / / / \ | / / / v | | v / __vlan_add / / | / / | / v | / __vlan_vid_add | / \ | / v v v br_switchdev_port_vlan_add The ranges UAPI was introduced to the bridge in commit bdced7ef7838 ("bridge: support for multiple vlans and vlan ranges in setlink and dellink requests") (Jan 10 2015). But the VLAN ranges (parsed in br_afspec) have always been passed one by one, through struct bridge_vlan_info tmp_vinfo, to br_vlan_info. So the range never went too far in depth. Then Scott Feldman introduced the switchdev_port_bridge_setlink function in commit 47f8328bb1a4 ("switchdev: add new switchdev bridge setlink"). That marked the introduction of the SWITCHDEV_OBJ_PORT_VLAN, which made full use of the range. But switchdev_port_bridge_setlink was called like this: br_setlink -> br_afspec -> switchdev_port_bridge_setlink Basically, the switchdev and the bridge code were not tightly integrated. Then commit 41c498b9359e ("bridge: restore br_setlink back to original") came, and switchdev drivers were required to implement .ndo_bridge_setlink = switchdev_port_bridge_setlink for a while. In the meantime, commits such as 0944d6b5a2fa ("bridge: try switchdev op first in __vlan_vid_add/del") finally made switchdev penetrate the br_vlan_info() barrier and start to develop the call path we have today. But remember, br_vlan_info() still receives VLANs one by one. Then Arkadi Sharshevsky refactored the switchdev API in 2017 in commit 29ab586c3d83 ("net: switchdev: Remove bridge bypass support from switchdev") so that drivers would not implement .ndo_bridge_setlink any longer. The switchdev_port_bridge_setlink also got deleted. This refactoring removed the parallel bridge_setlink implementation from switchdev, and left the only switchdev VLAN objects to be the ones offloaded from __vlan_vid_add (basically RX filtering) and __vlan_add (the latter coming from commit 9c86ce2c1ae3 ("net: bridge: Notify about bridge VLANs")). That is to say, today the switchdev VLAN object ranges are not used in the kernel. Refactoring the above call path is a bit complicated, when the bridge VLAN call path is already a bit complicated. Let's go off and finish the job of commit 29ab586c3d83 by deleting the bogus iteration through the VLAN ranges from the drivers. Some aspects of this feature never made too much sense in the first place. For example, what is a range of VLANs all having the BRIDGE_VLAN_INFO_PVID flag supposed to mean, when a port can obviously have a single pvid? This particular configuration _is_ denied as of commit 6623c60dc28e ("bridge: vlan: enforce no pvid flag in vlan ranges"), but from an API perspective, the driver still has to play pretend, and only offload the vlan->vid_end as pvid. And the addition of a switchdev VLAN object can modify the flags of another, completely unrelated, switchdev VLAN object! (a VLAN that is PVID will invalidate the PVID flag from whatever other VLAN had previously been offloaded with switchdev and had that flag. Yet switchdev never notifies about that change, drivers are supposed to guess). Nonetheless, having a VLAN range in the API makes error handling look scarier than it really is - unwinding on errors and all of that. When in reality, no one really calls this API with more than one VLAN. It is all unnecessary complexity. And despite appearing pretentious (two-phase transactional model and all), the switchdev API is really sloppy because the VLAN addition and removal operations are not paired with one another (you can add a VLAN 100 times and delete it just once). The bridge notifies through switchdev of a VLAN addition not only when the flags of an existing VLAN change, but also when nothing changes. There are switchdev drivers out there who don't like adding a VLAN that has already been added, and those checks don't really belong at driver level. But the fact that the API contains ranges is yet another factor that prevents this from being addressed in the future. Of the existing switchdev pieces of hardware, it appears that only Mellanox Spectrum supports offloading more than one VLAN at a time, through mlxsw_sp_port_vlan_set. I have kept that code internal to the driver, because there is some more bookkeeping that makes use of it, but I deleted it from the switchdev API. But since the switchdev support for ranges has already been de facto deleted by a Mellanox employee and nobody noticed for 4 years, I'm going to assume it's not a biggie. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Ido Schimmel <idosch@nvidia.com> # switchdev and mlxsw Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de> # hellcreek Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 00:01:46 +00:00
if (vlan->vid == vlanmc.vid) {
/* Remove this port from the VLAN */
vlanmc.member &= ~BIT(port);
vlanmc.untag &= ~BIT(port);
/*
* If no ports are members of this VLAN
* anymore then clear the whole member
* config so it can be reused.
*/
if (!vlanmc.member && vlanmc.untag) {
vlanmc.vid = 0;
vlanmc.priority = 0;
vlanmc.fid = 0;
}
ret = smi->ops->set_vlan_mc(smi, i, &vlanmc);
if (ret) {
dev_err(smi->dev,
"failed to remove VLAN %04x\n",
vlan->vid);
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
return ret;
}
net: switchdev: remove vid_begin -> vid_end range from VLAN objects The call path of a switchdev VLAN addition to the bridge looks something like this today: nbp_vlan_init | __br_vlan_set_default_pvid | | | | | br_afspec | | | | | | | v | | | br_process_vlan_info | | | | | | | v | | | br_vlan_info | | | / \ / | | / \ / | | / \ / | | / \ / v v v v v nbp_vlan_add br_vlan_add ------+ | ^ ^ | | | / | | | | / / / | \ br_vlan_get_master/ / v \ ^ / / br_vlan_add_existing \ | / / | \ | / / / \ | / / / \ | / / / \ | / / / v | | v / __vlan_add / / | / / | / v | / __vlan_vid_add | / \ | / v v v br_switchdev_port_vlan_add The ranges UAPI was introduced to the bridge in commit bdced7ef7838 ("bridge: support for multiple vlans and vlan ranges in setlink and dellink requests") (Jan 10 2015). But the VLAN ranges (parsed in br_afspec) have always been passed one by one, through struct bridge_vlan_info tmp_vinfo, to br_vlan_info. So the range never went too far in depth. Then Scott Feldman introduced the switchdev_port_bridge_setlink function in commit 47f8328bb1a4 ("switchdev: add new switchdev bridge setlink"). That marked the introduction of the SWITCHDEV_OBJ_PORT_VLAN, which made full use of the range. But switchdev_port_bridge_setlink was called like this: br_setlink -> br_afspec -> switchdev_port_bridge_setlink Basically, the switchdev and the bridge code were not tightly integrated. Then commit 41c498b9359e ("bridge: restore br_setlink back to original") came, and switchdev drivers were required to implement .ndo_bridge_setlink = switchdev_port_bridge_setlink for a while. In the meantime, commits such as 0944d6b5a2fa ("bridge: try switchdev op first in __vlan_vid_add/del") finally made switchdev penetrate the br_vlan_info() barrier and start to develop the call path we have today. But remember, br_vlan_info() still receives VLANs one by one. Then Arkadi Sharshevsky refactored the switchdev API in 2017 in commit 29ab586c3d83 ("net: switchdev: Remove bridge bypass support from switchdev") so that drivers would not implement .ndo_bridge_setlink any longer. The switchdev_port_bridge_setlink also got deleted. This refactoring removed the parallel bridge_setlink implementation from switchdev, and left the only switchdev VLAN objects to be the ones offloaded from __vlan_vid_add (basically RX filtering) and __vlan_add (the latter coming from commit 9c86ce2c1ae3 ("net: bridge: Notify about bridge VLANs")). That is to say, today the switchdev VLAN object ranges are not used in the kernel. Refactoring the above call path is a bit complicated, when the bridge VLAN call path is already a bit complicated. Let's go off and finish the job of commit 29ab586c3d83 by deleting the bogus iteration through the VLAN ranges from the drivers. Some aspects of this feature never made too much sense in the first place. For example, what is a range of VLANs all having the BRIDGE_VLAN_INFO_PVID flag supposed to mean, when a port can obviously have a single pvid? This particular configuration _is_ denied as of commit 6623c60dc28e ("bridge: vlan: enforce no pvid flag in vlan ranges"), but from an API perspective, the driver still has to play pretend, and only offload the vlan->vid_end as pvid. And the addition of a switchdev VLAN object can modify the flags of another, completely unrelated, switchdev VLAN object! (a VLAN that is PVID will invalidate the PVID flag from whatever other VLAN had previously been offloaded with switchdev and had that flag. Yet switchdev never notifies about that change, drivers are supposed to guess). Nonetheless, having a VLAN range in the API makes error handling look scarier than it really is - unwinding on errors and all of that. When in reality, no one really calls this API with more than one VLAN. It is all unnecessary complexity. And despite appearing pretentious (two-phase transactional model and all), the switchdev API is really sloppy because the VLAN addition and removal operations are not paired with one another (you can add a VLAN 100 times and delete it just once). The bridge notifies through switchdev of a VLAN addition not only when the flags of an existing VLAN change, but also when nothing changes. There are switchdev drivers out there who don't like adding a VLAN that has already been added, and those checks don't really belong at driver level. But the fact that the API contains ranges is yet another factor that prevents this from being addressed in the future. Of the existing switchdev pieces of hardware, it appears that only Mellanox Spectrum supports offloading more than one VLAN at a time, through mlxsw_sp_port_vlan_set. I have kept that code internal to the driver, because there is some more bookkeeping that makes use of it, but I deleted it from the switchdev API. But since the switchdev support for ranges has already been de facto deleted by a Mellanox employee and nobody noticed for 4 years, I'm going to assume it's not a biggie. Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Ido Schimmel <idosch@nvidia.com> # switchdev and mlxsw Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de> # hellcreek Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-01-09 00:01:46 +00:00
break;
net: dsa: realtek-smi: Add Realtek SMI driver This adds a driver core for the Realtek SMI chips and a subdriver for the RTL8366RB. I just added this chip simply because it is all I can test. The code is a massaged variant of the code that has been sitting out-of-tree in OpenWRT for years in the absence of a proper switch subsystem. This creates a DSA driver for it. I have tried to credit the original authors wherever possible. The main changes I've done from the OpenWRT code: - Added an IRQ chip inside the RTL8366RB switch to demux and handle the line state IRQs. - Distributed the phy handling out to the PHY driver. - Added some RTL8366RB code that was missing in the driver at the time, such as setting up "green ethernet" with a funny jam table and forcing MAC5 (the CPU port) into 1 GBit. - Select jam table and add the default jam table from the vendor driver, also for ASIC "version 0" if need be. - Do not store jam tables in the device tree, store them in the driver. - Pick in the "initvals" jam tables from OpenWRT's driver and make those get selected per compatible for the whole system. It's apparently about electrical settings for this system and whatnot, not really configuration from device tree. - Implemented LED control: beware of bugs because there are no LEDs on the device I am using! We do not implement custom DSA tags. This is explained in a comment in the driver as well: this "tagging protocol" is not simply a few extra bytes tagged on to the ethernet frame as DSA is used to. Instead, enabling the CPU tags will make the switch start talking Realtek RRCP internally. For example a simple ping will make this kind of packets appear inside the switch: 0000 ff ff ff ff ff ff bc ae c5 6b a8 3d 88 99 a2 00 0010 08 06 00 01 08 00 06 04 00 01 bc ae c5 6b a8 3d 0020 a9 fe 01 01 00 00 00 00 00 00 a9 fe 01 02 00 00 0030 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 As you can see a custom "8899" tagged packet using the protocol 0xa2. Norm RRCP appears to always have this protocol set to 0x01 according to OpenRRCP. You can also see that this is not a ping packet at all, instead the switch is starting to talk network management issues with the CPU port. So for now custom "tagging" is disabled. This was tested on the D-Link DIR-685 with initramfs and OpenWRT userspaces and works fine on all the LAN ports (lan0 .. lan3). The WAN port is yet not working. Cc: Antti Seppälä <a.seppala@gmail.com> Cc: Roman Yeryomin <roman@advem.lv> Cc: Colin Leitner <colin.leitner@googlemail.com> Cc: Gabor Juhos <juhosg@openwrt.org> Cc: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2018-07-14 09:45:55 +00:00
}
}
return 0;
}
EXPORT_SYMBOL_GPL(rtl8366_vlan_del);
void rtl8366_get_strings(struct dsa_switch *ds, int port, u32 stringset,
uint8_t *data)
{
struct realtek_smi *smi = ds->priv;
struct rtl8366_mib_counter *mib;
int i;
if (port >= smi->num_ports)
return;
for (i = 0; i < smi->num_mib_counters; i++) {
mib = &smi->mib_counters[i];
strncpy(data + i * ETH_GSTRING_LEN,
mib->name, ETH_GSTRING_LEN);
}
}
EXPORT_SYMBOL_GPL(rtl8366_get_strings);
int rtl8366_get_sset_count(struct dsa_switch *ds, int port, int sset)
{
struct realtek_smi *smi = ds->priv;
/* We only support SS_STATS */
if (sset != ETH_SS_STATS)
return 0;
if (port >= smi->num_ports)
return -EINVAL;
return smi->num_mib_counters;
}
EXPORT_SYMBOL_GPL(rtl8366_get_sset_count);
void rtl8366_get_ethtool_stats(struct dsa_switch *ds, int port, uint64_t *data)
{
struct realtek_smi *smi = ds->priv;
int i;
int ret;
if (port >= smi->num_ports)
return;
for (i = 0; i < smi->num_mib_counters; i++) {
struct rtl8366_mib_counter *mib;
u64 mibvalue = 0;
mib = &smi->mib_counters[i];
ret = smi->ops->get_mib_counter(smi, port, mib, &mibvalue);
if (ret) {
dev_err(smi->dev, "error reading MIB counter %s\n",
mib->name);
}
data[i] = mibvalue;
}
}
EXPORT_SYMBOL_GPL(rtl8366_get_ethtool_stats);