mt7621 refactor and mux phy0/4 to gmac1 to achieve 2 Gbps CPU bandwidth #10238

arinc9 · 2022-07-10T15:29:27Z

Hi, I maintain MT7621 SoC and Ralink Pinctrl driver on upstream Linux.
https://github.com/torvalds/linux/blob/4e23eeebb2e57f5a28b36221aa776b5a1122dde5/MAINTAINERS#L17040

This patch series brings thorough refactoring to the devicetrees of the mt7621 SoC and fix GB-PC1 and GB-PC2 devicetrees while at it.

It is possible to achieve 2 Gbps routing in total on devices with MT7621 SoCs.
Currently you can only get a maximum of 500/500 Mbps (download/upload) bandwidth on routing between networks.

With the last patch in the series, 1000/1000 Mbps routing or 2 Gbps total bandwidth to the CPU is possible.

Speed Test

Here are the results for iperf3 bidirectional speed test with hardware offloading.
Routing and NAT performance is identical.

Without the patch

[  5][RX-S]   0.00-20.00  sec   353 MBytes   148 Mbits/sec                  receiver
[  8][TX-S]   0.00-20.00  sec  1.79 GBytes   769 Mbits/sec    0             sender

With the patch

[  5][RX-S]   0.00-20.00  sec  2.16 GBytes   926 Mbits/sec                  receiver
[  8][TX-S]   0.00-20.00  sec  1.80 GBytes   775 Mbits/sec    0             sender

Devices That Benefit 2 Gbps CPU Bandwidth

There's a lot to make a list. Check the changelog for the last patch.

The Patch Series Was Tested On

Unielec U7621-06 16M by me and OpenWrt forum user user31894691
TP-Link RE650 v1 by me
Ubiquiti EdgeRouter X SFP by me
D-Link DIR-882 rev. A1 by OpenWrt forum user Warlock
D-Link DIR-878 rev. A1 by OpenWrt forum user Warlock
MTS WG430223 by @csharper2005
Xiaomi Mi Router 3 Pro by OpenWrt forum user SopaDeMacaco-UmaDeli
Beeline Smartbox Giga by OpenWrt forum user SopaDeMacaco-UmaDeli
Netgear WAC124 by OpenWrt forum user piaddic7ed
Beeline Smartbox Flash by OpenWrt forum user SopaDeMacaco-UmaDeli
Xiaomi Mi Router 4A Gigabit Edition by @ulp1an, @ubatuhan and @dkey21
Xiaomi Redmi AC2100 by @Stkrx
Ubiquiti EdgeRouter X by OpenWrt forum user d687r02j8g
ZyXEL NWA50AX by @ThxAndBye
Unielec U7621-01 by OpenWrt forum user Toomoch
Xiaomi Router 3G by @jokubasver and @eduardo010174
D-Team Newifi D2 by @csharper2005
H3C TX1801 Plus by @DragonBluep
Xiaomi Mi Router CR660x by @infiWang
Cudy X6 by OpenWrt forum user jedboy
Cudy 2100 by OpenWrt forum user jedboy
TP-Link Archer C6U v1 by @dreamer7
GnuBee GB-PC2 by Petr Louda
YouHua WR1200JS by @ayyyuki1
D-Link DIR-1960 by OpenWrt forum user adiastra
Netgear WAX202 by @looi
ipTIME T5004 by @odkrys
TOTOLINK X5000R by OpenWrt forum user Agent
Beeline SmartBox TURBO by @csharper2005
YunCore AX820 by @f00b4r0

Forum link for more information:
https://forum.openwrt.org/t/users-needed-to-test-2-gbps-wan-lan-nat-routing-on-ramips-mt7621-devices/131478

ptpt52 · 2022-07-10T17:23:33Z

mark

neheb · 2022-07-11T01:46:57Z

How interesting...

Do you think this would also work on mt7622?

edit: it won't.

So if I get this right, gmac0 is already connected to the switch. this PR connects gmac1 as well. Doesn't that require multi-CPU DSA?

edit2: reading this PR again looks like port 4 is taken out of the switch and assigned to gmac1. Hrm.

edit3: does raise the question whether it was a good idea for the PC2 to introduce another ethernet interface. After all there are two ethernet ports. If port 4 can be remapped to gmac1, what was the point?

arinc9 · 2022-07-11T05:42:21Z

edit3: does raise the question whether it was a good idea for the PC2 to introduce another ethernet interface. After all there are two ethernet ports. If port 4 can be remapped to gmac1, what was the point?

I asked the same question back in time. Best thing I could come up with is it was a community effort project and I don't really think they knew better. After all, you can connect up to 5 ports on the MT7530 switch in MT7621AT/MT7621ST. There was no need to add an external phy on the board just to add one more ethernet port.

So if I get this right, gmac0 is already connected to the switch. this PR connects gmac1 as well. Doesn't that require multi-CPU DSA?

edit2: reading this PR again looks like port 4 is taken out of the switch and assigned to gmac1. Hrm.

They implemented this feature in the MT7530 DSA subdriver. I'd rather want the phy muxing code to be run from the Mediatek Gigabit PHY driver instead. Maybe we can move the code from the DSA subdriver to there in the future.

arinc9 · 2022-07-11T07:02:34Z

v2:
mt7621_beeline_smartbox-giga.dts: fix phy reg offset
mt7621_linksys_ea6350-v4.dts: fix formatting
mt7621_tplink_eap615-wall-v1.dts: fix gmac1 label

mrkiko · 2022-07-11T14:46:29Z

Thanks for your work.
Has thisPR been already run-tested on a GB-PC1?

Thanks!

arinc9 · 2022-07-11T17:36:38Z

Thanks for your work. Has thisPR been already run-tested on a GB-PC1?

Thanks!

I had Sergio confirm the muxed phy working on mainline Linux on GB-PC1. However, I don't expect to see any issues on OpenWrt. It wouldn't hurt to test my builds if you have one.

target/linux/ramips/dts/mt7621.dtsi

rsalvaterra · 2022-07-13T08:10:38Z

Well, this one flew right under my radar. I will definitely give it a spin on my Redmi AC2100, though. Thanks, @arinc9!

csharper2005 · 2022-07-13T18:19:20Z

Great work. MTS WG430223 - runtests ok.

neheb · 2022-07-20T22:07:58Z

Can you include changes for the WAX202?

arinc9 · 2022-07-21T06:45:08Z

Can you include changes for the WAX202?

Done.

v3:
Include ZyXEL NWA50AX, NWA55AXE and Netgear WAX202.
Slightly change the patch log for better explanation.

neheb · 2022-07-21T06:55:40Z

Great. I'm currently running this patchset with no problems.

ptpt52 · 2022-07-21T07:45:13Z

tested this patch set, looks good.

All pin groups function as gpio unless set otherwise. Therefore, remove this unnecessary binding. Remove the ethernet node workaround for rgmii2. Tested on a range of devices documented on the GitHub PR. Link: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id=d9a683f57b5b4ba9814a49963346cadf562dad2f Link: openwrt#10238 Signed-off-by: Arınç ÜNAL <arinc.unal@arinc9.com>

Mux the MT7530 switch's phy0/4 to the SoC's gmac1 on devices where RGMII2 pins are available. This achieves 2 Gbps total bandwidth to the CPU using the second RGMII. The ports called "wan" are muxed where possible. On a minority of devices, this is not possible. Those cases: mt7621_ampedwireless_ally-r1900k.dts: lan3 mt7621_ubnt_edgerouter-x.dts: eth0 mt7621_gnubee_gb-pc1.dts: ethblue mt7621_linksys_re6500.dts: lan1 mt7621_netgear_wac104.dts: lan4 mt7621_tplink_eap235-wall-v1.dts: lan0 mt7621_tplink_eap615-wall-v1.dts: lan0 mt7621_ubnt_usw-flex.dts: lan1 The "wan" port is just what the vendor designated on the board/plastic chasis of the device. On a technical level, there is no difference between a lan and wan port on MT7621AT, MT7621DAT and MT7621ST SoCs. Prefer connecting to WAN via the port described above for these devices to benefit the feature brought with this patch. Tested on a range of devices documented on the GitHub PR. Link: openwrt#10238 Signed-off-by: Arınç ÜNAL <arinc.unal@arinc9.com>

csharper2005 · 2022-07-21T16:14:49Z

Pending Beeline SmartBox TURBO - runtest ok. The device dts has been updated in accordance with the best practice from this PR.

ulp1an · 2022-07-21T19:29:08Z

Tested on Xiaomi Mi Router 4A Gigabit Edition w/ 1 Gbps fiber connection and it definitely worked. Thank you.

… bandwidth ramips: fix GB-PC1 and GB-PC2 device support Change switch port labels to ethblack & ethblue. Change lan1 & lan2 LEDs to ethblack_act & ethblue_act and fix GPIO pins. Add the external phy with ethyellow label on the GB-PC2 devicetree. Do not claim rgmii2 as gpio, it's used for ethernet with rgmii2 function. Enable ICPlus PHY driver for IP1001 which GB-PC2 has got. Update interface name and change netdev function. Enable lzma compression to make up for the increased size of the kernel. Make spi flash bindings on par with mainline Linux to fix read errors. Tested on GB-PC2 by Petr. Tested-by: Petr Louda <petr.louda@outlook.cz> Signed-off-by: Arınç ÜNAL <arinc.unal@arinc9.com> ramips: mt7621-dts: do not claim rgmii2 group as gpio for certain devices These devices do not use rgmii2 as gpio, therefore remove rgmii2 pin group from state-default. Remove overwriting the ethernet node for these devices. Move claiming the rgmii2 group from mt7621_zyxel_nwa-ax.dtsi to mt7621_zyxel_nwa50ax.dts as it's only the latter using rgmii2 pins as gpio. Remove duplicate ethernet overwrite from mt7621_tplink_archer-x6-v3.dtsi. Claim rgmii2 group as gpio on mt7621_bolt_arion.dts as it uses an rgmii2 pin, 26, as gpio. Signed-off-by: Arınç ÜNAL <arinc.unal@arinc9.com> ramips: mt7621-dts: mux phy0/4 to gmac1 Mux the MT7530 switch's phy0/4 to the SoC's gmac1 on devices where RGMII2 pins are available. This achieves 2 Gbps total bandwidth to the CPU using the second RGMII. The ports called "wan" are muxed where possible. On a minority of devices, this is not possible. Those cases: mt7621_ampedwireless_ally-r1900k.dts: lan3 mt7621_ubnt_edgerouter-x.dts: eth0 mt7621_gnubee_gb-pc1.dts: ethblue mt7621_linksys_re6500.dts: lan1 mt7621_netgear_wac104.dts: lan4 mt7621_tplink_eap235-wall-v1.dts: lan0 mt7621_tplink_eap615-wall-v1.dts: lan0 mt7621_ubnt_usw-flex.dts: lan1 The "wan" port is just what the vendor designated on the board/plastic chasis of the device. On a technical level, there is no difference between a lan and wan port on MT7621AT, MT7621DAT and MT7621ST SoCs. Prefer connecting to WAN via the port described above for these devices to benefit the feature brought with this patch. mt7621_d-team_newifi-d2.dts cannot benefit this feature, although it looks like it should, because the rgmii2 pins are wired to unused components. Tested on a range of devices documented on the GitHub PR. Link: openwrt#10238 Signed-off-by: Arınç ÜNAL <arinc.unal@arinc9.com> ramips: Correct Unielec 01 and 06 dts labels

Ntemis · 2023-06-16T16:13:19Z

You forgot the MQmaker WiTi board as it has 2 wan ports as well but only gmac0 node is set in dts
Check known issues here
https://openwrt.org/toh/mqmaker/witi
With CONFIG_ICPLUS_PHY=y been set i think is time to bring the second wan to life

arinc9 · 2023-06-17T05:17:21Z

@Ntemis I did not forget, I did not have this information to properly address the devicetree for this device. Do you have this device to test WAN2? I will make a patch for this.

arinc9 · 2023-06-17T19:07:56Z

The known issues section on the link claims the external PHY for WAN2 is wired to the switch instead of the SoC's GMAC. I'll need to you test to confirm that.

Ntemis · 2023-06-18T15:19:39Z

i can test and yes i do own the device

arinc9 · 2023-06-18T17:37:37Z

Try the images here. See if the network connectivity using the wan1 and wan2 interfaces works.

Source code: arinc9@d84cb17

Ntemis · 2023-06-19T23:27:46Z

I let you know asap tomorrow, thanks for the image

Ntemis · 2023-06-20T00:03:48Z

@arinc9 you the man! It works perfectly, amazing thanks

arinc9 · 2023-06-20T09:02:59Z

parakallo! I'll send a patch with you stated as a tester soon.

Ntemis · 2023-06-20T13:42:07Z

büyük desteğin için teşekkürler komşu

Mux the MT7530 switch's phy0/4 to the SoC's gmac1 on devices where RGMII2 pins are available. This achieves 2 Gbps total bandwidth to the CPU using the second RGMII. The ports called "wan" are muxed where possible. On a minority of devices, this is not possible. Those cases: mt7621_ampedwireless_ally-r1900k.dts: lan3 mt7621_ubnt_edgerouter-x.dts: eth0 mt7621_gnubee_gb-pc1.dts: ethblue mt7621_linksys_re6500.dts: lan1 mt7621_netgear_wac104.dts: lan4 mt7621_tplink_eap235-wall-v1.dts: lan0 mt7621_tplink_eap615-wall-v1.dts: lan0 mt7621_ubnt_usw-flex.dts: lan1 The "wan" port is just what the vendor designated on the board/plastic chasis of the device. On a technical level, there is no difference between a lan and wan port on MT7621AT, MT7621DAT and MT7621ST SoCs. Prefer connecting to WAN via the port described above for these devices to benefit the feature brought with this patch. mt7621_d-team_newifi-d2.dts cannot benefit this feature, although it looks like it should, because the rgmii2 pins are wired to unused components. Tested on a range of devices documented on the GitHub PR. Link: openwrt#10238 Signed-off-by: Arınç ÜNAL <arinc.unal@arinc9.com>

… bandwidth ramips: fix GB-PC1 and GB-PC2 device support Change switch port labels to ethblack & ethblue. Change lan1 & lan2 LEDs to ethblack_act & ethblue_act and fix GPIO pins. Add the external phy with ethyellow label on the GB-PC2 devicetree. Do not claim rgmii2 as gpio, it's used for ethernet with rgmii2 function. Enable ICPlus PHY driver for IP1001 which GB-PC2 has got. Update interface name and change netdev function. Enable lzma compression to make up for the increased size of the kernel. Make spi flash bindings on par with mainline Linux to fix read errors. Tested on GB-PC2 by Petr. Tested-by: Petr Louda <petr.louda@outlook.cz> Signed-off-by: Arınç ÜNAL <arinc.unal@arinc9.com> ramips: mt7621-dts: do not claim rgmii2 group as gpio for certain devices These devices do not use rgmii2 as gpio, therefore remove rgmii2 pin group from state-default. Remove overwriting the ethernet node for these devices. Move claiming the rgmii2 group from mt7621_zyxel_nwa-ax.dtsi to mt7621_zyxel_nwa50ax.dts as it's only the latter using rgmii2 pins as gpio. Remove duplicate ethernet overwrite from mt7621_tplink_archer-x6-v3.dtsi. Claim rgmii2 group as gpio on mt7621_bolt_arion.dts as it uses an rgmii2 pin, 26, as gpio. Signed-off-by: Arınç ÜNAL <arinc.unal@arinc9.com> ramips: mt7621-dts: mux phy0/4 to gmac1 Mux the MT7530 switch's phy0/4 to the SoC's gmac1 on devices where RGMII2 pins are available. This achieves 2 Gbps total bandwidth to the CPU using the second RGMII. The ports called "wan" are muxed where possible. On a minority of devices, this is not possible. Those cases: mt7621_ampedwireless_ally-r1900k.dts: lan3 mt7621_ubnt_edgerouter-x.dts: eth0 mt7621_gnubee_gb-pc1.dts: ethblue mt7621_linksys_re6500.dts: lan1 mt7621_netgear_wac104.dts: lan4 mt7621_tplink_eap235-wall-v1.dts: lan0 mt7621_tplink_eap615-wall-v1.dts: lan0 mt7621_ubnt_usw-flex.dts: lan1 The "wan" port is just what the vendor designated on the board/plastic chasis of the device. On a technical level, there is no difference between a lan and wan port on MT7621AT, MT7621DAT and MT7621ST SoCs. Prefer connecting to WAN via the port described above for these devices to benefit the feature brought with this patch. mt7621_d-team_newifi-d2.dts cannot benefit this feature, although it looks like it should, because the rgmii2 pins are wired to unused components. Tested on a range of devices documented on the GitHub PR. Link: openwrt#10238 Signed-off-by: Arınç ÜNAL <arinc.unal@arinc9.com> ramips: Correct Unielec 01 and 06 dts labels

sanitariu · 2023-11-15T20:54:35Z

Hi, anyone have problems like double arping from ISP or packet losses when router is connected via epon optical device ?

hauke added the target/ramips pull request/issue for ramips target label Jul 10, 2022

arinc9 force-pushed the mt7621-dts-work branch from d5c168c to 4a1f68e Compare July 11, 2022 07:02

ptpt52 reviewed Jul 11, 2022

View reviewed changes

target/linux/ramips/dts/mt7621.dtsi Outdated Show resolved Hide resolved

SopaDeMacaco-UmaDelicia mentioned this pull request Jul 20, 2022

ramips: Add support for Beeline SmartBox TURBO+ #4108

Closed

arinc9 force-pushed the mt7621-dts-work branch from 4a1f68e to d8ecbda Compare July 21, 2022 06:42

This was referenced Jul 21, 2022

ramips: add support for Linksys E7350 #4470

Closed

ramips: Add support for Sercomm S1500 clones #4770

Closed

neheb mentioned this pull request Jul 21, 2022

ramips: add support for ZyXEL Keenetic Giga III #4587

Open

csharper2005 mentioned this pull request Jul 21, 2022

ramips: add support for Beeline SmartBox TURBO #10118

Closed

Djfe mentioned this pull request Feb 26, 2023

FS#2000 - MT7621 routing performance issue #5976

Open

MrHamel mentioned this pull request Mar 18, 2023

Possible AP Alternative: TP-Link EAP615-Wall socallinuxexpo/scale-network#594

Closed

henielssen mentioned this pull request Oct 16, 2023

[23.05.0-rc4] Ubiquiti EdgeRouter X has increaced cpu load and reduced network speed on eth0 when using it as an all ports Switch compared to 22.03.5 (closed) #13613

Closed

1 task

Npeca75 mentioned this pull request Oct 25, 2023

LLDP & gmac1 & DSA & mt7621 #13788

Open

1 task

InsaneKnight mentioned this pull request Nov 25, 2023

TP-Link TL-WDR4900 v1 have poor routing performance after switching to DSA. #14053

Open

1 task

DragonBluep mentioned this pull request Mar 3, 2024

ramips: Add support for Ruijie RG-EW1200G PRO #14757

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

mt7621 refactor and mux phy0/4 to gmac1 to achieve 2 Gbps CPU bandwidth #10238

mt7621 refactor and mux phy0/4 to gmac1 to achieve 2 Gbps CPU bandwidth #10238

arinc9 commented Jul 10, 2022 •

edited

Loading

ptpt52 commented Jul 10, 2022

neheb commented Jul 11, 2022 •

edited

Loading

arinc9 commented Jul 11, 2022 •

edited

Loading

arinc9 commented Jul 11, 2022

mrkiko commented Jul 11, 2022

arinc9 commented Jul 11, 2022 •

edited

Loading

rsalvaterra commented Jul 13, 2022

csharper2005 commented Jul 13, 2022 •

edited

Loading

neheb commented Jul 20, 2022

arinc9 commented Jul 21, 2022

neheb commented Jul 21, 2022

ptpt52 commented Jul 21, 2022

csharper2005 commented Jul 21, 2022

ulp1an commented Jul 21, 2022

Ntemis commented Jun 16, 2023 •

edited

Loading

arinc9 commented Jun 17, 2023 •

edited

Loading

arinc9 commented Jun 17, 2023

Ntemis commented Jun 18, 2023

arinc9 commented Jun 18, 2023 •

edited

Loading

Ntemis commented Jun 19, 2023

Ntemis commented Jun 20, 2023

arinc9 commented Jun 20, 2023 •

edited

Loading

Ntemis commented Jun 20, 2023

sanitariu commented Nov 15, 2023

mt7621 refactor and mux phy0/4 to gmac1 to achieve 2 Gbps CPU bandwidth #10238

mt7621 refactor and mux phy0/4 to gmac1 to achieve 2 Gbps CPU bandwidth #10238

Conversation

arinc9 commented Jul 10, 2022 • edited Loading

Speed Test

Devices That Benefit 2 Gbps CPU Bandwidth

The Patch Series Was Tested On

ptpt52 commented Jul 10, 2022

neheb commented Jul 11, 2022 • edited Loading

arinc9 commented Jul 11, 2022 • edited Loading

arinc9 commented Jul 11, 2022

mrkiko commented Jul 11, 2022

arinc9 commented Jul 11, 2022 • edited Loading

rsalvaterra commented Jul 13, 2022

csharper2005 commented Jul 13, 2022 • edited Loading

neheb commented Jul 20, 2022

arinc9 commented Jul 21, 2022

neheb commented Jul 21, 2022

ptpt52 commented Jul 21, 2022

csharper2005 commented Jul 21, 2022

ulp1an commented Jul 21, 2022

Ntemis commented Jun 16, 2023 • edited Loading

arinc9 commented Jun 17, 2023 • edited Loading

arinc9 commented Jun 17, 2023

Ntemis commented Jun 18, 2023

arinc9 commented Jun 18, 2023 • edited Loading

Ntemis commented Jun 19, 2023

Ntemis commented Jun 20, 2023

arinc9 commented Jun 20, 2023 • edited Loading

Ntemis commented Jun 20, 2023

sanitariu commented Nov 15, 2023

arinc9 commented Jul 10, 2022 •

edited

Loading

neheb commented Jul 11, 2022 •

edited

Loading

arinc9 commented Jul 11, 2022 •

edited

Loading

arinc9 commented Jul 11, 2022 •

edited

Loading

csharper2005 commented Jul 13, 2022 •

edited

Loading

Ntemis commented Jun 16, 2023 •

edited

Loading

arinc9 commented Jun 17, 2023 •

edited

Loading

arinc9 commented Jun 18, 2023 •

edited

Loading

arinc9 commented Jun 20, 2023 •

edited

Loading