ramips: Add support for HiWiFi HC5962 by gigibox · Pull Request #865 · lede-project/source

gigibox · 2017-02-22T12:53:22Z

This patch adds supports for the HiWiFi HC5962(gee4) http://www.hiwifi.com

Short specification:

MT7621AT + MT7612EN + 7603EN
256MB DDR3 RAM
128MB NAND flash
1+3 x 1000M Ethernet
1x USB 2.0 port. 1x USB 3.0 port
UART pad on PCB (JP3: TX, RX, GND, 3.3V)

Flash instruction:

Download lede-ramips-mt7621-hc5962-squashfs-factory.bin
Login as root via SSH on 192.168.199.1 and then copy factory.bin(using wget or nc or...) to /tmp/
use the following commands:
$ mtd write /tmp/lede-ramips-mt7621-hc5962-squashfs-factory.bin firmware
$ mtd erase firmware_backup && reboot
After reboot you should be able to login as root via SSH on 192.168.1.1

Signed-off-by: ZengFei Zhang zhangzengfei@kunteng.org

liudf0716 · 2017-02-22T12:54:45Z

good job

pepe2k

Hello @gigibox,

Thank you for your contribution, please find my comments below.

Please, include also some short description about the device you are adding support for and how to flash LEDE image on it, inside the commit message. You can find many different examples in commit history.

Cheers,
Piotr

pepe2k · 2017-02-22T16:54:00Z

target/linux/ramips/base-files/etc/board.d/01_leds

This will be misleading for the user as this LED is already set for status in diag.sh. Please, drop this line.

pepe2k · 2017-02-22T16:54:59Z

target/linux/ramips/base-files/etc/board.d/02_network

Please, keep boards in alphabetical order here.

pepe2k · 2017-02-22T16:58:20Z

target/linux/ramips/base-files/etc/diag.sh

Please, keep boards in alphabetical order here.

pepe2k · 2017-02-22T17:02:13Z

target/linux/ramips/base-files/etc/board.d/02_network

Please, keep boards in alphabetical order here (hc5*61 should be first).

pepe2k · 2017-02-22T17:11:27Z

target/linux/ramips/dts/HC5962.dts

Please, use only status here.

pepe2k · 2017-02-22T17:20:27Z

target/linux/ramips/dts/HC5962.dts

Remove this new line, please.

pepe2k · 2017-02-22T17:22:15Z

target/linux/ramips/image/mt7621.mk

Please, keep all boards in alphabetical order here.

pepe2k · 2017-02-22T17:22:27Z

target/linux/ramips/image/mt7621.mk

Drop this line, please.

pepe2k · 2017-02-22T17:22:34Z

target/linux/ramips/image/mt7621.mk

Drop this line, please.

pepe2k · 2017-02-22T17:26:18Z

target/linux/ramips/image/mt7621.mk

KiB is a wrong unit here.

mkresin

Please use git commit --amend to add changes to your commit. Use git push --force to overwrite your github repository/branch. The PR will afterwards update automatically.

mkresin · 2017-02-22T17:55:33Z

target/linux/ramips/base-files/etc/board.d/01_leds

Any specific reason why you switch the LED to on here? If not, drop it.

mkresin · 2017-02-22T17:57:57Z

target/linux/ramips/dts/HC5962.dts

+
+#include "mt7621.dtsi"
+
+#include <dt-bindings/input/input.h>


Please include <dt-bindings/gpio/gpio.h> here as well and keep alphabetical order..

Use the GPIO_ACTIVE_LOW and GPIO_ACTIVE_HIGH macros afterwards in stead of 1 and 0 in the gpio parameters.

Check the recent ramips board additions for examples.

mkresin · 2017-02-22T17:59:35Z

target/linux/ramips/dts/HC5962.dts

Drop this node. the spi node is disabled by default in mt7621.dtsi.

mkresin · 2017-02-22T18:02:24Z

target/linux/ramips/dts/HC5962.dts

Something is wrong here. The firmware and the kernel partition are having the same start address. Drop the firmware partition, you are having already the ubi partition.

mkresin · 2017-02-22T18:18:34Z

target/linux/ramips/image/mt7621.mk

Drop this line and add the following instead:

IMAGE_SIZE := $(ralink_default_fw_size_32M) IMAGE/sysupgrade.bin := sysupgrade-tar | append-metadata

That should create the correct image format for sysupgrade.

If the sysupgrade.bin you created is intended for the first installation of LEDE (maybe via bootloader) use the following:

IMAGE_SIZE := $(ralink_default_fw_size_32M) IMAGES += factory.bin IMAGE/sysupgrade.bin := sysupgrade-tar | append-metadata IMAGE/factory.bin := append-kernel | pad-to $$(KERNEL_SIZE) | append-ubi | check-size $$$$(IMAGE_SIZE)

Edit:

You need to add your board to the platform_pre_upgrade() function in target/linux/ramips/base-files/lib/upgrade/platform.sh to actually do the nand specific update.

all corrected.Thanks for you help!

heinzek · 2017-02-22T19:36:57Z

target/linux/ramips/image/mt7621.mk

You add "kmod-usb-ledtrig-usbport" module, but you not defined any led as usbport.

heinzek · 2017-02-22T19:38:39Z

target/linux/ramips/dts/HC5962.dts

Here is unnecessary tab.

heinzek · 2017-02-22T19:38:56Z

target/linux/ramips/dts/HC5962.dts

+			label = "hc5962:white:status";
+			gpios = <&gpio0 6 0>;
+		};	
+


Here is unnecessary tab.

heinzek · 2017-02-22T19:39:42Z

target/linux/ramips/dts/HC5962.dts

Unnecessary tab.

heinzek · 2017-02-22T19:40:20Z

target/linux/ramips/dts/HC5962.dts

Unnecessary tab.

heinzek · 2017-02-22T19:40:39Z

target/linux/ramips/dts/HC5962.dts

Unnecessary tab.

heinzek · 2017-02-22T19:40:56Z

target/linux/ramips/dts/HC5962.dts

Unnecessary tab.

heinzek · 2017-02-22T19:41:05Z

target/linux/ramips/dts/HC5962.dts

Unnecessary tab.

heinzek · 2017-02-22T19:41:20Z

target/linux/ramips/dts/HC5962.dts

Unnecessary tab.

liudf0716 · 2017-02-23T02:09:02Z

good advice

gigibox · 2017-02-23T14:16:17Z

Thanks a lot for the review and comments. ^^

mkresin

Please add a comment with a serial log which shows:

the boot of the OEM firmware
the switch to LEDE from the OEM software
the LEDE boot
sysupgrade lede-ramips-mt7621-hc5962-sysupgrade.bin

This way we can ensure that at least the basic stuff is tested.

mkresin · 2017-02-23T15:51:02Z

target/linux/ramips/base-files/lib/upgrade/platform.sh

 	local board=$(ramips_board_name)

 	case "$board" in
-    	ubnt-erx)


Please restore the former indentation!

mkresin · 2017-02-23T16:00:10Z

target/linux/ramips/base-files/lib/upgrade/platform.sh

I'm quite sure the sysupgrade will fail at this point since your sysupgrade image doesn't start with the uImage magic 0x27051956.

The whole image header validation isn't required for the hc5962 since a validation based on the metadata will be done. Please add a new block for the hc5962 containing:

hc5962) # these boards use metadata images return 0 ;;

mkresin · 2017-02-23T16:00:45Z

target/linux/ramips/base-files/etc/board.d/01_leds

What is the purpose of this blinking LED? I guess this can be removed, right?

Hi.
This device has only one indicator light, So, After the system starts, I hope the Indicator blink white to show system normal operation.

But you are already using status_led in /etc/diag.sh for this. Default behavior of status_led is to blink during system boot up process and stay on when system is booted.

It works like this:

when system boot, its red light blink, after booted, become white light blink.

Please, use then only the white LED for status_led in /etc/diag.sh instead of the red one and don't add anything else in /etc/board.d/01_leds. Using red LED could be misleading as this color is usually used to indicate some error and/or problem.

We have a common/general approach over whole code, for all targets, for status LED and there is no need to reproduce vendor firmware approach here.

Thanks!

Nice. Thanks! That explains a lot! I'll do that.

mkresin · 2017-02-23T17:26:59Z

target/linux/ramips/dts/HC5962.dts

drop this partition. it has still the same start address on flash (0x140000) as the kernel partition. For NAND board you don't need a firmware partition.

Sorry, I just found it submission failed, is now complete.

mkresin · 2017-02-23T17:30:57Z

target/linux/ramips/dts/HC5962.dts

+		reg = <0x340000 0x1E00000>;
+	};
+
+	partition@2140000 {


If this partition is required to revert to the OEM firmware, mark the partition as read-only. If this partition is not required, drop the node and use the extra space for the ubi partition.

gigibox · 2017-02-24T09:42:21Z

The-boot-of-the-OEM-firmware.txt
switch-to-LEDE-from-OEM-software.txt
do-sysupgrade-lede-sysupgrade-bin.txt

mkresin · 2017-02-26T21:07:06Z

With your latest changes you broke the sysupgrade again. Have you tested the sysupgrade with your latest changes?

You dropped the following requested changes with your last push:

It is not necessary to do the nand_do_platform_check for the hc5962, since the validation is done via metadata.

It was fine the way it was before your last change. You have just added to much in indentation.

gigibox · 2017-02-27T05:02:41Z

@mkresin Please forgive my ignorance. I have changed it as you suggested and has been tested. Thank you again for you time on this matter.

hc5962-lede-test-sysupgrade.txt

mkresin · 2017-02-27T21:36:35Z

Look really good to me. One last remark.

In the ethernet node of the DTS you read the mac address as hex values from the factory partition. But later in the boot process you replace this mac address with the ASCII values read from the bdinfo partition (target/linux/ramips/base-files/etc/board.d/02_network). Either of them seam to be wrong to me. Any explanation for this?

In your bootlog the following can be seen:

[   10.100000] mt76x2e 0000:02:00.0: ASIC revision: 76120044
[   10.100000] mt76x2e 0000:02:00.0: EEPROM data check failed: 7603
[   10.110000] mt76x2e 0000:02:00.0: Invalid MAC address, using random address 7e:dd:3f:8b:49:e8
[   10.130000] mt76x2e 0000:02:00.0: ROM patch already applied
[   10.130000] mt76x2e 0000:02:00.0: Firmware Version: 0.0.00
[   10.140000] mt76x2e 0000:02:00.0: Build: 1
[   10.140000] mt76x2e 0000:02:00.0: Build Time: 201507311614____
[   10.170000] mt76x2e 0000:02:00.0: Firmware running!
[   10.180000] mt7603e 0000:01:00.0: ASIC revision: 76030010
[   12.260000] mt7603e 0000:01:00.0: Firmware Version: _mt7603mp
[   12.270000] mt7603e 0000:01:00.0: Build Time: 20151020155902

Not sure about this, but could it be that something is wrong with the EEPROM of the mt76@1,0 (0000:02:00.0) wireless? Maybe you have defined the wrong position on flash?

This patch adds supports for the HiWiFi HC5962(gee4) http://www.hiwifi.com Short specification: - MT7621AT + MT7612EN + 7603EN - 256MB DDR3 RAM - 128MB NAND flash - 1+3 x 1000M Ethernet - 1x USB 2.0 port. 1x USB 3.0 port. - reset button - UART pad on PCB (JP3: TX, RX, GND, 3.3V) Flash instruction: 1, Download lede-ramips-mt7621-hc5962-squashfs-factory.bin 2, Login as root via SSH on 192.168.199.1 and then copy factory.bin(using wget or nc or...) to /tmp/ 3, use the following commands: $ mtd write /tmp/lede-ramips-mt7621-hc5962-squashfs-factory.bin firmware $ mtd erase firmware_backup && reboot After reboot you should be able to login as root via SSH on 192.168.1.1 Signed-off-by: ZengFei Zhang <zhangzengfei@kunteng.org>

gigibox · 2017-02-28T04:57:35Z

Yes, mkresin, you are right, I adjusted the reading position of the eeprom, Now it works very well, Signal is stable and reliable.

In addition, HiWiFi use bdinfo parition to save MAC address, The ethernet node of the DTS have since been deleted.

Thanks!

[    9.990000] mt7603e 0000:01:00.0: ASIC revision: 76030010
[   12.040000] mt7603e 0000:01:00.0: Firmware Version: _mt7603mp
[   12.040000] mt7603e 0000:01:00.0: Build Time: 20151020155902
[   12.090000] firmware init done
[   12.260000] mt76x2e 0000:02:00.0: ASIC revision: 76120044
[   12.280000] mt76x2e 0000:02:00.0: ROM patch already applied
[   12.340000] mt76x2e 0000:02:00.0: Firmware Version: 0.0.00
[   12.350000] mt76x2e 0000:02:00.0: Build: 1
[   12.350000] mt76x2e 0000:02:00.0: Build Time: 201507311614____
[   12.380000] mt76x2e 0000:02:00.0: Firmware running!

Full log:
hc5962-lede-upgrade-bootlog.txt

liudf0716 · 2017-02-28T05:29:49Z

@mkresin I have submitted a pull request for update libevent days ago, but why no member reviewed my PR, either no member gave me advise or suggestion. here is my PR
#848

mkresin · 2017-02-28T22:47:03Z

Merged! Thanks for the PR and thanks a lot to all reviewers.

SouHong · 2017-08-26T02:22:56Z

HC5962.dts-ok.tar.gz

#for usr 2.0:

gpio_export {
	compatible = "gpio-export";
	#size-cells = <0>;

	usbpower {
		gpio-export,name = "usbpower";
		gpio-export,output = <0>;
		gpios = <&gpio0 12 0>;
	};
};

pepe2k requested changes Feb 22, 2017

View reviewed changes

mkresin requested changes Feb 22, 2017

View reviewed changes

heinzek reviewed Feb 22, 2017

View reviewed changes

mkresin requested changes Feb 23, 2017

View reviewed changes

mkresin approved these changes Feb 27, 2017

View reviewed changes

pepe2k approved these changes Feb 27, 2017

View reviewed changes

mkresin self-assigned this Feb 28, 2017

mkresin closed this Feb 28, 2017

gigibox deleted the patch-hc5962 branch March 1, 2017 02:11

pepe2k mentioned this pull request Mar 10, 2017

ramips: add support for Netgear EX3700 #937

Closed


		#include "mt7621.dtsi"

		#include <dt-bindings/input/input.h>

Uh oh!

Conversation

gigibox commented Feb 22, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

liudf0716 commented Feb 22, 2017

Uh oh!

pepe2k left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

mkresin left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

mkresin Feb 22, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

liudf0716 commented Feb 23, 2017

Uh oh!

gigibox commented Feb 23, 2017

Uh oh!

mkresin left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

gigibox commented Feb 22, 2017 •

edited

Loading

mkresin Feb 22, 2017 •

edited

Loading

gigibox commented Feb 24, 2017 •

edited

Loading

mkresin commented Feb 26, 2017 •

edited

Loading