[as9726_32d] Add new platform by acctalex · Pull Request #883 · opencomputeproject/OpenNetworkLinux

acctalex · 2022-04-19T09:06:43Z

CPU: Intel Broadwell-DE D-1519 4 cores
RAM: DDR4 8GB x2 SO-DIMM (Expandable up to 16GBx2, w/i ECC)
SSD: M.2 MLC 50GB x 1
PHY: Marvell 88E1111
CPLD: Altera MAX 10 10M04DAU324C8G X1
CPLD: Altera Cyclone IV EP4CGX30CF23C7N X1
CPLD: Altera 5M1270ZF256C5N x1
Ethernet Port: 400G QSFP56-DD port x 32 and 10G SFP+ x 2
PSU: 1+1 Redundant/hot swappable 1500W AC/DC PSU (C14 ac inlet)
Cooling: 5+1 fan-tray modules with 40mm x40mm x 56mm 12V fans, hotswappable
Dimension: 17.26” x 23.23” x 1U (438.4x590x43.5mm)
Linux kernel version: 5.4, due to customer requirements

Signed-off-by: Alex Lai alex_lai@edge-core.com

CPU: Intel Broadwell-DE D-1519 4 cores RAM: DDR4 8GB x2 SO-DIMM (Expandable up to 16GBx2, w/i ECC) SSD: M.2 MLC 50GB x 1 PHY: Marvell 88E1111 CPLD: Altera MAX 10 10M04DAU324C8G X1 CPLD: Altera Cyclone IV EP4CGX30CF23C7N X1 CPLD: Altera 5M1270ZF256C5N x1 Ethernet Port: 400G QSFP56-DD port x 32 and 10G SFP+ x 2 PSU: 1+1 Redundant/hot swappable 1500W AC/DC PSU (C14 ac inlet) Cooling: 5+1 fan-tray modules with 40mm x40mm x 56mm 12V fans, hotswappable Dimension: 17.26” x 23.23” x 1U (438.4x590x43.5mm) Signed-off-by: Alex Lai <alex_lai@edge-core.com>

paulmenzel

Thank you for adding the device. Please mention the Linux kernel version in the commit message.

Did you send the Linux kernel patches upstream for review?

paulmenzel · 2022-04-19T09:40:08Z

+{LED_DIAG, LED_MODE_OFF,   ONLP_LED_MODE_OFF},
+{LED_DIAG, LED_MODE_GREEN, ONLP_LED_MODE_GREEN},
+{LED_DIAG, LED_MODE_AMBER, ONLP_LED_MODE_ORANGE},
+{LED_DIAG, LED_MODE_GREEN_BLINK, ONLP_LED_MODE_GREEN_BLINKING},
+{LED_LOC,  LED_MODE_OFF,   ONLP_LED_MODE_OFF},
+{LED_LOC,  LED_MODE_AMBER,   ONLP_LED_MODE_ORANGE},
+{LED_LOC,  LED_MODE_AMBER_BLINK,  ONLP_LED_MODE_ORANGE_BLINKING},
+{LED_FAN,  LED_MODE_AUTO,  ONLP_LED_MODE_AUTO},
+{LED_PSU1, LED_MODE_AUTO,  ONLP_LED_MODE_AUTO},
+{LED_PSU2, LED_MODE_AUTO,  ONLP_LED_MODE_AUTO}


Incorrect indentation?

paulmenzel · 2022-04-19T09:41:24Z

+
+		if (unlikely(result != data_len)) {
+			result = -EIO;
+			msleep(10);


Why such a high delay?

10 ms delay is not so high here. I think.

Everything about 1 ms is high, and should be justified.

paulmenzel · 2022-04-19T09:41:44Z

+		int status;
+		int power_good = 0;
+
+		dev_dbg(&client->dev, "Starting as9726_32d update\n");


Why update?

in order to return the most up to date value, an update is needed.

But what is updated? Please improve the debug message.

paulmenzel · 2022-04-19T09:42:04Z

+
+		if (status < 0) {
+			dev_dbg(&client->dev, "cpld reg 0x60 err %d\n", 
+				status);


Fits on line above.

paulmenzel · 2022-04-19T09:43:27Z

+		if (status < 0) {
+			dev_dbg(&client->dev, "cpld reg 0x60 err %d\n", 
+				status);
+		} else {
+			data->status = status;
+		}


No curly braces if all branches only consist of one single line.

paulmenzel · 2022-04-19T09:44:04Z

@@ -0,0 +1,6 @@
+KERNELS := onl-kernel-4.19-lts-x86-64-all:amd64


Why not the latest Linux kernel version?

our customers need the 4.19 kernel.

Please mention that in the commit message. (I am very uneasy about the fact, that users rely on such old versions, even if they are LTS series. For ONL the latest Linux kernel version should always be usable.)

…onality, and to use correct settings. 1. removed the old project name in fan.c. 2. removed unnecessary mutex_unlock() in fan.c. 3. added AIM_FREE_IF_PTR(mn) in platform_lib.c to prevent memory leak. 4. added support of Belpower psu. 5. added support of transciever LP mode sysfs and onlpdump. 6. modified the fan policy. 7. modified to use the fan air flow direction to detemine which set of thermal thersholds should be used. 8. modified to use the correct console port setting. 9. removed IR3570 related codes in __init__.py 10. modified some codes to fit the linux coding style. Signed-off-by: Alex Lai <alex_lai@edge-core.com>

1. used kernel 5.4 2. removed msleep(10) when failed to read correct PSU info. Signed-off-by: Alex Lai <alex_lai@edge-core.com>

1. initialize pca9548 idle_state in kernel 5.4.40 version, or the data read by the transceivers will be wrong. Signed-off-by: Alex Lai <alex_lai@edge-core.com>

1. Modify PSU related files to support new PSU model:FSJ035-610G and FSJ036-610G. 2. Modify to show correct fan direction for FSJ001, FSJ004, SPAACTN-01, and SPAACTN-02 3. Modify to show the current PSU is of AC or DC. 4. Modify to let vout, vin, iout, iin, pout, and pin return 0, if the PSU is present but unplugged. Signed-off-by: Alex Lai <alex_lai@edge-core.com>

1. In onlpdump, modified to read the fan direction of each psu and show, instead of one reffering to other. Signed-off-by: Alex Lai <alex_lai@edge-core.com>

1. used the comunity master branch's x86_64-all.config of kernel 5.4. Signed-off-by: Alex Lai <alex_lai@edge-core.com>

paulmenzel · 2024-01-17T07:56:29Z

Please mention the used Linux kernel version in the commit message description of commit [as9726_32d] Add new platform, and say why. The latest LTS version is 6.6.

acctalex · 2024-01-24T09:32:29Z

Please mention the used Linux kernel version in the commit message description of commit [as9726_32d] Add new platform, and say why. The latest LTS version is 6.6.

OK, done. Please review, thanks.

jeff97830133 · 2024-02-20T08:03:10Z

Hello @paulmenzel

Please help review this PR.
Thanks!!

1. Use the eeprom at 0x56 on the mainbaord, and the eeprom at 0x57 on the cpu board will no longer be used. Signed-off-by: Alex Lai <alex_lai@accton.com>

Signed-off-by: Roger Ho <roger530_ho@accton.com>

1. set/get the QSFP eeprom tx-disable field via onlp_sfp_control_set() and onlp_sfp_control_get(), respectively, to control the tx-disable functionality Signed-off-by: Alex Lai <alex_lai@accton.com>

…-disable 1. by default, checksum is done on page 0 in sff.c, so after accessing the apis for qsfp-dd tx-disable, page 0 must be set to upper memory. Signed-off-by: Alex Lai <alex_lai@accton.com>

paulmenzel suggested changes Apr 19, 2022

View reviewed changes

acctalex added 5 commits May 4, 2022 08:22

[accton/as9726_32d] used kernel 5.4 and corrected the PSU's delay

bdb39bc

1. used kernel 5.4 2. removed msleep(10) when failed to read correct PSU info. Signed-off-by: Alex Lai <alex_lai@edge-core.com>

[accton/as9726_32d] initialize pca9548 idle_state

8cf5492

1. initialize pca9548 idle_state in kernel 5.4.40 version, or the data read by the transceivers will be wrong. Signed-off-by: Alex Lai <alex_lai@edge-core.com>

Merge remote-tracking branch 'upstream/master' into as9726_32d

f08e7b0

Merge remote-tracking branch 'upstream/master' into as9726_32d

edf7b15

acctalex force-pushed the as9726_32d branch 3 times, most recently from 83e60f3 to edf7b15 Compare August 2, 2023 07:59

acctalex added 4 commits August 2, 2023 08:36

[Edgecore][as9726-32d] Modified PSU fan dir. detection method in onlpd

f2f24c3

1. In onlpdump, modified to read the fan direction of each psu and show, instead of one reffering to other. Signed-off-by: Alex Lai <alex_lai@edge-core.com>

[Edgecore][as9726-32d] changed the x86_64-all.config of kernel 5.4

5ac3975

1. used the comunity master branch's x86_64-all.config of kernel 5.4. Signed-off-by: Alex Lai <alex_lai@edge-core.com>

Merge remote-tracking branch 'upstream/master' into as9726_32d

b53e5e1

acctalex requested a review from paulmenzel January 16, 2024 03:48

acctalex added 2 commits September 25, 2024 01:49

Merge remote-tracking branch 'upstream/master' into as9726_32d

de8eccc

[Accton][as9726-32d] Use the eeprom on the mainboard

5ea910b

1. Use the eeprom at 0x56 on the mainbaord, and the eeprom at 0x57 on the cpu board will no longer be used. Signed-off-by: Alex Lai <alex_lai@accton.com>

roger530-ho force-pushed the as9726_32d branch from f7b532a to 72f3533 Compare June 3, 2025 03:38

[Accton][as9726-32d] update thermal policy/fan controller

cc70dd7

Signed-off-by: Roger Ho <roger530_ho@accton.com>

roger530-ho force-pushed the as9726_32d branch from 72f3533 to cc70dd7 Compare June 3, 2025 03:44

acctalex added 2 commits July 24, 2025 09:09

[Accton][as9726-32d] Add set/get QSFP tx-disable functionality

51a9713

1. set/get the QSFP eeprom tx-disable field via onlp_sfp_control_set() and onlp_sfp_control_get(), respectively, to control the tx-disable functionality Signed-off-by: Alex Lai <alex_lai@accton.com>

[Accton][as9726-32d] Set upper mem to page 0 after set/get QSFP-DD tx…

7f0b272

…-disable 1. by default, checksum is done on page 0 in sff.c, so after accessing the apis for qsfp-dd tx-disable, page 0 must be set to upper memory. Signed-off-by: Alex Lai <alex_lai@accton.com>

This was referenced Jun 2, 2026

accton/as9726-32d: support linux 4.19 accton/OpenNetworkLinux#175

Closed

accton/as9726-32d: support linux 4.19 accton/OpenNetworkLinux#177

Merged

		@@ -0,0 +1,6 @@
		KERNELS := onl-kernel-4.19-lts-x86-64-all:amd64

Conversation

acctalex commented Apr 19, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

paulmenzel 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!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

paulmenzel commented Jan 17, 2024

Uh oh!

acctalex commented Jan 24, 2024

Uh oh!

jeff97830133 commented Feb 20, 2024

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

acctalex commented Apr 19, 2022 •

edited

Loading