[device/celestica]: Add support for xcvrd daemon in silverstone #91

mudsut4ke · 2020-10-28T10:45:55Z

- What I did

Add function to sfputil support xcvrd daemon

- How I did it

Add get_transceiver_change_event function to support xcvrd daemon
Added a workaround command to make the link go up.

- How to verify it

Verify rsyslog when sfp plug / unplug event occurs.
- xcvrd_event.log
Verify port link status when event occurs
- xcvr_even_and_port_statust.log
AOI module log :
- AOI-xcvrd.log

stevenlu99 · 2020-10-28T18:51:59Z

at function _bring_up_port_link
It seems does not initialize CMIS-4.0 init sequence, instead it only put port in loopback. this will link up but would not able to pass traffic.

Also, it does not change back to page-0 at end of function

mudsut4ke · 2020-10-29T12:51:43Z

at function _bring_up_port_link
It seems does not initialize CMIS-4.0 init sequence, instead it only put port in loopback. this will link up but would not able to pass traffic.

Also, it does not change back to page-0 at end of function

@stevenlu99 , Please review a new init script in 9f33f91

zhenggen-xu · 2020-10-30T04:52:50Z

device/celestica/x86_64-cel_silverstone-r0/led_proc_init.soc

@@ -1,5 +1,8 @@
-#The Port LED of Silverstone SONIC can't work well, after the issue is fixed by SAI, The led will start. 
-#led auto on; led start
+# Enable all ports


why we enabled all ports by default? This should be controlled by configDB data.

I will remove it (this one should added on our debug image only)

zhenggen-xu · 2020-10-30T04:53:05Z

device/celestica/x86_64-cel_silverstone-r0/led_proc_init.soc

+port all en=1
+sleep 6
+
+# The Port LED of Silverstone SONIC can't work well, after the issue is fixed by SAI, The led will start.


This comment should be removed.

zhenggen-xu · 2020-10-30T05:04:01Z

device/celestica/x86_64-cel_silverstone-r0/Silverstone-128x100/cmis-init.sh

+#!/bin/bash
+
+
+# Configure Innolight DR4 transceiver module into 4x100G breakout mode


We need make this script works for multiple optical vendors mentioned in email.

zhenggen-xu · 2020-10-30T05:04:18Z

device/celestica/x86_64-cel_silverstone-r0/Silverstone-128x100/cmis-init.sh

+# so that each 100G ports can be operated independently
+
+# Usage:
+# ./cmis4_initial.sh $port_i2c_no


cmis4_init.sh

zhenggen-xu · 2020-10-30T05:04:44Z

device/celestica/x86_64-cel_silverstone-r0/Silverstone/cmis-init.sh

+#!/bin/bash
+
+
+# Configure Innolight DR4 transceiver module into 4x100G breakout mode


This HWSKU should use 1x400G?

zhenggen-xu · 2020-10-30T05:20:12Z

device/celestica/x86_64-cel_silverstone-r0/plugins/sfputil.py

+        (platform, hwsku) = self._get_platform_and_hwsku()
+        platform_path = "/".join([PLATFORM_ROOT_PATH, platform])
+        hwsku_path = "/".join([platform_path, hwsku])
+        port_config_file_path = "/".join([hwsku_path, "cmis-init.sh"])


change the name of port_config_file_path to cmis_init_file_path?

zhenggen-xu · 2020-10-30T05:27:41Z

device/celestica/x86_64-cel_silverstone-r0/plugins/sfputil.py

    EEPROM_OFFSET = 9
    PORT_INFO_PATH = '/sys/class/silverstone_fpga'
    QSFP_DD_DOM_OFFSET = 2304

+    POLL_INTERVAL = 1


Is this seconds or miliseconds, please make a comment.

zhenggen-xu · 2020-10-30T05:28:47Z

device/celestica/x86_64-cel_silverstone-r0/plugins/sfputil.py

+        int_sfp = {}
+
+        sleep_time = min(
+            timeout, self.POLL_INTERVAL) if timeout != 0 else self.POLL_INTERVAL


timeout is milliseconds, what is POLL_INTERVAL unit? If it is seconds, this min() parameters do not seem to be correct.

platform/broadcom/sonic-platform-modules-cel/silverstone/modules/switchboard.c

zhenggen-xu · 2020-10-30T05:35:34Z

device/celestica/x86_64-cel_silverstone-r0/plugins/sfputil.py

+            status = False
+        return status, result
+
+    def _bring_up_port_link(self, int_sfp, init_script):


This should be renamed to something related to initializing the optical modules.

…of silverstone

stevenlu99 · 2020-10-30T16:36:51Z

@mudsut4ke if you use cmis_init.sh to initialize cmis-4.0 transceivers to 4x100G mode, please keep in mind that this script was originally programmed to demonstrate cmis-4.0 init sequence for Innolight, for different vendors, the sequences might have differences.
Another 2 cents:

please be careful with sleep(s) in the script. it need fine tuned
How do you plan to check cmis-4.0 init was successes? In case of init failed, do you plan add retry?

mudsut4ke · 2020-11-03T03:55:03Z

@stevenlu99 Thanks for the feedback.
I will try to add another command to check port status and add retry for failed case.

zhenggen-xu · 2020-11-05T04:31:07Z

Please include the test results for at least Innolight and Eoptolink optics, and the ports should come up with 4x100G mode.

zhenggen-xu · 2020-11-05T04:22:36Z

device/celestica/x86_64-cel_silverstone-r0/plugins/sfputil.py

+            print "Error: unable to read file: %s" % str(e)
+        return None
+
+    def _is_port_device_present(self, port_idx):


make this function return true or false, not a number that could be anything.

mudsut4ke · 2020-11-18T08:46:50Z

@stevenlu99 , @zhenggen-xu

I have updated code follow your comment and I also attached log to PR description

zhenggen-xu · 2020-11-18T19:55:09Z

We mentioned many times in email, we are looking for 4x100G mode of the optics (Both Innolight and Eoptolink), I don't see that in the log, it seems to be a wrong HWSKU and wrong optical modes.

shivanangi · 2020-11-25T05:17:52Z

Hello Zhenggen,

I see a parallel PR #92 running. Do you still need the qualification on Innodisk and Eptolink?

Thanks.

mudsut4ke · 2020-11-25T07:33:35Z

We mentioned many times in email, we are looking for 4x100G mode of the optics (Both Innolight and Eoptolink), I don't see that in the log, it seems to be a wrong HWSKU and wrong optical modes.

@zhenggen-xu , We encountered an error (port link down) in 4x100G mode test with 1x400G port connected to 4x100G port and our team tried to figure out the root cause of this link down issue.

However, does the test result you want use the same port link topology as above?

And I also have one more question about #92,
Do I need to rebase sfputil.py code to use that modified version from Broadcom ?

zhenggen-xu · 2020-11-25T19:02:12Z

We mentioned many times in email, we are looking for 4x100G mode of the optics (Both Innolight and Eoptolink), I don't see that in the log, it seems to be a wrong HWSKU and wrong optical modes.

@zhenggen-xu , We encountered an error (port link down) in 4x100G mode test with 1x400G port connected to 4x100G port and our team tried to figure out the root cause of this link down issue.

However, does the test result you want use the same port link topology as above?

And I also have one more question about #92,
Do I need to rebase sfputil.py code to use that modified version from Broadcom ?

I suggest you work with BRCM to figure out the solution going forward. As for the use case, we wanted 4x100G mode, and at least Innolight and Eoptolink, both with CMIS 4.0.

…te submodules (sonic-net#4852) * src/sonic-platform-common 75698a8...82bbeab (9): > [sfputil] Make SfpUtilHelper.get_physical_to_logical noexcept as in SfpUtilBase (sonic-net#96) > [sfp_base] Update return value documentation of channel-specific methods (sonic-net#98) > [sfp] Tweak key names of some transceiver info fields (sonic-net#97) > fix typo: portconfig.ini to port_config.ini (#94) > [chassis_base] Add platform API support for system LED (#91) > Add PCIe check commad (#64) > [sfputilbase.py] Don't try to print EEPROM sysfs file name if we failed to read from it (#81) > [sfputilbase | sfputilhelper] Add support of platform.json (#72) > [eeprom] Add try-except to catch the IOError (#85) * src/sonic-platform-daemons 0f4fd83...abe115e (2): > [xcvrd] Tweak some transceiver info key names (#62) > [psud][thermalctld] Always get fan/PSU LED status from platform API to avoid status inconsistencies (#59) * src/sonic-utilities fd7781b...16a33f2 (9): > [config] Fix syntax error (sonic-net#966) > [config] Fix indentation level in _get_disabled_services_list() (sonic-net#965) > a4e64d1 [sonic_installer] Refactor sonic_installer code (sonic-net#953) > 90efd62 [Show | Command Reference] Add Port breakout Show Command (sonic-net#859) > [sfpshow][mock_state_db] Tweak key names of some transceiver info fields (sonic-net#958) > [show] Add missing verbose option to "show line" (sonic-net#961) > [filter-fdb] Check VLAN Presence When Filter FDB (sonic-net#957) > [master]fix sonic-net#4716 show ipv6 interfaces neighbor_ip is N/A issue (sonic-net#948) > Fix for command. show interface transceiver eeprom -d Ethernet (sonic-net#955) Note: sonic-utilities update fixes sonic-net#4716

- sonic-net/sonic-platform-daemons@1893c40 - Fix the xcvrd theowing error on sfprecover function on getKeys() not valid. - sonic-net/sonic-platform-daemons@65fa443 - Merge pull request #90 from abdosi/multiasic-fix - sonic-net/sonic-platform-daemons@7f812c9 - [xcvrd] Don't log unnecessary messages upon empty transceiver change event (#53) - sonic-net/sonic-platform-daemons@0969202 - Fix pcied daemon failure (#91) Signed-off-by: Petro Bratash <petrox.bratash@intel.com>

zhenggen-xu reviewed Oct 30, 2020

View reviewed changes

mudsut4ke force-pushed the 201811-silverstone-sfputil-xcvrd branch from 37b59e1 to 8f53880 Compare October 30, 2020 07:02

Wirut Getbamrung added 4 commits October 30, 2020 14:39

[platform/cel-silverstone]: Add sysfs for XCVR interrupt event status

f98ba0e

[device/celestica]: implement get_transceiver_change_event in sfutil …

e9f3fb1

…of silverstone

[device/silverstone]: add cmis init script

3d1b648

[device/celestica-silverstone]: update perssion for cmis-init script

8400496

mudsut4ke force-pushed the 201811-silverstone-sfputil-xcvrd branch from 8f53880 to 8400496 Compare October 30, 2020 07:39

[device/celestica-silverstone]: clean cmis init code

1c4e42f

mudsut4ke force-pushed the 201811-silverstone-sfputil-xcvrd branch from 8626456 to 1c4e42f Compare October 30, 2020 08:01

[device/celestica-silverstone]: clean sfputil code

3e5c896

mudsut4ke marked this pull request as ready for review November 5, 2020 03:36

zhenggen-xu reviewed Nov 5, 2020

View reviewed changes

Wirut Getbamrung added 2 commits November 17, 2020 18:19

[device/celestica]: update retry condition for cmis init

1f49835

[device/celestica]: add error detection to cmis init script

4f8dbca

[device/celestica]: update script to avoid error

3614955

Wirut Getbamrung added 3 commits January 12, 2021 10:03

update code follow an update from brcm

504e2a1

clean sfputil

0226ded

remove unused files

d6d0e5c

remove unused files2

5241e6a

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[device/celestica]: Add support for xcvrd daemon in silverstone #91

[device/celestica]: Add support for xcvrd daemon in silverstone #91

mudsut4ke commented Oct 28, 2020 •

edited

Loading

stevenlu99 commented Oct 28, 2020 •

edited

Loading

mudsut4ke commented Oct 29, 2020

zhenggen-xu Oct 30, 2020

mudsut4ke Oct 30, 2020

zhenggen-xu Oct 30, 2020

zhenggen-xu Oct 30, 2020

zhenggen-xu Oct 30, 2020

zhenggen-xu Oct 30, 2020

zhenggen-xu Oct 30, 2020

zhenggen-xu Oct 30, 2020

mudsut4ke Oct 30, 2020

zhenggen-xu Oct 30, 2020

zhenggen-xu Oct 30, 2020

stevenlu99 commented Oct 30, 2020

mudsut4ke commented Nov 3, 2020

zhenggen-xu commented Nov 5, 2020

zhenggen-xu Nov 5, 2020

mudsut4ke commented Nov 18, 2020

zhenggen-xu commented Nov 18, 2020

shivanangi commented Nov 25, 2020

mudsut4ke commented Nov 25, 2020 •

edited

Loading

zhenggen-xu commented Nov 25, 2020

		#!/bin/bash


		# Configure Innolight DR4 transceiver module into 4x100G breakout mode

[device/celestica]: Add support for xcvrd daemon in silverstone #91

Are you sure you want to change the base?

[device/celestica]: Add support for xcvrd daemon in silverstone #91

Conversation

mudsut4ke commented Oct 28, 2020 • edited Loading

stevenlu99 commented Oct 28, 2020 • edited Loading

mudsut4ke commented Oct 29, 2020

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

stevenlu99 commented Oct 30, 2020

mudsut4ke commented Nov 3, 2020

zhenggen-xu commented Nov 5, 2020

Choose a reason for hiding this comment

mudsut4ke commented Nov 18, 2020

zhenggen-xu commented Nov 18, 2020

shivanangi commented Nov 25, 2020

mudsut4ke commented Nov 25, 2020 • edited Loading

zhenggen-xu commented Nov 25, 2020

mudsut4ke commented Oct 28, 2020 •

edited

Loading

stevenlu99 commented Oct 28, 2020 •

edited

Loading

mudsut4ke commented Nov 25, 2020 •

edited

Loading