Re: kern/53814: wm0 device timeout in netbsd 7.1

[Masanobu SAITOH <msaitoh%execsw.org@localhost>]

The following reply was made to PR kern/53814; it has been noted by GNATS.

From: Masanobu SAITOH <msaitoh%execsw.org@localhost>
To: gnats-bugs%NetBSD.org@localhost, kern-bug-people%netbsd.org@localhost,
 gnats-admin%netbsd.org@localhost, netbsd-bugs%netbsd.org@localhost
Cc: msaitoh%execsw.org@localhost
Subject: Re: kern/53814: wm0 device timeout in netbsd 7.1
Date: Thu, 27 Dec 2018 12:38:45 +0900

 On 2018/12/26 15:50, aravind.ss1094%gmail.com@localhost wrote:
 >> Number:         53814
 >> Category:       kern
 >> Synopsis:       wm0 device timeout in netbsd 7.1
 >> Confidential:   no
 >> Severity:       serious
 >> Priority:       medium
 >> Responsible:    kern-bug-people
 >> State:          open
 >> Class:          sw-bug
 >> Submitter-Id:   net
 >> Arrival-Date:   Wed Dec 26 06:50:00 +0000 2018
 >> Originator:     Aravind Mani
 >> Release:        netbsd 7.1
 >> Organization:
 > private organization
 >> Environment:
 > chip type: I354
 >> Description:
 > We use WM_T_I354 chip type.When we reload continuously,we could able to observe device timeout issue. wm_init(),wm_reset() doesn't help to recover from problem state.The only way to recover is to reload the switch.There was no initialization error.
 >>From wm_print_stats() and wm_pkt_stats(),i don't see any value in the registers listed and the packets are not hitting the hardware.
 > wm_reset also didn't help to recover the issue.
 > 
 > Do you need any other output to investigate further?
 > Is there any other way to recover from this issue?.
 > Is there any other fix has been done around this area?.
 > 
 > 
 > 
 >> How-To-Repeat:
 > Reload the switch continuously that runs with NetBSD 7.1.
 >> Fix:
 > 
 
   Are you using modified version of if_wm.c? It has neither wm_print_stats()
 nor wm_pkt_stats().
 
 > Do you need any other output to investigate further?
 
 wm(4) has WM_EVENT_COUNTERS option, so it would be good to
 add "options WM_EVENT_COUNTERS" into your kernel configuration
 file and see vmstat -e.
 
 > Reload the switch continuously that runs with NetBSD 7.1.
 
 It's little hard to know what triggers the problem because
 I don't know what your switch implementation do in the reload.
 
 I have SGMII based C2000 machines. I've not tested on others
 (e.g. KX, PCIe SERDES or GMII). It would be good to check your
 PHY configuration and/or status if your system is not SGMII based.
 
 
 -- 
 -----------------------------------------------
                  SAITOH Masanobu (msaitoh%execsw.org@localhost
                                   msaitoh%netbsd.org@localhost)