OpenWrt/LEDE Project

  • Status Unconfirmed
  • Percent Complete
    0%
  • Task Type Bug Report
  • Category Base system
  • Assigned To No-one
  • Operating System All
  • Severity High
  • Priority Very Low
  • Reported Version openwrt-19.07
  • Due in Version Undecided
  • Due Date Undecided
  • Private
Attached to Project: OpenWrt/LEDE Project
Opened by eginnc - 01.12.2019

FS#2647 - EA6350v3 (IPQ4018) memory usage climbs until processes are killed and wifi is lost

Device is Linksys EA6350v3

Issue occurs on 19.07 rc1, 19.07 rc2 and snapshot r11595 (snapshot version is by memory, but pretty sure that’s what I had on it).

Information for this bug report is based on 19.07 rc2 with the following packages installed for future potential use, but most not in use:
luci-app-sqm luci-proto-wireguard luci-app-wireguard ca-bundle curl https_dns_proxy luci-app-https_dns_proxy luci-app-statistics luci-app-samba kmod-usb-printer p910nd luci-app-p910nd diffutils minidlna luci-app-minidlna

Note: sqm, wireguard, https_dns_proxy, samba, usb printer, p910nd server, and minidlna are NOT in use. These are just the typical packages I add to my
router for future use.

Steps to reproduce:
Problem occurs after router runs for a few hours. EA6350v3 is set up as a wired access point for an EdgeRouter X (also on 19.07 rc2). The Edgerouter provides sqm and DNS and DHCP for LAN, guest and IOT VLANs. There are ~3 “Guest” 2.4 G wifi clients (a smart switch, an IP camera, a DEEBOT vacuum); the IOT VLAN is mapped to a physical port with an Ooma Telo plugged in (I know - I need to map the guest devices to IOT wifi someday...). The LAN is mapped to a physical port with a Roku 3 plugged in and to normal wifi. There are another ~4 or 5 5GHz wifi clients on the LAN wifi (a laptop, a couple Amazon Echo’s, a Google Home Mini, a Samsung Orbit Android phone) and a 2.4G LG Android phone on the 2.4 GHz wifi. CPU loading is quite low.

Memory usage starts out rather high (~100MB, climbs and climbs, eventually drops, climbs again, repeats. Eventually (a day or two) processes start getting killed to free up memory and things start dying. I have an EA8500 setup almost identically as a second AP in the house, and it’s memory usage sits around 50 MB versus 100MB to 200+MB for the EA6350v3. Something’s just not right...

The final message in the logs that repeats until processes begin to get killed is:

kern.warn kernel: [15897.905286] ath10k_ahb a000000.wifi: failed to increase tx pending count: -16, dropping

But the weird memory usage pattern precedes this message in the logs.

I’ve attached memory usage graphs and system and kernel logs from 19.07 rc2 to illustrate the problem. rc1 and snapshot behaved the same. This post also has logs showing processes getting killed if that helps diagnose: https://forum.openwrt.org/t/ipq4018-linksys-ea6350v3-wifi-dead-after-24-48-hrs/49080/3?u=eginnc

Loading...

Available keyboard shortcuts

Tasklist

Task Details

Task Editing