Pluggy
December 24, 2022, 5:00pm
1
I have one Plusnet Hub 1 (PH1) as main VDSL router and 2 dumb WAP connected by cable. One is another PH1 and the other is an old TPLink device. I tried to transfer a file via the VDSL router to a PC on the PH1 WAP and it kept crashing. This was all cable. I have a hot spare and I swapped out the main VDSL router for another and it was still a bit flaky. Noticed that I had trouble getting the suspect faulty PH1 to boot normally when testing it.
It would try to boot and then the lights would go out for about 35 seconds when it would then start to boot again. This kept repeating . I have another PH1 (4 in total) and all were running openwrt-22.03.2. This fault occasionally happens on them all so not hardware related.
Searched the web for answers and it seems I am not alone:
opened 06:16PM - 03 May 22 UTC
target/lantiq
bug
release/22.03
I've witnessed an issue on two HH5a with v22.03.0-rc1 where the hubs 'may' some … times take a considerable time to boot up. (**Update Jan 2023:** Issue still exists in latest 22.03.3)
Normally for HH5a, the power light flashes green colour then changes to constant blue light after about 35 seconds with OpenWrt up and running.
However, with 22.03.0-rc1, the sequence starts off with flashing green colour, then the light may go out for considerable time, and the flashing green light appears, before eventually turning solid blue after about 2 minutes.
I reproduced the issue on HH5a which has a serial connection and captured the serial console output.
**Use a pin to perform a hard reset.
After OpenWRt on HH5a boots up, power cycle the router by turning power Off and ON**
Here is partial log:
`
[ 12.978795] random: ubusd: uninitialized urandom read (4 bytes read)
[ 12.987617] random: ubusd: uninitialized urandom read (4 bytes read)
[ 13.004014] procd: - init -
Please press Enter to activate this console.
[ 14.538961] kmodloader: loading kernel modules from /etc/modules.d/*
[ 14.938541] urngd: v1.0.2 started.
[ 15.231591] IFXOS, Version 1.7.1 (c) Copyright 2009, Lantiq Deutschland GmbH
[ 15.243113] NET: Registered protocol family 8
[ 15.246019] NET: Registered protocol family 20
[ 15.260745] PPP generic driver version 2.4.2
[ 15.282942] Lantiq (VRX) DSL CPE MEI driver, version 1.5.17.6, (c) 2007-2015 Lantiq Beteiligungs-GmbH & Co. KG
[ 15.293882]
[ 15.293882]
[ 15.293882] Lantiq CPE API Driver version: DSL CPE API V4.17.18.6
[ 15.311616]
[ 15.311616] Predefined debug level: 3
[ 15.333878] Loading modules backported from Linux version v5.15.33-0-g06f50ca83ace
[ 15.340137] Backport generated by backports.git v5.15.33-1-0-g183c4ab2
[ 15.353225] Infineon Technologies DEU driver version 2.0.0
[ 15.360608] IFX DEU DES initialized (multiblock).
[ 15.367953] IFX DEU AES initialized (multiblock).
[ 15.371762] IFX DEU SHA1 initialized.
[ 15.375306] IFX DEU MD5 initialized.
[ 15.388884] IFX DEU SHA1_HMAC initialized.
[ 15.391978] IFX DEU MD5_HMAC initialized.
[ 15.417049] NET: Registered protocol family 24
[ 15.423093] ath9k_pci_owl_loader 0000:00:0e.0: enabling device (0000 -> 0002)
[ 15.432846] ath9k_pci_owl_loader 0000:00:0e.0: fixup device configuration
[ 36.488592] rcu: INFO: rcu_sched detected stalls on CPUs/tasks:
[ 36.493101] rcu: 1-...!: (1 GPs behind) idle=6d2/1/0x40000000 softirq=759/760 fqs=3
[ 36.500915] (detected by 0, t=2102 jiffies, g=-111, q=26)
[ 36.506399] Sending NMI from CPU 0 to CPUs 1:
ROM VER: 1.1.4
CFG 06
NAND
NAND Read OK
ROM VER: 1.1.4
CFG 06
NAND
NAND Read OK
etc etc
`
**It eventually starts and completes booting of OpenWrt.**
**I power cycled the HH5a again. It booted up normally within 35 seconds**
**I power cycled the HH5a once more, and captured the problem again**
**Here is partial log up to the point where router appears to crash and reboot**
`
Please press Enter to activate this console.
[ 14.492110] kmodloader: loading kernel modules from /etc/modules.d/*
[ 14.923248] urngd: v1.0.2 started.
[ 15.192023] IFXOS, Version 1.7.1 (c) Copyright 2009, Lantiq Deutschland GmbH
[ 15.204926] NET: Registered protocol family 8
[ 15.207833] NET: Registered protocol family 20
[ 15.221231] PPP generic driver version 2.4.2
[ 15.243233] Lantiq (VRX) DSL CPE MEI driver, version 1.5.17.6, (c) 2007-2015 Lantiq Beteiligungs-GmbH & Co. KG
[ 15.254917]
[ 15.254917]
[ 15.254917] Lantiq CPE API Driver version: DSL CPE API V4.17.18.6
[ 15.272597]
[ 15.272597] Predefined debug level: 3
[ 15.293478] Loading modules backported from Linux version v5.15.33-0-g06f50ca83ace
[ 15.299607] Backport generated by backports.git v5.15.33-1-0-g183c4ab2
[ 15.312669] Infineon Technologies DEU driver version 2.0.0
[ 15.319876] IFX DEU DES initialized (multiblock).
[ 15.327406] IFX DEU AES initialized (multiblock).
[ 15.331094] IFX DEU SHA1 initialized.
[ 15.334913] IFX DEU MD5 initialized.
[ 15.338324] IFX DEU SHA1_HMAC initialized.
[ 15.342493] IFX DEU MD5_HMAC initialized.
[ 15.368097] NET: Registered protocol family 24
[ 15.373730] ath9k_pci_owl_loader 0000:00:0e.0: enabling device (0000 -> 0002)
[ 15.383367] ath9k_pci_owl_loader 0000:00:0e.0: fixup device configuration
[ 15.531332] random: crng init done
[ 36.442384] rcu: INFO: rcu_sched detected stalls on CPUs/tasks:
[ 36.446890] rcu: 1-...!: (0 ticks this GP) idle=716/1/0x40000000 softirq=707/707 fqs=0
[ 36.454966] (detected by 0, t=2102 jiffies, g=-135, q=51)
[ 36.460449] Sending NMI from CPU 0 to CPUs 1:
ROM VER: 1.1.4
CFG 06
NAND
NAND Read OK
ROM VER: 1.1.4
CFG 06
NAND
NAND Read OK
etc etc
`
**It eventually starts and completes booting of OpenWrt.**
When I install 21.02.01 on both HH5a, they behave and boot up normally in about 35 seconds when I power cycle the routers.
As suggested in the Github article I have reverted the service PH1 to openwrt-21.02.3 and they work perfectly.
The MAC problem I sorted out. One of the PH1 routers case label MAC address was different from the hardware MAC address. The spurious address is:
30:24:78:E8:DE:BO
Maybe there are more PH1 or HH5A routers using this ? It would only become a problem if like me you are using them as WAP on a network.