[mobian][bookworm] Data connection drops and modem disappears ocassionally

[fernando-c-m]

Hello,
First of all a big thank you for your hard work providing us with an open source userspace for the pinephone modem.
I am facing some problems with r0.4.1, mobian, bookworm stable and 5.14 kernel

• Data disconnects and won't come back
• Modem disappears (but with 5.14 seems to be coming back after a while)
  Then everything is OK.
  I am aware that this could be also a modemmanager or networkmanager problem, but I need your expertise to assess that.
  All in all with 5.14 the whole system behavior seems to be better than with 5.13 but I have been testing this only for some hours until now. : )
  Let me know if you need any additional information.
  modem.txt
  openqti.log
  dmesg.txt

[Biktorgj]

Hi,
This is the last remaining reset event that we're tracking down, it's still not fully known if the issue is on the modem side, the Pinephone side or both.

In any case, ModemManager should reconnect afterwards, but I see the connection getting stuck. Did you flash latest ADSP firmware? It's linked in the release but not bundled in the installation package to avoid mixing closed source stuff with the rest: https://github.com/Biktorgj/quectel_eg25_recovery/raw/EG25GGBR07A08M2G_01.003.01.003/update/NON-HLOS.ubi

That's exactly one of the things it fixes... you can check if you have it already installed by sending the following AT command to the Modem: AT+QGMR

[fernando-c-m]

Ok. I am getting as version 01.003.00.000
This is what I (re)did:

root@mobian:/home/mobian# wget https://github.com/Biktorgj/quectel_eg25_recovery/raw/EG25GGBR07A08M2G_01.003.01.003/update/NON-HLOS.ubi
--2021-10-04 09:07:51--  https://github.com/Biktorgj/quectel_eg25_recovery/raw/EG25GGBR07A08M2G_01.003.01.003/update/NON-HLOS.ubi
Resolving github.com (github.com)... 140.82.121.3
Connecting to github.com (github.com)|140.82.121.3|:443... connected.
HTTP request sent, awaiting response... 302 Found
Location: https://raw.githubusercontent.com/Biktorgj/quectel_eg25_recovery/EG25GGBR07A08M2G_01.003.01.003/update/NON-HLOS.ubi [following]
--2021-10-04 09:07:52--  https://raw.githubusercontent.com/Biktorgj/quectel_eg25_recovery/EG25GGBR07A08M2G_01.003.01.003/update/NON-HLOS.ubi
Resolving raw.githubusercontent.com (raw.githubusercontent.com)... 185.199.109.133, 185.199.111.133, 185.199.108.133, ...
Connecting to raw.githubusercontent.com (raw.githubusercontent.com)|185.199.109.133|:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: 41025536 (39M) [application/octet-stream]
Saving to: ‘NON-HLOS.ubi’

NON-HLOS.ubi                           100%[==========================================================================>]  39.12M  9.59MB/s    in 4.5s    

2021-10-04 09:07:56 (8.78 MB/s) - ‘NON-HLOS.ubi’ saved [41025536/41025536]

root@mobian:/home/mobian# echo -ne "AT+QFASTBOOT\r" > /dev/ttyUSB2
root@mobian:/home/mobian# lsusb
Bus 004 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
Bus 002 Device 015: ID 18d1:d00d Google Inc. Xiaomi Mi/Redmi 2 (fastboot)
Bus 002 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
Bus 003 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
Bus 005 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
root@mobian:/home/mobian# fastboot flash modem NON-HLOS.ubi 
Sending 'modem' (40064 KB)                         OKAY [  1.343s]
Writing 'modem'                                    OKAY [ 12.337s]
Finished. Total time: 13.833s
root@mobian:/home/mobian# fastboot reboot
Rebooting                                          OKAY [  0.004s]

but after this I still get:

AT+QGMR

EG25GGBR07A08M2G_01.003.00.000

OK

Are there any steps which I am missing?

[LD1DL]

Hello,
Thank you for the great job accomplished.
The data connection drop seemed to come with revision 0.4.1 of the firmware. It's not Mobian specific, I get the same with Manjaro.
The stability of the modem was greatly improved with the revisions of last month but sometimes the modem still disappear and doesn't come back. Then the result of "lsusb" tells it went in fastboot mode (Google Inc. Xiaomi Mi/Redmi 2 (fastboot)) but it doesn't react to command "fastboot reboot" and I have to reboot the phone. Is it the last remaining reset event that you are tracking down or I have another issue? ADSP version: 01.003.01.003.
BR

[Biktorgj]

Hi @fernando-c-m and @LD1DL

There's still an unresolved reset cause for the modem I'm actively trying to track down. sometimes the modem will reset the usb port instead of resuming. The issue is either totally random, or I haven't been able to reliably replicate it. It can happen the second time the phone sleeps or not happen at all for 600 suspend events.

This happens much more often with the 5.14 kernel, which also exhibits data getting stuck when this happens.

Which kernel versions are you running in your phones? Also, if in doubt, update to 0.4.2. it's a small fix, but it will avoid wasting more battery than needed when a reset happens since it won't try to recover until the PinePhone is actually active

[LD1DL]

@Biktorgj
The kernel version is 5.14.8 and the firmware is already 0.4.2. Yes, you reduced a lot the power consumption with the last versions of the firmware! The data connection drop started when I updated modem FW to 0.4.1 from 0.4.0. Do you want me to downgrade to confirm only the firmware version is involved?

[pizdjuk]

Using the firmware version 0.4.2, kernel 5.14.15-1, the modem disappears and wont return.

modem.txt

[Biktorgj]

That log doesn't tell me much except the modem is still alive and responding to AT commands. Can you get a full kernel log from the Pinephone? If ADB is enabled, modem's kernel log might be handy too

[pizdjuk]

Just another hang: the modem doesnt appear on the USB bus, the bus number increases iteratively: 2,3,....,127,2,3,4,...,127,2,3.....

[11230.616030] usb 4-1: device not accepting address 14, error -62
[11230.808065] usb 4-1: new full-speed USB device number 15 using ohci-platform
[11231.224044] usb 4-1: device not accepting address 15, error -62
[11231.230157] usb usb4-port1: unable to enumerate USB device
[11232.612115] usb 4-1: new full-speed USB device number 16 using ohci-platform
[11232.804042] usb 4-1: device descriptor read/64, error -62
[11233.108030] usb 4-1: device descriptor read/64, error -62
[11233.408037] usb 4-1: new full-speed USB device number 17 using ohci-platform
[11233.600054] usb 4-1: device descriptor read/64, error -62
[11233.904055] usb 4-1: device descriptor read/64, error -62
[11234.016131] usb usb4-port1: attempt power cycle
[11234.232029] usb 4-1: new full-speed USB device number 18 using ohci-platform
[11234.648239] usb 4-1: device not accepting address 18, error -62
[11234.840068] usb 4-1: new full-speed USB device number 19 using ohci-platform
[11235.256039] usb 4-1: device not accepting address 19, error -62
[11235.262145] usb usb4-port1: unable to enumerate USB device
[11235.544065] usb 4-1: new full-speed USB device number 20 using ohci-platform
[11235.736033] usb 4-1: device descriptor read/64, error -62
[11236.040524] usb 4-1: device descriptor read/64, error -62
[11236.340479] usb 4-1: new full-speed USB device number 21 using ohci-platform
[11236.532113] usb 4-1: device descriptor read/64, error -62

[B0bbyTheBuilder]

Hello @Biktorgj

I recently took a leap of faith and flashed my daily driver, a PinePhone pmOS CE edition running latest Edge version of postmarketOS.
So far, so good. I haven't had any issues with Calls, SMS nor MMS.

However I'm experiencing data issue where the network interface wwan0 end up loosing it's IP after a while:

wwan0     Link encap:Ethernet  HWaddr FE:27:7D:XX:XX:XX  
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:12 errors:0 dropped:0 overruns:0 frame:0
          TX packets:317 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000 
          RX bytes:962 (962.0 B)  TX bytes:100806 (98.4 KiB)

I have the latest ADSP:

AT+QGMR
EG25GGBR07A08M2G_01.003.01.003

OK

And I used the flashall script from version 0.4.2 to flash your firmware.
The only command that produced an error was this one:

fastboot oem stay

I don't have the exact error but I recall it said something like unknown command and it did proceed with the rest of the script.

So, yesterday evening I restarted the phone and let it sit on the charger overnight. I checked I had data connection right after restarting.
This morning the IP was gone. I was able to call and send SMS though.

I've added the modem's logs here:

dmesg.txt
messages.0.txt
messages.txt
openqti.log.txt

Thanks for taking the time to look into it and for the amazing work you and other involved are doing here.

[LD1DL]

Hello all,

I did the test to come back to firmware version 0.4.0 and I confirm data drop is a problem with firmware versions > 0.4.0. With version 0.4.0 the data connection is stable (thought the modem disappears more frequently and usually comes back after a few tens of seconds). With v. 0.4.2 the modem is more stable but data drops frequently and I have to reboot the modem to get data back.

BR

[eugenr]

Hi,

I'm getting this issue also with 0.4.5 @Biktorgj's modem firmware.

• Pinephone Mobian Edition
• upgraded to mobian/bookworm testing
• Linux kernel 5.13-sunxi64
• ModemManager 1.18.2-0.2
• AT+QGMR: EG25GGBR07A08M2G_01.003.00.000

Phosh announced (with the icon) the data connection lost at around 'Nov 21 15:51' time. This log time is UTC+2:
pp-journalctl-e.log

Two hours later I managed to also get all the modem's logs. All this time I've prevented the phone to suspend.
modem-boot.log
modem-dmesg.log
modem-messages.0.log
modem-messages.log
modem-thermal.log

openqti.log
I've truncated this one. It contained further only lines with "[..]USB is suspended." making the original file at 78MB.

ADDED: And with 'adb shell' I saw that all the network devices ('ip address') has only zeros for adresses. Just wanted to mention this too.

[eugenr]

On 0.4.5 data connection got lost as a side effect of debug messages flooding openqti.log file until the tmpfs was left with no space anymore.

Those debug messages were removed in 0.4.5-2, and using this version I could not reproduce this issue anymore. Yay! It works great!
Thanks!

[B0bbyTheBuilder]

I second @eugenr , 0.4.5-2 works like a charm. I haven't lost data in more than 48 hours :D

[fernando-c-m]

0.4.5-2 installled on Sunday
Today it happened again.
See attached logs:
dmesg_modem.txt
openqti.log
dmesg_211125_mobian.txt

[tiol11]

Hi,
I am experiencing a similar issue. Modem does not appear in lsusb answer. There is no answer to sudo sh -c 'echo -ne "AT+ADBON\r" > /dev/ttyUSB2'. Logs which I was able to extract are attached.
Modem gets back online by switching off its kill switch, then switch back on and sudo systemctl restart eg25-manager. Connection is back but modem does not answer AT commands (no AT port available to run command).
Phone is PinePhone 3GB with up-to-date Mobian Bookworm,
kernel 5.15-sunxi64 from Mobian repo.
Network manager 1.34.0-1 from Debian repo.
Modem firmware 0.5.3.
ADSP version 01.003.01.003 (note 7 in https://github.com/Biktorgj/pinephone_modem_sdk/blob/honister/docs/AT_INTERFACE.md suggests the answer to "AT+QGMR" is the fw version instead of the ADSP, but ADSP section in https://github.com/Biktorgj/pinephone_modem_sdk/releases release announcements tells the opposite)
networkmanager.log
ModemManager.log
eg25-manager.log
dmesg_pinephone.log

Hope this helps in tracing this issue, thank you for maintaining this firmware.