AM3356: Communication with /dev/ttyACM is not possible

Hi Ohya-san,

Please explain what module(s) geneate the /dev/ttyACM0 and ttyACM1 device modes?

Hello, Bin

We will provide you with information.

Is the information you need?

- dmesg

$dmesg | grep ttyACM0
cdc_acm 2-1:1.0: ttyACM0: USB ACM device

$dmesg | grep ttyACM1
cdc_acm 2-1:1.2: ttyACM1: USB ACM device

- lsmod

$lsmod | grep cdc_acm
cdc_acm                24576  0

- modinfo

$modinfo cdc_acm
filename:       /lib/modules/6.1.46-g1d4b5da681/kernel/drivers/usb/class/cdc-acm.ko
alias:          char-major-166-*
license:        GPL
description:    USB Abstract Control Model driver for USB modems and ISDN adapters
author:         Armin Fuerst, Pavel Machek, Johannes Erdfelt, Vojtech Pavlik, David Kubicek, Johan Hovold
alias:          usb:v1519p0452d*dc*dsc*dp*ic*isc*ip*in*
 .....(omit)
alias:          usb:v0424p274Ed*dc*dsc*dp*ic*isc*ip*in*
depends:
intree:         Y
name:           cdc_acm
vermagic:       6.1.46-g1d4b5da681 preempt mod_unload ARMv7 p2v8

Best regards,

Hi Ohya-san,

Thanks, I think I understand the /dev/ttyACM0 and ttyACM1 nodes are created by a USB modem attached.

Naomasa Ohya said:

After configuring GPS reception, if you do not read data from /dev/ttyACM1,

after a while you will not be able to send or receive AT commands on /dev/ttyACM0.

When the problem happens, does linux console print any message?

Hello, Bin

When this occurs, nothing is displayed on the Linux console.

Best regards,

Hi Ohya,

Please check your kernel .config to see if option "CONFIG_USB_TI_CPPI41_DMA" is enabled. If so, please try to disable it to see if the issue still happens.

Hello, Bin

The option "CONFIG_USB_TI_CPPI41_DMA" was enabled, so I disabled it and checked.

It did not solve the problem.

In the following versions, no problems occur regardless of whether it is enabled or disabled.

・SDK Version.08.02.00.24
・Linux Kernel Version.5.10.100(2021 LTS)

Best regards,

Hi Ohya-san,

Okay thanks for running the test.

For figuring out which kernel change causes the issue, you can use 'git bisect' tool on the kernel source code, which would narrow down the kernel commit id which causes the issue.

I recommend you do the 'git bisect' process on the upstream kernel instead of SDK kernel. TI has not done much of AM335x USB related kernel development in recent years, so the USB driver in upstream kernel should be the same as TI SDK kernel.

TI SDK v9.1 kernel is about equivalent to upstream kernel v6.1.46. So first check if this kernel has this issue;

Next check if upstream kernel v6.1-rc1 has the issue. If not, you can use 'git bisect' between upstream kernel v6.1-rc1 and v6.1.46.

Hello, Bin

I found out that it was caused by some kernel change.

Can you investigate and fix it?

This could be a fatal problem.

Best regards,

Naomasa Ohya said:

I found out that it was caused by some kernel change.

What did you find?

Hello, Bin

Sorry、I haven't found anything

I misunderstood Bin's answer.

We need to investigate where the cause is.

Best regards,

Hi Ohya-san,

As I explained how to use 'git bisect', I was expecting you to debug it.

I don't have an USB device which has two ACM interfaces to reproduce and debug the problem.