PROCESSOR-SDK-AM335X: USB Device mode

Hello Biser,

Yes we can put partial schematics.

Attached is the schematic for the USB0 circuit.  It functions as a host – no issue.

TVS – D3 is not-populated, It is show in in the schematic but has been removed from the board.

The caps C443 and C444 have also been not-populated, they have been removed from the board.

The resistors R651 and R653 have been replaced with 0 ohm resistors.

The resistor R652 has also been removed from the board.

Thanks

Maulin Lodhiausb_host_or_peripheral_comments.pdf

Hello DK,

Yes we had made a special cable Mini - AB compliant to verify Host functionality on USB0.

But our main issue is to make this USB0 port as peripheral. Our Hw expert verified and confirmed all 3 points highlighted in your last response. Thus it may be something else... SW configuration related perhaps ?

FYI - Linux version we are using is  3.12.17

Thanks

Maulin Lodhia

Maulin,

Please provide the console output of command 'cat /sys/class/udc/*/uevent'.

Maulin,

There are a few things probably wrong.

Maulin Lodhia said:

[ 80.347666] musb-hdrc musb-hdrc.0.auto: Falied to request rx1.

I am not sure about this error, never saw it before. Let's ignore it for now.

Maulin Lodhia said:

[ 80.388390] musb-hdrc musb-hdrc.0.auto: musb_init_controller failed with status -517

Very likely you didn't build kernel with CONFIG_MUSB_PIO_ONLY=y this time, and the musb driver is waiting for CPPI DMA driver, and not loaded properly.

Maulin Lodhia said:

Now when I tried installing individual gadget driver I get these errors
insmod g_serial.ko

Don't try to load a gadget driver at this point, because the usb controller driver is not ready.

Please avoid to use insmod command for loading usb drivers, and use modprobe instead. The usb drivers have many dependencies, and using insmod can easily get it wrong.

Please use command 'modprobe musb_am335x' to load all the usb controller related drivers, then use 'modprobe g_serial' to load the serial gadget driver.

BTY, you should'nt need 'modprobe musb_am335x' if you installed all the kernel modules correctly into your filesystem, the kernel boot process should load those usb drivers automatically for you.

Maulin Lodhia said:

[ 233.180214] USB Speed : USB 2.0
[ 235.365578] TE Flow Control Enabled
[ 235.657205] creating qcqmi0

But I cant still install individual modules on target since I am not sure if modprobe actually installed the modules correctly due to lack of feedback. However as mentioned in my previous message insmod g_serial and related modules gives me errors.

By now you only loaded usb controller related drivers. To load g_serial driver, please try the following - note that I didn't test it... (you really should use modprobe, if it doesn't load the drivers correctly, you should fix your filesystem - you didn't install all the kernel modules properly.)

# insmod libcomposite
# insmod g_serial

Maulin,

Maulin Lodhia said:

owever for individual USB test I did not want to create the entire rootfs again instead just changing uImage (kernel image) and dtb (device tree (if necessary).

This is not recommended. Unlike the old 2.6 kernel, now the kernel usb subsystem, especially the udc, is getting more and more complicated in terms of dependencies. You'd better update the kernel modules along with kernel image and dtb in the rootfs, unless you know how the kernel exactly works.

Maulin Lodhia said:

I tried installing it individually as per your suggestion but still get errors.
insmod libcomposite.ko
[ 2486.214639] libcomposite: Unknown symbol config_group_init (err 0)
[ 2486.221251] libcomposite: Unknown symbol configfs_unregister_subsystem (err 0)
[ 2486.229821] libcomposite: Unknown symbol config_item_set_name (err 0)
[ 2486.236944] libcomposite: Unknown symbol configfs_register_subsystem (err 0)
[ 2486.244593] libcomposite: Unknown symbol config_group_init_type_name (err 0)
[ 2486.252109] libcomposite: Unknown symbol config_item_put (err 0)
Error: could not insert module libcomposite.ko: Unknown symbol in module

All those missing symbols are in the fs/configfs, please check if you have 'CONFIG_CONFIGFS_FS=y' in your .config.

Maulin Lodhia said:

root@kona64:/lib/modules/3.12.17/kernel/drivers/usb# insmod g_serial.ko
[ 2540.942261] g_serial: Unknown symbol usb_remove_function (err 0)
...
Error: could not insert module g_serial.ko: Unknown symbol in module

Don't try to insmod g_serial if libcomposite.ko is not loaded properly.

Hello Bin

After making CONFIG_CONFIGFS_FS=y instead of 'm' orginally

I was able to successfully install external modules in this order

• udc-core
• musb_hdrc
• musb_dsps
• musb_am335x
• libcomposite
• u_serial
• usb_f_serial
• usb_f_acm
• usb_f_obex
• usb_g_serial
• finally... insmod g_serial.ko

Output: 

[ 292.069948] g_serial gadget: Gadget Serial v2.4
[ 292.074846] g_serial gadget: g_serial ready

But the original problem still persists when I plug USB0 (using a cable) to PC it doesn't even attempt to enumerate. 

Conclusion:

Both built-in and external loading on modules (gadget drivers) have the same behavior.

Thanks

Maulin 

Maulin Lodhia said:

[ 292.069948] g_serial gadget: Gadget Serial v2.4
[ 292.074846] g_serial gadget: g_serial ready

But the original problem still persists when I plug USB0 (using a cable) to PC it doesn't even attempt to enumerate. 

Please provide the output of 'cat /sys/class/udc/musb_hdrc.0.auto/uevent' in this case, before I start debug this.

And please attach you .config for my debug.

Yes I have all 3 CONFIG flags set to 'y'

8562.config.txt

Hello Bin,

Here is the output of gadget when it built in

dmesg | grep gadget
[ 1.473908] g_serial gadget: adding 'acm'/dd9c53c0 to config 'CDC ACM config'/c05ed5c4
[ 1.473964] g_serial gadget: acm ttyGS0: dual speed IN/ep1in OUT/ep1out NOTIFY/ep2in
[ 1.473992] g_serial gadget: Gadget Serial v2.4
[ 1.478759] g_serial gadget: g_serial ready

Here is the output of usb in general for the same linux load

 dmesg | grep usb
[ 0.257687] usbcore: registered new interface driver usbfs
[ 0.257962] usbcore: registered new interface driver hub
[ 0.258269] usbcore: registered new device driver usb
[ 1.322228] usbcore: registered new interface driver asix
[ 1.328253] usbcore: registered new interface driver ax88179_178a
[ 1.334946] usbcore: registered new interface driver cdc_ether
[ 1.341333] usbcore: registered new interface driver r815x
[ 1.347316] usbcore: registered new interface driver net1080
[ 1.353525] usbcore: registered new interface driver cdc_subset
[ 1.359995] usbcore: registered new interface driver zaurus
[ 1.366151] usbcore: registered new interface driver cdc_ncm
[ 1.376700] usbcore: registered new interface driver GobiNet
[ 1.384307] usbcore: registered new interface driver usbserial
[ 1.390698] usbcore: registered new interface driver GobiSerial
[ 1.397142] usbserial: USB Serial support registered for GobiSerial
[ 1.412177] of_get_named_gpio_flags: can't parse gpios property of node '/ocp/usb@47400000/usb-phy@47401300[0]'
[ 1.413706] musb-hdrc musb-hdrc.0.auto: Enabled SW babble control
[ 1.420142] musb-hdrc: ConfigData=0xde (UTMI-8, dyn FIFOs, bulk combine, bulk split, HB-ISO Rx, HB-ISO Tx, SoftConn)
[ 1.420159] musb-hdrc: MHDRC RTL version 2.0
[ 1.420171] musb-hdrc: setup fifo_mode 4
[ 1.420197] musb-hdrc: 28/31 max ep, 16384/16384 memory
[ 1.421668] of_get_named_gpio_flags: can't parse gpios property of node '/ocp/usb@47400000/usb-phy@47401b00[0]'
[ 1.423088] musb-hdrc musb-hdrc.1.auto: Enabled SW babble control
[ 1.429516] musb-hdrc: ConfigData=0xde (UTMI-8, dyn FIFOs, bulk combine, bulk split, HB-ISO Rx, HB-ISO Tx, SoftConn)
[ 1.429532] musb-hdrc: MHDRC RTL version 2.0
[ 1.429543] musb-hdrc: setup fifo_mode 4
[ 1.429567] musb-hdrc: 28/31 max ep, 16384/16384 memory
[ 1.429676] musb-hdrc musb-hdrc.1.auto: MUSB HDRC host driver
[ 1.435858] musb-hdrc musb-hdrc.1.auto: new USB bus registered, assigned bus number 1
[ 1.473348] udc musb-hdrc.0.auto: registering UDC driver [g_serial]
[ 1.554095] usbcore: registered new interface driver usbhid
[ 1.559955] usbhid: USB HID core driver
[ 2.053750] usb 1-1: new high-speed USB device number 2 using musb-hdrc
[ 2.217037] usb 1-1: config 1 has an invalid interface number: 8 but max is 3
[ 2.224572] usb 1-1: config 1 has no interface number 1
[ 2.294312] usb 1-1: GobiSerial converter now attached to ttyUSB0
[ 2.324194] usb 1-1: GobiSerial converter now attached to ttyUSB1
[ 2.354015] usb 1-1: GobiSerial converter now attached to ttyUSB2
[ 2.383926] GobiNet 1-1:1.8 wwan0: register 'GobiNet' at usb-musb-hdrc.1.auto-1, GobiNet Ethernet Device, 32:d1:59:7f:05:81

As per the highlighted g_serial is registered. However when I plug the cable in to PC it doesn't even attempt to enumerate.

Thanks

Maulin Lodhia

Maulin Lodhia said:

As per the highlighted g_serial is registered. However when I plug the cable in to PC it doesn't even attempt to enumerate.

Please don't forget to check the output of 'cat /sys/class/udc/musb-hdrc.0.auto/uevent', to ensure the gadget driver is bound t MUSB driver.

Then dump the following two registers.

# devmem2 0x47401401 b
# devmem2 0x47401460 b

Here is the requested output.

cat /sys/class/udc/musb-hdrc.0.auto/uevent

DRIVER=g_serial

USB_UDC_NAME=musb-hdrc

USB_UDC_DRIVER=g_serial

devmem2 0x47401401 b

/dev/mem opened.

Memory mapped at address 0xb6fbf000.

Read at address  0x47401401 (0xb6fbf401): 0xF0

devmem2 0x47401460 b

/dev/mem opened.

Memory mapped at address 0xb6fad000.

Read at address  0x47401460 (0xb6fad460): 0x89

Ok, everything on the sw side is good now, but

Maulin Lodhia said:

devmem2 0x47401460 b

Read at address  0x47401460 (0xb6fad460): 0x89

tells the AM335x is unable to sense VBUS - this register should be 0x99 by now, but it is 0x89 instead.

So I guess because AM335x MUSB controller is unable to detect VBUS, it doesn't enable D+ pullup, so the USB host is unable to detect the attach,

I will let the hw guru to help you to check your hw again, but just to double check, you removed R759 to not ground the ID pin, right?

Hello Bin,

Could you send be the register description for 0x47401460 and any other register that we should be checking ?

I tried looking at SPRUH73G document and couldn't find it.

Also as per SPRUH73G pg. 1686 (section 16.3.7)

The procedure for the USB2.0 OTG controller determining its operating modes (usb controller assuming a

role of a host or a peripheral) starts when the USB 2.0 controller goes into a session. The USB 2.0

controller is in session when either it senses a voltage (>= 4.4V) on the USBx_VBUSIN pin and the

controller sets its DEVCTL[SESSION] bit field or when the firmware sets the DEVCTL[SESSION] bit;

assuming that it will operating as a host.

I couldn't find the DEVCTL register in section 16.5 (USB Registers) of the same document

Regards,

Maulin Lodhia

Please verify USBn_VBUS is >4.4V at the processor.

Hello DK,

It is 5.1V which is > 4.4V

Thanks

Maulin

VBUS is being supplied by the Host when you are in Device mode, yes? Verify USBn_VBUS is 0V when we are disconnected from the Host, and >4.4V when we are connected to the Host.

-DK- said:

VBUS is being supplied by the Host when you are in Device mode, yes? Verify USBn_VBUS is 0V when we are disconnected from the Host, and >4.4V when we are connected to the Host.

In addition to DK's request above, please also dump register 'devmem2 0x47401460 b' in both cases.

When USB is Disconnected:

• USBn_VBUS is 0V
• devmem2 0x47401460 b

/dev/mem opened.

Memory mapped at address 0xb6f11000.

Read at address  0x47401460 (0xb6f11460): 0x80

When USB is Connected:

• USBn_VBUS is 5.1V
• devmem2 0x47401460 b

/dev/mem opened.

Memory mapped at address 0xb6f11000.

Read at address  0x47401460 (0xb6f11460): 0x89

Thanks

Maulin

Maulin Lodhia said:

When USB is Connected:

USBn_VBUS is 5.1V

devmem2 0x47401460 b

Read at address  0x47401460 (0xb6f11460): 0x89

I still think there is something wrong on the board. When the AM335x USB0 controller detects VBUS > 4.4v, bit[3,4] of register 0x47401460 should be 11b, but now it is 01b, which means the detected VBUS is < AVALID.

Hello DK/Bin,

What about USB_CTRL0 register? What are the bits set to for this register?

 Reference SPRUH73G–October 2011 pg. 780

Figure 9-24, Table 9-31

Bits 31 to 24 must always stay 0x3c?

Bit 23, datapolarity_inv what is that set to?

Bit 20, otgsessenden, session end detect enable

Bit 19, otgvdet_n, vbus detect enable? Should this be set to 1?

Bit 18, dmgpio_pd, DM is not being used in GPIO mode, no pulldown enabled?

Bit 17, dpgpio_pd, no pulldown enabled?

Bit 14 and bit 15, no UART or inversion?

Bit 12, gpiomode, USB mode = 0 and not GPIO mode (uart mode)?

Bit 10, cdet_extctl, charger detection is bypassed?

Bit 9 and bit 8, pullups enabled or no effect?

Bits 7 and 6 charger case bits?

Bits 5 and 4 sink or source?

Bit 2, charger detect disable, 1 disable?

Bit 1 and 0 power down bits?

 So when I read the value for this register it is showing 0x00000000

Tried reading this on 2 units with same results ?

Could this be the reason why VBUS is not getting detected ?

Thanks

Maulin Lodhia

Maulin Lodhia said:

 So when I read the value for this register it is showing 0x00000000

It shouldn't be 0. Are you sure you were reading from 0x44e10620?

What is the value of USB_CTRL1 which is 0x44e10628 for the 2nd usb port?

FYI

The following is a scope shot of the USB0_VBUS (Pin P15) on the AM3352. This scope shot is taken right under the device on the secondary side of the board. Since the part is populated on the primary side of the board. The scope shot shows the USB0_VBUS signal rising from close to ground to 5.07V as the USB cable is plugged in. There is no other point on the board that is closer to the device to be able to measure. I have checked this on a few boards so the enclosed waveform should be what is present on the input USB0_VBUS pin of the device. Even with the enclosed waveform the device is not showing that VBUS is being detected (via the devmem2 read that was mentioned earlier). The enclosed Vbus level 5.07V is > than 4.75V is > than 4.4V.

Why would the device not be able to see this VBUS signal which is at the correct level at its respective input pin (P15)?

VSSA_USB_0 (pin N14) and VSSA_USB_1 (pin M14) are connected to ground. VDDA3P3V_USB0 (pin N15) and VCCA3P3V_USB1 (pin R15) are connected to 3.3V. VDDA1P8V_USB0 (pin N16) and VDDA1P8V_USB1 (pin R16) are connected to 1.8V.

Hello,

Any update on this ? Could you please help answer Ron's concerns/questions for HW design ? As per the design and scope capture the VBUS level is 5.07v which is greater than 4.4V to be detected. So why would am335x now detect it ? 

Are we missing something here ?

Thanks

Maulin

You say this port works in Host mode. Please switch it to Host mode and repeat the previous experiment:

devmem2 0x47401460 b

Please post the result here.

Before connecting a device 

Read at address  0x47401460 (0xb6f9c460): 0x19

After connecting a device

[  632.123519] usb 1-1: new full-speed USB device number 5 using musb-hdrc

devmem2 0x47401460 b

/dev/mem opened.

Memory mapped at address 0xb6f66000.

Read at address  0x47401460 (0xb6f66460): 0x5D

[  642.273662] hid-generic 0003:2047:0301.0004: timeout initializing reports

[  642.280831] hid-generic 0003:2047:0301.0004: device has no listeners, quitting

Hello Bin,

SW Change:

I did not rebuild the kernel only thing I changed is device tree (only one line).

usb@47401000 {
status = "okay";
/* USB 0 is peripheral - trying host mode */
dr_mode = "host";
};

In regards to HW change I will summarize the change shortly.

Thanks

Maulin Lodhia