AM3358 USB Host Split NAK behavior

Yuji Kuwabara

Other Parts Discussed in Thread: AM3358

I'm working on AM3358 (rev2.1) USB1 as host mode.

NAK timeout works correctly (for full speed bulk IN endpoint) if a device is attached to root port (without intermediate hub).

However, it seems to me that NAK timeout does not occur (for same full speed bulk IN endpoint) if a device is attached via high speed hub.

(it means that USB core endpoint RX transaction does not complete until a full speed bulk IN endpoint (via hub) returns DATA)

Is this behavior by design?

Regards.

over 9 years ago

0 Biser Gatchev-XID over 9 years ago

TI__Guru**** 393215 points

Hi,

What software and which version are you using?

0 Yuji Kuwabara over 9 years ago in reply to Biser Gatchev-XID

Prodigy 190 points

It is my own software (no OS at all).

So there may be problem in my program.

Regards.

0 Biser Gatchev-XID over 9 years ago in reply to Yuji Kuwabara

TI__Guru**** 393215 points

I have asked the USB experts for help on this.

0 Bin Liu over 9 years ago

TI__Guru*** 146336 points

Yuji Kuwabara said:
(it means that USB core endpoint RX transaction does not complete until a full speed bulk IN endpoint (via hub) returns DATA)

I am not sure I understand the problem. Do you have a USB analyzer trace or other debug data which can explain the problem in details?

I can read Lecroy .usb files.

0 Yuji Kuwabara over 9 years ago in reply to Bin Liu

Prodigy 190 points

I'm sorry that I do not have USB analyzer.

I simply observe USB interrupt handler.

If you think such behavior is not by design, I will further investigate.

Regards.

0 Bin Liu over 9 years ago in reply to Yuji Kuwabara

TI__Guru*** 146336 points

The USB2.0 Spec does not define the NAK timeout, it is up to the implementation of design and handling the NAK timeout.

The MUSB controller on AM335x has a register RX_Interval (offset 0Bh) for each RX Endpoint to define the NAK timeout for IN transactions when MUSB is in host mode.

So please check how this register has been programmed in your software in both use cases to see if that will explain the difference in your test.

0 Yuji Kuwabara over 9 years ago in reply to Bin Liu

Prodigy 190 points

I set a register HOST_RXINTERVAL (offset 0x0D per endpoint) to "2".
By the way, I found a musb driver source code.
It seems to me that it is known behavior.

github.com/.../musb_host.c

/* Enable bulk RX NAK timeout scheme when bulk requests are
* multiplexed. This scheme doen't work in high speed to full
* speed scenario as NAK interrupts are not coming from a
* full speed device connected to a high speed device.
* NAK timeout interval is 8 (128 uframe or 16ms) for HS and
* 4 (8 frame or 8ms) for FS device.
*/

However, if I understand high speed hub Transaction Translator specification correctly, NAK is returned to USB host.

Regards.

Processors

Processors forum

AM3358 USB Host Split NAK behavior