Linux/AM5718: AM5718 USB1 detach problem

Mauro Palombini

Part Number: AM5718

Tool/software: Linux

Dear Sirs, We are having some trouble with USB1 interface of AM5718 used as HOST in HS mode (Only USB2.0 D+/D- signals are connectes). Randomly, the USB key connected to the port disconnects during a read or write operation. The conditions are:

1) The used USB Key is Data Traveler da 16GByte (see picture)
2) The linux Kernel is 4.9.45-rt23

More than one key of the same brand and model have been tested and all of them show the problem. The USB Key is a USB SS type key. Other two SS USB Key brands have been tested without any problem.

Have you had experience on this type of problem?

Thanks in advance

Best regards

over 7 years ago

0 Bin Liu over 7 years ago

TI__Guru**** 170521 points

Hi Mauro,

Do you have any kernel log showing the disconnect problem?
Does the issue happen on your custom board? If so, do you see the same problem on TI EVM?

0 Mauro Palombini over 7 years ago in reply to Bin Liu

Intellectual 265 points

Dear Liu, no we don't have any log, what happens is that the dev disappears. This occurs in our custom board and unfortunately we can't check on the EVM because it uses HUB and the proble is not present when the HUB is present.

Thanks

0 Mauro Palombini over 7 years ago in reply to Bin Liu

Intellectual 265 points

Sorry Liu, can be the problem related to the errata reported on "i845: USB2.0 False Detection of Disconnect Condition"?

Please let me know.

BR

0 Bin Liu over 7 years ago in reply to Mauro Palombini

TI__Guru**** 170521 points

Mauro,

does 'lsusb' command on the board still list the thumb drive or not?
Do you have access to AM5718 IDK evm? It doesn't have hub on the board.

0 Mauro Palombini over 7 years ago in reply to Bin Liu

Intellectual 265 points

Dear Liu, I'm not software eng. and I will forward your question to the SW dept.. We have not the AM5718 IDK EVM but you gave me a good suggestion. We will ask one.

BR

0 Bin Liu over 7 years ago in reply to Mauro Palombini

TI__Guru**** 170521 points

Mauro,

The software workaround mentioned in i845 has been implemented since kernel v3.15. So I doubt you issue is caused by i845, unless your board layout has a problem. have you checked your board signal integrity on usb?

0 Mauro Palombini over 7 years ago in reply to Bin Liu

Intellectual 265 points

Liu, yes we have had the signal integrity test and eye diagram. See attached file.

BRGRL6002286_usb_host-Compliance_Test.pdf

0 Mauro Palombini over 7 years ago in reply to Bin Liu

Intellectual 265 points

Dear Liu, we tested one more model from Kingston, the Datatraveler 100 g3 of 16GB and it had similar problem.

Is it possible to check the problem at your side using the two USB Keys of Keyston DataTraveler G4 of 16GB of capacity and Datatraveler 100 G3 of 16GB of capacity?

The Datatraveler 100 g3 of 16GB seems to show the problem more constantly.

Please let me know.

Thanks

0 Bin Liu over 7 years ago in reply to Mauro Palombini

TI__Guru**** 170521 points

Hi Mauro,

Can you please try the patch in the following post?

0 Mauro Palombini over 7 years ago in reply to Bin Liu

Intellectual 265 points

Sorry Liu, could you please send me the complete patch? I think I can't see all the tread contents!

Thanks

0 Bin Liu over 7 years ago in reply to Mauro Palombini

TI__Guru**** 170521 points

Here you go.

diff --git a/drivers/usb/core/hub.c b/drivers/usb/core/hub.c
index bcc1e1b..38adcd4 100644
--- a/drivers/usb/core/hub.c
+++ b/drivers/usb/core/hub.c
@@ -4229,7 +4229,6 @@ static void hub_set_initial_usb2_lpm_policy(struct usb_device *udev)
        struct usb_hub *hub = usb_hub_to_struct_hub(udev->parent);
        int connect_type = USB_PORT_CONNECT_TYPE_UNKNOWN;
 
-       if (!udev->usb2_hw_lpm_capable)
                return;
 
        if (hub)

0 Mauro Palombini over 7 years ago in reply to Bin Liu

Intellectual 265 points

Dear Liu, the patch seems to work. We are still continuing the check. I will let you know the result.

Processors

Processors forum

Linux/AM5718: AM5718 USB1 detach problem