USB not working

[leon_heller]

[Sean S]

Hi Shodruk

Thanks for the note on overloading.
I can accept that and the info has been rather clearly communicated. My bigger concern is that the simplest of USB peripherals - specifically keyboards and mice using much less than 500mA - are not working.

Hi 9600

The last update from Adapteva seems to give the impression that the ECO has fixed the problem when in fact it has not. If they are working on it at least they can advise us and let us know what is happening. There are likely many people out there struggling with USB and blissfully unaware that a huge problem exists. The issue needs to be clearly communicated. Likewise with progress... The same would apply to any other product that you bought.

All the best to you all

[ESI]

connecting serial cable to parallella brought a little light into the darkness. In my case, whenever usb was not working, I found a "phy not found" error message directly after EHCI-texts. Then lsusb also failed with -99.
Today I tried to connect just the empty OTG-cable. Then it worked. I connected (hub plus) keyboard and mouse, and still working. Booting with everything was ok withbeverything connected. I onlz tried a few times, but right now, I am having no problem with USB host anymore. My guess is, only host-mode of OTG is correctly configured, so that without OTG cable no phy is found....

[shodruk]

Thanks, ESI.
Maybe that is the point!

renatovitolo, ESI,

Could you post the output of these commands?

cat /proc/cmdline

cat /etc/X11/xorg.conf

[Sean S]

Hi ESI

I am not sure if I fully understand. If I am interpreting you properly, it's a matter of making sure that the USB cable is for USB OTG host:

correct (A device/host): OTG cable = microUSB pins 4 and 5 are shorted( pin 4 is ID, 5 is gnd)
Incorrect (B device/slave): data cable = microUSB pin 4 is unused (pin 4 is ID, 5 is gnd)

Thanks

[ESI]

bad news, it stopped working again today, when I started the cold parallella

@shodruk
root@linaro-nano:~# cat /proc/cmdline
console=ttyPS0,115200 root=/dev/mmcblk0p2 rw earlyprintk rootfstype=ext4 rootwait devtmpfs.mount=1

root@linaro-nano:~# cat /etc/X11/xorg.conf
cat: /etc/X11/xorg.conf: No such file or directory

@Sean S
with upper information it does not matter anymore, but what I meant with the OTG-cable was the short cable (microUSB to host port) which is in the assesory kit from Adapteva. (I guess it shortens pins 4 and 5..., never measured it, though)

but back to the facts. I came home, connected OTG-cable, network(1G) and microHDMI (oh, and serial port of course), than i powered it up. All accessories are from the official kit. But no usb
Reset button did not help, connecting a hub and the reset button neither. With nothing in the usb-port I disconnected power and reconnected it, and then there it was...ok again. Connecting, disconnecting, repowering, reset button it did not stop working.

So here is my next theory.
The USB-driver probing is maybe too sharp on timeout. Slightly changes in temperature, cable lentgh etc. is the difference between working and not working. Or the Phy does not like his voltage ramp up sometimes... Maybe too early releasing the Phy from reset...
So next step could be to make the EHCI-driver a module, to be able to load unload it again.(but today, only repowering helped to recover...)
Is there a possibility to reset the Phy via SW (GPIO pin or something?) Then a reset pulse at boot may help also...

[shodruk]

Thanks ESI,

1. Could you try this configuration?

bootargs = "root=/dev/mmcblk0p2 rw rootfstype=ext4 rootwait";

compile to dtb, sync, reboot, then confirm cmdline by

cat /proc/cmdline

This enables devtmpfs and disables serial console.
This should improve bootup stability.

2. What is your USB hub?
3. What if connect a keyboard only?

[Sean S]

lsusb also gives me the mentioned error message -99 of libusb. I also did as suggested in http://elinux.org/Parallella_Speed_Up_Ubuntu, however this did not solve the problem.

confirmed that my usb adapter has pins 4 and 5 shorted (USB OTG host)

[Sean S]

Hi Shodruk

I tried:

bootargs = "root=/dev/mmcblk0p2 rw rootfstype=ext4 rootwait";

Still unable to initi -99

[Sean S]

Well , that's it from me. Parallella going in the bin. You get what you pay for after all.

I am shocked that Adapteva have not responded. And Xilinx, please consider your partners more carefully - this has left a very sour taste in my mouth and it is an opportunity lost for you. I won't let my engineers touch Zynq.

Hello Microsemi and Altera!

Thank you to everyone that offered support and good luck in sorting this out. Hopefully Adapteva will realize the value of their loyal backers... and if not, I sincerely hope Xilinx pull their finger out and help Adapteva so as not to lose more opportunities.