[meta-intel] Atom E3800 USB3 Device Problem

[Yong, Jonathan]

On 06/21/2016 20:08, Anders Christensen wrote:
> Hi Jonathan,
> 
> Thanks for your reply.
> 
> We have developed our own mainboard based on the Bakersport/Valleyisland design.
> The BIOS is a Coreboot based BIOS version 610 (06/24/15) from Sage Electronics (which unfortunately is out of business).
> The BIOS FSP seems to be a modified version of the GOLD3 release.
> My CPU is a D0 stepping and I'm using microcode M0130679907.
> By default the USB3 device was disabled in the BIOS, so I have done the following configuration myself:
> 
> -          Enabled the OTG device on the SOC straps (document 514482, section A.2).
> 
> -          Enabled the OTG device in the coreboot devicetree.cb file.
> 
> -          Configured an interrupt in the irqroute.h file i.e. added PCI_DEV_PIRQ_ROUTE(OTG_DEV,  D, A, A, A).
> 
> -          Enabled the ULPI interface on the GPIO pins (document 526998, section 7.1).
> 
> -          By default the BIOS writes 0x107 to PCI configuration register 0x04 (PCI_COMMAND).
> 
> -          All other registers left unchanged by BIOS.
> Our hardware engineers have verified that the ULPI clocks (USB_ULPI_CLK and USB_ULPI_REFCLK) are running with correct frequencies.
> When cable is inserted we can see some additional communication on the ULPI data and control pins.
> I have built our own Yocto 2.0 image with both the default kernel 3.14 as well as the newer kernel 4.1.15. No change.
> I have also tried the Krogoth21/Corei7 BSP with kernel 4.4 as you suggested. No changes either.
> Here is the register dump just after the error occurred (just after loading the mass storage driver).
> By the way, I don't see any changes in register values whether the cable is inserted or not.

Our engineers are requesting an ACPI table (is there one? I'm not
familiar with coreboot) dump along with the kernel configuration.

Is the connection a USB uAB or type-C?