[PATCH] Intel xhci: Only switch the switchable ports

Keng-Yu Lin kengyu at canonical.com
Mon Aug 13 06:25:55 UTC 2012

On Thu, Aug 9, 2012 at 7:26 PM, Andy Whitcroft <apw at canonical.com> wrote:
> The patch _looks_ reasonable.  Has this been tested on anything without
> the affected hardware?  Is it going upstream?
> -apw

I did test the patch on various Panther Point-based machines. They all
work well with the patch. On the affected machines, the webcam issue
is fixed. On the unaffected machines USB works.

The previous upstream port switching patch made it only do the port
switching on Panter Point platform by checking the PCI IDs. So my
patch in consequence does this way too.

There are really affected and unaffected hardware, though they share
the same PCI device ID 0x1E26. HM70 platform is affected by this
issue, HM77 is not.
But at the moment I have no idea on the chipset mystery. (I've been
always assuming the same PCI IDs means the same thing, in fact it's
not in the case).

However in general, the XUSB2PRM and USB3PRM registers do contain the
information of which ports should not be altered from its original
value set-up by BIOS.
My patch just implements the check and make the kernel follow the
designed behaviour.


More information about the kernel-team mailing list