Re: [Intel-gfx] USB-C DP mode problem on linux

From: Imre Deak <imre.deak@intel.com>
To: Heikki Krogerus <heikki.krogerus@linux.intel.com>,
	Andrzej Kre <andrzej.kre@gmail.com>
Cc: intel-gfx@lists.freedesktop.org,
	Heikki Krogerus <heikki.krogerus@linux.intel.com>
Subject: Re: [Intel-gfx] USB-C DP mode problem on linux
Date: Tue, 20 Oct 2020 17:25:16 +0300	[thread overview]
Message-ID: <20201020142516.GC3370607@ideak-desk.fi.intel.com> (raw)
In-Reply-To: <20201019142459.GM1667571@kuha.fi.intel.com>

Hi,

On Mon, Oct 19, 2020 at 05:24:59PM +0300, Heikki Krogerus wrote:
> Hi Andrzej,
> 
> On Sat, Oct 17, 2020 at 01:34:54PM +0200, Andrzej Kre wrote:
> > Hi,
> > 
> > I know that You were involved in working on USB-C DP drivers.
> > You are my last chance to resolve my issue.
> > 
> > On my HP laptop I have Intel UHD Graphics 620.
> > When I'm connecting my 4K monitor to Display Port. It is assigning to
> > DP-2-2 socket and  I have full 3840x2160 with 60.00Hz
> > But, when I'm connecting the same monitor to the USB-C port, then it is
> > connecting to the DP-1 socket and the maximum that it can achieve is
> > 3840x2160 with only 30.00Hz.
> > But I'm making some trick: I'm connecting the same monitor through HDMI, so
> > it is connecting to DP-1 socket, and simultaneously I'm connecting USB-C,
> > and now USB-C is connecting to DP-2-2 socket (because DP-1 is occupied by
> > HDMI) and I can have full 4K with 60Hz.
> > Please, help me, how to force USB-C to connect always to DP-2-2 socket?
> > Or do You know maybe where is the problem?
> 
> Unfortunately we have no control over the mux in the operating system
> on Skylakes, at least in USB subsystem. It all happens in firmware.
> Maybe graphics side can do something.
> 
> Adding Jani, Imre, Ville and the Intel GFX list.

On SKL/KBL the USB-C -> native DP/HDMI conversion is done by an off-CPU
chip and the display driver doesn't have a way either to affect the
muxing.

Not sure why things work on DP-2 and not on DP-1, there is no port
specific limits on the CPU side that would explain this. There is a link
training failure in the log, so would be good to see more details on
that. Could you file a ticket at 
https://gitlab.freedesktop.org/drm/intel/-/issues
providing a full log booting with drm.debug=0x1e for the working and
non-working cases?

Thanks,
Imre
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx