From mboxrd@z Thu Jan 1 00:00:00 1970 From: ville.syrjala@linux.intel.com (Ville =?iso-8859-1?Q?Syrj=E4l=E4?=) Date: Tue, 3 Dec 2013 10:51:13 +0200 Subject: [PATCH 1/3] drm: Add LCD display clock polarity flags In-Reply-To: References: <1385998768-23512-1-git-send-email-marex@denx.de> <20131202200117.GN16735@n2100.arm.linux.org.uk> Message-ID: <20131203085113.GQ10036@intel.com> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On Mon, Dec 02, 2013 at 03:32:30PM -0500, Rob Clark wrote: > On Mon, Dec 2, 2013 at 3:01 PM, Russell King - ARM Linux > wrote: > > On Mon, Dec 02, 2013 at 04:39:26PM +0100, Marek Vasut wrote: > >> Add DRM flags for the LCD display clock polarity so the pixelclk-active DT > >> property can be properly handled by drivers using the DRM API. > > > > I still say that not even this should be part of the DRM mode API to > > userspace. The hint that you're changing the user API is that you're > > modifying a header file below a 'uapi' directory. > > > > The settings of double scan, sync polarity etc are all part of the > > display mode specification (check CEA-861 documents). Things like > > pixel clock polarity are not part of the mode specification, they're > > a property of the display itself and are independent of the mode. > > > > Therefore, they should not be part of struct drm_mode_modeinfo. > > Note that we could make them part of drm_display_mode (but > drm_crtc_convert_to_umode() should filter them out when copying from > drm_mode_modeinfo.. I agree this information should not be coming from > userspace). Seems like the sort of thing that the bridge or encoder > could set on the adjusted_mode. Probably better to not mix user visible and internal flags in drm_display_mode.flags. Just makes it more difficult to add real flags. Maybe just stick them into private_flags for the drivers that care. -- Ville Syrj?l? Intel OTC