Op 08-06-16 om 15:27 schreef Ville Syrjälä: > On Wed, Jun 08, 2016 at 02:15:25PM +0100, Chris Wilson wrote: >> On Wed, Jun 08, 2016 at 01:41:46PM +0300, ville.syrjala@xxxxxxxxxxxxxxx wrote: >>> From: Ville Syrjälä <ville.syrjala@xxxxxxxxxxxxxxx> >>> >>> Move the encoder cloning check to happen earlier in the modeset. The >>> main benefit will be that the debug output from a failed modeset will >>> be less confusing as output_types can not indicate an invalid >>> configuration during the later computation stages. >>> >>> For instance, what happened to me was kms_setmode was attempting one >>> of its invalid cloning checks during which it asked for DP+VGA cloning >>> on HSW. In this case the DP .compute_config() was executed after >>> the FDI .compute_config() leaving the DP link clock (1.62 in this case) >>> in port_clock, and then later the FDI BW computation tried to use that >>> as the FDI link clock (which should always be 2.7). 1.62 x 2 wasn't >>> enough for the mode it was trying to use, and so it ended up rejecting >>> the modeset, not because of an invalid cloning configuration, but >>> because of supposedly running out of FDI bandwidth. Took me a while >>> to figure out what had actually happened. >> Did it reject the 1.62 link clock when you simply removed the >> check_encoder_cloning()? Just checking... :) > Nope. The rejection only happened due to exceeding the max fdi lane > count. I think I had a warn in there somewhere when I originally > submitted the fdi==port_clock patch, but that apparently didn't > survive into the version that eventually got merged. > For patch 1-11: Reviewed-by: Maarten Lankhorst <maarten.lankhorst@xxxxxxxxxxxxxxx> _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/intel-gfx
