[PATCH] modesetting: Add common 16by9 and 16by10 resolutions to mode list

[Michael Cronenworth]

On 08/22/2017 02:28 PM, Adam Jackson wrote:
> On Tue, 2017-08-22 at 13:20 -0500, Michael Cronenworth wrote:
>> Related:https://bugzilla.redhat.com/show_bug.cgi?id=1339930
> Nak, at least in this form. If these are really common modes then the
> DMT spec or similar should define real timings for them, and we should
> use those timings rather than pretend GTF can handle it. For example:
>
>> +    { 15360, 8640 },
> hyoscyamine:~% gtf 15360 8640 60
>
>    # 15360x8640 @ 60.00 Hz (GTF) hsync: 536.16 kHz; pclk: 11675.42 MHz
>    Modeline "15360x8640_60.00"  11675.42  15360 16824 18568 21776  8640 8641 8644 8936  -HSync +Vsync
>
> An 11GHz dotclock is slightly unrealistic.
>
> Ideally xf86DefaultModes would be replaced by DMTModes (and the latter
> synced with a newer version of that list, eg from drm), because then
> you wouldn't have needed to change the ms driver at all.

Thanks for the review. It was adopted from the Intel driver.

Today, the ms driver and xf86GetDefaultModes() grabs 4:3 modes only and appends the 
current display mode on the end, which happens to be a 16:9 mode on the system I'm 
using. Using a 2560x1440 mode to play any 3D application on an Intel IGP does not 
net a lot of frames per second. Downscaling to a 4:3 mode to end up with black bars 
on the monitor is not ideal either.

What direction do you want this patch to take? Update the DMT list (if it needs it) 
and switch xf86GetDefaultModes() to use DMT?


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.x.org/archives/xorg-devel/attachments/20170822/df25a204/attachment.html>