[Intel-gfx] new behaviors: ?2.99.917?

[Felix Miata]

Chris Wilson composed on 2014-12-24 07:07 (UTC):

> On Wed, Dec 24, 2014 at 12:22:38AM -0500, Felix Miata wrote:

>> 4 series onboard video rev 03 (G41)
>> using DVI to HDMI cable on DVI motherboard connector, which Xorg reports as HDMI1
>> Display is 1920x1080 LCD TV, using HDMI connector #4
>> Rawhide
>> kernel 3.18.1
>> Xorg 1.16.3RC1
>> xorg-x11-drv-intel-2.99.917

>> 1-connected and unconnected absent from Xorg.0.log

> That's not in any -intel. You are misinterpreting verbiage in the
> logfile if you think that is what 2.21.15 says.

Methinks you misinterpreted my comment at the time you replied. Sometime
after v2.20.19 the string "connected" disappeared from Xorg.0.log files,
while remaining all along in logs created using radeon and nouveau drivers.
Since your reply, Rawhide's driver moved from v2.99.917-1 to v2.99.917-2, and
the string returned to the logs.

>> 2-EDID preferred absent from Xorg.0.log

> That's not used by any -intel.

I see "First detailed timing is preferred mode" in logs generated using
radeon and nouveau, and those generated using intel thru v2.20.19 and in
v2.99.917-2, so absence from logs generated in the interim using intel makes
me wonder why. Maybe only Xorg rather than driver is what uses it, but the
difference when looking at files only from the interim period when they were
absent made this the more logical place to ask.

>> 3-EDID preferred is not used by Xorg

> Indeed.

So if -intel does not and Xorg does not, what does?

>> 4-video= on cmdline if present and valid is used by Xorg instead of EDID
>> preferred

> Xorg inherits the current mode.

This is a primary reason for the OP, a distinctly different behavior from
using radeon and nouveau. I can understand the logic in it, but will be
bothered by it as long as the difference among drivers remains, and would
really rather it never happen that the others change to do the same as what
intel started doing . It's a long tradition that the kernel cmdline options
affecting framebuffer vtys (vga= pre-KMS, video= using KMS) not be inherited
by Xorg. Those have quickly and easily at boot time enabled the vty rows and
columns to be indirectly adjusted to provide a comfortable text size without
impacting the much more elaborately configured X desktops. Current intel
behavior requires an additional Xorg 'Section "Device"' option to get the
configured DisplaySize onto the default output in place of the automagic that
worked not so long ago in intel, and still can in radeon and nouveau.

Logs supporting my comments:
http://fm.no-ip.com/Tmp/Linux/Xorg/Igfx/
-- 
"The wise are known for their understanding, and pleasant
words are persuasive." Proverbs 16:21 (New Living Translation)

 Team OS/2 ** Reg. Linux User #211409 ** a11y rocks!

Felix Miata  ***  http://fm.no-ip.com/