From: nunojsilva@invalid.invalid
On 2026-01-15, Robert Riches wrote:
> On 2025-12-26, John-Paul Stewart wrote:
>> On 2025-12-24 11:15 p.m., Robert Riches wrote:
>>>
>>> Does anyone have favorite tutorials or other helps for dealing
>>> with a high-DPI transition?
>>>
>>> I'm Running Devuan (currently Daedalus, hopefully soon Excalibur)
>>> using plain X and fvwm2, no DE as such with 1920x1200 monitors.
>>> Just purchased a round of 4k monitors, so I'll be needing to
>>> adjust xterm fonts, web browser issues, other GUI application
>>> issues, etc.
>>
>> It's been a couple years since made a similar switch. But as far as I
>> recall, there's not much to do other than configuring X. I had to
>> manually specify the monitors' physical size (in millimetres) in my
>> xorg.conf so that X could calculate the correct DPI. Then everything
>> that's measured in points just kind of fell into place. (E.g., there
>> was nothing to be done for LibreOffice, etc.)
>
> After waiting for adapters to arrive (by "super saver" shipping)
> and getting the 1.1GB system logs after physically installing the
> 4K monitors, I got them running at 1920x1200 by putting
> "xrandr --output ... --mode 1920x1200" and a couple of sleep
> commands in my .Xclients file. Omitting the xrandr commands
> brings up X in 3840x2160 mode, but the DPI is stuck at 96.
>
> Xorg.0.log showed the EDID size information being fetched correct
> as 597 x 336 mm. However, Xorg.0.log showed the DPI being set to
> (96, 96).
>
> So, I created an xorg.conf. Evidently, X11 wants things
> specified in terms of two monitors side by side--thus the 1194 by
> 336. After some experimenting, I put this in /etc/X11/xorg.conf:
>
> vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv cut here vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv
> Section "Screen"
> Identifier "Default Screen Section"
> Monitor "Asus PA279CV"
> EndSection
>
> Section "Monitor"
> Identifier "Asus PA279CV"
> VendorName "Asus"
> ModelName "PA279CV"
> DisplaySize 1194 336
> EndSection
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ cut here ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>
> With that, grepping /var/log/Xorg.0.log for words EDID, mm, and
> DPI yields this:
>
> vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv cut here vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv
> [179297.003] (II) AMDGPU(0): EDID for output DisplayPort-0
> [179297.003] (II) AMDGPU(0): EDID Version: 1.3
> [179297.003] (II) AMDGPU(0): clock: 594.0 MHz Image Size: 597 x 336 mm
> [179297.003] (II) AMDGPU(0): clock: 148.5 MHz Image Size: 597 x 336 mm
> [179297.003] (II) AMDGPU(0): clock: 297.0 MHz Image Size: 597 x 336 mm
> [179297.003] (II) AMDGPU(0): clock: 241.5 MHz Image Size: 597 x 336 mm
> [179297.003] (II) AMDGPU(0): Number of EDID sections to follow: 1
> [179297.003] (II) AMDGPU(0): EDID (in hex):
> [179297.006] (II) AMDGPU(0): EDID for output DisplayPort-1
> [179297.006] (II) AMDGPU(0): EDID Version: 1.3
> [179297.007] (II) AMDGPU(0): clock: 594.0 MHz Image Size: 597 x 336 mm
> [179297.007] (II) AMDGPU(0): clock: 148.5 MHz Image Size: 597 x 336 mm
> [179297.007] (II) AMDGPU(0): clock: 297.0 MHz Image Size: 597 x 336 mm
> [179297.007] (II) AMDGPU(0): clock: 241.5 MHz Image Size: 597 x 336 mm
> [179297.007] (II) AMDGPU(0): Number of EDID sections to follow: 1
> [179297.007] (II) AMDGPU(0): EDID (in hex):
> [179297.007] (II) AMDGPU(0): EDID for output DisplayPort-2
> [179297.007] (II) AMDGPU(0): EDID for output HDMI-A-0
> [179297.007] (**) AMDGPU(0): Display dimensions: (1194, 336) mm
> [179297.007] (**) AMDGPU(0): DPI set to (163, 163)
> [179297.465] (II) event6 - Logitech Trackball: device set to 300 DPI
> [179297.561] (II) event6 - Logitech Trackball: device set to 300 DPI
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ cut here ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>
> The (163, 163) is the correct DPI figure. An earlier DisplaySize
> value of 597 336 had yielded a DPI figure of (326, 163). With
> either of the DisplaySize values, and even with Xorg.0.log
> showing the correct DPI, xdpyinfo still reports DPI as 96x96.
>
> So, it appears that even when Xorg.0.log is showing the correct
> DPI, but there's apparently a disconnect between that and
> xdpyinfo.
>
> Any suggestions? Thanks in advance!
I've seen this happen with radeon (the driver), logs showed the correct
information being retrieved via EDID, and then the resolution would be
reverted to some hardcoded 96.
For that reason, I've had either
xrandr --dpi VALUE
or
xrandr --dpi OUTPUT_NAME
in .xinitrc, to either force a specific DPI value or get it set again
from EDID information from the named output.
I think I never got around to figure out why is the driver - or some
other piece? - apparently overriding the EDID resolution like this
(makes one wonder if remarks about X11 not handling "hi-dpi" well could
be in part due to this nowadays... after all under X11 one could use
different resolutions two decades ago already).
--
Nuno Silva
--- SoupGate-Win32 v1.05
* Origin: you cannot sedate... all the things you hate (1:229/2)
|
