Hi
Am 22.10.25 um 15:12 schrieb Hans de Goede:
Hi Thomas,
On 11-Oct-25 5:17 PM, Thomas Zimmermann wrote:
Hi Hans
Am 11.10.25 um 12:02 schrieb Hans de Goede:
Hi,
A while ago I did a blogpost about not having the native GPU drivers in
the initrd: https://hansdegoede.dreamwidth.org/28291.html
With the Fedora /boot partition size increase there has been renewed
interest in this and as a result I got an interesting comment on
the blog post pointing out that there are EFI protocols to get EDID
info for the monitor (presumably for the one the GOP has lit up).
It would be useful to see if we can hook this up so that the simpledrm
device can report the physical dimensions of the screen and the boot
splash does not need to guess the DPI and thus does not need to guess
whether to do hiDPI scaling or not. The comment is here:
https://hansdegoede.dreamwidth.org/28291.html?thread=14467#cmt14467
and it points to:
https://uefi.org/sites/default/files/resources/2_-_AMD_UEFI_Plugfest_EDID_Spring2012.pdf
https://uefi.org/specs/UEFI/2.9_A/12_Protocols_Console_Support.html
Unfortunately I do not have time to work on this myself, hence this
email in the hope that someone else can pick this idea up.
There are now efidrm and vesadrm, which support specifically the features that
we get from the rsp. standards, hardware and bootloader. For VESA and EFI the
EDID can be transferred from grub via bootparams. A patch [1] is on its way
into upstream grub. The Linux kernel stores that data in edid_info, from where
it is picked up by efidrm and vesadrm. If the EFI framebuffer comes directly
from the EFI calls, we need to implement this in the kernel, but the principle
is the same. I also have ideas for making this work on Coreboot.
For simpledrm, I'd rather strip if from the ability to use EFI/VESA
framebuffers and focus on simple-framebuffer DT nodes. These sometimes come
with DPI information. Maybe we could extend the DT to support EDID as well.
[1] https://lists.gnu.org/archive/html/grub-devel/2025-10/msg00043.html
I have long been working on all this and it is now slowly coming together.
Cool, thank you for working on this!
I agree that if we are going to have efidrm and vesadrm that
making simpledrm a DT platform thing only makes sense.
I do wonder about the choice to primarily make the EDID passing
a grub thing though.
Some distros are actively trying to move away from grub since
grub is quite a complex beast and they are trying to move to
something simpler like systemd-boot. As such it would be a good
idea IMHO if efidrm could directly get the EDID from EFI itself.
Yes, the GRUB patch is mostly useful for VESA. For EFI on x86, I posted
a patch series to fetch the EDID in the EFI stub and forward it to the
kernel. Have a look at [1]. For non-x86 EFI systems, I also have a
prototype, but wanted to get the x86 done first. I have to some research
if systemd-boot requires anything extra.
Things are moving in the right direction.
Best regards
Thomas
[1]
https://lore.kernel.org/lkml/[email protected]/
Note I'm afraid I don't really have time to work on this myself.
Regards,
Hans
--
--
Thomas Zimmermann
Graphics Driver Developer
SUSE Software Solutions Germany GmbH
Frankenstrasse 146, 90461 Nuernberg, Germany
GF: Ivo Totev, Andrew Myers, Andrew McDonald, Boudien Moerman
HRB 36809 (AG Nuernberg)
