On 2025/12/26 12:57, Crystal Kolipe wrote: > On Thu, Dec 25, 2025 at 11:10:01PM -0000, Stuart Henderson wrote: > > Happens with some machines, we don't know why. > > If booting without a monitor connected is working with a regular kernel but > failing with the ramdisk kernel, then I'd guess that what is happening is that > the system is trying to use the efifb driver to connect to wsdisplay rather > than a more specific one which is not included in the ramdisk kernel config, > (E.G. inteldrm, radeondrm, or amdgpu). > > If it's using efifb, no EDID data will be being received as there is no > monitor providing it, (a dummy hdmi plug obviously supplies that data, > mitigating the problem). That might be confusing the efifb driver, for > example if the firmware supplies a value of zero for the x resolution then > the driver won't attach. Or maybe the firmware simply doesn't present an > efi framebuffer at all if it doesn't see EDID data from a monitor. > > If it's erroneously using the vga driver on a machine with no physical VGA > hardware or firmware emulation of VGA hardware, then it's quite plausible that > the machine could crash.
It needs someone with hardware to investigate really. Comparing GPT (UEFI, efifb) and MBR (BIOS) installs might give more clues. I wonder if serial console makes a difference. (If it still reboots with serial console that gives more debug opportunities).
