[Bug 288900] loader.efi: System won't boot without kvm or console=comconsole in loader.conf

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=288900

Vidith Hundekar <hundekarvidith@gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |hundekarvidith@gmail.com

--- Comment #2 from Vidith Hundekar <hundekarvidith@gmail.com> ---
Hi,

I’m Vidith, currently working on the EFI boot counter project as part of GSoC
exploration, and while going through the loader and early boot path I came
across this issue.

I was trying to understand how console initialisation works in loader.efi when
no display is attached.

From the description and the follow-up comment, it seems like when no monitor
is connected, the firmware may not expose a usable EFI console — no GOP or no
usable Simple Text Output Protocol. If loader.efi assumes ConsoleOut is always
valid and tries to use it during early initialisation without checking, that
would explain the hang or reboot loop, since this is all still happening in the
BootServices phase where the loader depends entirely on what the firmware
provides.

The comconsole workaround seems to support this — it sidesteps EFI console
usage entirely and takes a different code path, which is why it works even when
there's no serial port physically present.

My question is whether loader.efi currently verifies that a valid console
handle exists (for example, via HandleProtocol on ConsoleOut) before attempting
to use it, or whether it assumes one is always present. If it's the latter,
would it make sense for the loader to fall back more gracefully when firmware
exposes no console?

I'm likely missing some of the finer details of how console selection is
handled internally, so any clarification would be helpful.

Thanks

-- 
You are receiving this mail because:
You are the assignee for the bug.