> And when the user provides an EDID one should parse it and set the default
> resolution to match it. But that's a less important feature.
It's more complex than you might think, and (to me personally) it seems
to require more time than its importance justifies.
https://bugzilla.redhat.com/show_bug.cgi?id=1749250
Read the thread. Actually, I wrote some EDID parsing code a while ago, but that's before QEMU supporting EDID so I had to do it outside QEMU and pass my parsing result to ramfb as the now-removed starting_width / starting_height. In the context QEMU, the EDID actually reflects the user preference since the whole structure is usually made up from the user-specified resolution. And I think most guest OSes initialize first-time-seen monitors to their EDID resolution, which should have motivated QEMU to provide an EDID for a virtual monitor.
But at this point it's kind of awkward to do the EDID / resolution handling (that I need) in the ramfb driver as the kvmgt EDID has to be read out from the i915 MMIO just like a physical GPU. Guess now my use case is better covered with a fully functional i915 framebuffer driver for OVMF. If I had the time...