[Bug 215001] Regression in 5.15, Firmware-initialized graphics console selects FB_VGA16, screen corruption

[bugzilla-daemon at bugzilla.kernel.org]

https://bugzilla.kernel.org/show_bug.cgi?id=215001

--- Comment #5 from Javier Martinez Canillas (javier at dowhile0.org) ---
Created attachment 300235
  --> https://bugzilla.kernel.org/attachment.cgi?id=300235&action=edit
[PATCH] video: vga16fb: Only probe for EGA and VGA 16 color graphic cards

Can you try the attached (untested) patch?

The commit you mention in Comment 3 didn't change any logic but just moved the
platform device registration that match the vesafb and efifb drivers.

As far as I can tell, the problem is that the vga16fb driver doesn't check if
the video modes are one that's supported by it and just blindly attempts to
probe the driver.

This probably wasn't a problem before because the sysfb.o was in
arch/x86/kernel but no is in drivers/firmware. But the issue in vga16fb was
present before, just moving exposed it.

I believe that the correct thing is to check if screen_info says that's a
{E,V}GA 16 color mode and fail to probe the vga16fb driver otherwise.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are on the CC list for the bug.