iMac Retina Resolution Issue After Install

[Austin Brons]

On a late 2015 27" Retina iMac (MK482LL/A) with AMD R9 M390 running Sierra 10.12.3. I install refind successfully and when I boot back into Mac OS I find the Apple boot logo is larger, along with all of my icons and windows. I was able to install RDM and change resolutions but the display lacks the sharpness that it had previously. I'm curious if changing the resolution values in refind.conf will fix this, but I'm not hopeful.

[Austin Brons]

Restarting holding option corrected the resolution issue, changing the resolution in the conf file gave an error.

[Roderick W. Smith]

Resolution problems with recent "Retina" Macs are well-known, and affect not just rEFInd, but also GRUB and Windows. See this AskUbuntu question for (slightly) more on the topic. Unfortunately, without access to the hardware, I'd have a hard time investigating it, so I'm hoping somebody with the necessary programming skill and the hardware takes up the challenge; or maybe I'll eventually buy such a system -- but they aren't trivial purchases, as you know. FWIW, this problem is noted on "The Future of rEFInd" documentation page.

[Roderick W. Smith]

This is a long shot, but could you try the following test build of rEFInd:

http://www.rodsbooks.com/refind-bin-0.10.5.1.zip

...or if you prefer a USB image...

http://www.rodsbooks.com/refind-flashdrive-0.10.5.1.zip

This version adds code to make rEFInd more cautious about setting the screen resolution, which might be the source of the problem. To test it properly, it must:

• Be the first instance of rEFInd launched. Thus, it must either replace the current rEFInd on your disk or be launched from a USB flash drive or some other location. If another rEFInd instance (or another boot manager, like GRUB) has already run, then the test is invalid.
• Be configured to not set the screen resolution -- thus, no resolution or textmode option in refind.conf.

If this version fixes the problem, I'll continue testing it. (The resolution-setting code is very finicky and works differently on every system, so it's conceivable that my tentative fix will break something else on other computers.)

Once again, the fix I've implemented here is a long shot. I'm simply bypassing a function call that jumps out at me as being a possible source of the problem; but there are dozens of other system calls that could be the real source of the problem.