[Guide] Alternative to the minStolenSize patch with 32mb DVMT-prealloc

[gongzhen]

And there are something interesting that some graphics have cursors of 0mb. For HD620 that is 00002002 00000000 which is 34+0 by default. I tried 32+0 but it panics. Obviously 32-32=0 which leaves nothing for cursor and causes kernel panic. And for 22+0 it works just as 22+9. Will it help to explain that starting from SKL some codec may have floating cursors sizes?

And I also have a question that although I'm sure I have injected 0x59160000 and it works, but in system preference shows Device ID: 0x5916 and Revision ID: 0x0002. Is that only a detection for correct hardware or the real 02001659 works after DVMT pre-alloc changed?

It shows as Iris HD 640 or 650 (I forgot that) in some software required graphic acceleration such as Adobe lines. It that also true for HD630 hackintosh as MBP 15,3 have vanilla HD630 alongside dGPU? But I'm using MacBook10,1 and it should only feature a HD615. Some animate transition is glitch while that's also true for my friends using Mac, so I suppose the HD615 should have more natural transition however it failed and treat that as iris HD Graphics.

 

[RehabMan]

Hi, Rehabman. In July I had been disturbing you about the issue with HD620 graphics of 2880*1920 on my laptop. The old minStolenSize patch finally works so I abandon this new way. However, the minStolenSize patch broke since on 10.13.2 beta and I managed to figure it out even though I know less about reverse engineering.
Find: 896db8764648
Replace: 896db8ed4648

And now I have nearly finished courses on this semester and decide to re-install the system. So I started to dig into this new way patch. The so-called freeze is probably the graphics refused to recognize the high resolution at boot time(CoreDisplayFixup.kext is present for my situation). If I connect it to an external monitor, it will output normally(with 591b0000 it may glitch and crash soon, which is weird) and the built-in display is still blank. After several cycle of tests, I found by raising 19mb fb and 9mb cursor to 22+9 or 24+8 (I haven't test the minimum combination required but I can confirm 21+9 is not working so far)will get it working. It seems the fb sizes corresponding to the screen sizes and should be expanded for proper functional.

I wonder how will these affect graphics? You mentioned 19+9 comes from Azul, so it is true that the fb.kext will simply pass the remaining process to other part or other kexts as long as the fb and cursors sizes meet the need of built-in display? And the rest of 32-19-9 would have nothing to do so is it right to fill that fully with fb and cursor (for my case is 24+8)?

Yeah, I did some similar experiments with fbsize/cursorsize with a KBL laptop I had here for a while (it was a Lenovo 710-11). And definitely there is a requirement for larger fbsize with larger screen resolution (I was experimenting with HiDPI modes on 1080p native display).

We can only guess and experiment to discover what works for various screen resolutions.

And there are something interesting that some graphics have cursors of 0mb. For HD620 that is 00002002 00000000 which is 34+0 by default. I tried 32+0 but it panics. Obviously 32-32=0 which leaves nothing for cursor and causes kernel panic. And for 22+0 it works just as 22+9. Will it help to explain that starting from SKL some codec may have floating cursors sizes?

And I also have a question that although I'm sure I have injected 0x59160000 and it works, but in system preference shows Device ID: 0x5916 and Revision ID: 0x0002. Is that only a detection for correct hardware or the real 02001659 works after DVMT pre-alloc changed?

It shows as Iris HD 640 or 650 (I forgot that) in some software required graphic acceleration such as Adobe lines. It that also true for HD630 hackintosh as MBP 15,3 have vanilla HD630 alongside dGPU? But I'm using MacBook10,1 and it should only feature a HD615. Some animate transition is glitch while that's also true for my friends using Mac, so I suppose the HD615 should have more natural transition however it failed and treat that as iris HD Graphics.

I'm guessing that the 0 cursor bytes means "calculate automatically". It is all guessing as we don't have source to the kext. And yes, those zero cursor sizes started in the SKL kexts, from what I remember.

The revision ID has nothing to do with ig-platform-id.

 

[ttavyy]

First of all, thank you for all the awesome guides you have made @RehabMan.

1) Could you please tell me where I can find more details about "ig-platform-id".

2)
"
So... a patch for Broadwell 0x16260006:

Comment: 0x16260006, 32MB BIOS, 19MB framebuffer, 9MB cursor bytes (credit RehabMan)
Name: AppleIntelBDWGraphicsFramebuffer
Find: 06002616 01030303 00002002 00005001
Replace: 06002616 01030303 00003001 00009000
"

How do I find that "06002616 01030303 00002002 00005001" code? What are the tools needed for this guide?

3) My signature: Thinkpad T450, i5 5300u, intel HD5500.
Do you recommend High Sierra or Sierra for this hardware?

 

[RehabMan]

1) Could you please tell me where I can find more details about "ig-platform-id".

What kind of details are you looking for.
It is simply a mechanism to select a framebuffer configuration within Intel graphics framebuffer kexts, HD4000 and later.

How do I find that "06002616 01030303 00002002 00005001" code?

You can find it with a hex editor.
I use Hex Fiend.

3) My signature: Thinkpad T450, i5 5300u, intel HD5500.
Do you recommend High Sierra or Sierra for this hardware?

Your choice.

 

[gi10]

Hey @RehabMan

I've been trying to install sierra 10.12.6 all day

For the installation use ig-platform-id defined to 0x12345678

Now I am trying to use your patch 32mb. windows tells me that it is 128mb but I think it is not reliable ... but I only get kp

What I can be doing wrong?


Clover folder attached

 

[RehabMan]

Hey @RehabMan

I've been trying to install sierra 10.12.6 all day

For the installation use ig-platform-id defined to 0x12345678

Now I am trying to use your patch 32mb. windows tells me that it is 128mb but I think it is not reliable ... but I only get kp

What I can be doing wrong?


Clover folder attached

You must activate the patch correctly:
- boot with invalid ig-platform-id (it can be changed in the Clover Options -> Graphics Injector)
- DO NOT use FakeID!
- rebuild cache:

sudo kextcache -i /

- reboot normally

 

[gi10]

You must activate the patch correctly:
- boot with invalid ig-platform-id (it can be changed in the Clover Options -> Graphics Injector)
- DO NOT use FakeID!
- rebuild cache:

sudo kextcache -i /

- reboot normally

I did that but kp continue...

 

[RehabMan]

Im nm

Huh?
You will need to write questions in English.

 

[gi10]

Huh?
You will need to write questions in English.

Sorry for last answerd, my cat decided playing with keyboard haha. Solved using 0x16260004 its a problem use that and not 0x16260006??. Time to patch ssdt off topic, i will make new thread.

 

[RehabMan]

Sorry for last answerd, my cat decided playing with keyboard haha. Solved using 0x16260004 its a problem use that and not 0x16260006??. Time to patch ssdt off topic, i will make new thread.

The patch you have in your config.plist is specific to 0x16260006. Unless you changed it, it will not work with 0x16260004.
Generic patch is in post #1 and an even more generic patch is in the plists provided by my Clover guide.

But ig-platform-id varies... trial/error is the way.