Jump to content
559 posts in this topic

Recommended Posts

Posted (edited)

I'm so happy @jalavoui, now need to resolve this graphic problem...

12 minutes ago, Visual Ehrmanntraut said:

Sorry, but, you used leaked kexts and AI, "great milestone"... sure, if you ignore the fact apple already did 98% and the AI did the 0,9999% rest of the work for it. But it's always that last 1% that takes the longest.

I'm not trying to be toxic here, it's really just I'm not sure what the return would be for you, since you burned so much money and it's surely not going to be stable, at least not for long periods of time, even if you fix the swizzled-looking output.

But even if it all works out, at least it'd rid myself of additional work, that almost everyone expects me to do for Hackintosh.

 

 

I can confirm it has been a LOT OF WORK ... my legs are breaking
Also, IA is not so easy to make it work, sometimes you need to ask it to revert, sometimes you can collaborate with it, sometimes you need to show her the path...

Decide to resolve what route and what not it has been a big human work... in 2024 with jala hints I reached the cursor on black screen without ia in RPL machine, I think I'm the first in the world, like jala said.

With jala hint and lshbluesky works I did in 2024 my first binary patch... believe me I think jala is a master!

Edited by Stezza88
  • Like 1

But you don't understand, programming is an art form, and by using those tools we help those greedy evil soul-less companies/people replace what's essentially technology artists.
You see, I still do everything myself because this is my passion; it's like paintings for people's computers. I see my pure efforts bringing joy and smiles to people and that's what keeps me working on them, along with the learning journey. I don't even use Hackintosh anymore.

btw, I'm sure I would've done it myself quite fast, probably faster than this, provided you all waited and maybe also provided me with gen11 hardware. I just have a backlog of things I'm doing, including the iOS emulator, and a lack of budget (to buy hardware and equipment).

4 minutes ago, Stezza88 said:

I accept you... u can continue to talk in the lshbluesky post if not accept this  work...

Sorry, I'm not entirely sure what you're trying to say here. With the broken English I can only understand it as a "stop talking if you're not going to help or accepting what's happening here". However, I'm just here because I'm currently on break and bored, and I already intended to stop talking and move to only observing after that last comment.

Found a strong lead. PLANE_STRIDE 0xa0->0x14 — the driver swapped stride from 0xa0 (LINEAR: 25604 / 64 = 160) to 0x14 (Y-TILED: 25604 / 512 = 20 tiles). If PLANE_CTL tiling bits don't agree with that stride, scanout reads linear memory as if Y-tiled (or vice versa) — exactly your corruption pattern.

Posted (edited)
55 minutes ago, Visual Ehrmanntraut said:

Sorry, I'm not entirely sure what you're trying to say here. With the broken English I can only understand it as a "stop talking if you're not going to help or accepting what's happening here". However, I'm just here because I'm currently on break and bored, and I already intended to stop talking and move to only observing after that last comment.

 

To finish, I put everything open source hoping someone will improve all of this work to create a BIG STABLE LIBRARY FOR ALL INTEL PRODUCTS

 

AS NOOTEDRED already does..

Edited by Stezza88
Posted (edited)
12 minutes ago, Stezza88 said:

To finish, I put everything open source hoping someone will improve all of this work to create a BIG STABLE LIBRARY FOR ALL INTEL PRODUCTS

 

Should I interpret that as an "ok" to using the knowledge for the "Big Stable Library for all Intel Products"™ that I already intended for NootedRed to have? :-P

Edited by Visual Ehrmanntraut
Posted (edited)

Btw, the reason I'm intending to do that is because I'm already slowly building SDKs for doing this stuff through NootedRed, and if time allows it I'll eventually entirely replace all Acidanthera kexts with the kind of "ULTRA STABLE LIBRARY FOR ALL HACKINTOSH"-type kexts, but also the usual automagic configuration-less operation.

Edited by Visual Ehrmanntraut
  • Like 2
3 minutes ago, Stezza88 said:

Use this work for RPL machines....

Jumping from feature to feature will cause tech debt, it's not good for software stability and developer sanity.

I'm slowly designing and building tools to make everything easier and cleaner to do.

image.thumb.png.f0e671d4ca45d0a8c5caccf6eddbfae1.png

  • Thanks 1
22 minutes ago, Visual Ehrmanntraut said:

Jumping from feature to feature will cause tech debt, it's not good for software stability and developer sanity.

I'm slowly designing and building tools to make everything easier and cleaner to do.

image.thumb.png.f0e671d4ca45d0a8c5caccf6eddbfae1.png

 

I wish you all the best, but now we are in NootedGreen thread I remember you

Posted (edited)

I needed AI for the work I'm doing in the place I live... so I paid annual features.. rate limited but at least I have" a friend" which can help me also there...

Edited by Stezza88
12 minutes ago, Stezza88 said:

 

I wish you all the best, but now we are in NootedGreen thread I remember you

Look at the copyright comments in most of the files in your repository.

In other words, you use my code, so I assume mentioning future updates was relevant.

Posted (edited)
7 minutes ago, Visual Ehrmanntraut said:

Look at the copyright comments in most of the files in your repository.

 

Sources I used were deliberately given to me from this forum... I don't believe to have broken any copyright

8 minutes ago, Visual Ehrmanntraut said:

Look at the copyright comments in most of the files in your repository.

In other words, you use my code, so I assume mentioning future updates was relevant.

 

I never met you... so it's the first time I see you... ok I will add credits also to you

Edited by Stezza88
Posted (edited)
4 minutes ago, Stezza88 said:

I never met you... so it's the first time I see you... ok I will add credits also to you

You've never seen me before? But I made the AMD iGPU and Navi 22 dGPU support...

Edited by Visual Ehrmanntraut
Posted (edited)

I was in an hour pause.. now i start to produce..

The next move on this issue is the AppleIntelPlane::setupParams decomp (or whatever fills PLANEPARAMS[+0x18]) — the only piece of evidence that would actually tell us the real buffer layout. Until then, dp0 = corrupted login + HD cursor is the steady state.

 

SetupParams is the same function I analyzed earlier — it writes CRTCParams offsets only:

  • [r15+0] TRANS_CLK_SEL
  • [r15+4] TRANS_DDI_FUNC_CTL
  • [r15+8] TRANS_DDI_FUNC_CTL2
  • [r15+0Ch] TRANS_MSA_MISC
  • [r15+28h] PIPE_SRCSZ ✓ (matches paramsSurfCompare's CRTCParams [+0x28] = PIPE_SIZE)

But paramsSurfCompare's PLANE_STRIDE field is at PLANEPARAMS [+0x18] (different struct, different argument), and PLANE_CTL at PLANEPARAMS [+0x00]. Neither offset is touched in SetupParams. So this function — though large — still doesn't tell us what sets the stride or tiling.

The PLANEPARAMS-builder lives elsewhere. The most likely places:

  1. AppleIntelPlane::SetupParams — different class, same name pattern (this is AppleIntelBaseController::SetupParams)
  2. AppleIntelPlane::buildParams / AppleIntelPlane::setupParams
  3. AppleIntelPlane::SetupSurfaceInfo or similar surface-builder
  4. hwSetMode / hwSetStream themselves (the callers of paramsSurfCompare) — they may inline the PLANEPARAMS construction before calling compare
Edited by Stezza88
Posted (edited)

i'm gonna just focus on code as all other questions are not why i'm on insanlymac.

 

when u got time check nblue. the dam mtl bundle in /l/e is not chached so patches wont work

fine for now as i only did a small patch that i moved to original info.plist.

 

time will tell what direction we need togo.

gonna check logs but i think u gonna get the frameb to work and if so its world first

 

u are aware that if this work u can also fix icl ?

very good options ahead

Edited by jalavoui
  • Like 1
Posted (edited)

Fragmentation changed -> we have found the right spot to hook!

 

7 minutes ago, jalavoui said:

i'm gonna just focus on code as all other questions are not why i'm on insanlymac.

 

when u got time check nblue. the dam mtl bundle in /l/e is not chached so patches wont work

fine for now as i only did a small patch that i moved to original info.plist.

 

time will tell what direction we need togo.

gonna check logs but i think u gonna get the frameb to work and if so its world first

 

u are aware that if this work u can also fix icl ?

very good options ahead

 

We are here to make the INTEL HISTORY ON MACOSX dear jala

Edited by Stezza88

Spectacular result. Apple's stack has flipped to Y-tiled:

Before After V200
PLANE_CTL = 0x94000408 (bit 10 X-tiled) PLANE_CTL = 0x94001008 (bit 12 Y-tiled-legacy) ✓
PLANE_STRIDE = 0x14 (X-tile units) PLANE_STRIDE = 0x50 (Y-tiled 128B units = 80 × 128 = 10240 B/line) ✓

But now our V99R is destroying it:

  • V99R: 0x50 → 0x280 — multiplies by 8, gives wrong byte stride (640 × 128 = 81920 B/line, 8× too wide)
V99R[Cdp]: 0x94001008 → 0x94000008 — clears Y-tiled bit, makes scanout linear while buffer is Y-tiled V99S hardware override still forces STRIDE=0xa0, CTL=linear

Apple's now-correct Y-tiled values are being clobbered to wrong-linear values by our old code. Strip them so Apple's path passes through clean.

  • V200 (new): sets fb[+0x4A18] = 1 in AppleIntelFramebufferinit (both pre- and post-original) → Apple's stack now produces Y-tiled buffer + Y-tiled PLANE_CTL (0x94001008) + Y-tiled stride (0x50)
  • V99R/V99S/V99f/V99F: passthrough for STRIDE/CTL — no rewriting, just diagnostic logging
  • V99R[P] non-aperture redirect + V99G GGTT remap (dp0 plumbing) preserved
  • All other dp0 / cursor-relevant logic preserved (Path B/C, V55, V97, V103, V195)

Boot with the same args. Look for these in the next log:

V200[1]: fb=... fb[+0x4A18]=1 forced
V99R[S1]: PLANE_STRIDE 0x50 passthrough (Y-tiled 128B units)
V99R[C1]: PLANE_CTL 0x94001008 passthrough (tiling=4)
V99S[1]: SURF arm 0x...: pre-arm STRIDE=0x50 CTL=0x94001008

Visually: should be a coherent, non-fragmented login screen with HD cursor preserved. If anything regresses to corruption, the new log lines will show whether Apple is still writing Y-tiled values or not.

×
×
  • Create New...