Iris Xe iGPU on Tiger Lake successfully loaded ICLLP Frambuffer and VRAM also recognizes 1536MB! + However, some issues.

[jalavoui]

dam production also as this bug that i patched in debug

it's a small bug but gonna patch it anyway

 

looking at depthFromAttribute() - trying to guess why the first null come from here

Edited November 2, 2024 by jalavoui

[Visual Ehrmanntraut]

Just now, jalavoui said:

dam production also as this bug that i patched in debug

What's the bug?

Oh, regOff < rmmioSize - 4 instead of 8. 

[jalavoui]

yeah that was easy to fix.

 

this part of code is where all bad things start to happen and make me patch the plane/scaler calls

gonna waste some time with this

the china code (**(code **)(*(long *)in_RDI[0x948] + 0x130)) looks like a writereg() to me

 

gonna control values from this fastwritereg also

Edited November 2, 2024 by jalavoui

[Visual Ehrmanntraut]

My database says
0003b198            let field_4a08_1: *mut AppleIntelDisplayPath = ADJ(rax_29)->field_4a08;
0003b19b            data_d97c8 += 1;
0003b1ad            let rax_30: i32 = AppleIntelFramebufferController::getTranscoderOffset(fController_7, rsi_34, field_4a08_1->fPipe, 0x1000);
0003b1b2            let fRegAccessManager_1: *mut AppleIntelRegisterAccessManager = this->fRegAccessManager;
0003b1b9            data_d9050 += 1;
0003b1c0            data_dbeb8 += 1;
0003b1d3            let mut rcx_15: *mut *mut AppleIntelDisplayPath;
0003b1d3            
0003b1d3            if this->field_3fd8 == 0xffffffff {
0003b1ec                data_dbed8 += 1;
0003b1f3                rcx_15 = &this->field_49f0;
0003b1d3            } else {
0003b1d5                data_d9058 += 1;
0003b1dc                data_dbed0 += 1;
0003b1e3                rcx_15 = &this->field_4a08;
0003b1d3            }
0003b1d3            
0003b209            fRegAccessManager_1->vt->WriteRegister32(fRegAccessManager_1, (rax_30 + 0x70008) as u64, *(ADJ(rcx_15)->field_4a08.byte_offset(0x154) as *mut i64));
 

y'know what just take the bndb lol
AppleIntelTGLGraphicsFramebuffer.bndb.tar.zst

[jalavoui]

i do hope other devs decide to come here and write 2 lines of code...

 

omg why this happening

 

grrrr

Edited November 2, 2024 by jalavoui

[Visual Ehrmanntraut]

1 minute ago, jalavoui said:

i do hope other devs decide to come here and write 2 lines of code...

That's a good one. Wait, you're joking right? 😄 

[jalavoui]

no, this is what i get from trying to read code - bugs !

 

 

gonna download bninja again

 

note to self: b4 donwload a new release read the pop up - i need a old release

Edited November 2, 2024 by jalavoui

[Visual Ehrmanntraut]

17 minutes ago, jalavoui said:

omg why this happening

 

grrrr

I'm using BN 4.2.6335

[jalavoui]

this is latest

 

[Visual Ehrmanntraut]

2 minutes ago, jalavoui said:

this is latest

 

Oh, no. I'm using latest development version of Binary Ninja Personal, not free

Edited November 2, 2024 by Visual Ehrmanntraut

[jalavoui]

think you can port db to free version ? other devs might use it

[Visual Ehrmanntraut]

Not really, sorry. Go buy Binary Ninja : -)

[jalavoui]

k i'll think on something else

 

this better not be a bug - i hate conflicts

Edited November 2, 2024 by jalavoui

[joevt]

Where is the ApplelntelTGLGraphicsFramebuffer binary?

[jkbuha]

3 hours ago, joevt said:

Where is the ApplelntelTGLGraphicsFramebuffer binary?

 

@Chito Kim posted a snapshot here

[jalavoui]

hi joevet

 

it's in the sle_Internal.zip page 24

 

in that folder /lep is the production version that visual is studying

/sle are debug versions

/sle are signed unmodified versions

 

btw i've check your patches and i can't guess how you find out this

are you overwritting apple default in /S/L with user default in /L/E ?

this is very interesting

 

speaking about patches u giys need to follow this 

https://github.com/moraea/non-metal-frameworks

 

this is a very good work

 

also

https://github.com/ChefKissInc/NootedRed

is bug free

 

btw visual do you remenber howto config this ? i mean os x boot args

 

i know there's other tool but i'd like to add this code to nblue

i wrote about it long ago and dont remenber howto setup boot args and most important

which tool to use in terminal to link to network and get live logs

idk why u remove such a great tool from nred

 

i'm also thinking on adding this to nblue

idk why but can't get lilu to print results...

old sources from https://github.com/NyanCatTW1/NootedRed/blob/nyanDbg2/NootedRed/kern_nred.cpp

 

any1 as an Apple ethernet driver with kdp protocol ? i kinda miss this

 

Edited November 2, 2024 by jalavoui

[joevt]

15 hours ago, jalavoui said:

btw i've check your patches and i can't guess how you find out this

are you overwritting apple default in /S/L with user default in /L/E ?

this is very interesting

 

User overrides are in /L/Displays and system overrides are in /S/L/Displays.

 

SwitchResX modifies user overrides. It can override EDID and some other stuff related to the display. User overrides is used by CoreDisplay.framework to create display modes. If a User override doesn't exist, then it uses system overrides.

 

System overrides is used by displaypolicyd for .mtdd files and for some other settings.

 

I wanted to test changes to .mtdd files. I tried making mtdd files in the user overrides location but nothing happened, so I grepped for the overrides location to find that displaypolicyd was responsible for loading mtdd files and that it wouldn't load mtdd files from the user overrides location. I needed a method to modify system overrides. I figured it would be easier to make displaypolicyd always use user overrides (requires copying system overrides to the user overrides location). I don't think what I wanted to achieve actually worked though (dual tile 4K 144Hz). But there might be a use for this patch.

 

I have a script to get a list of all the properties that exist in the overrides. I need to make a list of all the properties that CoreDisplay and displaypolicyd use (search for CFDictionaryGetValue?) and compare to see which are unique to each.

 

 

15 hours ago, jalavoui said:

any1 as an Apple ethernet driver with kdp protocol ? i kinda miss this

 

Maybe you can add a serial COM port?

https://github.com/acidanthera/bugtracker/issues/1954

 

Or maybe Firewire kprintf or fwkdp or fwkpfv?

I haven't tried either recently.

 

[jalavoui]

this is from old thread. most of us don't have a apple driver for our etherner cards and as such we miss iokerneldebugger and can't do proper debugging

visual can you check what parts of this code is needed so it will work on RealtekRTL8111 ? it's mieze driver

i can't decompile this better...

 

omfg i just can't belive this. the kext still works on ventura

i loaded in opencore. just added my card id

but acer physically disable the cable connector so i can't use this

 

 

here's the kext if someone wanna use iokerneldebugger

parameters are in the 2016 thread above

 

AppleRTL8169Ethernet.kext.zip

 

Edited November 3, 2024 by jalavoui

[jalavoui]

i'm doing a review on nblue dependencies

 

so to build nblue first get some files from (type this in terminal)

git clone --recursive https://github.com/ChefKissInc/NootedRed.git

 

files will be on your home folder (NootedRed). next move Lilu + MacKernelSDK from nred to nblue folder (replace)

then build nblue. no need build lilu

 

 

bug, etc complain to this guy

 

great now can't load nblue... fixing...

so kp+kp+kp -> fixed by using nred base xcode.proj

...

Edited November 3, 2024 by jalavoui

[Visual Ehrmanntraut]

8 hours ago, jalavoui said:

visual can you check what parts of this code is needed so it will work on RealtekRTL8111 ? it's mieze driver

 

AppleRTL8169Ethernet_high_level_lang.txt

[Visual Ehrmanntraut]

I would assume you need to override the ::enable(IOKernelDebugger*) variant. From a quick look in RTL8111 I don't see that existing. There are also calls to ::lock and ::unlock.

[jalavoui]

tks. now that i ported nred base to nblue i'm getting all sort of lilu bugs.

i wonder how it was working b4. hmm maybe cause i have al kexts installed and lilu didnt crash cause of that

 

pain continues

 

got this theory and i would like an opinion.

all working igpus have this property with a non zero value (this is nred sample)

 

this apple driver at start() do something to config space (same as nred as seen in logs)

 

i there a way to enable this iopciexpresslinkstatus() for the igpu. maybe this is checked by os x and a lot of things depend on this setting

 

this is my new lilu issue...

 

Edited November 3, 2024 by jalavoui

[Visual Ehrmanntraut]

1 hour ago, jalavoui said:

this is my new lilu issue...

accessed invalid or read-only memory (you probably forgot to call MachInfo::setKernelWriting)

[jalavoui]

well old nblue as this guy name

 

 

and new nblue

 

 

and ofc it is crashing

 

renaming didnt solve the issue. same kp

gonna try debug it

 

wonder what changed from old xcodeproj to new nred proj

 

hmm i doubt this kp address is correct

 

 

code is fresh from nred

 

 

 

 

 

any1 with older source plz upload it -  i got no backup of old code

Edited November 6, 2024 by jalavoui

[joevt]

All Lilu based kexts need to be built with the same version of Lilu that is injected by Open Core.

I use sym links for MacKernelSDK and Lilu for building each of the Lilu based kexts with my forks.

 

myprojects=/Volumes/Work/Programming/KextProjects
lilukext="$myprojects/Lilu/joevt-Lilu/DerivedData/Lilu/Build/Products/Debug/Lilu.kext"
# build lilu with Xcode
mkdir -p "$myprojects/WhateverGreen"
cd "$myprojects/WhateverGreen"
git clone https://github.com/joevt/WhateverGreen.git "joevt-WhateverGreen"
cd "$myprojects/WhateverGreen/joevt-WhateverGreen"
rm Lilu.kext    2> /dev/null ; ln -s "$lilukext" "Lilu.kext"
rm MacKernelSDK 2> /dev/null ; ln -s "$myprojects/MacKernelSDK/joevt-MacKernelSDK" "MacKernelSDK"