[Acer PT14-51 Laptop] HowTo

[benmacfreak]

i did test last night the config and acpi you did for me jala, it tries to boot (im using sonoma 14.7 from a usb) without nblue or weg fork it goes to a blank screen only. should i add reset nvram driver?

[jalavoui]

the config i did was based on my oc old 9.6 version. try to change it for the version you have

 

this is my oc confg

 

it's old just get OC + OC confgurator that can let you do proper configs

 

if you read this thread you will see everyone does mistakes. so we talk about it here and try to find a solution

 

asus for instance took 5 pages to figure out his nvram delete mistake

 

so ben i think if you can fix the new config your system will work just fine

Edited November 2, 2024 by jalavoui

[benmacfreak]

1 minute ago, jalavoui said:

the config i did was based on my oc old 9.6 version. try to change it for the version you have

ok ill do that now.

[Stezza88]

1 hour ago, jalavoui said:

you dont patch the binary directly use wg or nblue todo it.

 

did u make ghidra analyse the binary 1st?

 

just testing for this

 

 

if i try this in the ventura binary from decompkernelcachede

 

 

ghidra failed to decompile the code and i get this

 

 

 

this is bad luck but you can load on ida pro

 

 

Hope IDA 7.7 could do its dirty job too.. now i'm setupping some things..

Edited November 2, 2024 by ASUS Vivobook

[jalavoui]

dont try in windows get ida pro for mac os it's free. better not use windows ever

[Stezza88]

.

 

Edited November 3, 2024 by ASUS Vivobook

[Stezza88]

I have found this... it was "com.apple.driver.AppleIntelICLGraphics : __ZN16IntelAccelerator10getGPUInfoEv + 0x3ed"

And maybe i thought that 0x3ed is the offset inside this function..

 

Edited November 3, 2024 by ASUS Vivobook

[Stezza88]

In my code for example, i don't have this sequence of byte so the lshbluesky patch won't apply

// Sku Bypass IntelAccelerator::getGPUInfo
static const uint8_t f2[] = {
	0x0F, 0x87, 0x17, 0x01, 0x00, 0x00, 0x48, 0x8D, 0x0D, 0x96, 0x02, 0x00, 0x00
};

when i find, it's empty the search

 

Edited November 3, 2024 by ASUS Vivobook

[Stezza88]

I did my patch, and it worked!! I've followed lshbluesky methodology (pointing to a safe location) in the other post to do it..

// Sku Bypass IntelAccelerator::getGPUInfo
static const uint8_t f2[] = {
	0xE8, 0x38, 0x0C, 0x06, 0x00, 0xE9, 0x2D, 0x01, 0x00, 0x00
};
static const uint8_t r2[] = {
	0xC7, 0x83, 0x98, 0x11, 0x00, 0x00, 0x08, 0x00, 0x00, 0x00
};

Noi i have got a new kp... "Graphics Firmware Load failed boot hash check!\"

Kernel-2024-11-03-115328.panic

 

But this has no instructions associated... it could be a problem..

 

Edited November 3, 2024 by ASUS Vivobook

[Stezza88]

With -disablegfxfirmware boot arg kp disappeared but It loops on "busy timeout iGPU"

 

Edited November 3, 2024 by ASUS Vivobook

[jalavoui]

that's interesting you're using scheduler for apple firmware and icl is trying to load it. kinda like linux

are you using wg from other thread?

i hacked it long ago to load tgl guc but it sure miss lots of work cause this wg patches are for old os x version. try scheduler 4 also it's non apple firmware. 5 is nblue tgl default

 

 

it's interesting your card is trying todo same as linux (firmware loading)

did you try with AppleIntelTGLGraphics also ?

 

igpu hang also happens on wg for framebuffer only? fix this first

you can also try to load icl or AppleIntelTGLGraphics without the framebuffer driver for testing (will produce some logs)

 

hope you can make nblue to work - much easier way to patch kexts

 

this is the correct. the offset of panic is always printed in kp msg - i think i posted something about this with a calculator print

 

make sure you check for the bytes in the binary for repeating. if f2 repeats in binary on some other function the patch will probably fail

sometimes is better to use bigger pattern to avoid that

Edited November 3, 2024 by jalavoui

[Stezza88]

1 hour ago, jalavoui said:

that's interesting you're using scheduler for apple firmware and icl is trying to load it. kinda like linux

are you using wg from other thread?

i hacked it long ago to load tgl guc but it sure miss lots of work cause this wg patches are for old os x version. try scheduler 4 also it's non apple firmware. 5 is nblue tgl default

 

 

it's interesting your card is trying todo same as linux (firmware loading)

did you try with AppleIntelTGLGraphics also ?

 

igpu hang also happens on wg for framebuffer only? fix this first

you can also try to load icl or AppleIntelTGLGraphics without the framebuffer driver for testing (will produce some logs)

 

hope you can make nblue to work - much easier way to patch kexts

 

[WEG] I've forked the forked weg source code to work for Raptor Lake too..

kern_igfx.cpp

kern_model.cpp

Edited November 3, 2024 by ASUS Vivobook

[Stezza88]

1 hour ago, jalavoui said:

that's interesting you're using scheduler for apple firmware and icl is trying to load it. kinda like linux

are you using wg from other thread?

i hacked it long ago to load tgl guc but it sure miss lots of work cause this wg patches are for old os x version. try scheduler 4 also it's non apple firmware. 5 is nblue tgl default

 

 

it's interesting your card is trying todo same as linux (firmware loading)

did you try with AppleIntelTGLGraphics also ?

 

igpu hang also happens on wg for framebuffer only? fix this first

you can also try to load icl or AppleIntelTGLGraphics without the framebuffer driver for testing (will produce some logs)

 

hope you can make nblue to work - much easier way to patch kexts

 

this is the correct. the offset of panic is always printed in kp msg - i think i posted something about this with a calculator print

 

[NBlue] But the .bundle files must be loaded too? In folder GPUBundle? How can I copy the AppleIntelTGLGraphics into S L E ?

 

[EDIT] With nblue (this is proof with com.apple.driver.AppleIntelICLLPGraphicsFramebuffer from S/L/E) is more complicated to me because i get no kp (= can't patch) .. i hang on this (same with *LE drivers)

 

Edited November 3, 2024 by ASUS Vivobook

[jalavoui]

you gotta break os x seal to install to /S/L/E. bad idea for now as you can't load any framebuffer

you can try /sle/AppleIntelTGLGraphics.kext + bundles +hookcase into /L/E just for testing. patches are in nblue

 

idk if putting bundles in /Library/GPUBundles would to anything put in /L/E 1st

 

hope you can solve that igpu hang issue either with wg or nblue

 

found a link error with nblue. gonna apply as nred - read other thread

Edited November 3, 2024 by jalavoui

[Stezza88]

[WEG] Tryed scheduler 4 and 5 .. nothing changed... same hang.. here some logs dump ..

Lilu_1.6.9_23.6.txt

Edited November 3, 2024 by ASUS Vivobook

[Stezza88]

What i obtain if i do this boot arg? Could be important?

1. Use the -igfxfbdump boot flag to dump native and patched framebuffer table to ioreg and then File->Import->IOReg Dump menu

[jalavoui]

i think wg framebuffer settings dont work for icl but give it a try

[Stezza88]

[NBlue] with ICLLE or TGLLE

"Iokit daemon stall 'iGPU'"

"Iokit daemon stall 'iGPU'"

 

Edited November 4, 2024 by ASUS Vivobook

[Stezza88]

[NBlue] Using latest test NBLue kext setting only Gen7ICL IOPRimaryMatch 0xA7A08086, GraphicScheduler 4 : got an unsupported ICL SKU kp

Kernel-2024-11-04-221337.panic

Edited November 4, 2024 by ASUS Vivobook

[jalavoui]

nice kp. check nblue::init it uses a specific card id. also patched on icl frame

u might wanna change this values if trying to load icl graph kext

 

 

for icl graph i use to test with this card id 

 

icl graph code as SYSLOG_COND. i bet the patched bytes won't match your os x version. change to panic_cond and test

 

latest nblu bin src is updated- check it

 

btw this functions is not working correctily. 

Gen11::getGPUInfo(void *that)

 

check tgl graph patch for num slices,etc. and use same logic for icl graph kext

you card values sould show up in linux logs

Edited November 4, 2024 by jalavoui

[Stezza88]

NootedBlue     nblue: @ stolen_size 0x8000000
NootedBlue     nblue: @ kextG11HW Failed to apply patches!
NootedBlue     NBlue: @ Failed to apply AGDP board-id patch

Edited November 4, 2024 by ASUS Vivobook

[Stezza88]

I applied my icl sku bypass patch ant it worked!!

Now I got new kp (GraphicsSchedulerSelect = 4)

GUC Binary Load Failure

Kernel-2024-11-04-234536.panic

Edited November 4, 2024 by ASUS Vivobook

[Stezza88]

- with GraphicsSchedulerSelect = 5 no kp but stucks here

 

 

- with GraphicsSchedulerSelect = 2 got kp unsupported scheduler select

 

- with GraphicsSchedulerSelect = 0 it booted! but same problems.. but I got a faster pc! Always with no QE/CI No metal

 

Edited November 4, 2024 by ASUS Vivobook

[jalavoui]

very interesting. what about scheduler 0 (apple)

 

Failed to apply AGDP board-id patch

gonna check this. no idea why it fails. it's a very old wg path also used on nred

 

btw your nblue source is old. download from other thread (page 24). the one you have says "it's bad"

they're together

 

i wonder what icl graph is doing cause the ig-platform should be unsupported 0x9a49 is ment for tgl graph kext

 

i got what you did. you load the graph kext without the framebufer. its is possible but only to get some logs.

u will need to have the framebuffer loaded. or it might just do work for something

Edited November 4, 2024 by jalavoui

[Stezza88]

Same result with newer version.. btw I've got always IONDRVSupport.kext.zipas loaded framebuffer

and always

NootedBlue     nblue: @ stolen_size 0x8000000
NootedBlue     nblue: @ kextG11HW Failed to apply patches!
NootedBlue     NBlue: @ Failed to apply AGDP board-id patch

btw it boots...............

Edited November 4, 2024 by ASUS Vivobook