[80% Solved] Iris Xe iGPU on Tiger Lake successfully loaded ICLLP Frambuffer and VRAM also recognizes 1536MB! + However, some issues.

Mastachief · September 12

Its not supposed to hang, its supposed to panic, take a look below at mine

jalavoui · September 12

guys who as this dam lcd display with

image.png.22a65bfbb7ec019b9d7c1dc8d870e1ab.png

the driver uses this "feature" that we don't need

i'm checking linux code

i think we need to get rid of this bs and use part of old icl code

image.png.1bbc90a5dcea23d268fd9b0e09dad385.png

SO I'LL W8 FOR INTEL

image.png.2e382a217946b23ffa051907b5112b22.png

Edited September 12 by jalavoui

ArtikDiamond · September 12

@Mastachiefis there a feature in nblue to disable dgpus? Like -wegnoegpu but for nblue. Without that macOS will try to use our dgpu and we can't test it...

BitBass · September 12

7 hours ago, ArtikDiamond said:

Wait a second! Me and some other people managed to boot, install and use macOS normally, you should try too

Sure, but some guidance would be appreciated. Let's take it somewhere else though, so we don't muddy the thread with non-dev talk.

Edited September 12 by BitBass

jalavoui · September 12

a feature in nblue to disable dgpus? i think can be done with opencore injection. i write about it but dont remenber where.

i do hope someone get xcode and try a few patches so we can get this done

as for non devs just try install it and post logs

this patch isnt helping

image.png.91e98bc822bf30d89932db137aaf124d.png

doing a review as it seems tobe setting state to zero

need to fix this code plus

add extra

if (conditon)

setportmode(0)

return 0

uint32_t AppleIntelPortHAL::probePortMode()
{
    uint64_t rmmio = getMember<uint64_t>(this, 0x9B0);
    unsigned int uVar3 = *reinterpret_cast<volatile uint32_t *>(rmmio + 0x1638a0);
    unsigned int *puVar2 = getMember<unsigned int*>(this, 0x548);
    unsigned long PVar5;
    unsigned long uVar4;
    
    if (getMember<uint64_t>(this, 0x1be0) != 0)
    {
        unsigned int uVar2 =  *reinterpret_cast<volatile uint32_t *>(rmmio + 0xc4000);
        unsigned int uVar3 = *reinterpret_cast<volatile uint32_t *>(rmmio + 0x44470);
        
        switch(*puVar2) {
        case 0:
                uVar4 = (unsigned long)(uVar2 >> 0x10 & 1);
                break;
        case 1:
          uVar4 = (unsigned long)(uVar2 >> 0x11 & 1);
          break;
        default:
          uVar4 = 0;
          break;
        case 3:
          uVar4 = (unsigned long)((uVar3 >> 0x10 | uVar2 >> 0x18 | uVar3) & 1);
          break;
        case 4:
          uVar4 = (unsigned long)((uVar3 >> 0x11 | uVar3 >> 1 | uVar2 >> 0x19) & 1);
          break;
        case 5:
          uVar4 = (unsigned long)((uVar3 >> 0x12 | uVar3 >> 2 | uVar2 >> 0x1a) & 1);
          break;
        case 6:
          uVar4 = (unsigned long)((uVar3 >> 0x13 | uVar3 >> 3 | uVar2 >> 0x1b) & 1);
          break;
        case 7:
          uVar4 = (unsigned long)((uVar3 >> 0x14 | uVar3 >> 4 | uVar2 >> 0x1c) & 1);
          break;
        case 8:
          uVar4 = (unsigned long)((uVar3 >> 0x15 | uVar3 >> 5 | uVar2 >> 0x1d) & 1);
        }


        if (uVar4==0) {
            //setPortMode)((AppleIntelPort *)this,0);
            if (*(char *)(*(long *)(puVar2 + 0x14) & 8) == 0) {
                //(*this->vt->resetSoftwareState)(this);
                *getMember<uint64_t*>(this, 0x114)=0;
                *getMember<uint64_t*>(this, 0x11c)=0x2000000;
                *getMember<uint64_t*>(this, 0x124)=0;
                *getMember<uint64_t*>(this, 0x12c)=0;
                *getMember<uint64_t*>(this, 0x134)=0;
                *getMember<uint64_t*>(this, 0x140)=0;
                *getMember<uint64_t*>(this, 0x550)=0;

                bzero(getMember<uint64_t*>(this, 0x140),0x400);
                bzero(getMember<uint64_t*>(this, 0xc),0x108);
                bzero(getMember<uint64_t*>(this, 0x558),0x1c);
            }
            *puVar2 = 0;
            
            return 0;
        }
    }
    
    switch(*puVar2) {
    case 0:
    case 1:
            PVar5 = 1;
      break;
    case 2:
            PVar5 = 0;
      if (uVar3 != 0xffffffff) {
          if ((uVar3 & 0x40) == 0) {
          if ((uVar3 & 0x20) != 0) {
          LAB_000b2d13:
            PVar5 = 3;
          }
        }
        else {
        LAB_000b2d3d:
          PVar5 = 2;
        }
      }
      break;
    case 3:

            PVar5 = 0;
      if (uVar3 != 0xffffffff) {
          if ((uVar3 >> 0xe & 1) != 0) {
              goto LAB_000b2d3d;
        }
        if ((uVar3 >> 0xd & 1) != 0) {
            goto LAB_000b2d13;
        }
      }
      break;
    case 4:

            PVar5 = 0;
      if (uVar3 != 0xffffffff) {
          if ((uVar3 >> 0x16 & 1) != 0) {
              goto LAB_000b2d3d;
        }
        if ((uVar3 >> 0x15 & 1) != 0) {
            goto LAB_000b2d13;
        }
      }
      break;
    case 5:

            PVar5 = 0;
      if (uVar3 != 0xffffffff) {
          if ((uVar3 >> 0x1e & 1) != 0) {
              goto LAB_000b2d3d;
        }
        if ((uVar3 >> 0x1d & 1) != 0) {
            goto LAB_000b2d13;
        }
      }
      break;
    default:
      PVar5 = 0;

    }

LAB_000b2d4a:
    
    return PVar5;
    
};

Edited September 12 by jalavoui

ArtikDiamond · September 12

Me and ben are stuck because of the dgpu unfortunately...

jalavoui · September 12

tooke me 3 seconds to find this

try the opencore device injection patch

that was waht i ment

https://dortania.github.io/OpenCore-Install-Guide/extras/spoof.html

this part is what i used and tested

it works very well to disable the dgpu but make sure you put the right path

if working the device will vanish from ioreg

do this using opencore config tool

image.png.64cf3260444781c2a86d1437a724bf1d.png

Edited September 12 by jalavoui

jkbuha · September 12

35 minutes ago, jalavoui said:

tooke me 3 seconds to find this

try the opencore device injection patch

that was waht i ment

https://dortania.github.io/OpenCore-Install-Guide/extras/spoof.html

i've used that spoof method previously to inject the 9A49 platform-id instead of my native one, but same results. Did you intend to use it for another purpose?

Edit: got it, the injection method is to disable dGPU. note that might work for ˜50% of dGPUs, for the rest you will need to apply an ACPI _PS3 or _OFF call depending on your BIOS. My github has more details on how to do it precisely, it follows the OC-Little method...

Edited September 12 by jkbuha

benmacfreak · September 12

2 hours ago, jalavoui said:

a feature in nblue to disable dgpus? i think can be done with opencore injection. i write about it but dont remenber where.

i do hope someone get xcode and try a few patches so we can get this done

as for non devs just try install it and post logs

this patch isnt helping

doing a review as it seems tobe setting state to zero

need to fix this code plus

add extra

if (conditon)

setportmode(0)

return 0

uint32_t AppleIntelPortHAL::probePortMode()
{
    uint64_t rmmio = getMember<uint64_t>(this, 0x9B0);
    unsigned int uVar3 = *reinterpret_cast<volatile uint32_t *>(rmmio + 0x1638a0);
    unsigned int *puVar2 = getMember<unsigned int*>(this, 0x548);
    unsigned long PVar5;
    unsigned long uVar4;
    
    if (getMember<uint64_t>(this, 0x1be0) != 0)
    {
        unsigned int uVar2 =  *reinterpret_cast<volatile uint32_t *>(rmmio + 0xc4000);
        unsigned int uVar3 = *reinterpret_cast<volatile uint32_t *>(rmmio + 0x44470);
        
        switch(*puVar2) {
        case 0:
                uVar4 = (unsigned long)(uVar2 >> 0x10 & 1);
                break;
        case 1:
          uVar4 = (unsigned long)(uVar2 >> 0x11 & 1);
          break;
        default:
          uVar4 = 0;
          break;
        case 3:
          uVar4 = (unsigned long)((uVar3 >> 0x10 | uVar2 >> 0x18 | uVar3) & 1);
          break;
        case 4:
          uVar4 = (unsigned long)((uVar3 >> 0x11 | uVar3 >> 1 | uVar2 >> 0x19) & 1);
          break;
        case 5:
          uVar4 = (unsigned long)((uVar3 >> 0x12 | uVar3 >> 2 | uVar2 >> 0x1a) & 1);
          break;
        case 6:
          uVar4 = (unsigned long)((uVar3 >> 0x13 | uVar3 >> 3 | uVar2 >> 0x1b) & 1);
          break;
        case 7:
          uVar4 = (unsigned long)((uVar3 >> 0x14 | uVar3 >> 4 | uVar2 >> 0x1c) & 1);
          break;
        case 8:
          uVar4 = (unsigned long)((uVar3 >> 0x15 | uVar3 >> 5 | uVar2 >> 0x1d) & 1);
        }


        if (uVar4==0) {
            //setPortMode)((AppleIntelPort *)this,0);
            if (*(char *)(*(long *)(puVar2 + 0x14) & 8) == 0) {
                //(*this->vt->resetSoftwareState)(this);
                *getMember<uint64_t*>(this, 0x114)=0;
                *getMember<uint64_t*>(this, 0x11c)=0x2000000;
                *getMember<uint64_t*>(this, 0x124)=0;
                *getMember<uint64_t*>(this, 0x12c)=0;
                *getMember<uint64_t*>(this, 0x134)=0;
                *getMember<uint64_t*>(this, 0x140)=0;
                *getMember<uint64_t*>(this, 0x550)=0;

                bzero(getMember<uint64_t*>(this, 0x140),0x400);
                bzero(getMember<uint64_t*>(this, 0xc),0x108);
                bzero(getMember<uint64_t*>(this, 0x558),0x1c);
            }
            *puVar2 = 0;
            
            return 0;
        }
    }
    
    switch(*puVar2) {
    case 0:
    case 1:
            PVar5 = 1;
      break;
    case 2:
            PVar5 = 0;
      if (uVar3 != 0xffffffff) {
          if ((uVar3 & 0x40) == 0) {
          if ((uVar3 & 0x20) != 0) {
          LAB_000b2d13:
            PVar5 = 3;
          }
        }
        else {
        LAB_000b2d3d:
          PVar5 = 2;
        }
      }
      break;
    case 3:

            PVar5 = 0;
      if (uVar3 != 0xffffffff) {
          if ((uVar3 >> 0xe & 1) != 0) {
              goto LAB_000b2d3d;
        }
        if ((uVar3 >> 0xd & 1) != 0) {
            goto LAB_000b2d13;
        }
      }
      break;
    case 4:

            PVar5 = 0;
      if (uVar3 != 0xffffffff) {
          if ((uVar3 >> 0x16 & 1) != 0) {
              goto LAB_000b2d3d;
        }
        if ((uVar3 >> 0x15 & 1) != 0) {
            goto LAB_000b2d13;
        }
      }
      break;
    case 5:

            PVar5 = 0;
      if (uVar3 != 0xffffffff) {
          if ((uVar3 >> 0x1e & 1) != 0) {
              goto LAB_000b2d3d;
        }
        if ((uVar3 >> 0x1d & 1) != 0) {
            goto LAB_000b2d13;
        }
      }
      break;
    default:
      PVar5 = 0;

    }

LAB_000b2d4a:
    
    return PVar5;
    
};

i have an apple dev account and sequoia on my laptop, BUT im unsure what to do to patch the kext, i can build it but unsure HOW to patch the code though..

jalavoui · September 12

i deleted old sources and binarys to clean the thread.

latest nblue binary work with icl framebuffer (few posts back)

as for tgl this are latest sources - hope some devs helps by publishing new code

update sle_internal to allow in future double class injection (currently only links tgl portmode)

sle_Internal.zip

Nblue will now allow all possible tgl/icl tests by editing the info.plist

HookCase-master.zip

NootedBlue-master.zip

Edited Monday at 01:39 AM by jalavoui

jkbuha · September 13

16 hours ago, jalavoui said:

i deleted old sources and binarys to clean the thread.

latest nblue binary work with icl framebuffer (few posts back)

as for tgl this are latest sources - hope some devs helps by publishing new code

NootedBlue-master.zip 410.74 kB · 12 downloads HookCase-master.zip 191 kB · 6 downloads

ok, let me try and contribute here. after some configuration, have successfully set up the source code, and built both kexts under xcode.

happy to start looking at which functions need to be looked at/uncommented first. just let me know which are the most important ones to start.

ArtikDiamond · September 13

22 hours ago, jalavoui said:

tooke me 3 seconds to find this

try the opencore device injection patch

that was waht i ment

https://dortania.github.io/OpenCore-Install-Guide/extras/spoof.html

this part is what i used and tested

it works very well to disable the dgpu but make sure you put the right path

if working the device will vanish from ioreg

do this using opencore config tool

I tried both yesterday and today, but igfxutils can't detect my dGPU...

jalavoui · Sunday at 06:35 PM

while i was playing with those apple internal kexts i managed to load the icl framebuufer.

the log is huge depending on IGLogLevel=8 number

but the advantage of getting proper logs + usage of many parameters not present on production version is huge

here's the bin and logs. you need hookcase + previous sle_internal installed to /L/E

now it's so clear why many cards dont work - all is in the new logs

NootedBlue.kext.zip

x.log.zip HookCase.kext.zip

Edited Monday at 01:40 AM by jalavoui

jkbuha · Monday at 11:39 AM

this latest version doesnt detect the iGPU (46a6 or 9a49), and remains looping in text mode without any second boot. no KPs, no blank screen

benmacfreak · Monday at 01:33 PM

1 hour ago, jkbuha said:

this latest version doesnt detect the iGPU (46a6 or 9a49), and remains looping in text mode without any second boot. no KPs, no blank screen

i get the same with my laptop 9a68 device ID

jalavoui · Tuesday at 02:54 PM

theory

image.png.1311d223180089ba3c97baa8e4bed934.png

then in code DDIx is "index"

image.png.61f7a9e8d7e2e59c11e836cef7507342.png

after this auto portindex we get pipes

image.png.07171bb6f88277db6af39dde6f631fdc.png

idk where bus id is used

checking icl v tgs

then the rest of code fails

trying to apply this patch

Edited Tuesday at 08:56 PM by jalavoui

Mastachief · Tuesday at 07:09 PM

From Hackintool BUSID and IndexSection

Below is from the BUSID Section:

Proven by ApplelntelFramcbufferControllor::MapFBToPort, by a call to ApplelntelFramebufferController::getGMBuslDfromPort.

This is GMBUS (Graphic Management Bus) ID described in

Gen 9

https://cdrdv2-public.intel.com/ 690989/intel-gfx-prm-osrc-hsw-display.pdf

Gen 11

https://cdrdv2-public.intel.com/ 705833/intel-gfx-prm-osrc-tgl-vol-12-display-engine.pdf

The use could be found in Intel Linux Graphics Driver source code:

https://github.com/torvalds/linux/blob/6481d5ed076e69db83ca75e751ad492a6fb669a7/drivers/gpu/drm/i915/intel_lrc.c

https://github.com/torvalds/linux/blob/6481d5ed076e69db83ca75e751ad492a6fb669a7/drivers/gpu/drm/i915/i915_reg.h

However, it should be noted that Apple identifiers are slightly different from Linux driver.

In Linux 0 means disabled, however, for Apple it has some special meaning and is used for internal display.

Other than that the values are the same:

• GMBUS_PIN_DPC (4) HDMIC

• GMBUS_PIN_DPB (5) SDVO, HDMIB

• GMBUS_PIN_DPD (6) HDMID

• GMBUS_PIN_VGADDC (2) VGA untilBroadwell inclusive.

So basically you could use 4, 5, 6 for arbitrary HDMI or DisplayPort displays.

Since 5 supports SDVO (https://en.wikipedia.org/wiki/Seriat_Digital_Video_Out), it may also be used to support DVI displays.

Starting with Skylake VGA works via SOVO too (instead of a dedicated GMBUS_PIN_VGAODC id).

Below is from the Index Section:

Watch out, this is really messy (see ApplelntelFramebutferController::MapFBToPort}.

I am not fully sure why this exists, and recommend setting index to array index (i.e. the sequential number from 0).

The only accepted values are 0, 1, 2, 3, and-1 (0xFF).

When index is equal to arrayindex the logic is simple: Port with index 0 is always considered built-in {of LVOS type) regardless of any other values.

Ports with indexes 1-3 are checked against type, HDMI will allow the use of digital audio, otherwise DP is assumed.

Port with index 0xFF is ignored and skipped.

When index!= array index port type will be read from connector[index].type

Say, we have 2 active ports:

0 - [1] buslD 4 type LVDS

1 - [2] buslD 5 type DP

2 - [3] buslD 6 type HDMI

3 - [-1] buslD 0 type Dummy

This will result in 2 framebuffers which types will be shifted:

0 - buslD 4 type DP

1 - buslD 5 type HDMI

In fact BuslD values are also read as connector[index].buslD, but are later mapped back via ApplelntelFramebufferController::getGMBuslDfromPort by looking up a connector with the specified index. The lookup will stop as soon as a special marker connector (-1) is found.

To illustrate, if we have 2 active ports:

0 - [1] buslD 4 type LVDS

1 - [2] buslD 5 typoDP

2 - [-1] buslD 6 type HDMI

3 - I·1] buslD 0 type Dummy

The result will be 2 framebuffers which types and the second buslD will be shifted:

0 - buslD 4 typoDP

1 - buslD 6 typeHDMI

It is also used for port-number calculation.

- LVDS displays (more precisely, displays withCNConnectorAlwaysConnected flag set) get port-number 0.

- Other displays go through index - port-number mapping: 1 - 5, 2 - 6, 3 - 7, or fallback to 0.

By following the above guide, thats how i came up with the Device properties for Intel Plus Graphics G7 and the alldata connectors patch using whatevergreen. instead of Zero the connectors i just assign different ID's to null them out. as the connector information differed between cannonlake and icelake, there were differences with the alldata, it was half the actual amount of values.

<key>PciRoot(0x0)/Pci(0x2,0x0)</key>
			<dict>
				<key>AAPL,GfxYTile</key>
				<data>AQAAAA==</data>
				<key>AAPL,ig-platform-id</key>
				<data>AABSig==</data>
				<key>AAPL,slot-name</key>
				<string>Internal@0,2,0</string>
				<key>AAPL00,override-no-connect</key>
				<data>AP///////wAw5HycAAAAAAAfAQSVIhNgA2+xp1VMniUMUFQAAAABAQEBAQEBAQEBAQEBAQEBLjaAoHA4H0AwIDUAWMIQAAAaHySAoHA4H0AwIDUAWMIQAAAaAAAAAAAAAAAAAAAAAAAAAAAAAAAAAgAMOv8KPH0SFCR9AAAAAEY=</data>
				<key>complete-modeset</key>
				<data>AQAAAA==</data>
				<key>device-id</key>
				<data>U4oAAA==</data>
				<key>device_type</key>
				<string>VGA compatible controller</string>
				<key>disable-agdc</key>
				<data>AAAAAA==</data>
				<key>dpcd-max-link-rate</key>
				<data>CgAAAA==</data>
				<key>enable-cdclk-frequency-fix</key>
				<data>AQAAAA==</data>
				<key>enable-dbuf-early-optimizer</key>
				<data>AQAAAA==</data>
				<key>enable-dpcd-max-link-rate-fix</key>
				<data>AQAAAA==</data>
				<key>enable-dvmt-calc-fix</key>
				<data>AQAAAA==</data>
				<key>force-online</key>
				<data>AQAAAA==</data>
				<key>framebuffer-con0-alldata</key>
				<data>AAAAAAAAAAAAAAAAAAAAAAIAAAAYAAAA</data>
				<key>framebuffer-con0-enable</key>
				<data>AQAAAA==</data>
				<key>framebuffer-con1-alldata</key>
				<data>AgAAAAAAAAABAAAAAQAAAAAEAADBAgAA</data>
				<key>framebuffer-con1-enable</key>
				<data>AQAAAA==</data>
				<key>framebuffer-con2-alldata</key>
				<data>AwAAAAoAAAABAAAAAQAAAAAEAADBAgAA</data>
				<key>framebuffer-con2-enable</key>
				<data>AQAAAA==</data>
				<key>framebuffer-patch-enable</key>
				<data>AQAAAA==</data>
				<key>framebuffer-unifiedmem</key>
				<data>AAAAgA==</data>
				<key>model</key>
				<string>Intel Iris Xe Graphics</string>
			</dict>

Edited Tuesday at 07:33 PM by Mastachief

jalavoui · Tuesday at 09:07 PM

tks. for future ref i found the condition that is used to match port number to gmbus

image.png.71aba61a304d8b9a67bb69f31009228b.png

in code this value is used if not zero

image.png.eeeb3fde2a8fe7deb4baee9cfbe24a74.png

the requestes it calls are

image.png.44425de47ffec5e8278e65650979058e.png

wonder why i dont see this log line anywere

Edited Tuesday at 09:11 PM by jalavoui

[80% Solved] Iris Xe iGPU on Tiger Lake successfully loaded ICLLP Frambuffer and VRAM also recognizes 1536MB! + However, some issues.

Recommended Posts

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites